*Article* **KYP-2047, an Inhibitor of Prolyl-Oligopeptidase, Reduces GlioBlastoma Proliferation through Angiogenesis and Apoptosis Modulation**

**Sarah Adriana Scuderi <sup>1</sup> , Giovanna Casili <sup>1</sup> , Alessio Ardizzone <sup>1</sup> , Stefano Forte <sup>2</sup> , Lorenzo Colarossi <sup>3</sup> , Serena Sava <sup>3</sup> , Irene Paterniti <sup>1</sup> , Emanuela Esposito 1,\* , Salvatore Cuzzocrea <sup>1</sup> and Michela Campolo <sup>1</sup>**


**Citation:** Scuderi, S.A.; Casili, G.; Ardizzone, A.; Forte, S.; Colarossi, L.; Sava, S.; Paterniti, I.; Esposito, E.; Cuzzocrea, S.; Campolo, M. KYP-2047, an Inhibitor of Prolyl-Oligopeptidase, Reduces GlioBlastoma Proliferation through Angiogenesis and Apoptosis Modulation. *Cancers* **2021**, *13*, 3444. https://doi.org/10.3390/ cancers13143444

Academic Editor: Stanley Stylli

Received: 1 June 2021 Accepted: 7 July 2021 Published: 9 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/).

**Simple Summary:** Glioblastoma (GB) is the most aggressive brain tumor characterized by necrosis, excessive proliferation, and invasiveness. Despite relevant progress in conventional treatments, the survival rate for patients with GB remains low. The present study investigated the potential effect of KYP-2047, an inhibitor of the prolyl-oligopeptidase (POP or PREP), in an in vivo U87-xenograft model and in an in vitro study on human GB cells. This study demonstrated the abilities of KYP-2047 to counteract and reduce GB progression through angiogenesis and apoptosis modulation.

**Abstract:** Glioblastoma (GB) is the most aggressive tumor of the central nervous system (CNS), characterized by excessive proliferation, necrosis and invasiveness. The survival rate for patients with GB still remains low. Angiogenesis and apoptosis play a key role in the development of GB. Thus, the modulation of angiogenesis and apoptosis processes represent a possible strategy to counteract GB progression. This study aimed to investigate the potential effect of KYP-2047, an inhibitor of the prolyl-oligopeptidase (POP), known to modulate angiogenesis, in an in vivo U87-xenograft model and in an in vitro study on human GB cells. Our results showed that KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg was able to reduce tumor burden in the xenograft-model. Moreover, KYP-2047 significantly reduced vascular endothelial-growth-factor (VEGF), angiopoietins (Ang) and endothelial-nitric-oxide synthase (eNOS) expression. In vitro study revealed that KYP-2047 at different concentrations reduced GB cells' viability. Additionally, KYP-2047 at the concentrations of 50 µM and 100 µM was able to increase the pro-apoptotic protein Bax, p53 and caspase-3 expression whereas Bcl-2 expression was reduced. Thus, KYP-2047 could represent a potential therapeutic treatment to counteract or reduce GB progression, thanks its abilities to modulate angiogenesis and apoptosis pathways.

**Keywords:** glioblastoma (GB); prolyl-oligopeptidase (POP); vascular endothelial growth factor (VEGF); transforming growth factor-β (TGF-β); angiopoietin (Ang); endothelial nitric oxide synthase (eNOS)

## **1. Introduction**

Gliomas are the main neoplastic diseases affecting the central nervous system (CNS) [1]. Among gliomas, glioblastoma (GB) is the most common primary malignant tumor of CNS, with an incidence of about 3–4 cases per 100,000 people per year [2]. GB is classified by the World Health Organization (WHO) as grade IV astrocytoma, characterized by poorly

differentiated neoplastic astrocytes with high mitotic activity, necrosis and vascular proliferation [2]. GB occurs more frequently in mature people aged between 45 and 75 years with a higher incidence in men than in women, associated with a poor quality of life [3]. GB is characterized by abnormal angiogenesis, apoptosis alteration and invasiveness [4]. Genome-wide expression studies in glioblastomas revealed that GB is associated with chromosomic alterations which can include deletions, amplifications or mutations which contribute to the development of GB [5]. In addition to genetic risk factors, other risk factors involved in the development of GB have been identified, such as exposure to ionizing radiation, ultraviolet rays, smoke, and pesticides [3,6]. The symptomatology of GB is varied, as it is related to the location and degree of infiltration of the tumor mass. Currently, standard treatment for GB includes surgical removal of the tumor, followed by the concomitant administration of chemotherapeutic agents such as temozolomide (TMZ) and radiotherapy [3]. However, the survival rate for patients with GB still remains low [7]; consequently, the identification of new therapeutic targets and new molecules able to reduce or arrest the progression of GB represents an important goal for cancer research. Many studies have focused on the role of angiogenesis and apoptosis in the development of GB [8,9]. It has been proposed that therapeutic resistance of GB is due to an up-regulation of anti-apoptotic proteins such as Bcl2 and a downregulation of pro-apoptotic proteins, leading to activation of oncogenes that promote tumor cell survival [9]. Moreover, also angiogenesis represents a key event for tumor growth and progression [10]; in fact, it has been demonstrated that several angiogenic factors such as vascular endothelial growth factor (VEGF) and angiopoietins (Ang) are up-regulated in GB that generate highly permeable and functionally immature blood vessels which contribute to tumor growth [8,10]. Recently, different studies have focused on the effect of KYP-2047 [11,12], a specific and potent inhibitor of the prolyl-oligopeptidase (POP or PREP), a serine protease involved in the angiogenesis process [11,12]. POP is present both in the brain and in peripheral tissues; it is involved in the hydrolysis of proline and in many other physiological functions [13]. KYP-2047 demonstrated the ability to modulate the angiogenesis process, but also cell cycle and differentiation [11–13]. Therefore, considering the key roles of angiogenesis and apoptosis in GB pathology, the aim of this study was to investigate the potential effect of KYP-2047 in an in vivo U87-xenograft model and in vitro model on human GB cells to counteract or reduce GB progression.

#### **2. Materials and Methods**

#### *2.1. In Vivo Studies*

#### 2.1.1. Cell Line

The human GB cell line U-87 (U-87MG ATCC® HTB-14™ Homo sapiens brain Likely glioblastomas) was obtained from ATCC (American Type Culture Collection, Rockville, MD, USA). U-87 cells were cultured in 75 cm<sup>2</sup> flask with respectively Dulbecco's modified Eagle's medium (DMEM—Sigma-Aldrich® Catalog No. D5030; St. Louis, MO, USA) supplemented with antibiotics (penicillin 1000 units—streptomycin 0.1 mg/L, Sigma-Aldrich® Catalog No. P4333; St. Louis, MO, USA), L-glutamine (GlutaMAX™, ThermoFisher Scientific® Catalog No. 35050061; Waltham, MA, USA) and 10% (*v*/*v*) fetal bovine serum (FBS, Sigma-Aldrich® Catalog No. 12103C St. Louis, MO, USA) in a humidified atmosphere containing 5% CO<sup>2</sup> at 37 ◦C.

#### 2.1.2. Animals

Wild-type nude male mice C57BL/6J were purchased from Jackson Laboratory (Bar Harbor, Hancock, ME, USA) and housed in microisolator cages under pathogen-free conditions on a 12 h light/12 h dark schedule for a week. Animals were fed a standard diet and water ad libitum. Animal experiments were in compliance with Italian regulations on protection of animals used for experimental and other scientific purposes (DM 116192) as well as European Union (EU) regulations (OJ of EC L 358/1 18 December 1986).

#### 2.1.3. Experimental Design

The Xenograft tumor model was performed as previously described by Deng et al. [14]. The mice were inoculated subcutaneously with 3 <sup>×</sup> <sup>10</sup><sup>6</sup> human glioblastoma U-87 cells in 0.2 mL of phosphate buffered saline (PBS) and 0.1 mL matrigel (BD Bioscience, Bedford, MA, USA). Animals were treated with KYP-2047 at doses of 1 mg/kg, 2.5 mg/kg and 5 mg/kg every three days from day 7. KYP-2047 was dissolved in PBS with 0.001% of dimethyl sulfoxide (DMSO). After tumor cell inoculation, animals were monitored daily for morbidity and mortality [15]. At the thirty-fifth day, the animals were sacrificed and their tumors were excised and processed for analysis. Tumor volumes were measured non-invasively by using an electronic calliper. The tumor burden was calculated using the following formula: 0.5 × length × width. The tumor size was measured every four days for 28 days. The tumor volume was calculated using an empirical formula, V = 1/2 × ((the shortest diameter) 2 × (the longest diameter)). The experiments were performed three times to verify the data, using 25 animals for each experimental group.

Experimental groups:

The mice were randomly divided into four groups, as described below:


Furthermore, the control group + KYP-2047 1 mg/kg was only subjected to histological evaluation, mean tumor burden and mean tumor weight, because it did not induce any beneficial effect; therefore, we decided to continue analyzing only KYP-2047 2.5 mg/kg and 5 mg/kg groups.

#### 2.1.4. Histological Evaluation

Histological evaluation was performed as previously described by Esposito et al. [16]. Tumor samples were fixed with 10% neutral formalin, embedded in paraffin, and sectioned at 7 µm. Sections were deparaffinized with xylene and stained with hematoxylin and eosin. The slides were analyzed by a pathologist blinded to the treatment groups. All sections were analyzed using an Axiovision microscope (Zeiss, Milan, Italy).

#### 2.1.5. Western Blot Analysis

Tumor samples from each mouse were suspended in extraction Buffer A (0.2 mM PMSF, 0.15 mM pepstatin A, 20 mM leupeptin, 1 mM sodium orthovanadate), homogenized at the highest setting for 2 min, and centrifuged at 12,000× *g* rpm for 4 min at 4 ◦C. Supernatants are the cytosolic fraction, whereas the pellets, containing enriched nuclei, were resuspended in Buffer B (1% Triton X-100, 150 mM NaCl, 10 mM TrisHCl pH 7.4, 1 mM EGTA, 1 mM EDTA, 0.2 mM PMSF, 20 mm leupeptin, 0.2 mM sodium orthovanadate) and centrifuged at 12,000× *g* rpm for 10 min at 4 ◦C; supernatants are the nuclear fraction. Protein concentration was estimated by the Bio-Rad protein assay using bovine serum albumin as standard. Then, tumor samples, in equal amounts of protein, were separated on 12% SDS-PAGE gel and transferred to nitrocellulose membrane as previously described [17]. The following primary antibodies were used: anti-vascular endothelial growth factor (VEGF) (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7269); anti-endothelial nitric oxide synthase (eNOS) (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-376751); antiangiopoietin 1 (Ang1) (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-517593); antiangiopoietin 2 (Ang2) (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-74403); anti-Ki-67 (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-23900); anti-Bax (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7480); anti-Bcl2 (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7382). Antibody dilutions were made in PBS/5% *w*/*v* nonfat dried

milk/0.1% Tween-20 (PMT) and membranes incubated overnight at 4 ◦C. Membranes were then incubated with secondary antibody (1:2000, Jackson ImmunoResearch, West Grove, PA, USA) for 1 h at room temperature. To ascertain that those blots were loaded with equal amounts of protein lysate, they were also incubated with β-actin antibody (for cytosolic fraction 1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-8432) or lamin A/C (for nuclear fraction 1:500, Santa Cruz Biotechnology, Dallas, TX, USA; sc-376248). Signals were detected with an enhanced chemiluminescence (ECL) detection system reagent according to the manufacturer's instructions (Thermo Fisher, Waltham, MA, USA). The relative expression of the protein bands was quantified by densitometry with BIORAD ChemiDocTMXRS + software.

#### 2.1.6. Immunohistochemical Localization of Vascular Endothelial-Growth-Factor (VEGF), Endothelial Nitric Oxide Synthase (eNOS), CD34, Ki-67, Bcl2 and Caspase-3

Immunohistochemical localization was performed as previously described by Esposito et al. [16]. Slides were incubated overnight using the following primary antibodies: VEGF (Santa Cruz Biotechnology, Dallas, TX, USA; 1:100 in PBS, *v*/*v*; sc-7269), eNOS (Santa Cruz Biotechnology, Dallas, TX, USA, 1:100 in PBS, *v*/*v*; sc-376751) anti-Bcl2 (1:100; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7382); anti-caspase-3 (1:100, Santa Cruz Biotechnology, Dallas, TX, USA; sc-56053); anti-Ki-67 (1:100; Santa Cruz Biotechnology, Dallas, TX, USA; sc-23900); anti-CD34 (1:100; Santa Cruz Biotechnology, Dallas, TX, USA; sc-74499). At the end of the incubation with the primary antibodies, the sections were abundantly washed with PBS and incubated with a secondary antibody (Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h at room temperature. The reaction was revealed by a chromogenic substrate (brown DAB), and counterstaining with NUCLEAR FAST-RED. The percentage of positive staining was measured using a computerized image analysis system (Leica QWin V3, Cambridge, UK). The images were acquired using an optical microscope (Zeiss, Axio Vision, Feldbach, Schweiz). For immunohistochemistry, the images were shown at a magnification of 20 × (50 µm of the bar scale).

#### 2.1.7. Caspase-3 Activity Measurement

Caspase-3 activity in tumor lysate was measured using a colorimetric Assay Kit (cat#ab39401, Abcam, Cambridge, UK) as suggested by manufacturer's instruction.

#### 2.1.8. RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction (RT-qPCR)

Total RNA of tumor samples was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. RNA isolation was performed as previously described by Weinert et al. [18]. First-strand cDNA obtained from RNA samples was stored at −80 ◦C until use.

The mRNA expression levels of VEGF and eNOS in each sample, was measured using Power Up Sybr Master Mix (Applied Biosystems) and a QuantStudio Flex Real-Time Polymerase Chain Reaction (PCR) System (Applied Biosystems) [19]. The primer used for reverse transcriptase PCR were for VEGF: forward 50 -GAGCAGAAGTCCCATGAAGTGA-30 and reverse 50 -CACAGGACGGCTTGAAGATGT-30 ; eNOS: forward 50 -CCTGTGAGACCTT CTGTGTGG-30 and reverse 50 -GGATCAGACCTGGCAGCAACT-30 . The mRNA expression levels were normalized to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH): forward: 50 -GGGCTGGCATTGCTCTCA-30 , reverse: 50 -TGCTGTAGCGTATTCATTG-30 . Each sample was analyzed in triplicate, and all tests were repeated at least three times.

#### 2.1.9. Enzyme-Linked Immunosorbent Assay (ELISA) Kit

An enzyme-linked immunosorbent assay (ELISA) kit was performed to evaluate PREP expression in serum of each mice using Mouse PREP ELISA kit (cat#Q9QUR6 RayBiotech, Peachtree Corners, GA, USA) as suggested by manufacturer's instructions. The serum of each animal was collected and measured by ELISA kit once a week.

#### 2.1.10. Immunofluorescence Assay

Immunofluorescence staining was performed as previously described by Campolo et al. [20]. Tumor samples were collected and processed for immunofluorescence staining. Tissue sections of 7 µm were incubated with the following primary antibody anti-CD34 at 37 ◦C overnight (1:100; Santa Cruz Biotechnology, Dallas, TX, USA; sc-74499). Then, tissue sections were washed with PBS and incubated with secondary antibody anti-mouse Alexa Fluor-488 antibody (1:1000 *v*/*v*, Molecular Probes, Altrincham, UK) for 1 h at 37 ◦C. For nuclear staining, 40 ,60 -diamidino-2-phenylindole (DAPI; Hoechst, Frankfurt, Germany) (2 µg/mL) in PBS was added. Sections were observed and photographed at 40× magnification using a Leica DM2000 microscope.

#### *2.2. In Vitro Studies*

#### 2.2.1. Cell Lines

U-87 MG (U-87 MG ATCC® HTB-14™ Homo sapiens brain likely glioblastomas), U-138MG (U-138 MG ATCC® HTB-16™ Homo sapiens brain glioblastoma IV grade), A-172 (A-172 ATCC® CRL-1620™ Homo sapiens brain glioblastoma) were obtained from ATCC (American Type Culture Collection, Rockville, MD, USA). The human GB cell lines were seeded in 75 cm<sup>2</sup> flask with respectively Dulbecco's modified Eagle's medium (DMEM—Sigma-Aldrich® Catalog No. D5030; St. Louis, MO, USA) supplemented with antibiotics (penicillin 1000 units—streptomycin 0.1 mg/L, Sigma-Aldrich® Catalog No. P4333; St. Louis, MO, USA), L-glutamine (GlutaMAX™, ThermoFisher Scientific® Catalog No. 35050061; Waltham, MA, USA) and 10% (*v*/*v*) FBS (Sigma-Aldrich® Catalog No. 12103C St. Louis, MO, USA) in a humidified atmosphere containing 5% CO<sup>2</sup> at 37 ◦C.

#### 2.2.2. Cell Treatment

Human GB cells were plated on 96-well plates at a density of 4 <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well to a final volume of 150 µL. After 24 h, GB cells were treated with KYP-2047 (Sigma-Aldrich®) for 24 h at increasing concentrations 0.01 µM, 0.1 µM, 0.5 µM, 1 µM, 10 µM, 30 µM, 50 µM and 100 µM dissolved in PBS.

Experimental Groups:


The experiments were repeated three times to verify the data.

For western blot analysis and immunofluorescence assay on U-87, A-172 and U-138 cells, we decided to continue to analyze only KYP-2047 at the concentrations of 50 µM and 100 µM because represented the most cytotoxic concentrations revealed by MTT assay.

#### 2.2.3. Cell Viability Assay

Cell viability assay on U-87, U-138 and A-172 cells were performed using a mitochondriadependent dye for live cells (tetrazolium dye; MTT) to formazan [20]. GB cells were pre-treated with increasing concentrations of KYP-2047 for 24 h. After 24 h, cells were incubated at 37 ◦C with MTT (0.2 mg/mL) for 1 h. The medium was removed, and the cells lysed with dimethyl sulfoxide (DMSO) (100 µL). The extent of reduction in MTT to formazan was quantified by measurement of optical density (OD) at 550 nm with a microplate rider.

#### 2.2.4. Western Blot Analysis

Western blot analysis on U-87, A-172 and U-138 cell lysates was performed as previously described by Campolo et al. [20]. Human GB cells were washed with ice-cold PBS harvested and resuspended in Tris-HCl 20 mM pH 7.5, NaF 10 mM, 150 µL NaCl, 1% Nonidet P-40 and protease inhibitor cocktail (Roche). After 40 min, cell lysates were centrifuged at 16,000× *g* for 15 min at 4 ◦C. Protein concentration was estimated by the Bio-Rad protein assay using bovine serum albumin as standard. Samples were then heated at 95 ◦C for 5 min and equal amounts of protein separated on a 10–15% SDS-PAGE gel and transferred to a PVDF membrane (Immobilon-P). The membranes were incubated overnight at 4 ◦C with primary antibodies: anti-Bax (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7480); anti-Bcl2 (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-7382); anti-p53 (1:500; Santa Cruz Biotechnology, Dallas, TX, USA; sc-126). To ascertain that blots were loaded with equal amounts of protein lysate, they were also incubated with the antibody β-actin for cytosolic fraction (1:500; Santa Cruz Biotechnology; Dallas, TX, USA. sc-8432) and lamin A/C for nuclear fraction (1:500; Santa Cruz Biotechnology; Dallas, TX, USA, sc-376248). Signals were detected with enhanced chemiluminescence (ECL) detection system reagent according to the manufacturer's instructions (Thermo Fisher, Waltham, MA, USA). The relative expression of the protein bands was quantified by densitometry with BIORAD ChemiDocTMXRS + software.

2.2.5. Immunofluorescence Assay for Transforming Growth Factor-β (TGF-β) and Caspase-3

Immunofluorescence assay was performed on U-87, A-172 and U-138 cells as previously described by Donaldson [21]. GB cells on glass cover slips were rinsed briefly in phosphate-buffered saline (PBS:0.15 M NaCl, 10 mM Na2HPO4, pH 7.4), permeabilized in 0.2% Triton X-100/PBS and blocked with 10% goat serum. The cells were stained overnight (O/N) at 4 ◦C with primary antibodies: anti-transforming growth factor-β (TGFβ, 1:50, Santa Cruz Biotechnology, Dallas, TX, USA; sc-130348) and anti-caspase-3 (1:50, Santa Cruz Biotechnology, Dallas, TX, USA; sc-56053). At the end of the incubation with the primary antibody, the sections were abundantly washed with PBS and incubated with a secondary antibody anti-mouse Alexa Fluor-488 antibody (1:1000 *v*/*v* Molecular Probes, UK) for 1 h at 37 ◦C. Sections were washed in PBS and for nuclear staining 40 ,60 -diamidino-2-phenylindole (DAPI; Hoechst, Frankfurt; Germany) 2 µg/mL in PBS was added. Sections were observed and photographed at 40× magnification using a Leica DM2000 microscope (Leica, Axio Vision, Feldbach, Schweiz). All images were digitalized at a resolution of 8 bits into an array of 2560 × 1920 pixels. Optical sections of fluorescence specimens were obtained using a HeNe laser (543 nm), an ultraviolet (UV) laser (361–365 nm), and an argon laser (458 nm) at a 1 min, 2 s scanning speed with up to eight averages; 1.5 µm sections were obtained using a pinhole of 250. Contrast and brightness were established by examining the most brightly labeled pixels and applying settings that allowed clear visualization of structural details while keeping the highest pixel intensities close to 250.

#### *2.3. Materials*

KYP-2047 and all other chemicals were obtained by Sigma-Aldrich (Milan, Italy). All stock solutions were prepared in non-pyrogenic saline (0.9% NaCl, Baxter, Milan, Italy).

#### *2.4. Statistical Analysis*

All values are expressed as mean ± standard error of the mean (SEM) of "n" observations. Each analysis was performed three times with three samples replicates for each one. The results were analyzed by one-way analysis of variance (ANOVA) followed by a Bonferroni post hoc test for multiple comparisons. A *p*-value of less than 0.05 was considered significant.

#### **3. Results** *3.1. In Vivo Studies*

**3. Results** 

significant.

*2.3. Materials* 

*2.4. Statistical Analysis* 

*3.1. In Vivo Studies* 3.1.1. Effect of KYP-2047 on Tumor Growth

3.1.1. Effect of KYP-2047 on Tumor Growth The histological analysis of the control group (Figure 1A) showed a significant sub-

*Cancers* **2021**, *13*, x 7 of 20

The histological analysis of the control group (Figure 1A) showed a significant subcutaneous tumor mass, associated to an increase in necrosis and neutrophil infiltration; while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg showed a reduction in tumor sections as well as neutrophil infiltration (Figure 1C,D), much more than KYP-2047 at the dose of 1 mg/kg (Figure 1B). Furthermore, we observed a marked reduction of mean tumor burden, tumor volume and tumor weight following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg, much more than KYP-2047 1 mg/kg (Figure 1E–G). Moreover, to better understand if the expression levels of PREP changed during the course of treatment in the tumors, we decided to verify the expression of PREP during the treatment with KYP-2047 by ELISA kit. The results showed that KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg was able to reduce significantly PREP levels particularly from day 14 (Figure 1H). During the course of treatment, no important change in animals' weight was seen (Figure 1I). cutaneous tumor mass, associated to an increase in necrosis and neutrophil infiltration; while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg showed a reduction in tumor sections as well as neutrophil infiltration (Figure 1C,D), much more than KYP-2047 at the dose of 1 mg/kg (Figure 1B). Furthermore, we observed a marked reduction of mean tumor burden, tumor volume and tumor weight following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg, much more than KYP-2047 1 mg/kg (Figure 1E–G). Moreover, to better understand if the expression levels of PREP changed during the course of treatment in the tumors, we decided to verify the expression of PREP during the treatment with KYP-2047 by ELISA kit. The results showed that KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg was able to reduce significantly PREP levels particularly from day 14 (Figure 1H). During the course of treatment, no important change in animals' weight was seen (Figure 1I).

by examining the most brightly labeled pixels and applying settings that allowed clear visualization of structural details while keeping the highest pixel intensities close to 250.

KYP-2047 and all other chemicals were obtained by Sigma-Aldrich (Milan, Italy). All stock solutions were prepared in non-pyrogenic saline (0.9% NaCl, Baxter, Milan, Italy).

All values are expressed as mean ± standard error of the mean (SEM) of "n" observations. Each analysis was performed three times with three samples replicates for each one. The results were analyzed by one-way analysis of variance (ANOVA) followed by a Bonferroni post hoc test for multiple comparisons. A *p*-value of less than 0.05 was considered

**Figure 1.** Effect of KYP-2047 on tumor growth. An elevated tumor mass was observed in the control group (**A**) while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced tumor mass and neutrophil infiltration (**C**,**D**) more than KYP-2047 at dose of 1 mg/kg (**B**). Moreover, the panel (**E**,**F**) showed a reduction in tumor volume and tumor weight respectively following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg without encountering important weight differences (Panel **I**). Additionally, the panel H showed a decrease of PREP expression following KYP-2047 treatment particularly from day 14. Data are representative of at least three independent experiments. Sections were observed and photographed at 10x magnification. (**E**) # *p* < 0.05 vs. CTR; ## *p* < 0.01 vs. CTR; (**F**) ## *p* < 0.01 vs. CTR; ### *p*< 0.001 vs. CTR. (**G**) \*\*\* *p* < 0.001 vs. CTR. (**H**) \*\* *p* < 0.01 vs. CTR; \*\*\* *p* < 0.001 vs. CTR.

