*Review* **Tumor and Peripheral Immune Status in Soft Tissue Sarcoma: Implications for Immunotherapy**

**Luana Madalena Sousa 1,2, Jani Sofia Almeida 3,4,5,6,7, Tânia Fortes-Andrade <sup>1</sup> , Manuel Santos-Rosa 3,4,5,6,7 , Paulo Freitas-Tavares 7,8, José Manuel Casanova 4,5,6,7,8 and Paulo Rodrigues-Santos 1,3,4,5,6,7,\***


**Simple Summary:** Soft Tissue Sarcomas are a rare and heterogeneous group of tumors, which have a characteristic complexity, leading to a difficult diagnosis and a lack of response to treatment. The aim of this review is to summarize the role of immune cells, soluble plasmatic factors, immune checkpoints; and the expression of immune-related genes predicting survival, response to therapy, and potential immunotherapeutic agents or targets in Soft Tissue Sarcomas.

**Abstract:** Soft Tissue Sarcomas (STS) are a heterogeneous and rare group of tumors. Immune cells, soluble factors, and immune checkpoints are key elements of the complex tumor microenvironment. Monitoring these elements could be used to predict the outcome of the disease, the response to therapy, and lead to the development of new immunotherapeutic approaches. Tumor-infiltrating B cells, Natural Killer (NK) cells, tumor-associated neutrophils (TANs), and dendritic cells (DCs) were associated with a better outcome. On the contrary, tumor-associated macrophages (TAMs) were correlated with a poor outcome. The evaluation of peripheral blood immunological status in STS could also be important and is still underexplored. The increased lymphocyte-to-monocyte ratio (LMR) and neutrophil-to-lymphocyte ratio (NLR), higher levels of monocytic myeloid-derived suppressor cells (M-MDSCs), and Tim-3 positive CD8 T cells appear to be negative prognostic markers. Meanwhile, NKG2D-positive CD8 T cells were correlated with a better outcome. Some soluble factors, such as cytokines, chemokines, growth factors, and immune checkpoints were associated with the prognosis. Similarly, the expression of immune-related genes in STS was also reviewed. Despite these efforts, only very little is known, and much research is still needed to clarify the role of the immune system in STS.

**Keywords:** soft tissue sarcoma; immune monitoring; immunophenotyping; cytokines; immune checkpoints; gene expression

**Citation:** Sousa, L.M.; Almeida, J.S.; Fortes-Andrade, T.; Santos-Rosa, M.; Freitas-Tavares, P.; Casanova, J.M.; Rodrigues-Santos, P. Tumor and Peripheral Immune Status in Soft Tissue Sarcoma: Implications for Immunotherapy. *Cancers* **2021**, *13*, 3885. https://doi.org/10.3390/ cancers13153885

Academic Editors: Dimosthenis Andreou and Joanna Szkandera

Received: 19 July 2021 Accepted: 30 July 2021 Published: 1 August 2021

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

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

### **1. Soft Tissue Sarcoma**

Soft Tissue Sarcomas (STS) are a heterogeneous group of diseases of mesenchymal origin. STS represent approximately 1% of solid tumors [1]. This group comprises over 50 different histologic subtypes that affect patients of all ages [2]. Although they can occur anywhere in the body, the most common anatomic sites are the extremities (60–70%) and the abdomen and retroperitoneum (20%) [3]. In addition to being highly heterogeneous in anatomical localization and histology, they are also heterogeneous in terms of molecular characteristics and prognosis [4].

STS diagnosis is mainly based on histological interpretations, including immunohistochemistry, cytogenetic, and molecular analysis [5]. However, due to their rarity and heterogeneity, the diagnosis is challenging and requires expert analysis [6]. Therefore, a consensus and reproducible diagnostic criteria are crucial. The WHO classification provides an organization by tumor type, considering morphologic, immunohistochemical, and genetic features [7,8]. This classification also stratifies STS according to clinical behavior into benign, intermediate locally aggressive, intermediate rarely metastasizing, and malignant [7,8].

The increased availability of genomic technologies has provided a better understanding of sarcoma biology. STS can be divided into two groups based on genetic profiles: STS associated with specific genetic alterations and STS with nonspecific and nonrecurrent genetic alterations [5]. The first group includes chromosomal translocations that produce chimeric fusion genes, often encoding aberrant transcription factors, oncogenic mutations, or recurrent gene amplifications. These alterations may be tumor-specific or shared by several histological tumors with different histomorphologies and behaviors. In contrast to the STS associated with specific genetic alterations, the second group tends to have complex karyotypes, such as changes in chromosome number, unbalanced translocations, genetic deletions, and amplifications [5]. Concerning etiology, even though the majority is unknown, there are some genetic predisposal syndromes, such as Li-Fraumeni syndrome, Von Recklinghausen disease, or RB1 tumor-suppressor gene mutations that can lead to STS. Environmental factors, such as ionization, radiation, and chemical exhibitors, may also promote these sarcomas [6].

For localized STS, surgical resection with or without radiotherapy is the standard treatment. Unfortunately, STS recurs frequently as a locally inoperable or metastatic disease. For a locally advanced or metastatic disease, the usual treatment is chemotherapy [9]. Single-agent anthracycline is the first-line therapy and, for the second-line treatment, trabectedin and eribulin have demonstrated efficacy for some subtypes of STS [4].

Despite the remarkable improvement in cancer diagnosis and treatment, many patients do not respond to therapy. This limited effectiveness of current strategies is often attributed to the complexity of the disease. That is, at least partly, supported by the complex microenvironment where the tumor is growing and defeating the immune system.

