**Molecular Pathways in Cancers**

Editors

**Ion Crist ´obal Marta Rodr´ıguez**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Ion Cristobal ´ Translational Oncology Division Health Research Institute-Fundacion Jimenez Diaz-UAM, Oncohealth Institute Madrid Spain Marta Rodr´ıguez Pathology Health Research Institute-Fundacion Jimenez Diaz-UAM Madrid Spain

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Cancers* (ISSN 2072-6694) (available at: www.mdpi.com/journal/cancers/special issues/MP-Cancers).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-3682-8 (Hbk) ISBN 978-3-0365-3681-1 (PDF)**

© 2022 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**


Expression of Phosphorylated BRD4 Is Markedly Associated with the Activation Status of the PP2A Pathway and Shows a Strong Prognostic Value in Triple Negative Breast Cancer Patients Reprinted from: *Cancers* **2021**, *13*, 1246, doi:10.3390/cancers13061246 . . . . . . . . . . . . . . . . **171**


Reprinted from: *Cancers* **2020**, *12*, 3362, doi:10.3390/cancers12113362 . . . . . . . . . . . . . . . . **247**

### **Preface to "Molecular Pathways in Cancers"**

Despite continuous advances in anticancer therapies, the survival rates in most tumor types remain very poor, especially in those patients with advanced stages of the disease, due to a complex network of alterations that change and increase the oncogenic behaviour of tumor cells. Therefore, a better understanding of the main molecular mechanisms that govern the different human cancer types is and will be the best strategy to improve patient outcomes.

> **Ion Crist ´obal and Marta Rodr´ıguez** *Editors*

### *Editorial* **Advances in the Knowledge of the Molecular Pathogenesis of High-Prevalence Tumors and Its Relevance for Their Future Clinical Management**

**Marta Rodríguez 1,2, \* ,† and Ion Cristóbal 3,4, \* ,†**


**Citation:** Rodríguez, M.; Cristóbal, I. Advances in the Knowledge of the Molecular Pathogenesis of High-Prevalence Tumors and Its Relevance for Their Future Clinical Management. *Cancers* **2021**, *13*, 6053. https://doi.org/10.3390/ cancers13236053

Received: 22 November 2021 Accepted: 24 November 2021 Published: 1 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

This Special Issue aims to include relevant works that increase our knowledge about the molecular pathways that govern the development and progression of high-prevalence human cancers, which are responsible for most cancer-related deaths worldwide. This is one of the ways to provide oncologists with novel therapeutic tools that can improve the clinical management and outcome of cancer patients. In addition to original articles providing relevant results that will be commented upon in more detail below, this Special Issue also contains several review articles that summarize the current state of the art in crucial aspects of human cancer. Thus, the work by Martinez-Useros et al. [1] reviewed epigenetic pathways and treatments (several under study in current clinical trials) that target epigenetic modifications in highly aggressive tumors. Gundamaraju et al. [2] focused in their manuscript on the molecular mechanisms by which the mitochondria influence cancer biology and their usefulness to develop therapeutic strategies. Another relevant challenge, reviewed by Isaguliants et al. [3], is the increased risk of developed cancer observed in people living with human immunodeficiency virus type 1 (HIV-1), despite a long-term successful implementation of antiretroviral therapy. The authors focused on the oncogenic properties of five viral proteins: envelope protein gp120, accessory protein negative factor Nef, matrix protein p17, transactivator of transcription Tat, and reverse transcriptase RT. All proteins either led to the proliferation of pre-existing malignant cells or induced the malignant transformation of normal cells, which is responsible for the carcinogenic effects of HIV-1. Moreover, Senent et al. [4] reported the importance of the complement system in ovarian cancer, highlighting how certain elements of this system play tumor-promoting roles that decrease the efficacy of distinct therapeutic approaches, and discussing the potential usefulness of the complement as a target of treatments for ovarian cancer. In addition, the work by Nishida [5] reviewed the role of oncogenic signaling pathways on the cancer immunosuppressive microenvironment in hepatocellular carcinoma. Interestingly, this manuscript summarizes the molecular factors that could be determining the efficacy of therapies based on immune checkpoint inhibition. Finally, Guijarro-Hernández and Vizmanos [6] carried out a systematic review summarizing the signaling pathways affected in Ph-negative mieloproliferative neoplasms (MPNs). MPNs are driven not only by a constitutive activation of the JAK2/STAT signaling and JAK2 related pathways, but a complex network of non-canonical pathways that affects key cellular functions such as epigenetic and transcriptional regulation, splicing and additional pathways that confer a highly complex and coordinated program in the tumor cells of these blood disorders.

Regarding the contribution of original articles, our Special Issue contains four pieces of work about breast cancer focused on the identification of molecular aberrations that can serve as novel molecular targets and prognostic markers. Thus, Lee et al. [7] reported that the use of a CD99-derived agonist ligand inhibited EGF-induced EGFR dimerization. This issue involved a PTPN12-dependent c-Src/FAK inactivation that impaired cytoskeletal reorganization and suppressed tumor growth in vivo of the triple negative breast cancer cell line MDA-MB-231. Furthermore, Noblejas-López et al. [8] carried out a genomic mapping that evaluated the presence of alterations in 304 splicing-related genes and their prognostic value in luminal breast cancer patients. They identified that amplifications in *CLNS1A*, *LSM1*, and *ILF2* determined poor outcome. At the functional level, they found that these alterations conferred enhanced proliferation in luminal cell lines that can be pharmacologically reversed by using BET inhibitors. In this line of thinking, an increasing number of publications have shown that the use of BET inhibitors could be a therapeutic approach in triple negative breast cancer (TNBC), and that PP2A is a tumor suppressor that directly targets the bromodomain-containing protein 4 (BRD4) regulating its stabilization and activation. The work of Sanz-Alvarez et al. [9] evaluated the clinical impact of BRD4 phosphorylation levels in TNBC patients. Notably, they observed BRD4 hyperphosphorylation in around 34% of cases, and strongly associated with PP2A inhibition status. Moreover, this alteration was markedly associated with patient recurrence and predicted unfavorable prognosis, suggesting the clinical relevance of the PP2A/BET axis as a potential novel marker in TNBC. Considering these results, and the fact that the PP2A pathway has also been previously reported to be affected in luminal breast cancer, it seems that the PP2A/BET interplay could represent a plausible druggable target to develop alternative therapeutic strategies in certain breast cancer patient subgroups from different molecular subtypes. The same research group also published another study in this Special Issue, in this case about HER2-positive breast cancer models. In their work, Sanz-Alvarez et al. [10] evaluated the efficacy of three different PI3K/AKT/mTOR inhibitors (BEZ235, everolimus, and TAK-228) in a panel of HER2-positive breast cancer cell lines with primary and acquired resistance to Trastuzumab. They found promising results combining TAK-228 with Trastuzumab in all resistant cell lines, observing decreased cell proliferation together with increased apoptosis and G0/G1 cell cycle arrest. Considering these results, the combination of Trastuzumab with PI3K/AKT/mTOR inhibitors emerges as a potential alternative strategy to overcome Trastuzumab resistance in HER2-positive breast cancer.

In the context of prostate cancer, the work by Martínez-Martínez et al. [11] provides novel findings about the role of dual specificity phosphatase 1 (DUSP1). The authors demonstrated that this tumor suppressor leads to Snail downregulation and decreased migration and invasion capabilities of prostate cancer cells through the inhibition of c-Jun N-terminal Kinase (JNK) and extracellular-signal-regulated kinase (ERK). Notably, they also found that the subgroup of prostate cancer patients with an expression pattern DUSPhigh/activated JNKlow/activated ERKlow/Snaillow showed better clinical outcome, suggesting its potential utility as molecular marker in this disease. Moreover, Khalil et al. [12] showed relevant results suggesting that the TLK1/NEK1/YAP1 signaling axis plays a key role during the process of androgen-sensitive to androgen-independent conversion, facilitating progression to metastatic castration-resistant prostate cancer. Finally, Papadaki et al. [13] published a comprehensive study in bladder cancer. They found that two secreted extracellular matrix proteins, osteomodulin (OMD), and proline/argininerich and leucine repeat protein (PRELP), were selectively expressed in bladder umbrella epithelial cells but markedly downregulated in bladder cancer cells. These two proteins act as tumor suppressors, regulating epithelial to mesenchymal transition (EMT), which was mediated by the inhibition of the TGF-β and EGF pathways.

Altogether, this Special Issue includes several reviews and unique articles with novel, interesting findings that allow the readers to improve their knowledge about the molecular mechanisms involved in high-prevalence tumors and the recent advances in targeted therapies for these diseases.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Epigenetics of Most Aggressive Solid Tumors: Pathways, Targets and Treatments**

**Javier Martinez-Useros 1, \* , Mario Martin-Galan 1 , Maria Florez-Cespedes <sup>2</sup> and Jesus Garcia-Foncillas 1, \***


**Simple Summary:** The large amount of knowledge regarding epigenetic pathways has opened a broad range of treatments that provide hope for adult patients with highly aggressive forms of solid tumors. The most commonly used treatments for epigenic modifications are based on the specific inhibitors of DNA methyltransferases, azacitidine and decitabine (5-AZA-dC), and on histone deacetylases inhibitors, such as trichostatin A (TSA) or vorinostat (SAHA). However, many other compounds are under investigation, and some are being evaluated in clinical trials. In this review, we have extracted relevant information about epigenetic pathways and treatments that target epigenetic modifications in highly aggressive tumors, as a new hope for these patients.

**Abstract:** Highly aggressive tumors are characterized by a highly invasive phenotype, and they display chemoresistance. Furthermore, some of the tumors lack expression of biomarkers for target therapies. This is the case of small-cell lung cancer, triple-negative breast cancer, pancreatic ductal adenocarcinoma, glioblastoma, metastatic melanoma, and advanced ovarian cancer. Unfortunately, these patients show a low survival rate and most of the available drugs are ineffective. In this context, epigenetic modifications have emerged to provide the causes and potential treatments for such types of tumors. Methylation and hydroxymethylation of DNA, and histone modifications, are the most common targets of epigenetic therapy, to influence gene expression without altering the DNA sequence. These modifications could impact both oncogenes and tumor suppressor factors, which influence several molecular pathways such as epithelial-to-mesenchymal transition, WNT/β–catenin, PI3K–mTOR, MAPK, or mismatch repair machinery. However, epigenetic changes are inducible and reversible events that could be influenced by some environmental conditions, such as UV exposure, smoking habit, or diet. Changes in DNA methylation status and/or histone modification, such as acetylation, methylation or phosphorylation, among others, are the most important targets for epigenetic cancer therapy. Therefore, the present review aims to compile the basic information of epigenetic modifications, pathways and factors, and provide a rationale for the research and treatment of highly aggressive tumors with epigenetic drugs.

**Keywords:** epigenetic; methylation; acetylation; non-coding RNA; small-cell lung cancer; triplenegative breast cancer; pancreatic ductal adenocarcinoma; glioblastoma; metastatic melanoma; advanced ovarian cancer

#### **1. Introduction**

DNA is organized inside the nucleus, in a very complex structure called chromatin. The negative charge of DNA is supported by basic proteins that are rich in arginine and lysine residues, called histones. There are five families of histones and according to their function they are called core histones (H2, H3, and H4) that form the nucleosome core, or linker histones (H1 and H5), which contribute to the condensation of the nucleosome.

**Citation:** Martinez-Useros, J.; Martin-Galan, M.; Florez-Cespedes, M.; Garcia-Foncillas, J. Epigenetics of Most Aggressive Solid Tumors: Pathways, Targets and Treatments. *Cancers* **2021**, *13*, 3209. https:// doi.org/10.3390/cancers13133209

Academic Editor: Michael J. Spinella

Received: 4 May 2021 Accepted: 24 June 2021 Published: 27 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The nucleosome core is composed by two H2A–H2B dimers and a H3–H4 tetramer. The electrostatic attraction between the positively charged histones and negatively charged DNA allows the complex structure of chromatin to form [1,2]. Chromatin is composed of nucleosomes wrapped by 146–147 bp DNA [3]. The H1 histone serves as a linker between the nucleosomes, in order to provide a highly stable chromatin structure [4]. Histones possess amino-terminal tails that allow gene regulation, by epigenetic modifications, due to their flexible shaping [4]. Deregulation in the deposition of histone modification is associated with several human diseases, such as cancer [5]. Moreover, some epigenetic modifications could be influenced by specific molecular pathways involved in cancer, such as epithelial-to-mesenchymal transition (EMT) [6], Wnt/β-catenin signaling [7], the MAPK signaling pathway [8], DNA repair [9], hypoxia [10], and the PI3K–mTOR pathway [11]. Interestingly, some environmental conditions, such as UV exposure or diet, are also able to induce epigenetic changes. For example, compounds such as folate, choline, betaine, and methionine act as cofactors or methyl donors for DNA methylation reactions. A diet rich in resveratrol, curcumin, genistein, epigallocatechin-3-gallate, sulforaphane, and quercetin is able to reactivate certain tumor suppressive genes by inducing DNA demethylation; however, fungi-contaminated agricultural foods contain mycotoxins that may also lead to cancer [12].

Clinical research has achieved several advances in cancer treatment that have led to a longer survival of patients. However, treatment strategies for highly aggressive tumors remains almost constant, without any significant improvements. In the new era of targeted therapy, epigenetic therapies appear as a potential approach for the treatment of highly aggressive tumors, offering new hope for these patients. Methylation and hydroxymethylation of DNA, and histone modifications, are the most common targets of epigenetic therapy, to influence gene expression without any DNA alteration. On the other hand, increasing reports support the use of non-coding RNA as epigenetic treatment to intercept translation, and negatively regulate the expression of oncogenes.

#### *1.1. DNA Methylation*

DNA methylation plays a crucial role in normal cell metabolism; therefore, changes in the methylation status of cells, by methyltransferases, can lead to cell transformation and represent the difference between normal and tumor cells [13] (Figure 1). Cytosine and adenine are the only bases susceptible to methylation. DNA methylation consists of the transfer of methyl groups (-CH3) to the cytosine in position C5, which is followed by a guanine (G). These sites are termed CpG dinucleotides and result in 5-methylcytosine. These sites occur with high frequency in CpG genomic regions. Non-cytosine methylation, such as the methylation of adenine or thymine, appears in very low probability [14]. CpG islands are located in ~60% of human promoters, and methylation of these sites results in a transcriptional repression of the genes [5,15]. Furthermore, 60–80% of CpG islands of somatic cells genome are methylated [16]. The DNA methyltransferase (DNMT) (Figure 1) family regulates the process of DNA methylation [17]. This protein family is composed of the following five members: DNMT1, DNMT2, DNMT3a, DNMT3b and DNMT3L. Interestingly, mutations in some of these members are usually associated with some types of cancer [18]. For example, the DNMT3b subtype is significantly overexpressed in some tumors [19,20]. The methylation status of DNA can be read by MBD (methyl-CpG binding domain) proteins, which are divided into three families. The first family includes MeCP2, MBD1, MBD2, MBD3 and MBD4 [21]; although, MBD3 can only detect hydroxymethylated DNA [22]. The second family is characterized by a BTB domain (also called as the POZ domain) and comprises ZBTB33, ZBTB4 and ZBTB38 [23]. The third family includes the following two proteins: UHRF1 and UHRF2 [24]. Some drugs are able to modulate the expression levels of these proteins. Decitabine and 5-azacytidine trigger calciumcalmodulin kinase (CamK) activity, leading to MeCP2 nuclear export, which induces the epigenetic reactivation of some tumor suppressive genes in colorectal cancer [25]. Other drugs, such as 5-azacytidine, doxorubicin, vorinostat, paclitaxel, or cisplatin, regulate the

expression of different MBD proteins. MBD1 was upregulated after treatment with all those drugs. Downregulation of MBD2 was observed after 5-azacytidine, doxorubicin, or vorinostat treatment, MBD3 downregulation after vorinostat, and the inhibition of MBD4 varied in a time- and drug-dependent manner [26]. Another study reported the decrease in ZBTB4 levels after roscovitine treatment [27]. Concerning UHRF1, its downregulation enables the demethylation, and the subsequent reactivation, of some epigenetically silenced tumor-suppressive genes [28]. Giovinazzo et al. reported the pharmacological inhibition of UHRF1 by the anthracycline derivatives, idarubicin and mitoxantrone [29]. Therefore, several drugs allow the negative modulation of these MBD proteins, implying a high potential to be used as target therapies.

Aberrant DNA methylation has been associated with drug resistance, and as predictive biomarker [30]. Also, inadequate methylation is associated to inflammatory diseases, premalignant lesions and cancer led by chromatin instability [31]. Hypermethylation and hypomethylation of DNA are usual phenomena in cancer; indeed, tumor-suppressive genes are hypermethylated in cancer cells, while they remain hypomethylated in normal cells [32]. Therefore, the demethylation of target genes could be a promising approach in clinical practice. Physiologically, demethylation of DNA sequences is carried out by the ten-eleven translocation (TET) proteins. The three mammalian TET proteins, called TET1, TET2 and TET3, enable the oxidation of 5-methylcytosine (5mC) of nucleic acids, to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC) or 5-carboxylcytosine (5caC) [33]. The mutation or inhibition of TET proteins is associated with aging and tumorigenesis [34]. Indeed, mutation in *TET2* is frequently found in hematopoietic malignancies [35], and the downregulation of TET proteins has been observed in several solid tumors, such as breast cancer, gastric, glioblastoma, liver, lung, melanoma and prostate [34,36–38].

**Figure 1.** Schematic representation of the DNA methylation process. DNMTis inhibits DNA methylation by downregulation of DNMTs. The action of DNA methylation inhibitors (DNMTs inhibitors and TET proteins) triggers a chromatin-remodeling process and chromatin structure becomes transcriptionally accessible to RNA polymerase II, which will begin the transcription process. DNMTis: DNA methyltransferases inhibitors. DNMTs: DNA methyltransferases. TET: ten-eleven translocation proteins. RNAPII: RNA polymerase II. Me: methyl. Ac: acetyl.

#### *1.2. Histone Modification*

Histone modification can take place in the following two locations: the flexible tails of the nucleosomes and the internal sites in the core of the histone (Figure 2) [39]. The residues most susceptible for modification are lysine and arginine residues, and hydroxyl group-containing serine/threonine/tyrosine residues [40]. Histone modification includes several reactions, such as the methylation and acetylation of lysine and arginine residues, phosphorylation of threonine and serine residues, SUMOylation of lysine residues, isomerization of proline residues, ADP-ribosylation, ubiquitylation, citrullination, deamination, formylation, O-GlcNAcylation, propionylation, butyrylation and crotonylation [41]. Histone acetylation of lysine limits the interactions between the histones H3 and H4, and DNA; while deacetylation leads to gene inactivation [42]. Acetylation is associated with active transcription, and facilitates the recruitment of co-regulators and elements to promote transcription. Modifications of histones are driven by protein effectors and are crucial in the regulation of gene expression. HATs (histone acetyltransferases) are a group of effectors that transfer the acetyl groups to lysine residues of histones [43]. Notably, aberrations in the histone modification pattern may induce cancer [44]. For example, tumor cells present a loss of Lys16 acetylation and Lys20 trimethylation of histone H4 at the early phase of tumor initiation [45]. In contrast, histone deacetylases are another group of effectors that remove the acetyl groups from acetyl-lysine residues, which allows DNA to wrap tightly to histones [46]. Histone deacetylases (HDACs) have been recently reported as a target for cancer therapy (Figure 2) [46]. HDAC1-11 and other histone deacetylases, termed sirtuins, normally play a role as gene silencers [47]. Other effectors are histone demethylases that remove methyl groups from lysine residues. The lysine-specific demethylase 1 (LSD1) exhibits tumor-prone abilities in glioblastoma, and its inhibition sensitizes tumor cells to vorinostat, increasing apoptosis [48]. Other histone demethylases, such as KDM4, produce genome instability, while KDM6 is considered a tumor-suppressive factor [49].

On the other hand, readers of these modifications determine the functional outcome of specific epigenetic change. Some of the proteins involved in the recognition of histone modifications are BET (bromodomain and extraterminal domain-containing). This family is composed of four proteins (BRD2, BRD3, BRD4 and BRDT), and plays important roles in tumor development, since they also lead to transcriptional activity [50,51]. For this reason, BET inhibitors have been evaluated as anti-tumor therapies, showing encouraging results in several malignancies, without significant toxicities or adverse events (Figure 2) [51].

**Figure 2.** Schematic representation of the main histone modification processes. Both acetylation and methylation positively and negatively regulate gene transcription according to the methylated or acetylated residue (up). Several drugs have been designed to allow chromatin remodeling by the inhibition of BET, HDAC or HMT proteins that condense chromatin and hamper transcription (down). Histones acetylation and cytosines unmethylation will result in an open chromatin structure and gene transcription is active. BET: bromodomain and extra-terminal motif (BET) proteins. HDAC: histone deacetylases. HMT: histone methyltransferase. DNMTs: DNA methyltransferases. HAT: histone acetyltransferase. HDM: histone demethylase.

#### *1.3. Non-Coding RNA*

This family includes several factors, but the most notable, in regards to cancer, are small interfering RNA (siRNAs), microRNAs (miRNAs), PIWI-interacting RNA (piRNAs), and long non-coding RNAs (lncRNAs) (Figure 3) [5].

The small interfering RNA (siRNA) transcripts are double-stranded RNA fragments, about 21–25 base pairs long. The function of siRNA is thought to be related to erasing viral double-stranded sequences to avoid infection. SiRNA is cleaved by Dicer from long double-stranded RNA sequences [52]. The double-stranded siRNA is processed by the RNA-induced silencing complex (RISC), to produce single-stranded siRNA [53]. This strand is able to recognize the target mRNA. The perfect match induces mRNA degradation, and a partial match results in translational repression [54].

MiRNA are the most known non-coding RNA and they are involved in several cell functions. Several miRNAs are linked to cancer initiation and development. Furthermore, miRNAs can be tumor-prone or tumor-suppressive factors [55]. MiRNAs are very similar to siRNAs; however, miRNAs originate from double-stranded RNA hairpins, rather than long double-stranded RNA that need additional manipulation by DROSHA [56].

P-element-induced wimpy testis (PIWI) proteins belong to the Argonaute (AGO) family and were discovered in the germline [57]. They also bind a unique type of noncoding small RNAs, called piRNAs (PIWI-interacting RNAs). This tandem, composed of PIWI and piRNAs, constitute the piRNA-induced silencing complex (piRISC). PiRNAs

are special mediators, because depending on the factors that modulate, some piRNAs are considered oncogenic, while others are considered tumor-suppressive factors [58].

Long non-coding RNAs (lncRNAs) constitute a huge subgroup of ncRNAs, defined as RNA transcripts, with more than 200 nucleotides [59]. LncRNAs play an important role in the development of various cancers [60]. The lncRNA, HOTAIR, is closely related to epigenetic modifications. The knockdown of HOTAIR activates transcription-reducing H3K27 trimethylation [61]. Moreover, HOTAIR is able to interact with lysine-specific histone demethylase 1A (LSD1) [62]. Aberrant HOTAIR expression has been observed in several tumors, and its positive expression has been associated with several hallmarks of cancer, such as high cell proliferation, angiogenesis or drug resistance, by the direct regulation of several downstream factors involving multiple signaling pathways [63–65]. Another crucial lncRNA is MALAT-1 (metastasis-associated lung adenocarcinoma transcript-1), which is aberrantly upregulated in multiple tumor types, and yields high proliferative and metastatic profiles [66]. High expression of MALAT-1 has been associated with high-grade and advanced-stage melanoma, glioma and lung cancers [67–69].

**Figure 3.** Schematic representation of non-coding RNA. LncRNA targets mRNA to inhibit translation or degrade mRNA (left). PIWI proteins stabilize piRNAs and lead to post-translational control (middle). MiRNA are originated from doublestranded RNA hairpins. The ribonuclease III enzyme, DROSHA, binds and cleaves hairpin structures in primary RNA transcripts into precursor miRNAs. Once transported to cytoplasm, precursor miRNAs are processed by DICER into mature miRNAs that regulate expression of mRNA (right). miRNA: microRNA. DICER: ribonuclease III enzyme. DROSHA: ribonuclease III enzyme. RISC: RNA-induced silencing complex. piRNA: PIWI-interacting RNA.

> The large amount of knowledge regarding epigenetic modifications has opened a broad palette of treatment strategies for the most aggressive solid tumors in adulthood. Thus, the objective of this review is to compile basic knowledge about epigenetic pathways and treatments, and provide a rational for further clinical trials, based on the use of these treatments in highly aggressive solid tumors.

#### **2. Epigenetic Modulation in Highly Aggressive Solid Tumors**

The most commonly drugs used as hypomethylating agents are specific inhibitors of DNA methyltransferase (DNMT), for example, azacitidine and decitabine (5-AZA-dC) (Figure 1) [70]. These drugs lead to a reduction in whole DNA methylation status [71], and damage DNA by inducing genomic instability that hampers DNA synthesis [72]. Trichostatin A (TSA) and vorinostat (SAHA) are the most used inhibitors for the class I and II histone deacetylases (HDAC), demonstrating a broad spectrum of epigenetic activities [73]. Sodium phenylbutyrate is also a histone deacetylase inhibitor that is under investigation for its potential use in malignant brain tumors [74]. Although epigenetic drugs have a great potential to improve patient prognosis, there are also important considerations concerning global transcriptional effects. Epigenetic modifications by drugs may result in an aberrant gene expression pattern, leading to a global transcriptional alteration that will drive severe genome instability and cancer [75]. At the molecular level, several studies have reported the upregulation of P21 after epigenetic treatment [76]. On the other hand, since germ cells drive broad epigenetic reprogramming, these drugs could influence histone modifications and alterations in the non-coding RNAs of sperm and oocytes, which may influence progeny development [77]. Furthermore, these drugs have been demonstrated to impair normal hematopoiesis. Indeed, some of the adverse events in the clinical evaluation of epigenetic drugs are hematologic toxicity [76], as well as severe cardiac toxicity, as previously reported with the administration of the histone deacetylase inhibitor [78]. Nevertheless, these drugs exhibited promising results for cancer patients, and due to the growing interest and hope in epigenetic modulation in the clinical practice, we focus this review on different pathways and treatments for the most aggressive solid tumors, specifically small-cell lung cancer, triple-negative breast cancer, pancreatic ductal adenocarcinoma, glioblastoma, metastasic melanoma and ovarian cancer (Figure 4).







**Figure 4.** Summary of the most representative epigenetic modifications of most aggressive solid tumors observed in adulthood. UV: ultraviolet. SCLC: small-cell lung cancer. PDAC: pancreatic ductal adenocarcinoma. AOC: advanced ovarian cancer. TNBC: triple-negative breast cancer. GBM: glioblastoma.

#### *2.1. Epigenetic Modulation in Small-Cell Lung Cancer*

Small-cell lung cancer (SCLC) incidence over time has decreased, reducing by 10–11% in all the cases of lung cancer, which may reflect decreases in smoking habits and changes in the type of cigarettes [79]. One of the causes that leads to a malignant phenotype in lung cancer is the exposure to polycyclic aromatic hydrocarbons, such as benzo (a) pyrene. This induces *TRIM36* hypermethylation, and its subsequent inhibition is associated with the acquisition of an aggressive phenotype [80]. SCLC is the highest aggressive subtype of lung cancer, since tumor cells are highly proliferative, and they spread and metastasize quickly throughout the body [81].

The methylation status of bronchial washings from different types of lung cancers provided a signature, based on four DNA methylated factors (*P16*, *TERT*, *WT1*, and *RASSF1*), which could improve the efficiency of SCLC diagnosis when compared with cytologic evaluation [82]. Another study found that SCLC frequently express thyroid transcription factor 1 (TTF1) at high levels, due to hypomethylation of its promoter [83]. TTF1 overexpression has been reported to confer high tumor cell proliferation and survival [84]. Also, the hypermethylation status in *DCLK1*, which has been associated to colorectal cancer and cholangiocarcinoma, has been found in liquid biopsies in 75% of SCLC patients, and has been associated with poor survival; therefore, this could represent a promising biomarker for early diagnosis and disease prognostic for this cancer subtype [85]. Several other genes have also been found methylated in SCLC, for example *ITK*, *RUNX3*, *CTLA4*, *PLG*, *EMR3*, *SLC22A18*, *TRIP6IL10*, *PECAM1*, *S100A2*, *MMP9*, *ERCC1*, *CSF3R* and *CAV1* [86].

In the treatment scenario, one study reported that 5-AZA-dC and the HDAC inhibitors, LBH589 or MGCD0103, synergistically reduced proliferation in five out of nine SCLC cell lines in vitro [87]. Interestingly, the authors observed higher expression of IFN-stimulated genes in the resistant cell lines after treatment, which determine SCLC cell sensitivity to epigenetic modulators [87]. Another study describes that TSA is able to induce an increase in ABCB1, a protein that confers drug resistance to tumor cells [88]. In clinical trials, a new epigenetic treatment, called RRx-001, is under investigation (NCT02489903; Table 1; Figures 1 and 2). RRx-001 is an alkylating agent based on a dinitroazetidine derivative that inhibits DNA methyltransferase (DNMT) and induces DNA damage via ATM/γ-H2AX, and apoptosis by the activation of caspases [89]. This drug is being tested in platinum refractory or resistant SCLC patients, with 3.8% complete responses and 23.1% partial responses, which increased the overall survival OS [90].

The progress in the treatment of SCLC has been very limited in the last decade, especially when compared to the numerous results that arise for NSCLC. Although the FDA approved the use of immunotherapy anti-PD-L1 in combination with carboplatin and etoposide as an induction therapy in extensive-stage SCLC, much remains to be done to achieve a cure for SCLC patients. In fact, the combination of immunotherapy plus chemotherapy has only represented an improvement in the overall survival of two months [91]. Therefore, there is much left to be done, and, in this sense, drugs directed against epigenetic targets may represent potential treatment approaches.

#### *2.2. Epigenetic Modulation in Triple-Negative Breast Cancer*

Triple-negative breast cancers (TNBC) comprise 7–14% of all breast cancers [92]. TNBC is considered the most aggressive subtype due to the lack of expression of estrogen receptors (ER), progesterone (PR), and HER2 receptors that make the currently used drugs ineffective. One study reported a highly methylated promoter region in the ER gene [93]; thus, a correlation has been suggested with the downregulation of ER expression levels in TNBC patients and the absence of a response [94]. Histone H3 methylation and deacetylation lead to a less compact chromatin structure, which facilitates DNA access to transcription protein machineries. For example, one of the activated genes, due to histone modification in TNBC that provides proliferative features, is NF-κB and its NF-κB-inducing kinase (NIK) [95].


**Table 1.** Current clinical trials developed with epigenetic-based therapies in highly aggressive solid tumors in adulthood. 


**Table 1.***Cont.* 


**Table1.***Cont.*

Drugs in **bold** are the epigenetic-based therapies. N/A: not available. ET: epigenetic therapy. DNMT: DNA methyltransferases. HDAC: histone deacetylases. BET: bromodomain and extra-terminal motif proteins. ORR: overall response rate. CR: complete response. PR: partial response. SD: stable disease. A: active, not recruiting. C: completed. R: recruiting. T: terminated.

A high percentage of TNBC patients carry germline/somatic mutations or epigenetic silencing in *BRCA1*, which implies a deficient DNA repair machinery. Genome-wide DNA methylation analysis in TNBC supports that hypermethylation causes the downregulation of PRSS8, VAMP8 and CLDN4 factors, which confer mesenchymal features [96]. One study revealed a high incidence of *BRCA1* methylation in a TNBC basal-like subtype. This finding could imply resistance to PARP inhibitors for the treatment of *BRCA*-mutant basal-like TNBC [97]. As most of the cases carry mutations in *TP53*, one study has demonstrated that the use of zinc metallochaperones (ZMCs) is efficient to reactivate zinc-deficient mutant *TP53*, by restoring its zinc binding. The use of ZMC1 with a mutation in *TP53R175H* restores *TP53* reactivation [98]. Another mechanism altered by epigenetic modifications in TNBC is the epithelial-to-mesenchymal transition (EMT). The combination of the methyltransferase inhibitor, SGI-110, with the histone deacetylase inhibitor, MS275, has shown a high antitumor ability against TNBC, by epigenetically targeting EMT. Here, TNBC cells showed a marked upregulation of the epithelial protein E-cadherin, and WNT inhibition, and reduced nuclear translocation of EpCAM, which reversed the mesenchymal phenotype after treatment [99]. CD24 overexpression is associated with histone acetylation and is an independent poor prognostic factor in TNBC; importantly, CD24 may be a potential therapeutic target for this type of breast cancer [100]. Mutation analysis revealed that a novel carbazole, SH-I-14, disrupted the STAT3 –DNMT1 interaction and led to the re-expression of tumor-suppressive genes such *PDLIM4* or *VHL*, through demethylation, and showed a high anti-proliferative effect in TNBC models [100].