#### 3.1.2. Effect of KYP-2047 on Angiogenesis

Angiogenesis is an essential process for tumor growth [22]. GB is characterized by a deregulation of angiogenic growth factors as VEGF and eNOS expression, which play a key role in maintaining vascular homeostasis and vessel integrity [22–24]. Therefore, in this study we decided to investigate by immunohistochemical staining the levels of VEGF and eNOS. Our results demonstrated a significant increase of VEGF and eNOS levels in the control group (Figures 2A and 3A respectively); however, the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced their expression (Figure 2B,C, see immunohistochemistry score Figure 2D; Figure 3B,C, see immunohistochemistry score Figure 3D respectively) in a dose-dependent manner. These results were confirmed also by Western blot analysis and RT-qPCR, showing a significantly reduction of VEGF and eNOS expression in the groups treated with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg compared to control group (Figure 2M, see densitometric analysis Figure 2M1,N and Figure 3E, see densitometric analysis Figure 3E1,F).

Additionally, we evaluated the expression of CD34, a transmembrane glycoprotein involved in the process of newly-forming tumour vessels [25] by immunohistochemistry and immunofluorescence analysis. In this context, our results showed a significant reduction of CD34 expression in the groups treated with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg compared to control group (Figure 2E–G; see immunohistochemistry score Figure 2H) (Figure 2I–K; see CD34 ratio positive cells score Figure 2L). *Cancers* **2021**, *13*, x 9 of 20

**Figure 2.** Effect of KYP-2047 on vascular endothelial-growth-factor (VEGF) and CD34 expression. Immunohistochemical staining showed a marked expression of VEGF and CD34 in the control group (**A**,**E**) whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced their expression (**B**,**C**,**F**,**G**). Sections were observed and photographed at 10×, 20× and 40× magnification The data for VEGF were confirmed also by western blot analysis and quantitative real-time polymerase chain reaction (RT-qPCR), showing a decrease of VEGF expression following KYP-2047 treatment (**M**,**N**). Moreover, the data for CD34 were confirmed also by immunofluorescence assay (**I**,**J**,**K**). Data are representative of at least three independent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**H**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. (**L**) ## *p* < 0.01 vs. CTR; (**M**) ### *p* < 0.001 vs. CTR. (**N**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **Figure 2.** Effect of KYP-2047 on vascular endothelial-growth-factor (VEGF) and CD34 expression. Immunohistochemical staining showed a marked expression of VEGF and CD34 in the control group (**A**,**E**) whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced their expression (**B**,**C**,**F**,**G**). Sections were observed and photographed at 10×, 20× and 40× magnification The data for VEGF were confirmed also by western blot analysis and quantitative real-time polymerase chain reaction (RT-qPCR), showing a decrease of VEGF expression following KYP-2047 treatment (**M**,**N**). Moreover, the data for CD34 were confirmed also by immunofluorescence assay (**I**,**J**,**K**). Data are representative of at least three independent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**H**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. (**L**) ## *p* < 0.01 vs. CTR; (**M**) ### *p* < 0.001 vs. CTR. (**N**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR.

**Figure 3.** Effect of KYP-2047 on endothelial-nitric-oxide synthase (eNOS) expression. Immunohistochemical staining showed a marked expression of eNOS in the control group (**A**) whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced its expression (**B**,**C**). Sections were observed and photographed at 10× magnification. The data were confirmed by Western blot analysis and RT-qPCR, showing a decrease of eNOS expression following KYP-2047 treatment (**E**,**F**). Data are representative of at least three independent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**E**) ### *p* < 0.001 vs. CTR. (**F**) # *p* < 0.05 vs CTR; ## *p* < 0.01 vs. CTR. **Figure 3.** Effect of KYP-2047 on endothelial-nitric-oxide synthase (eNOS) expression. Immunohistochemical staining showed a marked expression of eNOS in the control group (**A**) whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced its expression (**B**,**C**). Sections were observed and photographed at 10× magnification. The data were confirmed by Western blot analysis and RT-qPCR, showing a decrease of eNOS expression following KYP-2047 treatment (**E**,**F**). Data are representative of at least three independent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**E**) ### *p* < 0.001 vs. CTR. (**F**) # *p* < 0.05 vs CTR; ## *p* < 0.01 vs. CTR.

Studies on angiogenesis have emphasized the importance of others angiogenic factors involved in tumor growth such as angiopoietins, in particular angiopoietin 1 (Ang1) and angiopoietin 2 (Ang2), currently proposed as biomarkers of GB [26,27]. Therefore, we detected Ang1 and Ang2 expression by Western blot analysis on tumor samples. Our results showed a significantly decrease of Ang1 and Ang2 levels following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg compared to control group (Figure 4A, see densitometric analysis 4A1; Figure 4B, see densitometric analysis 4B1) in a dose-dependent Studies on angiogenesis have emphasized the importance of others angiogenic factors involved in tumor growth such as angiopoietins, in particular angiopoietin 1 (Ang1) and angiopoietin 2 (Ang2), currently proposed as biomarkers of GB [26,27]. Therefore, we detected Ang1 and Ang2 expression by Western blot analysis on tumor samples. Our results showed a significantly decrease of Ang1 and Ang2 levels following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg compared to control group (Figure 4A, see densitometric analysis 4A1; Figure 4B, see densitometric analysis 4B1) in a dose-dependent manner.

manner. Furthermore, we investigated the role of Ki-67, a nuclear protein associated with tumor proliferation and progression [28,29]. As shown in the Figure 4C, the blot revealed a marked expression of Ki-67 in the control group whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced its expression (see densitometric analysis 4C1). Moreover, Ki-67 was evaluated also by immunohistochemistry assay con-Furthermore, we investigated the role of Ki-67, a nuclear protein associated with tumor proliferation and progression [28,29]. As shown in the Figure 4C, the blot revealed a marked expression of Ki-67 in the control group whereas the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced its expression (see densitometric analysis 4C1). Moreover, Ki-67 was evaluated also by immunohistochemistry assay confirming the results obtained as showed in the Figure 4D–F (see immunohistochemistry score 4G).

firming the results obtained as showed in the Figure 4D–F (see immunohistochemistry

score 4G).

**Figure 4.** Effect of KYP-2047 on Ang1, Ang2 and Ki-67 expression. The blots revealed a significant increase of Ang1 and Ang2 expression in the control group while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced their expression (**A**,**B**). Sections were observed and photographed at 10× magnification Moreover, the panel **C** revealed a significant increase of Ki-67 in the control group while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly decreased its expression. The data for Ki-67 was confirmed also by immunohistochemistry (**D**–**F**). Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) ### *p* < 0.01 vs. CTR; (**C**) ## *p* < 0.01 vs. CTR. (**G**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **Figure 4.** Effect of KYP-2047 on Ang1, Ang2 and Ki-67 expression. The blots revealed a significant increase of Ang1 and Ang2 expression in the control group while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly reduced their expression (**A**,**B**). Sections were observed and photographed at 10× magnification Moreover, the panel (**C**) revealed a significant increase of Ki-67 in the control group while the treatment with KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg significantly decreased its expression. The data for Ki-67 was confirmed also by immunohistochemistry (**D**–**F**). Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) ### *p* < 0.01 vs. CTR; (**C**) ## *p* < 0.01 vs. CTR. (**G**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR.

> 3.1.3. Effect of KYP-2047 on Apoptosis Pathway 3.1.3. Effect of KYP-2047 on Apoptosis Pathway

Considering the key role of apoptosis in GB progression [30], we evaluated the proapoptotic Bax, and anti-apoptotic Bcl2 protein by western blot analysis on tumor samples. The results showed that KYP-2047 was able to increase Bax expression and reduce Bcl2 expression (Figure 5A; see densitometric analysis 5A1; Figure 5B, see densitometric analysis 5B1). Moreover, the ability of KYP-2047 to modulate Bcl2 expression was confirmed by immunohistochemistry as shown in Figure 5C–E (see immunohistochemistry score Figure 5F). Furthermore, we detected caspase-3 levels by immunohistochemistry and by a colorimetric assay kit on tumor samples, showing that KYP2047 at doses of 2.5 and 5 mg/kg significantly increased caspase-3 activity compared to the control group (Figure 6A–C; see immunohistochemistry score Figure 6D,E Considering the key role of apoptosis in GB progression [30], we evaluated the proapoptotic Bax, and anti-apoptotic Bcl2 protein by western blot analysis on tumor samples. The results showed that KYP-2047 was able to increase Bax expression and reduce Bcl2 expression (Figure 5A; see densitometric analysis 5A1; Figure 5B, see densitometric analysis 5B1). Moreover, the ability of KYP-2047 to modulate Bcl2 expression was confirmed by immunohistochemistry as shown in Figure 5C–E (see immunohistochemistry score Figure 5F). Furthermore, we detected caspase-3 levels by immunohistochemistry and by a colorimetric assay kit on tumor samples, showing that KYP2047 at doses of 2.5 and 5 mg/kg significantly increased caspase-3 activity compared to the control group (Figure 6A–C; see immunohistochemistry score Figure 6D,E

**Figure 5.** Effect of KYP-2047 on apoptosis pathway in the U87-xenograft model. The blots revealed an increase of proapoptotic Bax expression and a decrease of Bcl2 expression following KYP-2047 treatment compared to control group (**A**) (**B**). Additionally, immunohistochemistry staining confirmed a decrease of Bcl2 expression after KYP-2047 treatment. (**C**– **F**). Sections were observed and photographed at 10× magnification. Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**F**) ### *p* < 0.001 vs. CTR. **Figure 5.** Effect of KYP-2047 on apoptosis pathway in the U87-xenograft model. The blots revealed an increase of proapoptotic Bax expression and a decrease of Bcl2 expression following KYP-2047 treatment compared to control group (**A**,**B**). Additionally, immunohistochemistry staining confirmed a decrease of Bcl2 expression after KYP-2047 treatment. (**C**–**F**). Sections were observed and photographed at 10× magnification. Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**F**) ### *p* < 0.001 vs. CTR. **Figure 5.** Effect of KYP-2047 on apoptosis pathway in the U87-xenograft model. The blots revealed an increase of proapoptotic Bax expression and a decrease of Bcl2 expression following KYP-2047 treatment compared to control group (**A**) (**B**). Additionally, immunohistochemistry staining confirmed a decrease of Bcl2 expression after KYP-2047 treatment. (**C**– **F**). Sections were observed and photographed at 10× magnification. Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**F**) ### *p* < 0.001 vs. CTR.

**Figure 6.** Effect of KYP-2047 on caspase-3 expression. Immunohistochemistry assay revealed an increase of cleavedcaspase-3 expression following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg compared to control group (**A**–**C**). Sections were observed and photographed at 10× magnification. Additionally, the data for caspase-3 were confirmed also by a colorimetric assay kit as shown in the panel E. Data are representative of at least three independent experiments. (**D**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**E**) ## *p* <0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **Figure 6.** Effect of KYP-2047 on caspase-3 expression. Immunohistochemistry assay revealed an increase of cleavedcaspase-3 expression following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg compared to control group (**A**–**C**). Sections were observed and photographed at 10× magnification. Additionally, the data for caspase-3 were confirmed also by a colorimetric assay kit as shown in the panel E. Data are representative of at least three independent experiments. (**D**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**E**) ## *p* <0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **Figure 6.** Effect of KYP-2047 on caspase-3 expression. Immunohistochemistry assay revealed an increase of cleaved-caspase-3 expression following KYP-2047 treatment at doses of 2.5 mg/kg and 5 mg/kg compared to control group (**A**–**C**). Sections were observed and photographed at 10× magnification. Additionally, the data for caspase-3 were confirmed also by a colorimetric assay kit as shown in the panel E. Data are representative of at least three independent experiments. (**D**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**E**) ## *p* <0.01 vs. CTR; ### *p* < 0.001 vs. CTR.

#### *3.2. In Vitro Studies 3.2. In Vitro Studies*

3.2.1. Effect of KYP-2047 on Cell Viability 3.2.1. Effect of KYP-2047 on Cell Viability

KYP-2047 cytotoxicity was evaluated incubating U-87, A-172 and U-138 cells with growing concentrations of KYP-2047 (0.01 µM, 0.1 µM, 0.5 µM, 1 µM, 10 µM, 30 µM, 50 µM and 100 µM) for 24 h. KYP-2047 treatment showed a significant decrease of cell viability in all three cell lines in a concentration dependent-manner as shown in the Figure 7A–C. Therefore, based on MTT results, we decided to continue testing for other analysis only KYP-2047 at concentrations of 50 µM and 100 µM on U-87, A172 and U138 cells because they represented the most cytotoxic concentrations. KYP-2047 cytotoxicity was evaluated incubating U-87, A-172 and U-138 cells with growing concentrations of KYP-2047 (0.01 µM, 0.1 µM, 0.5 µM, 1 µM, 10 µM, 30 µM, 50 µM and 100 µM) for 24 h. KYP-2047 treatment showed a significant decrease of cell viability in all three cell lines in a concentration dependent-manner as shown in the Figure 7A–C. Therefore, based on MTT results, we decided to continue testing for other analysis only KYP-2047 at concentrations of 50 µM and 100 µM on U-87, A172 and U138 cells because they represented the most cytotoxic concentrations.

**Figure 7.** Effect of KYP-2047 on U-87, U-138 and A-172 cell viability. Cell viability was evaluated using MTT assay 24 h after KYP-2047 treatment at the concentrations of 0.01 µM, 0.1 µM, 0.5 µM, 1 µM, 10 µM, 30 µM, 50 µM and 100 µM. U-87, U-138 and A-172 cells showed a similar decrease of cell viability following KYP-2047 treatment in a concentrationdependent manner (**A**–**C**). Data are representative of at least three independent experiments. **Figure 7.** Effect of KYP-2047 on U-87, U-138 and A-172 cell viability. Cell viability was evaluated using MTT assay 24 h after KYP-2047 treatment at the concentrations of 0.01 µM, 0.1 µM, 0.5 µM, 1 µM, 10 µM, 30 µM, 50 µM and 100 µM. U-87, U-138 and A-172 cells showed a similar decrease of cell viability following KYP-2047 treatment in a concentration-dependent manner (**A**–**C**). Data are representative of at least three independent experiments.

#### 3.2.2. Effect of KYP-2047 on Apoptosis Pathway 3.2.2. Effect of KYP-2047 on Apoptosis Pathway

Apoptosis plays a key role in the development of cancer including GB [31]. Deregulation of apoptotic process is a relevant hallmark of a tumor [31], responsible not only for its progression but also for tumor resistance to therapies [32]. Therefore, we investigated the effect of KYP-2047 on the apoptosis pathway in U-87, A-172 and U-138 cell lysates evaluating the pro-apoptotic Bax, tumor suppressor p53 and anti-apoptotic Bcl2 protein by Western blot analysis (The original Western blot can be found in Figure S5). Our results revealed an increase of Bax and p53 levels following KYP-2047 treatment in U87 cell lysates performed for 24 h at the concentrations of 50 µM and 100 µM compared to control group (Figure 8A, see densitometric analysis 8A1; Figure 8B, see densitometric analysis 8B1, respectively); while Bcl2 expression was significantly reduced following KYP-2047 treatment compared to control group (Figure 8C; see densitometric analysis 8C1) in a concentration-dependent manner. The same results appear for A-172 and U-138 cell lysates, confirming an increase of pro-apoptotic Bax and p53 expression following KYP-2047 treatment compared to control group (Figure S1A, see densitometric analysis 1A1; Figure S1B, see densitometric analysis 1B1, respectively) (Figure S2A, see densitometric analysis 2A1; Figure S2B, see densitometric analysis 2B1, respectively) and a decrease of anti-apoptotic Apoptosis plays a key role in the development of cancer including GB [31]. Deregulation of apoptotic process is a relevant hallmark of a tumor [31], responsible not only for its progression but also for tumor resistance to therapies [32]. Therefore, we investigated the effect of KYP-2047 on the apoptosis pathway in U-87, A-172 and U-138 cell lysates evaluating the pro-apoptotic Bax, tumor suppressor p53 and anti-apoptotic Bcl2 protein by Western blot analysis (The original Western blot can be found in Figure S5). Our results revealed an increase of Bax and p53 levels following KYP-2047 treatment in U87 cell lysates performed for 24 h at the concentrations of 50 µM and 100 µM compared to control group (Figure 8A, see densitometric analysis 8A1; Figure 8B, see densitometric analysis 8B1, respectively); while Bcl2 expression was significantly reduced following KYP-2047 treatment compared to control group (Figure 8C; see densitometric analysis 8C1) in a concentration-dependent manner. The same results appear for A-172 and U-138 cell lysates, confirming an increase of pro-apoptotic Bax and p53 expression following KYP-2047 treatment compared to control group (Figure S1A, see densitometric analysis 1A1; Figure S1B, see densitometric analysis 1B1, respectively) (Figure S2A, see densitometric analysis 2A1; Figure S2B, see densitometric analysis 2B1, respectively) and a decrease of anti-apoptotic Bcl2 protein expression (Figure S1C; see densitometric analysis 1C) (Figure S2C; see densitometric analysis 2C1).

sitometric analysis 2C1).

Bcl2 protein expression (Figure S1C; see densitometric analysis 1C) (Figure S2C; see den-

**Figure 8.** Effect of KYP-2047 on apoptosis pathway in U-87 cell lysates. The blots on U87 cell lysates revealed an increase of pro-apoptotic Bax and p53 expression following KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to control group (**A**,**B**). Moreover, KYP-2047 at the concentrations of 50 µM and 100 µM reduced significantly Bcl2 expression compared to control group (**C**). Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) # *p* < 0.05 vs. CTR; ## *p* < 0.01 vs. CTR; (**C**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **Figure 8.** Effect of KYP-2047 on apoptosis pathway in U-87 cell lysates. The blots on U87 cell lysates revealed an increase of pro-apoptotic Bax and p53 expression following KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to control group (**A**,**B**). Moreover, KYP-2047 at the concentrations of 50 µM and 100 µM reduced significantly Bcl2 expression compared to control group (**C**). Data are representative of at least three independent experiments. (**A**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR; (**B**) # *p* < 0.05 vs. CTR; ## *p* < 0.01 vs. CTR; (**C**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR.

3.2.3. Effect of KYP-2047 on TGF-β and Caspase-3 Expression by Immunofluorescence Assay 3.2.3. Effect of KYP-2047 on TGF-β and Caspase-3 Expression by Immunofluorescence Assay

Current studies have focused on the role of TGF-β in the tumor microenvironment suggesting that it plays a key role for GB progression [33,34]. Therefore, we investigated TGF-β expression by immunofluorescence assay on U-87, A-172 and U-138 cell lines. Our results confirmed a significant reduction of TGF-β expression after KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to the control group in U-87 cells (Figures 9A–C, see TGF-β ratio positive cells score Figure 9D), as well as in A-172 and U-138 cell lines (Figure S3A–C; see TGF-β ratio positive cells score 3D); (Figure S4A–C; see TGFβ ratio positive cells score 4D). Current studies have focused on the role of TGF-β in the tumor microenvironment suggesting that it plays a key role for GB progression [33,34]. Therefore, we investigated TGF-β expression by immunofluorescence assay on U-87, A-172 and U-138 cell lines. Our results confirmed a significant reduction of TGF-β expression after KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to the control group in U-87 cells (Figure 9A–C, see TGF-β ratio positive cells score Figure 9D), as well as in A-172 and U-138 cell lines (Figure S3A–C; see TGF-β ratio positive cells score 3D); (Figure S4A–C; see TGF-β ratio positive cells score 4D).

In addition to the regulation of the cell cycle and differentiation, TGF-β is able to induce apoptosis [35] promoting the activation of pro-apoptotic caspase-3, a member of the cysteine-aspartic acid protease family [36]. Thus, in this study we detected caspase-3 expression by immunofluorescence assay in all three GB cell lines. The results obtained showed an increase of caspase-3 expression following KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to the control group in U-87 cells (Figure 9E–G, see caspase-3 ratio positive cells score 9H) as well as in A-172 and U-138 cell lines in a concentration-dependent manner (Figure S3E–G; see caspase-3 ratio positive cells score 3H); (Figure S4E–G; see caspase-3 ratio positive cells score 4H). In addition to the regulation of the cell cycle and differentiation, TGF-β is able to induce apoptosis [35] promoting the activation of pro-apoptotic caspase-3, a member of the cysteine-aspartic acid protease family [36]. Thus, in this study we detected caspase-3 expression by immunofluorescence assay in all three GB cell lines. The results obtained showed an increase of caspase-3 expression following KYP-2047 treatment at the concentrations of 50 µM and 100 µM compared to the control group in U-87 cells (Figure 9E–G, see caspase-3 ratio positive cells score 9H) as well as in A-172 and U-138 cell lines in a concentration-dependent manner (Figure S3E–G; see caspase-3 ratio positive cells score 3H); (Figure S4E–G; see caspase-3 ratio positive cells score 4H).

**Figure 9.** Effect of KYP-2047 on transforming growth factor-β (TGF-β) and caspase-3 expression in U-87 cells. Immunofluorescence assay performed on U-87 cells revealed a marked expression of TGF-β in the control group (**A**), while the treatment with KYP-2047 at the concentrations of 50 µM and 100 µM reduced significantly TGF-β expression (**B**,**C**). Additionally, immunofluorescence staining showed an increase of caspase-3 levels in the groups treated with KYP-2047 at the **Figure 9.** Effect of KYP-2047 on transforming growth factor-β (TGF-β) and caspase-3 expression in U-87 cells. Immunofluorescence assay performed on U-87 cells revealed a marked expression of TGF-β in the control group (**A**), while the treatment with KYP-2047 at the concentrations of 50 µM and 100 µM reduced significantly TGF-β expression (**B**,**C**). Additionally, immunofluorescence staining showed an increase of caspase-3 levels in the groups treated with KYP-2047 at the concentrations of 50 µM and 100 µM (**F**,**G**) compared to control group (**E**). Data are representative of at least three independent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**H**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR.

#### concentrations of 50 µM and 100 µM (**F**,**G**) compared to control group (**E**). Data are representative of at least three inde-**4. Discussion**

pendent experiments. (**D**) ### *p* < 0.001 vs. CTR; (**H**) ## *p* < 0.01 vs. CTR; ### *p* < 0.001 vs. CTR. **4. Discussion**  Glioblastoma (GB) is the most common and aggressive primary brain tumor in adults [37]. GB arise from glial cells but can also develop from astrocytic or neural stem/progenitor cells [38]. GB can be classified into primary and secondary subtypes, based on preexisting lesion [38]. The primary GB subtype develops rapidly de novo in elderly patients without clinical or histologic evidence, whereas the secondary subtype develops from evolution of low-grade astrocytic tumours over the course of 4–5 years [38]. In the last decade, many studies have focused on the role of genetic mutations which contribute to GB initiation as TP53 and isocitrate dehydrogenase (IDH) mutations [37,39]. GB is characterized by a high degree of invasiveness, cell proliferation, angiogenesis and apoptosis alteration [2]. Despite scientific advances, the survival rate for patients with GB remains low and additional therapies are needed [12]. Previous studies have demonstrated that angiogenesis and apoptosis play a key role in GB pathology promoting cell survival and proliferation [10,31]. Therefore, the modulation of angiogenesis and apoptosis processes represent a valid strategy to counteract or reduce GB progression. KYP-2047 (4-phenylbutanoyl-L-prolyl-2(S)-cyanopyrrolidine) was developed as a highly specific and potent POP (or PREP) inhibitor, a serine protease involved in the angiogenesis process [40]. Recent studies revealed that KYP-2047 was able to modulate not only angiogenesis [11] but Glioblastoma (GB) is the most common and aggressive primary brain tumor in adults [37]. GB arise from glial cells but can also develop from astrocytic or neural stem/progenitor cells [38]. GB can be classified into primary and secondary subtypes, based on pre-existing lesion [38]. The primary GB subtype develops rapidly de novo in elderly patients without clinical or histologic evidence, whereas the secondary subtype develops from evolution of low-grade astrocytic tumours over the course of 4–5 years [38]. In the last decade, many studies have focused on the role of genetic mutations which contribute to GB initiation as TP53 and isocitrate dehydrogenase (IDH) mutations [37,39]. GB is characterized by a high degree of invasiveness, cell proliferation, angiogenesis and apoptosis alteration [2]. Despite scientific advances, the survival rate for patients with GB remains low and additional therapies are needed [12]. Previous studies have demonstrated that angiogenesis and apoptosis play a key role in GB pathology promoting cell survival and proliferation [10,31]. Therefore, the modulation of angiogenesis and apoptosis processes represent a valid strategy to counteract or reduce GB progression. KYP-2047 (4-phenylbutanoyl-L-prolyl-2(S)-cyanopyrrolidine) was developed as a highly specific and potent POP (or PREP) inhibitor, a serine protease involved in the angiogenesis process [40]. Recent studies revealed that KYP-2047 was able to modulate not only angiogenesis [11] but also cell cycle and differentiation [12,13]. Therefore, in this study we investigated the potential effect of KYP-2047 on angiogenesis and apoptosis pathways in an in vivo U87-xenograft model and in vitro study on the human GB cell line.

also cell cycle and differentiation [12,13]. Therefore, in this study we investigated the potential effect of KYP-2047 on angiogenesis and apoptosis pathways in an in vivo U87-xenograft model and in vitro study on the human GB cell line. Firstly, we evaluated the ability of KYP-2047 to inhibit tumor growth in the xenograft model. Our results showed a high-grade necrosis and neutrophil infiltration in the control Firstly, we evaluated the ability of KYP-2047 to inhibit tumor growth in the xenograft model. Our results showed a high-grade necrosis and neutrophil infiltration in the control group, while KYP-2047 at higher doses significantly reduced subcutaneous tumor mass as well as neutrophil infiltration. Moreover, KYP-2047 significantly decreased mean tumor burden and tumor weight at higher doses, without encountering important weight differences.

group, while KYP-2047 at higher doses significantly reduced subcutaneous tumor mass Interestingly, treatment with KYP-2047 was able to reduce. PREP levels in serum of animals, particularly from day 14.