There is a growing interest in studying the immunological status of STS patients. The tumor microenvironment (TME) includes different populations of non-tumor cells, such as endothelial, stromal, cancer-associated fibroblasts and adipocytes, and immune cells [9]. The study of tumor-infiltrating and peripheral immune cells and mediators of the immune response may help to reveal the mechanisms related to tumor immunity. Moreover, such a study could identify potential biomarkers that favor an accurate prognosis, effective therapy response monitoring, and a refined approach to treatment. Recently, a transcriptomic analysis of >10,000 patients identified four distinct TME subtypes conserved across 20 different cancers: immune-enriched, fibrotic (IE/F); immune-enriched, non-fibrotic (IE); fibrotic (F); and immune-depleted (D). This TME subtyping strongly correlated with survival in most of the cancer types analyzed. The IE/F and IE TME were correlated with a better prognosis, while the F TME was linked to a worse prognosis. Furthermore, this study has also showed that patients with immune-favorable TME subtypes could benefit the most from immunotherapeutic approaches [10].

Concerning sarcomas, critical elements of peripheral blood and TME also play an essential role in predicting the response to therapy and are potential therapeutic agents or targets. Furthermore, a study from The Cancer Genome Atlas (TCGA) consortium proposed an association of the TME with prognosis in different STS histotypes [11]. Regarding the TME, the immune cells play an important role in controlling the progression of multiple tumor types. Nevertheless, in human STS, their characterization remains poorly defined. In a later study, Petitprez et al. developed a new classification and stratification of STS based on the composition of the immune microenvironment [12]. This classification was made up of five sarcoma immune classes with clearly different profiles and significantly different TME compositions. Each histological subtype was identified in each class, making it clear that the immune profile varies even between tumors with the same histology. This work also confirmed that the simplistic characterization of STS as "non-immunogenic" tumors does not apply to all, given that two sarcoma immune classes showed an elevated expression of genes specific to immune populations and the expression of immune-checkpoint-related genes. Furthermore, they also demonstrated that the immune microenvironment could be used to evaluate the prognosis and predict the response to immunotherapy.

The aim of this review was to summarize the prognostic and therapy response prediction value of immune cells, soluble plasmatic factors, immune checkpoints, and the expression of immune-related genes in STS patients, as well as their role in immunotherapeutic approaches.

### **2. The Role of Immune Cells in STS**

### *2.1. Tumor-Associated Macrophages*

Macrophages are vital innate immune cells present in tissues, and it has been suggested that they play a role in tumor development and progression [13]. They are differentiated by the local microenvironment into M1 or M2 macrophages, developing a pro- or antiinflammatory response, respectively. Macrophages that are differentiated by the TME are called tumor-associated macrophages (TAMs). Due to several factors, for example, IL-4 and IL-13, an M2-like differentiation occurs in the TME, which facilitates tumor immune escape and metastasis [14,15]. M2-like TAMs block CD8 T cell-mediated antitumor immune response either directly, through their expression of inhibitor ligands, such as the programmed death-ligand 1 (PD-L1), or indirectly, via the C-C motif chemokine ligand 22 (CCL-22)-mediated recruitment of regulatory T cells (Tregs). A recent study detected, through immunohistochemistry, M2-like TAMs in all STS samples, while M1 like TAMs were only found in a few tumors and in a low density [16]. The presence of TAMs polarized toward a pro-tumoral phenotype in all the STS samples analyzed supports the possibility of targeting TAMs for STS treatment. TAMs could also be used to predict the clinical outcome. In several tumor types, this prognostic significance has already been shown [17,18]. However, concerning sarcomas, little is currently known. Still, the high density of M2-like macrophages, expressing CD163, and M1-like macrophages, identified by CD68 staining, were both significantly correlated with a poor outcome in non-gynecologic leiomyosarcomas [19]. Later, Kostine et al. also evaluated M2 and M1-like macrophages, and only the M2 phenotype was associated with worse survival rates for leiomyosarcoma [20]. Similarly, in myxoid liposarcoma (MLS), high levels of TAMs were also associated with poor survival [21]. More recently, a study performed with different types of STS identified TAMs as a poor prognostic for local recurrence, confirming the negative prognostic value of TAMs [22].

### *2.2. Tumor-Associated Neutrophils*

Neutrophils make up a substantial proportion of the immune infiltrate in cancer, and their role has long been a matter of controversy. Similar to TAMs, in mouse models, it has been demonstrated that tumor-associated neutrophils (TANs) can retain some functional plasticity and can acquire different phenotypes based on specific features of the TME. In a TGF-β-rich environment, neutrophils usually acquire an N2 phenotype associated with a pro-tumor activity. On the contrary, in the presence of IFN-β or inhibition of TGF-β, neutrophils switch to an N1 profile, which is usually associated with anti-tumor activity. Although the tumor-promoting effects of N2 TANs have been demonstrated, human TANs remain underexplored [23]. Ponzetta et al. have shown that mice with profound neutropenia presented an earlier tumor development compared with wild-type mice [24]. Moreover, the adoptive cell transfer of neutrophils into sarcoma-bearing mice restores tumor growth to the level of the control group. These results prove that TANs are essential to restrain sarcomagenesis. The same study also showed a correlation between the high density of TANs and a better outcome in undifferentiated pleomorphic sarcomas (UPS). However, this correlation was not observed in other STS subtypes, such as dedifferentiated liposarcoma, leiomyosarcoma, and myxofibrosarcoma.

### *2.3. Tumor-Infiltrating Lymphocytes*

Tumor-infiltrating lymphocytes (TILs) are strong indicators of tumor immunogenicity. TILs have been described in various malignant tumors, including STS, and some studies support the influence of TILs on the progression of some tumors [25]. It was observed that most STS patients had low TIL infiltration. However, in STS, TILs have been only reported considering a few STS subtypes in limited sample size studies. For these reasons, although the presence of TILs and their impact on positive outcomes have been demonstrated in several sarcoma subtypes, these reports may not be representative of all STS [12,26].