Concerning histone acetylation, one study showed high levels of H3K9 acetylation in the *TGFβR2* promoter in the TNBC cell line, MDA-MB-231. Moreover, the inhibition of *TGFβR2* decreased migration of the cell line [101]. Another factor, the enhancer of zeste homolog 2 (EZH2), is a type of histone methyltransferase that is highly expressed in TNBCs, and its expression implies shorter disease-free survival in TNBC patients [102]. EZH2 works together with HDACs to mediate transcription repression, by increasing histone H3 Lys27 trimethylation (H3K27me3). One study reported that the inhibition of EZH2 increases H3 Lys27 acetylation, which promotes open chromatin transcription activation, and induces apoptosis in TNBC, through the upregulation of B-cell lymphoma-2-like 11 (BIM) [103].

In respect to ncRNA, the presence of hypermethylation at miR-31 loci in TNBC has been described. Moreover, miR-31 maps to the sequence of a novel long non-coding RNA, LAOT554202 [104]. Both are downregulated in TNBC; however, epigenetic treatment was shown to increase both miR-31 and LAOT554202 expression [104].

Also, the deregulation of some lncRNAs has been associated with the progression of different breast tumors [105]. It has been described that high levels of MALAT1 have correlated with tumor aggressiveness and poor survival of TNBC patients [106,107]. Another lncRNA, HOTAIR, is commonly upregulated in TNBC and associated with the invasive phenotype [108] and lymph node metastasis [109]. In contrast, GAS5 has a protective effect against TNBC, and its overexpression suppressed tumor progression [110], and increased sensitivity to paclitaxel and the subsequent apoptosis ratio [111]. A meta-analysis from 21 studies reported that patients with upregulation of HOTAIR and MALAT1, among others, and downregulation of GAS5 and another three lncRNAs, presented poor survival rates [112]. Another meta-analysis supported that the expression of some lncRNAs, such as MALAT1 and HOTAIR, are associated with positive lymph nodes, while the expression of GAS5 exhibited the opposite effect [113]. Although the FDA has approved epigenetic agents to overcome chemoresistance, to reverse DNA methylation (e.g., 5-azacytidine), and to reverse histone deacetylation (e.g., Trischostatin A and vorinostat (SAHA)), the efficacy of 5-azacytidine has not been consistent in breast cancers. Currently, a new BET inhibitor, ZEN-3694, is being tested in clinical trials because of its ability to prevent the interaction between the BET proteins and acetylated histones (Figure 2). ZEN-3694 is being evaluated in TNBC patients without germline mutations of *BRCA1* or *BRCA2* (NCT03901469; Table 1). Another phase I/II clinical trial is based on the reactivation of ER by decitabine and the

histone deacetylase inhibitor, LBH589, in order to enhance the subsequent tamoxifen treatment (NCT01194908; Table 1).

Modification of the epigenetic machinery is a new tool for the treatment of TNBC, especially BET inhibitors. These drugs have already shown positive effects in preclinical models, and they have yet to be evaluated in clinical trials. These new drugs against epigenetic targets have the potential to decrease tumor aggressiveness and increase sensitivity to standard treatments. Maybe in the foreseeable future, these treatments will improve patient prognosis.

#### *2.3. Epigenetic Modulation in Pancreatic Ductal Adenocarcinoma*

Pancreatic ductal adenocarcinoma (PDAC) shows the lowest five-year survival rate, around 3%, and it is the fourth leading cause of cancer-related deaths in men and women [114]. It is often misdiagnosed and the symptoms are commonly treated by ambulatory care, leading to a late diagnosis; thus, patients present metastatic disease in ~80% of cases at diagnosis. Furthermore, it exhibits chemoresistance due to a complex link between the tumor cells and their microenvironment [115]. In PDAC, most of the studies are centered on mutations in *SMAD4*, *TP53*, *KRAS* or *CDKN2A*, which happen in more than 50% of patients [116]. Furthermore, the mutation in *MBD4* has been found in PDAC, with microsatellite instability [117]. A recent study discovered mutations and genetic variants in several epigenetic regulators, such as *ARID1B*, *PBRM1*, *SMARCA2*, *KDM6A*, *ARID1A*, *SMARCA4*, and *MLL2* [118]. In addition, PDAC has a broad epigenetic signature, which activates oncogenes and inactivates tumor-suppressive genes [119]. Both high- and lowgrade PDAC exhibit specific epigenetic features associated with gene expression patterns. In low-grade PDAC, a highly enhanced H3K4me3 domain has been found, while in highgrade PDAC, a higher H3K4me1 signal was found [120]. Increased expression of DNMT1, DNMT3A and DNMT3B has been detected in PDAC, which suggests direct involvement in the epigenetic regulation of tumor progression [121]. In fact, hypermethylation has been found in *APC* (47.9% of cases), *BRCA1* (45.8%), P16/*INK4a* (35.4%), P15*/INK4b* (35.4%), *RARβ* (35.4%), and *P73* promoters (33.3%) in PDAC patients. Moreover, other genes are methylated to impair several signaling pathways, such as TGF-β, WNT, integrin or ROBO [122].

Concerning histone-modifying enzymes, aberrant HATs and HDACs have been found in PDAC. One study, performed in PDAC-derived cell lines, showed an inhibition of the expression of HAT, P300, and a secondary upregulation of several miRNAs [123]. The supplementary missense mutation in *P300* supports its role as a tumor-suppressive gene in PDAC [124]. The aberrant expression of HDACs is frequently observed in PDACs. For example, HDAC2 and HDAC7 expressions are increased in PDACs, especially in poorly differentiated cases [125,126]. Also, the overexpression of HDAC7 clearly differentiates PDAC from other benign pancreatic neoplasms. A study found that HDAC1 was overexpressed in 56% of PDAC and PanIN lesions [127]. Other studies suggest that RNF2 allows ubiquitination of H2A and downregulation of RNF2, which inhibits tumor proliferation in PDAC in vitro [128]. Histone acetyltransferase (HAT) inhibitors impact genome-wide H3K27ac patterns of PDAC cells [120]. The HAT inhibitors ICG-001 and C646 also impair gene expression and inhibit tumor growth in PDAC [129].

Concerning miRNA, one study with PDAC patients revealed a poor prognosis signature based on the deregulation of 64 miRNAs, and the upregulation of miR-21, miR-196a-2, miR-203, miR-155, miR-210, and miR-222 [130]. Further studies confirmed a decreased expression of miR-132 in PDAC by promoter methylation [131]. Also, lncRNAs have appeared as important regulators for PDAC tumorigenesis [132]. HOTAIR, HOTTIP, MALAT1, and PVT1 are the most studied oncogenic lncRNA in PDAC [133], while LINC00673 and H19 are potential tumor suppressors [134,135]. PIWI-interacting RNAs (piRNAs) and their association with the PIWI subfamily of Argonaute proteins are crucial in pancreatic cancer progression. Indeed, PIWIL1 and PIWIL2 proteins are downregulated in PDAC, probably due to CpG island methylation [136].

The impact of bromodomain inhibitors has also been evaluated in PDAC. BRD4770 is an inhibitor of G9a that induces PDAC autophagy [137]. Moreover, histone methylation regulatory genes, such as KDM6A, are expressed and considered a new candidate in PDAC tumorigenesis [118]. KDM6A is an H3K27me3 demethylase, which is necessary for endoderm differentiation [138]. Another study reported that regions with loss of KDM6A sensitize PDAC cells to bromodomain inhibitors [139]. Other factors have been involved in the progression of PDAC. For example, EZH2 is an H3K27 methyltransferase that has been shown to be overexpressed in PDAC cell lines and patients [140]. The high expression of EZH2 is associated with an aggressive, poorly differentiated subgroup, which shows a shorter survival of patients [141]. Treatments based on the EZH2 inhibitor, DZNep, enhanced the effect of gemcitabine in tumor-derived cell lines and primary cultures from PDAC [142]. Small-molecule inhibitors against EZH2, which are currently being investigated as target therapies against PDAC, are as follows: EPZ-6438, GSK126, CPI-169 and UNC-1999 [143]. High expression of KDM2B is found in PDAC, and it associates with *KRASG12D* to promote tumor initiation in in vivo models [144]. It has been reported that histone H3 modification of the *MUC2* promoter region regulates *MUC2* gene expression, and this expression could be positively modulated by treatment with trichostatin A (TSA) and 5-aza [145]. Another significant treatment is based on the inhibition of telomerase, through the following epigenetic mechanism: methyl-2-cyano-3,12-dioxooleana-1,9(11) dien-28-oate (CDDO-Me). This drug is able to decrease cell proliferation and induce apoptosis in PDAC, through the inhibition of the DNA methyl transferases DNMT1 and DNMT3a [146]. Another strategy with 5-aza-dC in combination with a MEK inhibitor is able to induce cell cycle arrest [147]. Interleukin-13 receptor α2 (IL-13Rα2) is a tumorassociated antigen and a potential target for cancer therapy. Indeed, histones at the IL-13Rα2 promoter region are highly- acetylated; thus, treatment with HDAC inhibitors enhanced the expression of IL-13Rα2 and allowed sensitization for a second treatment [148].

In clinical trials, a pilot study with relapsed patients (NCT02847000; Table 1) tested decitabine in combination with tetrahydrouridine, a cytidine deaminase inhibitor, to avoid catabolism of decitabine. In this study, investigators found scarce effect, due to the local and systemic overexpression of cytidine deaminase in metastatic patients; the resectable patients did not overexpress this protein. This suggested a need for even higher tetrahydrouridine doses in advanced stages [149]. Another phase II trial with resectable PDAC is ongoing, to improve survival with oral azacitidine (CC-486); it includes high-risk patients that have positive lymph nodes, positive margins and/or elevated CA19-9 levels (NCT01845805; Table 1; Figure 1). In another study, with advanced or metastatic PDAC patients, only the patients treated with the combination of azacitidine plus nab-paclitaxel completed the treatment [150]. Previously, other studies have set the bases for the use of romidepsin with small-molecule inhibitors, to target both the MAPK and PI3K signaling pathways to increase apoptosis in RAS-mutated tumors, such as PDAC [151]. Currently, a new clinical trial against PDAC is active, to determine the safety and tolerability of azacitidine and/or romidepsin, combined with nab-paclitaxel/gemcitabine, followed by anti-PD-L1 and lenalidomide (NCT04257448; Table 1). Despite the vast epigenetic landscape of PDAC, clinical and translational research is opening broad treatment perspectives with hopeful results, which involve modulation of the immune response, or administration of epigenetic therapies alone or in combination with standard chemotherapy, to improve patients survival.

#### *2.4. Epigenetic Modulation in Glioblastoma*

Glioblastoma (GBM) is the most commonly diagnosed tumor in elderly Caucasian men [152]. Unfortunately, there is no effective treatment for GBM and the standard treatment for such brain tumors comprises surgical resection with concomitant chemoradiotherapy with temozolomide, followed by adjuvant chemotherapy [153]. However, the main handicaps achieving a successful recovery are tumor heterogeneity, chemoresistance of cancer stem cells, and diffusion of drugs through the blood–brain barrier. Based on molecular

profiling, GBMs are classified into the following three major groups: (1) the 1p/19q codeletion status group, consisting of the IDH-mutant-1p/19q co-deletion status low-grade group; (2) the G-CIMP-low group, including IDH-mutant non-co-deletion status with low DNA methylation status; and (3) the G-CIMP-high group, including the IDH-mutant non-co-deletion group with higher global levels of DNA methylation. IDH mutants lead to major epigenetic changes, because they produce the onco-metabolite 2-hydroxyglutarate that hampers iron-dependent hydroxylases, which includes the 5′ -methylcytosine hydroxylases belonging to the TET family [154]. Among these, the second group, G-CIMP-low, has the worst prognosis [155].

*MGMT* (O-6-methylguanine DNA methyltransferase) hypermethylation predicts BCNU (carmustine) and temozolomide response in gliomas [156,157]. Moreover, patients with hypermethylation of MGMT showed longer overall survival than patients without methylation (43 vs. 16 months, respectively), and a longer time to progress (36 vs. 11 months, respectively) [158]. Treatment with temozolomide combined with the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) delayed temozolomide resistance when compared with treatment with temozolomide alone, by MGMT overexpression [159]. Some HDAC inhibitory prodrugs of butyric acid and valproic acid increased the antitumor efficacy of doxorubicin, without cardiotoxicity, in mouse models of GBM (Figure 2) [160].

Recently, it has been described that a specific GBM subtype, with high levels of MGMT, expresses methyl-CpG binding domain 3 (MBD3) protein, which targets CK1A. Therefore, this subtype of patients may obtain benefit from CK1A activator pyrvinium pamoate (Pyr-Pam), leading to MBD3 degradation [161]. The new histone deacetylase inhibitor CKD5 is a derivative of 7-ureido-N-hydroxyheptanamide, and it revealed strong antitumor effects in GBM, both in in vitro and in vivo models. The use of the demethylases KDM1 and KDM5A was also evaluated as a potential therapeutic target [161]. A study demonstrated that the inhibition of KDM1 and KDM5A showed a significant antitumor effect in wild-type and temozolomide-resistant GBM cells [162]. Another study tested the multi-KDM inhibitor JIB-04, which has strong anti-clonogenic activity in wild-type and temozolomide-resistant GBM cell lines [163]. Another potent HDAC6 inhibitor, CAY-10603, is able to induce apoptosis in several GBM primary and stem cell-like cell lines [164]. Another study, with small molecules such as EZH2 and HDACi, achieved proliferation arrest of GBM [165]. Treatment with vorinostat (HDAC inhibitor) and tranylcypromine (histone lysine demethylase KDM1A inhibitor) (Figure 2) decreased GBM stem cell proliferation and led to significant tumor regression in mouse models [166]. Also, the use of bromodomain inhibitors have risen in popularity, due to enhanced tumor lethality [167]. In fact, the BET inhibitor caused downregulation of the lncRNA HOTAIR, which induced cell cycle arrest in GBM cells [168]. Several signaling pathways, such as WNT/β-catenin, mTOR, or P53-HIF, are found to be activated in gliomas, due to the downregulation of several lncRNAs [63]. The inhibition of HOTAIR leads to the increased expression of miR-326, which induces the expression of FGF-1 [169]. Another lncRNA, MALAT1, which is upregulated in temozolomide-resistant GBM, has been seen to promote miR-101, miR-203 and thymidylate synthase expression when downregulated [170,171].

Concerning clinical trials, the use of temsirolimus has obtained interesting improvement in 36% of treated patients; furthermore, the treatment achieved a significantly longer time to progress [172]. In contrast, panobinostat administration with bevacizumab did not show any significant improvement in progression-free survival compared to bevacizumab alone [173]. A phase I/II trial with a histone deacetylase inhibitor, romidepsin, found this drug to be inefficient for patients with recurrent GBM [174]. Currently, a phase I clinical trial is ongoing, to test whether folic acid is able to lead to MGMT methylation and improve temozolomide plus radiation treatment in grade IV tumors (NCT01700569; Table 1). This trial was based on the fact that folate could induce DNA methylation and increase the sensitivity to temozolomide in in vivo models [175].

In conclusion, although molecular diagnosis has brought new options to identify and treat patients, therapeutic options remain without any significant changes. Currently, the best standard treatment is the maximum safe resection, followed by chemoradiation and adjuvant chemotherapy. We hope that new clinical trials with epigenetic target therapies could improve the responses to conventional treatments.

#### *2.5. Epigenetic Modulation in Metastatic Melanoma*

The main issue with metastatic melanoma lies in its chemoresistance. Currently, the new immunecheckpoint inhibitors against CTLA-4, PD-1 or PD-L1 have improved patient outcome. However, secondary genomic aberrations make tumor cells acquire rapid resistance to these therapies [176]. One of the risk factors associated with melanoma is UV radiation; this is due to changes in DNA methyltransferase and in histone acetylation, which leads to silencing of tumor-suppressive genes. In contrast, some dietary consumption of green tea and proanthocyanidins from grape seeds has the ability to block UV-induced epigenetic modification in the skin of *CIP1*/P21 or P16/*INK4a* [177]. The epigenetic modifications of melanoma are well defined; in fact, malignant transformation of peritumoral skin is due to epigenetic changes [178]. CC chemokine receptor 7 (*CCR7*) and CXC chemokine receptor 4 (*CXCR4*) are epigenetically upregulated in melanoma cells, and have the ability to induce metastasis of melanoma [179]. The following four tumor-suppressive genes are frequently hypermethylated in advanced melanoma: death-associated protein kinase (*DAPK*), O6-methylguanine DNA methyl-transferase (*MGMT*), *RAS* association domain family protein 1A (*RASSF1A*), and retinoic acid receptor-β2 (*RAR-β2*). The hypermethylation of *DAPK*, *MGMT* and *RASSF1A* is significantly lower in the early stages than in the advanced stages, whereas the incidence of hypermethylation of *RAR-β2* is highly similar in the early and advanced stages [180]. The HDAC inhibitor dacinostat (LAQ824) is able to restore retinoid sensitivity by reverting *RAR-β2* methylation in melanoma cells, and it achieved the highest benefits in combination with retinoids [181]. Also, TET proteins have been reported to play a crucial role in melanoma, since their ectopic expression of TET2 eradicates tumor proliferation and increases survival in vivo [37]. It has been described that the loss of histone acetylation and H3K4 (histone H3 Lysine 4) methylation in *BRAFV600E* and *PTEN* promote malignant transformation of melanocytes [182]. EZH2 is another factor expressed in metastatic melanoma; its depletion has been shown to restore P21/*CDKN1A* expression and arrest cell proliferation [183].

Concerning ncRNA, several studies have reported the importance of miRNA regulation in melanoma. For example, miRNA-125b is involved in the regulation of vitamin D receptor (VDR), and in the resistance of 1,25-dihydroxyvitamin D3, a potential therapy for metastatic melanoma [184]. Moreover, the expression of other miRNAs, from a large cluster of parentally imprinted regions located on chromosome 14q32, is significantly downregulated in melanoma, by epigenetic modulation. Interestingly, this miRNA cluster can be re-expressed with a combination of demethylating agents and histone deacetylase inhibitors. In this region, re-expression of mir-376a and mir-376c delayed cell growth and migration; moreover, one of the targets of both miRNAs is IGF1R, which is a tumor-prone factor in melanoma [185].

Since the largest clinical issue in the treatment of advanced melanoma patients is chemoresistance, the effort of researchers is centered around the discovery of a new treatment method to improve drug sensitivity. Interleukin-2 has exhibited potent antitumor activity in the fight against melanoma; nevertheless, its high toxicity has limited its use [186]. Treatment with SAHA is able to induce H3 and H4 hyperacetylation of P14/*ARF* promoter, and upregulate its expression [187]. Treatment with 5-aza-dC prevents the induction of DNMT1 and DNMT3b at the P16/*INK4A* promoter, leading to its subsequent activation [187]. Another treatment evaluated is allyl isothiocyanate (AITC), which has been reported to reduce cell proliferation and decrease the activation of HDACs, HATs, and other histone methyl transferases (HMTs). This approach is a very promising epigenetic therapy for advanced melanoma [188]. Some isothiocyanates, such as sulforaphane and iberin, could act over the epigenetic modulation of melanomas, and are currently under investigation [189]. Immune checkpoint-based therapy has improved patient lifespan from

nine months to 2 years [190]. Perhaps, in the near future, the combination of anti-CTLA4 or anti-PD1 immune checkpoint inhibitors and epigenetic therapy could suppress the chemoresistance of metastatic melanoma [191].

Clinical trials with epigenetic therapy in metastatic melanoma have been mostly based on decitabine and other epigenetic modulating drugs, such as histone deacetylase inhibitors. A phase I clinical trial has explored the safety and tolerability of two epigenetic drugs, decitabine and panobinostat (a histone deacetylase inhibitor), in combination with temozolomide, to overcome chemoresistance in advanced melanoma (NCT00925132; Table 1). However, in this study, most of the patients exhibited disease progression [192]. Another clinical trial is testing the efficacy of oral azacitidine (CC-486) combined with pembrolizumab (NCT02816021; Table 1; Figure 1). Here, PD-1-naïve patients achieved a partial response (55% ORR), and accrual to this arm A continues; however, none of the patients with progression on prior PD-1 therapy, in arm B, have responded [193]. Other investigators have tested whether the action of vemurafenib (BRAF inhibitor) is more effective in combination with decitabine in low doses (NCT01876641; Table 1). Although the trial was terminated, due to a loss of funding, 3/14 patients achieved a complete response, 3/14 had a partial response, and 5/14 had stable disease. Moreover, its preclinical assessment demonstrated effectiveness of the combination, and a high potential in delaying chemoresistance [194]. Another clinical trial, performed in non-inflamed stage III/IV melanoma, is recruiting patients (NCT03765229; Table 1), and its clinical rationale is based on the induction of PD-L1 expression by the action of entinostat (HDAC inhibitor; Figure 2) [195]. The addition of anti-PD-1/anti-PD-L1 checkpoint inhibitors to HDAC inhibitors has been demonstrated to enhance the antitumor effect when compared to monotherapy, both in in vitro and in vivo models [196,197]. Another phase I clinical trial has evaluated the safety and efficacy of decitabine in combination with temozolomide (NCT00715793; Table 1). Here, there were 2/35 complete responses (CR), 4/35 partial responses (PR), 14/35 stable diseases (SD), 13/35 progressive diseases (PD), and the median overall survival was 12.4 months [198]. Another drug combination under investigation is tinostamustine with the anti-PD-L1 antibody nivolumab (NCT03903458; Table 1). Tinostamustine is an alkylating histone deacetylase inhibitor (HDACi), which resulted from the fusion of the alkylating agent bendamustine to the pan-HDACi vorinostat (Figure 2). This combination is expected to enhance the antineoplastic effect in refractory, locally advanced, or metastatic melanoma patients [199]. Also, the alkylating agent dacarbazine is the only drug approved by the Food and Drug Administration (FDA) as a therapy for advanced melanoma, with response rates between 7 and 13% [200].

Epigenetic therapies allow the reversibility of epigenetic modifications and are drawing attention to metastatic melanoma research, to prevent or delay the emergence of resistance to current standard treatments. Therefore, new discoveries in epigenetic therapies are expected to be evaluated in further clinical trials.

#### *2.6. Epigenetic Modulation in Ovarian Cancer*

Aggressive ovarian tumors (AOT) are the gynecological cancers with the highest mortality rate, probably because most AOT patients present advanced stages at diagnosis (stage III or IV), due to the lack of symptoms or unavailable specific screening biomarkers [201]. While response in the early stages is frequently acceptable, advanced tumors present a short progression driven by chemoresistance. Some translational research has shown that epigenetic aberrations are quite important in tumor initiation and development [202]. For example, the expression of HDAC2 hampers the DNA damage responses induced by platinum compounds, and contributes to the pathogenesis and chemoresistance of AOT [203]. In addition, the inhibition of H4K16 acetylation has been observed in AOT [204]. Further, hMOF, a member of the HATs family that acetylates H4K16, could also serve as an epigenetic biomarker for the diagnosis of malignant AOT, since patients with high expression levels of hMOF present improved survival when compared to those with low hMOF levels [205]. The presence of class I HDACs are able to induce the pro-

gression of AOT, and high expression of class I HDACs has also been detected in AOT patient samples. Furthermore, the expression of class I HDAC proteins has been considered a poor prognostic biomarker in AOT [206]. Cacan et al. have demonstrated that the downregulation of RGS2, an inhibitor of G-protein-coupled receptor proteins (GPCRs), confers chemoresistance of AOT cells, which is in part due to the repression of the promoter region of *RGS2* by class I HDACs [207]. Also, chemoresistance to platinum-based drugs has been associated with SIRT1 upregulation through the BRCA1–SIRT1–EGFR axis [208]. SIRT1 upregulation correlates to *TP53* inactivation by deacetylation [209]. SIRT3, in contrast, inhibits AOT cell migration via TWIST downregulation [210]. Other factors, such as EZH2, are overexpressed and have a direct positive correlation with AOT histological grade and tumor stage [211]. Further, 3-deazaneplanocin A (DZNEP) is a target for EZH2, with a promising anticancer efficacy against AOT [211]. Another EZH2 inhibitor, GSK126 (Figure 2), has demonstrated a better response in *ARID1A*-mutated patients [212]. Another study has associated LSD1 overexpression with AOT [213], and the combination of LSD1 with sodium butyrate increases most of the hallmarks of AOT [214,215]. Other factors, such as KDM3A, are crucial for AOT progression, undifferentiation, and platinum resistance, and have been identified as a potential target for AOT [216].

It is known that cancer modifies the microenvironment to inhibit the immune system. In this context, the overexpression of HLA-class I and II has been associated to AOT [217]. Epigenetically silenced hMLH1, together with cisplatin, could be an effective treatment, alongside decitabine and other HDAC inhibitors, such as belinostat (Figure 2), against AOT [218]. Chemoresistant tumor cells have inhibited the expression of OX-40L and 4- 1BBL, two stimulator receptors of the immune system, with the concomitant overexpression of the immunosuppressive factor PD-L1 [219]. Indeed, HDAC1 and HDAC3 showed a strong association with OX-40L and 4-1BBL promoters, which contributes to OX-40L and 4-1BBL repression [219].

The inhibition of histone acetyltransferase is a new approach for the treatment of malignant AOT and its chemoresistance. The following three HDAC inhibitors have been approved by the FDA: romidepsin, panobinostat, and vorinostat (Figure 2). Trichostatin A (TSA), which exhibits a significant inhibition of class I and II HDACs, is able to activate P73 and trigger apoptosis in AOT cells [220]. Another study evaluated belinostat with carboplatin in platinum refractory AOT patients. However, the lack of drug activity concluded in the termination of the study [221]. Other authors initiated a phase Ib/II trial with recurrent AOT patients, to evaluate the clinical benefit of paclitaxel, carboplatin and belinostat [222]. Here, 3/35 patients presented a complete response, while 12/35 exhibited a partial response, with an ORR of 43%. It is remarkable that the median overall survival was not reached; thus, the results showed that paclitaxel + carboplatin + belinostat regimen demonstrated a clinical benefit. In a phase II study, vorinostat was evaluated for the treatment of recurrent AOT; however, vorinostat exhibited minimal activity as a single agent [223]. Another phase II trial evaluated the effect of hydralazine and magnesium valproate (NCT00404508; Table 1; Figures 1 and 2). The clinical benefit with these epigenetic agents was observed in 80% of patients, which supported their use as epigenetic therapy to overcome chemoresistance in recurrent patients [224]. Another study tested decitabine as an epigenetic chemosensitizer to carboplatin plus a paclitaxel regimen (NCT02159820; Table 1). The study is supported by the fact that 5-aza-dC treatment is able to restore P27 expression and increases the sensitivity of tumor cells to cisplatin [225]. Another study aims to determine the optimal dose of oral azacitidine (CC-486) in combination with pembrolizumab, for the treatment of platinum-resistant or refractory AOT (NCT02900560; Table 1; Figure 1).

AOC is strongly influenced by epigenetic changes that affected DNA methylation and histone modifications. The first attempts to modify the epigenetic of AOC with drugs have achieved low response rates as single agents; thus, their combination with targeted therapies, based on the mutational burden of tumors, must be evaluated.

#### **3. Conclusions**

In the clinic, patients with highly aggressive tumors are presented with different prognoses, despite having a similar stage and grade of cancer. These observations could be explained by the tumor heterogeneity that is characterized by several epigenetic modification profiles [226]. Firstly, we must highlight several oncogenic point mutations associated with epigenetic regulators, such as *IDH1/2*, *EZH2* or *DNMT3A*. Moreover, not all mutations are tumor-prone, and we must consider tumor-suppressive factors such as KDM6A and CREBBP/P300 [227]. Finally, another important element is when DNA epigenetic modifications emerge with histone modifications, to inactivate the action of tumor-suppressive factors [228]. All these actions are crucial in the regulation of tumor initiation and development. Overall, these alterations could serve as molecular biomarkers to stratify high-risk patients into different groups and provide the best treatment strategy in each case. We are confident that all the positive results, obtained in hematopoietic malignancies in preclinical studies, provide a strong rationale for further trials in highly aggressive solid tumors, to improve patient survival and prevent chemoresistance. Most of the clinical trials with epigenetic drugs are in combination with standard chemotherapies; however, further research is needed with the combination of epigenetic drugs and targeted therapies.

**Author Contributions:** Conceptualization, J.M.-U. and J.G.-F.; writing—original draft preparation, J.M.-U.; figures and tables, M.M.-G.; writing—review and editing, M.F.-C. and J.M.-U.; visualization, M.M.-G. and M.F.-C.; supervision, J.G.-F.; funding acquisition, J.M.-U. and J.G.-F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work has been carried out thanks to "V Becas de Investigación Carmen Delgado/Miguel Pérez-Mateo" (25791/001) by Asociación Cáncer de Páncreas (ACanPan) y la Asociación Española de Pancreatología (AESPANC).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We especially thank the oncologist Aberto Orta-Ruiz from the Medical Oncology Department (Fundacion Jimenez Diaz University Hospital) for his appreciated revision, suggestions and criticism for the present review article. All figures have been designed with BioRender.com.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Revisiting Mitochondria Scored Cancer Progression and Metastasis**

**Rohit Gundamaraju 1, \*, Wenying Lu <sup>2</sup> and Rishya Manikam 3, \***


**Simple Summary:** The indispensible role of mitochondria has been described over a century ago by Otto Warburg which has been serving the fields of cell biology and cancer biology immensely. Mitochondria are the principal site for vital mechanisms which vastly dictate the physiology. The intricacy of mitochondria's role cancer have been noticed and well addressed in recent times. The underlying mechanisms are surfacing to unveil the nature of mitochondria and its participation in tumor cell motility and metastasis. This addressing may unravel novel therapeutic options. This review summarizes and reweighs the key aspects like underlying and emerging mechanisms which might be useful in designing novel chemotherapy.

**Abstract:** The Warburg effect has immensely succored the study of cancer biology, especially in highlighting the role of mitochondria in cancer stemness and their benefaction to the malignancy of oxidative and glycolytic cancer cells. Mitochondrial genetics have represented a focal point in cancer therapeutics due to the involvement of mitochondria in programmed cell death. The mitochondrion has been well established as a switch in cell death decisions. The mitochondrion's instrumental role in central bioenergetics, calcium homeostasis, and translational regulation has earned it its fame in metastatic dissemination in cancer cells. Here, we revisit and review mechanisms through which mitochondria influence oncogenesis and metastasis by underscoring the oncogenic mitochondrion that is capable of transferring malignant capacities to recipient cells.

**Keywords:** mitochondria; metastasis; OXPHOS; cancer; Warburg effect; cancer therapeutics

#### **1. Good and Bad Mitochondria**

Tumor cell metabolic reprogramming dictates the difference between normal and tumor cells. Mitochondria play a major role in metabolic reprogramming:It has been shown that tumor mitochondria not only change their structure but also decrease the potential of oxidative phosphorylation (OXPHOS) and apoptosis [1]. Considering the countless functions of mitochondria, including in the tricarboxylic acid (TCA) cycle, OXPHOS, etc., it is no surprise that mitochondria are directly involved in cancer progression (Figure 1) [2]. Metastasis is a hallmark of cancer and includes several steps: detachment of local tumors, intra-invasion, circulation in the blood, extra-invasion, and colonization in the secondary sites for survival. In all the above stages, mitochondrial metabolism is tuned for tumor cell adaptation to facilitate metastasis [3]. In addition, several postulations have been proposed on the vital role of mitochondria in metastasis, where mitochondria help overcome perturbations in metastasis environments. mtDNA single nucleotide polymorphisms (SNPs) and a few mutations might lead to distinctions in metastatic susceptibilities in cancer histotypes or patient groups. Many studies have revealed that mitochondria are involved in a chain of events including modulation of the the microenvironment, motility and invasion, plasticity,

**Citation:** Gundamaraju, R.; Lu, W.; Manikam, R. Revisiting Mitochondria Scored Cancer Progression and Metastasis. *Cancers* **2021**, *13*, 432. https://doi.org/10.3390/cancers 13030432

Academic Editor: Ion Cristóbal Received: 17 December 2020 Accepted: 21 January 2021 Published: 23 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

and consolidation [4]. Therefore, in the current review, we consolidated the essential contributing factors of mitochondria in cancer progression and, specifically, metastasis. We also discuss several questions that address the underlying mechanisms of context-dependent contributions of mitochondria in metastasis.