GB is one of the most highly angiogenic solid tumor [41]. Its tumor vasculature is both structurally and functionally abnormal, characterized by a dense network of vessels tortuous with increased diameter and thickened basement membranes [41]. Thus, angiogenesis is considered as a pathologic hallmark of GB, leading to VEGF activation, an

angiogenic growth factor that promotes glioblastoma proliferation and CD34 activation, a transmembrane glycoprotein involved in the process of newly-forming tumour vessels [25,42]. Therefore, in this study we investigated VEGF and CD34 expression, showing that KYP-2047 at higher doses was able to reduce their expression significantly compared to the control group. Moreover, we investigated the role of eNOS, a relevant endothelial enzyme that modulates vascular homeostasis and vessel integrity [24]. In this context, our results showed that the control group was characterized by an increase of eNOS expression, whereas KYP-2047 significantly reduced eNOS expression.

The formation of new blood vessels is an essential process for GB growth [22]. In addition to VEGF, this process requires the involvement of other angiogenic factors as the angiopoietins, in particular angiopoietin 1 (Ang1) and angiopoietin 2 (Ang2) which have similar functions [26,43]. Previous studies revealed that Ang1 and Ang2 regulate vascular development and remodelling, promoting tumor growth [43,44]. Therefore, we decided to investigate the expression of Ang1 and Ang2 in GB; our results showed that the control group was characterized by an increase of Ang1 and Ang2 expression, while the treatment with KYP-2047 was able to significantly reduce their expression, inhibiting GB proliferation.

An increased vascularization provides to the tumor cells more oxygen and nutrients, promoting metastatic spread and cell proliferation [22]. In this context, Mastronardi et al. evaluated the correlation between angiogenesis and proliferation processes, through Ki-67 evaluation, a nuclear protein that regulates the cell cycle and differentiation [45]. Ki-67 is considered a relevant marker of tumor proliferation in GB [45,46]. Thus, we decided to evaluate Ki-67 expression, demonstrating that the control group was characterized by an increase of Ki-67 level, while KYP-2047 treatment was able to significantly reduce its expression.

Moreover, considering the key role of apoptosis in GB progression [3], we decided to investigate Bax, Bcl2 and caspase-3 expression in the U87-xenograft model, showing that KYP-2047 at doses of 2.5 mg/kg and 5 mg/kg was able to increase pro-apoptotic Bax and caspase-3 expression while Bcl2 expression was significantly reduced following KYP-2047 treatment.

To confirm the promising results obtained by an in vivo U87-xenograft model, we decided to conduct an in vitro model of GB on U-87, A-172 and U-138 cell lines. Firstly, we evaluated the cytotoxicity of KYP-2047 at different concentrations on U-87, A-172 and U-138 GB cells, demonstrating that KYP-2047 was able to significantly reduce cell viability in all three GB cell lines in a concentration-dependent manner.

Previous studies revealed that also apoptosis plays a key role in the development of GB [31,47]. It has been demonstrated that a down-regulation of apoptosis is associated with tumor survival [31,47]. Therefore, in this study we decided to evaluate the effect of KYP-2047 on the apoptosis pathway by evaluating protein levels of pro-apoptotic Bax, p53 and anti-apoptotic Bcl2 on U-87, A-172 and U-138 cell lysates. Our results revealed that KYP-2047 reduced Bcl2 expression, while Bax and p53 expression were significantly increased following KYP-2047 treatment in a concentration-dependent manner in all three GB cell lines, confirming apoptosis modulation.

Tumor proliferation is associated with an increase of TGF-β expression [48].

TGF-β regulates cell differentiation and apoptosis, promoting caspase-3 activation, a key regulator in apoptotic pathway [36]. Thus, we investigated the expression of TGF-β and caspase-3 in the in vitro model [49].

The results showed an increase of TGF-β expression in the control group while KYP-2047 treatment significantly reduced its expression in U-87, A-172 and U-138 cell lines. In addition, a marked increase of pro-apoptotic caspase-3 expression was revealed following KYP-2047 treatment, highlighting the ability of KYP-2047 to modulate apoptosis in all three GB cell lines.

Thus, the results obtained in an in vivo xenograft model and in an in vitro study on human GB cell lines revealed that KYP-2047 was able to reduce GB progression and growth by modulating angiogenesis and apoptosis pathways. Therefore, KYP-2047 could be considered as an alternative therapeutic strategy to counteract or reduce GB progression.

#### **5. Conclusions**

The data obtained revealed the ability of KYP-2047 to modulate angiogenesis and apoptosis pathways in an in vivo xenograft model and in an in vitro model of GB, reducing tumor progression. Therefore, on the basis of these results, KYP-2047 could represent an available strategy for the treatment of GB.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/cancers13143444/s1, Figure S1: Effect of KYP-2047 on apoptosis pathway in A-172 cell lysates., Figure S2: Effect of KYP-2047 on apoptosis pathway in U-138 cell lysates., Figure S3: Effect of KYP-2047 on TGF-β and Caspase3 expression in A-172 cells., Figure S4: Effect of KYP-2047 on TGF-β and Caspase3 expression in U-138 cells., Figure S5: Original western blot.

**Author Contributions:** S.A.S. prepared the manuscript; S.A.S., G.C., and A.A. performed experiments; S.F., L.C., and S.S. carried out formal analysis; I.P., E.E., and S.C. planned the experiments and critically revised the manuscript. M.C. supervised the research and revised the manuscript. All authors have read and agreed to the published version of the manuscript.

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

**Institutional Review Board Statement:** This study was approved by the University of Messina Review Board for the care of animals, in compliance with Italian regulations on protection of animals (n◦ 368/2019-PR released on 14 May 2019). Animal care was in accordance with Italian regulations on the use of animals for the experiment (D.M.116192) as well as with the Council Regulation regulations (EEC) (O.J. of E.C. L 358/1 12/18/1986).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** All data generated or analyzed during this study are included in this article.

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

#### **References**


## *Review* **FAM72, Glioblastoma Multiforme (GBM) and Beyond**

**Nguyen Thi Thanh Ho 1,† , Chinmay Satish Rahane 2,†, Subrata Pramanik <sup>3</sup> , Pok-Son Kim <sup>4</sup> , Arne Kutzner <sup>5</sup> and Klaus Heese 1,\***


**Simple Summary:** Glioblastoma multiforme (GBM) is a serious and aggressive cancer disease that has not allowed scientists to rest for decades. In this review, we consider the new gene pair |-SRGAP2–FAM72-| and discuss its role in the cell cycle and the possibility of defining new therapeutic approaches for the treatment of GBM and other cancers via this gene pair |-SRGAP2–FAM72-|.

**Abstract:** Neural stem cells (NSCs) offer great potential for regenerative medicine due to their excellent ability to differentiate into various specialized cell types of the brain. In the central nervous system (CNS), NSC renewal and differentiation are under strict control by the regulation of the pivotal SLIT-ROBO Rho GTPase activating protein 2 (SRGAP2)—Family with sequence similarity 72 (FAM72) master gene (i.e., |-SRGAP2–FAM72-|) via a divergent gene transcription activation mechanism. If the gene transcription control unit (i.e., the intergenic region of the two sub-gene units, SRGAP2 and FAM72) gets out of control, NSCs may transform into cancer stem cells and generate brain tumor cells responsible for brain cancer such as glioblastoma multiforme (GBM). Here, we discuss the surveillance of this |-SRGAP2–FAM72-| master gene and its role in GBM, and also in light of FAM72 for diagnosing various types of cancers outside of the CNS.

**Keywords:** brain cancer; cell cycle; differentiation; glioblastoma; proliferation; RAS; SRGAP2; stem cell; TP53

## **1. Introduction**

The human brain is a unique organ that can perform higher cognitive functions and is therefore different from all other species. Its uniqueness is reflected in the expression of four paralog gene pairs |-SRGAP2–FAM72-| (A–D) [1,2]. FAM72 is active in proliferating neural stem cells (NSCs) found in the brain hippocampus [1–5]. There are four specific FAM72 (A–D) paralogs associated with four respective SRGAP2 paralogs on human chromosome 1 (chr 1), but only one such gene pair co-exists as the |-SRGAP2–FAM72-| master gene in all other notochord containing vertebrates (Figure 1a) [1,2,6,7].

**Citation:** Ho, N.T.T.; Rahane, C.S.; Pramanik, S.; Kim, P.-S.; Kutzner, A.; Heese, K. FAM72, Glioblastoma Multiforme (GBM) and Beyond. *Cancers* **2021**, *13*, 1025. https:// doi.org/10.3390/cancers13051025

Academic Editors: Stanley Stylli and Giulio Cabrini

Received: 25 January 2021 Accepted: 22 February 2021 Published: 1 March 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/).

**Figure 1.** Overview scheme of the |-SRGAP2–FAM72-| master gene expression across the phylogenetic tree. (**a**) While humans express four master genes on chr 1, all other notochord containing vertebrates carry only one such master gene; other species do not show any such master gene, and thus far, no species have been found that show two or three such master genes. FAM72 shows four exons (149 amino acids (aa)), SRGAP2 is composed of 22 exons (1071 aa), and both sub-genes are separated by a 4-kbp intergenic region (IGR). The four paralogous gene pairs A–D are located on opposite strands from one another [1,2,5]. (**b**) Simplified divergent gene transcription paradigm scheme of the novel pivotal |-SRGAP2–FAM72-| master gene in the brain. The |-SRGAP2–FAM72-| master gene resides within a nucleosome-depleted region with the IGR (blue), containing potential transcription factor (TF)-binding sites (BS) (TFBS) between the SRGAP2 (red) and FAM72 (green) genes indicated. Reverse-oriented SRGAP2 (red) and FAM72 (green) genes are expressed from opposite DNA strands [1,8]. The dual IGR promotor controls the two reverse-oriented reciprocal functional-dependent genes FAM72 and SRGAP2, respectively, located on opposite DNA strands. If FAM72 gene is activated by TFs, then the transcription of the SRGAP2 gene is activated until it is actively terminated early and vice versa for neuronal differentiation; accordingly, if FAM72 is in the 'on' modus, SRGAP2 is switched off and vice versa [8–11]. Through this mechanism, FAM72 maintains renewal and proliferation of a critical mass of NSCs during brain development while SRGAP2 promotes escape of the cell cycle fostering neuronal differentiation and brain plasticity [5,8]. This structure represents a novel paradigm for controlling the transcription of divergent genes in regulating NSC gene expression and may allow for novel therapeutic approaches to restore or improve higher cognitive functions and cure cancers (Figure 2).

**Figure 2.** Overview of the |-SRGAP2–FAM72-| master gene expression in GBM [12]. As long as FAM72 remains in the on modus, NSCs keep proliferating. For neuronal differentiation and brain plasticity, FAM72 needs to be switched off to allow SRGAP2 activation and brain development. The activity of the |-SRGAP2–FAM72-| master gene expression during glia cell differentiation is less clear. Since glia cells have the capacity to proliferate, FAM72 might be switched on or off [13–15]. Eventually, mutations in GBM-specific driver genes: epidermal growth factor receptor (EGFR), tumor protein p53 (TP53), phosphatase and tensin homolog (PTEN), neurofibromin 1 (NF1), spectrin alpha, erythrocytic 1 (SPTA1) and phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) or sodium voltage-gated channel alpha subunit 9 (SCN9A), matrix remodeling associated 5 (MXRA5), a disintegrin and metalloprotease domain 29 (ADAM29), kinase insert domain receptor (KDR), phosphatidylinositol-4-phosphate 3-kinase catalytic subunit type 2 gamma (PIK3C2G), and low-density lipoprotein receptor related protein 1B (LRP1B) induce NSC transformation into cancer stem cells (CSCs) while FAM72 is still in the on modus [12].

#### **2. Physiological Function of the |-SRGAP2–FAM72-| Master Gene**

Endogenous FAM72 expression has been shown in the hippocampal dentate gyrus [3], where the |-SRGAP2–FAM72-| master-gene regulates NSC renewal, neurogenesis and brain plasticity [4,5]. Here, the |-SRGAP2–FAM72-| master-gene is under a divergent gene expression control (Figure 1b) [5,8]. Thus, FAM72 expression is switched on to promote NSC renewal and proliferation and is switched off (concomitantly SRGAP2 is switched on) to foster differentiation, neuritogenesis, synaptic plasticity, and brain development (Figure 1b) [4,5,8,16–21]. However, this divergent expression paradigm is currently restricted to neural tissue [5,8] and apoptosis is induced if it gets out of control (i.e., neuronal expression of FAM72 forces reentry into the cell cycle) [3].

#### **3. Pathophysiological Function of the |-SRGAP2–FAM72-| Master Gene—FAM72 Expression in Various Types of Cancer**

Early studies revealed that FAM72 was overexpressed outside the nervous system in various types of cancer with the protein kinase C signaling pathway activated in neuroblastoma and breast adenocarcinoma (e.g., MCF-7 and MDA-MB-231 cells) [3] and uracil DNA glycosylase-2 as a binding partner in malignant colon cancers [22]. FAM72B was identified as a member of a 7-gene signature in prostate cancer [23], and it was also upregulated in multiple non-neuronal tissues as well [12]. FAM72B, C, and D were also among the highly upregulated genes in B-cell lymphoma [24]. Recently, FAM72D has been identified as a specific proliferation marker in multiple myelomas [25]. Moreover, we reported increased mean expression of FAM72 paralogs across human tumors compared to control tissues, except in cases of skin cutaneous melanoma, kidney chromophobes, and pheochromocy-

tomas. This indicates that neuronal FAM72 paralogs are being expressed in non-neuronal proliferating tumor tissue cells [12].

#### *3.1. The |-SRGAP2–FAM72-| Master Gene in Brain Cancer*

Previously, we correlated FAM72 (A–D) mRNA expression z-scores and highly mutated protooncogenes as well as unique mutated genes in deceased GBM patients. mRNA expression and mutation data for GBM was retrieved from cBioportal. Normalized mRNA expression z-score data were computed for all GBM samples and the data for FAM72 (A–D) paralogs were grouped in bins with a size of 0.7 z-score units and correlated with genes showing high numbers of tissue-specific gene mutations. Linear regression was determined first between the FAM72 (A–D) paralogs and then between all available genes in the GBM study, then visualized using online Python-based Bokeh software. A complex brain-specific gene-mutation signature: EGFR, TP53, PTEN, NF1, SPTA1, PIK3CA or SCN9A, MXRA5, ADAM29, KDR, PIK3C2G, and LRP1B was identified that correlated with high FAM72 expression and may lead to cell cycle activation, cell transformation, and cell proliferation. This led to the identification of several pivotal driver genes responsible for the transformation of NSCs into CSCs and GBM (Figure 2) [12].

On the other hand, the partner gene SRGAP2 showed no change in expression in GBM. SRGAP2 is reported to be a tumor suppressor [26], and its expression is usually induced when FAM72 expression is blocked. NSCs stop proliferating during neural differentiation and neuronal synaptogenesis [4,5,8,16–21], but may lead to apoptosis in non-neuronal tissue or proliferating cancerous cells [3,5]. Genomic rearrangements causing loss of physiological functions of SRGAP2 may enhance cell motility and metastasis [26].

#### *3.2. The |-SRGAP2–FAM72-| Master Gene in Other Cancerous Tissues*

Our recent large-scale tissue analysis demonstrated that the Ki-67 gene (MKI67) and FAM72 paralogs are co-expressed in proliferating cells in NSCs and also outside neuronal tissue (i.e., in cancer cells across various tissues) (Figure 3, Supplementary Materials Figure S1). FAM72 does not appear to be a protooncogene and the reciprocal expression dependency of SRGAP2 and FAM72 seems to be limited to the nervous system. Outside the nervous system, FAM72 expression appears to be induced by a different cancer-causing oncogene [12,27,28].

#### *3.3. FAM72 in Adrenocortical Carcinoma*

Our understanding of the molecular mechanism driving ACC has advanced. Alterations in the components of the WNT1/β-catenin, EGFR, and TP53 pathways are prominent markers in ACC [29–32]. CTNNB1 and TP53 mutations are mutually exclusive in aggressive adrenal cancers [36]. Activating mutations in CTNNB1 have been observed in approximately 25% of adrenocortical cancers [37]. TP53 mutations have been observed in more than 50% of child patients, but only in 4% of adult patients of ACC [38,39].

Recently, we identified a complex novel ACC-specific gene signature: CRIPAK, DGKZ, GARS1, LRIG1, ZFPM1, and ZNF517, which was significantly, specifically, and most repeatedly mutated in ACC and correlated with high FAM72 expression (Figure 3) [28]. This gene set is involved in tumor suppression and cellular proliferation and thus could be useful for the prognosis and development of therapeutic approaches for the treatment of ACC.

Experimental evidence indicates that EGFR signaling is an anchor body through which proliferative pathways can be initiated and most of the proto-oncogenes in ACC act downstream of EGFR. Moreover, in ACC, LRIG1 mutations would cause a continuous expression of the EGFR signaling cascade, thereby causing cellular proliferation. Inhibition of EGFR via tumor suppressor LRIG1 is thus a key step in regulating (either partially or fully) the consequent signaling cascades. Mutations in GARS1 also serve to increase proliferation via a cascade that is, however, independent of the phosphoinositide-3-kinase (PI3K)/mitogen-activated protein kinase 1 (MAPK1)/WNT1 signaling pathways. Muta-

tions in our novel gene set thus appear to be more influential in ACC tumorigenesis than those described in earlier studies and could serve as a powerful therapeutic target [28,29].

**Figure 3.** Experimental evidence-based schematic illustration of FAM72 and MKI67 co-activation in adrenocortical carcinoma (ACC). (**a**) Mutations in ACC-specific driver proto-onco- or tumor-suppressor-genes (red and green color) push the cell through the cell cycle and mediate MKI67 as well as FAM72 expression during the M-phase. Red-colored proto-oncogenes (or tumor-suppressor genes) are from Rahane et al. [28], while green-colored proto-oncogenes (or tumor-suppressor genes) are from Zheng et al. [29]; additional ACC-specific cell cycle information are from Assié et al. [30], Lippert et al. [31], and Pereira et al. [32]. Tumor suppressor LRIG1 interferes with EGFR signaling and might be a druggable protein of primary interest [33–35]. (**b**) Schematic illustration of mRNA expression correlation between FAM72A on the one hand and M-phase cell cycle genes, including MKI67, on the other hand. FAM72A expression correlates with the expression of cell cycle phase-specific genes across various human cancer tissue. Genes specifically associated with the late G2- to M-phase of the cell cycle, including ASPM, BUB1, CENPE, CENPF, CEP55, KIF14, KIF23, NEK2, NUF2, and SGO1 (ASPM, BUB1, CEP55, KIF14, KIF23, and NEK2 are involved either with spindle formation or with regulation; CENPE, CENPF, NUF2, and SGO1 are involved in the centromere-kinetochore complex) [12,28]. ASPM, Assembly factor for spindle microtubules; BUB1, Budding uninhibited by benzimidazoles 1 mitotic checkpoint serine/threonine kinase; CENPE, Centromere protein E; CENPF, Centromere protein F; CEP55, Centrosomal protein 55; CRIPAK, Cysteine-rich p21-activated protein kinase 1 inhibitor; CTNNB1, Catenin beta 1; DGKZ, Diacylglycerol kinase zeta; FZD, Frizzleds; GARS1, Glycyl-tRNA synthetase 1; KIF14/23, Kinesin family member 14/23; LRIG1, Leucine rich repeats and immunoglobulin-like domains 1; NEK2, Never in mitosis gene a-related kinase 2; NUF2, NUF2 component of NDC80 kinetochore complex; RPL22, Ribosomal protein L22; PRKAR1A, Protein kinase cAMP-dependent type I regulatory subunit alpha; RAS, Rat sarcoma; SGO1, Shugoshin 1; WNT1, Wingless and Int-1 family member 1; ZFPM1, Zinc finger protein, friend of GATA family member 1; ZNF517, Zinc finger protein 517; ZNRF3, Zinc and ring finger 3.

#### **4. FAM72 and Its Role in the Cell Cycle**

#### *4.1. FAM72 in the M-Phase of the Cell Cycle*

FAM72 (A–D) is highly expressed when promoting NSC and cancer cell proliferation and are present in the G2/M phase of the cell cycle [2,5,12,28]. It has been shown that knock-down of FAM72A in NSCs blocks cell proliferation and causes cell differentiation [4]. In line with this, FAM72B knockdown experiments showed that cell proliferation was reduced in human fibroblasts [40], suggesting that FAM72B also has a common role in promoting cell proliferation, similar to the other FAM72 members. Cell cycle specific expression analysis revealed that FAM72 (A–D) activity occurred particularly during the G2/M-phase, but not during the G1/S-phase (Figure 3b) [12,28].

NSC or cancer cell fate is determined based on specific E2 factor transcription factor E2Fx TFs (x = 1, 2, 3, 4 and 6, i.e., E2F1, E2F2, E2F3, E2F4, E2F6 such as E2F6 in a complex with transcription factor dimerization partner 1 [TFDP1]) bound to the promoter within the IGR of the |-SRGAP2–FAM72-| master gene. We found that FAM72 expression correlates with the expression of a baculoviral inhibitor of apoptosis protein (IAP) repeat (BIR) containing 5 (BIRC5, also known as survivin), Forkhead box M1 (FOXM1), LIN9, LIN54

(partially), and retinoblastoma binding protein 4 (RBBP4) (Lin53, partially) and also with pivotal E2Fx TFs in various cancer tissues including brain glioma. Other genes showed either weak (TFDP1 and TFDP2) or no correlation (oligodendrocyte marker OLIG2, tumor suppressor family with sequence similarity 107 member A [FAM107A]), paired box protein Pax-6 (PAX6), and ten eleven translocation protein 2 (tet methylcytosine dioxygenases 2, TET2), and LIN37) (Figure 4, Supplementary Materials Figures S2–S16).

**Figure 4.** Schematic illustration of mRNA expression correlation of FAM72A compared with several other GBM-relevant genes, including E2Fx TFs. (**a**) FAM72A expression correlates with the expression of selected genes and TFs (Supplementary Materials Figures S2–S17). (**b**) FAM72A expression does not correlate with the expression of OLIG2, FAM107A nor with PAX6 (Supplementary Materials Figures S7, S12, and S13). (**c**) FAM72A expression correlates with neuroblastoma rat sarcoma proto-oncogene (NRAS), TP53, and weakly with sex determining region Y (SRY) box transcription factor 2 (SOX2) in glioma (Supplementary Materials Figures S18–S20). (**d**) FAM72A expression correlates with RE1 silencing transcription factor (REST) in glioma (Supplementary Materials Figure S21).

#### *4.2. FAM72 in the G0 Stage of the Cell Cycle*

Some studies showed that retinoblastoma transcriptional corepressor 1 (RB1) may cause the cell to go into the G0 phase with different cell fates: Quiescent G0 with reversible return option to reenter the cell cycle for proliferation, post-mitotic G0 with irreversible cell differentiation, or cell senescence G0, eventually leading to apoptosis [41,42]. Our data suggest that the |-SRGAP2–FAM72-| master gene induces the RB1 pathway (eventually via TP53 acetylation) to push cells into the G0 stage concomitantly with SRGAP2 expression, supporting neural survival and stabilizing a neuronal phenotype at stage G0 [8].

As we reported recently, the dual IGR promoter has an important role in regulating divergent gene transcription of both directions of the |-SRGAP2–FAM72-| master gene [8]. In the context of rat PC12 cells (a well-known neuronal cell model to study neurogenesis [8,43–48]), Fam72a expression (in proliferating PC12 cells stimulated by the mitogen Egf) or Srgap2 expression (in differentiating PC12 cells stimulated by nerve growth factor (Ngf)) was enhanced upon growth factor (Ngf or Egf)-mediated stimulation. Strikingly, under serum-withdrawal-induced stress and bi-directional IGR control, Egfstimulated PC12 cells were kept alive for a long period of time with Fam72a expression,

while Ngf-stimulated PC12 cells remained in a G0 stage co-expressing Srgap2 and Fam72a without proliferation [5,8].

#### *4.3. Governance of FAM72 Expression: The IGR and Its TFBSs*

A comparative genome analysis of the IGR (located between the SRGAP2 and FAM72 genes within the |-SRGAP2–FAM72-| master gene on the one hand and the gene promoters of several G2/M-phase-specific cell cycle genes on the other hand) revealed potential common regulatory elements (i.e., common TFBSs), driving the expression of those cell cycle genes and FAM72 to promote and maintain cell proliferation (Figures 3b, 4 and 5a,b). We found that many genes with increased expression during the late G2/M-phase of the cell cycle including all human FAM72 paralogs shared the same TFBS motifs for GATA binding protein 2 (GATA2) [12], E2F4, E2F6, and TFDP1 (Figures 4 and 5a,b). This indicates that their expression is co-regulated in concert with the FAM72 paralogs and implies a common temporal and spatial function, particularly fostering cell proliferation, eventually associated with the RAS signaling pathway [49–53].