### 2.3.1. T Cells

To explore the level of T cell infiltration in STS, two studies analyzed the expression profile of CD3E. The former suggested that T-cell infiltration could depend on the STS subtype and proposed that a highly mutated tumor type may have greater immunogenicity and a robust T-cell infiltrate [27]. In the latter, CD3E was highly expressed in some STS samples, such as rhabdomyosarcoma and alveolar soft part sarcoma, corroborating the idea that T-cell infiltration depends on the STS subtype [28].

CD8 T cells can mediate the lysis of neoplastic cells. For that reason, these cells are usually associated with a direct anti-tumor immune response. Furthermore, there is an influence of these cells on the clinical course of several types of tumors. However, the excessive and constant exposure of CD8 T cells to cancer antigens and inflammatory signals leads to a progressive loss of the T cell effector function; this is called "exhaustion". Exhausted T cells can be characterized by the presence of inhibitory receptors; PD-1 and LAG3 are among them [29]. The analysis of CD8 T cells in the TME, including their receptor repertoire, has been increasing, given the availability of new activating drugs [30].

CD4 T cells are also required for anti-tumor immunity. They comprise diverse subsets with different and sometimes opposing roles in TME, upregulating or downregulating the immune response. Regarding their anti-tumor activity, they are responsible for enhancing the cytotoxic function of CD8 T cells, increasing clonal expansion, functioning as antigenpresenting cells, for example [31,32]. Fresh tumors resected at surgery and analyzed by flow cytometry have shown a greater prevalence of CD4 than CD8 T cells in well differentiated and dedifferentiated retroperitoneal liposarcoma [30]. The majority of tumor-infiltrating lymphocytes were CD4 'helper' T cells, and most CD8 T cells expressed their programmed cell death protein 1 (PD-1). This information suggests that CD8 T cells have been triggered by tumor antigen but are suppressed.

On the contrary, D'Angelo et al. described a greater prevalence of CD8 than CD4 T cells in STS tumors [33]. Those tumors were more likely to express PD-L1 and PD-1, once more suggesting the inactivation of these cells. Another study analyzed the density of T cells in 28 tumors diagnosed as undifferentiated sarcoma [34]. They observed a positive correlation between the density of CD8 T cells and the density of macrophages. Since some studies have indicated that TAMs suppress the cytotoxic functions and chemotaxis of CD8

T cells in other tumors, it would be interesting to know whether TAMs also affect CD8 T cells in undifferentiated sarcomas [35,36]. T cells in other tumors, it would be interesting to know whether TAMs also affect CD8 T cells in undifferentiated sarcomas [35,36].

studies have indicated that TAMs suppress the cytotoxic functions and chemotaxis of CD8

*Cancers* **2021**, *13*, 3885 5 of 21

Several studies have been trying to correlate the frequency of immune cells with the prognosis in STS (Figure 1). An association between CD8 T cells and improved outcomes has been observed [26,37,38]. However, conflicting studies have also observed an association with poor outcomes [39]. Moreover, there are also other studies that state that there is no statistical significance in this correlation [34,40]. Concerning CD4 T cells, the controversy remains. Although in some studies, CD4 T cells have been associated with a positive outcome [40,41], the opposite, an association with a poor prognosis, has also been observed [33,39]. In addition, some studies do not observe any significant prognostic value [38]. These discrepancies between studies may be due to the differences in methodology, antibody clones, and cutoff values used [39]. Furthermore, studies have indicated that these cell frequencies vary between STS subtypes and treatments [42,43]. For these reasons, the differences in sarcoma subtypes and the limited size of patient cohorts may also explain the discrepancies in the results. Several studies have been trying to correlate the frequency of immune cells with the prognosis in STS (Figure 1). An association between CD8 T cells and improved outcomes has been observed [26,37,38]. However, conflicting studies have also observed an association with poor outcomes [39]. Moreover, there are also other studies that state that there is no statistical significance in this correlation [34,40]. Concerning CD4 T cells, the controversy remains. Although in some studies, CD4 T cells have been associated with a positive outcome [40,41], the opposite, an association with a poor prognosis, has also been observed [33,39]. In addition, some studies do not observe any significant prognostic value [38]. These discrepancies between studies may be due to the differences in methodology, antibody clones, and cutoff values used [39]. Furthermore, studies have indicated that these cell frequencies vary between STS subtypes and treatments [42,43]. For these reasons, the differences in sarcoma subtypes and the limited size of patient cohorts may also explain the discrepancies in the results.

**Figure 1.** Expression levels of immune cell subtypes, cytokines, chemokines, growth factors, and soluble receptors and their prognostic value in STS. The TME has been associated with the prognosis in several tumors. However, in STS, this association is still underexplored. Immune cells such as B cell, DC, TANs, and NK have been associated with a positive prognosis (green). On the contrary, TAMs, and some soluble factors: LIF, IL-8, HGF, IL-2R, VEGF, MCP-3, TNF-R, IL-6, and M-CSF, have been associated with a negative prognosis (red). The prognostic value of MDSCs, Tregs, CD4 T cells, and CD8 T cells is not clear yet (gray) [11,22,24,26,40,42,44–47]. **Figure 1.** Expression levels of immune cell subtypes, cytokines, chemokines, growth factors, and soluble receptors and their prognostic value in STS. The TME has been associated with the prognosis in several tumors. However, in STS, this association is still underexplored. Immune cells such as B cell, DC, TANs, and NK have been associated with a positive prognosis (green). On the contrary, TAMs, and some soluble factors: LIF, IL-8, HGF, IL-2R, VEGF, MCP-3, TNF-R, IL-6, and M-CSF, have been associated with a negative prognosis (red). The prognostic value of MDSCs, Tregs, CD4 T cells, and CD8 T cells is not clear yet (gray) [11,22,24,26,40,42,44–47].