**2021**, , x FOR PEER REVIEW 2 of 13

**Figure 1.** Epithelial–mesenchymal transition (EMT) by mitochondria.TCA: tricarboxylic acid; OXPHOS: oxidative phosphorylation. Mitochondrial metabolites are accumulated upon mutation of the indicated TCA cycle enzymes which activates the EMT. In cancer cells, the TCA cycle not only serve to produce reducing equivalents to fuel the electron transport chain, but also to generate biosynthetic intermediates that are necessary for cell proliferation and migration.

#### **2. Can Mitochondrial Dynamics Dictate Cancer Spread?**

The balance of mitochondrial fusion and fission is necessary for the regulation of various processes, including the quality of mitochondria, cell metabolism, cell death, proliferation, and cell migration, and is maintained by numerous mitochondrial-shaping proteins. Negative modulations or malfunctions in these processes resulting in changes in mitochondrial dynamics lead to diseases like cancer [5]. Mitochondrial dynamics are correlated with various diverse disease pathologies. For instance, a high nutrientdeprived state triggers mitochondrial fission and hence results in programmed cell death. A significant number of proteins such as the GTPases (Mfn1, Mfn2, Opa1, and Drp1) have strong regulatory effects in balancing mitochondrial fusion and fission. Any perturbations or failure in managing the correct dynamic state leads to cancer [6]. It is noteworthy that mitochondrial dynamics have a deep role in cancer cell migration (Figure 2). The discovery of susceptibility to cancer associated with altered or modulated mitochondrial dynamics could result in new targeted therapies. Some of these altered dynamics, such as mitochondrial fission, are discussed in this review.

Mitochondrial dynamics also massively impact apoptosis. Studies have correlated them to dynamic homeostasis and tumor growth. Strikingly, signaling downstream of mutant KRAS in pancreatic cancers leads to mitochondrial fragmentation and increased activation of Drp1, processes that are required for KRAS-driven tumor growth in vivo. In addition, recent studies also suggest that mitochondrial dynamics are important for regulating metastatic phenotypes such as invasion and migration in breast and thyroid cancers [7].

Examples of high-to-low expressions of Drp1 and Mfn1 have been implicated in metastatic breast cancer [8]. Moreover, Drp1 is regarded as crucial for apoptosis due to its informed role in releasing cytochrome-c. The Drp1/Mfn1 expression ratio correlates to aggressive cancers and cell proliferation. The Drp1/Mfn1 expression ratio was found to be increased in hepatocellular carcinoma (HCC) tissues and associated with poor prognosis. Escalated mitochondrial fission mediated by imbalanced reactive oxygen species (ROS) production was found to be the primary reason for the pro-survival ability of the HCC cells

under both in vitro and in vivo conditions [9]. High Drp1 expression was also observed in ovarian cancers where Drp1 was found coexisting with cell cycle-related genes, thereby facilitating cancer cell proliferation [10]. Mitochondrial fission also aided in cisplatin resistance in ovarian cancers [11]. Drp1 inhibitors like Driptor1 were employed against breast cancer cells, which show not only that the mitochondrial dynamics-mediated pathway is useful in designing anti-cancer therapy [12] but also that mitochondrial fission facilitates the survival, apoptosis, and drug resistance of breast cancer cells [13].

**Figure 2.** The differential role of mitochondrial dynamics in normal and cancer cells.OXPHOS: oxidative phosphorylation; ATP: adenosine tri phosphate; mROS: mitochondrial reactive oxygen species. The morphology and physiology of the mitochondria and its healthy functioning is governed by fission proteins. Mitochondria fission promotes glycolysis, mitophagy, and apoptosis and is also necessary for cell division. In contrast, mitochondria fusion promotes ATP and ROS production via OXPHOS. In normal cells (left area of the figure), mitochondria fusion and fission are well-balanced, which results in healthy homeostasis. Whereas in cancer cells, an imbalance of the fusion and fission is favored to drive proliferation, metastasis, and the maintenance of cancer stem cell phenotypes.

Mammalian target of rapamycin complex 1 (mTORC1), which is a trigger factor in cancers, stimulates translation of mitochondrial fission process 1 (MTFP1), which is coupled to pro-fission phosphorylation and mitochondrial recruitment of DRP1 in melanoma cells [14]. This shows that DRP1 couples with pro-cancer pathways. Other very recent evidence of mitochondrial dynamics and their role in cancer promotion was recorded in a study where Drp1 increased prostate cancer cell survival under metabolic stress conditions [15]. Further, knock down of Nestin, which is one of the classic markers in gastrointestinal cancers, downregulated recruitment of Drp1 to mitochondria in gastrointestinal stromal tumor cells [16]. To support this, another novel study consisting of Paris Saponin II (PSII), a major steroidal saponin extracted from Rhizoma Paris polyphylla, was employed against Drp1, which aided the modulation of Drp1-mediated mitochondrial fission [17]. The PSII in [17] surprisingly downsized the xenograft tumor size and impeded the phosphorylation of ERK1/2 and Drp1 at Ser616. Mitochondrial dynamics also enormously influence survival and stemness maintenance of cancer stem cells (CSCs), which are responsible for tumor recurrence and other malignant traits. Blockade of fission debilitated the self-renewal capacity of CSCs and led to CSC exhaustion. In addition, the reliability and functionality of T cells in the cancer microenvironment depend vastly on the mitochondrial dynamics

balance, which hints at the essentiality and usefulness of targeting mitochondrial dynamics in anti-cancer treatment [18].

#### **3. Mitochondria's Vital Role in Numerous Cancers**

Mitochondria have been well-known as crucial factors in various characteristics of cancer biology, including cancer development, metastasis, and drug resistance [19,20]. The alteration of mitochondria dynamics can affect the regulation of cancer cells. Mitochondria duties in dynamic networks include changes in size and distribution of sub-cellular components, and these dynamics are maintained by two main opposing processes: fission and fusion [21], regulated by dynamin-related protein 1 (Drp1) and mitofusins (Mfns) [22], respectively. Unbalanced mitochondrial fission or fusion dysregulates the cellular processes that contribute to tumorigenesis [23,24]. In breast cancer, increased mitochondrial fragmentation intensifies the capabilities of breast cancer cells to metastasize by activating Drp1 or silencing Mfns [8].The imbalance of Drp1/Mfn expression has also been found to cause additional mitochondrial fission and impaired mitochondrial fusion inhuman lung cancer cell lines, which is a key process for cell cycle progression [25].In addition, cancer cells are involved in the mitochondrial respiration chain to gain an obvious increase in ATP production [26]. Cancer cells generate invasion or metastasis by utilizing energy, powered through the transcription co-activator, PGC-1α, to promote OXPHOS, mitochondrial biogenesis, and oxygen consumption rate [27]. The association between PGC-1α expression in the invasion and metastasis of human invasive breast cancer was found in previous study [27]. Furthermore, the dysregulation of mitochondrial respiratory chains prompts ROS-induced integrin β5 expression and results in an increase in tumor cell invasion and metastasis in gastric cancer cells [28]. In addition, mitochondrial respiratory chain complexes are involved in cell apoptosis processes, in particular, complexes I and III are key regulators of cell apoptosis and major sources of ROS generation [29]. In most aggressive breast cancer, the most remarkable activity of complexes is observed. ROS-associated signaling pathways can be a potential suppressor for the tumor treatment target. On the other hand, mitochondrial dysfunction is identified as being associated with cancer progression. mtDNA mutations have been frequently encountered in cancer cells. Mitochondrial fusion activity is essential for mtDNA maintenance, a loss of mtDNA has been correlated with the drug resistance of anti-estrogen therapy in breast cancer [30]. Moreover, the mutation of mtDNA is one of the key factors that stimulate mitochondrial-mediated metastasis. For instance, mutated ROS-generating mtDNA promotes invasion and metastasis in lung cancer cells and breast cancer cells [31,32]. In addition, declined OXPHOS gene expression was found to result in metastasis in cancer cell lines and in metastatic melanoma in renal cancer specimens [33].

Other aspects to be considered are cross-links with mitochondrial dysfunction and promotion of tumor cells metastasis. Epithelial–mesenchymal transition (EMT) enables cancer cells to obtain the migration abilities to move out of the primary tumor and translocate to new target organs [34,35]. EMT transfers the epithelial cell to mesenchymal phenotypes in many epithelial tumor cells that are affected by mitochondrial dysfunction [33]. Mitochondrial dysfunction initiates EMT via EMT signaling pathways. TGF-β is known as a key growth factor controlling EMT progression through TGF-β/SMAD/SNAIL, phosphatidylinositol-3-kinase (PI3K)/AKT signaling pathways [36]. TGF-β phosphorylates TGF-β receptor-regulated Smad2 and Smad3, then upregulates the expression of their downstream gene, Snail-1, which is a positive regulator of EMT and metastasis [36]. Activated PI3K/AKT signaling can also upregulate the intracellular expression of Snail, thereby inducing the EMT [37]. The depleted mtDNA induces mitochondrial dysfunction and further triggers EMT induction, the prostate and breast adenocarcinoma cells show mesenchymal phenotypes with TGF-β overexpression [38].Moreover, in hepatocellular carcinoma cells, mtDNA depletion induces EMT via TGF-β/SMAD/SNAIL signaling [39]. In the tumor microenvironment, the hypoxia-induced accumulation of HIF-1 alpha activates the expression of TWIST which ultimately induces EMT [40]. The co-expression

of HIF-1 alpha, TWIST, and Snail in primary tumors of head and neck cancer patients correlates with the poorest prognosis [41]. mtDNA depletion also can induce mitochondrial dysfunction and promotes EMT induction via mitochondrial reversed signaling. Mitochondrial reversed signaling triggers transcriptional activation of EMT signaling pathways, such as SNAIL, TWIST, and mesenchymal markers, such as vimentin, N-cadherin, with a corresponding loss of epithelial marker E-cadherin [42]. mtDNA-depletioncan also cause a loss of mesenchymal phenotypes of ESPR, such as ESPR1 in breast cells and expressed stem-cell phenotypes, suggesting a generation of cancer stem cells [42]. On the other hand, mutated mitochondrial metabolic enzymes are closely correlated with EMT-induced metastasis, which contributes to the initiation of oncogenic signaling cascades in cancers [43,44]. Another link with mitochondria in cancer cell metastasis is epidermal growth factor receptor (EGFR). EGFR was found intensively expressed in the mitochondria of highly invasive non-small cell lung cancer (NSCLC) cells [45]. EGF is a growth factor that initiates the EMT by activating the RAS/RAF/MEK/ERK MAPK signaling cascade. The activated ERK1/2-MAPK induces EMT, promoting the regulation of cell motility and invasion [46]. EGF initiates cancer cell invasion by regulating mitochondrial functions. EGF activates the mitochondrial translocation of EGFR, mitochondrial fission, and redistribution, upregulates cellular ATP production, and enhances cancer cell motility in vitro and in vivo. Furthermore, EGFR can regulate mitochondrial dynamics by interchanging with Mfn1 and disturbing Mfn1 polymerization, therefore, overexpression of Mfn1 reverses the phenotypes resulting from EGFR mitochondrial translocation to induce mitochondrial fragmentation [45].

#### **4. Multiple Mechanisms of Metastasis by Mitochondria**

Deciphering the mechanisms of metastasis involving mitochondria is extremely important in establishing therapeutics. The tumor microenvironment plays a prominent role in the progression of cancer, and it has a similar role in cancer chemoresistance via a mechanism called mitochondrial transfer, which broadly favors further invasion and metastasis. Mitochondrial transfer occurs in cells that fail to perform aerobic respiration due to mtDNA malfunction [47]. On the other hand, a horizontal mitochondrial transfer is also associated with chemoresistance. In the tumor microenvironment, horizontal transfer is regarded as lethal, since the transfer of mtDNA from the host cell to the cancer cell leads to escalated tumor-initiation ability because the cancer cells possess reduced respiratory function, and horizontal transfer in such instances improves the aggressiveness of cancer cells. Studies have exhibited that mtDNA transfer protects the cells from chemotherapeutic drugs. In a study involving acute myelogenous leukemia (AML), cells took up functional mitochondria from the bone marrow-derived stromal cells, which lead to protection of the cells from the drug effect and evasion of cell death [48]. Mitochondrial transfer under an in vivo setting not only leads to chemoresistance but also disease relapse. This entire concept of mitochondrial transfer endorses the notion of tumor plasticity and highlights the ability of the tumor cells to overcome unfavorable conditions by altering energy metabolism [19]. Further, mitochondrial transfer has been implicated in murine tumor models with essential functional consequences for tumor growth and metastasis. This has also been supported by studies where the mitochondrial transfer rescued cancer cells that were suffering deficiencies in OXPHOS and were prone to therapeutic apoptosis [49]. It is a proven phenomenon that dysregulated mitochondrial trafficking leads to metastasis of cancer cells.

Ubiquitination of syntaphylin (SNPH) (Figure 3) is regarded as a vital regulator of mitochondrial trafficking. Studies show that SNPH aids in binding the mitochondria to the microtubule. Mechanistic studies hint that SNPH is modified by the ubiquitin ligase CHIP/STUB1 and deubiquitinated in a USP7-dependant manner, which suggeststhat ubiquitination of SNPH isa pivotal regulator of mitochondrial trafficking and tumor cell invasion [50]. Apart from the SNPH mechanics, hypoxia also governs mitochondria localization in cancer cells. Tumor cells under the influence of hypoxia downregulate

SNPH protein and mRNA levels, which inturn leads to increased invasiveness in glioblastoma cells. Surprisingly, tumors with stabilized HIFα or with deletions resulted in lower expression of SNPH, denoting SNPH's principal role in metastasis [50].

Mitochondria can enormously influence malignant transformation and dictate the tumor plasticity of cancer cells and govern several mechanisms to address tough environmental conditions. Mitochondria are the major source of ROS utilized in OXPHOS. Mitochondrial enzymes, such as pyruvate dehydrogenase (PDH), a-ketoglutarate-dehydrogenase (a-KGDH), acyl-CoA dehydrogenase, and glycerol-3-phosphate dehydrogenase, are involved in ROS generation [51]. The huge difference between normal cells and cancer cells is the controlled levels of mitochondrial ROS (mROS). The levels of mROS are properly regulated in cancer cells in order to play a role in essential cellular processes. On the other hand, cancer cells have functions like oncogene activation, tumor suppressor loss, and hypoxia, which lead to uncontrollable mROS levels that aid in sustaining cancer cell. mROS levels participate in multiple steps of oncogenesis and induce mtDNA mutations. They also influence apoptosis evasion, metabolic reprogramming, and cellular proliferation [52]. mROS is responsible for the activation of several important oncogenic signaling pathways such as the epidermal growth factor receptor (EGFR) signaling pathway [53]. Mitochondria help the epithelial cells in gaining migration speed by providing energy, as it was demonstrated that insufficient energy with deficiencies of mitochondria inhibited cell motility. Similarly, mitEGFR enhances mitochondria fission and cancer cell motility, independent of its phosphorylation status [45]. In order to drive towards a proliferative state and escape mitochondrial permeability transition-mediated cell death, cancer cells intelligently maintain high levels of anti-oxidant proteins to prevent ROS accumulation. On the other hand, the interrelationship between mROS and hypoxia inducible factor-1 (HIF-1) is complex. Hypoxia-mediated mROS leads to HIF-1 activation, which facilitates metastasis because of the metabolic shift from OXPHOS to glycolysis by increasing the expression of glycolyticenzymes. In contrast, HIF-1 decreases mROS production, promotes tumor growth, and facilitates the survival of metastatic cells, denoting the vibrant and functional role of mROS in various cancers [51].

Similar to the above, Sirtuin 3 (SIRT3) is involved in several key processes such as the response to oxidative stress and mitochondrial metabolism regulation. SIRT3, a NAD+ dependent mitochondrial deacetylase that promotes efficient oxidative metabolism, is a key regulator of mitochondrial ROS production and detoxification [54]. Literature suggests that SIRT3 has a role in regulating mitochondrial quality control and affects genes involved in homeostasis such as PGC-1α and TFAM. SIRT3 silencing results in making breast and colon cancer cell lines prone to cytotoxic treatment-mediated sensitivity via escalated oxidative stress and altered biogenesis [55]. SIRT3's role in malignancy was also assessed in a study where silencing of SIRT3 resulted in a reduction of visible clones by 64%, when assessed by a clonogenicity assay. This enumerates the fact that SIRT3 downregulation leads to compromised mitochondrial metabolism and increased sensitivity to oxidative stress [56]. Promotion of metastasis by ROS is tricky. High levels of ROS lead to inhibition of metastasis in melanoma, whereas in other cancers, ROS promotes metastasis [54]. Considering the role of SIRT3 in regulating ROS homeostasis, studies have shown that SIRT3 is essential in extinguishing Src oxidation and Src/Fak signaling to inhibit cell migration and metastasis in breast cancer cells via ROS adjustment [54].

Mitochondrial fission is a process commonly implicated in tumor progression, where dynamin-related protein-1 (Drp1) is bonded to one of its receptors, mitochondrial fission factor (MFF), on the mitochondrial outer membrane. Mitochondrial fission has been widely correlated with cell death and mitochondrial integrity [57]. MFF is overexpressed in numerous cancers. MFF is linked to VDAC1 in the mitochondrial outer membrane, which partially explains its association with cancer progression. However, mechanistically, MFF silencing leads to an upsurge of mitochondrial outer membrane permeability and oxidative stress, which inturn leads to the triggering of mitochondrial-mediated cell death, thereby impeding tumor proliferation and metastasis in mice [57]. A very recent study has

determined the role of mitochondrial fission in cancer [58]. This study utilized phosphatidyl serine decarboxylase (PISD), an enzyme that orchestrates mitochondrial fission. It was evidenced that mitochondrial fission inhibits metastasis in triple-negative breast cancer cells. The study also enumerated that the alterations in mitochondrial fission not only inhibited cancer metastasis, cell migration, and cell invasion, but also repressed cancer cell signaling via ERK and Akt [58]. **2021**, , x FOR PEER REVIEW 6 of 13 α

**Figure 3.** Regulation of EMT by mitochondria.ROS: reactive oxygen species; PI3K: Phosphoinositide 3-kinases; SNPH: syntaphilin; HIF: hypoxia inducible factor. Tumorigenesis calls for hypoxicmediated reprogramming for metastasis. Bcl-2 family members regulate the PI3K pathway involved in metastasis progression. ROS generated during the metabolic process play a critical role in metastasis. Syntaphilin (SNPH), which generally arrests the mitochondrial trafficking in neurons, inhibits metastasis. In tumors with high expression of SNPH, mitochondria are anchored perinuclearly, resulting in lessened cell invasion and inhibited metastatic dissemination. In tumors with loss of SNPH expression, mitochondria are free to move to the cortical cytoskeleton via Kinesin/MIRO1 complexes. These cortical mitochondria fuel enhanced tumor cell invasion and correlate with poorer prognoses. Hypoxia can increase eNOS phosphorylation by activating the PI3K/AKT pathway. HIF-1α can also directly influence the expression of eNOS, which can be activated by phosphorylation of the serine 1177 residue, thereby, triggering migration and angiogenesis.

#### **5. Mitochondrial Stress Response in Cancer Spread**

Mitochondria are responsible for more than just energy production. Recently discovered mechanisms like mitochondrial unfolded protein response (mtUPR) and mitochondrial precursor over accumulation stress (mPOS) are paving new avenues for therapeutics and for understanding diseases better via mitochondria [59]. A genetic study in yeast denoted a novel protective mechanism named mPOS against mitochondrial protein import deficiency. mPOS is a newly-discovered pathway of proteostatic stress-mediated cell death due to mitochondrial dysfunction. mPOS is triggered by mitochondrial damage and the aberrant accumulation of mitochondrial precursors in the cytosol [60]. In parallel to this, mitoCPR was spotted in budding yeast. mitoCPR is a novel cellular response to defective mitochondrial protein import that protects mitochondrial functions [61]. The mitochondrion has an inherently stressful internal environment and it is speculated that dysregulation of stress signaling or an inability to switch on these adaptations during times of mitochondrial stress may underpin mitochondrial dysfunction and amount to pathological states overtime.

The role of mitochondrial chaperones in the cell stress response is quite intriguing. Gamitrinib, for instance, is a mitochondrial targeted HSP90 inhibitor with potential anti-cancer activity. Glioblastoma cells induced with low doses of gamitranib revealed accumulation of unfolded proteins in the mitochondria and a stress response gene characterized by upregulation of chaperones, especially Hsp70. Utilizing this target (mitoUPR) in mitochondria, TRAP-1 or CypD were ablated by genetic or chemical inhibitors. This resulted in the downregulation of NF-kB and related genes. Furthermore, there was an upregulation of pro-apoptotic genes, which aided in mitochondrial-mediated cell death. NF-kB has a wide role in tumor promotion and endorsement in the metastatic environment. Additionally, NF-kB plays a major role in treatment resistance and poor outcomes in cancer. Hence, targeting mitoUPR aids in concomitant loss of NF-kB, which inturn results in exposing the tumors to apoptosis-based therapies [62]. This clearly shows that mitoUPR can be a potential target for cancer therapy. Supplemental to this, new study evidence shows that mitoUPR under the absence of stress, as a part of an adaptive mechanism by cancer cells, results in reduced oxidative stress and is called mitohormesis. mitUPR has an axis with SIRT3, which supports invasion and metastasis. In addition, changes in the mtUPR gene resulted in poor clinical outcomes in patients with breast cancer [63].

#### **6. Mitochondrial Ion Channels as a Target in Combating Cancer**

The mitochondrial channels, characterized as either outer or inner membrane channels, are widely targeted in cancer therapies. The outer membrane channels include VDAC and the inner membrane channels include mtKATP, mtBKCa, mtIKCa, mtKv1.3, mtTASK-3, and the nonselective permeability transition pore (MPTP) [64]. Mitochondrial outer membrane channels participate in mitochondrial outer membrane permeabilization, while inner membrane channels modulate changes in membrane potential and thereby influence reactive oxygen (ROS) production and efficiency of the respiratory chain. ROS in turn may activate MPTP or the caspase-independent ROS-triggered parthanatos (poly (ADPribose) polymerase-1 dependent cell death). In addition, MPTP can also be triggered by Ca2+overload in the mitochondrial matrix or by IMM depolarization and by several other factors (for example oxidative stress) [65].

The basis for mitochondrial ion channels being targeted is due to their role in cancer metastasis. A brief description of the role of potassium channels and their role in cancer progression is that channels like IKCa control OXPHOS. Inhibition of the channel has no or only minor effects on cell proliferation in the presence of glucose, but forcing the cells to generate ATP exclusively via oxidative phosphorylation by culturing them in galactose, allowed researchers to understand that inhibition of the channel decreased proliferation. Kv1.3 is another channel that modulates the cell cycle. Mitochondrial calcium fluxes have also been shown to regulate cancer proliferation. Additionally, calcium channels also drive proliferation. The constitutively active Ca2+transfer from the endoplasmic reticulum (ER)

to mitochondria plays a crucial role inensuring viability of tumorigenic cells, and defects in this uptake into mitochondria lead to cancer cell death. The crosstalk between potassium and calcium channels isnot completely clear, but a putative K+/H+transporter, LETM1, has been shown on calcium influx/efflux into/from mitochondria, and silencing ofLETM1 promoted AMPK activation, cell cycle arrest, and autophagy [66].

MPTP can be activated indirectly by different drugs eliciting changes in inner membrane potential, causing ROS production, or leading to calcium overload in the matrix. MPTP opening leads to rupture of the mitochondrial outer membrane (MOM), which contributes to cytochrome-c release, a process required for apoptosome formation and subsequent activation of effector caspases.

#### **7. Mitochondria as a Therapeutic Target in Cancers**

The energy required for cancer cell migration, invasion, and metastasis is supplied by mitochondria. Suppression of the mitochondrial energy function can reduce the frequency of tumor cell metastasis and invasion (Table 1). Targeting dysregulated Drp1-dependent mitochondrial fission could supply a novel scenario for defeating breast cancer metastasis [8]. On the other hand, another novel therapeutic strategy to limit or prevent cancer metastasis is by potentially blocking EMT through targeting specific EMT biomarker genes that are correlated with mitochondria health, signal proteins of the mitochondrial reverse signaling pathway, specific metabolic enzymes, or metabolism-dependent epigenetic reprogramming [33]. As proof, the PI3K/AKT signaling pathway plays a key role in EMT progress and is considered to be a principal signaling pathway in cancer that prompts extensive transcriptional and metabolic reprogramming, specifically in mitochondria. PI3K has been considered a potential target for the prevention and treatment of metastatic tumors. Inhibitors of PI3K have been utilized in tumor treatment to inhibit mitochondrial ATP production and diminish glycolysis [67–69]. A recent study highlighted a compound called NSC130362, which belongs to the class of 1,4-naphthoquinones (NQs) and has vibrant pharmacological properties [70]; it has been shown to possess anti-cancer effects, including anti-proliferative and anti-angiogenesis activity [71,72], suppress glycolysis and mitochondrial function [73], and inhibit NF-κB signaling [74]. Natural products on the other hand have also received attention in cancer chemotherapy. Honokiol (HNK) is a potent anti-tumor agent that affects EGFR and mitochondrial function to inhibit the cancer cells' genesis and metastasis. A study has shown that HNK inhibits mitochondrial respiration, which leads to the induction of apoptosis in lung cancer cells [75]. There are other natural products that have been identified with direct or indirect effects on mitochondrial function in cancers (Table 2 [76]).




**Table 2.** Natural compounds affecting mitochondrial function and acting as cancer therapeutics [76].

#### **8. Conclusions**

Recent advances in the field of cancer biology have delineated mitochondrial dysfunctions in cancer. Tumors take advantage of the modulated mitochondrial function to escalate invasiveness. Key mechanisms like respiration are not only essential for tumor growth but also for navigating tumor cells into the circulatory system, facilitating metastasis. Mechanisms connecting mitochondrial dynamics to the development of metastasis remain a puzzle. Moreover, the capability of the mitochondria in allowing cancer cells to adapt to stress should be considered. Consequently, mitochondrial biogenesis might answer these questions and unravel the mechanisms useful for therapeutic strategies for cancer treatment.

**Author Contributions:** R.G. designed, conceptualized and drafted the review; W.L. drafted the review; R.M. reviewed, read and finalized the review. All authors have read and agreed to the published version of the manuscript.

**Funding:** This review received no funding.

**Acknowledgments:** Research Management Training Centre: RMTC-LKL- TAHMAZ Meditech; Dato Seri Harlem Shah-Lim Kon Lian (LKL Sinovac Genesis).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**



#### **References**


### *Review* **Oncogenic Effects of HIV-1 Proteins, Mechanisms Behind**

**Maria Isaguliants 1,2,3,4, \* , Ekaterina Bayurova 1,2 , Darya Avdoshina 1,2 , Alla Kondrashova 2 , Francesca Chiodi 3 and Joel M. Palefsky 5**

	- francesca.chiodi@ki.se

**Simple Summary:** People living with human immunodeficiency virus type 1 (HIV-1) (PLWH) are at increased risk of developing cancer despite successful antiretroviral therapy (ART). Here, authors suggest novel mechanism behind this phenomenon. HIV proteins, namely envelope protein gp120, accessory protein negative factor Nef, matrix protein p17, transactivator of transcription Tat and reverse transcriptase RT, are known to be oncogenic per se, to induce oxidative stress and to be released from the infected or expressing cells. These properties are proposed to underlie their capacity to affect bystander epithelial cells causing their malignant transformation, and to enhance tumorigenic potential of already transformed/cancer cells. HIV proteins can act alone or in collaboration with other known oncoproteins, specifically originating from the oncogenic human viruses such as human hepatitis B and C viruses, and human papilloma viruses of high carcinogenic risk, which cause the bulk of malignancies in people living with HIV-1 on ART.

**Abstract:** People living with human immunodeficiency virus (HIV-1) are at increased risk of developing cancer, such as Kaposi sarcoma (KS), non-Hodgkin lymphoma (NHL), cervical cancer, and other cancers associated with chronic viral infections. Traditionally, this is linked to HIV-1-induced immune suppression with depletion of CD4+ T-helper cells, exhaustion of lymphopoiesis and lymphocyte dysfunction. However, the long-term successful implementation of antiretroviral therapy (ART) with an early start did not preclude the oncological complications, implying that HIV-1 and its antigens are directly involved in carcinogenesis and may exert their effects on the background of restored immune system even when present at extremely low levels. Experimental data indicate that HIV-1 virions and single viral antigens can enter a wide variety of cells, including epithelial. This review is focused on the effects of five viral proteins: envelope protein gp120, accessory protein negative factor Nef, matrix protein p17, transactivator of transcription Tat and reverse transcriptase RT. Gp120, Nef, p17, Tat, and RT cause oxidative stress, can be released from HIV-1-infected cells and are oncogenic. All five are in a position to affect "innocent" bystander cells, specifically, to cause the propagation of (pre)existing malignant and malignant transformation of normal epithelial cells, giving grounds to the direct carcinogenic effects of HIV-1.

**Keywords:** human immunodeficiency virus type 1; epithelial cells; carcinogenicity; oxidative stress; reactive oxygen species; gp120; Tat; Nef; matrix protein p17; reverse transcriptase

#### **1. Introduction**

Immune suppression and related dysfunctions result in a high prevalence in people living with human immunodeficiency virus (PLWH) of HIV-1/AIDS-associated disorders,

**Citation:** Isaguliants, M.; Bayurova, E.; Avdoshina, D.; Kondrashova, A.; Chiodi, F.; Palefsky, J.M. Oncogenic Effects of HIV-1 Proteins, Mechanisms Behind. *Cancers* **2021**, *13*, 305. https://doi.org/10.3390/ cancers13020305

Received: 1 December 2020 Accepted: 4 January 2021 Published: 15 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

including so called AIDS-defining cancers (ADC)—Kaposi sarcoma (KS), non-Hodgkin lymphoma (NHL) and cervical cancer. In the era of antiretroviral therapy (ART), their rates have sharply declined: KS by 60–70% and NHL, by 30–50% compared to the pre-ART era. Still, the incidence of KS in PLWH remains elevated 800-fold, of NHL 10-fold and of ADC 4-fold, compared to their rates in the general population. There is also a significant increase in the number of yearly diagnosed cases of non-AIDS-defining cancers [1].

The incidence of these malignancies among PLWH remains elevated compared to that in uninfected population despite successful ART. Traditionally, this is linked to HIVinduced immune suppression with depletion of CD4+ T-helper cells, and exhaustion of lymphopoiesis, however, the immune suppression is much more complex than HIV-1 induced loss of CD4+ T cells. HIV-1 causes dysregulation of the innate immune system, persistent immune activation, dysfunction of the inflammatory response and immune system aging (senescence) early in HIV-1 infection. Successful ART ameliorates, but does not completely correct the major immune dysfunctions [2–6], substantial immunological impairment pertains even on the background of the successful ART [7,8] (for the latest review, see [9]). Hyper-immunoactivation and inflammation persisting in PLWH is recognized as a major cause of HIV-1 associated malignances. This abnormal immunoactivation emerges as the cumulative effect of thymic dysfunction, ART toxicity, persistent antigen stimulation caused by co-infections, microbial translocation, residual viremia and dysbiosis [10], aggravated by incomplete recovery of CD4+ T cell functions and intrinsic B and T cell defects on the background of persistent aberrant activation of monocytes, natural killer cells (NK) and innate lymphoid cells [7,11,12].The immune deficiency and dysfunction of the immune system may not be the only cause [13]. Under successful ART, HIV-1 should become latent, however, a study of HIV-1 integration sites in latently infected cell lines evidenced an ongoing viral replication [14], demonstrating that ART cannot fully suppress the process. Massive data have accumulated on the crucial role in high incidence of malignancies among PLWH of the residual virus production and circulating viral proteins. This review concentrates on their role in the high prevalence of cancers among individuals living with HIV-1.