**Figure 5.** Integrated diagram for putative TFBSs in the intergenic region IGR between the transcription start sites (TSS) of FAM72A and SRGAP2, using the Ensembl and JASPAR databases, and the effect on the cell cycle. (**a**) Putative TFBSs on the IGR between SRGAP2 and FAM72A coding sequences in *Homo sapiens*. Multiple TFBSs are present for binding of the TFs GATA2, SPI1, MZF1, EGR1, SP1, and E2Fx (x = 1, 2, 3, 4 and 6). The open reading frames (ORFs) for FAM72A and SRGAP2 are indicated on the right and left sides, respectively. TFs that are common between FAM72 (A–D) and selected M-phase cell cycle genes are in pale blue. Investigation of the potential TFBSs on the IGR shows that FAM72A is a cell cycle gene particularly active in mitosis and under control of the DREAM and MMB-FOXM1 complexes acting on the CHR BS to regulate |-SRGAP2–FAM72A-|. The DREAM complex is composed of TFDP1, RBL2, or RBL1, the repressor E2F TF E2F4 or E2F5 and the MuvB core complex (containing LIN9, LIN37, LIN52, RBBP4 (LIN53), LIN54). The MMB–FOXM1 complex is composed of the MuvB complex (dissociated from the DREAM complex), MYBL2, and FOXM1. Notably, the CHR site (pale blue) located next to the TSS of the FAM72A gene has the highest potential to be targeted for driving FAM72A gene expression. (**b**) The crucial E2F4/E2F6/TFDP1 BS in cell fate decision. The consensus E2F4/E2F6/TFDP1 BS within the IGR could become occupied by an E2Fx family member depending on cell demand during specific cell phase stages and may be crucial for cell fate decision to activate either FAM72A (for cell proliferation and renewal) or SRGAP2 (for neural differentiation). Chr, chromosome; CHR, Cell cycle gene homology region; E2F1/2/3/4/6, E2 factor TF 1/2/3/4/6; EGR1, early growth response 1; EHF, ETS homologous factor; ETS1, E26 transformation specific proto-oncogene 1; GATA2, GATA binding protein 2; MGA, MAX dimerization protein; MZF1, Myeloid zinc finger 1; NFIC, Nuclear factor I C; SP1-1, Specificity protein 1 TFBS 1; SPI1, Spleen focus forming virus proviral integration oncogene 1; TBX15/TBX1/TBX4, T-box TFBS 15/1/4; TFDP1, TF dimerization partner 1; ZNF345C, Zinc finger protein 345C.

Additional comparative genome analysis between the FAM72 and MKI67 gene promoters also revealed common potential TFBSs for the TFs GATA2, E26 transformation specific proto-oncogene 1 (ETS1), myeloid zinc finger 1 (MZF1), and nuclear factor I C (NFIC), zinc finger protein 345C (ZNF354C) (Figures 3b and 5a) [12].

To further understand the mechanism of IGR-controlled |-SRGAP2–FAM72-| master gene expression, we performed bioinformatic analysis of TFBSs on the IGR (Figure 5a) [54]. The predicted TFBSs appear to partly explain our questions raised based on their ability to control the cell cycle and transcription regulation of this |-SRGAP2–FAM72-| master gene pair via its IGR. Specifically, we discovered E2F4, E2F6, and TFDP1 TFBSs present on the IGRs of the |-SRGAP2–FAM72A-|, |-SRGAP2C–FAM72B-|, |-SRGAP2D–FAM72C-|, and |-SRGAP2B–FAM72D-| gene pairs (Figure 5a,b). This indicates the participation of a heterodimeric E2Fx/TFDP1 complex, which may contribute to the divergent gene transcription control of FAM72A and SRGAP2, respectively (Figure 5a,b). The E2Fx family is known to consist of TF members, which all play important roles in the cell cycle control. The E2F4/E2F6/TFDP1 predicted sites on the IGR are assumed as binding sites for E2Fx family members with both gene activation or repression abilities [55–57]. Interchangeable roles of E2Fx family members were revealed by a comprehensive ChIP analysis of E2F1 (e.g., E2F1-3a activators), E2F4 (e.g., E2F4-5 canonical repressors), and E2F6 (e.g., E2F6-8 atypical repressors) in normal and tumor cells [55], while loss of one E2F member could cause a function compensation by the other E2Fs to ensure cell cycle operation [58,59]. Specifically, E2F6 encodes a member of a family of TFs that plays a crucial role in the control of the cell cycle, of which the protein lacks the transactivation and tumor suppressor protein association domains found in other E2Fx family members, and it contains a modular suppression domain that functions in the inhibition of transcription. It interacts in a complex with chromatin modifying factors. Moreover, TFDP1 encodes a member of a family of TFs that heterodimerize with E2Fx proteins to enhance their DNA-binding activity and promote transcription from E2Fx target genes. The encoded protein functions as part of this complex to control the transcriptional activity of numerous genes involved in cell cycle progression from G1 to the S phase.

In the CNS, E2Fx TFs such as E2F1, E2F2, E2F3, and E2F4, along with the pocket proteins (PPs including RB1, RB-like pocket proteins RBL1 (p107) and RBL2 (p130)), regulate NSC self-renewal via pivotal genes including SOX2, PAX6, fibroblast growth factor 2 (FGF2), distal-less homeobox 1 and 2 (DLX1, DLX2), neogenin 1 (NEO1), and neuropilin 1 (NRP1) as well as the Notch and sonic hedgehog (SHH) pathways [60–70]. Interestingly, E2F4 establishes a proper cell fate both in conjunction with or without RB1 [71,72].

Detailed spatiotemporal expression analysis of E2Fx TFs unraveled specific E2Fx activators (E2F3A) and canonical (E2F4) and atypical (E2F8) E2Fx TF repressors during the cell cycle [73]. An orchestrated accumulation of different E2Fx TF combinations control gene expression in proliferating (E2F3A-8-4) and differentiating (E2F3A-4) cells. The sequential nuclear accumulation and disappearance of E2F3A, E2F8, and E2F4 form an E2F module used to drive waves of activation and repression that support cell-cycle-dependent oscillations in gene expression necessary for cell proliferation and cell divisions. Another E2Fx TF module composed of E2F3A and E2F4 is used to extinguish cell-cycle-dependent gene expression in cells programmed to exit the cell cycle and differentiate. With an activity in the G2 phase and a TFBS within the IGR, E2F4 seems to be among the pivotal TFs controlling the |-SRGAP2–FAM72-| master gene (Figure 5).

Since E2F4 may have both functions of gene activation and repression, we assume that E2F4 could be the key that could repress FAM72A and support SRGAP2 expression during differentiation. This is consistent with the finding that E2F4 permanently accumulates in the nucleus of differentiating and differentiated cells [73].

With these important discoveries about the |-SRGAP2–FAM72-| master gene and its regulatory role regarding cell fate decision [5,8], we looked for partners to cooperate with these two genes. Among the possible candidates, E2Fx TFs and their regulatory partners, the PPs RB1, RBL1 and RBL2, are widespread and dynamic epigenetic stem cell

regulators [69]. The E2Fx consensus TFBS on the IGR shows an ability to interact with various E2Fx TFs, which in turn, can bind to the IGR and govern FAM72 as well as SRGAP2 expression (Figure 5). The RB1 and E2Fx TFs make complexes called RB-E2Fx, which cooperate with a protein complex called DREAM (dimerization partner (DP), RB-like, E2F and multi-vulval class B (MuvB)), repressing G1/S cell cycle genes to move the cell cycle forward to the G2/M phase [74–76]. The conserved human DREAM complex thus has been described as an important master regulator of cell cycle genes with a decisive role in coordinating cell cycle progression [75–81].

The DREAM complex comprises TFDP1, RBL1, or RBL2, the repressor E2Fx TFs E2F4 or E2F5 and the MuvB core complex (containing LIN9, LIN37, LIN52, RBBP4 (also known as LIN53), LIN54).

LIN9, a component of the DREAM core complex, encodes a tumor suppressor protein that inhibits DNA synthesis and oncogenic transformation through association with the RB1 protein. It also interacts with a complex of other cell cycle regulators to repress cell cycle-dependent gene expression in non-dividing cells [82].

RBBP4 is a chromatin remodeling factor that encodes a ubiquitously expressed nuclear protein that belongs to a highly conserved subfamily of WD-repeat proteins [83,84]. It is involved in histone (de-) acetylation and chromatin assembly and remodeling. RBBP4 is also part of co-repressor complexes, which are integral components of transcriptional silencing. It is found among several cellular proteins that bind directly to RB1 to regulate cell proliferation and also seems to be involved in transcriptional repression of E2Fxresponsive genes [85,86].

As an integral subunit of the DREAM complex, LIN54 is a pivotal regulator of cell cycle genes, which binds to the cell division control 2 (CDC2) promoter for cell cycle progression [87].

Previously, we described FAM72 (A–D) expression specifically during the G2/M phase [12,28]. Other scientists have verified that the DREAM and MMB-FOXM1 complexes can bind genomic cell cycle gene homology region (CHR) motifs, suggesting that DREAM and MMB-FOXM1 are crucially involved in regulating FAM72 (A–D) expression during the G2/M phase (Figures 5 and 6) [12,75,76,88]. Indeed, the DREAM complex was verified to bind to the FAM72 promoter, most probably via the CHR BS on the IGR. Notably, the CHR element is conserved on the IGR across all FAM72 (A–D) (Figure 5a) [75,76]. Genome-wide association studies and experimental validation have verified FAM72D as a G2/M cell cycle gene modulated by the DREAM and MMB-FOXM1 complexes [75,76]. These complexes bind and regulate FAM72D through a CHR BS on the IGR.

The DREAM complex interacts with the CHR element and E2Fx TFBSs to inhibit G1/S cell cycle gene expression until MuvB dissociates away to associate with MMB-FOXM1 to push the cell cycle into the G2/M phase [75,76]. During quiescence and early G1 phase of the cell cycle, the DREAM–MuvB complex represses cell cycle-promoting gene expression. When the stages end, it becomes deactivated, while the MuvB complex dissociates away to associate with v-myb avian myeloblastosis viral oncogene homolog-like 2 (MYBL2) and FOXM1, forming the MMB–FOXM1 complex. This new complex promotes late cell cycle gene expression and is required to pass through the G2/M phases [80]. FOXM1 gets phosphorylated during the M phase and regulates the expression of several cell cycle genes such as cyclin B1 (CCNB1) and cyclin D1 (CCND1). It is a crucial TF also found in fostering GBM development and progression by regulating key factors involved in cell proliferation, epithelial to mesenchymal transition (EMT), invasion, angiogenesis, and upregulating WNT1/β-catenin signaling [89].

**Figure 6.** FAM72 paralog-specific cell cycle signaling mediated by various TFs. DREAM (blue rectangle) is composed of TFDP1, RBL1 or RBL2, the E2Fx TFs E2F4, or E2F5 and the MuvB core complex (consists of LIN9, LIN37, LIN52, RBBP4 (LIN53), LIN54). During quiescence/G0 and early G1 phases of the cell cycle, DREAM represses cell cycle gene expression. When these G0/G1 stages end, DREAM gets inactivated so that the MuvB complex dissociates away to form a new complex with the MYBL2 and FOXM1 called the MMB-FOXM1 complex (red triangle, MuvB, MYBL2 and FOXM1). This new complex promotes late cell cycle gene expression and is required to pass through the G2/M phases [80]. At the end of the M-phase, REST inhibits neuronal gene expression (such as SRGAP2) to allow re-entry into a new cycle, thus maintaining NSC renewal and FAM72 expression. Once it receives a neurogenic signal, REST is degraded, FAM72 expression is blocked, and SGRAP2 expression is initiated for neuronal differentiation. ATOH1, Atonal basic helix-loop-helix (bHLH) TF 1; BMI1, B cell-specific Moloney murine leukemia virus integration site 1; BTG2, B-cell translocation gene 2;CDK2/4/6, Cyclin dependent kinase 2/4/6; CDKN1A/1B/2A, Cyclin dependent kinase inhibitor 1A/1B/2A; FOXM1, Forkhead box M1; GLI1, Glioma-associated oncogene family zinc finger 1; MDM2, Murine double minute 2; MuvB, Multi-vulval class B complex; MYBL2, v-myb avian myeloblastosis viral oncogene homolog-like 2; MYCN, v-myc avian myelocytomatosis viral oncogene neuroblastoma derived; NEUROG2, Neurogenin 2; REST, Transcriptional repressor RE1 silencing transcription factor; SHH, Sonic hedgehog signaling molecule.

The cell cycle promoting regulation indeed comes from the interaction between the FOXM1 protein—a part of the MMB-FOXM1 complex—and the FAM72A [75,76,78,81], FAM72B [40,75,76,78,81], and FAM72D [25,75,76,78,81] promotors, confirming that all FAM72 (A–D) paralogs are regulated by this pathway during the G2/M phase in proliferating cells (i.e., NSCs and cancer cells). Since the FAM72 function may contribute to the mitotic spindle or the kinetochore-centromere complex formations and activities, loss of MMB-FOXM1 or FAM72 (A–D) function may cause spindle assembly chaos and mitotic catastrophe (Figure 6) [12,25].

Taken together, FAM72A, FAM72B, and FAM72D might be regulated by the DREAM complex as well as the RB-E2F3b/4/5 complex to be suppressed for a while by interacting with the putative E2F4/E2F6/TFDP1 TFBS, until the E2F1/2/3a/activators promote essential G1/S gene expressions and thereby foster cell cycle progression into G2/M phases and FAM72 activation via the MMB–FOXM1 complex. Thus, cell cycle progression and control depend on targeting the genomic E2F4/E2F6/TFDP1 TFBS (for G1/S phase) and the CHR

motif (for G2/M phase) on the IGR with pivotal regulators involved such as DREAM, the RB family members-E2Fx-, and the MMB-FOXM1 complexes.

#### *4.4. FAM72 Expression and the RE1 Silencing Transcription Factor*

REST was initially identified as a transcriptional repressor that represses neuronal genes in non-neuronal tissues [90,91]. However, depending on the cellular context, this gene can act as either an oncogene or a tumor suppressor, and its specific role in glioma remains controversial [92,93]. The encoded protein is a member of the Kruppel-type zinc finger transcription factor family. It represses transcription by binding a DNA sequence element called the neuron-restrictive silencer element [94,95]. The protein is also found in undifferentiated neuronal progenitor cells and it is thought that this repressor may act as a master negative regulator of neurogenesis [96–98]. Alternatively-spliced transcript variants have been described [99]. Expression correlation analyses showed a weak correlation of FAM72A with REST in glioma (Figure 4d).

#### *4.5. FAM72 Expression and Long Non-Coding RNAs*

Additionally, it has been hypothesized that IGR regulation of the |-SRGAP2–FAM72-| master gene is susceptible to long non-coding RNAs (lncRNAs) [1,5,8]. Long non-coding RNAs (LncRNAs) are of particular interest due to the wide variety of roles they play in gene regulation. LncRNAs have been reported to regulate transcription (via epigenetic mechanisms [100,101]) as well as pluripotency and cellular reprogramming [102] and have been implicated in a variety of diseases, notably cancers of the breast [103], colon [104], stomach [105], lymph [24], and the CNS [106].

Recent reports about the oncogenic role of lncRNA revealed interactions between a lncRNA and the centrosomal protein CEP112 as well as the breast cancer type 1 susceptibility protein BRCA1, which resulted in mitotic abnormalities and malignancies [107]. The particular lncRNA, called genomic instability inducing RNA (Ginir), functions normally during embryonic development and is enriched in the brain. The expression of Ginir, along with its partner genomic instability inducing RNA antisense (Giniras), was regulated in a spatio-temporal manner and overexpression of Ginir led to tumorigenesis [107]. This ties in with the role of lncRNA in FAM72 expression. Since FAM72 is also expressed predominantly in NSCs, it is likely that the transcription of the |-SRGAP2–FAM72-| master gene is regulated by a similar pair of lncRNAs on the IGR. FAM72 co-expresses with centrosomal proteins in cancer tissues [12], and it is possible that dysfunction of the lncRNA on the IGR would lead to loss of control over FAM72 expression, thereby leading to cellular proliferation.

#### *4.6. Anti-Apoptotic Features of |-SRGAP2–FAM72-| via TP53*

Our previous study showed an early anti-apoptotic rescue program activated via the IGR-based expression of the |-Srgap2–Fam72a-| master gene under serum-free stress conditions in rat PC12 cells. Tp53 was thought to influence Fam72 activities in this stress response to rescue cells from apoptosis by driving them into the G0 phase, a possible new anti-apoptotic functional ability of Fam72a [8]. This anti-apoptotic activity of Fam72a was recently consolidated with its highly correlated expression with BIRC5 (Figure 4, Supplementary Materials Figure S2) [108], a member of the family of IAPs that prevent apoptotic cell death [109]. IAP family members usually contain multiple BIR domains, but BIRC5 encodes a protein with only a single BIR domain. The encoded protein also lacks a C-terminus RING finger domain. Along with FAM72A, the FOXM1 protein was also found to be similarly co-regulated with BIRC5 [108]. BIRC5 expression is high in most tumors; however, its usefulness as a prognostic marker is still a controversial issue [110,111].

Although TP53-mediated impact on FAM72 might be indirect, we found a TATA box and a SP1-1 TFBS on the IGR, which could be bound by TP53 with high affinity, thereby eventually affecting FAM72 directly or indirectly by blocking those positions for other TFs (Figure 5) [112–114].

On the other hand, TP53 could also bind SP1-1 TFBS on the IGR for transcription regulation by competing with the SP1 protein. To enter G0, downregulation of cell division cycle 25C (CDC25C), another key molecule for cell cycle progression through the G2/M phase [115–117], is mediated by TP53 via two independent mechanisms. One of these involves direct binding to the CDC25C promoter [114].

In another scenario, FAM72 expression was regulated by both DREAM and MMB– FOXM1 complexes under the control of TP53, particularly in cancer cells [75,76]. Through inhibition of cyclin-dependent kinase (CDK) activity by the CDK inhibitor 1A (encoded by CDKN1A), FAM72A and FAM72D were downregulated by TP53 in response to DNA damage via interfering with the DREAM and MMB-FOXM1 complex binding via the CHR BS motifs on the IGR. This prevents the FAM72A and FAM72D expressions, respectively, thus confirming FAM72A and FAM72D as G2/M-phase-promoting cell cycle genes (Figures 5 and 7) [75].

**Figure 7.** The influence of TP53 on FAM72 paralogs directly regulates the cell cycle in cancer. Upon a stressful DNAdamaging signal (e.g., gamma irradiation), TP53 gets activated to mediate cell arrest in G0 to give the cell quiescence for cell repair or, if impossible, to induce the alternative pathway for apoptosis. TP53-mediated cell cycle arrest is conveyed by CDKN1A (p21) causing inhibition of the cell-cycle promoting the CDK4/6-E2Fx pathway; consequently, the G1/S phase genes remain blocked. The TP53-CDKN1A-CDK4/6 pathway also causes activation of DREAM, which in turn blocks FAM72 expression via the CHR element within the IGR of the |-SRGAP2–FAM72-| master gene.

Notably, investigation of TP53 interaction network showed that FAM72A, FAM72B, and FAM72D expressions positively and negatively correlated with TP53 expression in multiple types of cancer under unknown pathways. All FAM72 and TP53 are expressed increasingly in kidney renal papillary-cell carcinoma, but are only highly expressed in FAM72B and TP53 in pancreatic adeno-carcinoma, pheochromocytoma, and paraganglioma. In contrast, the TP53 expression went down, while the three FAM72 were upregulated in lung adeno-carcinoma and prostate adeno-carcinoma [118]. The correlations indicate that all members of the FAM72 family have important roles in tumorigenesis and crossing regulation with the tumor suppressor TP53.

Taken together, FAM72A, FAM72B, and FAM72D can be regulated by the DREAM complex by interacting with the putative E2Fx BS from cell cycle G0/G1/S phases, which

are controlled by the RB-E2Fx complex—a specific E2Fx TF complex for cell cycle gene expressions. This pathway was also regulated through inhibition of CDKN1A coordinated by TP53 (Figures 5 and 7). In summary, all FAM72 (A–D) paralogs and TP53 appear to have strictly correlated expression patterns with a possible crucial functional impact on each other. On the topic of interfering with FAM72 expression in the context of tumor cell proliferation, the tumor suppressor TP53-FAM72 linked pathway could be important as an option to inhibit cancer cell proliferation.

The TP53-CDKN1A pathway also takes away the MYBL2 phosphorylation by CDK2, resulting in the activation of MMB-FOXM1, which in turn, could act on the CHR element on the IGR of |-SRGAP2–FAM72-| master gene for FAM72 activation [119,120]. If the DNA damage is too strong (e.g., causing a mutation in a cancer driver proto-oncogene or even in TP53 itself), G2/M phase genes and FAM72 expression remains at high level fostering cancer cell proliferation [12,75,76,78,81]. AKT1, AK strain transforming serine/threonine kinase 1; ATM, Ataxia-telangiectasia mutated serine/threonine kinase; ATR, Ataxia telangiectasia and Rad3-related serine/threonine kinase; BAK1, BCL2 antagonist/killer 1; BAX, BCL2 associated X; BCL2, B cell lymphoma 2; BCL2L1, BCL2 like 1; BID, Bcl-2 homology 3 interacting domain death agonist; CASP3/6/7/9, Caspase 3/6/7/9; CHEK1/2, Checkpoint kinase 1/2; CYCS, Cytochrome c, somatic; DIABLO, Direct inhibitor of apoptosis binding protein with low pI; MCL1, Myeloid cell leukemia 1; MOMP, Mitochondrial outer membrane permeabilization; PIK3CG, Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit gamma.

#### **5. Methylation of FAM72 in Cancerous Tissues**

DNA methylation is a well-studied epigenetic modification and involves the covalent attachment of a methyl group to the 5-carbon residue of cytosine [121]. These attachments usually occur on genomic regions with a high density of CpG nucleotides, called CpG islands, but methylation has also been reported in non-CpG regions [122]. Modifications in DNA methylation have been reported from various disorders including multiple sclerosis, diabetes, multiple human cancers as well as neurological disorders [123–127]. Both hyperand hypomethylation at the CpG islands have been associated with cancers, and there has been a lot of work to understand the mechanisms regulating this behavior [128–130].

We verified the hypomethylation of FAM72A in GBM, which revealed that expression of FAM72A in GBM could depend on its methylation status [12]. Investigation of the methylation status of FAM72A in non-neuronal tissues revealed that increased expression of FAM72A in lung and uterine cancer tissues appeared to be rather independent of its methylation status (Figure 8a). However, methylation-expression analysis of breast and liver cancer tissues showed an increase in mRNA expression corresponding to a decrease in promoter methylation. The methylation status of the FAM72 promoter thus appears to be important—to a certain extent—in some tissues, namely GBM, breast, and liver cancers, whereas other factors come into play in other non-neuronal tissues. The increased FAM72 expression in non-neuronal tissues is driven by somatic mutations in oncogenes, which would then trigger the signaling cascade for promoting cellular proliferation and fostering tumorigenesis and metastasis [12]. Another factor responsible for increased FAM72 expression could be binding of TFs, which regulate other proliferative genes. We described GATA2 as one of the candidates that could regulate both FAM72 as well as prophase/metaphase cell cycle genes [12].

Comparing the corresponding methylation-expression statuses in SRGAP2 revealed that there is no clear difference in its methylation status, which correlates to no or minor changes in its expression status across the same cancer tissues. Indeed, SRGAP2 itself shows no changes in its comparatively higher expression (with FAM72A) with slight differences in promoter methylation of SRGAP2 in breast, liver, lung, and uterine cancers (Figure 8). However, a decrease of methylation (demethylation) on the SRGAP2 gene body in GBM with no changes in its gene expression indicates that this genomic methylation does not affect SRGAP2 gene expression itself, but may rather have an impact on the other coupled

gene FAM72A (Figure 8). This also fits with SRGAP20 s established role in neuronal cell differentiation, synaptic maturation as well as neuronal migration [5,8,16,17,26,131]. Usually, SRGAP2 is mobilized to foster neuronal differentiation and synaptic plasticity [4,5,8,16–21]. However, in non-neuronal cells, SRGAP2 expression is needed for rearranging the cytoskeleton required for cell-specific locomotion and motility and, if genomic rearranging occurs within the genomic SRGAP2 body, its tumor suppressor function is abolished, and metastasis is induced [26,132–134]. Overall, however, it appears that in cancer tissues including GBM [135], the methylation status does not have a major impact on the |-SRGAP2–FAM72-| master gene.

In the brain, the hypothalamus, cerebral cortex, and hippocampus have been reported to be rich sources of oxidized 5-methylcytosine (5mC), thus converting it to 5 hydroxymethylcytosine (5hmC) [139] on enhancers [140]. Demethylation (5hmC) is significantly increased when NSCs and neural progenitor cells (NPGs) differentiate into neurons [141]. This is in contrast to the hypomethylation observed during oncogenesis in FAM72 or the lack of methylation changes observed in SRGAP2, thus confirming that the proliferative and neurogenic mechanisms occur via completely different mechanisms under normal and pathophysiological conditions in the |-SRGAP2–FAM72-| master gene. 5mC-loss/5hmC-gain loci are enriched in active enhancers and motifs for key binding factors involved with neurogenic genes during neurogenesis, as is expected for neurogenic SRGAP2 expression during neuronal differentiation under physiological conditions [142]. The TET1-3 proteins are connected with neural fate decisions [141–144]. TET2 is a key protein involved in the development and cancer regulating gene expression via oxidization of 5mCs, thereby promoting locus-specific reversal of DNA methylation [145,146]. Thus, TET2 mutations are associated with multiple neurodegenerative diseases [147] and variations of the TET2 gene in either non-coding or coding regions might cause alterations of the homeostasis of key aging-related processes [147].