### 2.3.2. B Cells

Recent data have shown that B cells can shape immune responses in tumors [48]. However, the association of these cells with disease prognosis has been a reason for disagreement.

In several tumors, it described an association with a good prognosis. However, the opposite has been reported, too [16]. In well-differentiated and dedifferentiated retroperitoneal liposarcoma, B cells were found, generally with a low frequency, in some of the tumors analyzed [30]. In 2011, it was suggested that B cells could be an independent favorable prognostic factor in STS patients with wide resection margins [40]. Later, the association of B cells with a good prognosis was supported by Tsagozis et al. [16]. This study also observed an absence of B cells in many tumor areas, corroborating previous works.

Recently, Petitprez et al. published an integrative analysis dedicated to B cells and their influence on sarcoma survival and immunotherapy response [12]. They found that B cells are a key discriminative feature of a group of patients with improved survival and a better response to PD-1 blockade therapy, confirming their role as a positive prognostic factor. In addition, Helmink et al. found that B cell markers were the most differentially expressed genes in the tumors of STS responsive patients versus tumors of patients that did not respond to immunotherapy [49]. This data confirmed once more the potential of B cells as biomarkers.

### 2.3.3. Natural Killer Cells

Natural killer cells (NK) have the ability to lyse transformed cells [50]. Therefore, these cells play an important role in cancer immunosurveillance [51]. Studies of other tumors, such as clear cell renal cancer and non-small cell lung cancer, have evaluated the role of NK cells in the TME and the relationship between the infiltration of NK cells and the clinical outcome [52–54].

There have been a few studies of the NK cell function in STS. One of them used flow cytometry to detect infiltrating NK cells, generally in a low density, in some welldifferentiated and dedifferentiated retroperitoneal liposarcoma tissues [30]. Another study analyzed the tumor immune microenvironment signatures of 206 STS patients [11]. Regarding NK cell infiltrate, they reported that these immune cells were the only cells to correlate significantly with better disease-specific survival (DSS) in several sarcoma types. Later, Judge et al. also correlated tumor-infiltrating NK cells with improved survival in STS [26].

Although NK cells display an even higher cytolytic activity compared to CD8 T cells, their cytolytic function may be drastically dependent on the balance of activating and inhibiting surface receptors [55]. One activating receptor, NKp30, was found to be particularly downregulated in peripheral and tumor-infiltrating NK cells in gastrointestinal sarcoma (GIST) when compared to the circulating NKp30+ NK cells of healthy volunteers [56]. Nevertheless, the levels of total NK cells were similar in GIST and healthy volunteers. These results highlight the importance of further studies focused on NK cell receptors, since they affect the functions of these cells without affecting their frequency.

### *2.4. Dendritic Cells*

Dendritic cells (DCs) also play an essential role in the immunological environment. The TCGA analyzed the immune cell infiltrates based on tumor gene expression signatures and showed a correlation between the presence of tumor-infiltrating DCs and improved DSS in UPS and myxofibrosarcoma [11]. Although there is a lack of studies concerning DCs in STS, this conclusion suggests an important role of antigen presentation in immune responses against these tumors.

### *2.5. Suppressor Cells*

### 2.5.1. Regulatory T Cells

Regulatory T cells (Tregs) are physiologically suppressive cells and play an important role in maintaining the homeostasis of the immune response. They can produce immunosuppressive cytokines such as interleukin 10 (IL-10) and tumor growth factor-β (TGFβ), they can express negative costimulatory molecules such as cytotoxic T-lymphocyteassociated protein 4 (CTLA-4), PD-1, or PD-L1, and they consume cytokine interleukin 2 (IL-2). These functions lead to an inhibition of T lymphocytes and the promotion of im-

mune escape [57]. Studies of other tumors have associated high density of tumor-infiltrating Tregs with a poor outcome. However, the opposite has also been demonstrated [58]. In STS, D'Angelo et al., using immunohistochemistry, observed a high density of tumor-infiltrating Tregs in 75% of STS patients, most of them of GIST histology [33]. Later, another study evaluated tumor-infiltrating Tregs by immunohistochemistry and showed an association between the increased infiltration of these cells and a poor prognosis in STS [44]. However, an association has also been found between a greater percentage of Tregs, analyzed by multiplex immunofluorescence, and a better outcome [59]. The same study also correlated the increased tumor-infiltrating Tregs with a better response to pembrolizumab, anti-PD-1 monotherapy. Despite this, it has also been suggested that Tregs are not associated with STS prognosis [26,38]. Due to these controversial results and the limited number of studies, the prognostic significance of Tregs remains undefined.

### 2.5.2. Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells (MDSCs) are another subset of suppressive cells that can facilitate tumor immune escape, impairing the function of T cells, NK cells, and DCs. These immature myeloid cells can be phenotypically divided into early-MDSCs (e-MDSCs), monocytic MDSCs (M-MDSCs), and polymorphonuclear MDSCs (PMN-MDSCs) [60,61].

A study performed by Highfill et al. sought to investigate whether there was an expansion of MDSCs in rhabdomyosarcoma, the most common soft tissue sarcoma of childhood [60]. They used mice bearing rhabdomyosarcoma and observed, by flow cytometry, an expansion of MDSCs, preferentially PMN-MDSCs, localized at the tumor site. It was demonstrated that PMN-MDSCs have an essential role in rhabdomyosarcoma immune escape. Preventing the trafficking of these cells to the tumor could also improve the efficacy of checkpoint blockade. The role of MDSCs in human STS tumors remains underexplored.