#### **2. Prevalence of Non-AIDS Defining Cancers Increases Despite Successful Antiretroviral Therapy**

The category of non-AIDS-defining cancers (NADC) includes liver cancer related to infections with hepatitis B and C viruses (HBV and HCV), brain cancer, and cancers associated with infection with human papillomaviruses of high oncogenic risk (HR HPVs), specifically, the anal cancer.

#### *2.1. Liver Cancer*

Hepatocellular carcinoma (HCC) is the third-largest cause of cancer-related mortality on a global scale. It constitutes nearly the majority of liver cancer cases, followed by intrahepatic cholangiocellular carcinomas [15]. HCC is a recognized complication of liver cirrhosis, developing stepwise from regenerative to low-grade, then high-grade dysplastic nodules, although in some cases it may also develop de novo [15]. The burden of HCC is expected to increase worldwide in the next few decades, due to the population growth and aging expected in coming years [16]. Treated HIV-1 infection is associated with decreased survival in HCC, independent of stage, anticancer treatment, and geographical origins of the patients [16]. HIV-1 is not sufficient to cause liver cancer on its own, but may promote development of liver cancer by multiple mechanisms not yet fully understood [17]. Although the role of immune suppression in HCV-related HCC is not clear [17], mechanistic evidence suggests an accelerated progression of chronic liver disease to fibrosis and ultimately malignancy mediated by HIV-1-mediated impairment of antiviral CD4+ and CD8+ T-cell responses [18–20].

HIV-1 infection is characterized by increased microbial translocation resulting in elevated levels of circulating lipopolysaccharides (LPS) in the portal and systemic circulations [21]. LPS are well-known inducers of the innate immune activation. Increased levels

of LPS and/or soluble CD14 (sCD14; reflects LPS-induced monocyte activation) in PLWH on ART correlate with impaired recovery of CD4+ T cells. They also tightly correlated with multiple markers of immune activation, specifically, high levels of type I interferons and activated CD8+ T cells. In HIV-1 infection, these two parameters strongly predict disease progression [21]. In the liver, LPS activate hepatic stellate (HSC) and Kupffer cells (KFC), resulting in the generation of superoxide and release of proinflammatory and profibrogenic cytokines such as TNF-α, IL-1, IL-6, and IL-12 that induce liver damage and accelerated liver fibrosis [22]. Activation of Kupffer cells by LPS involves signaling through TLR-4, shown to govern the transition from chronic hepatic inflammation to hepatocellular carcinoma [23]. Another product released from bacterial cell walls, (1→3)-β-D-Glucan (βDG), emerges as an additional significant source of monocyte and NK cell activation, further contributing to immune dysfunction and inflammation [24].

Growing evidence accumulates of HIV-1 grossly affecting the liver. HIV-1-monoinfected patients demonstrate markers of liver fibrinogenesis/liver injury (by transient liver elastography) correlated with high plasma levels of HIV-1 RNA [25]. HIV-1 RNA has been detected in primary human hepatocytes both ex vivo and in vitro [26,27]. Also many hepatocyte cell lines are permissive to a low level HIV-1 infection although the nature of receptor(s) for HIV-1 on liver cells is unclear [28]. HIV-1 can also directly infect Kupffer cells; infectious replication-competent HIV-1 has been isolated from KFC obtained from liver at autopsy from three HIV-1-infected individuals who died while on ART [29]. Another target of HIV-1 are hepatic stellate cells (HSC), the primary cells involved in liver fibrogenesis, affected through both direct HIV-1 infection and HIV-1 exposure [30]. Interactions of HIV-1, specifically its envelope protein gp120, with chemokine receptors CCR5 and CXCR4 induce cell signaling in HSCs and immune cells within the liver promoting inflammatory responses [31,32]. Direct HIV-1 infection of KFC results in the amplification of proinflammatory responses to LPS [33], enhanced fibrosis and cirrhosis, and exhaustion of virus-specific T-cells. HIV-1-infected HSCs produce collagen I and release monocyte chemoattractant protein-1 (MCP-1) [34]. Exposure of HSCs to HIV-1 results in the production of reactive oxygen species (ROS), and expression of collagen and tissue inhibitor of matrix metalloproteinases-1 (TIMP1) [35]. These events, together with abnormalities in the gut microbial communities, significantly contribute to the high rates of liver cancer in PLWH [36] (Figure 1).

Even more important driving force of hepatocellular carcinogenesis in PLWH is coinfection with HBV and HCV [37]. In HIV-1/HBV and HIV-1/HCV co-infected patients, HIV-1 infection decreases the rate of spontaneous viral clearance from the liver, accelerates fibrogenesis and increases the rates of liver-related morbidity and mortality, including the development of HCC [30,38]. In HIV-1/HCV co-infected individuals HCC occurs at a younger age and after a shorter period of HCV infection than in HIV-1 negative individuals [39], with the risk to develop HCC increasing each year by 11% [19]. Important risk factors for the progression to liver cancer are high HBV and HCV viral loads [18,19,40]. They are associated with (over)expression of viral oncoproteins known to induce oxidative stress and chromosomal instability/genomic damage, promote chronic inflammation with liver damage resulting in the malignant transformation of liver cells [41,42] (Figure 1).

#### *2.2. Brain Cancer*

PLWH are highly predisposed to developing brain cancer, including primary central nervous system lymphomas (PCNSL) and glioblastomas (GBM) [43,44]. In pre-ART era, brain tumors were registered in 10% of PLWH [43]. Prevalence of PCNSL in AIDS patients was 3600-fold greater than in the general population, reaching 12% in AIDS patients [44]. ART has dramatically reduced these rates, possibly due to the effect of protease inhibitors [45]. Still, the prevalence of brain tumors in PLWH appears to be higher than in general population: in USA; recorded prevalence of PCNSL in HIV-1 infected is 8.4% compared to <3.3% in the general US population [45,46] Also GBM occurs in PLWH (in various stages of HIV-1 infection) at a younger age and at a frequency 5.4- to 45-fold higher than in

the general population [47]. Furthermore, the median survival rate in patients with GBM for PLWH is shorter than for HIV-1-negative patients receiving same treatment (an average of 8 compared to 14 months, respectively) [48].

**Figure 1.** The effect of HIV-1 on cells of the liver. Infection with HIV-1 and even exposition of hepatocytes (HP), hepatic stellate cells (HSC), Kupffer cells (KFC) to HIV-1 leads to production of reactive oxygen species (ROS) and induction of proinflammatory microenvironment, which in turn, promote/enhance replication of HBV, HCV, as well as HIV-1 itself, resulting in enhanced fibrosis, cirrhosis and development of hepatocellular carcinoma (HCC). Infections are depicted in red, secondary effects in dashed black, and events leading to tissue damage in ochre-colored lines.

The nature of the brain tumor-HIV-1 relationship is not fully understood. The majority of these tumors are central nervous system lymphomas but gliomas may develop as well. GBM tumors appear approximately three years after HIV-1 infection [43]. The stimulatory effect of HIV-1 infection on the development of GBM has been associated with reduced immune surveillance. However, survival of PLWH after GBM diagnosis is not associated with CD4+ cell counts [47]. The absence of a correlation between GBM development & progression with immune incompetence [47,49–51] indicates that aggressive tumor behavior is not a direct consequence of the immune deficiency and suggests direct involvement of HIV-1 in the initiation and progression of brain cancers. Importantly, HIV-1 infection in the brain is not limited to microglia/macrophages, but also affects astrocytes, which can then serve as a potential reservoir for further productive infection, viral persistence, and latency [52,53].

#### *2.3. Squamous Cell Carcinomas*

PLWH suffer from squamous carcinomas at numerous sites including the lung, anogenital region, oral cavity, epiglottis and cervix. Many of these malignancies are associated with infection by human papillomaviruses of high carcinogenic risk (HR HPVs). Similar to the rates of liver and brain cancer, the rates of HR HPV-associated cancers in PLWH are steadily growing despite successful ART [54–56]. CD4+ levels and resulting immune

suppression play a prominent role in controlling HPV replication and development of early disease, particularly, the development of pre-cancerous intraepithelial neoplasia: in PLWH, the probabilities of HPV acquisition and development of intraepithelial neoplasia increase in proportion to the loss of CD4 T cells [57]. However, progression to high-grade lesions and further to cancer is not predetermined by CD4+ depletion, i.e., is not a straightforward outcome of HIV-1-induced immune suppression [58,59], but rather an outcome of the accumulated changes in the host cell genome and transcriptome involving tumor suppressor genes, apoptosis-related genes, DNA damage-repair genes, and cell cycle regulatory genes [58,60,61].

Question arises how could this rely to the epithelial cells which are considered to be non-susceptible to HIV-1 infection and non-permissive to HIV-1 replication? HIV-1 infects a variety of immune cells, such as CD4+ T lymphocytes and monocytes/macrophages. However, several studies show that it may also infect or rather "trespass" other cell types, in which HIV-1 virions and individual HIV-1 proteins were repeatedly detected. In primate models, application of HIV-1 to the surfaces of oropharyngeal [62], anal/rectal [63], cervicovaginal and foreskin/penile [62,64–66] epithelia was shown to lead to subsequent systemic infection of HIV-1-susceptible immune cells, indicating that HIV-1 travels through these tissues to reach its targets. Indeed, application of HIV-1 to human foreskin, vaginal and cervical tissue explants ex vivo leads to the transmission of HIV-1 across these epithelia [64,66–71].

These findings are not restricted to the epithelial cells of the reproductive tract. HIV-1 antigens and RNA were detected in gastric epithelial cells in the biopsy and autopsy samples of HIV-1-infected patients; furthermore, TEM analysis visualized HIV-1 particles in the cytoplasm of gastric epithelial cells [72]. Interestingly, HIV-1 load in blood positively correlated with the number of HIV-1-infected gastric epithelial cells. The latter increased with progression of chronic infection, being significantly higher at the AIDS compared to the asymptomatic stage. HIV-1 infection of gastric epithelial cells associated with a severe inflammatory response in the gastric mucosa manifested by infiltration and aberrant activation of the immune cells [72].

Another example is presented by human mammary epithelial cells (MEC). MEC express HIV-1 receptors CD4, CCR5, CXCR4, and galactosyl ceramide (GalCer). Although the evidence for direct MEC infection by HIV-1 was missing, HIV-1 virions were found in the endosomal compartments of these cells. Furthermore, activated CD4+ T cells co-cultured with HIV-1-exposed MEC were productively infected with HIV-1 [73]. This confirmed that mammary epithelial cells can endocytose HIV-1 and facilitate its transfer to CD4+ T lymphocytes [73]. At the other end, a contact-dependent HIV-1 transfer was shown from HIV-1-infected macrophages to both primary and immortalized renal tubule epithelial cells (RTE). Live imaging of HIV-1 infected RTE cells revealed four different fates: latency, hypertrophy, cell death, and proliferation [74]. HIV-1 can also enter airway epithelial cells and alter their function by increasing the expression of inflammatory mediators [75]. This data unequivocally demonstrate that HIV-1 could be internalized and/or sequestered by human epithelial cells of different origins.

#### **3. Mechanisms Underlying HIV-1 Pathogenicity in Epithelial Cells**

In CD4+ cells HIV-1 was reported to preferably integrate into cancer-associated genes or cell cycle regulation genes dysregulation of which can lead to cancer formation as was described for other retroviruses [76–78]. Replication of HIV-1 in epithelial cells has not been shown except for the early findings of human uterine epithelial cells productively infected by HIV-1 with reverse transcription of viral RNA, transcription of viral DNA, and secretion of infectious virus [79]. Of note, co-cultivation of human CD4+ T cell lines with HIV-1-infected uterine epithelial cells (and also by virions released by these cells) led to HIV-1 infection of the CD4+ T cells [79]. Bulk of data accumulated so far evidence sequestration of HIV-1 by human epithelial cells of different origins without evidence of productive replication or integration. However, a "real" infection can take place as well.

HIV-1 was shown to hijack other viral Envs to directly enter CD4-negative cells through pseudotyping [80–82]. Lately, Tang Y. et al. have shown that HIV-1 infected T cells can fuse to and transfer the virus to placental trophoblasts, if the later express on their surfaces the envelope glycoprotein of human endogenous retrovirus family W1, syncytin [83]. This leads to the formation of an HIV-1 reservoir in the epithelial cells [83]. Syncytin-1 derives from a family of endogenous retroviruses and originates from HERVW1 infection of human germ cells [84]. Expression of syncytin could be a common feature of an epithelial cell which make them susceptible to HIV-1 via a "non-canonical" route of HIV-1 infection. These are not necessarily the epithelial cells of placenta. According to the recent preliminary report published in bioRxiv, HIV-1 can infect human bronchial epithelial cells; after exposition to HIV-1 they were shown to express p24 and contain latent HIV-1 provirus [85]. These findings along with the data by Asin SN et al. [79] indicate that in certain cases epithelial cells can be infected with HIV-1, possibly as a one-round abortive infection with reverse transcription of RNA and integration of the proviral DNA governed by respective enzymes constituting HIV-1 virion. Such integrated proviral HIV-1 DNA would not only serve as an HIV-1 reservoir, but would also give progeny to the genetically modified cells (with proviral DNA inserts) susceptible to malignant transformation. The observation by Hughes K et al. of a proliferation of HIV-1 infected epithelial cells consistent with clonal expansion of individual cells ideally fits this scenario [74].

HIV-1 antigens may also affect epithelial cells without infecting them. Epithelial cells may respond to the defective virions incapable of productive infection or freely circulating HIV-1 antigens shed by the infectious or defective virions. Addition of HIV-1/HIV-1 antigens to the epithelial cells generates an inflammatory microenvironment or rather microenvironmental immune abnormalities [86–88] (as those associated with HR HPV infection). Microenvironment of B-cell lymphomas in PLWH is characterized by expression of CD3, CD4, CD8, CD56, CD68, CD163, FOXP3, TIA1, granzyme B, perforin, CD57, CD34 and PD-1 [89], and enrichment with soluble factors, including cytokines IL-1, IL-2, IL6, IL10, and chemokines of the CCL and CXCL families [89,90]. Such microenvironment was also found in the intraepithelial cancerous lesions of PLWH [91]. Studies on the mucosaassociated lymphoid tissue system (MALT) in PLWH have shown abnormal immune responses in the mucosal milieu, including upregulation of expression of multiple regulatory cytokines such as IL-8, IL-23, TNF-α, IL-17A, and IFN-γ (TNF-α/IL-17A/IFN-γ triad), the depletion of Langerhans cells and CD3+ lymphocytes, increases in Foxp3+ T-regulatory cells, and in local lymphocyte infiltrates composed by CD8+ T cells, associated with the development of high-grade squamous intraepithelial lesions (HSILs) [91–94].

Furthermore, presence in the epithelial cells of HIV-1/HIV-1 proteins modulates their capacity to express E-cadherin, a marker of epithelial to mesenchymal transition (EMT) [69,75,95]. HIV-1 interaction with the surface of mucosal epithelial cells was also shown to activate the transforming growth factor-beta (TGF-β) and mitogen-activated protein kinase signaling pathways [96]. When activated, these pathways may lead to the disruption of epithelial junctions and EMT [97]. Indeed, EMT was induced by exposure of oral keratinocytes from HIV-1-negative individuals to HIV-1 virions as well as Tat and gp120 proteins [98]. Within premalignant cells or in the environment of the malignant cells, HIV-1 driven EMT would promote motile/migratory cells and accelerate the neoplastic process.

Altogether these observations imply direct carcinogenic effect(s) of HIV-1 virions and/or antigens. This concept, proposed in 2002 by B Clarke & R Chetty [58] and four years later by Palefsky JM [59], is now supported by considerable experimental proof. It brings up several issues of importance for epithelial cells: (i) could malignant transformation be promoted by cooperation of HIV-1 with other oncogenic viruses; (ii) which HIV-1 antigens are implicated; and finally (iii) what are the underlying molecular mechanisms?

#### **4. Potentiation of Carcinogenesis by Interactions of HIV-1 with Other Oncogenic Viruses**

Oncogenic transformation associated with virus infection was for a long time considered to result from a mono-infection (infection with a single virus). However, it is now established that in many cases induction of cancer depends on the simultaneous presence and interactions of multiple viral agents in diverse combinations. Viruses co-infecting human tissues may have synergistic or regulatory effects on carcinogenesis, targeting existing neoplastic cells as well as their microenvironment including reactive T-cells, B cells and macrophages, and non-immune cells such as endothelial cells. HIV-1, in particular, potentiates the effects of EBV, KSHV, HCV, and HPV oncogenes, promoting carcinogenesis in individuals co-infected with HIV-1 and EBV, KSHV, HCV, HBV, and HPV. Here, we will focus on molecular interactions of HIV-1 with HBV, HCV, and HPVs.

Progression to liver cancer/HCC in HIV-1/HBV and HIV-1/HCV co-infected patients is promoted by direct and indirect interactions between these viruses and their antigens within the cells harboring HIV-1 due to infection or sequestration of the virion (viral proteins). HIV-1 infection of hepatic cell lines increases the expression of HBV antigens [27]. HIV-1 gp120 causes intracellular accumulation of HBV DNA as well as HBsAg causing hepatotoxicity [99]. Direct interaction of HIV-1 and HBV in liver cells has been demonstrated, with the HBV X protein interacting with HIV-1 Tat to facilitate HIV-1 replication [99]. Upon co-cultivation of HIV-1 infected Jurkat cells with hepatocytes, up-to 16% of the latter acquire Nef. Sequestered Nef alters the size and numbers of lipid droplets (LD), inducing 1.5 to 2.5 fold up-regulation of replication of HCV subgenomic replicon, a remarkable finding in relation to the initially indolent viral replication. Nef also dramatically enhances the ethanol-mediated up-regulation of HCV replication accelerating progression to HCC [100]. HIV-1 gp 120 also causes TGF-β mediated up-regulation of HCV replication [86]. Taken together, these data indicate that HIV-1 and single HIV-1 proteins are critical elements in accelerating progression of liver pathogenesis by enhancing HBV and HCV replication and coordinating production of key intra- and extra-cellular molecules that orchestrate liver decay [100].

One of the mechanisms of HIV-1 potentiation of liver cancer is the induction of oxidative stress. HCV, HIV-1 (and antiretroviral therapy) act together to activate production of ROS in HSCs and hepatocytes. ROS promote phosphorylation of the major mitogenactivated protein kinases active in human cells, p38 kinase, c-JUN N-terminal kinase (JNK) and extracellular signal-regulated kinase (ERK) that control cell growth, differentiation and apoptosis. In their turn, the phosphorylated p38 MAPK, JNK, and p42/44 ERK phosphorylate nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) protein complex, mastering transcriptional regulation of inflammation and cell death [31]. Following these events, phosphorylated NF-κB translocates to the nucleus, and where it normally modulates the production of both pro- and antifibrogenic/antiapoptotic genes, ensuring that liver cells are protected from apoptosis, but are capable to build the required inflammatory and immune responses [101]. In the presence of LPS, NF-κB can upregulate the expression of profibrogenic genes, such as procollagen α1, transforming growth factor β1 (TGF-β1) and tissue inhibitor of MMPs (TIMP-1) [31,101]. This process is accelerated by HIV-1/HIV-1 proteins: exposure of hepatocytes to HIV-1/HIV-1 proteins results in the elevated production of ROS and increased expression of collagen and TIMP1, further amplified by HCV infection, and even exposure to infectious HCV [35]. Taken together, these data indicate that HIV-1-mediated potentiation of hepatocellular carcinogenesis reflects a concerted action of HIV-1, HBV and HCV as viruses and/or individual viral proteins (Figure 1). Based on compelling data, McGivern & Lemon even suggested that the path to hepatocellular carcinoma in chronic hepatitis C shares important features with the carcinogenesis induced by HPV [102].

The increased risk of PLWH developing HPV-associated cancer can also, at least in part, be due to the interactions between HIV-1 and HPV. In general, epithelial cells of PLWH show loss of E-cadherin, and upregulation of vimentin and TGF-b1 expression with spindle-like morphology indicating induction of TGF-b1-dependent EMT, critical for malignant transformation. As noted above, EMT is induced not only due to HIV-1 infection, but also through exposition of epithelial cells to HIV-1 proteins [69,75,95,97]. EMTinduced keratinocytes can then be infected with pseudoviral HPV16 particles (HPV-16 PsVs) and whole HPV16 virus, with infected cells expressing viral oncogenes E6/E7, whereas unexposed keratinocytes could not be infected with either PsVs, or infectious HPV16. Furthermore, "HIV-1-induced" EMT keratinocytes could be transformed with HPV16 DNA, transformed cells showing active proliferation and migration [103]. This confirms that prolonged exposure to and interaction of HIV-1 with oral and anal epithelial cells induces EMT. EMT-induced loss of cell adhesion and increased proliferation and mobility of epithelial cells play a critical role in HPV infection and HPV-associated transformation. HIV-1-induced EMT in the orogenital mucosa may promote progression of pre-cancerous HPV-associated neoplasia to cancer in HIV-1-infected individuals [103].

"Molecular" cooperation between HIV-1 and HPV has not been sufficiently well characterized, but there are relevant examples in this field. Tat protein was shown to transactivate the HPV long control region and increase expression of oncoprotein E7 of HPV18 in HeLa cells [104,105]. Tat can upregulate the expression of E6 and E7 oncoproteins of HPV type 16 in HPV 16-infected human oral keratinocytes, notably enhancing the in vitro proliferative capacity of these cells [106,107], and increase the transcription of E2 modulating HPV replication [108]. The direct angiogenic effects of Tat [109] or its capacity to up-regulate the expression of E6 and E7 of HR-HPVs [110] allows Tat to favor the angiogenic switch in high-grade CIN. We have shown that gp120 and reverse transcriptases (RT) derived from various HIV-1 strains, can increase the expression of HPV 16 E6 in a cervical cancer cell line containing full-length HPV 16 genome Ca Ski (Figure 2), while HIV-1 p24 exerts no effect. In similar conditions, gp120 increases the expression of HPV16 E6 also in HPV16 immortalized anal epithelial AKC2 cells [104,106,111,112]. Furthermore, Tugizov et al. have shown that in the HPV-immortalized anal and cervical epithelial cells Tat and gp120 proteins induce the EMT phenotype, leading to increased migration of cells via collagen membranes [103]. The data on the interaction(s) between HPV and other HIV-1 proteins is missing.

Overall, these findings indicate that the increased incidence of AIDS-defining and non-AIDS defining forms of cancer in PLWH may reflect the direct or indirect, often concerted, carcinogenic effect(s) of HIV-1 and/or individual HIV-1 proteins on diverse infected as well as uninfected bystander cells. Furthermore, some HIV-1 proteins appear to be directly involved in cell transformation and propagation of malignant cells.

**Figure 2.** Transcription of oncoproteins E6 and E7 of HPV 16 in Ca Ski cells treated with HIV-1 proteins. Ca Ski cells harboring 600 full genomic copies of HPV 16 (ATCC CRL-1550) were cultured in RPMI-1640 medium (PanEco, Moscow, Russia) supplemented with 10% FBS and 100 mg/mL penicillin/streptomycin mix at 37 ◦C in an 5% CO<sup>2</sup> and split every 4 days. A panel of recombinant HIV-1 proteins: gp120 [113]; p24 (NIBSC ARP 694.1); RT of HIV-1-1 clade B HXB2 strain [114], drug resistant (dr) RT of HIV-1-1 clade B isolated from patient with multiple drug resistance mutation (RT1.14; [114]) and RT of Moloney murine leukemia virus (MMTV) (CRIE, Moscow, Russia) were added to the culture medium, typically in concentration of 1 ng/mL, and incubated for 48 h, according to the methodology described previously by Lein K. et al. [115] Total RNA was extracted and reverse transcribed as described by Jansons et al. 2020 [116]. Gene-specific PCRs were performed on Rotor-Gene 6000 (Qiagen, Darmstadt, Germany) with SYBR Green kit (Evrogen, Moscow, Russia) with primers specific to HPV 16 E6 and E7 [117]. Expression of mRNA, assessed by the standard ddCt method, was normalized to expression of 18S RNA (18Srna\_rt\_f: GTAACCCGTTGAACCCCATT; 18Srna\_rt\_r: CCATCCAATCGGTAGTAGCG), and presented as fold change compared to cells treated with p24, as was recommended earlier [118]. Values represent mean ± SD from two independent assays run in duplicates. \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05 by the ordinary two-way ANOVA with Sidak's multiple comparisons test.

#### **5. HIV-1 Antigens Involved in Cell Transformation and Tumor Propagation**

#### *5.1. Transactivator of Transcription (Tat)*

Tat has long since been known to influence cell cycle progression. In HeLa cells, Tat induces a significant increase in the levels of proliferation markers together with the reduction in the expression of cell cycle inhibitors of transcription [119]; it inhibits epithelial differentiation, blocks apoptosis in vitro and accelerates tumor formation in vivo [119]. In addition, Tat significantly increases in vitro migration in the absence of fetal calf serum [119]. These results suggest that HIV-1 may enhance carcinogenesis by promoting cell cycle progression [111]. Furthermore, it has been shown that binding of Tat to Tat-interacting promoter 30 (TIP30) enhanced EMT and metastasis of non-small cells lung cancer cells by regulating the nuclear translocation of Snail [120]. One of the possible mechanisms of Tat induced carcinogenesis is blocking at the mRNA level of the expression of a Rb family member pRb2/130 and cyclin-dependent kinase inhibitors p21 and p17 [111]. The transduc-

tion domain of Tat specifically attenuates growth of polyamine-deprived tumor cells [121]. Tat is also known to modulate VEGF and targets VEGFRs which increases angiogenesis and supports tumor growth [122]. Furthermore, Tat alters DNA repair in host cells, potentially leading to genomic instability [123,124]. Specifically, Tat induces expression of the DNA polymerase beta gene, which codes for a central mediator in the DNA base-excision repair pathway [125]. It also interferes with double-strand break DNA repair, as cellular extracts containing Tat possess a reduced capacity to re-join linearized DNA [126], indicating that Tat, as well as cellular co-factors of Tat, interfere with repair of double-stranded DNA breaks [123].

#### *5.2. Envelope Glycoprotein gp120*

Glioma cells were shown to interact with the HIV-1 envelope protein gp120. This interaction promotes proliferation, migration, survival and stimulates glycolysis in glioma cell lines and tumor growth in animal models [127]. Increased glycolysis, also known as the Warburg effect characteristic of malignancy [128], results in increased protein and lipid synthesis, and promotes uncontrolled propagation (both proliferation and invasion) of tumor cells, as it provides them with glycolytic intermediary precursors required for the synthesis of DNA, proteins and lipids [127,129]. As Tat, gp120 induces EMT and cell migration through the TGF-B1 and MAPK signaling pathways [115,130].

#### *5.3. Accessory Protein Negative Factor (Nef)*

Nef is one of the earliest and most abundantly expressed HIV-1 proteins. Nef has the ability to modulate multiple cellular signaling pathways in both CD4+ lymphocytes and macrophages. Nef inhibits the apoptotic function of p53 due to its ability to decrease p53 protein half-life and, consequently, p53 DNA binding activity and transcriptional activation [131]. Both internalized and ectopic expression of Nef in endothelial cells synergizes with Kaposi's sarcoma (KS) KSHV oncoprotein K1 to facilitate vascular tube formation and cell proliferation, and enhance angiogenesis in the chicken chorioallantoic membrane (CAM) model. In vivo experiments further indicate that Nef can accelerate K1-induced angiogenesis and tumorigenesis in athymic nu/nu mice [132]. On non-small lung cancer A549 cells, Nef promotes cell proliferation, migration, anchor independent growth and reduces the levels of expression of p53, increasing the aggressiveness of cancer cells [133].

#### *5.4. Reverse Transcriptase (RT)*

We have shown that constitutive expression of HIV-1 RT in murine mammary gland adenocarcinoma 4T1 cells leads to upregulation, in a concentration-dependent manner, of the expression of the transcription factors Twist and Snail tightly involved in EMT [134]. In vivo, expression of RT by 4T1 cells results in enhanced tumor growth and potentiates formation of metastasis in distal organs of immunocompetent syngenic mice [134]. Interestingly, this is not a common property of the reverse transcriptases, as constitutive expression of enzymatically active reverse transcriptase domain of telomerase reverse transcriptase, on contrary, suppressed both tumor growth and metastatic activity of 4T1 cells [116].

#### *5.5. Matrix Protein p17*

Matrix/p17 protein induces expression of chemokines [135], exerts pro-angiogenic [136] and lymphangiogenic [137] activities, and deregulates the biological activity of diverse cells of the immune system [138]. Overall, p17 generates a prolymphangiogenic microenvironment, predisposes the lymph node to lymphoma growth and metastasis [137] and promotes the aggressiveness (propagation) of human triple-negative breast cancer cells [139]. In a HIV-1 transgenic mouse model of lymphoma, only expression of HIV-1 p17, but not of other HIV-1 proteins, induced spontaneous B-cell lymphomas in HIV-1 transgenic mice, with p17 expressed at high levels in the early stages of the disease [140]. Murine lymphoma tissues exhibited enrichment in expression of the recombination-activating genes (Rag1/2) [140].

The latter suggests that intracellular signaling induced by p17 leads to genomic instability and promotes the transformation [140].

Thus, several HIV-1 proteins are directly or indirectly oncogenic, stimulating transformation of healthy cells and propagation and aggressiveness of already existing cancer cells. These oncogenic properties are linked to two essential characteristics of these proteins: their capacity to induce oxidative stress with production of reactive oxygen species and their ability to exit HIV-1-infected cells (active or passive transport).

#### **6. Oncogenic HIV-1 Proteins Induce Oxidative Stress**

Virally-induced cancer evolves over long periods of time in the context of a strongly oxidative microenvironment, on the background of chronic inflammation. Oxidative stress induced by chronic viral infection is one of the factors driving neoplastic transformation, ultimately leading to oncogenic mutations in many cellular signaling cascades that drive cell growth and proliferation [42,141]. Oxidative damage of chromosomal DNA and chronic immune-mediated inflammation are key features of HBV, HCV, HPV, and HIV-1 infections [42,141]. As we have earlier reviewed, numerous lines of evidence show that HIV-1 infection triggers pronounced oxidative stress in both laboratory models and the context of in vivo infection by deregulation of oxidative stress pathways with escalation of ROS production and by inducing mitochondrial dysfunction [141]. As a result, PLWH exhibit multiple markers of oxidative stress including DNA damage [134,142]. The enhancement of ROS production is mediated by the envelope protein Gp120, Tat, Nef, RT, and p17 [141–146].

#### *6.1. Transactivator of Transcription*

Tat induces oxidative stress both directly and indirectly via several independent mechanisms. The first involves the NADPH oxidases [147], and in the second, an enzyme involved in the catabolism of biogenic polyamines, spermine oxidase (SMO) [148], and the third, a mitochondrial dysfunction [149]. A detailed analysis of the levels of ROS in different subcellular compartments of the HIV-1 infected cells revealed a strong increase in the levels of H2O<sup>2</sup> in the endoplasmic reticulum (ER), demonstrating with the help of genetically encoded ratiometric sensor HyPER [150,151]. This indicated the involvement in H2O<sup>2</sup> production of NOX4 which primarily resides in ER [152]. The levels of H2O<sup>2</sup> in the cytoplasm and mitochondria were not elevated [151]. The above activities of Tat are thought to underlie the onset of HIV-1-associated dementia [109,150].

#### *6.2. Envelope Protein Gp120*

Early findings indicated that gp120 increases free radical production from monocytederived macrophages (MDM) detected by spin-trapping methods, and that the spin trap adduct results from a reaction involving nitrogen oxide NO or its closely related oxidized derivatives [153]. We have earlier summarized a profound role of gp120 in the induction of oxidative stress [141], namely gp120 induces ROS production in cell lines of lymphoid origin, in the endothelial brain cells, astrocytes, neurons and microglia. In astrocytes, it enhances ROS production by several parallel mechanisms: via Fenton–Weiss–Haber reaction, NOX2 and NOX4, and cytochrome P450 2E1 (CYP2E1) [154,155]. The latter is mediated through the upregulation of CYP2E1 expression. In cancer (neuroblastoma) cells, gp120 induces proline oxidase that synthesized pyroline-5-carboxylate with concomitant generation of ROS (reviewed in [141]).