This indicates that TET2 and the 5mC/5hmC mechanism may contribute to |-SRGAP2– FAM72-| master gene activity during neurogenesis (e.g., the 5hmC-gain of neurogenic SRGAP2 during neural differentiation) (Figure 9). Partially differentiated NSCs going into NPGs might be able to concurrently express FAM72 and SRGAP2, thus not resulting in complete loss of 5mC and still gaining some 5hmC. The full 5hmC-gain needed is met only once neurogenic commitment is accomplished, when SRGAP2 is sufficiently expressed and FAM72A expression is completely blocked (e.g., in post-mitotic differentiated neurons).

Moreover, multiple myeloma [25] and breast cancers [25] showed that FAM72 expression may be dependent on its methylation status. Demethylation (5mC → 5hmC) of FAM72D occurs mainly in intronic enhancers (but outside the IGR area) and could activate FAM72D to maintain mitotic fidelity. This probably also works for the other FAM72 member expressions such as FAM72A, FAM72B, and FAM72C (due to high homology (99%) in amino acid sequences) [25]. As for GBM, Kan et al. could not identify epigenetically affected FAM72, though our data showed a change in the methylation status (Figure 8) [12,135].

This observation is also in line with our results obtained in PC12 cells [8]. The hypothesis here is whether GATA2 TFBS (present three times on the IGR of the human |-SRGAP2–FAM72-| master gene, Figure 5) can act as a binding target for this pioneering TF since GATA2 may not directly mediate DNA methylation, but may have an impact on DNA packaging controlled by histone methylation and acetylation [142,148]. GATA TFs can either activate gene expression by synergy with another co-activator (which recruits a histone methyltransferase and/or a histone acetyl transferase) or repress gene expression by cooperating with a co-repressor to recruit a histone demethylase and/or a histone deacetylase [148]. Thus, the GATA2 TFBSs on the IGR may play important roles in the governance of the |-SRGAP2–FAM72-| master gene expression and require further investigation.

**Figure 8.** (**a**) Comparison between methylation status and expression levels of FAM72A and SRGAP2 across normal and tumor tissues in GBM, breast, lung, uterine, and liver cancers. Mean methylation beta values were plotted against mean RNA-sequencing by expectation-maximization (RSEM) expression values (log2-transformed normalized RSEM values). Circles indicate normal tissues and triangles indicate cancer tissues. Green symbols indicate GBM data, purple symbols indicate breast invasive carcinoma data, black symbols indicate liver hepatocellular carcinoma data, blue symbols indicate lung adenocarcinoma data, red symbols indicate lung squamous cell carcinoma data, and orange symbols indicate uterine corpus endometrial carcinoma data. In the case of FAM72A, the differences in methylation status between normal and cancer tissues vary among GBM, breast, and liver cancer, where less methylation leads to a two-fold difference in FAM72A expression in cancer tissues. Methylation status between normal and cancer tissues is similar in lung and uterine cancer tissues. In the case of lung tissues, the cancer samples show higher FAM72A expression and a higher methylation status as well, indicating that the FAM72A promoter methylation alone may not be responsible for its increased expression but other factors such as mutations in cancer driver-oncogenes may promote increased FAM72 expression and foster cancer cell proliferation [12,28]. For SRGAP2, the differences in methylation status between normal and cancer tissues is not significant, except eventually for GBM. However, there are no significant changes in SRGAP2 expression in most tissues. Hence, the change of methylation levels within the genomic SRGAP2 area in GBM does not affect SRGAP2 expression. Mean beta values as well as mean mRNA expression values were retrieved from the Wanderer database [136]. BRCA, breast cancer (breast invasive carcinoma); GBM, glioblastoma multiforme; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; UCEC, uterine corpus endometrial carcinoma. (**b**) Investigation of the specific methylation probes on the genomic |-SRGAP2–FAM72A-| master gene in both normal and GBM tissues. Demethylation is described as a decrease in the methylation score of all probes bound to cancer tissue genomes compared to normal tissue genomes. Unfortunately, most probes are focused on the genomic SRGAP2 gene body and IGR area, while no probes could be identified to bind to the genomic FAM72 gene body area. As above, almost no change of SRGAP2 expression level was observed throughout many cancer types including GBM. In contrast, the discovered demethylations may have an impact on regulating the other part of the |-SRGAP2–FAM72-| master gene (i.e., modulating FAM72A expression). The probe information was retrieved from the HumanMethylation450 v1.2 manifest file on the Illumina database (https://support.illumina.com/downloads/infinium\_humanmethylation450\_product\_files.html) (accessed date: 20 December 2020) and aligned the source sequences to genome reference consortium human build 38 patch release 13 (GRCh38.p13) using the BLAST-like alignment tool (BLAT) function from the Integrative Genomics Viewer (IGV) [137,138].

**Figure 9.** Effects of methylation and demethylation on the expression of the master gene |-SRGAP2–FAM72-|. The epigenetic modifications during neurogenesis can control FAM72 expression for cell fate decision. Dysregulation causes CSC formation and tumorigenesis. In proliferating (non-cancerous) NSCs, demethylation or hypomethylation (such as 5-hydroxymethylcytosine (5hmC) or the loss of the methyl group in the 5-methylcytosine nucleotide (5mC)), were demonstrated to activate neurogenic genes such as SRGAP2 to mediate neural differentiation [142]. As a consequence, FAM72 is deactivated. In cancer, demethylation is crucial for FAM72 activation during CSC proliferation. In the case of glioma genesis, FAM72 is silenced in glia progenitor cells until activated TP53 replaces the methylation factors DNA methyltransferase 3 alpha/beta (DNMT3A/DNMT3B) for demethylation factors TET1/TET2/TET3 so that FAM72 is activated for proliferation and forming GBM cells, which is in line with the genomic hypomethylation of the FAM72 promoter region in our previous study (Figure 8) [12,149,150]. FAM72 expression is activated outside the CNS only under cancerous conditions by mutated protooncogenes and genomic FAM72 demethylation by TET family members to support the proliferation of cancerous cells including multiple myeloma [25,151].

#### **6. FAM72 and FAM107A in GBM**

FAM107A (also known as downregulated in renal cell carcinoma 1 [DRR1]) is a novel unique protein family that exhibits functional similarity with heat shock proteins (HSPs) during the cellular stress response with diverse functions in cancer and the nervous system [152]. Recent evidence indicates that FAM107A is involved in GBM invasion and progression, possibly through the induction of EMT activation by phosphorylation of AKT1 [153]. Accordingly, antibody (against glioblastoma stem cells surface markers glycoprotein cluster of differentiation 44 (CD44) and ephrin receptor A2 (EPHA2)-antisense oligodeoxynucleotides (ASOs) strategy against FAM107A) were established for the treatment of GBM [154].

In agreement with FAM107A as a tumor suppressor gene [152,155,156], FAM72A shows a negative expression correlation in GBM (Figure 4b).

#### **7. FAM72 and Its Role as a Potential Biomarker in Clinical Cancer Diagnostics**

Liquid biopsies carrying circulating tumor-derived material, also called the "tumor circulome," consist of circulating tumor DNA (ctDNA), circulating tumor RNA (ctRNA), circulating tumor proteins (ctPs), tumor-derived extracellular vesicles (EVs), tumor-educated platelets (TEPs), and circulating tumor cells (CTCs), among others, which have promising diagnostic potential at each stage of cancer [157]. Liquid biopsies have a great potential to overcome existing limitations of tissue biopsies, particularly in light of sampling and analysis of such liquid biological sources, typically blood, for cancer diagnosis, screening, and prognosis. The 'tumor circulome' can be directly or indirectly used as a source of cancer biomarkers in liquid biopsies, particularly ctDNA, ctRNA, and ctPs. FAM72, at the

ctDNA, ctRNA, and ctP level, could possibly serve as biomarkers for clinical diagnostics of cancer as its expression is usually limited to proliferating NSCs.

#### **8. FAM72 and Its Role in Cancer Therapy: Therapeutic Options against Tumorigenic FAM72**

Targeting FAM72 could thus be a viable treatment method for several cancer types outside the CNS because knockout of neural-specific FAM72 gene function in non-neuronal tissue may cause spindle assembly defects outside the CNS, followed by cell differentiation, senescence, or death by mitotic catastrophe in all non-neuronal proliferating cancer cells. FAM72 is an attractive target for therapy as it is a proliferative marker expressed during the late G2/M-phase of the cell cycle as well as its low expression in normal non-neuronal tissues [3,12,158], and multiple potential approaches are possible.

#### *8.1. Therapeutic Options against Tumorigenic FAM72: RNA Interference (RNAi)*

RNAi has emerged as a very effective tool for in vivo selective silencing of gene transcription, and substantial progress has been made in analyzing the therapeutic potential of various RNAi products. There are certain advantages of using RNAi for cancer therapy including the ability to target any gene including FAM72A [4], low dosages, and extended inhibition after a single dose [159]. Recently conducted clinical trials against solid tumors are promising, with the RNAi being delivered via nanoparticles [159,160]. Short hairpinloop RNAs (shRNAs) have been demonstrated to knockdown FAM72A activity, leading to differentiation in NSCs [4]. This proves the efficacy of the approach in developing therapy against FAM72. Another approach would be to target both small interfering RNAs (siRNAs) as well as telomerase reverse transcriptase and/or MKI67 [161]. Briefly, the authors constructed adenovirus containing siRNAs targeting both MKI67 as well as the telomerase reverse transcriptase. Gene silencing for multiple oncogenes using more than one siRNA have been demonstrated before [162], and the experiment by Fang et al. [161] inhibited renal cancer cells in vitro. An oncolytic vector containing siRNAs targeted toward FAM72A as well as telomerase reverse transcriptases could prove effective without affecting normal cells, especially in non-neuronal tissues.

Another approach would be the application of ASOs. ASOs are synthetically generated nucleotide sequences, about 12–25 bases long, which can be tailored according to the target sequence of interest. Intracellular binding of the ASO to its target mRNA results in RNAse cleavage, thereby leading to a lack of mRNA translation and protein formation. Currently, there are approximately 90 ongoing clinical cancer trials evaluating treatment with ASOs, with a majority being in phase I [163,164]. Animal models have proved the efficacy in inhibiting tumor formation using MKI67 ASOs, however, issues remain with optimizing dosage and nuclease degradation susceptibility [165,166]. There have been some successes using ASO cancer trials. OT-101, a phosphorothioate ASO designed for the targeted inhibition of human transforming growth factor beta 2 (TGFβ2) mRNA, has proceeded to the phase I/II clinical trial and demonstrated encouraging results [167]. AZD9150, a STAT3 inhibiting ASO, has demonstrated tumor suppressive activity in lung and lymphoma models as well as in a phase1b trial of pretreated lymphoma patients [168,169]. Another group reported that AZD9150 increases drug sensitivity and decreases tumorigenicity in neuroblastomas [170]. Recruitment for AZD9150 trials in colorectal, pancreatic, and lung cancer is ongoing (NCT02983578) [171].

Although RNAi-based drug therapeutic trials have been ongoing for some time, it was only in 2018 that the Food and Drug Administration (FDA) approved the first RNAibased drug ONPATTRO, which is used to treat transthyretin amyloidosis. Due to a better understanding of the clinical development process required for RNAi therapeutics, more candidates are presently in development and trials, especially for cancer [172]. Selection and design of a delivery vector for RNA duplexes targeted toward FAM72 would be critical. Benayoun et al. have already demonstrated RNA silencing for FAM72, utilizing shRNA lentiviral constructs [4]. Alternatively, gRNA delivery via any of the methods above-mentioned could be performed to knockout FAM72.

#### *8.2. Therapeutic Options against Tumorigenic FAM72: CRISPR-Cas9*

An alternative mechanism to knockout FAM72 in cancer tissues would be to use the clustered regularly interspersed short palindromic repeats (CRISPR)-CRISPR-associated protein (Cas) 9 gene editing tool. Briefly, CRISPR and Cas target foreign viral DNA as part of the adaptive immune system in bacteria [173]. A combination of trans-activating RNA (tracrRNA) and CRISPR targeting RNA (crRNA), together known as small guide RNA (sgRNA or sg FAM72-RNA), guide Cas proteins to the targeted foreign viral (or tumorigenic FAM72) DNA, which is then degraded [174]. The sg FAM72-RNA in combination with the Cas9 protein from Streptococcus pyogenes form the popular CRISPR-Cas9 gene editing tool [175–177]. A nuclease deficient Cas9 (dCas9) system combined with a transcriptional repressor protein such as the Kruppel-associated box (KRAB) [178,179] that target the transcription start site for FAM72 would be ideal to knockdown FAM72 in vivo at the site of the tumor [179–183]. Since FAM72 is overexpressed in non-neuronal cancer tissues, such a system would only affect the cancer tissues, leading to greater specificity. The delivery mechanism could be via lipid nanoparticles, similar to siRNA (Figure 10) [184].

**Figure 10.** (**a**) Mechanisms of FAM72 knockdown using RNAi and CRISPR for the possible treatment of various types of cancer. Exogenous double-stranded RNA (dsRNA) or siRNA can be delivered via microinjection or lipid nanoparticles. The dsRNA or siRNA is released from the endosome after which it binds to the RNA-induced silencing complex (RISC). This complex then binds to the FAM72 mRNA, leading to the degradation of the whole complex. If shRNA is delivered via plasmid or viral vectors, the RNA is processed in the nucleus and exported into the cytoplasm. The Dicer enzyme processes shRNA into siRNA and then binds it to the RISC, followed by loading onto the target mRNA, and the resulting complex is degraded as before. Alternatively, the CRISPR-dCas9 with a transcriptional repressor protein is delivered via lipid nanoparticles. After entering the endosome, the CRISPR-dCas9 complex is released and it enters the nucleus. The Cas9 nuclease is directed to the target DNA by its bound sgRNA. Following binding of the dCas9 complex with the FAM72 target DNA, the repressor will attach to the transcriptional start site of FAM72, thereby resulting in a knockdown of transcription

and thus, prevention of spindle formation causing mitotic catastrophe followed by cell death. ASOs delivered into the cell binding directly to the mRNA transcript, resulting in RNAse degradation. Cas, CRISPR-associated proteins; CRISPR, clustered regularly interspersed short palindromic repeats; dCas9, nuclease deficient Cas9; dsDNA/RNA, double stranded DNA/RNA; KRAB, Kruppel-associated box; sgRNA, single guide RNA; shRNA, short hairpin loop RNA; siRNA, small interfering RNA. (**b**) Double-strand break positions of CRISPR/Cas9 application on the human FAM72A gene. The sgFAM72A#1, sgFAM72A#2, sgFAM72A#3, sgFAM72A#4 have been used to target and break within exon 1 of FAM72A to interrupt FAM72A gene transcription. sgFAM72A, single guide FAM72A: target DNA positions to be recognized and cleaved by the CRISPR/Cas9 system for FAM72A gene expression knockout.

#### *8.3. Therapeutic Options against Tumorigenic FAM72: Chemotherapy*

FAM72 and its paralogs could also be targeted via chemotherapy options using targeted drugs. We conducted an *in silico* binding study to predict potential ligand binding sites on FAM72A [185]. We found potential Zn2+ and Fe3+ binding sites along with possible binding for the organic compound RSM: (2s)-2-(acetylamino)-N-methyl-4-[(R) methylsulfinyl] butanamide) [185].

Structure-based drug design (SBDD) is rapidly growing with the development of new technologies (e.g., high-throughput screening, molecular docking, pharmacophore mapping, quantitative structure-activity/property/toxicity relationship (QSAR/QSPR/QSTR), and virtual screening) to interpret, guide, and advance experimental biomedical research to achieve success in anti-cancer drug discovery [186–190]. SBDD methods analyze threedimensional (3D) structures of macromolecule, typically of proteins or RNA, to identify key sites and interactions, which are important for their specific biological functions [187]. Understanding key sites and interactions can be used to design potential drug candidates that can interfere with essential interactions of the target protein and thus interrupt signaling pathways for survival and progression of cancer cells [187,191]. This requires knowledge of the 3D structure of the drug candidate and how its shape and charge cause it to interact with its biological target, ultimately revealing a therapeutic effect [187,192].

As discussed above in this review, increasing evidence indicates that FAM72 is a potential therapeutic target for the treatment of cancers [1,3,8,158], especially GBM [12] and ACC [28]. In essence, 3D protein structures and understanding ligand–protein interactions of FAM72 represent the key and even obligatory steps in FAM72-targated drug design for the development of a useful treatment for GBM and ACC. There is an urgent need to advance the FAM72-targeted drug design process, and we employed a comprehensive in silico 3D protein determination strategy to determine the 3D protein structure of FAM72A and further identify potential ligand–protein interactions of FAM72A (Figure 11) [185]. An integrated approach combining homology modeling and de novo modeling was applied to obtain a reliable 3D protein structure of FAM72A [185]. In the homology modeling, a homologous template search was performed in various databases (e.g., National Center for Biotechnology Information-Protein Data Bank (NCBI-PDB), Phyre2, 3D-JIGSAW, Swiss Model, and RaptorX) [185]. Additionally, 3D FAM72A protein structure models were also obtained from Phyre2, 3DJIGSAW, Swiss Model, and RaptorX tools. Furthermore, an optimized prediction with the Modeller program [193–195] using templates, 1YQ3\_D, 4OGC\_A, 4OGE\_A, 3GA3\_A, 3MCA\_B, 1I8D\_B, 4M0M\_A, 2FJA\_A, and 3UK7\_A (obtained from NCBI-PDB, Phyre2, 3D-JIGSAW, Swiss Model, and RaptorX) revealed that the monomeric 3D FAM72A protein structure, based on the 3GA3\_A template, was the most reliable model in terms of stereochemical parameter evaluations (i.e., G-factor, Ramachandran plot analysis, and additional comparative iterative threading assembly refinement (I-TASSER) analysis) [185]. To this end, protein-ligand binding site prediction based on BioLiP protein function database screening (based on COACH, TM-SITE, S-SITE, COFACTOR, and ConCavity methods) [196,197] revealed that FAM72A is a Zn2+- or Fe3+-containing protein, which could potentially interact with the organic molecule RSM (Figure 11) [185]. Taken together, these data suggest a theoretical view of the 3D structure model of FAM72A and its ligand-binding sites [185]. In our view, these structural and protein–ligand interaction data provide a basis of FAM72A protein ligand-binding sites, which require further investigation

of FAM72A-driven cancers (e.g., GBM and ACC) [185].

**Figure 11.** Structure based anti-cancer drug screening for the treatment of FAM72A-mediated cancers. Based on an in silico 3D protein structure model of FAM72A and its ligand-binding sites, the potential hit molecule RSM has been proposed for possible further therapeutic activity evaluations via in vitro and in vivo experiments.

#### **9. Conclusions**

The |-SRGAP2–FAM72-| master gene appears to be a pivotal genomic unit involved in brain development and synaptic plasticity. However, in light of the tissue-specific governance of this master gene, it remains to be seen what differentiates and regulates the expression of the |-FAM72–SRGAP2-| master gene across neuronal and non-neuronal tissues. This knowledge might be crucial for the specific biomedical interference with tumorigenic cell proliferation targeting FAM72.

using well-defined in vitro and in vivo experiments to confirm the therapeutic activity of the suggested compound as potential leads for drug discovery screenings for the treatment

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2072 -6694/13/5/1025/s1, Supplementary Figures S1–S21: Correlation of Fam72A mRNA expression with mRNA expression of MKI67, BIRC5 (survivin), E2F1, E2F2, E2F4, E2F6, FAM107A, FOXM1, LIN9, LIN37, LIN54, OLIG2, PAX6, RBBP4, TET2, TFDP1, TFDP2, TP53, SOX2, NRAS, and REST, respectively, across various TCGA human cancer tissues.

**Funding:** This study was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), which was funded by the Ministry of Education (2019R1F1A1056445). This work was partly supported by an Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korean government (MSIT) (No. 2020-0-01373, Artificial Intelligence Graduate School Program (Hanyang University)) and the research fund of Hanyang University (HY- 202000000700017, 202000000790014).

**Acknowledgments:** We would like to thank Hanyang University for providing a scholarship to N.T.T.H.

**Conflicts of Interest:** The authors have no conflict of interests to declare.

#### **References**


## *Review* **Glioblastoma: Emerging Treatments and Novel Trial Designs**

**Vincenzo Di Nunno 1,\*, Enrico Franceschi <sup>1</sup> , Alicia Tosoni <sup>1</sup> , Lidia Gatto <sup>1</sup> , Raffaele Lodi <sup>2</sup> , Stefania Bartolini <sup>1</sup> and Alba Ariela Brandes <sup>1</sup>**


**Simple Summary:** Nowadays, very few systemic agents have shown clinical activity in patients with glioblastoma, making the research of novel therapeutic approaches a critical issue. Fortunately, the availability of novel compounds is increasing thanks to better biological knowledge of the disease. In this review we want to investigate more promising ongoing clinical trials in both primary and recurrent GBM. Furthermore, a great interest of the present work is focused on novel trial design strategies.

**Abstract:** Management of glioblastoma is a clinical challenge since very few systemic treatments have shown clinical efficacy in recurrent disease. Thanks to an increased knowledge of the biological and molecular mechanisms related to disease progression and growth, promising novel treatment strategies are emerging. The expanding availability of innovative compounds requires the design of a new generation of clinical trials, testing experimental compounds in a short time and tailoring the sample cohort based on molecular and clinical behaviors. In this review, we focused our attention on the assessment of promising novel treatment approaches, discussing novel trial design and possible future fields of development in this setting.

**Keywords:** glioblastoma; newly diagnosed glioblastoma; recurrent glioblastoma; GBM; new trial design

## **1. Introduction**

Glioblastoma (GBM) is the most common primary brain tumor, with an estimated incidence of 3.22/100,000 persons in the United States and a five-year overall survival of only 6.8% [1,2]. Nowadays, GBM can be diagnosed as a diffuse astrocytic glioma without IDH and H3R gene mutations, with microvascular proliferation, necrosis, and/or peculiar molecular features such as TERT mutation, EGFR amplification, and/or gain of chromosome 7 combined with the loss of chromosome 10 [3–6]. According to the EANO guidelines for the diagnosis and treatment of diffuse gliomas of adulthood, isocitrate dehydrogenase (IDH)-mutated glioblastoma should be better defined as a grade 4 IDHmutant astrocytoma [6].

Current management of patients with GBM employs maximal safe resection surgery followed by radiation and chemotherapy [2,7–10].

Recurrent GBM can be managed by different approaches [11–13], including locoregional treatment and systemic treatments [2,14–19].

The prognosis of patients with GBM remains poor, with an estimated overall survival (OS) of 12–18 months from primary diagnosis and a life expectancy of 5–10 months after the diagnosis of recurrent GBM [20–22].

Since treatments provided are not curative, guidelines strongly recommend the patient's inclusion in clinical trials [2,23].

**Citation:** Di Nunno, V.; Franceschi, E.; Tosoni, A.; Gatto, L.; Lodi, R.; Bartolini, S.; Brandes, A.A. Glioblastoma: Emerging Treatments and Novel Trial Designs. *Cancers* **2021**, *13*, 3750. https://doi.org/10.3390/ cancers13153750

Academic Editor: Stanley Stylli

Received: 25 June 2021 Accepted: 21 July 2021 Published: 26 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/).

In the last decade, several novel discoveries about the molecular, genomic, and biological background of the disease have been determined. Nonetheless, none of these improvements translated into a significant progress in terms of therapeutic options. Indeed, several drugs and approaches showing promising results in early studies failed to confirm a clinical improvement on large randomized trials. Furthermore, the enrollment on clinical trials is limited, with only 10% of GBM patients being enrolled in a clinical study [24,25].

The purpose of the present paper is to investigate possible reasons related to the lack of therapeutic improvements on GBM, focusing on possible improvements in terms of trial planning and design. We also reviewed more promising experimental systemic treatments for patients in early phase of development, as well as in patients with newly diagnosed and recurrent GBM.

#### **2. Therapeutic Targets on GBM**

Several biological obstacles make the development novel effective drugs difficult [26,27]. These are represented by (1) the blood–brain and blood– tumor - brain barrier which makes the passage of therapeutic compounds difficult, (2) the extreme heterogeneity of the disease, and finally (3) the capacity to develop molecular mechanisms able to promote treatment resistance to antitumoral treatment. All these elements reduce the development of novel target agents. Nonetheless, the increasing knowledge of the molecular mechanisms related to disease development and progression has allowed the identification of several attractive targets for the systemic management of GBM (Table 1) [26,27]. The majority of these targets are represented by tyrosine kinase (TK) receptors.


**Table 1.** Clinical trials cited in the text. MGMT: methylation of the O(6)-methylguanine-DNA methyltransferase, TMZ: temozolomide.

The amplification of the epidermal growth factor receptor can be found in about 50% of GBMs [28], and several agents targeting this pathway have been investigated in GBMs (a discussion of treatments proposed for EGFR inhibition is included in Section 4. Recurrent GBM). Other than EGFR, some other TK receptors have gained particular interest.