### **3. Soluble Factors: Cytokines, Chemokines, Growth Factors, and Others**

The network of pro- and anti-inflammatory cytokines and chemokines orchestrates the immune cell signaling and function and, as such, largely contributes to the complexity of the TME. Cytokines have been studied in a broad range of tumors, and their involvement in cancer development, progression, and recurrence has been suggested. Moreover, the cytokine profile might be a prognostic factor for clinical outcome [62,63]. The prognostic value of cytokines, chemokines, growth factors, and soluble receptors in STS is summarized in Figure 1.

As well as cytokines, chemokines have multifaceted roles in tumor development and progression, promoting malignancy or restricting tumor growth [64]. Likewise, growth factors and soluble receptors also play a significant role in TME [65,66].

Preliminary studies have found an elevated serum level of some cytokines, growth factors, and immune-related soluble receptors in patients with STS. Higher serum levels of vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) have been reported. They promote angiogenesis, facilitating the tumor's growth and increased metastatic spread. Furthermore, VEGF also promotes the proliferation of immunosuppressive cells and T cell exhaustion, contributing largely to immune escape and cancer development [67–69]. In addition, increased serum levels of interleukin 6 (IL-6), receptors for TNF (TNF-RI and TNF-RII), interleukin 2 receptor α (IL-2Rα), interleukin 10 (IL-10), macrophage-colony stimulating factor (M-CSF), and interleukin-8 (IL-8) were also found in STS patients [45,70,71].

Rutkowski et al. analyzed the serum levels of 13 cytokines and soluble receptors in STS patients before treatment [45]. The results confirmed the elevated levels of VEGF, FGF, IL-6, TNF RI, TNF RII, IL-2Rα, IL-10, M-CSF, and IL-8 stated above. Furthermore, they tried to correlate the serum levels of these cytokines with clinic-pathological features. IL-2Rα, TNF RI, M-CSF, and VEGF correlated with tumor size, IL-8 was associated with tumor grade, and IL-6 appeared to be correlated with tumor size, grade, and metastases. Additionally, it was proved that IL-6 and IL-8 were correlated with decreased survival [45].

In relation to IL-6, a few more studies have confirmed the association of its serum levels with survival. Hagi et al. observed high levels of IL-6 associated with the presence of STS and proposed that IL-6 could be used as a marker for the differential diagnosis [72]. Furthermore, they confirmed the correlation between elevated IL-6 serum levels and decreased survival [69].

Wysoczynski et al. proposed that leukemia inhibitory factor (LIF) promotes the progression and the metastatic behavior of rhabdomyosarcoma cells, contributing to the resistance of rhabdomyosarcoma to conventional treatment [46]. Later, Wysoczynski found that IL-8 was a pivotal pro-angiogenic factor in rhabdomyosarcoma cells during hypoxia [73]. Still, in rhabdomyosarcoma cells, another study showed that tumor cell progression seemed to be regulated by the interleukin-4 receptor (IL-4R)-dependent signaling pathway, highlighting the role of IL-4 in this common type of STS [74].

TNF was also found in high levels in STS patient serum [45]. Similar to IL-6, the correlation between TNF and tumor grade, size, metastases, or recurrence was investigated. However, there was no significant association between the serum levels of TNF and these clinic-pathological features. Similarly, no association between these features and serum levels of IL-10 and granulocyte colony-stimulating factor (G-CSF) was demonstrated in STS patients [45].

Regarding IL-2Rα, its higher level in STS patients has been correlated with tumor size. Another study performed in 2012 suggested that a low serum level of IL-2Rα was associated with prolonged overall survival (OS) [47]. In this same study, Sleijfer et al. also indicated that low monocyte chemotactic protein-3 (MCP3) and hepatocyte growth factor (HGF) levels were associated with extended progression-free survival (PFS). However, they mentioned that these associations might be false-positive ones, so these results should be interpreted with caution and confirmed by more studies.

### **4. Expression of Immune Checkpoints and Their Ligands in STS**

Immune checkpoints are essential in regulating the immune response. In cancer, they can be dysregulated, working as an immune resistance mechanism [75].

In 2013, the impact of the immune checkpoints PD-1 and PD-L1 in STS (Figure 2) was evaluated for the first time [76]. The result from immunohistochemistry showed an intratumoral infiltration of PD-1 positive lymphocytes and the expression of PD-L1 in most STS samples. Additionally, PD-1 positivity, PD-L1 positivity, and the combined PD-1/PD-L1 pattern were independent prognostic indicators of OS and event-free survival. Furthermore, more studies have evaluated these immune checkpoints, the majority by immunohistochemistry, and confirmed the presence of PD-1 and PD-L1, and their association with a negative prognosis [77–86]. However, in some studies, PD-1 and PD-L1 expression appear to be low or absent, and the PD-L1 expression has not been associated with the outcome in STS [26,33,42,59,77,81,87]. Wunder et al. showed recently that the PD-1 and PD-L1 expression depended on the STS subtype and the prognostic value of PD-L1, justifying the discrepancies between studies with different subtypes of STS [88]. In addition, these discrepancies may also be due to the use of different methods of expression assessment, cutoff values, antibody clones, and tissue samples analyzed before and after therapeutical interventions [37,88].

PD-1 and PD-L1 expression levels have also been correlated in some studies with T-cell infiltration, and PD-L1 expression has been associated with more PD-1 positive TILs [27,79].

**Figure 2.** Prognostic value of immune checkpoints in STS. Several studies have been trying to correlate the presence of immune checkpoints with the prognosis of patients with STS. These studies have showed a negative prognostic value for B7-H3, PD-1, PD-L1, NKp30, B7-H6, Sirpα, CD47, CD155, LAG3, and IDO (red). A positive prognostic value was associated with the immune checkpoint E-Cadherin (green) [15,26,39,76–79,89–92]. **Figure 2.** Prognostic value of immune checkpoints in STS. Several studies have been trying to correlate the presence of immune checkpoints with the prognosis of patients with STS. These studies have showed a negative prognostic value for B7-H3, PD-1, PD-L1, NKp30, B7-H6, Sirpα, CD47, CD155, LAG3, and IDO (red). A positive prognostic value was associated with the immune checkpoint E-Cadherin (green) [15,26,39,76–79,89–92].