The effect of HIV-1/HIV-1 proteins on the cellular antioxidant defense system is controversial. They can both suppress and enhance antioxidant defense pathways [141]. Gp120 was shown to induce oxidative stress response. It up-regulates functional expression in cultured astrocytes of multidrug resistance protein 1 (Mrp1) which effluxes endogenous substrates glutathione and glutathione disulphide involved in cellular defense against oxidative stress [156]. It also upregulates the expression of nuclear factor erythroid derived 2-related factor 2 (Nrf2), a basic leucine zipper transcription factor which is known to regulate antioxidant defensive mechanisms) in human astrocytes, stimulating expression of

key antioxidant defensive enzymes hemoxygenase (HO-1) and NAD(P)H dehydrogenase quinone1 (Nqo1) [157]. Pre-treatment of astrocytes with antioxidants or a specific calcium chelator BAPTA-AM, significantly blocks the upregulation of Nrf2, HO-1 and Nqo1 [157].

#### *6.3. Accessory Protein Negative Factor*

Nef protein has pro-oxidant activity in microglial cells and in neutrophils. It first induces phosphorylation and then translocation of the cytosolic subunit of NADPH oxidase complex p47(phox) into the plasma membrane which in turn induces superoxide anion release from macrophages [158,159]. As a multifunctional HIV-1 protein, Nef also activates the Vav/Rac/p21-activated kinase (PAK) signaling pathway involved in activation of phagocyte NADPH oxidase (thus, Nef indirectly activates NADPH oxidase) [160]. This leads to the dramatic augmentation of the production of ROS [100], and enhancement of cell responses to a variety of stimuli (Ca(2+) ionophore, formyl peptide, endotoxin) [160]. It also leads to decreased tolerance of the cells to hydrogen peroxide, specifically in astrocytes which normally support neuronal function and protects them against cytotoxic substances including ROS [161]. Rac1-dependent NOX2-mediated reactive oxygen species production was shown to contribute to ongoing HIV-1-related vascular dysfunction [162].

#### *6.4. Reverse Transcriptase*

We have previously demonstrated that expression of RT by human cells induces production of ROS [163]. Later studies demonstrated that this is a property of different RT variants, including drug resistant variants, and variants retargeted for lysosomal processing and secretion [114,163]. Expression of all RT variants led to an increase in the levels of expression of Phase II detoxifying enzymes HO-1 and Nqo-1. Artificial secretion of RT resulted in a decrease of RT capacity to induce oxidative stress with a decrease in the production of ROS compared to the parental enzyme [114].

#### *6.5. Matrix Protein p17*

There is no direct evidence of p17-induced oxidative stress. However, p17 possesses specific structural motifs defined as "*coiled coil*" sequences, and has a high propensity to form multimers, mis-fold and aggregate, forming amyloidogenic assemblies [164,165]. This is typical to amyloidogenic proteins actively involved in the pathogenesis of many human diseases, such as Alzheimer's disease and Parkinson' disease. Amyloidogenic assemblies are toxic, specifically to neural cells. Experiments in the invertebrate nematode *Caenorhabditis elegans* as a "biosensor" demonstrated that p17 significantly inhibits its pharyngeal contractions as do the amyloidogenic proteins [166]. Intrahippocampally injected into mice, p17 induced neurocognitive disorders, comparable in strenght to the effects of other known amyloidogenic proteins [166]. Interestingly, amyloidogenic proteins (typically amyloid-beta peptide Aβ) bound to redox active metal ions, such as copper, catalyse the production of ROS, in particular the most reactive one, hydroxyl radical. This effect may underlie the observed oxidative damage exerted by Aβ peptide on itself and on the surrounding molecules (proteins, lipids, DNA) [167]. One can hypothesize that matrix protein p17 with its amyloidogenic assemblies may trigger the production of ROS through a similar mechanism.

Thus, HIV-1 proteins with known oncogenic/mitogenic potential, Tat, gp120, Nef, RT, and potentially p17, have a potential to directly or indirectly induce oxidative stress, which could be one of the mechanisms by which they induce and potentiate carcinogenesis (Figure 3). Interestingly, HIV-1 proteins with an oncogenic potential involved in the induction of oxidative stress, such as Tat, gp120, Nef, RT, and possibly p17, can be found outside of the cells in which they are expressed.

**Figure 3.** Suggestive mechanism of direct carcinogenic effects of HIV-1 proteins. HIV-1 infected cells express and release gp120, Tat, Nef, p17, RT, each capable of the induction of oxidative stress. (1) p17 may trigger the production of ROS through binding of redox active metal ions by its amyloidogenic assemblies [167]. (2) Nef may indirectly activate NADPH oxidase by activating the Vav/Rac/p21-activated kinase (PAK) signaling pathway involved in phagocytic NADPH oxidase activation and produce peroxynitrite [160]. (3) Tat induces oxidative action through several independent mechanisms via NADPH oxidase, spermine oxidase (SMO) induction and mitochondrial dysfunction [148]. (4) RT induces ROS through unknown mechanisms. There is ROS –dependent activation of the Twist [134], which regulates the expression of Nrf2, which stimulating the expression of antioxidant enzymes (HO1, Nqol1). In addition, the Twist regulates the expression of the Snail. Both transcription factors, Twist and Snail, are involved in epithelial to mesenchymal transduction (EMT). (5) Gp120 increases free radical production from monocyte-derived macrophages (MDM) inducing nitrogen oxide (NO). In astrocytes (AS), it enhances ROS production by several parallel mechanisms: via cytochrome P450 2E1 (CYP2E1), NOX2 and NOX4, and the Fenton-Weiss-Haber reaction. Multidrug resistance proteins (Mrps) involved in cellular defense against oxidative stress. Mrp4 (isoform of Mrp) involved in the regulation of ROS and it acts against ROS [156]. In neuroblastoma cells (NB) gp120 was shown to induce proline oxidase that produces pyroline-5-carboxylate with a concomitant generation of ROS [141]. Production of ROS, which damage of bystander cells inducing oxidative damage of DNA, proteins and lipids, apoptosis and inflammation. DNA damage drives genomic instability and promotes transformation of healthy cells, and propagation and dissemination of malignant cells [168]. Arrows indicate: purple arrows—secretion/entering the intercellular space; black arrows—relationships and interactions; red arrows—production of ROS; blue arrows—oxidative stress response. Text above arrows designates the processes leading to the production of ROS, and text below the arrows, forms of ROS.

#### **7. Oncogenic HIV-1 Proteins Inducing Oxidative Stress Are Found in the Extracellular Space** *7.1. Transactivator of Transcription*

Tat protein can be produced and released into the extracellular space by cells harboring actively replicating HIV-1 as well as by latently infected cells, with further uptake by the neighboring uninfected cells. Uptake of Tat would result in upregulation of inflammatory genes and cytotoxicity; this scenario was observed in a number of HIV-1 associated comorbidities, specifically, in neurocognitive disorders, cardiovascular impairment and accelerated aging [169]. Dangerously, the process may occur on the background of successful ART, in the absence of active HIV-1 replication and viral production. Considering that approximately 2/3 of all Tat expressed by infected T cells is secreted [170], the ac-

tivities of Tat described above make a considerable contribution into HIV-1 associated pathologies [171,172].

Soluble Tat, in the absence of the virus, has been shown to cause induction of apoptosis, release of neurotransmitters, oxidative stress and inflammation [169]. Uptake of Tat has been shown to lead to activation of several transcription factors [173,174] including Sp1, NF-κB, and others, resulting in the modulation of expression of both HIV-1 and host genes, including pro-inflammatory cytokines (like TNF-α, CCL2, IL-2, IL-6, and IL-8), adhesion molecules and sometimes, and pro- and anti-apoptotic factors [175–179], p53 and HPV oncoprotein E6 [107].

#### *7.2. Envelope Protein gp120*

Envelope protein gp120 is known to be secreted by chronically infected cells [180,181], particularly from the intraepithelial immune cells even in presence of ART [98]. A subset of PLWH demonstrate persistent circulation in plasma of gp120 [182] and in saliva [98]. Moreover, gp120 was found in tissues of PLWH [183]. Brain cells can be directly exposed to gp120 secreted by infiltrated and infected microglia and astrocytes [127]. Gp120 is internalized by bystander cells through receptor-independent mechanisms [184]. Internalization of gp120 leads to the release of several proinflammatory, angiogenic, and lymphangiogenic factors from affected cells [185].

#### *7.3. Accessory Protein Negative Factor*

Accessory protein negative factor Nef is found in the serum of PLWH [186,187]. Nef can stimulate its own export via the release of extracellular vesicles (exosomes) from HIV-1 infected cells [188]. Of note, exosomes serve as a marker and confirmation of the systemic oxidative stress [189]. Secreted in exosomes, Nef triggers apoptosis in bystander cells. Extracellular Nef has deleterious effects on CD4+ T cells [188,190]; on bystander B cells by suppressing immunoglobulin class switching [191]; and on astrocytes [192] and endothelial cells [162].

#### *7.4. Reverse Transcriptase (RT)*

In our lab, we have shown secretion of RT into cell culture fluids of cells transiently expressing RT [114]. Recently, RT was also detected in the exosomes detected in the uterine of PLWH [193].

#### *7.5. Matrix Protein p17*

Matrix protein p17 is continuously released into the extracellular space from HIV-1 infected cells, and can be detected in the plasma of PLWH and in different organs and tissue specimens [138]. Cellular aspartyl proteases promote the unconventional secretion of biologically active p17 [194]. HIV-1 secretion of biologically active p17 takes place at the plasma membrane and occurs following its interaction with phosphatidylinositol- (4,5)-bisphosphate and its subsequent cleavage from precursor Gag (Pr55Gag) by cellular aspartyl proteases [194]. Extracellularly, p17 deregulates the function of different cells involved in AIDS pathogenesis. Importantly, p17 accumulates and persists in different organs and tissues of PLWH on ART, even in the absence of any replicative activity [136,195,196]. These findings strongly suggest that p17 may be chronically present in HIV-1-I infected cells and tissues, even under ART-associated suppression of HIV-1 replication.

Thus, gp120 and Tat are actively secreted into the endothelial cell micro-environment, Nef can be neighboring uninfected cells including cells which cannot be infected with HIV-1, modulating their metabolism, cell cycle progression, ability to differentiate, motility, and, importantly, the genomic stability, through induction of ROS. Some HIV-1 proteins such as matrix p17 and gp120 can accumulate and persist in lymphoid tissues for at least 1 year after the on-start of ART on the background of successful suppression of viral replication [196]. These proteins are involved in different processes associated with malignant transformation and tumor growth with significant direct and indirect adverse effects on the epithelial cells. These include a range of responses that contribute to endothelial dysfunction, including enhanced adhesiveness, permeability, cell proliferation, apoptosis, as well as activation of cytokine secretion [86], eventually leading to malignant transformation (Figure 3). In this respect, their effect would resemble oncogenesis mediated by known viral oncoproteins originating from EBV, HTLV-1, KSHV, HCV, HBV, HPV, and identified as causative agents of both AIDS-defining and non-AIDS defining forms of cancer.

#### **8. Conclusions**

People living with human immunodeficiency virus receiving antiretroviral therapy are characterized by high prevalence of different forms of cancer affecting epithelial cells. HIV-1 does not infect epithelial cells, however both HIV virions and proteins were shown to be sequestered into epithelial cells and affect their functions. These proteins have three specific properties:


Sequestered/internalized by innocent bystander cells, these proteins modulate their metabolism, cell cycle progression, ability to differentiate, motility, redox balance (induce ROS) and genomic stability. Through this, they can trigger malignant transformation of normal cells. Another outcome is propagation (proliferation and dissemination) of already existing precancerous and cancer cells, and enhanced growth and metastatic activity of tumors expressing or exposed to HIV-1 proteins.

Altogether, we present a new mechanism of HIV-associated malignant transformation of epithelial cells driven by individual HIV proteins through the induction of reactive oxygen species. In this scenario, HIV-1 proteins act in a manner similar to the known viral oncogenes, and can cooperate with them promoting KSHV, EBV, HBV, HCV, and HPVassociated carcinogenesis. Such pathway of HIV associated carcinogenesis can co-occur together with carcinogenesis driven by persistent immune inflammation, and dysfunction of B cells, T cells and cellular components of the innate immune system.

**Funding:** This study was supported the Russian Fund for Basic Research grants 17\_54\_30002 and 20-04-01034 to M.I., Latvian Science Council grants LZP-2018/2-0308 and LZP-2020/2-0376 to M.I., and NCI R01CA 217715 to J.P. The work of Francesca Chiodi is supported by a grant from the Swedish Medical Research Council (Francesca Chiodi; Vetenskapsrådet 2019-01169).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **The Complement System in Ovarian Cancer: An Underexplored Old Path**

**Yaiza Senent 1,2,3, Daniel Ajona 1,2,3,4, \*, Antonio González-Martín 1,5 , Ruben Pio 1,2,3,4,† and Beatriz Tavira 1,3,6,†**


**Simple Summary:** Ovarian cancer is one of the leading causes of death among women and the most lethal cause of death from gynecological malignancy in developed countries. The immune system plays an essential role in ovarian cancer progression, and its modulation may be used as an effective therapeutic tool. In this review, we examine the relevance of the cellular and humoral components of the adaptive and innate immune responses in ovarian cancer, focusing on the role of an essential component of innate immunity, the complement system. Elements of this system show tumor-promoting activities that impede the efficacy of developing treatment strategies. We discuss evidence that suggests a role of complement components in the progression of ovarian cancer and provide a rationale for evaluating the inhibition of complement components in combination with immunotherapies aimed to reactivate antitumor T-cell responses.

**Abstract:** Ovarian cancer is one of the most lethal gynecological cancers. Current therapeutic strategies allow temporary control of the disease, but most patients develop resistance to treatment. Moreover, although successful in a range of solid tumors, immunotherapy has yielded only modest results in ovarian cancer. Emerging evidence underscores the relevance of the components of innate and adaptive immunity in ovarian cancer progression and response to treatment. Particularly, over the last decade, the complement system, a pillar of innate immunity, has emerged as a major regulator of the tumor microenvironment in cancer immunity. Tumor-associated complement activation may support chronic inflammation, promote an immunosuppressive microenvironment, induce angiogenesis, and activate cancer-related signaling pathways. Recent insights suggest an important role of complement effectors, such as C1q or anaphylatoxins C3a and C5a, and their receptors C3aR and C5aR1 in ovarian cancer progression. Nevertheless, the implication of these factors in different clinical contexts is still poorly understood. Detailed knowledge of the interplay between ovarian cancer cells and complement is required to develop new immunotherapy combinations and biomarkers. In this context, we discuss the possibility of targeting complement to overcome some of the hurdles encountered in the treatment of ovarian cancer.

**Keywords:** ovarian cancer; adaptive immunity; innate immunity; complement system; immunotherapy; cancer immunology; tumor microenvironment

**Citation:** Senent, Y.; Ajona, D.; González-Martín, A.; Pio, R.; Tavira, B. The Complement System in Ovarian Cancer: An Underexplored Old Path. *Cancers* **2021**, *13*, 3806. https://doi.org/10.3390/ cancers13153806

Academic Editors: Ion Cristóbal and Marta Rodríguez

Received: 2 July 2021 Accepted: 26 July 2021 Published: 28 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Current Status of Ovarian Cancer: Clinical Perspective and Needs**

Ovarian cancer is the most lethal gynecological cancer in developed countries [1]. According to data from the US National Cancer Institute (NIH), the five-year survival rate for ovarian cancer is 49.1% [2]. This can be attributed to a delay in the diagnosis due to the lack of specific symptoms; 70% of cases are diagnosed in stage III or IV, making it difficult to treat with curative intent [3]. Ovarian cancer is a complex disease that comprises different tumor types, of which epithelial ovarian cancer represents 90–95% of all cases [4]. The current standard treatment includes surgery and platinum-based chemotherapy followed by a maintenance period with the anti-angiogenic therapy bevacizumab [5]. Initial responses to chemotherapy are frequently high, but unfortunately, up to 70% of patients experience recurrence within the first three years, especially patients who are late-diagnosed [5]. Survival rates have recently improved with the introduction of a new generation of poly (ADP-ribose) polymerase inhibitors (PARP inhibitors (PARPi)). These drugs, administered after chemotherapy, prolong the time during which the disease does not progress, mainly in patients carrying BRCA mutations [6]. Despite this great advance, the overall survival of patients with ovarian cancer is still low. There are a variety of factors associated with chemoresistance and relapse, including interactions between ovarian cancer cells and their surrounding immune microenvironment [7]. Ovarian cancers are considered "immunogenic tumors" in which spontaneous antitumor immune responses have been demonstrated [8,9]. The presence of tumors infiltrating CD8<sup>+</sup> lymphocytes in the tumor microenvironment (TME) is associated with longer recurrence-free and overall survival [10,11], whereas the recruitment of regulatory T (Treg) cells is correlated with a poor outcome [12]. These associations indicate that ovarian cancers could respond to immunotherapy. However, immune checkpoint inhibitors (anti-CTLA-4 or anti-PD-1/PD-L1) have yielded modest clinical results in ovarian cancer patients [13,14]. A better understanding of the interplay between ovarian tumor cells and the immunological players in innate and adaptive immunity is critical for developing strategies to overcome the resistance of ovarian cancers to immunotherapy [15,16].

A major effector of innate immunity is the complement system, which represents one of the first lines of defense that distinguish "self" from "non-self" [17]. This system is composed of more than 50 soluble or membrane-bound effectors, regulators, and receptors, and it plays a relevant role in numerous physiological and pathological processes, including cancer [18]. Some evidence suggests that the modulation of complement activation may be exploited for the development of successful treatments against cancer [19,20]. In this review, we discuss the role played by components of adaptive and innate immunity on the development and progression of ovarian cancer. We mainly focus on the complement system, its role in the TME, and the rationale behind the use of complement modulators for the treatment of ovarian cancer.

#### **2. Cellular and Humoral Immune Components of the Ovarian Tumor Microenvironment**

The continuous feedback between tumor cells and the immune system is now recognized as a distinguished cancer hallmark [21]. Neoplastic transformation is characterized by the acquisition of tumor-associated molecular patterns that can be detected by the immune system. It is believed that upon recognition, innate and adaptive immunity can eliminate the vast majority of incipient cancer cells, avoiding tumor formation. However, the immune system is unable to eliminate all emerging malignant cells. When transforming cells escape from immune-mediated elimination, a dynamic interplay is established between tumor cells and the immune system, resulting in tumor-associated immune responses that may facilitate the development and progression of cancer [22]. In the case of ovarian tumors, a plethora of immune and non-immune cell types and non-cellular elements are found in the TME, not only in primary tumors but also in ascites and metastases [23]. The co-existence of multiple distinct tumor immune microenvironments within a single individual highlights the high plasticity and adaptability of ovarian cancers [24]. Herein, we summarize the

main roles of the cellular and humoral elements of the immune system in ovarian tumor progression.

#### *2.1. Cellular Immune Components*

Tumor cells co-exist with non-immune and immune cells, and this relationship determines the natural history of the tumor and its resistance or response to therapy. The cellular immune components of the ovarian TME include T and B lymphocytes, natural killer (NK) cells, dendritic cells (DCs), polymorphonuclear cells, and macrophages.

T cells are a prominent component of the ovarian TME. Infiltration by CD8<sup>+</sup> T cells is indicative of an ongoing immune response and is associated with a favorable prognosis [25]. Upon activation, tumor-specific CD8<sup>+</sup> T cells secrete IFN-γ, tumor necrosis factor (TNF) α, and cytotoxic mediators. However, in the ovarian TME, CD8<sup>+</sup> T-cell responses are often dysfunctional. The autologous recognition of ovarian tumor antigens is limited to approximately 10% of the intratumoral CD8<sup>+</sup> T receptor (TCR) repertoire [26]. This state can be attributed to the upregulation of T-cell exhaustion molecules by persistent antigen exposure and the existence of a hostile TME characterized by nutrient deprivation, hypoxia, oxidative stress, high concentrations of pro-inflammatory molecules, and the presence of immunosuppressive cell subsets [27]. In fact, ovarian cancers are highly enriched in Treg cells [28], a subset of lymphocytes that hamper tumor immunosurveillance by fostering peripheral tolerance to tumor antigens. Treg cells release and metabolize ATP to adenosine by the action of CD39 and CD73, a process that mediates immunosuppression via the adenosine and A2A pathways [29]. Consequently, depletion of Treg cells in ovarian cancerbearing mice effectively restores antitumor antigen-specific T-cell responses [30]. Other lymphoid subsets are important elements of the ovarian cancer immune infiltrate. In an orthotopic syngeneic mouse model, antitumor immunity was driven by CD4<sup>+</sup> T cells [15]. A study identified a novel tumor-infiltrating NK subset characterized by a high expression of PD-1, reduced proliferative capability in response to cytokines, low degranulation, and impaired cytokine production upon interaction with tumor targets [31]. The presence of CD20<sup>+</sup> B cells was associated with increased survival in ovarian cancer patients [32]. In human metastases of high-grade serous ovarian cancer, B cells develop memory responses in the TME and promote antitumor immune responses [33].

DCs are a diverse group of innate immune cells that infiltrate tumors and present tumor-derived antigens to naïve T cells. High densities of tumor-infiltrating DC-LAMP<sup>+</sup> mature DCs suggest the establishment of an antitumor immune response, which is associated with a favorable prognosis in ovarian cancer patients [34]. However, this immune response is often rendered dysfunctional because of a variety of mechanisms, such as the upregulation of B7-H1 [35], the activation of the endoplasmic reticulum stress response factor X-box binding protein 1 (XBP1) [36], the attenuation of the toll-like receptor-mediated DC activation [37], and the activation of the cyclooxygenase 2 (COX2)/prostaglandin E<sup>2</sup> (PGE2) axis to redirect the development of DCs toward the formation of myeloid-derived suppressor cells (MDSCs) [38].

MDSCs represent a heterogeneous population of immature myeloid cells that fail to differentiate into granulocytes, macrophages, or DCs. Two main subsets of MDSCs have been identified: polymorphonuclear MDSC (PMN-MDSC; CD11b+Ly6G+Ly6Clo in mice and CD11b+CD14−CD15+CD66b+LOX-1<sup>+</sup> in humans) and monocytic MDSC (M-MDSC; CD11b+Ly6G−Ly6Chi in mice and CD14+CD15−HLA−DR−/lo in humans). PMN-MDSCs and M-MDSCs are morphologically and phenotypically similar to neutrophils and monocytes, respectively [39]. These cells potently inhibit the anti-tumor immune response and reshape the TME to promote tumor growth and metastatic spread. The differentiation of myeloid precursors toward an MDSC phenotype is mediated by the inflammatory factor PGE<sup>2</sup> via DNA methyltransferase 3A (DNMT3A)-dependent hypermethylation and the downregulation of a subset of myeloid genes [40]. The infiltration of MDSCs into ovarian tumors is associated with the Snail-mediated upregulation of CXCL1 and CXCL2 chemokines that attract MDSCs to the tumor via CXCR2 [41]. In the tumor niche, granulocyte–monocyte colony-stimulating factor (GM-CSF), through the signal transducer and activator of transcription 5 (STAT-5) pathway, upregulates AMP-activated protein kinase alpha 1 (AMPKα-1) in MDSCs to suppress antitumor CD8<sup>+</sup> T-cell responses [42]. Both the presence of TNF-α and the production of NO by MDSCs sustain Th17 responses in the TME and myeloid cell recruitment in an IL-17-dependent manner [43,44].

Tumor-associated neutrophils, a cell population difficult to distinguish from PMN-MDSCs, are also involved in ovarian cancer-associated immune responses. In a KRASdriven ovarian cancer mouse model, neutrophils reduced the amount of tumor-associated Treg cells and M-MDSCs while increasing the antitumor immune response via the upregulation of CD8<sup>+</sup> T-cell function [45]. By contrast, the activation of neutrophils by mitochondrial DNA from ascites obstructs anti-tumor immunity and is associated with worse outcomes in patients with advanced ovarian cancer [46]. This study also reported the formation of neutrophil extracellular traps (NETs), networks of neutrophil decondensed chromatin fibers that are capable of binding tumor cells to support metastatic progression [47]. These contrasting roles of neutrophils in ovarian cancer have been attributed to different polarization states induced by the presence of transforming growth factor (TGF)-β and type-1 interferons in the TME [48].

Tumor-associated macrophages (TAMs) play a major role in the pathogenesis of ovarian cancer [49]. Macrophages constitute over 50% of the cells in peritoneal ovarian tumor nodules and malignant ascites and are involved in ovarian cancer initiation, progression, and metastasis [50]. TAMs are highly plastic cells that can exhibit two main phenotypes: anti-tumorigenic M1-like (F4/80hi and CD86<sup>+</sup> or CD80<sup>+</sup> or iNOS<sup>+</sup> in mice; CD68+HLA-DR+CD11c<sup>−</sup> and CD86<sup>+</sup> or CD80<sup>+</sup> or iNOS<sup>+</sup> in humans) and pro-tumorigenic M2-like (F4/80hi and CD163<sup>+</sup> or CD206<sup>+</sup> or arginase<sup>+</sup> in mice; CD68+HLA-DR+CD11c<sup>−</sup> and CD163<sup>+</sup> or CD206<sup>+</sup> in humans). Analyses of TAM polarization in ovarian cancer show that M2 TAMs are associated with a poor prognosis [51,52]. Malignant cells direct TAM differentiation to facilitate tumor progression. The activation of the ovarian TAM protumor phenotype requires the expression of zinc finger E-box binding homeobox 1 (ZEB1), a driver of the epithelial-mesenchymal transition (EMT), and involves direct crosstalk with tumor cells [53]. Tumor-expressed CD24 interacts with the inhibitory receptor sialicacid-binding Ig-like lectin 10 (Siglec-10) expressed by ovarian cancer-inhibiting TAMs to avoid their antitumor effects [54]. Ovarian cancer cells skew co-cultured macrophages to a phenotype similar to that found in ovarian tumors [55]. Ovarian cancer cells promote membrane-cholesterol efflux and depletion of lipid rafts to polarize TAMs toward a tumor-promoting phenotype characterized by the upregulation of IL-4 signaling [56]. In return, TAMs enhance the malignant potential of ovarian cancer cells. Endothelial growth factor (EGF) secreted from TAMs promoted tumor growth at early stages of transcoelomic metastasis in a mouse model of ovarian cancer [57]. Moreover, TAMs enhance ovarian cancer invasiveness through activation of the nuclear factor kappa B (NF-kB) and Jun N-terminal kinase (JNK) pathways in tumor cells [58].

#### *2.2. Humoral Immune Components*

The crosstalk between the different cellular components of the TME is essential to reprogram tumor-associated immune responses. This process is orchestrated by complex networks interconnected by sets of soluble factors and extracellular structures, such as cytokines, chemokines, small metabolites, and microvesicles, among others [59]. In particular, cytokines mediate key interactions between immune and non-immune cells in the TME [60], and cytokine-based immunotherapy is a promising strategy to modulate the host's immune response toward the induction of apoptosis in tumor cells [61]. To date, there are two FDA-approved treatments for melanoma and metastatic renal cell cancer based on the administration of TNF-α and interleukin (IL)-2 [62]. In the case of ovarian cancer, the proinflammatory cytokine IL-6 has been established as a key immunoregulator [63]. IL-6, along with other cytokines, activates pathways such as STAT and NF-kB, whose modulation could be used as a potential therapeutic tool [63].

Many years ago, Bjørge et al. found elevated levels of complement C1q, C3, C3a, and soluble C5b-9 in ascites from ovarian cancer patients, suggesting that local complement activation may constitute an important soluble component of the ovarian TME [64]. More recently, ovarian cancer has been classified as a cancer type with "upregulated complement" [65]. Interestingly, over the last decade, the complement system has emerged as a major non-cellular regulator of the TME in cancer immunity. Tumor-associated complement activation may support chronic inflammation, promote an immunosuppressive microenvironment, induce angiogenesis, and activate cancer-related signaling pathways [66]. In the case of ovarian cancer, complement dysregulation may even participate in the onset of tumors since complement molecules are already overexpressed in precursor lesions [67]. In the following section, we summarize the evidence supporting the involvement of the complement system in ovarian cancer progression.

#### **3. The Complement System and Its Dual Role in Ovarian Cancer**

In 1896, the complement system was first described as a heat-labile component in the serum able to "complement" heat-stable factors (antibodies). Now, the complement system is broadly known as a central part of the innate immune response composed of soluble and membrane-bound proteins that can coordinate a nonspecific inflammatory response against microbes and unwanted host elements [18]. Complement-circulating effectors are predominantly synthesized in the liver and are distributed throughout the body in an inactivated state. Complement can be activated by three main distinctive pathways: the classical pathway (CP), the lectin pathway (LP), and the alternative pathway (AP) (Figure 1). The three pathways converge in the cleavage of the complement component C3 into C3a and C3b. The CP is initiated in foreign, damaged, or dying cells when the C1 complex, which includes C1q, C1r, and C1s, recognizes antibody clusters, pathogenassociated molecular patterns (PAMPs), or danger-associated molecular patterns (DAMPs), among other molecules [68]. The LP is initiated by the recognition of carbohydrate patterns by mannose-binding lectin (MBL) or ficolins, along with the mannan-binding lectin serine proteases MASP1 and MASP2 [68]. The initiation of both the CP and the LP leads to the cleavage of C4 into C4a and C4b and, subsequently, C2 into C2a and C2b. The complex formed by C4b and C2b (C4bC2b, formerly C4b2a) constitutes the classical C3 convertase, which is responsible for the cleavage of C3 into C3a and C3b [68]. The AP is initiated by the spontaneous hydrolysis of C3 into C3(H2O), followed by its binding to factor B. This complex is recognized by factor D, which catalyzes the cleavage of factor B to form the fluidphase alternative C3 convertase C3(H2O)Bb. This convertase can mediate the cleavage of C3 into C3a and C3b to form the membrane-bound alternative C3 convertase C3bBb [69]. Subsequently, C3b is able to bind to C4bC2b (in the CP and LP) or C3bBb (in the AP), leading to the formation of C5 convertase. This complex catalyzes the cleavage of C5 into C5a and C5b. The later fragment sequentially binds to C6, C7, C8, and C9 to form the cytolytic membrane attack complex (MAC) [18,68]. Many complement functions are mediated by the anaphylatoxins C3a and C5a, which act as potent inflammatory modulators [70]. These peptides signal through their respective G-protein-coupled receptors C3aR and C5aR1 [71]. A second, lesser-known C5a receptor, C5aR2, also participates in C5a responses, though its role remains unclear. Finally, an array of membrane and soluble complement regulatory proteins (CRPs) protects normal cells from the overactivation of complement [68] (Figure 1).

**Figure 1.** Schematic representation of the effectors and regulators of the complement cascade. Complement is initiated by three distinctive pathways: the classical (blue arrows), the lectin (green arrows), and the alternative (brown arrows) pathways. All three pathways converge in the formation of C3 and C5 convertases, which in turn generate the inflammation modulators C3a and C5a. The terminal steps, which culminate in the assembly of the membrane attack complex (MAC), are common to the three pathways. Inhibitory proteins of the three pathways are shown in red boxes.