Altered tumor vascularization is one of the hallmarks of the disease and there are at least two TK receptors whose inhibition could be associated with angiogenesis regression and tumor responses. These are the vascular endothelial growth factor receptor (VEGFR)

and the platelet-delivered growth factor receptor (PDGFR) [16,29]. Several small TK inhibitors (TKIs) targeting one or both of these two receptors have been tested without significant benefit [16]. Imatinib, pazopanib, cediranib, sunitinib, sorafenib, nintedanib, tivozanib, dovitinib, crenolanib, and cabozantinib are all oral TKIs that failed to show a significant clinical benefit on patients with GBM [16]. The mesenchymal–epithelial transition (MET) receptor is another pathway that could be activated in GBM cells [30]. Although the multi-target and MET inhibitor cabozantinib showed only a modest effect on GBM [31] (weighted by a high adverse events rate), the oral MET inhibitor capmatinib is under investigation for patients with GBM, in combination with bevacizumab (NCT02386826).

The epidermal growth factor receptor 2 (HER2) amplification is a driver molecule that is well-targeted by several target compounds in breast cancer. This receptor can be amplified also in GBM cells [32]. However, to date, no agents targeting HER2 have shown clinical efficacy on patients with GBM. Indeed, the oral inhibitors lapatinib and neratinib failed to show a significant impact on patients with GBM and in patients with brain metastases from solid tumors [33,34]. The novel oral TKI tucatinib has been shown to pass through the blood–brain barrier, reaching therapeutic concentrations in the brain [35]. Although this could be an effective treatment on patients with HER2-altered GBM, no trials are investigating this agent.

The management of several solid tumors has been revolutionized by the advent of immune-checkpoint inhibitors. Briefly, these agents can restore an inhibited immune response against tumors and are effective also on brain metastases from solid malignancies [36]. Their role will be further discussed in the next paragraph. Nonetheless, some other immunological approaches are assuming particular interest in the hematological and solid tumor treatment field [37]. Chimeric antigen receptor T cells (CAR-Ts) and chimeric antigen receptor macrophage (CAR-Ms) are surely two of the most enthusiastic approaches, involving the genomic recombination of T cells or macrophages which are oriented against tumor cells. Although there is little data regarding the safety and efficacy of this approach on GBM, several early phase studies are assessing this strategy on patients with GBM (NCT04077866, NCT04741984). Nonetheless, it is still unclear which could be the optimal cell manufacture and administration process. Of interest, some data are suggesting that CAR-Ms could be key strategies for GBM management, mainly thanks to the better penetration of the macrophage into the tumor-associated microenvironment (NCT04741984) [38].

Indoleamine 2,3-dioxygenase (IDO) 1 and 2 are catabolic enzymes involved in the degradation of tryptophan. IDO is supposed to promote a negative regulation of immune response, and it has the potential to inhibit both innate and adaptive responses against the tumor [39]. Although these agents have been already tested on GBM, the combinations of IDO inhibitors such as indoximod, epacadostat, BMS-986205, [40,41] and immune checkpoint, chemotherapy, and/or radiation treatment is under assessment in several different trials (NCT04047706).

Oncolytic viruses are reprogrammed viruses able to specifically target tumor cells, replicating and killing them [42–45]. Previous studies suggested a potential effective role of these agents against GBM in preclinical models and within early clinical studies [42–45]. Thus, several trials are testing these agents on GBM patients (NCT03294486, NCT03714334, NCT02062827).

#### **3. Newly Diagnosed GBM**

Since 2005, the post-surgical standard treatment of GBM is surgical resection followed by temozolomide (TMZ), concomitant with and adjuvant to radiotherapy (60 Gy over six weeks), leading to a median survival time of 14.6 months [9]. The benefit from TMZ is greater in patients who present MGMT promoter methylation, which epigenetically silences the gene [7].

Different improvements of the current protocol have been tested in recent years.

Two trials demonstrating an improvement in overall survival with standard treatment have not been fully incorporated in the actual therapeutic scenario for different reasons. Tumor-treating fields (TTFields) is an antimitotic treatment modality, which acts by delivering a low-intensity (200 kHZ) electric field within the brain, alternating electric fields to the tumor. Through this action, TTFields interferes with GBM cell division and organelle assembly. The efficacy of incorporating TTFields in the standard first line treatment has been explored in the EF-14 trial [46]. In this randomized trial, 695 GBM patients, after completed concomitant radio chemotherapy, were randomized to TTFields plus maintenance TMZ or TMZ alone. The addition of TTFields lead to a significant increase in PFS (6.7 vs. 4.0 months, HR, 0.63; 95% CI, 0.52–0.76; *p* < 0.001) and OS (20.0 vs. 16.0 months HR, 0.63; 95% CI, 0.53–0.76; *p* < 0.001) over standard treatment, without significant difference in adverse events. Despite the FDA approving TTFields for newly diagnosed GBM in 2015, the use in clinical practice remains limited (3–12% of patients with newly diagnosed GBM) due to patients declining to wear the device, combined with difficulty in understanding the mechanism of action, doubts about the favorable outcome of existing studies, and the high costs of the treatment (to date, this treatment strategy is mainly adopted by USA, Israel, and Switzerland).

The CeTeG/NOA-09 German trial has randomized 141 MGMT-methylated GBM patients to standard TMZ concomitant with and adjuvant to radiotherapy, or to six cycles of a lomustine and TMZ combination in addition to radiotherapy [47]. Median OS was 31.4 months in the TMZ group, compared to 48.1 months in the lomustine–TMZ group (hazard ratio (HR) 0.60; 95% CI, 0.35–1.03; *p* = 0.0492). There was no difference in terms of progression-free survival (PFS), while adverse events of grade 3 or higher were observed in 51% and 59% of patients in the TMZ group and lomustine–TMZ group, respectively. However, the study presents some significant limitations. First, the small cohort of patients limits the validity of the results and presents the possibility of biases. Furthermore, the low number of randomized patients is in contrast with the high number of MGMT-methylated screened patients, with an accrual rate of only 60%. Another interesting issue was the improvement in OS which was not associated to a PFS benefit. This was not observed in previous newly diagnosed GBM phase III trials [9,46], and was not explained by differences in subsequent treatments at recurrence/progression.

Moreover, no survival benefit has been demonstrated with TMZ dose-dense regimens [48] or with extension of maintenance treatment up to 12 cycles [49]. To further explore this setting, the ANOCEF group proposes a randomized trial (NCT03663725) comparing standard treatment versus an intensified arm consisting of one TMZ cycle started between day 2 and 15 after surgery, followed by TMZ concomitant to radiotherapy, followed by maintenance TMZ until progression, intolerance, the patient's or the physician's decision.

Given the potential role of hypoxia in the biology of GBM, the addition of antiangiogenic therapy with bevacizumab has been investigated in two large phase III randomized trials in the first line setting [50,51]. Despite prolonging PFS in both trials, the addition of bevacizumab failed to demonstrate an overall survival improvement. Moreover, bevacizumab was associated with an increase in adverse events.

The introduction of immune checkpoint inhibitors (ICIs) has recently revolutionized the therapeutic scenario in a number of different cancer types. ICIs act as inhibitors of immune-checkpoints, restoring an inhibited immune-response against the tumor. Two phase III clinical trials investigated nivolumab (a programmed death receptor-1 (PD-1) inhibitor) in combination with radiation therapy in patients with unmethylated MGMT GBM (CheckMate-498; NCT02617589), and in association with radiation therapy plus concomitant and adjuvant temozolomide in patients with methylated MGMT glioblastoma CheckMate-548; NCT02667587). Unfortunately, none of these trials showed significant improvement in terms of OS and PFS for patients receiving nivolumab.

Another immunotherapy first line phase II trial, PERGOLA (NCT03899857) is evaluating the addition of pembrolizumab to standard treatment in newly diagnosed GBM patients.

The ICIs combination with ipilimumab and nivolumab has been initially studied in exploratory phase I cohorts. In these patients there was a significant rate of high-grade adverse events, with a discontinuation rate due to toxicity accounting for 20.30 [52]; thus, this combination strategy has not being further assessed in the subsequent phase III trials. The ICI ipilimumab and nivolumab combination is now being retested in a phase II/III study in newly diagnosed MGMT unmethylated GBM patients, comparing the usual treatment with radiation therapy and TMZ to radiation therapy in combination with ipilimumab and nivolumab (NCT04396860).

The role of active immunotherapy via vaccine injection is being explored in the ongoing study of dendritic cell (DC) immunotherapy against cancer stem cells. In this study, newly diagnosed GBM patients are vaccinated during standard treatment with ex vivo generated DCs transfected with mRNA from autologous tumor stem cells, survivin, and hTERT.

Enzastaurin (enz) inhibits protein kinase C-beta, angiogenesis, and has a direct cytotoxic activity against glioma cells [53]. Previous phase II studies carried out in recurrent high-grade glioma and in newly diagnosed MGMT unmethylated GBM patients did not show any significant single-agent activity [54,55]. However, the recent discovery of a novel biomarker, de novo genomic marker 1 (DGM1), a germline polymorphism on chromosome 8, highly correlated with response to enz in both lymphoma and GBM [56], has prompted the clinical development of this drug. In particular, GBM patients with DGM1+ assessment receiving enz had a median OS of 18 months versus 12.8 months in DGM1− patients (HR 0.68; 95% CI, 0.25–1.81; *p* = 0.12). Given these data, a randomized double-blind, placebo-controlled phase III study of enz added to temozolomide during and following radiotherapy in newly diagnosed GBM with or without DGM1 has been recently launched in the US (NCT03776071).

The phase III trial EORTC 1709 evaluating the addition of marizomib, a novel brainpenetrant pan-proteasome inhibitor, to standard TMZ/RT→TMZ in newly diagnosed GBM has been prematurely closed by IDMC, after evidence of more frequent grade 3/4 treatmentemergent adverse events compared to the standard therapy group (42.6% vs. 20.5%), including ataxia, hallucinations, and headache. The study did not show a significant impact on OS or PFS over standard treatment in [57].

Adaptive platform trials allow the testing of several experimental drugs at the same time, developing a more efficient and cost-effective mechanism for accelerating treatment approval for patients. In the neuro-oncology field, the GBM AGILE study (NCT03970447) is evaluating several experimental compounds on patients with newly diagnosed and recurrent GBM, tailoring each experimental arm according to the molecular assessment of the disease. AGILE opened for patient enrollment in 2019, and site activation is ongoing in the US, whereas expansion to Canada, Europe, and China are under progress. The trial is evaluating a new treatment arm using regorafenib, paxalisib, and VAL-083 in maintenance period in newly diagnosed GBM after concomitant treatment [58].

Despite available treatments, GBM inevitably recurs, demonstrating a poor overall prognosis with a two-year survival rate of less than 20%. Nevertheless, it should be highlighted that a small proportion of patients achieve a long survival of over three years, but the molecular prognostic and predictive background dividing long-term (LTS) from shortterm survivors (STS) is still poorly understood. Nonetheless, some studies investigated the clinical and molecular behaviors of LTS. Overall, LTS were younger at diagnosis, female, and presented MGMT methylation. The sphingomyelin metabolism was also increased in these patients [59–61]. With the aim to understand biological background of LTS, EORTC is conducting the EORTC 1419 Eternity trial (NCT03770468). This prospective and retrospective multicentric clinical epidemiological study will evaluate the molecular genetics, and host-derived and clinical determinants of GBM patients with an overall survival of more than five years.

#### **4. Recurrent GBM**

Effective treatment options are limited, and new therapeutic strategies are desperately needed. As of yet, nitrosoureas are still considered the standard of care for recurrent GBM. Several tyrosine kinase inhibitors (TKIs) and monoclonal antibodies (mAbs) targets have been investigated in the last few years with limited results [14,15,17,18,62–65], while many others are in clinical development in recent clinical trials.

About 50% of all GBM patients present an amplification of the epidermal growth factor receptor (*EGFR*) gene which represents a driver mutation in GBM. Most frequent EGRF mutations are represented by EGFRA289D, EGFRA289T, and EGFRA289V [28]. Nonetheless, agents targeting this receptor failed to show a significant survival impact on patients with GBM [66–68].

Recently, depatuxizumab mafodotin (depatux−m, ABT414), an antibody-drug conjugate that consists of an antibody directed against EGFR and EGFRvIII, conjugated to a toxin (monomethyl auristatin F), was evaluated in the INTELLANCE-2/EORTC\_1410 randomized phase II study [69]. Patients receiving depatux-m and TMZ had a trend towards improved survival (primary analysis: HR 0.71; 95% CI, 0.50–1.02; *p* = 0.06; second follow up analysis: HR 0.66; 95% CI, 0.48–0.93; *p* = 0.024), corresponding to a median OS difference of 9.6 months (deatux-m + TMZ) versus 8.2 months (TMZ). The presence of EGFR single-nucleotide variations (SNVs) was shown to predict an improved outcome in the depatux−m + TMZ arm. These SNVs result in a receptor that is hypersensitive to low-affinity EGFR ligands, which can explain the increased activity of depatux−m and TMZ [70].

Antiangiogenic approaches have been investigated since 2007, with bevacizumab being the most studied agent [14,19,71–78]. Despite promising results in terms of progressionfree survival across multiple studies, these results did not translate into an overall survival benefit in the randomized phase III EORTC 26101 trial that compared bevacizumab and lomustine with lomustine alone (9.1 vs. 8.6 months, hazard ratio for death, 0.95; 95% CI, 0.74–1.21; *p* = 0.65)

More recently, regorafenib, an oral multi-kinase inhibitor targeting VEGFR-1, -2, -3, TIE 2, PDGFR, FGFR, KIT, RAF-1, RET, and BRAF has been investigated in the randomized phase II trial REGOMA, which has been approved for the management of recurrent glioblastoma by the EMA (European Medicines Agency) [17]; this trial showed a median OS of 7.4 months in the regorafenib arm vs. 5.6 in the lomustine arm. Thus, this agent has been included in other ongoing trials (i.e., the AGILE study) or in combination with other agents (i.e., Nivolumab, NCT04704154). Alteration of the cyclin-dependent kinase 4–6 (CDK4–6) pathway is a common event in GBM. A phase II trial evaluated the role of palbociclib in recurrent GBM patients with RB1 proficiency. Despite adequate penetration in tumor tissue, palbociclib showed limited activity with a median PFS of 5 weeks and a median survival of 15.4 weeks [79]. Similarly, in another phase II trial in patients with recurrent GBM and with evidence of CDKN2A/B loss and intact RB, abemaciclib showed a six-month PFS of 9.37% (95% CI, 2.4–22.7%), a median PFS was 55 days (95% CI, 49–56 days), and a median OS of 384 days (95% CI, 228–488).

Larotrectinib is a selective TRK inhibitor that showed an impressive response rate and also durable disease control in GBM patients. The study [80], presented at the 2019 ASCO meeting, evaluated 18 cases with primary brain tumors, including six (32%) patients with GBM. A disease control rate was achieved in 100% of patients (in 14 evaluable patients), with a disease control rate ≥16 and 24 weeks in 79% and 71% of patients, respectively; the median PFS was 11 months (95% CI, 2.8–Not Reached). At the recent 2021 ASCO meeting, data regarding larotrectinib suggested that better results were obtained in pediatric patients with brain tumors, while no partial responses were seen in adult glioma patients.

Agents targeting BRAF inhibit the downstream altered MAPK pathway, which is often altered in solid tumors and is also an important driver of cell proliferation in glioma patients. V600E is the most frequent mutation in the BRAF gene described in gliomas, occurring in about 5% of adults [81]. Vemurafenib and dabrafenib, selective oral tyrosine kinase inhibitors of the oncogenic BRAF V600 kinase, have been tested in BRAF mutant melanoma patients. The role of vemurafenib in BRAF V600-mutant gliomas has been investigated in the VE-BASKET trial [82], which evaluated 24 patients (six GBM, five anaplastic astrocytoma, one high grade glioma not otherwise specified, and twelve with other histologies). For high-grade glioma patients, the response rate was 9%, the median PFS was 5.3 months, and the median survival was 11.9 months. Combined inhibition of BRAF and MEK in gliomas was also investigated in the ROAR basket trial [83]; in the group of high-grade gliomas, response rate was 27%, and the disease control rate was 57%.

Immunotherapies have also been investigated in recurrent GBM. The Check-Mate-143 trial evaluating nivolumab (a PD-1 inhibitor) versus bevacizumab in recurrent GBM was negative in the general population [18]. Nonetheless, the response duration was longer in the nivolumab (11.1 months) arm as compared to the bevacizumab arm (5.3 months). The corticosteroid use did not impact survival in the bevacizumab arm, while reduced doses were associated with an improved clinical outcome in the nivolumab treatment arm (HR, 0.59; 95% CI, 0.36–0.95). Moreover, a trend toward a longer survival was observed in MGMT-methylated patients without any baseline corticosteroids receiving nivolumab over bevacizumab (17.0 vs. 10.1 months; HR, 0.58; 95% CI, 0.30–1.11).

Pembrolizumab was also evaluated as a "neoadjuvant" treatment for recurrent GBM in an Ivy Foundation Early Phase Clinical Trials Consortium randomized study. Cloughesy and Colleagues evaluated the survival and immune response obtained when using pembrolizumab before and/or after surgery in 35 recurrent GBM patients [84]. Patients in the "neoadjuvant" arm with continued adjuvant therapy following surgery reported a significant increase in survival compared to patients treated with pembrolizumab only after surgery, with a median survival of 13.7 months in the "neoadjuvant/adjuvant" arm vs. 7.5 months in the "adjuvant"-only arm (HR: 0.39; 95% CI, 0.17–0.94; *p* = 0.04). Interestingly, treatment with pembrolizumab before surgery was associated with upregulation of T celland interferon-γ-related gene expression, but downregulation of cell cycle-related gene expression within the tumor.

Another immunotherapy approach consists of vaccination against EGFRvIII, a GBMspecific EGFR driver mutation [85]. Rindopepimut in combination with bevacizumab, or a control injection of keyhole limpet hemocyanin in combination with bevacizumab, were investigated in a randomized phase II trial in recurrent EGFRvIII-positive GBM patients. The primary endpoint was PFS at six months, which was 28% for rindopepimut and 16% for the control (*p* = 0.12); the analysis of survival, a secondary endpoint, showed a statistically significant advantage in the rindopepimut–bevacizumab arm (HR 0.53; 95% CI, 0.32–0.88; *p* = 0.01). Additionally, in a randomized phase III study investigating rindopepimut in patients with newly diagnosed GBM, this agent did not improve OS compared to the standard of care [86].

Another immunotherapy approach consists of active immunization (i.e., dendritic cells or peptide vaccines). Dendritic cells (DCs) are antigen-presenting cells able to induce adaptive immunity. Due to promising results from a phase III trial in a newly diagnosed setting [87] with an autologous tumor lysate-pulsed dendritic cell vaccine (DCVax®-L), similar approaches are now under investigation in phase III trials in the recurrent setting GBM (NCT04277221).

#### **5. Problematic Issues on Interventional Trials: The Glioblastoma Paradox**

The lack of therapeutic improvements in the last years appears even more disappointing considering the increasing scientific understanding of the disease and the large availability of novel potential active compounds to test.

This paradox makes GBM a unique disease in which the availability of key molecular and biological insight does not translate into the development of new drugs. The presence of the blood–brain barrier, the heterogeneous and complex biology of the disease, and the lack of sufficient investment are possible explanations of this failure. Nonetheless, some concerns emerge about the modality by which these novel compounds are tested therefore the clinical trial landscape (Table 2).


**Table 2.** Challenges and innovations of trial design planning for patients with glioblastoma.

In 2018, Vanderbeek A.M. et al. published the results of a survey of clinical trials reported on clinicaltrials.gov, including GBM patients in the United States from 2005 to 2016 [25]. Interestingly, they reported over 400 clinical trials of which the majority were represented by phase I/II and phase II studies (60%) [25].

Of note, the authors found a very high rate of uncompleted and terminated trials with one to ten studies concluded due to lack of accrual, funding, or futility (no clinical advantage emerging at early assessment) [25]. Moreover, there was a median time to study completion of three to four years in phase II studies. These data appear even more surprising considering that only 5 of 249 phase I/II and phase II trials were randomized. Phase III trials were a minority, representing only 7% of all clinical trials assessed. Twelve of sixteen phase III trials were supported by a previous phase II study, and the overall population enrolled in these trials represented 26% of the total population assessed on clinical trials between 2005 and 2016 [25].

The authors concluded that only one to ten (8–11%) patients entered into clinical trials, which is a very frustrating result considering the rate of terminated trials due to lack of accrual [25].

Another well-known problematic issue related to interventional trials on GBM is the weakness of surrogate efficacy endpoints [50,51,95,99,100]. Indeed, progression-free survival (PFS) and overall response rate (ORR) is successfully adopted in clinical trials assessing novel compounds on solid malignancies as they provide a reliable prediction of other outcomes of interest, such as clinical improvement and overall survival (OS). The use of surrogate endpoints of OS could be important as they can reduce the time of the study. Nonetheless, the relationship between OS and surrogate such as PFS and ORR is extremely uncertain on GBM as survival benefit cannot reflect the improvement of PFS or ORR [50,51,95,99,100], especially in the case of antiangiogenic treatments. The postprogressive survival is a composite outcome, which has been assessed in a large series of over a thousand patients with GBM, and represents an interesting surrogate endpoint [101]. The research of reliable surrogates of OS acquires great importance in the assessment of novel agents in GBM.

The availability of novel potentially active drugs is increasing as biological and genomic assessment of the disease becomes even more clear. Furthermore, it has been demonstrated that GBM is not a unique disease as its molecular behaviors can drastically modify the clinical presentation, progression, and response to treatment. The larger the availability of novel compounds, the higher the need for interventional trials. This can be complicated considering the low incidence of the disease.

To date, only a few patients benefit from interventional clinical trials. The rate of terminated study due to lack of accrual is relatively high even if patients are strongly required, considering the increasing availability of novel agents. Additionally, the time to study completion is long, requiring years due to the absence of reliable surrogate endpoints of overall survival. Finally, the distribution of patients could be unbalanced, since the majority of them are enrolled in phase III trials with a relatively small number of patients enrolled in early phase I and II studies. This can lead to an unpowered early efficacy study, exposed to the risk of unclear information. The result is the early termination of potentially active compounds and a further unsuccessful test (on phase III) of unactive drugs.

Excluding financial and biological problems related to the development of new effective compounds, these issues may represent a strong limitation to the clinical progress of GBM management.

#### *Improving Interventional Clinical Trials Design on GBM*

The primary field of improvement is represented by organizational improvement and the need for investment in the research of active compounds, trial planning, and patients on trial tutelage [88–90].

Patients with GBM should be referred to reference centers and the development of inter-center networks providing early information about active trials should be encouraged. Similarly, the participation of patients in clinical trials could be encouraged through facilities allowing patient mobility, permanence in the experimental center during the trial course and follow up, and job and economic safeguarding of patients and caregivers. These elements could reduce the number of early terminated trials, as well as increase the number of patients who could benefit from a clinical trial (Table 2).

From the organizational point of view, there are several fields of improvement of clinical trials in GBM [88–90].

The introduction of a comparator arm in phase II study has provided a more accurate estimation of the efficacy of the novel compound under investigation [91], however, again, the transition from a positive randomized phase II [14,15,17,18,64,68,73,86,102–105] trial with a limited number of patients to a large phase III trial was negative [77]. Nonetheless, early randomized studies require more time for their completion and a higher number of patients as compared to single-arm phase II studies. To avoid these limitations, the incorporation of Bayesian statistics in trial design is a winning strategy [92]. Classical trials test a hypothesis among a distinct population, in a study with a pre-planned sample size dimension which conditions the power of the study.

The hypothesis of the Bayesian model is not fixed, but its probability (for example to be true or false) is constantly modified during the study due to the increasing amount of data acquired. For example, the Bayesian adaptive randomized (AR) study can use the data accumulating in the course of the same trial to modify the treatment allocation according to the potentially more efficient interventional arms [92].

In 2012, Trippa L. et al. acquired data from different phase II trials assessing four different compounds. In their simulation, authors allocated these same patients into a Bayesian AR study, assessing the same interventional arms [92]. Results of this simulation were surprising, as the same findings of the previous phase II studies were confirmed without loss in statistical power and with a significantly lower number of patients required [92].

Nowadays, Bayesian AR is commonly adopted in clinical trial design and represents a significant improvement in terms of quality of the research due to the possibility of testing more treatments with a shared comparator arm, at the same time reducing the number of patients required.

Another commonly adopted strategy to overcome the need for a comparator arm, and thus the randomization, is the adoption of a historical cohort based on previous findings in clinical trials [91]. This strategy exposes the risk of several biases for different reasons. First, outcomes such as the survival of patients with GBM are not static values, as there is a trend showing increases in time even if there is no modification of treatment standards [91,106]. In addition, it has been well demonstrated that the inter-trial variability reflects a variable distribution of the outcome of interest, which significantly increases the risk of underestimating or overestimating the benchmarks [91]. This final result poses a very high risk of achieving false positive or negative observations in phase II trials, leading to a subsequent assessment of inactive compounds or the early termination of the study of an active drug [91].

Even if OS remains the best available clinical endpoint, the research of a novel surrogate endpoint is still a clinical need.

The PFS improvement failed to show an improvement in OS across different clinical trials [50,51,95,99,100]. Nonetheless, PFS expressed as the rate of patients progressing at a specified interval of time is commonly adopted in GBM clinical trials [50,51,95,99,100]. Again, Bayesian AR trials can offer a possible solution to this problem [96]. Thanks to the flexibility of the Bayesian AR trial, the incoming data provided in the course of the clinical trial can allow early determination of whether concordance between OS and PFS exists, therefore allowing, in case of concordance, decision-making results based on the assessment of PFS alone [96].