PD-1 and PD-L1 expression levels have also been correlated in some studies with Tcell infiltration, and PD-L1 expression has been associated with more PD-1 positive TILs [27,79]. Although the presence and prognostic value of these immune checkpoints has been controversial and underexplored in this type of tumor, they might still have a role in predicting the prognosis of STS patients. Furthermore, the expression of these immune check-Although the presence and prognostic value of these immune checkpoints has been controversial and underexplored in this type of tumor, they might still have a role in predicting the prognosis of STS patients. Furthermore, the expression of these immune checkpoints may also indicate the patients who will benefit from PD-1 therapies. In 2020, a study concluded that STS patients who responded to pembrolizumab, an anti-PD-1 monotherapy, exhibited more PD-L1-expressing macrophages than non-responders [59].

points may also indicate the patients who will benefit from PD-1 therapies. In 2020, a study concluded that STS patients who responded to pembrolizumab, an anti-PD-1 monotherapy, exhibited more PD-L1-expressing macrophages than non-responders [59]. Other immune checkpoints have been studied in several tumors, but there are only a few reports for STS. A recent study analyzed the expression of the B- and T-lymphocyte attenuator (BTLA) in sarcoma and found a lower expression mainly in CD4 TIL [77]. The same study also showed a high expression of lymphocyte-activation gene 3 (LAG3) on CD8 TILs. Other studies analyzed the expression of LAG3 by immunohistochemistry [39]. They confirmed its overexpression on TILs and found a significant association of LAG3 expression with a poor clinical outcome. Ishihara et al. suggested that a lower expression of indoleamine-pyrrole 2,3-dioxygenase 1 (IDO-1) was associated with a better prognosis in UPS [89]. E-Cadherin has also been studied in STS. It has been suggested that E-Cadherin has a possible role in the maintenance of epithelial architecture [93]. Furthermore, it was observed that upregulated E-Cadherin expression was associated with a better prognosis in STS patients [90,94]. The expression of B7-H6 and B7-H3 has also been evaluated in metastatic gastrointestinal stromal tumors and rhabdomyosarcoma, respectively [91,92]. In both studies, the expression of these molecules was associated with a worse Other immune checkpoints have been studied in several tumors, but there are only a few reports for STS. A recent study analyzed the expression of the B- and T-lymphocyte attenuator (BTLA) in sarcoma and found a lower expression mainly in CD4 TIL [77]. The same study also showed a high expression of lymphocyte-activation gene 3 (LAG3) on CD8 TILs. Other studies analyzed the expression of LAG3 by immunohistochemistry [39]. They confirmed its overexpression on TILs and found a significant association of LAG3 expression with a poor clinical outcome. Ishihara et al. suggested that a lower expression of indoleamine-pyrrole 2,3-dioxygenase 1 (IDO-1) was associated with a better prognosis in UPS [89]. E-Cadherin has also been studied in STS. It has been suggested that E-Cadherin has a possible role in the maintenance of epithelial architecture [93]. Furthermore, it was observed that upregulated E-Cadherin expression was associated with a better prognosis in STS patients [90,94]. The expression of B7-H6 and B7-H3 has also been evaluated in metastatic gastrointestinal stromal tumors and rhabdomyosarcoma, respectively [91,92]. In both studies, the expression of these molecules was associated with a worse prognosis. Dancsok et al. evaluated the immune checkpoints CD47 and Sirpα expression in sarcomas for the first time [15]. Through immunohistochemistry, the expression of both macrophagerelated immune checkpoints was correlated with an adverse prognostic factor. Recently, the expression of the exhaustion marker T cell immunoreceptor with Ig and ITIM domains

1

(TIGIT) was assessed in STS samples [26]. Although TIGIT expression was not associated with survival, the expression of its dominant ligand CD155 was associated with worse OS using the TCGA.

### **5. Immune-Related Gene Expression in STS**

Studies of lung cancer, ovarian cancer, head and neck squamous cell carcinoma, and renal cancer have suggested that immune-related genes (IRGs) may be used as prognostic biomarkers [95–98]. The IRG expression is underexplored in STS, and its prognostic significance remains unclear (Figure 3).

**Figure 3.** STS studies correlating the expression levels of immune-related genes and their prognostic significance. In STS, the expression of immune-related genes remains underexplored, and consequently, the prognostic value of these genes is still unclear. However, five main studies aimed at understanding this correlation, and their results are represented in this figure. Immune-related genes correlated with a good prognosis in STS are represented in green. On the other hand, immune-related genes associated with a bad prognosis are represented in red. From the peripheral to the center, circles represent genes encoding extracellular proteins, genes encoding transmembrane proteins, genes encoding intracellular proteins, the method used, and the respective study's first author and publication year [99–103]. <sup>1</sup> Prognostic value in synovial sarcomas; <sup>2</sup> Prognostic value in gastrointestinal stromal tumors; <sup>3</sup> Prognostic value in myxoid liposarcomas; <sup>4</sup> Prognostic value in sarcomas with complex genetics.

> In STS, high and low transcription levels of IL33 and its receptor ST2 were associated with the recruitment of CD8 T cells and the recruitment of Tregs and MDSCs, respec

tively [99]. Moreover, in the same report, both IL33 and ST2 levels were associated with a better outcome.