Complement plays an essential role in the control of cellular immunity [18], participating in the regulation, differentiation, and trafficking of several immune cell types [17,72]. C3 and C4 depletion impair humoral immune responses in vivo [73–75]. It has been postulated that antigen–antibody clusters interact with complement and are recognized by DCs, B lymphocytes, and macrophages [76]. Further evidence of the interplay between adaptive and innate immunity is the CD21(CR2)-CD19-CD81 complex on B cells, which enhances B-cell receptor function [77,78], partially by recognizing C3d-tagged surfaces [79]. Anaphylatoxins also play an important role in immune regulation. Most immune cell types express C3aR and/or C5aR1 on their surfaces [65]. On B cells, C3a impairs polyclonal immune responses and TNF-α and IL-6 production [80,81]. C5a has been extensively reported to induce the migration of several cell types [82–86]. Interestingly, C5a fosters antigen cross-presentation and the maturation of DCs [87–89]. Moreover, C3a-C3aR and C5a-C5aR1

signaling promote the activation [90] and expansion [91] of T cells and divert their differentiation from Treg cells [92,93]. Complement inhibitory proteins, such as CD46, have been shown to modulate T-cell fate depending on the isoform expressed and the presence of IL-2 [94,95]. Moreover, negative modulation of the inhibitor CD59 was demonstrated to ameliorate antigen-specific T-cell responses [96]. Overall, the information gathered during the past few decades illustrates the interconnections between the complement system and adaptive and innate immunity and endorses the hypothesis that complement's role extends beyond its traditional non-specific, first-defense function.

Dysregulation of complement can lead to the development of several pathologies. Kidney diseases, such as atypical hemolytic uremic syndrome (aHUS) and C3 glomerulopathies, are closely related to complement anomalies. C3 glomerulopathies are characterized by the production of C3 fragments in the fluid phase via the alternative pathway and abnormal complement consumption that leads to the damage of the glomerular basement membrane [97]. Activation of the complement system is also involved in the pathogenesis of systemic autoimmune diseases [98]. Alterations in regulatory proteins can trigger serious conditions as well. Paroxysmal nocturnal hemoglobinuria (PNH) is a hematological disorder caused by a deficiency in glycosylphosphatidylinositol anchor synthesis that negatively affects the expression of the CRPs CD55 and CD59 [99]. More recently, cancer progression has been associated with complement activation [66].

In the next sections, we review studies that have reported the participation of components of the complement system in the biology of ovarian cancer or its potential clinical use. The findings of these studies are summarized in Tables 1 and 2.




**Table 1.** *Cont.*

EMT: epithelial-mesenchymal transition, SCID: severe combined immunodeficient, mAb: monoclonal antibody, CDC: complementdependent cytotoxicity.

**Table 2.** Summary of the studies performed with clinical samples reporting the potential clinical use of the determination of complement components.



**Table 2.** *Cont.*

IHC: immunohistochemistry, mAb: monoclonal antibody, CDC: complement-dependent cytotoxicity.

#### *3.1. Complement Initiation Components in Ovarian Cancer*

C1q, the first component of the classical complement activation pathway, links innate and adaptive immunity [123]. Both promoting and inhibitory roles have been reported for C1q in cancer progression, but most studies associate C1q expression with poor clinical outcomes in cancer, as is the case for gliomas and osteosarcomas [124,125]. C1q may act as a tumor-promoting factor through both complement-dependent and complementindependent mechanisms [126,127]. In ovarian cancer, the role of C1q appears to be context-dependent. In vitro, C1q displays an anti-tumor effect in SKOV3 cells by promoting apoptosis through the upregulation of the TNF-α pathway and the downregulation of the mammalian target of rapamycin (mTOR) survival pathway [100]. Conversely, expression levels of C1q in circulating extracellular vesicles isolated from ovarian cancer patients in stages III–IV are significantly elevated compared with those isolated from healthy individuals [115]. Discrepancies have also been observed in the case of the globular C1q receptor (gC1qR), a cell surface receptor for C1q. This molecule is upregulated in tumor cells [128], and its overexpression induces mitochondrial dysfunction and p53-dependent apoptosis in human cervical squamous carcinoma cells in vitro [101]. Consistently, the induction of gC1qR expression by paclitaxel in ovarian cancer cell lines SKOV3 and CAOV3 results in mitochondrial dysfunction and cell apoptosis [102]. However, this consistency observed in vitro disappears when clinical samples from ovarian cancer patients at different stages of the disease are analyzed. gC1qR downregulation was observed in ovarian cancer patients in the early stages of the disease (stages I–II) [102]. By contrast, gC1qR seems to be overexpressed in tumor tissue from ovarian cancer patients in stages III and IV, and this is associated with a poor prognosis and cisplatin resistance [116]. These data suggest an increase in complement activation during ovarian cancer progression. Consistent with this assumption, C4 was detected in ascitic fluid from late-stage patients, while it was undetectable in ascitic fluid from healthy donors [64]. Moreover, C4 levels were found to be upregulated in plasma samples from chemoresistant compared with chemosensitive ovarian cancer patients [119]. In the same study, complement factor I and C3 were found to be downregulated [119]. Finally, MBL and MASP2 serum levels are altered in ovarian cancer patients, and MBL levels are associated with advanced disease stages [117]. The ovarian tumor antigen cancer antigen 125 (CA-125), a highly glycosylated protein, may be a target for pattern recognition molecules, such as collectins and ficolins, which may mediate the interaction with MBL and the activation of the lectin pathway [129]. Serum ficolins have been reported to be elevated in ovarian cancer patients despite their lower tumor expression [118]. In conclusion, several studies have reported the presence of complement initiation factors in ovarian cancer. However, the contribution of these factors

to ovarian cancer progression and response to treatment is still unclear and requires further investigation.

#### *3.2. C3 and C5 in Ovarian Cancer*

The C3- and C5-derived fragments C3a and C5a participate in the establishment of a chronic inflammatory state that may favor tumorigenesis and cancer progression [70]. In ovarian cancer, the implication of C3a and C5a seems to depend on multiple factors, although most of the evidence suggests a tumor-promoting effect. Nuñez-Cruz et al. assessed the role of complement in ovarian tumor progression using C3 and C5aR1-deficient mice. Complement inhibition impaired both tumor vascularization and growth [103]. Some molecular mechanisms have been associated with the tumor-promoting function of C3 and C5 in ovarian cancer tumor cells. These mechanisms include the activation of the phosphatidylinositol-3-kinase (PI3K) pathway and the induction of EMT [104,105]. C3 and C5 and their effector fragments also influence tumor progression by acting on immune cells. Circulating polymorphonuclear cells from ovarian cancer patients can acquire an immunosuppressive phenotype capable of restraining T-cell proliferation after exposure to ascites in a process dependent on C3 [130]. This T-cell non-responsiveness is associated with the production of C5a and is mediated by mTOR signaling and nuclear factor of activated T-cells (NFAT) translocation [131]. Interestingly, C5a may function in a dosedependent manner. Thus, in a SKOV-3 tumor model, low local doses of C5a reduced tumor growth in association with the recruitment of M1 TAMs and NK cells, while high doses promoted tumor progression [107]. Ovarian cancer cells overexpress ribosomal protein S19 (RPS19), which leads to tumor growth through its interaction with C5aR1 in MDSCs [132]. By contrast, the local production of C3 and the release of C5a disrupt the tumor endothelial barrier, facilitating the homing of T cells and their tumor recruitment [106]. This study further stresses the contrasting effects associated with complement effectors in different models of ovarian cancer. Unfortunately, the results reported in patients do not clarify the matter. High levels of C3 or C5aR1 have been associated with decreased overall survival [104,133]. By contrast, reduced expression of C3 was observed in the blood of ovarian cancer patients [134], and this factor was downregulated in the serum of platinum-resistant patients [119].

#### *3.3. Complement Regulatory Proteins in Ovarian Cancer*

CRPs protect host cells from autologous complement attack, but they can render complement ineffective at eliminating cancer cells. Membrane-bound CRPs (mCRPs), such as CD46, CD55, and CD59, are expressed by ovarian cancer tumors [121,135] and cell lines [108,122,135]. These regulators are linked to worse clinical outcomes and may constitute an obstacle for cancer immunotherapy [121,136–138]. Their presence has also been associated with the development of multi-drug resistance in ovarian cancer cells [139]. Neutralization of mCRPs increases the sensitivity to complement-dependent cytotoxicity [111,113,139], reduces ovarian tumor growth [110], and enhances the anti-tumor efficacy of therapeutic antibodies [108,112]. In line with these findings, CD55 silencing restores cisplatin sensitivity to chemotherapy in resistant ovarian cancer cells [109]. Regarding soluble complement regulators, a range of studies has demonstrated their importance in several tumor types [140–143]. In ovarian cancer, some soluble complement inhibitors, such as factor H and factor H-like 1 (FHL-1), have been found in ascitic fluid and primary tumors [64,114]. However, the role of these regulators in ovarian cancer progression has not been defined yet.

In conclusion, the evidence suggests that complement dysregulation drives ovarian cancer progression. Complement effectors, receptors, and regulators have been implicated in different aspects of ovarian cancer biology (Figure 2). Although there are inconsistencies in the description of the role of complement components in some clinical or experimental contexts, the majority of studies point toward a tumor-promoting activity of complement in

well-established tumors. These findings have paved the way for studies aimed to potentiate cancer therapies through the modulation of the complement system.

**Figure 2.** Complement-related mechanisms associated with ovarian cancer progression. Complement components have been implicated in different biological processes associated with ovarian cancer progression. They include modulation of immunosuppression in the tumor microenvironment; regulation of angiogenesis and endothelial permeabilization; autocrine and paracrine effects in tumor cells mediated by C1q, C3a, or C5a; and inhibition of complement-dependent cytotoxicity (CDC) by membrane-bound complement regulators.

#### **4. Therapeutic Potential of Targeting Complement in Ovarian Cancer**

Complement inhibition may be a useful therapeutic strategy against cancer [19]. Agonists of C5aR1 and C3aR increase ovarian tumor cell proliferation, migration, and invasion, suggesting that receptor antagonists could be used to block cancer growth [104]. Complement targeting may also impair angiogenesis, a highly relevant biological process in ovarian cancer. Elevated levels of serum VEGF after chemotherapy treatment have been associated with lower overall survival in ovarian cancer patients [144], and the anti-VEGF antibody bevacizumab has shown therapeutic activity in both patients and animal models [145–147]. Genetic or pharmacological inhibition of C3 or C5aR1 results in smaller and poorly vascularized ovarian tumors in vivo [103], and C5a is able to promote endothelial cell tube formation and migration [103,148]. Therefore, it can be speculated that inhibition of complement may potentiate the efficacy of anti-angiogenic agents.

Another scenario in which complement modulation may be of special relevance is immunotherapy. We previously described the implication of effectors and regulators of the complement system in the ability of T cells to infiltrate tumors and the response against tumor-associated antigens [149]. Using various models of lung cancer, we proposed that the modulation of complement activation can improve the antitumor efficacy of monoclonal antibodies targeting the PD-1/PD-L1 pathway [150]. This synergistic effect has also been reported in other tumor models targeting C5a/C5aR1 [151,152] or C3a/C3aR [151,153]. To our knowledge, these combinations have not been tested yet in models of ovarian cancer, and we can only hypothesize about the outcome of these studies. The inhibition of C3 or C5aR1 abrogates the suppressor phenotype of MDSCs in the ovarian TME [130,131], suggesting that complement inhibition may have a positive effect on the efficacy of anti-PD-1/PD-L1 therapies. Conversely, antitumor T cells require the production of C3 and the release of C5a in the endothelium in order to infiltrate ovarian tumors [106]. The targeting of mCRPs should also be considered in light of their relevance in the TME [136]. The inhibition of mCRPs may be used to sensitize tumors to other drugs. In ovarian cancer, the neutralization of CD46, CD55, and CD59 in combination with the anti-HER2 monoclonal antibodies trastuzumab and pertuzumab induces tumor cell killing in vitro [113]. Nevertheless, considering the dual role of complement molecules in ovarian tumors, in vivo studies are needed to determine whether complement inhibition has any impact on the response to checkpoint-based or antibody-based immunotherapies, and in which direction.

#### **5. The Need for Preclinical Models to Better Delineate the Role of Complement in Ovarian Cancer**

In this review, we discussed the functions that complement components exert in the biology of ovarian tumors. Many questions remain regarding the conflicting results observed in different experimental settings. To address these questions, in vivo models that faithfully recapitulate the complexity of the disease are needed. Currently, there are a few animal models established for the study of ovarian cancer. These include genetically engineered mouse models, xenograft cell transplants of human cell lines, and patient-derived xenografts [154]. These models have facilitated the study of many mechanisms associated with ovarian cancer progression and have allowed the evaluation of many therapeutic molecules [155]. For the study of complement-related mechanisms or treatments, mouse models that capture the complexity of the TME are required. Models based on syngeneic tumor cells injected intraperitoneally in immunocompetent mice represent a practical option. Some studies have used the syngeneic intraperitoneal injection of ID-8-MOSEC, a mouse epithelial ovarian cancer cell line originating in C57BL/6 mice, to evaluate the roles of C3, C5, and C5aR1 in ovarian cancer development and progression (Table 1) [104,106]. This cell line was developed by Dr. Katherine F. Roby in the Department of Anatomy and Cell Biology of the University of Kansas in the early 2000s, and it is one of the most frequently used ovarian cancer cell lines since it has the capacity to induce tumor peritoneal implants observed in stages III and IV [156]. Because of its slow growth rate, some strategies have been developed to increase the aggressiveness of this cell line, including the overexpression of dendritic cell chemoattractant beta-defensin 29 (Defb29) or VEGF [157], two factors associated with increased invasiveness. Nevertheless, this model does not completely recapitulate the human pathophysiology of the disease and does not exactly reproduce the TME [154]. The development of better ovarian cancer models is needed to unravel the mechanisms by which complement components modulate ovarian cancer progression and to evaluate complement-based therapeutic combinations.

#### **6. Conclusions**

A growing body of literature suggests that the complement system is involved in ovarian cancer progression. Nevertheless, the specific role of the different complement components in different clinical scenarios has just started to be unraveled, and many answers remain elusive. The molecular heterogeneity of ovarian cancers and the complexity of the biological interactions in the ovarian TME pose a challenge to our understanding of the mechanisms underlying the complement-associated immune responses and the identification of adequate therapeutic targets. The situation is aggravated by the lack of preclinical models that reliably recreate ovarian cancer traits. Therefore, further studies are needed to better delineate the complement-related mechanisms associated with ovarian cancer progression as well as to determine how complement activation should be modulated to treat ovarian cancer patients.

**Author Contributions:** Y.S., D.A., R.P., and B.T. conceived and designed the manuscript. All authors (Y.S., D.A., A.G.-M., R.P., and B.T.) wrote the manuscript. Y.S. prepared the figures. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors' work was supported by the Foundation for Applied Medical Research (FIMA), Centro de Investigación Biomédica en Red Cáncer (CIBERONC), Grupo Español de Investigación en Cáncer de Ovario (GEICO), Fundación Ramón Areces, and Instituto de Salud Carlos III-EU FEDER "Una manera de hacer Europa" (PI20/00419 and PI20/00959). YS was supported by a predoctoral fellowship from Ministerio de Ciencia, Innovación y Universidades (FPU18/02638).

**Conflicts of Interest:** D.A. and R.P. are authors of patents and patent applications claiming the determination of complement fragments for lung cancer diagnosis. No potential competing interests are disclosed by the other authors.

#### **References**


### *Review* **Role of Oncogenic Pathways on the Cancer Immunosuppressive Microenvironment and Its Clinical Implications in Hepatocellular Carcinoma**

**Naoshi Nishida**

Department of Gastroenterology and Hepatology, Kindai University Faculty of Medicine, 377-2 Ohno-Higashi, Osaka-Sayama 589-8511, Japan; naoshi@med.kindai.ac.jp; Tel.: +81-72-366-0221

**Simple Summary:** Hepatocellular carcinoma is known to become resistant to treatments easily by mutations in genes involved in the key cellular pathways targeted by current molecular targeted agents (MTAs). However, the immune checkpoint inhibitor (ICI) is a promising modality for cancer treatment, in which the cancer cells are made recognizable by the immune system. Blockade of the PD1/PDL1 proteins, which help cancers evade the immune system, is currently being tested in clinical trials in combination with MTAs. In this review, several cellular signaling pathways that can alter the immune processes within the tumor and can subsequently affect the patient's response to ICIs are detailed. This review may help scientists and clinicians to better understand the molecular factors that can influence ICI-based therapy and will help in identifying suitable cases for this type of treatment.

**Abstract:** The tumor immune microenvironment, including hepatocellular carcinoma (HCC), is complex, consisting of crosstalk among tumor components such as the cancer cells, stromal cells and immune cells. It is conceivable that phenotypic changes in cancer cells by genetic and epigenetic alterations affect the cancer–stroma interaction and anti-cancer immunity through the expression of immune checkpoint molecules, growth factors, cytokines, chemokines and metabolites that may act on the immune system in tumors. Therefore, predicting the outcome of ICI therapy requires a thorough understanding of the oncogenic signaling pathways in cancer and how they affect tumor immune evasion. In this review, we have detailed how oncogenic signaling pathways can play a role in altering the condition of the cellular components of the tumor immune microenvironment such as tumor-associated macrophages, regulatory T cells and myeloid-derived suppressor cells. The RAS/MAPK, PI3K/Akt, Wnt/β-catenin and JAK/STAT pathways have all been implicated in anti-tumor immunity. We also found that factors that reflect the immune microenvironment of the tumor, including the status of oncogenic pathways such as the volume of tumor-infiltrating T cells, expression of the immune checkpoint protein PD-1 and its ligand PD-L1, and activation of the Wnt/β-catenin signaling pathway, predict a response to ICI therapy in HCC cases.

**Keywords:** cancer; hepatocellular carcinoma; immune evasion; immunotherapy; immune checkpoint inhibitors; oncogenic signaling pathway; molecular targeted agents; genome; epigenome; tumor immune microenvironment

#### **1. Introduction**

Hepatocellular carcinoma (HCC) is highly refractory and is the third leading cause of cancer-related deaths worldwide [1]. Recent advancements in molecular targeted agents (MTAs) for HCC have dramatically improved the prognosis for patients with this disease. Following the approval of sorafenib as the first MTA for advanced HCC, lenvatinib has also been applied as a first-line systemic chemotherapeutic for HCC, while regorafenib, cabozantinib and ramucirumab have been approved as second-line agents [2]. Because MTAs primarily target molecules involved in oncogenic signaling pathways that play an

**Citation:** Nishida, N. Role of Oncogenic Pathways on the Cancer Immunosuppressive Microenvironment and Its Clinical Implications in Hepatocellular Carcinoma. *Cancers* **2021**, *13*, 3666. https://doi.org/10.3390/ cancers13153666

Academic Editors: Ion Cristóbal and Marta Rodríguez

Received: 8 June 2021 Accepted: 19 July 2021 Published: 21 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

important role in the development of cancer cells, the development of clones resistant to MTAs can happen easily by genetic mutations and modifications in the specific molecular pathways [3,4]. Hence, additional chemotherapeutic agents would be required.

In contrast, immune checkpoint inhibitors (ICIs) play a role in tumor regression by a different mechanism from that of MTAs [5]. They are known to interfere with the immunosuppressive mechanism to enhance the anti-tumor immune response [6]. Because the target molecules of ICIs are primarily expressed in the stromal cells as well as the cancer cells, ICIs can be effective even for patients who fail to respond well to MTAs or acquire resistance to them, potentially enabling ICIs to complement treatment with MTAs [5,7–9]. Although the clinical trial of anti-programmed cell death-1 (PD-1) monotherapy failed to show a significant difference in the survival of patients with advanced HCC compared with conventional MTAs, synergic effects of the combination of different kinds of agents can be expected in several ongoing clinical trials of ICI-based therapy. Based on a successful Phase III clinical trial, the combination of the ICI atezolizumab (an anti-PD-1 antibody) with MTA bevacizumab (an anti-VEGF-A antibody) was approved as a first-line therapy for unresectable HCC [10,11].

Because of the complexity of cancer immunity, where immune cells, tumor cells and other types of stromal cells affect each other, understanding the immune microenvironment of the tumor is difficult [5]. While it has been considered that oncogenic mutations in tumor cells do not directly affect the outcome of ICI therapy, recent reports have suggested that mutation-induced changes in the tumor phenotype can affect the tumor–stroma interactions through alterations in the expression of immunosuppressive cytokines, chemokines, receptors and metabolites, thereby potentially affecting the tumor immune microenvironment [5]. Thus, the anti-tumor effect of MTAs in combination therapy with ICIs can be attributed to the direct action of MTAs on the HCC cells, as well as the reduction in the immunosuppressive nature of the tumor microenvironment through the inhibition of specific oncogenic signals [12].

To understand the significance of oncogenic signaling in the establishment of an immunosuppressive tumor microenvironment, and for the application of this knowledge to the treatment of HCC, this review focused on the role of specific genetic mutations involved in the oncogenic pathways responsible for anti-tumor immunity, and the current status of and perspectives on the combination of ICIs and MTAs for the treatment of HCC.

#### **2. Cellular Components and Molecules Associated with an Inhibitory Tumor Immune Microenvironment**

Oncogenic signals affect the expression of several immune-related molecules, including immune regulatory receptors, ligands, growth factors and other humoral factors, which affect diverse stromal cells as well as cancer cells. The cellular components of tumors and their states are major players in the regulation of the tumor immune microenvironment. Therefore, to better understand the impact of oncogenic signals on anti-cancer immunity, the functions of the stromal cells involved in the immune microenvironment of tumors are briefly discussed here.

#### *2.1. Regulatory T-Cells*

Regulatory T-cells (Tregs) are CD4<sup>+</sup> T-cells characterized by the expression of the transcription factor Foxp3. They can be induced in tumor tissues through growth factors and cytokines, such as transforming growth factor β (TGF-β), interleukin 10 (IL-10) and vascular endothelial growth factor (VEGF), and inhibit immune responses through various mechanisms [13]. In particular, Tregs express the inhibitory immune checkpoint molecule cytotoxic T-lymphocyte (associated) antigen 4 (CTLA-4), which plays a critical role in the regulation of T cell-mediated anti-tumor immunity. Generally, T cell activation occurs through binding of the co-stimulatory factor B7 (CD80/CD86) on antigen-presenting cells and CD28 on T-cells, in addition to T-cell receptor (TCR) recognition of major histocompatibility complex (MHC)-presented antigens. Binding of CD80/CD86 on dendritic cells (DCs) with CTLA-4 on Tregs results in the inhibition of DC maturation. In addition, the

membrane molecule CD25 (IL-2 receptor subunit) on Tregs induces the depletion of IL-2 and suppression of cytotoxic T-cells (CTLs) by immunosuppressive cytokines such as TGF-β and IL-10, and cytotoxic secretions such as granzyme B and perforin released by Tregs [13]. A subtype of HCC that showed predominant expression of an mRNA related to Treg response has been reported [14]. Tregs also secrete the epidermal growth factor receptor (EGFR) ligand amphiregulin, which can promote the growth of HCC cells carrying EGFR in an autocrine manner [15]. Tregs also express VEGF receptor 2 on their surface, and the VEGF signal induces the expansion of this type of T cell [16].

#### *2.2. Myeloid-Derived Suppresor Cells*

Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid cells that suppress tumor immunity and can be induced by VEGF [16]. Via their increased arginase activity, degradation of arginine, and uptake of tryptophan, cysteine and other amino acids required for T-cell activation, MDSCs reduce the concentrations of these amino acids in the tissue microenvironment, thereby inhibiting the propagation and activation of T cells [17]. In addition, MDSCs produce TGF-β and IL-10, inducing Tregs and inhibiting natural killer (NK) cell function [18]. Furthermore, MDSCs induce the immunosuppressive M2 macrophages by secreting IL-10, which, in turn, downregulates IL-12 production by tumor-associated macrophages (TAMs) [19,20].

#### *2.3. Tumor-Associated Macrophages*

Generally, two types of macrophages exist in tumor tissues: M1 macrophages and M2 macrophages. Interferon-γ (IFN-γ) and Type 1 helper cell (Th1) cytokines induce the differentiation of inflammatory monocytes into M1 macrophages. Meanwhile, Type 2 helper cell (Th2) cytokines such as IL-4 and IL-13 promote the differentiation of tissue-resident monocytes into M2 macrophages [20]. In tumor immunity, M1 macrophages produce inflammatory cytokines such as tumor necrosis factor α (TNF-α), IL-6 and IL-12, and exert an anti-tumor effect, whereas M2 macrophages produce immunosuppressive cytokines such as IL-10 and TGF-β, and inhibit anti-tumor immune reactions [20]. The microenvironment in cancer is prone to inducing M2 polarization, which is a characteristic phenotype called tumor-associated macrophages (TAMs). The crosstalk between MDSCs and TAMs induces high IL-10 and low IL-12 levels. In addition, naïve CD4<sup>+</sup> T cells differentiate into Th2 cells that can produce IL-4 [21]. These processes result in the development of M2 macrophages, which is a disadvantageous state for tumor immunity. A high level of IL-10 induces the downregulation of human leukocyte antigen (HLA) Class II antigens and reduces the antigen presentation capacity of DCs [21]. It also expands the Treg population and inactivates natural killer (NK) cells. Additionally, the TGF-β secreted by MDSCs induces the expression of the inhibitory receptor T-cell immunoglobulin and mucin domain 3 (TIM-3) on TAMs [20].

#### *2.4. Cancer-Associated Fibroblasts and Vascular Endothelial Cells*

Cancer-associated fibroblasts (CAFs) have proangiogenic activity through the production of extracellular matrix and matrix metalloproteinases; they play a role in tissue remodeling [22]. They also inhibit NK cell function through the production of prostaglandin E2 (PGE2) and indoleamine-2,3-dioxygenase (IDO) [23]. Indoleamine-2,3-dioxygenase is an enzyme involved in tryptophan metabolism, and a reduced level of tryptophan in tumors inhibits local T-cell activation. Hence, crosstalk between CAFs and TAMs also plays a role in immunosuppression. CAFs produce IL-8 and cyclooxygenase-2 (COX2), which lead to the release of TNF and platelet-derived growth factor (PDGF) from TAMs, and further activation of CAFs.

#### *2.5. Other Stromal Cells*

The vascular endothelium is stimulated by angiogenic growth factors such as VEGF and PDGF. It stimulates Tregs and MDSCs in tumor tissues via the production of TGF-β, VEGF and the chemokine C-X-C motif chemokine 12 (CXCL12) [20]. Hepatic satellite cells (HSCs), which generally play a critical role in liver fibrogenesis, also participate in the induction of Tregs and MDSCs by releasing hepatocyte growth factors [24,25]. HSCs also produce amphiregulin and CXCL12, which induce Tregs and MDSCs, respectively [26]. A subset of DCs with high expression of CTLA-4 was also observed in HCC tissues, which may carry immune tolerogenic effects through the production of IL-10 and IDO.

#### *2.6. Immunosuppressive Metabolites*

As shown above, the concentrations of metabolites from cancer cells and stromal cells strongly affect the immune state of the tumor. Cyclic adenosine monophosphate (cAMP), which accumulates in tumor tissues, inhibits CD4<sup>+</sup> and CD8<sup>+</sup> T-cell responses and macrophage activation, and enhances the Treg response by binding to adenosine A2A receptors [27]. In addition, due to the hypoxic environment in tumor tissues, cAMP upregulates the enzyme COX-2, which synthesizes PGE2 from arachidonic acid. Subsequently, PGE2 binds to prostaglandin E receptor 4 on T-cells and affects T-cell activation and cytokine production [20].

MDSC- and TAM-derived arginase hydrolyzes arginine in the urea cycle and inhibits the function of CTLs via this deficiency in L-arginine. In tumor tissues, a hypoxic environment results in the expression of hypoxia-inducible factor 1α (HIF-1α), which is known to activate arginase [17]. Additionally, IDO is reported to be produced by DCs, macrophages, CAFs, vascular endothelial cells and HCC cells via inflammatory cytokines [28]. As previously stated, IDO inhibits T-cell activation and amplification via depletion of tryptophan and stimulates the differentiation of naïve CD4<sup>+</sup> T-cells into Tregs [20,29].

#### *2.7. Immune Checkpoint Molecules*

Immune checkpoint molecules regulate excessive T-cell activation and help to maintain immune homeostasis. In cancer cells, however, these immune checkpoint molecules help tumors evade the immune response. Many immune checkpoint molecules and their ligands have been identified, as summarized in Figure 1. Of these, CTLA-4, programmed cell death-1 (PD-1) and its ligand, programmed cell death-ligand 1 (PD-L1), are believed to play central roles in tumors' immune evasion [5,20]. The induction of immune checkpoint molecules is regulated by environmental factors as well as cell signaling. For example, extracellular stimulation of IFN-γ and hypoxia-induced HIF-1 can enhance the expression of PD-L1 in cancer cells, MDSCs and TAMs [30]. The binding of PD-L1 to PD-1 on TAMs induces the release of the immunosuppressive cytokine IL-10. Additionally, activation of the phosphatidylinositol-3 kinase (PI3K)–Akt pathway also reportedly induces PD-L1, and the loss-of-function mutation of phosphatase and tensin homolog deleted from chromosome 10 (PTEN), a regulator of the PI3K–Akt pathway, is associated with the expression of PD-L1 in cancer [31,32]. We have also shown that activating mutations in PI3KCA are associated with PD-L1 expression in HCC cells [33]. TIM-3, lymphocyteactivation gene 3 (LAG-3) and the B- and T-lymphocyte attenuator (BTLA, CD272) are also known as co-inhibitory molecules on activated T-cells, based on their association with galectin-9, MHC Class II and herpesvirus entry mediator (HVME), respectively (Figure 1). These suppressive receptors are observed in tumor-infiltrating lymphocytes (TILs) in HCC tissues and are considered to be markers of exhausted T-cells [33–35].

**Figure 1.** Immune checkpoint molecules and their ligands.

#### **3. Unique Aspects of Immunological Characteristics in the Liver and Hepatocarcinogenesis**

Although ICIs are becoming one of the key agents for the treatment of HCC, the response to this type of agent is still unsatisfactory in the majority of HCC cases compared with other types of malignancies [36]. The relatively low response rate to the ICIs can probably be attributed to the low antigenicity of HCC, as tumor mutation is not high in this type of tumor [36,37]. In addition, as the liver needs to be immunotolerant to nonpathological and persistent inflammation, it carries tolerance mechanisms to immune reactions, including cancer immunity.

The liver is continually exposed to the pathogen and microbe components from the gut via the blood supply of the portal vein. In this situation, the liver limits hypersensitivity to food-derived antigens and components of the intestinal flora to prevent excessive tissue damage and maintain systemic tolerance [38]. Chronic infection with hepatitis virus and persistent stimulation by metabolites further induce immune suppression in the liver, which is one of the unique characteristics of the underlying condition of hepatocarcinogenesis [39]. Resident macrophages (Kupffer cells) play a key role in hepatic tolerance through the production of anti-inflammatory cytokines, leading to downregulation of co-stimulatory molecules. This immunological environment of the liver results in the development of fully exhausted T-cells, where suppressive anti-tumor immunity is not susceptible to rescue by ICIs [36,39]. It has also been reported that CD8<sup>+</sup> PD-1<sup>+</sup> T-cells in NASH livers show a lack of immune surveillance and tissue-damaging function, which contribute to the increase in HCC emergence upon anti-PD-1 treatment in a NASH mouse model [40]. Augmentation of CD8<sup>+</sup> PD-1<sup>+</sup> T-cells was also observed in human NASH; a worse outcome in HCC patients treated with anti-PD-1 antibodies was observed [40]. Although the details of the difference in the response to ICIs between virally induced and NASH-induced HCCs are still unknown, it is possible that different amounts and quality of antigens and the difference in the liver microenvironment, such as the balance between partially exhausted and fully exhausted T-cells, may be involved in the outcome on ICIs [39,41,42].

In addition, ICI may not be effective or may even exacerbate the disease in some HCC patients. It is reported that blockade of the PD-1 and PD-L1 interaction may induce an expansion of PD-1<sup>+</sup> Tregs isolated from the liver of patients with chronic hepatitis C, because PD-1 on Tregs generally plays a role in the regulation of the CD4+CD25+FoxP3<sup>+</sup> T-cells [43]. Therefore, a blockade of the binding of the ligand with PD-1 on Tregs may result in further suppression of anti-tumor immunity [44]. More importantly, PD-1<sup>+</sup> Tregs may be involved in the hyperprogression of tumors in gastric cancer patients treated with anti-PD-1 antibodies [44]. As hyperprogression on anti-PD-1 antibodies has also been reported in HCC cases, ICI can be even detrimental in such cases [45].