There are several problems related to response assessment in patients with GBM [107–110]. Indeed, response assessment must involve other data in addition to dimensional and imaging criteria. The type of treatment provided and the molecular background of the disease are mandatory elements to estimate response to treatment. Integration of molecular and clinical data with imaging improves ORR estimation; nonetheless, functional imaging provided by magnetic resonance imaging (MRI) and positron emission tomography (PET) is increasing as to allow a more reliable distinction of progression/response to treatment [107–113]. Criteria of response assessment have been modified and reflects the type of treatment provided [107,108].

Novel technologies are currently employing the use of artificial intelligence algorithms which can, based on the data provided and learned, assess the disease [114,115], and improve the use of this endpoint.

Innovative trials on GBM are represented by AGILE, INSIGhT, and N2M2 trials [93,94,98].

The Adaptive Global Innovative Learning Environment (GBM AGILE) is a novel, multi-arm, platform trial which is composed of two different statistical designs [93]. The first phase is a Bayesian AR stage in which several compounds are tested with a common control. Through this phase, the aim is to isolate the active compounds and determine the population in which this is expected to be more effective, preventing and reducing the number of patients receiving ineffective treatments. Regarding this last point, results of the experimental arm investigating the CC-115 compound within the INSIGhT trial have been recently reported [116]. Thanks to the adaptive study design of the INSIGhT trial, a reduced number of patients received the experimental treatment which showed significant toxicity and lack of clinical efficacy [116]. Once that a promising active compound has been established, it proceeds to the second phase which involves classical fixed randomization to confirm the result of the Bayesian step [93]. Other advantages of this platform are represented by the inclusion of novel compounds at any time during the study. In addition, biomarkers can be assessed during each phase of the study allowing a fast discovery and validation of prognostic/predictive biological markers [93].

In addition, the INSIGhT trials employed a Bayesian AR in the first step [94]. Different from AGILE, in the INSIGhT trial, only patients with newly diagnosed unmethylated GBM without the isocitrate dehydrogenase (IDH) R132H gene mutation have been included. A key inclusion criterion is also represented by complete genomic data for biomarker groupings [94]. This trial is currently testing three different compounds simultaneously and comparing them to the standard represented by radiation and adjuvant temozolomide [94]. Preliminary results of the abemaciclib treatment arm have been recently reported showing no OS advantage for patients receiving the CDK inhibitor [117].

The paradigm of the precision medicine era is the administration of drugs tailored based on the biological background of tumor disease. AGILE and INSIGhT offer the possibility to test more drugs rapidly, isolating the population where the novel agent is more effective.

The NCT Neuro Master Match (N2M2) offers a different solution, as the goal of this trial is to primarily identify the target population, then provide the drug which can result in a clinical improvement based on the biological background of the disease [98]. This is an umbrella trial for patients with unmethylated IDH wild-type GBM [98]. The design of the study is composed of two parts; the discovery phase provides a molecular and neuropathological assessment of the disease to detect predefined biomarkers for targeted treatments, while the treatment phase employs a stratification of the population based on the results obtained in the discovery phase. The Bayesian model is employed to provide continuous monitoring of toxicity in phase I, while the efficacy endpoint is represented by six-month progression-free survival [98].

Despite umbrella and molecular tailored designs being extremely attractive, it should be noted that GBM is a heterogeneous disease and that the isolation of potentially predictive biomarkers may not reflect a sensitivity to defined novel compounds.

One proposed type of trial that specifically aims to target agents is represented by ''phase 00 ' studies [97] (Figure 1).

Due to the protection offered by the blood–brain barrier (BBB) and the blood-tumor– brain barrier, several drugs failed to show a clinical effect on GBM. Phase 0 studies can rapidly assess pharmacological effects of the compounds on a patient's tumors, also discovering if and how much the compounds pass the BBB and penetrate the tumor tissue. Briefly, the study design requires that patients assume the study drug one to two weeks before preplanned surgery. After surgery, there is an in vivo assessment of tumor tissue, cerebrospinal fluid, and/or blood. Vogelbaum M.A. et al. recently reviewed all phase 0 and phase 0-like studies carried out between 1993 and 2018, establishing that phase 0 study in neuro-oncology should include patients in which tumor resection is planned, and involve clinical doses of the investigational agent, a tissue sample from each part of the tumor (including enhancing and non-enhancing portions of the tumor), and the assessment of specific drug-related target effects [97].

**Figure 1.** Phase 0 study overall design. **Figure 1.** Phase 0 study overall design.

#### **6. Conclusions**

Due to the protection offered by the blood–brain barrier (BBB) and the blood - tumor– brain barrier, several drugs failed to show a clinical effect on GBM. Phase 0 studies can rapidly assess pharmacological effects of the compounds on a patient's tumors, also discovering if and how much the compounds pass the BBB and penetrate the tumor tissue. Briefly, the study design requires that patients assume the study drug one to two weeks before preplanned surgery. After surgery, there is an in vivo assessment of tumor tissue, cerebrospinal fluid, and/or blood. Vogelbaum M.A. et al. recently reviewed all phase 0 and phase 0-like studies carried out between 1993 and 2018, establishing that phase 0 study in neuro-oncology should include patients in which tumor resection is planned, and Glioblastoma represents a clinical challenge for oncologists and researchers. The increasing knowledge of molecular mechanisms related to disease onset and progression has allowed the development of several novel compounds which should be assessed among clinical trials. The need to test more and more compounds at the same time led to the development of a next generation of trials adopting a Bayesian design. In addition, phase 0 trials can detect early and perform an in vivo assessment of drugs able to penetrate the tumor tissue, stopping further development of drugs unable to cross the blood–brain barrier. All these elements will surely contribute to the development of effective treatments against the disease, as well as to allow patients access to experimental compounds.

tumor (including enhancing and non-enhancing portions of the tumor), and the assessment of specific drug-related target effects [97]. **Author Contributions:** Conceptualization, V.D.N., E.F., A.T., and A.A.B.; writing original draft preparation, V.D.N., E.F., A.T., A.A.B., L.G., and S.B.; editing, R.L. and A.A.B.; supervision, R.L. and A.A.B. All authors have read and agreed to the published version of the manuscript.

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

Glioblastoma represents a clinical challenge for oncologists and researchers. The increasing knowledge of molecular mechanisms related to disease onset and progression **Conflicts of Interest:** The authors declare no conflict of interest.

involve clinical doses of the investigational agent, a tissue sample from each part of the

#### **References**

among clinical trials. The need to test more and more compounds at the same time led to the development of a next generation of trials adopting a Bayesian design. In addition, phase 0 trials can detect early and perform an in vivo assessment of drugs able to penetrate 1. Ostrom, Q.T.; Cioffi, G.; Gittleman, H.; Patil, N.; Waite, K.; Kruchko, C.; Barnholtz-Sloan, J.S. CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2012–2016. *Neuro Oncol.* **2019**, *21*, v1–v100. [CrossRef] [PubMed]

has allowed the development of several novel compounds which should be assessed

the tumor tissue, stopping further development of drugs unable to cross the blood–brain barrier. All these elements will surely contribute to the development of effective treatments against the disease, as well as to allow patients access to experimental compounds. 2. Wen, P.Y.; Weller, M.; Lee, E.Q.; Alexander, B.M.; Barnholtz-Sloan, J.S.; Barthel, F.P.; Batchelor, T.T.; Bindra, R.S.; Chang, S.M.; Chiocca, E.A.; et al. Glioblastoma in adults: A Society for Neuro-Oncology (SNO) and European Society of Neuro-Oncology (EANO) consensus review on current management and future directions. *Neuro Oncol.* **2020**, *22*, 1073–1113. [CrossRef] [PubMed]


## *Review* **Advances in Immunotherapy for Adult Glioblastoma**

**Chirayu R. Chokshi <sup>1</sup> , Benjamin A. Brakel <sup>1</sup> , Nazanin Tatari <sup>1</sup> , Neil Savage <sup>1</sup> , Sabra K. Salim <sup>1</sup> , Chitra Venugopal <sup>2</sup> and Sheila K. Singh 1,2,\***


**Simple Summary:** Therapy failure and disease recurrence are hallmarks of glioblastoma (GBM), the most common and lethal tumor in adults that originates in the brain. Despite aggressive standards of care, tumor recurrence is inevitable with no standardized second-line therapy. Recent clinical studies evaluating therapies that augment the anti-tumor immune response (i.e., immunotherapies) have yielded promising results in subsets of GBM patients. Here, we summarize clinical studies in the past decade that evaluate vaccines, immune checkpoint inhibitors and chimeric antigen receptor (CAR) T cells for treatment of GBM. Although immunotherapies have yet to return widespread efficacy for the majority of GBM patients, critical insights from completed and ongoing clinical trials are informing development of the next generation of therapies, with the goal to alleviate disease burden and extend patient survival.


**Citation:** Chokshi, C.R.; Brakel, B.A.; Tatari, N.; Savage, N.; Salim, S.K.; Venugopal, C.; Singh, S.K. Advances in Immunotherapy for Adult Glioblastoma. *Cancers* **2021**, *13*, 3400. https:// doi.org/10.3390/cancers13143400

Academic Editor: Stanley Stylli

Received: 9 June 2021 Accepted: 30 June 2021 Published: 7 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/).

**Abstract:** Despite aggressive multimodal therapy, glioblastoma (GBM) remains the most common malignant primary brain tumor in adults. With the advent of therapies that revitalize the anti-tumor immune response, several immunotherapeutic modalities have been developed for treatment of GBM. In this review, we summarize recent clinical and preclinical efforts to evaluate vaccination strategies, immune checkpoint inhibitors (ICIs) and chimeric antigen receptor (CAR) T cells. Although these modalities have shown long-term tumor regression in subsets of treated patients, the underlying biology that may predict efficacy and inform therapy development is being actively investigated. Common to all therapeutic modalities are fundamental mechanisms of therapy evasion by tumor cells, including immense intratumoral heterogeneity, suppression of the tumor immune microenvironment and low mutational burden. These insights have led efforts to design rational combinatorial therapies that can reignite the anti-tumor immune response, effectively and specifically target tumor cells and reliably decrease tumor burden for GBM patients.

**Keywords:** glioblastoma; immunotherapy; vaccine; immune checkpoint inhibitors; chimeric antigen receptor (CAR) T cells

## **1. Introduction**

Glioblastoma (GBM) remains the most aggressive and prevalent malignant primary brain tumor in adults [1]. Unchanged since 2005, patients undergo standard of care (SoC) that consists of gross total resection to remove the tumor bulk, followed by radiation therapy (RT) with concurrent and adjuvant chemotherapy with temozolomide (TMZ) [2,3]. Despite these aggressive therapeutic efforts, tumor relapse is inevitable, and patients face a median overall survival of 14.6 months and a 5-year survival rate of 5.5–6.8% [1,2,4]. A major contributor to treatment failure is intra-tumoral heterogeneity that gives rise to tumor cell populations distinct at the genomic, transcriptomic, proteomic and functional levels [5–9]. In addition to SoC, two therapeutics have received approval from the Food and

Drug Administration, including (1) an anti-vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab, and (2) tumor-treating fields that target proliferating tumor cells. However, these therapies have yet to be incorporated into SoC for GBM patients. liferating tumor cells. However, these therapies have yet to be incorporated into SoC for GBM patients. Emerging therapeutics for GBM have shifted towards reconfiguring the patient's im-

and Drug Administration, including (1) an anti-vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab, and (2) tumor-treating fields that target pro-

Emerging therapeutics for GBM have shifted towards reconfiguring the patient's immune system to generate an anti-tumor response. Here, we will summarize clinical findings and highlight promising preclinical studies of three major immunotherapeutic modalities designed to treat GBM, including vaccines, antibodies and chimeric antigen receptor (CAR) T cells (Figure 1). For a recent review of advances in oncolytic virotherapy for gliomas, refer to Rius-Rocabert et al. [10]. Given that resistance to SoC and disease relapse are inevitable for GBM patients, preclinical and clinical advancement of immunotherapeutic modalities, combined with recent insights into the tumor immune microenvironment, are poised to improve clinical outcomes for this patient population. mune system to generate an anti-tumor response. Here, we will summarize clinical findings and highlight promising preclinical studies of three major immunotherapeutic modalities designed to treat GBM, including vaccines, antibodies and chimeric antigen receptor (CAR) T cells (Figure 1). For a recent review of advances in oncolytic virotherapy for gliomas, refer to Rius-Rocabert et al. [10]. Given that resistance to SoC and disease relapse are inevitable for GBM patients, preclinical and clinical advancement of immunotherapeutic modalities, combined with recent insights into the tumor immune microenvironment, are poised to improve clinical outcomes for this patient population.

*Cancers* **2021**, *13*, x FOR PEER REVIEW 2 of 22

**Figure 1.** Overview of current immunotherapeutic modalities being investigated to treat GBM. (**a**) CAR T cells recognize antigens through a genetically engineered extracellular receptor which triggers intracellular T cell activation and degranulation upon antigen binding. (**b**) Inhibitors of immune checkpoint proteins prevent their attenuation of immune responses upon activation and exhaustion. (**c**) Vaccines expose antigen-presenting cells to tumoral antigens, stimulating a target-specific immune response. Boxes indicate therapeutic targets or mediators being pursued for each modality. CAR: chimeric antigen receptor; CTL: cytotoxic T lymphocyte. **Figure 1.** Overview of current immunotherapeutic modalities being investigated to treat GBM. (**a**) CAR T cells recognize antigens through a genetically engineered extracellular receptor which triggers intracellular T cell activation and degranulation upon antigen binding. (**b**) Inhibitors of immune checkpoint proteins prevent their attenuation of immune responses upon activation and exhaustion. (**c**) Vaccines expose antigen-presenting cells to tumoral antigens, stimulating a target-specific immune response. Boxes indicate therapeutic targets or mediators being pursued for each modality. CAR: chimeric antigen receptor; CTL: cytotoxic T lymphocyte.

#### **2. Vaccines 2. Vaccines**

Cancer vaccines function by exposing tumor-associated antigens to antigen-presenting cells (APCs), which activate immune effector cells to achieve an anti-cancer immune response. Several promising vaccines targeting both single and multiple antigens have shown varying degrees of clinical response (Table 1); however, vaccines for GBM have yet to translate to SoC. While GBM-specific targets are sparse, several have been identified that are expressed exclusively or enriched in tumor cells. Perhaps the most explored to date, epidermal growth factor receptor variant III (EGFRvIII) is a mutant version of the Cancer vaccines function by exposing tumor-associated antigens to antigen-presenting cells (APCs), which activate immune effector cells to achieve an anti-cancer immune response. Several promising vaccines targeting both single and multiple antigens have shown varying degrees of clinical response (Table 1); however, vaccines for GBM have yet to translate to SoC. While GBM-specific targets are sparse, several have been identified that are expressed exclusively or enriched in tumor cells. Perhaps the most explored to date, epidermal growth factor receptor variant III (EGFRvIII) is a mutant version of the EGFR receptor specificically-expressed in GBM and has been targeted extensively through a variety of immunotherapeutic efforts, including vaccination. Similarly, the

cytomegalovirus (CMV) tegument phosphoprotein 65 (pp65) and IDH1 (R132H)-mutant peptides are frequently and specifically expressed in GBM, in contrast to healthy brain tissues [11,12]. Vaccination strategies targeting these proteins have shown efficacy in clinical trials and often elicit strong immune responses; however, no targets identified to date are expressed on all GBM cells, likely allowing clonally driven recurrence to evade such treatments. In contrast, multi-targeted vaccines initiating an immune response to multiple tumor-associated antigens better address intratumoral heterogeneity; however, these treatments have shown limited clinical success.

Antigen presentation and the following activation and regulation of effector cells is another important process in achieving an effective immune response, which involves several proteins such as those mediating suppression of T cells, macrophages and other tumor-infiltrating lymphocytes. Current efforts acting on this front, such as antibodies against these suppressors, have shown preclinical promise but have fallen short in clinical trials. Additionally, success seems to vary greatly upon the combination of these inhibitors, underlining the importance of understanding and enhancing synergistic interactions among treatments.

*Cancers* **2021**, *13*, 3400


#### *2.1. Single-Target Vaccines*

Several vaccines have been developed for GBM targeting a single, tumor-specific antigen. One such vaccine is rindopepimut, a peptide vaccine targeting EGFRvIII which has been identified as a tumor-specific mutant expressed in roughly one-third of GBM specimens [27]. This protein enhances GBM tumorigenicity [28,29] and is highly immunogenic [30], altogether providing a promising target for immunotherapy. Early preclinical studies have confirmed its immunogenicity and shown it to be effective in mice [31]; however, the protein's heterogeneous and unstable expression leaves room for EGFRvIII-negative tumor cells to drive therapy resistance and recurrence. A series of phase II rindopepimut trials, named "ACTIVATE, ACT II and ACT III," have shown promise (NCT00643097, NCT00458601), achieving median survival times between 22 and 26 months [13–15]. To validate these findings, a large phase III, trial termed "ACT IV", was completed with 371 patients (NCT01480479); however, no survival benefit was seen among vaccinated patients compared to controls, with median survivals of 20.1 and 20 months, respectively [16]. Interestingly, patients with significant residual disease received a greater benefit from the vaccine, perhaps due to a greater antigen load. Patients in the trial also showed strong humoral immune responses, suggesting resistance to the therapy was enabled at least in part by the heterogeneity of EGFRvIII expression. Indeed, those who underwent post-treatment biopsies of the recurrent tumor in both control and vaccinated groups showed loss of EGFRvIII expression in a majority of patients. This loss of expression highlights the limitations of single-target therapies in a heterogeneous tumor and underlines the importance combinatorial therapies will have in the future [32]. Additionally, the improved survival of the placebo group compared to historical controls was surprising, and future trials should account for this difference or change in control performance over time.

The complex interplay among therapies and the immune response must also be considered. For instance, rindopepimut was given along with TMZ, which induces lymphopenia [33]. While an accompanying increase in regulatory T cells suggests this may hinder the response to rindopepimut, previous findings have shown it can enhance it [14]. An additional study on rindopepimut was completed in 72 recurrent GBM patients in a phase II trial, termed "ReACT" (NCT01498328), combining the vaccine with bevacizumab, a monoclonal antibody against VEGF that has been shown to enhance immune responses [34]. The trial showed improvement upon the ACT IV trial, with 20% of treated patients surviving for 24 months compared to 3% for control-treated patients, in addition to a potential for rindopepimut to be combined with bevacizumab [17].

Another promising vaccination effort is the CMV dendritic cell (DC) vaccine. While rare in the healthy brain, viral proteins and nucleic acids of CMV are present in approximately 90% of GBM tumors [11]. The implications of CMV in tumor initiation and therapy resistance are not well understood; however, these viral antigens pose a potential immunotherapeutic target specific to cancerous cells. Of these antigens, CMV pp65 is highly expressed in glioma tumors and is the main target of current CMV vaccination strategies, as it elicits a strong cytotoxic T lymphocyte response following infection [35]. The CMV pp65 DC vaccine consists of autologous DCs pulsed with pp65 RNA fused in frame with the human Lysosomal Associated Membrane Protein (hLAMP) gene shown to enhance antigen processing [36]. A series of large phase II trials were recently completed with the vaccine in patients with newly diagnosed GBM following SoC treatment.

The initial "ATTAC" trial (NCT00639639) and subsequent "ATTAC-GM" trial (NCT00639639) both showed long-term survival in approximately one-third of patients. The initial trial also revealed that pre-conditioning with tetanus-diphtheria (Td) toxoid significantly increased DC migration to the lymph nodes, which correlated with increased survival, leading to half of the pre-conditioned patients remaining progression-free >36.6 months post diagnosis [18]. The second trial instead administered dose-intensified TMZ (DI-TMZ) with the vaccination, as DI-TMZ-induced lymphopenia has previously been shown to enhance both humoral and cellular immune responses [37]. While DI-TMZ increased immunosuppressive regulatory T cells, the group had a median survival of 41.1 months, greatly exceeding matched historical controls [19]. Excitingly, four patients remained progression-free at 59–64 months postdiagnosis, and overall, the trial showed the vaccine to be effective at targeting GBM based on the presence of CMV pp65. A subsequent phase II trial termed "ELEVATE" is ongoing to validate the benefit of Td toxoid pre-conditioning on DC migration and to evaluate synergy among vaccination, Td toxoid pre-conditioning and the anti-tumor antibody basiliximab (NCT02366728). To date, the trial has confirmed increased migration of DCs to the lymph nodes following pre-conditioning; however, analysis of other aims is not yet complete [20].

Vaccines have also been developed targeting the IDH1 subtype of gliomas, consisting of the IDH1 (R132H)-mutated peptide, which is present in <15% of GBM patients [12]. The vaccine was previously found to be effective in a mouse model transgenic for human MHC class I and II with IDH1 (R132H), showing MHC class II presentation of the epitope and mutation-specific T cell and antibody responses [38]. A phase I clinical trial termed "NOA-16" (NCT02454634) was recently completed for the vaccine delivered concurrently with topical imiquimod, a myeloid-activating TLR7 agonist. Results of the trial were extremely promising, with 93% of grade III-IV glioma patients showing a vaccine-specific immune response and 84% surviving >3 years [21]. A second phase II trial called "RESIST" is underway, adjuvating the vaccination with granulocyte-macrophage colony-stimulating factor (GM-CSF) in combination with TMZ and Td toxoid (NCT02193347).

#### *2.2. Multi-Target Vaccines*

To treat a heterogeneous disease such as GBM, targeting a single antigen can lead to clonal evolution and drive resistance. One way of overcoming this is by targeting multiple antigens concurrently. Interestingly, the greatest progress in therapeutic development has thus far been observed for single antigen-targeting vaccines, likely due to tumor-specific expression of these antigens. Regardless, the importance of targeting the molecular heterogeneity of GBM tumors is well established, and several multi-targeted GBM vaccines have shown promising results, such as personalized neoantigen-based vaccination strategies [39]. One such multi-targeted vaccine is DCVax-L, a personalized approach to peptide vaccination that uses autologous, or patient-derived, DCs pulsed with resected tumor lysate to target a variety of tumor antigens. In rat models, the vaccine was found to significantly increase survival and T cell infiltration [40], leading to several clinical trials. In a phase III trial (NCT00045968), a subset of patients (*n =* 232) were vaccinated and given concurrent TMZ, while all patients (*n =* 331) were given the vaccine upon tumor recurrence. The overall study population had a median survival of 23.1 months, with a large group (*n =* 100) having a particularly long median survival of 40.5 months unexplained by any prognostic factors, suggesting clinical efficacy related to vaccination [22]. A trial is now ongoing in patients who were previously ineligible due to post-chemoradiotherapy progression or insufficient vaccine production (NCT02146066). As an alternative approach to pulsing DCs with tumor lysate, DCs pulsed with a synthetic cocktail of tumor-associated antigens have shown promising preliminary results, with 5 of 16 vaccine-treated GBM patients surviving 6 years post-diagnosis [41,42].

Vaccines relying on heat shock proteins (HSP) are also being explored for GBM treatment. There have been several trials investigating HSP vaccines for glioma, which consist of HSPs and tumor-associated peptides. These vaccines primarily rely on tumor-derived HSP glycoprotein 96 (gp96), which binds tumor antigens forming the HSP protein complex-96 (HSPPC-96). This complex mediates presentation of antigens in antigen-presenting cells and can bind different peptides for a multi-targeted approach. An initial trial of a multi-peptide HSPPC-96 vaccine with TMZ (NCT00293423) confirmed strong peripheral and local immune responses specific to HSPPC-96-bound antigens in 11 of 12 treated patients [23]. These responders had a median survival of 11.8 months post-vaccination and surgery compared to 4 months for the single non-responding patient, and in the phase II portion of this trial, patients showed a median survival of 10.7 months, significantly

exceeding controls [24]. Additionally, patients with pre-vaccination lymphopenia had decreased survival compared to those with higher lymphocyte counts, likely due to worsened immune function and thus decreased responses. Addressing this question and further validating effectiveness of this vaccine, another trial (NCT02122822) revealed those with strong tumor-specific immune responses indeed had longer median survival than those with weak responses (>40.5 months and 14.6 months, respectively), with the overall patient population reaching a median survival of 31.4 months and again exceeding controls [25].

Another phase II trial was recently completed with the HSPPC-96 vaccine and TMZ following SoC (NCT00905060), achieving a median survival of 23.8 months, further validating efficacy of this vaccine [26]. Interestingly, this trial found expression of the T cell-suppressing immune checkpoint PD-L1 in myeloid cells to be indicative of survival, with high expression leading to shorter survival as compared to patients with lower PD-L1 expression (18 months and 44.7 months, respectively). While a promising lead, no HSPPC-96 vaccines have been combined with anti-PD-L1 therapies to date. However, a trial is currently investigating the vaccine when combined with standard TMZ, radiotherapy and the antibody pembrolizumab targeting the PD-L1 receptor, which is ongoing (NCT03018288).

#### **3. Antibodies Modulating the Tumor Immune Microenvironment**

A complex system of stimulatory and inhibitory regulators functions to maintain immune homeostasis. An important part of this system is immune checkpoints, which regulate activation to avoid autoimmunity. Upon activation or exhaustion, several immune cells upregulate these inhibitory checkpoints, thus limiting the immune response. Cancer cells express immune checkpoint proteins as well, allowing them to suppress the anti-cancer immune response. As a result, antibodies against these checkpoints, known as immune checkpoint inhibitors (ICI), have shown success in several cancers such as melanoma and non-small-cell lung cancer [43], and several are being tested for GBM (Table 2). Of these antibodies, the greatest progress has been noted for ICIs blocking programmed cell death protein 1 (PD-1) and cytotoxic T lymphocyte antigen 4 (CTLA-4), which are expressed on T cells to inhibit T cell activation and killing of tumor cells [44,45].