Recently, the gene expression of 364 differentially expressed IRGs was analyzed [100]. It was established that 18 of these genes were significantly associated with overall OS or/and with PFS, validating their value as prognostic biomarkers. Likewise, Dufresne et al. analyzed the expression of 93 genes encoding for immune checkpoints and membrane proteins in 253 STS samples [101]. This analysis showed a correlation between the immune signature and each sarcoma subgroup, concluding that the prognostic value could depend on the group. Another study constructed an immune gene-related prognostic model using five immune-related prognostic genes: IFIH1, CTSG, STC2, SECTM1, and BIRC5 [102]. These five genes had an effective performance in risk stratification of patients, showing their potential as biomarkers for predicting the response of STS patients to immunotherapy. In addition, in 2020, the analysis of high-grade STS tissue samples, divided according to OS, identified seven genes (C3, CD36, DOCK9, FCER2, FOS, HLA-DRB4, and NCAM1) correlated with a poor prognosis, and six genes (BIRC5, DUSP4, FOXP3, HLA-DQA1, HLA-DQB1, and LAG3) correlated with a good prognosis [103].

### **6. Peripheral Blood Immune Status**

The immunological status of peripheral blood in patients with STS remains unclear, just as its role as a prognostic indicator.

The circulating monocyte count has been studied recently as a marker of poor prognosis in several tumors [104]. In addition, the correlation between the increasing monocytes and decreasing lymphocytes with tumor growth and progression has already been proved in cancer populations [104,105]. In 2014, the lymphocyte/monocyte ratio (LMR) was studied for the first time in STS patients [106]. They concluded that the pre-treatment LMR ratio could act as a negative prognostic factor. Jiang et al. also analyzed the monocyte ratio in 124 STS patients [107]. Their analysis observed a significant association between poor prognosis for OS and PFS, and the presence of a monocyte ratio > 1, which is in line with studies concerning other tumors. In addition to being a poor prognosis factor, a low LMR indicates systemic inflammation in cancer, including STS. However, the association between inflammation indexes and the prognosis has been challenging and controversial. A study performed in 2019 evaluated 26 cases of STS and did not find significant differences in OS and PFS associated with the LMR [108].

Two meta-analyses aimed at evaluating the effect of neutrophil-to-lymphocyte ratio (NLR) in STS [109,110]. Both concluded that higher NLR was associated with poor OS, disease-free survival (DFS), and PFS. Although multiple studies have proved an association between different cellular ratios with the prognosis for several tumors, data for STS are still sparse [106]. The peripheral immunological status of STS was investigated by Kim et al. in 2021; they observed that a high level of M-MDSCs was associated with poor DFS and PFS [111]. In the same way, high levels of T-cell immunoglobulin and mucin-domain containing-3 (Tim-3) positive CD8 T cells were associated with lower DFS. On the contrary, high levels of NKG2D positive CD8 T cells were significantly associated with longer DFS times. The collection of tumor samples is usually difficult; therefore, more studies based on a minimally invasive method, such as collecting peripheral blood, are needed.

The aim of another study was to analyze the immune cells in both peripheral blood and tumor tissue [26]. The data showed that NK and T cells are both more activated and exhausted in tumor tissue than in circulation when comparing these two locations. Concerning NK cells, both CD56bright and CD56dim subsets were found in peripheral blood. However, in tumor tissues, CD56bright, the less mature and cytotoxic subset, appears to be less prevalent. The activation marker CD69 was also evaluated, and it is more expressed in both NK subsets in the tumor, compared to the peripheral blood. Similarly, the expression of the receptor of NK and T cell exhaustion TIGIT was increased in the tumor.

Regarding NK cells from peripheral blood, Bücklein et al. analyzed this cell subset in two groups of STS patients: chemotherapy-naïve STS patients and STS patients with a progression or relapse after chemotherapeutical treatment [112]. In both, NK cells were found to be dysfunctional during a chromium release assay using K562 cells as targets. The CD56dim NK cell subset frequency, studied using flow cytometry, was significantly lower in the blood from STS patients with a progression or relapse after therapy when compared to healthy donors. These conclusions could be specific to STS patients, since these alterations were not found in NK cells from renal cell carcinoma patients. In addition, a decreased expression of NKG2D, CD3ζ, and perforin was found and associated with the activation of NK cells in the second group of patients. On the contrary, Delahaye et al. did not find significant differences in the levels of peripheral NK cells nor in the NKG2D expression in GIST patients when compared to healthy volunteers [56]. However, they showed that a predominant expression of the immunosuppressive NKp30c isoform of the NKp30 receptor was associated with an unfavorable outcome.

### **7. Immunotherapy in STS**

In 1891, William B. Coley injected streptococcal organisms into a patient with sarcoma. The injection stimulated the immune system, and the sarcoma disappeared. After this successful experiment, he treated hundreds of patients with sarcomas, including STS. Coley initiated the discipline of cancer immunotherapy and demonstrated the possible use of this type of therapy for this disease [113].

It is now clear that the immune microenvironment is highly variable in STS, and this variability is frequently justified by STS heterogenicity. Despite this heterogenicity, clinical trials continue to incorporate various sarcoma subtypes to obtain the minimum number of patients required. Although there have been hints of positive responses to immunotherapy trials for STS, most trials have been negative or are not representative of all STS subtypes. Currently (July 2021), there are 85 clinical trials focused on immunotherapy in STS. Phase II and phase III clinical trials that have been completed and targeting the immune system in STS are shown in Table 1.