#### **4. Signaling Pathways and the Immune Microenvironment of Tumors**

Alterations in oncogenic signaling in cancer not only trigger abnormal differentiation and cell proliferation, but also play a crucial role in the immune evasion of tumors [12]. Cancerrelated signaling affects the state of the tumors' immune components via cytokine, chemokine and growth factor production. To date, genetic alterations in several signaling pathways observed in cancers have been reported to affect the tumor immune microenvironment.

#### *4.1. RAS/MAPK Signaling Pathway*

In malignant melanomas, activating mutations in BRAF (BRAFV600E) induce constitutive activation of the mitogen-activated protein kinase (MAPK) pathway, which stimulates immune-tolerant DCs and inhibits CD8<sup>+</sup> T-cells via the expression of the immunosuppressive cytokines IL-6 and IL-10, as well as via VEGF [46]. This effect has been reported to be inhibited by BRAF inhibitors and VEGF inhibitors. Furthermore, RAS/MAPK signaling inhibits antigen presentation on tumor cells, and inhibition of this pathway is associated with the recovery of MHC expression by IFN-γ in malignant melanomas and breast cancer [47,48]. Meanwhile, in a murine model of pancreatic cancer, activating mutations in KRAS (KRASG12D) induced MDSCs via the production of granulocyte macrophage colonystimulating factor (GM-CSF) and inhibition of CD8<sup>+</sup> T-cell infiltration into tumor tissues, which contributed to the establishment of an immunosuppressive tumor microenvironment [49]. In fact, GM-CSF is known to be upregulated in human pancreatic intraepithelial neoplasia and pancreatic cancer cells [49]. An association between activating mutations in KRAS and resistance to ICIs has also been reported in colorectal cancer. KRAS-mediated repression of interferon regulatory factor 2 (IRF) results in the high expression of the chemokine C-X-C motif ligand 3 (CXCL3), which induces MDSCs that express C-X-C motif chemokine receptor 2 (CXCR2) as the receptor of CXCL3 in tumor tissues [50]. In this manner, activation of KRAS induces MDSC-mediated resistance to antitumor immunity in patients with colorectal cancer (Figure 2).

**Figure 2.** Effects of oncogenic signal activation on the tumor immune environment.

#### *4.2. PI3K/Akt Signaling Pathway*

Activation of the P13K/AKT signaling pathway is involved in critical cellular functions, including survival, inhibition of apoptosis and proliferation. Activation of P13K/AKT is one of the common features of cancer. PTEN, which regulates this pathway, demonstrates loss-of-function mutations in various cancers. In melanomas, mutations in PTEN are related to resistance to anti-PD-1 antibody treatment and are correlated with a reduced volume of CD8<sup>+</sup> TILs [51]. In a murine model, PI3K-β inhibitors improved sensitivity

to treatment with anti-PD-1 antibodies and anti-CTLA4 antibodies when the resistance was induced by loss-of-function mutations in PTEN [51]. While loss of PTEN function is associated with the induction of various immunosuppressive cytokines, it also induces VEGF, which is presumed to be the mechanism by which immunosuppression is induced. In bladder cancer, activating mutations in PI3KCA are associated with a reduction in TIL volume, while PI3K inhibitors lead to an increase in TILs [52]. PI3K inhibitors have also been reported to inhibit Tregs and induce the differentiation of M2 macrophages into the M1 phenotype [53]. In addition, by inducing PD-L1 expression, both the RAS/MAPK and the PI3K/AKT signaling pathways can be involved in the suppression of anti-tumor immunity. We have previously reported that while activating mutations of the PI3K/AKT signaling pathway are associated with increased expression of PD-L1 in HCC, and the volume of TILs is generally high in PD-L1-positive HCC, TILs are deficient in HCCs with activating mutations of the PI3K/AKT signaling pathway [33]. Therefore, aside from external stimulation such as by IFN-γ, induction of PD-L1 is likely to be attributable to genetic mutations in the PI3K/AKT pathway in this setting (Figure 2).

BRAFV600E induces immune-tolerant dendritic cells (DCs) via the induction of IL-6, IL-10 and VEGF, ultimately inhibiting the action of CD8<sup>+</sup> T-cells. In addition, KRASG12D induces MDSCs via GM-CSF production, thereby inhibiting CD8<sup>+</sup> T-cells from infiltrating tumor tissues. In colorectal cancer, KRAS activation is known to induce CXCL3 expression and the induction of MDSCs with CXCR2, the receptor of CXCL3. In contrast, PI3K/AKT signaling activation is associated with VEGF expression and a decrease in tumor-infiltrating lymphocyte (TIL) volume, and has been reported to induce Tregs and inhibit the shift of M2 macrophages into M1 macrophages. In addition, PI3K/AKT signaling induces PD-L1 expression. Under the activation of WNT/β-catenin signaling, activating transcription factor 3 (ATF3)-mediated CCL4 downregulation is considered to reduce the migration of CD103<sup>+</sup> dendritic cells into the tumor and reduce CD8<sup>+</sup> TILs in melanoma. CCL5 was suggested to be involved in HCC. The activation of the WNT/β-catenin signaling pathway has also been reported to be involved in the formation of an immune suppressive tumor microenvironment through the upregulation of IL-10. Activation of the transcription factors YAP/TAZ, which are regulated by Hippo signaling, upregulate PD-L1 and are involved in CXCL5-mediated induction of MDSCs. In addition, YAP is involved in the induction of M2 macrophages by enhancing the transcription of CCL2.

#### *4.3. Wnt/β-Catenin Signaling Pathway*

ICIs are presumed to be insufficiently effective in cases where infiltration of CD8<sup>+</sup> T-cells in tumors is lacking, and an analysis of human melanomas has revealed that activation of the Wnt/β-catenin signaling pathway is associated with reduced TILs in tumors. In melanomas with activating mutations in the Wnt/β-catenin signaling pathway, C-C chemokine ligand 4 (CCL4) is downregulated, which reduces the migration of CD103<sup>+</sup> DCs and leads to a deficiency in CD8<sup>+</sup> TILs. Furthermore, the transcriptional repression of CCL4 was attributed to the activation of activating transcription factor 3 (ATF3) as a result of the activation of the β-catenin signaling pathway [54]. Activating mutations of the Wnt/β-catenin pathway occur frequently in HCC, and immunosuppression in the tumor microenvironment based on activation of this signal in liver cancer is presumed to occur via downregulation of CCL5 [55]. In a murine HCC model, induction of CCL5 increased the number of DCs and CD8<sup>+</sup> T-cells in tumors (Figure 2). In melanomas, activation of Wnt/β-catenin signaling also reportedly led to the upregulation of IL-10 through the binding of β-catenin/T-cell factor (TCF) on the IL-10 promoter, thereby contributing to the formation of an immunosuppressive environment [56].

Aside from the altered Wnt/β-catenin signaling in tumor cells, activation of this signaling pathway reportedly disturbs the effector function of CD8<sup>+</sup> T-cells and induces the exhausted T-cell phenotype in HCC and colorectal cancers, which contributes to the establishment of immune suppressive tumor microenvironment [41]. Interestingly, neutralization of a canonical Wnt ligand, Wnt 3a, enhances the T-cell response through the rescue of DC activation, resulting in tumor regression in a mouse model [42].

#### *4.4. MYC Gene*

The transcription factor c-myc regulates the expression of genes necessary for cell proliferation and survival. In many cancers, amplification and overexpression of c-myc has been observed; these are involved in inducing the expression of immune checkpoint molecules such as PD-L1 and CD47 [57]. CD47 is a cell surface glycoprotein that regulates phagocytosis by binding to signal regulatory protein alpha (SIRP-α), which is specific to macrophages and DCs. Thus, overexpression of c-myc is involved in the immune evasion of cancer cells through CD47 and PD-L1.

#### *4.5. Chromatin Remodeling Pathway*

Genomic DNA is stored in the nucleus as chromatin. During transcription, replication or repair, alterations of the chromatin structure by chromatin remodeling regulate the access of transcription factors to the DNA. The SWItch/sucrose non-fermentable (SWI/SNF) complex is a chromatin remodeling factor that induces the alteration of nucleosomes via ATP hydrolysis. Genetic abnormalities in SWI/SNF complex subunits are frequently observed in HCC and other human tumors [33], and loss-of-function mutations in the polybromo 1 (PBRM1) gene involved in the SWI/SNF complex are common in renal cancer. Intriguingly, this PBRM1 mutation is associated with the therapeutic effect of ICIs in renal cancer. *PBRM1*-deficient renal cancers show altered transcriptional expression in the JAK/STAT (Janus kinase/signal transducers and activators of transcription) and immune signaling pathways [58].

#### *4.6. JAK/STAT Signaling Pathway*

The JAK/STAT pathway, which transmits signals that are crucial for growth, differentiation, survival and immunity, is altered in many types of malignancy. The downstream transcription factor, STAT3, acts on the PD-L1 promoter, thereby inducing upregulation of PD-L1 in cancer cells. In melanomas, JAK1 and JAK2 mutations inhibit signals from interferon receptors and reduce antigen presentation on tumor cells, which results in resistance to ICI therapy [59]. Meanwhile, β2-microglobulin gene mutations have been reported to induce resistance to ICI treatment via the loss of MHC Class I antigen expression on the cell surface [60].

#### *4.7. Hippo Signaling Pathway*

Hippo signaling, which is involved in the regulation of growth and differentiation as well as in controlling organ size, is dysfunctional in many malignancies. Reduced Hippo signaling is also associated with cancer's immune evasion. Hippo signaling regulates yes-associated protein (YAP) and "transcriptional coactivator with PDZ-binding motif" (TAZ), the activation of which leads to the expression of PD-L1 and stimulates MDSCs carrying CXCR2, by upregulation of its ligand, CXCL5 [61]. In a murine model of HCC, YAP was reported to be associated with tumor immunosuppression via the induction of M2 macrophages resulting from enhanced transcription of CCL2 (Figure 2) [62].

#### *4.8. DNA Repair Pathway*

It is well known that cancers carrying mutations in DNA mismatch genes induce a large number of neoantigens that are attributed to the emergence of a variety of passenger mutations that occur in the microsatellite sequences of the DNA, where anti-tumor immunity is enhanced. Therefore, microsatellite instability is a biomarker for efficacy in the treatment of ICIs [63]. Similarly, cancers with a high mutation burden (TMB) are also markers of tumors with an active immune microenvironment because of their high antigenicity [64]. Recently, it was reported that loss-of-function mutations in the breast cancer susceptibility (*BRCA*) 1 and *BRCA 2* genes, which are involved in the homologous

recombination pathway of DNA repair, are also markers of a high TMB and could be predictors of the outcome of ICI-based treatment [65]. From this perspective, alterations in DNA repair pathways are critical for the establishment of high antigenicity and "immune hot" status in cancer.

#### *4.9. VEGF Signaling*

In tumors, external stimulation can lead to the production of growth factors. In HCC, tissue hypoxia leads to the production of VEGF via the activation of hypoxia-inducible factor 1 (HIF-1), resulting in tumor angiogenesis. The cellular components of tumors that suppress tumor immune responses, such as MDSCs, Tregs and TAMs, express VEGF receptors; therefore, inhibition of VEGF/VEGFR can alter anti-tumor immunity [11]. Anti-PD-1 antibodies and anti-VEGFR-2 antibodies have been reported to have a synergistic effect in murine models of HCC. Anti-VEGFR-2 antibodies induce an increase in CTLs and a decrease in TAMs and Tregs [66]. Atezolizumab + bevacizumab is expected to combine the effects of ICIs with inhibition of VEGF signaling to alter immunity. Pembrolizumab + lenvatinib, a multikinase inhibitor (MKI) with a powerful antiangiogenic effect, and atezolizumab + cabozantinib, which is capable of blocking angiogenesis through the inhibition of VEGFR and AXL, and camrelizumab (an anti-PD-1 antibody) + apatinib (a selective VEGFR2-tyrosine kinase inhibitor) are in Phase III clinical trials, while avelumab (an anti-PD-L1 antibody) + axitinib (which strongly inhibits VEGFR) are undergoing Phase I/II trials (Table 1). ICIs and agents with an anti-VEGF/VEGFR effect are currently the most promising combination therapies for HCC because of their synergistic effect on cancer immunity [6,11].


**Table 1.** Clinical trials for combinations of molecular targeted agents and immune checkpoint inhibitors in hepatocellular carcinoma.


**Table 1.** *Cont.*

<sup>1</sup> National Clinical Trial number (ClinicalTrials.gov registry number). <sup>2</sup> MTA: molecular targeted agent; ICI: immune checkpoint inhibitor; TKI: tyrosine kinase inhibitor; Ab: antibody.

#### **5. Signaling Pathway Abnormalities and the Immune Microenvironment in HCC**

In a mouse model of HCC, it has been shown that activation of Wnt/β-catenin signaling induces reduced migration of CD103<sup>+</sup> DCs and CD8<sup>+</sup> TIL deficiency via downregulation of CCL5. Previous reports have also shown that Wnt/β-catenin activation is associated with the reduced expression of T cell-derived genes in HCC tissues. Therefore, HCC with activated Wnt/β-catenin signaling is unlikely to respond to ICIs because of the "immune cold" phenotype [67]. In fact, post-ICI therapy outcomes are reported to be poor in cases of HCC with Wnt/β-catenin activation [68]. Using a cohort of HCC cases from The Cancer Genome Atlas (TCGA), we determined that the expression of T cell-related genes was low in cases of HCC with activating mutations in Wnt/β-catenin (Figure 3). In addition, in an analysis of HCC tissues, we determined that HCCs with activating mutations in Wnt/β-catenin pathway genes are significantly deficient in CD8<sup>+</sup> TILs [33]. However, we did not find CD8<sup>+</sup> TILs to be associated with mutations in any other oncogenic signaling pathways (Table 2).

**Figure 3.** Tumor-infiltrating lymphocyte status and abnormal WNT/β-catenin activation. Hierarchical cluster analysis was used to classify HCCs based on the expression of eight T cell-related cytokine mRNAs obtained from the TCGA transcriptome dataset (RNA-seq V2 RSEM) (**a**) "Immune hot" and "immune cold" refer to HCCs with high and low levels of T cell-related gene expression, respectively. Thus, "immune hot" indicates a large volume of tumor-infiltrating lymphocytes (TILs), while "immune cold" suggests that TILs are deficient. (**b**) The presence or absence of activating mutations in WNT/β-catenin pathway genes are determined based on the presence or absence of *CTNNB1, AXIN1* and *APC* mutations in the TCGA provisional dataset obtained. Activating mutations in the WNT/β-catenin pathway are more frequently detected in "immune cold" HCCs than in "immune hot" tumors (*p* = 0.0001). The TCGA dataset used in the present study was downloaded in September 2019.


**Table 2.** Association between alterations in oncogenic signaling pathways and the degree of CD8<sup>+</sup> tumor infiltrating lymphocytes.

Degree of tumor infiltrating lymphocytes (TILs) are compared between HCCs with mutations in each oncogenic pathway and those without mutations. In total, 154 HCCs were examined for mutations using the Ion AmpliSeq Comprehensive Cancer Panel, and the degree of CD8<sup>+</sup> cells was examined using immunohistochemistry. <sup>1</sup> Median: median number of CD8+ TILs/high power field. <sup>2</sup> *p* value by Wilcoxon's rank-sum test.

In a transcriptome analysis, we reported that Wnt/β-catenin signaling activation was associated with the decreased expression of gene sets related to T-cell priming/activation, IFN-γ response, immunosuppression and Tregs; it was most significantly associated with the downregulation of genes related to the IFN-γ response in multivariate analysis [69]. These data are consistent with the deficiency in CD8<sup>+</sup> T-cells in HCC tissues. In addition, we also reported that activating mutations in the Wnt/β-catenin pathway is negatively associated with PD-L1 expression in HCC [33]. As the expression of PD-L1 can be induced by the stimulation of IFN-γ, the lack of PD-L1 expression in HCC with Wnt/β-catenin activation can probably be attributed to the low degree of CD8<sup>+</sup> TILs that should secrete IFN-γ [69]. On the other hand, a previous study found that mutations in genes involved in chromatin remodeling, such as AT-Rich Interaction Domain 2 (*ARID2*), were also associated with an immunosuppressive tumor microenvironment through the expression of genes involved in the induction of M2 macrophages [14], although there were no associations between mutations of the genes involved in chromatin remodeling and the degree of CD8<sup>+</sup> TILs as well as PD-L1 expression [33]. As mutations of *ARID2* are reportedly associated with the TAM subclass of HCC, the immune suppressive mechanism in HCCs with *an ARID2* mutation should be different from that of CTNNB1 [14]. In contrast, PD-L1-positive HCCs often have high levels of CD8<sup>+</sup> TILs [33]. This may be due to the fact that PD-L1 expression in HCC cells can be mainly attributed to stimulation by the IFN-γ from TILs. It is possible that, under continuous immune response to cancer cells, many CD8<sup>+</sup> TILs are prone to expressing multiple inhibitory receptors (PD1, TIM-3, LAG-3) that result in the exhausted phenotype of T-cells [33]. In many cases, PD-L1 expression is considered to be a favorable prognostic factor of ICI therapy, suggesting that blockade of the PD-1/PD-L1 response could, at least partially, activate the T-cell immune response, even if the immune cells express additional inhibitory receptors. Indeed, we found that the absence of activating mutations in Wnt/β-catenin pathway genes, a high CD8<sup>+</sup> TIL volume and

PD-L1 expression were associated with long progression-free survival of HCC patients on anti-PD-1 antibody therapy, regardless of the expression of other inhibitory receptors, such as TIM-3 and LAG-3 [69]. In this way, assessments of gene alterations in cellular signaling pathways are not only useful for finding suitable MTAs that act on the altered cellular signal, but may also, theoretically, serve to predict the response to ICI therapy, based on the tumor immune microenvironment.

#### **6. Conclusions**

Alterations of cell signaling pathways play a critical role not only in the development of a malignant phenotype in cancer cells but also in the determination of anti-cancer immunity. In a Phase III clinical trial with HCC patients, ICI monotherapy failed to yield a significant anti-cancer response, suggesting that ICIs will be used primarily in combination therapy [6]. It has been speculated that the "immune cold" phenotype of the tumor microenvironment is critical for poor prognosis with ICIs, where activation of the Wnt/β-catenin signaling pathway plays an important role. From this point of view, understanding the response of HCCs carrying Wnt/β-catenin mutations to combination therapy with ICI and MTA is clinically important but has not been clarified yet. Currently, a combination of atezolizumab + bevacizumab is applicable for unresectable HCC; the efficacy of this combination on HCCs showing the "immune cold" phenotype is now under investigation [36]. In addition, aside from atezolizumab + bevacizumab, many ongoing clinical trials have examined combinations of ICIs and MTAs for HCCs that are refractory upon ICI monotherapy, mainly with MTA showing an anti-angiogenic effect (Table 1). However, future trials are likely to examine combinations of ICIs with agents that inhibit other oncogenic pathways that are critical for hepatocarcinogenesis, such as the Wnt/β-catenin pathway, the RAS/MAPK pathway and the PI3K/AKT pathway.

**Funding:** This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (KAKENHI: 21K07184, N. Nishida) and a grant from the Smoking Research Foundation (N. Nishida).

**Conflicts of Interest:** The author has not nothing to declare.

#### **References**


### *Systematic Review* **A Broad Overview of Signaling in** *Ph***-Negative Classic Myeloproliferative Neoplasms**

**Ana Guijarro-Hernández <sup>1</sup> and José Luis Vizmanos 1,2, \***


**Simple Summary:** There is growing evidence that *Ph*-negative myeloproliferative neoplasms are disorders in which multiple signaling pathways are significantly disturbed. The heterogeneous phenotypes observed among patients have highlighted the importance of having a comprehensive knowledge of the molecular mechanisms behind these diseases. This review aims to show a broad overview of the signaling involved in myeloproliferative neoplasms (MPNs) and other processes that can modify them, which could be helpful to better understand these diseases and develop more effective targeted treatments.

**Abstract:** *Ph*-negative myeloproliferative neoplasms (polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF)) are infrequent blood cancers characterized by signaling aberrations. Shortly after the discovery of the somatic mutations in JAK2, MPL, and CALR that cause these diseases, researchers extensively studied the aberrant functions of their mutant products. In all three cases, the main pathogenic mechanism appears to be the constitutive activation of JAK2/STAT signaling and JAK2-related pathways (MAPK/ERK, PI3K/AKT). However, some other non-canonical aberrant mechanisms derived from mutant JAK2 and CALR have also been described. Moreover, additional somatic mutations have been identified in other genes that affect epigenetic regulation, tumor suppression, transcription regulation, splicing and other signaling pathways, leading to the modification of some disease features and adding a layer of complexity to their molecular pathogenesis. All of these factors have highlighted the wide variety of cellular processes and pathways involved in the pathogenesis of MPNs. This review presents an overview of the complex signaling behind these diseases which could explain, at least in part, their phenotypic heterogeneity.

**Keywords:** myeloproliferative neoplasms; signaling pathways; JAK2; CALR; MPL; TPOR

#### **1. Introduction**

Myeloproliferative neoplasms (MPNs) are rare hematological malignancies characterized by the clonal expansion of mature myeloid cells. MPNs arise from certain somatic mutations in hematopoietic stem cells (HSCs) which provide a selective advantage and lead to the expansion of aberrant clones.

Classic MPNs consist of chronic myeloid leukemia (CML), polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF). In the last few years, the advances in molecular biology have provided key insights into the molecular mechanisms behind these diseases. CML is genetically defined by the *Philadelphia* (*Ph*) chromosome, the result of t(9;22)(q34;q11). This translocation leads to the production of a chimeric BCR-ABL1 protein with constitutive tyrosine kinase activity. The description of the *Ph* chromosome as a disease-initiating event in CML revolutionized the diagnosis and treatment of this disease [1]. The targeted therapy imatinib showed a specific inhibitory capacity against

**Citation:** Guijarro-Hernández, A.; Vizmanos, J.L. A Broad Overview of Signaling in *Ph*-Negative Classic Myeloproliferative Neoplasms. *Cancers* **2021**, *13*, 984. https:// doi.org/10.3390/cancers13050984

Academic Editor: Alain Piché

Received: 8 February 2021 Accepted: 22 February 2021 Published: 26 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the tyrosine kinase activity of BCR-ABL1 [2–4] that, despite not being curative [5], increased the 10-year survival of CML patients in chronic phase to more than 83%–84% [6,7].

This review is focused on PV, ET and PMF, all of them *Ph*-negative MPNs that share similar and mostly mutually exclusive driver mutations affecting *JAK2*, *MPL* and *CALR*. The aberrant functions of the mutant products encoded by these genes have been extensively studied and the main mechanisms that lead to the myeloproliferation described. Currently, it is considered that the major hallmark of *Ph*-negative MPNs is the constitutive activation of JAK2-related signaling pathways. In fact, at this time, the only targeted therapy approved in MPNs is the JAK1/2 inhibitor ruxolitinib, which can reduce splenomegaly and other common symptoms in patients with PMF, post-PV/ET MF [8,9] and PV resistant or intolerant to hydroxyurea [9,10]. Although a reduction in the mutant allele burden is rare [9], it could be achieved in long-term treatment [11]. However, the improvement in the overall survival of ruxolitinib-treated patients has been questioned [12–14]. Actually, malignant cells can still survive in these patients and the clinical response could be mainly due to the downmodulation of proinflammatory cytokines derived from the JAK2 inhibition [15]. These arguments have led researchers to question whether JAK2 is really the best drug target in these diseases or not [16].

In the meantime, some non-canonical mechanisms of mutant JAK2 [17–24] and CALR [25–33] have been described. Chronic inflammation [34–62] and the bone marrow microenvironment [63–72] also seem to contribute to the heterogeneous phenotypes found among MPN patients.

Additionally, mutations in disease-modifying genes that seem to increase the risk of leukemic transformation or progression from ET to myelofibrosis have also been identified [73–75]. The products encoded by these genes are involved in epigenetic modification, tumor suppression, transcription regulation, splicing, and some other signaling pathways [76,77]. Other factors, such as genetic predisposition, age or environment have also been shown to influence the heterogeneity of MPN phenotypes [78].

This review presents an overview of the signaling behind *Ph*-negative MPNs attending not only to the activation of JAK2-related canonical signaling pathways, but also to other non-canonical pathways, disease-modifying signaling, and additional factors that have been found to be involved in the pathogenesis of these diseases.

#### **2. JAK2-Related Canonical Signaling Pathways**

JAK2 signaling is activated through a variety of receptors such as those for erythropoietin (EPOR), thrombopoietin (TPOR), and granulocyte/macrophage colony-stimulating factor (GM-CSFR). They regulate the production of the erythroid, megakaryocytic, and granulocytic lineages, respectively. When stimulated by ligands, receptors dimerize and bring JAK2 kinases into proximity. JAK2 is phosphorylated upon receptor binding and induces the phosphorylation of the cytoplasmic portion of the receptor and downstream factors.

In 2005, several research groups simultaneously published the presence of the somatic mutation p.V617F (JAK2V617F) in the exon 14 of *JAK2* in patients with PV (96%), PMF (65%) and ET (55%) [79–84]. This mutation impairs the physiological inhibitory function of the JH2 pseudokinase domain upon the JH1 kinase domain, which acquires a constitutive activation that promotes JAK2 phosphorylation in the absence of ligand stimulation (Figure 1). In 2007, four additional gain-of-function somatic mutations in the exon 12 of *JAK2* were detected in 3% of patients with PV [84,85]: p.N542-E543del (30%), p.K539L (14%), p.E543- D544del (12%), and p.F537-K539delinsL (10%). All of them are located upstream of the JH2 pseudokinase domain and promote an increased phosphorylation of JAK2 compared to p.V617F [86].

**Figure 1.** JAK2-related canonical signaling pathways active in *Ph*-negative myeloproliferative neoplasms (MPNs). Mutations in *CALR* (CALRMUT), *JAK2* (JAK2V617F), and *MPL* (TPORW515) lead to the constitutive activation of JAK2/STAT, PI3K/AKT, and MAPK/ERK signaling that promotes the transport to the nucleus of several transcription factors such as STATs and FOXO. There, they regulate transcription of their target genes, causing increased proliferation and survival of mutant cells. Mutant proteins are depicted in red.

In 2006, the gain-of-function mutation p.W515L in the exon 10 of *MPL* was identified in a minor proportion of MPN patients [87]. p.W515L and p.W515K are the most commonly reported mutations, identified in approximately 5% of PMF patients and 1% of ET patients [88]. *MPL* encodes the thrombopoietin receptor (TPOR), which depends on JAKs to mediate signal transduction. *MPL* mutations (TPORW515) promote the dimerization and activation of TPOR, leading to transphosphorylation and activation of the previously bound JAK2 proteins (Figure 1) [89].

The molecular alteration that causes the 60–90% of PMF and ET cases in patients not harboring *JAK2*/*MPL* mutations was described in 2013 [90]. During that year, two research groups identified mutations in *CALR* [91,92], a gene that encodes calreticulin, a ubiquitous protein found in the endoplasmic reticulum (ER) of all nucleated cells with multiple functions inside and outside this organelle. CALR is a Ca2+-binding chaperone mainly involved in the regulation of intracellular Ca2+ homeostasis and a regulator of protein folding in the cellular response to ER stress (unfolding protein response (UPR)) [93]. However, this protein has been also found associated with other cytoplasmic, nuclear and extracellular proteins, so it could be involved in a wide variety of signaling pathways [94]. In fact, CALR has been associated with cellular stress responses, adipocyte differentiation, cardiogenesis, proliferation, wound healing, apoptosis and immunogenic cell death [90,94].

The structure of wild-type CALR consists of a signal peptide and three domains: an amino-terminal N-domain, a proline-rich P-domain and a carboxy-terminal C-domain, which contains an ER retention signal (KDEL). The *CALR* mutations described to date are insertions or deletions in exon 9 that shift the reading frame by one base pair (+1), mainly a 52-bp deletion (c.1902\_1143del) or type 1 mutation (CALRdel52), and a 5-bp insertion (c.1154\_1155insTTGTC) or type 2 mutation (CALRins5). As a result, mutant CALRs (CALRMut) show a novel C-terminal end that lacks the ER retention motif (KDEL) [91,92] and some Ca2+-binding sites [95]. In 2016, it was published that CALRMut is transported to the cellular membrane where it activates TPOR in a ligand-independent manner (Figure 1) [96–99]. The characterization of the TPOR binding capacity has revealed that the C-terminal end of CALRMut blocks the P-domain of the protein, which constitutively exerts an inhibitory effect on the N-domain. Consequently, the N-domain can bind to immature N-glycans on TPOR [96]. This mechanism is consistent with the observation that the Nglycan binding motif located in N-domain of CALRMut is required for TPOR activation [97]. In fact, blocking N-glycosylation on asparagine 117 of TPOR diminishes CALR-dependent TPOR activation [97,100]. Both wild-type and mutant CALR recognize immature forms of N-glycans and fold the protein correctly, but CALRMut fails to dissociate from the targeted protein [101]. Thus, the CALRMut-TPOR complex moves from the ER to the plasma membrane through the Golgi apparatus and is secreted out of cells [102]. However, secreted CALRMut is only capable to activate TPOR on the cell surface of cells expressing CALRMut since only these cells have the immature N-glycans on TPOR [96,102,103]. Stimulation of TPOR leads to the activation of JAK2-dependent signaling in a similar way to the rest of the *Ph*-negative MPNs.

In conclusion, the mutations described to date in *JAK2*, *MPL* and *CALR* lead to a constitutive activation of JAK2, which ultimately causes the aberrant proliferation and survival of malignant myeloid clones. The three major downstream signaling pathways that are activated by JAK2 are JAK2/STATs, MAPK/ERK, and PI3K/AKT (Figure 1). The evidence suggests that each of these pathways plays an important role in MPNs, although the JAK2/STAT pathway appears to be the main one. In fact, dysregulation of JAK2/STAT signaling has been identified in all MPNs regardless of mutational status [104].

#### *2.1. JAK2/STAT Pathway*

In MPNs, JAK2 phosphorylates and activates STATs (mainly STAT1, STAT3 and STAT5). It seems that STATs are differentially activated depending on the type of MPN. For example, *MPL* mutations increase STAT3 and STAT5 signaling. In PV patients, JAK2V617F binds to EPOR promoting STAT5 activation. In ET patients, both JAK2V617F and CALRMut bind to TPOR; JAK2V617F enhances the phosphorylation of STAT1 and STAT3 but CALRMut promotes STAT3 and STAT5 activation. However, in PMF, phosphorylation of STAT3 is decreased in both JAK2V617F and mutant CALRs. To date, the precise mechanisms that explain differential activation of the STATs remain unclear [78].

Once the STATs are phosphorylated, they form a dimer that enters the nucleus to activate the transcription of target genes (Figure 1). In this way, JAK2/STAT signaling stimulates cell proliferation, differentiation and survival.

#### *2.2. MAPK/ERK Pathway*

The activated JAK2 can also lead to the phosphorylation of ERK, a serine threonine kinase that activates multiple proteins in both the cytoplasm and the nucleus. ERK is a key regulator of a wide variety of signaling pathways, so its deregulation could disrupt multiple processes. In the cytoplasm, ERK contributes to ion transport, apoptosis, and regulation of metabolism, among others. In the nucleus, it targets regulators of cell cycle and multiple transcription factors (Figure 1) [105].

#### *2.3. PI3K/AKT Pathway*

JAK2 activation also stimulates the PI3K/AKT pathway. AKT is a cell survival kinase which inhibits apoptosis by phosphorylating the proapoptotic protein BAD and the transcription factor FOXO3A. In addition, AKT can activate a wide range of mechanisms such as protein translation through mTOR or the cell cycle machinery (Figure 1) [105].

#### **3. Non-Canonical Signaling Pathways**

#### *3.1. JAK2-Related Non-Canonical Signaling*

In 2009, activated JAK2 was described to be in the nucleus of hematopoietic cells and to phosphorylate Y41 on histone 3 (H3Y41). This event prevents the binding of heterochromatin protein 1 alpha (HP1α) to H3Y41 [17]. HP1α shows a proliferationdependent regulation and is involved in gene silencing, genome stability, and chromosome segregation (Figure 2). It has been found overexpressed in some tumors, and it has been proposed as a potential hallmark of cell proliferation that could be relevant in clinical oncology [18].