#### *3.1. Immune Checkpoint Inhibitors*

PD-1 targeting antibodies pembrolizumab and nivolumab have been approved to treat various solid tumors [43]; however, widespread clinical efficacy in GBM has yet to be achieved. Combination of an anti-PD-1 antibody and radiotherapy has shown preclinical success in vivo [53], leading to the phase III CheckMate 143 trial of nivolumab (NCT02017717) comparing it to the approved VEGF-A inhibitor bevacizumab in recurrent GBM. The trial results showed a median survival of around 10 months for both groups and identical 12-month survival rates of 42% [46]. Additionally, preliminary safety data of an earlier cohort of patients revealed high toxicity of a previously considered anti-PD-1/anti-CTLA-4 combination arm [54], leading to the discontinuation of this dual ICI therapy. Nivolumab has also been explored in other combinations such as the phase III CheckMate 498 trial (NCT02617589) delivered with radiotherapy, as compared to SoC (TMZ and radiotherapy); however, the trial showed no survival advantage of nivolumab treatment with similar median survivals around 14 months for both groups. Another phase III trial, CheckMate 548 (NCT02667587), is combining nivolumab, radiotherapy and TMZ. While still ongoing, an announcement was made that the trial failed to meet its primary endpoints of overall survival and progression-free survival [47].

Pembrolizumab is another anti-PD-1 antibody currently in trial for treatment of gliomas. In a phase I trial of 24 recurrent, high-grade glioma patients treated with pembrolizumab, bevacizumab and hypofractionated stereotactic irradiation (NCT02313272), more than half the patients achieved significant responses, and median survival was 13.5 months [48]. However, another phase I trial of pembrolizumab with bevacizumab compared to pembrolizumab alone in recurrent GBM patients (NCT02337491) showed a median survival of 8.8 months and 10.3 months, respectively [49]. The reduced survival upon lack of radiotherapy emphasizes the potential synergy of radiotherapy with anti-PD-1 therapies.

The interplay among chemotherapy and ICIs can also impact therapeutic efficacy, with preclinical studies showing that the order, timing and administration of chemotherapy relative to anti-PD-1 therapy drastically alter responsiveness of GBM tumors [55]. Additional efforts have been made to enhance the anti-tumor response, including neoadjuvant ICI administration prior to surgery, which has enhanced and prolonged the anti-tumor immune response and increased survival in other cancers [56,57]. A phase II trial using this approach with pembrolizumab in recurrent GBM patients showed increased survival with neoadjuvant and post-surgery adjuvant treatment, as compared to post-surgery adjuvant-only treatment (13.2 months and 6.3 months, respectively) [58]. Neoadjuvant administration also led to an upregulation of T cell- and interferon-γ-related gene expression and down-regulation of cell cycle-related genes. In a similar phase II trial (NCT02550249), neoadjuvant nivolumab was shown to enhance chemokine expression, T cell receptor (TCR) clonal diversity among tumor-infiltrating lymphocytes (TILs) and immune-cell infiltration in the tumor; however, median survival of treated patients was only 7.3 months [50]. Interestingly, two patients in the neoadjuvant cohort had complete surgical resection and remained disease-free for 33.3 and 28.5 months, which was not explainable by any recorded prognostic factors.

CTLA-4 (CD152) is another ICI that reduces CD28 co-stimulatory signaling by competitively binding to its natural ligands CD80 and CD86, suppressing T cell stimulation. Anti-CTLA-4 therapy has been approved for several cancers [43], extending survival of glioma-bearing mice [59], and in combination with anti-PD-1 therapy, shown eradication of tumors in a majority of mice [60]. Clinical trials have recently begun assessing anti-CTLA-4 therapies in treating gliomas (NCT02311920, NCT02829931), though no trials have been completed with glioma patients to date.

PD-L1, the ligand of PD-1 regularly expressed on APCs, is also expressed in cancer cells and mediates suppression of tumor-infiltrating T cells. Anti-PD-L1 antibodies have been approved in other cancers [43]; however, their efficacy in gliomas remains poor. An ongoing phase II trial is evaluating the anti-PD-L1 antibody durvalumab with radiotherapy and bevacizumab in GBM (NCT02336165), with preliminary results of the recurrent, bevacizumab-refractory cohort showing only 36% survival at 5.5 months [51].

Another phase I trial is looking at a different combination of ICIs, treating recurrent glioma patients with durvalumab and an anti-CTLA-4 antibody (NCT02794883); however, no updates have been given. Combinations of the anti-PD-L1 ICI avelumab are also being investigated, with ongoing phase II trials testing combinations with both hypofractionated radiation therapy (NCT02968940) and chemoradiotherapy (NCT03047473). Previous trials have found low expression of PD-L1 in GBM, with the CheckMate-143 trial finding only 10 of 37 patients with evaluable PD-L1 expression showing ≥10% [54]. This inherently limits any PD-L1 targeted therapies and may partially explain poor clinical outcomes thus far.

LAG-3 is another immune checkpoint receptor expressed on exhausted T cells that negatively regulates T cell responses. While anti-LAG-3 therapies have shown preclinical success [61], LAG-3 is expressed in a small percentage of tumor-infiltrating lymphocytes [62], thus limiting the potential impact of these therapies on stimulating the immune response. Regardless, a phase I trial evaluating the anti-LAG-3 antibody "BMS 986016" is underway, assessing its efficacy alone and in combination with the anti-PD-1 antibody nivolumab in recurrent GBM patients (NCT02658981). A recent update revealed a median survival of 8 months for the anti-LAG-3 group and 7 months for the anti-LAG-3, anti-PD-1 combination group. The trial also assessed an agonistic antibody targeting the 4-1BB (CD137) immune checkpoint protein. 4-1BB is a co-stimulatory receptor expressed by T cells upon activation, which augments activation signaling. The anti-4-1BB group had a promising median survival of 14 months [52]; however, while preclinical investigations support this therapy [63,64], further trials with anti-4-1BB antibodies are required.

TIM-3 is a receptor expressed on lymphocytes that can suppress the immune response by inducing T cell exhaustion, such that expression of TIM-3 in GBM has been linked with poor patient prognosis [65]. Anti-TIM-3 antibody therapy for GBM has shown success preclinically in combination with anti-PD-1 therapy and stereotactic radiosurgery (SRS). SRS drives the release of antigens from the tumor, enhancing the immune response, which is further stimulated by concurrent checkpoint inhibitors. While neither anti-TIM-3 nor SRS alone prolonged survival of GBM-bearing mice, combining the two increased median survival from 22 to 100 days, an effect similarly obtained using an anti-TIM-3 and anti-PD-1 combination [66]. When combining all three treatments, 100% of mice were alive 100 days post-engraftment, revealing great synergy and prompting a phase I trial of this combinational therapy, which is underway (NCT03961971).

#### *3.2. Macrophage-Targeted Antibodies*

Response to ICIs varies among tumor types and may depend on immune infiltrates such as TILs. Recently, mass cytometry and single-cell RNA sequencing of patient tumor specimens from various ICI-responding and non-responding cancers, such as GBM, revealed enrichment of CD73-high macrophages in GBM, which persist through anti-PD-1 treatment and limit ICI efficacy by inhibiting T cell infiltration [67]. Prevalence of these CD73-expressing macrophages correlated with a low response to ICIs, and genetic perturbation of CD73 in mice improved efficacy of anti-CTLA-4 and anti–PD-1 combination therapy, which correlated with greater T cell infiltration. These results show a promising and novel immunotherapeutic target to combine with existing ICIs.

CD47 is an enzyme that suppresses macrophage activation through binding the signal regulatory protein α (SIRPα). CD47 is overexpressed in many tumors [68], allowing cancer cells to avoid phagocytosis. Anti-CD47 antibodies have been developed to shift macrophages to an immunostimulatory phenotype, promoting an anti-tumor response [69] and effectively reducing growth of several tumors [70,71]. Preclinical studies of anti-CD47 therapies for glioma have shown that, while anti-CD47 therapy is sometimes effective at stimulating glioma cell phagocytosis [72], chemotherapy and radiotherapy are synergistic with treatment and may be required to enhance phagocytosis and extend survival in mice [73,74]. This enhanced phagocytosis also leads to increased antigen cross-presentation and T cell priming [74], and anti-CD47 therapies have shown synergy with autophagy inhibition [75,76], as well as other ICIs and tumor-specific antibodies [77]. The potential for synergistic co-therapies sophisticates treatment with anti-CD47 antibodies, and effective combinations should be compared prior to therapeutic development efforts.

#### **4. Chimeric Antigen Receptor (CAR) T Cells**

Chimeric antigen receptor (CAR) T cells represent an efficacious form of adoptive T cell therapy, in which peripheral T cells are genetically engineered to express a fusion receptor protein (i.e., CAR) that recognizes and targets a tumor-specific or -enriched antigen. Rapid and rational evolution of receptor design has transformed the first-generation CAR composed of a ligand-binding domain, extracellular spacer, transmembrane domain and an intracellular signaling domain—that suffered from limited signaling strength to highly efficacious second- and third-generation CARs that incorporate one or more intracellular co-stimulatory domains, respectively, to initialize and sustain T cell signaling [78–81]. Irrespective of design principles, an antigen-bound CAR T cell activates a potent cytokine release and cytolytic degranulation response that kills antigen-expressing tumor cells and results in T cell proliferation [82]. CAR T cell therapy has been highly effective against hematological malignancies, achieving remission rates of up to 90% in patients with relapsed or refractory B cell malignancies with anti-CD19 CAR T cells [83]. However, widespread clinical responses of CAR T cells have yet to be seen for solid tumors, including GBM. Here, we summarize lessons learned from clinical evaluation of CAR T cell therapies in GBM patients, highlight promising preclinical candidates and discuss approaches to improving clinical efficacy.

Unlike hematological malignancies, CAR T cell therapy design and administration require unique considerations in the context of GBM, including factors such as intratumoral antigen heterogeneity, bypassing the blood–brain barrier (BBB) and exerting a potent antitumor response in a highly immunosuppressive microenvironment [84]. Two schools of thought have guided the delivery of CAR T cell therapy to the brain thus far, one which supports systemic intravenous administration, and the other prefers intracavitary or intraventricular dosing to bypass the BBB. Supported by reports of a dysregulated BBB in GBM patients [85,86], investigators evaluating CAR T cell therapies targeting EGFRvIII and HER2 preferred intravenous delivery of their modality [87,88]. Although no dose-limiting toxicities were observed for either modality when delivered intravenously, three grade 2–4 adverse events were possibly associated with HER2 CAR T cell therapy, including headache (*n =* 1) and seizure (*n =* 2). In contrast, intracavitary (or intratumoral) delivery of CAR T cells is not functionally restricted by the BBB. Using a reporter gene system, preliminary clinical evidence supports trafficking of intracerebrally administered anti-IL13Rα2 CAR T cells to the tumor region using [18F]FHBG PET-based imaging [89]. Intracavitary treatment of GBM patients with anti-IL13Rα2 CAR T cells resulted in no dose-limiting toxicities [90,91]. However, similar to intravenous delivery of anti-EGFRvIII CAR T cells, two grade 3 adverse events were associated with the treatment, including headache (*n =* 1) and a neurologic event (*n =* 1). Unfortunately, an empirical and clinical comparison among CAR T cell delivery routes has yet to be performed for GBM.

To varying extents, clinical studies have evaluated CAR T cells for GBM targeting interleukin-13 receptor subunit alpha-2 (IL13Rα2), human epidermal growth factor receptor 2 (HER2) and EGFRvIII (Table 3), with follow-up studies targeting IL13Rα2 and HER2 underway. In addition, investigators have initiated clinical studies to evaluate CAR T cells targeting matrix metallopeptidase 2 (MMP2) [92], B7 family member B7-H3 [93–95], CD147 and NKG2-D type II integral membrane protein (NKG2D) [96,97]. Here, we outline clinical advances in CAR T cell therapies for the treatment of GBM.


**Table 3.** Summary of clinical trials for CAR T cells against GBM.

#### *4.1. IL13Rα2-Specific CAR T Cells*

IL13Rα2 is a monomeric high-affinity receptor for interleukin 13 (IL13) that is enriched in GBM specimens compared to normal brain tissue [100,101]. In fact, IL13Rα2 expression correlates moderately with the mesenchymal signature [100], a subtype of GBM associated with greater proliferation, tumorigenicity and resistance to conventional chemoradiotherapy as compared to other subtypes [102,103]. Supported by these findings, IL13Rα2 CAR T cells were designed using a mutated IL13-zetakine binding domain (IL13.E13K.R109K), engineered to provide greater specificity for IL13Rα2 over IL13Rα1/IL4Rα and attached to a CD28 co-stimulation and CD3ζ signaling domain [104]. These IL13-zetakine CAR T cells were specifically and potently activated in the presence of IL13Rα2-expressing glioma cells, whereas no appreciable effect was seen in the absence of IL13Rα2 expression. Strikingly, a single intracranial injection of IL13-zetakine CAR T cells into mice with orthotopic glioma xenografts led to a robust decrease in tumor burden and increased median overall survival from 35 to 40 days in control mice to 88 days in IL13-zetakine CAR T cell-treated mice. These promising preclinical results led to the first-in-human pilot safety and feasibility study of IL13-zetakine CAR T cells in three patients with relapsed GBM [90]. In the study, IL13-zetakine CAR T cells were administered via an implanted reservoir/catheter system

and led to treatment-induced inflammation at the tumor site. Although this treatment was well tolerated and led to decreased expression of IL13Rα2, two grade 3 headaches and a grade 3 neurologic event were observed following CAR T cell administration. A mean survival of 11 months after relapse was noted for these three patients, with one patient surviving 14 months.

Following this study, the group engineered second-generation IL13-targeted CAR T cells with a 4-1BB (CD137) co-stimulation domain and a mutated IgG4-Fc linker to improve anti-tumor potency and increase T cell persistence, while improving the safety profile [91,105]. These reengineered IL13BBζ-CAR T cells were administered to a patient with highly aggressive recurrent GBM with multifocal leptomeningeal disease and high IL13Rα2 expression. Although intracavitary infusions of IL13BBζ-CAR T cells did not cause any grade 3 or higher toxic effects and inhibited disease progression locally, distal non-resected tumors and new tumors progressed. Prompted by distant disease progression, IL13BBζ-CAR T cells were delivered via intraventricular infusions and led to dramatic reductions of all tumors after the fifth infusion, with a 77–100% decrease in tumor burden, a systemic anti-tumor inflammatory response and an absence of systemic toxic effects, allowing the patient to return to normal life and work activities. Unfortunately, disease recurrence was observed after 7.5 months with tumor formation in new locations and decreased expression of IL13Rα2, elucidating a common antigen loss response to targeted therapies and advocating for rational combinational or adjuvant therapies. Recently, preclinical efforts to improve IL13Rα2-directed CAR T cell therapy have included the incorporation of an IL13Rα2-specific single-chain variable fragment (scFv) [106], complementary IL15 expression to enhance T cell effector function [107], characterization of the tumor immune microenvironment following CAR T cell therapy [108] and optimal selection of T cell subsets for sustained CAR activity [109].

#### *4.2. EGFRvIII-Specific CAR T Cells*

Expressed heterogeneously in ~30% of GBM specimens [110], investigators have engineered and evaluated EGFRvIII-targeted CAR T cells in two in-human trials. A phase I study of EGFRvIII-targeted CAR T cells, previously tested in orthotopic xenograft models of EGFRvIII+ glioma for efficacy and specificity to EGFRvIII over EGFR [111,112], was conducted in 10 patients with EGFRvIII+ recurrent GBM to evaluate safety and feasibility as the primary endpoints [87]. Although no subjects experienced dose-limiting toxicities, including systemic cytokine release syndrome, tumor regression was not observed in any patients based on magnetic resonance (MR) imaging. A median overall survival of ~8 months was noted after CAR T cell infusion, with one long-term survivor exhibiting stable disease for >18 months. Of 10 treated patients, 7 underwent tumor resection post-infusion, and analysis of tumor tissue indicated a decrease or ablation of EGFRvIII expression. A second phase I clinical trial leveraged a third-generation EGFRvIII-targeted CAR with 4-1BB and CD38 co-stimulation domains to conduct a dose-escalation study in 18 patients with EGFRvIII+ GBM [99]. No dose-limiting toxicities were observed with EGFRvIII-targeted CAR T cells until the highest dose of <sup>≥</sup>1010, at which point a patient developed acute dyspnea and experienced oxygen desaturation, eventually succumbing to severe hypotension. Despite efforts to increase CAR T cell persistence and tumor localization, no objective responses were noted using MR imaging, with 16 of 17 remaining patients showing signs of disease progression <3 months after infusion and a median survival of 6.9 months post-treatment. Interestingly, a single patient remained alive up to 59 months post-CAR therapy, and an additional two patients survived >1 year. In addition to further preclinical studies on thirdgeneration anti-EGFRvIII CAR T cells by multiple groups [113–115], recent studies have augmented their approach to increase efficacy and decrease toxicity, including an approach to combine anti-EGFRvIII CAR T cells with anti-EGFR bispecific T cell-engager (BiTE) antibodies to treat EGFR-positive/EGFRvIII-negative GBM [116]. There are bispecific antibodies, such as BiTEs, that are synthetic antibody structures that bind to two separate epitopes, with intentions such as bridging tumor-immune cell interactions or increasing

antibody specificity. An in-depth review of bispecific antibodies, including BiTEs, was recently presented by Lim et al. [117]. Moreover, investigators recently developed multiantigen prime-and-kill synNotch-CAR T cells that use a dual receptor circuit, the first of which detects EGFRvIII or a brain-specific myelin oligodendrocyte glycoprotein to induce expression of CARs against EphA2 and IL13Rα2 [118]. In comparison to constitutively active anti-EGFRvIII/EphA2/IL13Rα2 CAR T cells, synNotch-CAR T cells showed greater anti-tumor efficacy without off-tumor toxicity.

#### *4.3. HER2-Specific CAR T Cells*

The human epidermal growth factor receptor 2 (HER2), originally discovered as a tumor-associated antigen in breast cancer, is a transmembrane glycoprotein with an intracellular tyrosine kinase domain [88]. HER2 is a sparsely expressed antigen in GBM, detected in up to 17% of specimens and indicative of poor prognosis [119,120]. With promising preclinical results of a second-generation anti-HER2 CAR engineered with a CD28 co-stimulatory domain [88], a clinical trial was undertaken to treat 17 patients with HER2-positive GBM with virus-specific anti-HER2 CAR T cells [98]. Although no doselimiting toxicity was observed and CAR T cell persistence was noted up to 12 months post-infusion, no significant survival benefit was noted for treated patients with a median overall survival of 11.1 months.

#### **5. Discussion**

Immunotherapy has yet to significantly improve clinical outcomes for GBM patients, and clinical studies have been disappointing thus far. Here, we detailed clinical and preclinical advances in immune checkpoint blockade, vaccination strategies and emerging CAR T cell therapies for the treatment of GBM (Figure 1). Among the major hurdles to clinical efficacy are immense intratumoral heterogeneity [6,7], parallel modes of immunosuppression by tumor cells [121–123] and low mutational burden in GBM [124]. With these factors in mind, investigators and clinicians are shifting their focus to combinatorial and personalized treatment strategies to achieve synergistic effects, reduce treatment resistance and overcome immunosuppression.

Given their effectiveness in other cancers such as melanoma [125], ongoing clinical studies are combining ICIs with conventional chemoradiotherapy and experimental therapeutics to increase efficacy. A rational advancement of ICI therapy is co-targeting multiple immune checkpoints, with clinical trials initiated to test the following combinations in GBM: anti-CTLA4 and/or anti-PD-1 with TMZ in newly diagnosed GBM (NCT02311920), anti-CTLA-4 and anti-PD-L1 in recurrent GBM (NCT02794883), anti-LAG-3 and anti-PD-1 in recurrent GBM (NCT02658981), anti-IDO with anti-CTLA4 or anti-PD-1 in GBM (NCT02327078). In addition, hypofractionated stereotactic radiotherapy (NCT0289931, NCT02313272 and NCT02530502) and MRI-guided laser ablation (NCT02311582) are also being combined with ICI. As reviewed by Rius-Rocabert, Garcia-Romero, Garcia, Ayuso-Sacido and Nistal-Villan [10], oncolytic viruses are another form of immunotherapy that preferentially infect tumor cells, thereby activating the innate immune system and increasing T cell trafficking to the tumor bed. Based on promising preclinical data [126–128], clinical studies are evaluating a combination of adenovirus-based therapy DNX-2401 with anti-PD-1 blockade for recurrent GBM (NCT02798406). Furthermore, a preclinical study has confirmed the usefulness of an anti-PD-1 antibody at augmenting DC vaccination in glioma-bearing mice, showing a significant improvement in survival attributed to the strong T cell response enabled by ICI treatment [129]. Given that genetically engineered CAR T cells are exposed to the same immunosuppressive microenvironment as endogenous tumor-infiltrating lymphocytes, ICIs are being combined with CAR T cells to augment their performance. A phase I clinical trial is evaluating anti-IL13Rα2 CAR T cells as a single modality and in combination with ICIs Nivolumab and Ipilimumab (NCT04003649). Synergy among ICIs and other immunotherapeutic modalities will likely

play a key role in advancing future therapies through addressing the immunosuppressive nature of the tumor.

Although CAR T cell therapy is a newer adaptation for GBM treatment, advancements to increase its clinical utility are rapidly progressing. Currently, 12 clinical trials are recruiting GBM patients to evaluate CAR T cell therapy against B7 family member B7-H3 (NCT04385173, NCT04077866), CD147, HER2 (NCT03389230), IL13Rα2 (NCT04003649, NCT04661384, NCT02208362), matrix metallopeptidase 2 (MMP2; NCT04214392) and NKG2D (NCT04717999). Furthermore, a recent clinical letter outlined the administration of B7-H3 CAR T cells to a 56-year-old woman with recurrent GBM, highlighting a potent but short-term anti-tumor response in situ, absent of grade 3 or higher toxicities associated with CAR T cell infusion [94]. Unfortunately, target antigen heterogeneity was predicted as the reason for treatment failure, as noted previously for CAR T cell therapy targeting EGFRvIII and IL13Rα2 [87,91]. Additionally, novel therapeutic targets for CAR T cell therapy are quickly emerging, including antigens such as the disialoganglioside GD2 [130], CD70 [131,132], CD133 [133], carbonic anhydrase IX (CAIX) [134], EphA2 [135,136], podoplanin (PDPN) [137], chondroitin sulfate proteoglycan 4 (CSPG4) [138,139] and adhesion molecule L1-CAM (CD171) [140]. Of these antigens, EphA2 is part of the EphR receptor tyrosine kinase family that coordinates positioning and patterning during early development [141]. Given that EphA2 is overexpressed in GBM specimens, especially in post-therapy GBM stem-like cells [142], anti-EphA2 CAR T cells [135,136] may be suited to target GBM at tumor recurrence. While current trials are focused on targeting single tumor-associated antigens, this increased repertoire of targets will allow multiple antigens to be targeted concurrently to overcome intertumoral heterogeneity. This approach has yielded fruitful results in preclinical glioma models, as shown by the development of tandem CAR T cells that bind HER2 and IL13Rα2 [143], as well as trivalent CAR T cells targeting HER2, IL13Rα2 and EphA2 [144]. In fact, these trivalent CAR T cells were able to eradicate nearly 100% of tumor cells from multiple GBM samples.

In addition to tumor-targeted CAR T cells and ICIs, modalities acting on other parts of the tumor immune microenvironment may play a vital role in achieving effective antitumor responses in a clinical setting. We summarized macrophage-targeted antibodies in Section 3.2 of this article. Another approach stems from a recent study that found natural killer cell function to be altered upon tumor infiltration, showing impairing lytic function as a possible mechanism of tumor immune evasion [145]. Strategies aimed at restoring natural killer cell activity against GBM are being investigated and have shown preclinical promise.

#### **6. Conclusions**

Emerging trends towards rational combinatorial therapies are likely to include a systemic reignition of the tumor immune microenvironment. The continued discovery of novel tumor-associated and tumor-specific antigens, paired with the improvement of therapeutic modalities to increase efficacy and reduce toxicity, are necessary for the clinical efficacy of immunotherapies. Overall, a combinatorial therapy delivered at various stages throughout SoC may reliably improve clinical outcomes in GBM patients.

**Author Contributions:** Conceptualization, S.K.S. (Sheila K. Singh), C.V., C.R.C. and B.A.B.; data curation, C.R.C. and B.A.B.; writing—original draft preparation, C.R.C. and B.A.B.; writing—review and editing, S.K.S. (Sheila K. Singh), C.V., C.R.C., B.A.B., N.T., N.S. and S.K.S. (Sabra K. Salim); visualization, B.A.B. and C.R.C.; supervision, S.K.S. (Sheila K. Singh); project administration, S.K.S. (Sheila K. Singh); funding acquisition, S.K.S. (Sheila K. Singh). All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Brain Tumor Foundation of Canada, BioCanRX, the McMaster University Department of Surgery, and Brain Cancer Canada.

**Conflicts of Interest:** S.K.S. is a consultant for and owns shares of Century Therapeutics Inc. This company, however, played no role in the design or writing of this review article.

## **References**