**Figure 2.** Main non-canonical signaling pathways activated by JAK2V617F in *Ph*-negative MPNs. In PV patients, JAK2V617F (depicted in red) activates the adhesion receptor Lu/BCAM through the RAP1-AKT signaling pathway, making their erythrocytes more adhesive. JAK2V617F has also been described to promote aberrant signaling in the nucleus, where it prevents the binding of heterochromatin protein 1 alpha (HP1α) and inhibits the methylation of histones via protein arginine methyltransferase 5 (PRMT5) impairment. MPN patients with JAK2V617F also seem to be insensitive to the anti-inflammatory cytokine IL-10, increasing TNF-α production through Toll-Like Receptor (TLR) signaling.

JAK2V617F also acquires the ability to phosphorylate the protein arginine methyltransferase 5 (PRMT5) leading to an impairment in its ability to methylate histones (Figure 2). When PRMT5 is knocked down in CD34+ cells, an increased colony formation and erythroid differentiation can be observed [19].

A recent study also suggests that erythrocytes from PV patients are more adhesive since JAK2V617F activates the erythrocyte adhesion receptor Lu/BCAM through an EPORindependent RAP1/AKT signaling pathway (Figure 2) [20].

Finally, monocytes from MPN patients with JAK2V617F have been found to have a defective negative regulation of toll-like receptor (TLR) signaling leading to increased production of the inflammatory cytokine TNF-α. These monocytes are insensitive to the anti-inflammatory cytokine IL-10, which in turn negatively regulates TNF-α production through TLR (Figure 2) [21]. Studies on TNF-α knockout mice have demonstrated that this cytokine is required for the development of an MPN-like disease [22]. Unrestrained production of TNF-α has been observed in an MPN patient but also in his identical twin, suggesting that it may be a genetic feature rather than a consequence of the disease [21]. In any case, the inflammatory environment can favor the maintenance and expansion of the JAK2V617F mutant clone since these cells are resistant to inflammation whereas non-mutant cells are not [22]. Thus, the JAK2V617F mutant clone seems to induce non-mutant cells to produce inflammatory cytokines, reinforcing the self-perpetuating environment for its continuous selection [23]. Finally, CD34+ progenitors of a PV patient with JAK2V617F have been reported to use dual-specificity phosphatase 1 (DUSP1) to protect themselves against inflammatory stress and DNA damage, promoting their proliferation and survival in this microenvironment (Figure 2) [24].

#### *3.2. CALR-Related Non-Canonical Signaling*

Several studies have identified novel mechanisms that collaborate with the activation of TPOR in CALR-mediated cellular transformation (Figure 3). CALRMut seems to cause reduced activation of the UPR pro-apoptotic pathway and to have an increased sensitivity to oxidative stress by the down-modulation of oxidation resistance 1 (OXR1) in K562 cells. These mechanisms lead to resistance to UPR-induced apoptosis and genomic instability, respectively [25]. Moreover, CALRdel52 causes increased recruitment of the friend leukemia integration 1 (FLI1) transcription factor to the *MPL* promoter to enhance transcription [26], which suggests a promotion of tumorigenesis by modulating transcription through interactions with transcription factors in the nucleus.

Bioinformatic analyses of CALRMut revealed the appearance of potential phosphorylation sites for kinases that may have a role in the regulation of multiple cellular activities [27] and recent studies have shown that CALRMut causes increased binding affinities for proteins involved in the activation of the UPR (HSPA5, HSPA9, and HSPA8) and cytoskeletal (MYL9 and APRC4) and ribosomal proteins (RP17, RSP23, and RPL11), as well as reduced binding to MSI2, a transcriptional regulator that targets genes mainly involved in cell cycle regulation [26].

On the other hand, CALR is an integral part of the peptide loading complex (PLC), which mediates the loading of cellular antigens onto major histocompatibility complex class I (MHC-I) molecules. In addition to CALR, the PLC is composed of PDIA3, TAPbinding protein, TAP1, and TAP2. Specifically, CALR interacts with PDIA3 in a glycandependent manner and preserves steady-state levels of TAP-binding protein and MHC-I heavy chains. Besides, it rescues suboptimally assembled MHC-I molecules from post-ER compartments [28]. HEK293T cells lacking CALR expression show a reduction of properly loaded MHC-I on the cell surface, a defect that is not restored by expression of CALRMut [29]. Consistent with this, cells with CALRMut show reduced antigen presentation on MHC-I (Figure 3) [54] and decreased binding affinities for PDIA3 [26]. These results suggest that *CALR* mutations have a loss-of-function effect on PLC and, therefore, may contribute to the development of MPN by promoting immunoevasion after loss of tumor antigenicity [28]. Additionally, CALR operates as a key damage-associated molecular

pattern (DAMP) when it is translocated to the outer cell membrane of dying cancer cells. CALR-exposing cancer cells deliver pro-phagocytic signals to antigen presenting cells (APCs) and activate dendritic cell efferocytosis. Mutations in *CALR* increase the secretion of the protein both in vitro and in vivo since the ER retention motif (KDEL) is compromised. Soluble CALR binds to CALR receptors in the APCs and limit their ability to phagocytise, leading to immunosuppressive effects (Figure 3) [30].

**Figure 3.** Major non-canonical mechanisms derived from CALRMut. CALRMut (depicted in red) shows different binding affinities for proteins implicated in the unfolding protein response (UPR) (HSPA5, HSPA9, and HSPA8), proteins of the cytoskeleton (MYL9 and APRC4), and ribosomal proteins (RP17, RSP23, and RPL11), as well as reduced binding to MSI2, a transcriptional regulator that target genes mainly involved in cell cycle regulation. Additionally, CALRMut seems to reduce the activation of the pro-apoptotic pathway of the UPR and increases oxidative stress and DNA damage through the downmodulation of oxidation resistance 1 (OXR1). CALRMut also shows decreased binding affinities for PDIA3 and has a loss-of-function effect on the peptide loading complex (PLC), which mediates the loading of cellular antigens onto major histocompatibility complex class I (MHC-I) molecules, favoring immunoevasion. Mutations in *CALR* increase the secretion of the protein and bind to CALR receptors in antigen presenting cells (APCs), limiting their ability to phagocytize wild-type CALR-exposing cancer cells. The main differences between the phenotypes observed in patients with type 1 (del52) and type 2 (ins5) mutations have been attributed to thrombopoietin receptor (TPOR)-independent cytosolic calcium fluxes and the binding affinity for the transcription factor FLI1.

The wild-type CALR protein also regulates the activation of the stored-operated calcium entry (SOCE) machinery by interacting with PDIA3 and STIM1. Concretely, STIM1 is a protein of the SOCE machinery that leads to calcium mobilization. CALRMut has been shown to trigger TPOR-independent cytosolic calcium fluxes in megakaryocytes through defective interactions between CALRMut, PDIA3 and the SOCE machinery. This results in uncontrolled proliferation of megakaryocytes that can be reversed with a SOCE inhibitor [31].

The type of *CALR* mutation has been associated with different disease features. Thus, type 1 mutations are more often associated with PMF (70%) or progression from ET to a myelofibrotic state [32], while type 2 mutations are more often associated with ET [91]. The mechanisms underlying this phenomenon have not been fully elucidated, but it has been demonstrated that type 2 mutants retain longer stretches of the negatively charged amino acids of wild-type CALR than type 2 mutants, which may neutralize the positive electronic charge generated at the C-terminal end. Additionally, type 1 mutant C-terminus generates greater changes in megakaryocyte cytosolic calcium flux than type 2 mutants [33].

#### *3.3. Additional Non-Canonical Signaling*

Non-canonical mechanisms affecting inflammatory signaling pathways and the bone marrow microenvironment have been widely observed in all MPNs, regardless of subtype and driver mutation.

#### 3.3.1. Inflammatory Signaling Pathways

As previously noted, chronic inflammation is a characteristic feature of MPNs (Figure 4). In fact, MPN patients typically exhibit increased levels of inflammatory cytokines [34,106]. The impaired JAK2/STAT signaling is not the only contributor to inflammation in these diseases, as the inhibition of JAK2 is not sufficient to normalize the levels of inflammatory cytokines [35]. On the contrary, a significant enrichment of the NF-κB signaling pathway has been observed in both malignant and non-malignant cells in MPNs [36].

NFE2 overexpression has also been reported in most patients and seems to play a role in chronic inflammation [37,38]. NFE2 participates in inflammatory cascades by increasing IL-8 transcription and promotes proliferation by activating the expression of CDK4, CDK6 and cyclin D3 [39,40]. In addition, it produces reactive oxygen species (ROS), a group of highly reactive oxygen-containing molecules which participate in numerous biological processes [41]. This results in lipid and protein oxidation, increased oxidative DNA damage (8-oxo-G), and subsequent double-stranded DNA breaks that induce instability [38]. Excessive ROS production and subsequent oxidative stress confer a proliferative advantage to JAK2V617F clones and activate proinflammatory pathways (NF-κB) that create more ROS. In this way, MPNs have recently been described as "a human inflammation model for cancer development", as they are characterized by a self-perpetuating circle in which inflammation creates ROS which in turn creates more inflammation [42].

Multiple inflammatory signaling pathways such as IFN-α and IL-1β have been also found to be involved in the pathogenesis of MPN. Interferons are key regulators of HSCs. Data from murine PV JAK2V617F models have shown that hematopoietic stem progenitor cells (HSPCs) become more proliferative and lose quiescence when treated with IFN-α, leading to their depletion [43,44]. The ability to deplete previously dormant malignant stem cells together with the enhancement of the immune response have made IFN-α one of the most efficient treatment options in MPNs [107]. On the other hand, IL-1β is a proinflammatory cytokine released by myeloid cells in response to TLR stimulation, that activates multiple downstream pathways such as NF-kB and p38 MAPK [45]. The preleukemic niche of MPNs secretes high levels of IL-1, which drives granulocyte/macrophage differentiation [46]. IL-6 and IL-8 also seem to participate in MPN pathogenesis. IL-6 is a proinflammatory cytokine produced by monocytes, macrophages and T-cells that signals via JAK1/STAT3 [45]. Several mouse models for MPNs have shown a high expression of IL-6 in both mutant and wild-type HSCs [23]. Additionally, elevated IL-6 levels have been observed in JAK2V617F PV and PMF patients [47]. In fact, some studies point that IL-6 may participate in the progression of MPN to AML [45]. IL-8 is also a proinflammatory

chemokine released in response to IL-1 or TNF-α that binds to CXCR1 or CXCR2 and activate JAK/STAT, PI3K/AKT, MAPK, PLC/PKC and FAK [45]. Elevated levels of IL-8 have been found in PV and ET patients [48].

**Figure 4.** Non-canonical inflammatory signaling pathways affected in *Ph*-negative MPNs, regardless of subtype and driver mutation. The preleukemic niche of MPNs secretes high levels of IL-1, a proinflammatory cytokine that activates multiple downstream pathways, such as p38 MAPK and NFκB. NF-κB, in turn, generates high levels of IL-8, a proinflammatory cytokine that binds to CXCR1 or CXCR2 and activates STAT3, PI3K/AKT, p38 MAPK, FAK and PLC/PKC. NFE2 overexpression has also been reported in most MPN patients and has been associated with high IL-8 levels and increased ROS and DNA damage. On the other hand, IL-6 is a proinflammatory cytokine produced by monocytes, macrophages and T-cells that signals via JAK1/STAT3, whose levels have been found elevated in JAK2V617F PV and PMF patients. Finally, IFN-α is a key regulator of hematopoietic stem cells (HSCs) that depletes previously dormant hematopoietic stem progenitor cells (HSPCs) and enhances the immune response. A pathogenic role of oncostatin-M, TGF-β1, platelet-derived growth factor (PDGF), basic fibroblast growth factor (BFGF), VEGF, bone morphogenic proteins, inhibitors of matrix metalloproteinases (MMPs) and lipocalin-2 has been suggested in *Ph*-negative MPNs.

Numerous cytokines have been implicated in mediating fibrosis, osteosclerosis and angiogenesis in PMF patients. Thus, several studies have suggested a pathogenic role for oncostatin-M [49], TGF-β1 [50,51], platelet-derived growth factor [51], basic fibroblast growth factor [50], VEGF [52], bone morphogenetic proteins [53], and inhibitors of matrix metalloproteinases [54,55].

Myeloid cells have been reported also to produce elevated levels of lipocalin-2 in PV, ET, and PMF patients. This protein increases the growth of bone marrow cells in PMF patients, but not in healthy donors. On the contrary, it increases reactive oxygen species, DNA damage, and apoptosis in normal cells, but not in PMF patients. Lipocalin-2 also induces the expression of factors that contribute to fibrosis, such as VEGF, TGF-β1, bone morphogenetic protein-2, RUNX2, osteoprotegerin and collagen type I [56,108].

Heat shock proteins (HSPs) are key players during inflammation. HSP90 stabilizes numerous proteins, such as JAK2. The HSP70 family is composed of some proteins (HSPA5, HSPA8, and HSPA8) that have been found to be enriched in fractions bound to CALRMut [26]. HSP70 also seems to contribute to cell proliferation through regulation of JAK2/STAT signaling. In fact, the inhibition of HSP70 expression in an ex vivo model of PV and ET increased apoptosis of the erythroid lineage and decreased JAK2 signaling [57]. HSP70 also activates TLR2 and TLR4, leading to NF-κB activation, rapid calcium flux, and TNF-α, IL1-β and IL-6 production [58]. Moreover, HSP70 can be secreted as a "danger signal" and bind peptides to form a complex that binds to cell surface receptors, such as CD91 and Lox-1 [59].

Finally, there is also evidence for a link between inflammation and thrombosis. Thrombosis in MPN patients has been associated with an increased platelet-leukocyte interaction. While MPN leukocytes overexpress the surface protein CD11b, its receptor (CD62p) is upregulated on platelets. This results in increased formation of leukocyte-platelet complexes [60–62].

#### 3.3.2. Bone Marrow Microenvironment

The bone marrow microenvironment is a complex and dynamic structure composed of multiple cell types. Clonal HSCs in MPNs interact with other cells in this microenvironment and remodel it allowing further malignant expansion (Figure 5). There is a growing evidence that endothelial cells, mesenchymal stem cells, stromal cells, osteoblasts, and osteoclasts may contribute to the pathogenesis of these diseases in the bone marrow [63].

In a mouse model, endothelial cells expressing JAK2V617F have been shown to be capable of causing the expansion of hematopoietic stem and progenitor cells, which could be caused by increased expression of the cytokines CXCL12 (C-X-C motif chemokine ligand 12) and SCF (stem cell factor) by endothelial cells [64,65].

Mesenchymal stem cells (MSCs) also seem to be important in the pathogenesis of MPNs. In contrast to endothelial cells expressing JAK2V617F, MSCs negative for JAK2V617F have been reported to reduce the expression of CXCL12 and SCF [109,110]. They also support HPSC proliferation [66] and overexpress galectin-1 in all MPN subtypes and galectin-3 in PV patients [67]. Galectins mediate cell adhesion and stimulate cell migration, proliferation and apoptosis through interactions with integrins, laminin and fibronectin. In addition, MSCs promote the expansion of osteoblasts by cell contact and excessive TGF-β1, Notch, IL-6, IL-1β, and TNF-β signaling. Abnormal osteoblasts overproduce inflammatory cytokines, promote fibrogenesis and reduce CXCL12 expression [88]. By contrast, monocytes with JAK2V617F seem to increase osteoclast forming ability in MPN patients, favoring the survival of clonal HSCs [68].

**Figure 5.** Role of the bone marrow microenvironment in the pathogenesis of *Ph*-negative MPNs. Endothelial cells expressing JAK2V617F increase the expression of CXCL12 and stem cell factor (SCF) and cause the expansion of HSCs and progenitor cells. On the other hand, mesenchymal stem cells (MSCs) negative for JAK2V617F show a reduced expression of CXCL12 and SCF. MSCs also overexpress galectin-1 in all MPN subtypes and galectin-3 in PV patients, and promote the expansion of osteoblasts by cell contact and excessive TGF-1β, Notch, IL-6, IL-1β, and TNF-β signaling. Osteoblasts overproduce inflammatory cytokines and reduce CXCL12 expression. By contrast, monocytes with JAK2V617F seem to increase osteoclast forming ability and favor the survival of clonal HSCs. Meanwhile, clonal HSCs produce high levels of IL-1β, which induces nestin-positive MSCs death. Additionally, the sympathetic nerve fibers supporting Schwann cells are reduced in the bone marrow of MPN patients. Regarding the extracellular matrix, MPN patients with JAK2V617F show increased levels of MMP-2 and MMP-9 and patients with primary myelofibrosis (PMF) have increased levels of all LOX family members.

> A recent study has recently found numerous differences between the bone marrow niche of ET and PV patients. In ET, the HSPCs move faster and more frequently towards the endosteal niche and the number of osteoblasts and osteoclasts increases. However, in PV, only the non-endosteal sinusoids are dilated [69]. Other studies have demonstrated that the sympathetic nervous system has a role in the bone marrow niche of MPN patients. Specifically, sympathetic nerve fibers supporting Schwann cells and nestin-positive MSCs are reduced in the bone marrow of MPN patients. In a murine MPN model harboring JAK2V617F, stem cells secreted IL-1β, which induces nestin-positive MSCs death and enables disease expansion [70].

> Regarding the extracellular matrix, several studies have also pointed to a role of matrix metalloproteinases (MMPs) and lysyl oxidase (LOX) in the pathogenesis of MPNs. MPN patients with JAK2V617F show increased levels of MMP-2 and MMP-9 [71] and patients with PMF have increased levels of all LOX family members. LOX is involved in collagen cross-linking and promotes fibrogenesis [72].

#### **4. Disease Modifiers**

Several non-driver somatic mutations have been identified in MPN patients. According to recent studies, more than 80% of patients with PMF [73] and over 50% of PV/ET patients have at least one additional somatic mutation of this type [74]. These mutations occur in genes affecting a wide variety of processes like epigenetic regulation, tumor suppression, transcription regulation or splicing, but also additional signaling pathways (Figure 6). They often modify the course of the disease and the presence of more than one such aberration has been associated with a worse survival [75].

**Figure 6.** Overview of the disease-modifying genes mutated in *Ph*-negative MPNs and their molecular consequences. These mutations occur in genes affecting epigenetic regulation (*ASXL1*, *EZH2*, *DNMT3A*, *TET2*, *IDH1*, and *IDH2*), tumor suppression (*TP53* and *PPM1D*), transcription regulation (*RUNX1* and *NFE2*), splicing (*SRSF2, U2AF1, SF3B1*, and *ZRSR2*), and other signaling pathways (*SH2B3*, *CBL*, *NRAS*/*KRAS*, *PTPN11*). Mutant proteins are depicted in red.

#### *4.1. Epigenetic Regulation*

The most common non-driver somatic mutations affect epigenetic regulation and have been identified in *ASXL1* (ASXL transcriptional regulator 1), *EZH2* (enhancer of zeste polycomb repressive complex 2 subunit), *DNMT3A* (DNA methyltransferase 3 alpha), *TET2* (TET methylcitosine dioxygenase 2), *IDH1* and *IDH2* (isocitrate dehydrogenase NADP+, 1 and 2).

The products of *ASXL1* and *EZH2* are involved in chromatin modification (Figure 6, upper box). Normal ASXL1 interacts with the polycomb repressor complex 2 (PRC2) and enhances its function as methylator of H3K27. H3K27 trimethylation results in the silencing of the *HOXA* gene family which participates in chromatin remodeling. Additionally, ASXL1 interacts with BRCA1-associated protein 1 (BAP1), creating the polycomb group repressive deubiquitinase complex, which globally removes monoubiquitin from H2AK119 and locally at *HOXA* and *IRF8* in HSCs [76,111,112]. *ASXL1* mutations are almost exclusively frameshift and nonsense mutations in exon 12, decrease H3K27 trimethyla-

tion [111] and enhance the activity of the ASXL1-BAP deubiquitinase complex [113]. This causes the deregulated expression of genes critical for HSC self-renewal and differentiation, as well as more open chromatin in c-Kit+ cells. Mutant ASXL1 also binds to the bromodomain-containing protein 4 (BDR4), resulting in the phosphorylation of RNA polymerase II and the acetylation of H3K27 and H3K122, which lead to the upregulation of genes governing myeloid differentiation [76,114]. Another mechanism reported for mutant ASXL1 consists of the repression of TGF-β pathway through H3K and H4K deacetylation [115]. Although normal ASXL1 activates the retinoic acid receptor [116] and interacts with the peroxisome proliferator activated receptor gamma (PPARγ) to repress lipogenesis [117], the effects of *ASXL1* mutations on these mechanisms are still unknown. In summary, the consequences of *ASXL1* mutations are diverse and are not fully elucidated; the mutant protein shows a loss of function in some mechanisms but a gain of function in others.

*EZH2* encodes a histone lysine N-methyltransferase that constitutes the catalytic component of PRC2. The majority of *EZH2* mutations are missense with loss of function effects resulting in the silencing of *HOXA9.* This supports myeloid progenitor self-renewal and leukemic transformation [118,119].

*DNMT3A*, *TET2*, *IDH1* and *IDH2* encode DNA methylation modifiers (Figure 6). *DNMT3A* encodes a *de novo* DNA methyltransferase responsible for DNA methylation at CpG dinucleotides. The mutation most frequently observed is p.R882H, that impairs the CpG specificity, flanking sequence preference and DNMT3A enzymatic activity [120]. Mechanistic studies in mice indicate that mutant DNMT3A decreases PRC2 recruitment at H3K27 favoring accessibility at enhancer chromatin marks and persistent HSC gene expression. JAK2V617F patients also harboring *DNMT3A* mutations show aberrant selfrenewal and altered inflammatory signaling pathways [121].

*TET2* encodes an enzyme that catalyzes the oxidation of 5-methylcytosine (5mC) to 5 hydroxymethylcytosine (5-hmC). Mutations in *TET2* are nonsense or missense changes that lead to a loss of function [122] and DNA hypermethylation due to decreased production of 5-hmC. Mutant TET2 increases the expression of HSC self-renewal genes and sensitizes hematopoietic cells to acquire other mutations and leads to significant myeloid lineage skewing [123] and increased IL-6 production [124]. The order of mutation acquisition can influence the MPN phenotype; mutations in *TET2* arising prior to JAK2V617F favors the ET phenotype, but the acquisition of JAK2V617F in a *TET2* non-mutated background is more likely to result in the PV phenotype [125].

*IDH1* and *IDH2* encode isocitrate dehydrogenases that catalyze decarboxylation of isocitrate into alpha ketoglutarate (α-KG). While IDH1 acts in the cytosol, IDH2 works in the mitochondria. The most common *IDH1* (p.R132H and p.R132C) and *IDH2* mutations (p.R140Q) increase 2-hydroxyglutarate (2-HG) production. 2-HG prevents histone demethylation and the expression of lineage-specific differentiation genes, leading to a block to cell differentiation [126–128]. This compound can also bind ten-eleven translocation (TET) and Jumonji proteins, inhibiting their functions [129]. IDH mutations have also been associated to enhanced aberrant splicing of mutant SRSF2, leading to genomic instability and risk of leukemic transformation [130].

#### *4.2. Tumor Suppression*

*TP53* (tumor protein P53) and *PPM1D* (protein phosphatase, Mg2+/Mn2+ dependent 1D or P53-induced protein phosphatase 1) are involved in tumor suppression (Figure 6). TP53 is a transcription factor that responds to DNA damage inducing transcriptional programs that result in cell cycle arrest or apoptosis [131]. *TP53* mutations are missense changes with several non-mutually exclusive effects: loss of function, gain of function, and dominant-negative effect on normal TP53 [77]. It has been also demonstrated that mutant TP53 increases HSC self-renewal and resistance to cellular stress [132]. There are several upstream regulators of TP53, which are overexpressed in MPNs, such as MDM2 and

MDM4. Both of them inhibit TP53 function by facilitating nuclear export and by inducing its degradation [133].

PPM1D is a serin-threonine phosphatase which negatively regulates TP53 and is transcriptionally upregulated on TP53 induction [134]. Mutations in *PPM1D* are truncating and frameshift changes in exon 6 that lead to a protein that lacks a carboxyterminal degradation domain. This results in altered cell cycle progression, decreased apoptosis and reduced mitochondrial priming [135].

#### *4.3. Regulation of Transcription*

*RUNX1* (RUNX family transcription factor 1) and *NFE2* (nuclear factor, erythroid 2) encode transcription factors and have been also found mutated in MPNs (Figure 6). RUNX1 contains a runt homology domain (RHD) responsible for DNA binding and heterodimerization with core binding factor β (CBF-β). Through this interaction, RUNX1 controls key hematopoietic transcriptional programs. Specifically, RUNX1 participates in hematopoietic differentiation, cell cycle regulation, ribosome biogenesis, and p53 and TGF-β pathways [136]. RUNX1 mutations are missense, frameshift, and non-sense changes that inactivate the protein leading to a reduced myeloid differentiation and an increase in HSC self-renewal [77].

Mutations described in *NFE2* are a 4-amino acid in-frame deletion and frameshift changes that lead to a carboxy-terminally truncated protein [40]. Mutant NFE2 promotes myelopoiesis and causes elevated expression of wild-type NFE2 and histone demethylase JMJD1C maybe by a decreased binding of the repressor HP1α [137].

#### *4.4. Splicing*

Pre-mRNA splicing is catalyzed by the spliceosome, a complex of five snRNPs and multiple proteins. Mutually exclusive mutations in RNA splicing factors encoded by *SRSF2* (serine and arginine rich splicing factor 2), *U2AF1* (U2 small nuclear RNA auxiliary factor 1), *SF3B1* (splicing factor 3b subunit 1) and *ZRSR2* (zinc finger CCCH-type, RNA binding motif and serine/arginine rich 2) have been reported in MPNs (Figure 6, lower box).

SRSF2 contains a ribonucleoprotein with an RNA binding motif and a carboxylterminal serine/arginine rich domain [138], both involved in the recognition and binding to the RNA sequences GGNG and CCNG in exon splicing enhancers (ESEs). The most frequent mutation is p.P95H, that leads to a preferential recognition of CCNG motifs and alters the balance of splicing of multiple pre-mRNAs, which cause downregulation of *EZH2* [139], as well as the mis-splicing of *CASP8,* which activates NF-κB signaling [140]. The expression of mutant SRSF2 has also been demonstrated to cause accumulation of R loops, replication stress and activation of ATR-Chk1 signaling [141,142]. Additionally, mutant SRSF2 seems to predominantly form RUNX1a over RUNX1b and regulate DNA stability [143,144].

U2AF1 recognizes pyrimidine-rich tracts with a conserved terminal AG in 3′ splice sites [145]. The most prevalent somatic mutations affect Q157 and its surroundings; p.Q157 mutants generate mis-splicing of ARID2 and EZH2 [123] and are associated with a worse outcome [146]. Patients can also harbor mutations in serine 34 (p.S34F/Y) that cause different expression and splicing patterns than p.Q157 mutations and have been associated with increased splicing, accumulation of R loops and exon skipping [142,147]. Both types of mutations are located within the CCCH zinc fingers of U2AF1, that are critical for RNA binding [148]. This protein has also been shown to bind to mRNA and repress translation; p.S34F mutation seems to affect the translation of hundreds of mRNAs, but the effect of the other mutations on translation is still unknown [149].

ZRSR2 heterodimerizes with U2AF2 and participates in the recognition of the 3′ splice site. Mutations in this gene are mostly frameshift and nonsense loss-of-function changes that affect splicing and lead to intron retention. Mutant ZRSR2 has been reported to cause increased MAPK and ERBB signaling in myelodysplastic syndromes [150].

SF3B1 forms part of the spliceosome complex. Mutations in *SF3B1* are missense changes (p.K700E and p.H662Q) that cause alternative 3′ splice site selection [151]. These mutations block erythroid maturation [152] and modify the expression of genes involved in RNA processing, cell cycle, heme metabolism and nonsense-mediated RNA decay [77].

#### *4.5. Additional Signaling Pathways*

Finally, other mutations have been found in *SH2B3* (SH2B adaptor protein 3, previously known as *LNK*), *CBL* (CBL proto-oncogene), *NRAS* and *KRAS* (NRAS and KRAS proto-oncogene, GTPase) and *PTPN11* (protein tyrosine phosphatase non-receptor type 11), all of them encoding elements involved in signaling (Figure 6).

SH2B3 is an adaptor protein that interacts with and inhibits signaling through cytokine receptors and kinases such JAK2 [153–155] decreasing the proliferation of hematopoietic cells [156–158]. In addition, this protein can recruit the E3-ubiquitin ligase CBL for degradation of receptors and other molecules [157]. Mutations in *SH2B3* are mainly missense changes that disrupt the negative-feedback loops on growth stimulation [155,157].

CBL recognizes and ubiquitinates activated tyrosine kinase receptors and JAK2 leading to their proteasomal degradation. Mutations in *CBL* are mostly missense changes that reduce the E3 ligase activity and the degradation of its substrates [159–161]. However, they are not merely loss-of-function mutations since *CBL* knockout cells show increased cytokine sensitivity [162].

Missense substitutions affecting *NRAS*/*KRAS* favor the GTP-bound state of RAS, causing a constitutive activation of growth signaling [163].

Finally, *PTPN11* encodes a protein tyrosine phosphatase which dephosphorylates RAS [164]. *PTPN11* mutations increase its phosphatase activity [165], leading to a high dephosphorylation of RAS which increases the activation of RAS-RAF-MEK-ERK pathway [166].

#### **5. Additional Factors Involved in Disease**

There are several factors that have been shown to influence heterogeneity in MPN phenotypes, such as the HSC in which the mutation appears first, genetic background, gender, age, and environmental factors.

HSCs are highly heterogeneous and carry a lineage-bias [167]. It has been demonstrated that the acquisition of a driver mutation in a platelet-biased HSC may drive to an ET phenotype, whereas the PV phenotype is more probable when mutation is acquired in balanced/myeloid-biased HSCs [168].

It is well known that there is an association between the *JAK2* haplotype 46/1 or GGCC and MPNs. This haplotype is found in 24% of the population and in the 56% of MPN patients [169] increasing the susceptibility of developing a *JAK2* mutation, but also to *CALR* mutations and weakly to *MPL* mutations [169,170]. Recent studies have identified several SNPs in different loci which have been associated with an increased risk of developing some MPN subtypes [171–173].

Regarding gender, the ET phenotype has been mostly reported in women, while PV/PMF are more prevalent in men [174,175]. Women seem to have a greater symptom burden than men [175], but the male gender has been associated with a higher likelihood of myelofibrotic transformation in ET patients [176].

The incidence of MPNs also increases with age, and this factor is the strongest predictor of death in PV and ET [177,178]. This phenomenon has been related to the influence of aging in hematopoiesis, maybe due to a greater probability of acquiring somatic mutations in HSCs [78] favored by a pro-inflammatory state due to the accumulation of inflammatory cytokines associated with age [45]. This higher probability would also explain the increased risk of disease progression in these patients [78].

Retrospective observational studies have reported that the occupational exposure to benzene and/or petroleum, prior blood donation (specifically for PV) [179], and smoking [180,181] are associated with a higher risk of MPNs.

#### **6. Conclusions**

The understanding of the pathogenesis of MPNs has undergone a complete revolution in the last 15 years, especially since the p.V617F mutation in *JAK2* was characterized. Since then, MPNs have basically been considered signaling disorders, especially affecting the JAK2/STAT pathway, but also the MAPK/ERK and PI3K/AKT pathways. Further characterization of mutations in *MPL*, and the mechanism by which *CALR* mutations activate TPOR, reinforced this view. However, although the pathogenic mechanisms of the JAK2, TPOR, and CALR mutants seem quite straightforward and simple, various studies have shown that these alterations can cause more complex disturbances in the cell through non-canonical mechanisms. This, together with the characterization of new somatic genetic alterations that affect genes involved in other processes and signaling pathways, have revealed the complexity of the pathogenesis of MPN, which could partly explain the phenotypic heterogeneity observed among patients.

**Author Contributions:** Conceptualization: J.L.V. and A.G.-H.; review of the literature: A.G.-H.; writing—original draft preparation: A.G.-H.; writing—review and editing: J.L.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** J.L.V. and A.G.-H. research are supported by PIUNA 2020 program of the University of Navarra (Code 15058203).

**Acknowledgments:** The authors acknowledge Cristina Hurtado by her technical support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