As was mentioned before, the expression of PD-1 and PD-L1 were present in some studies and absent in others, which appears to depend on the STS subtype. The presence of these immune checkpoints in some subtypes offers a promise for immunotherapy based on checkpoint inhibitors in these specific subtypes. Unfortunately, clinical trials testing immune checkpoint inhibitors in STS have not showed the impressive results achieved for many other cancers. The intention of the first study was to analyze the efficacy of targeting the immune checkpoint CTLA-4 with ipilimumab in synovial sarcoma, but neither a clinical benefit nor immunological activity was demonstrated [114]. Similarly, uterine leiomyosarcoma patients did not respond to anti-PD-1 antibody nivolumab in a phase II study [115]. Later, the clinical trial SARC028 tested the anti-PD-1 therapy with pembrolizumab. Promising responses for specific subtypes were observed in this trial, such as UPS and dedifferentiated liposarcoma. Moreover, the response to pembrolizumab was correlated to higher tumor-infiltrating lymphocytes at the baseline. Based on these promising results for specific subtypes of STS and in specific immune microenvironments, further research and correlative studies are required to improve the selection of patients for future clinical trials with immune checkpoint blockade in STS.


**Table 1.** Completed phase II and III clinical trials for immunotherapy in soft tissue sarcomas.

Adoptive cell therapy is based on the manipulation, modulation, and selection of immune cells to eliminate the tumor, overcoming the immune system's tolerance to cancer cells. As sarcomas appear to be one of the tumors most vulnerable to NK cell cytotoxicity, NK cell-based therapies seem to be a promising alternative treatment [116]. In 2010, it was demonstrated that rhabdomyosarcoma is sensitive to expanded NK cells [117], and phase I and II clinical trials of expanded haploidentical NK cells in rhabdomyosarcoma patients have begun (NCT02409576). The aim of another ongoing clinical trial is to com-

bine cryosurgery and multiple NK immunotherapies (NCT02849366) (Table 1). Similar to NK cells, lymphocytes could also be harvested from the patient or a donor, expanded, and then reinfused into the patient. Although the use of TILs against STS is poorly investigated, two ongoing phase II clinical trials have started. One of them proposes a donor lymphocyte infusion in patients with relapsed malignancies, including sarcoma (NCT00003887). The other is trying to eradicate minimal residual disease in sarcomas, including alveolar rhabdomyosarcoma, with autologous T cell transplantation concomitant with the tumor-specific peptides vaccine (NCT00001566). Alternatively, genetically engineered T cells expressing receptors for specific recognition of the cancer testis antigen New York esophageal squamous cell carcinoma-1 (NY-ESO-1) could be a promising strategy, since the expression of NY-ESO-1 in some subtypes of STS has been demonstrated, especially in synovial sarcomas [118,119]. In this STS subtype, a T-cell receptor-based gene therapy against NY-ESO-1 demonstrated promising results [120]. In another pilot study, an autologous T-cell expressing T-cell receptor recognizing NY-ESO-1 confirmed previous results with an anti-tumor response in 50% of metastatic synovial sarcoma [121]. Considering all these previous promising results, the aim of an ongoing clinical trial is to create an immune response against NY-ESO-1 antigen with a CDX-1401 cancer vaccine (NCT00948961) (Table 1). Cancer vaccines are a strategy to treat tumors. These vaccines attempt to elicit an immune response against tumor cells through the active manipulation of DCs. However, in addition to other limited reports of DC-based vaccination in STS, a study performed in 2017 indicated that the treatment is effective only in a small number of patients [122]. Several current clinical trials use vaccination with autologous dendritic cells to try to strengthen the immune system against sarcomas, including STS (NCT01347034; NCT02496520; NCT00365872). Peptide vaccination could also be an approach to treat STS, and clinical trials are testing peptide vaccines to enhance the immune response in STS (Table 1).

Clinical trials concerning immunotherapy for STS have, so far, shown limited and inconclusive results, which is largely due to the lack of representativity of several STS histologic types in the studies. However, attempts are still ongoing to identify biomarkers for monitoring immunotherapy and predict clinical outcome [123,124].

### **8. Future Perspectives**

Beyond the necessity of large-scale studies on tumor-infiltrating immune cells and their role in clinical features, it is also necessary to pair the analysis of tumor samples with peripheral blood samples to understand whether the information obtained about the circulating immune cells could be used to predict disease outcome or the response to treatment. The collection of peripheral blood is a minimally invasive procedure, which facilitates sample harvesting and consequently increases the number of patients who could undergo such a process and would allow patient monitoring during the treatment.

Regarding the soluble factors, there is still much to be learned about the array of these factors secreted by the tumor and their activity and interactions in TME. Given the pleiotropic and redundant nature of the soluble factors, the therapeutical target should be the balance between pro- and anti-inflammatory ones instead of the inhibition or activation of one in particular.

Treatments targeting immune checkpoints may represent a promising approach for other types of cancers as well. Nevertheless, it is necessary to select the patients who will benefit from this type of therapy carefully. Regarding IRGs, there are still only very few studies, so more research is required to understand the potential functional mechanisms of IRGs and their role in STS. The dual role of immunity in cancer leads us to believe that combination approaches that both stimulate protective host responses and inhibit immune subversion tactics might be more efficacious. The heterogenicity of STS implies that a "one size fits all" approach may be less successful. Furthermore, comprehensive immune profiling in combination with the evaluation of clinical features will be important to predict the response to therapy and survival. Lastly, the immune profiling of each patient might lead to personalized therapy.

The knowledge accumulated regarding tumor and peripheral immune status could be helpful in designing novel immunotherapeutic approaches for STS.

### **9. Conclusions**

STS have been treated as "non-immunogenic" tumors until now. However, this current work has proved that this characterization did not apply to all of them, since elements of the immune system were highly expressed in some STS samples. These elements, including immune cells, soluble plasmatic factors, immune checkpoints, and the expression of immune-related genes have been correlated with STS prognosis. Furthermore, their role in predicting the response to therapy and their potential as therapeutical agents or targets has been proven in STS. The infiltration of B cells, NK cells, TANs, and DC in STS tumors were correlated with a better outcome. On the contrary, TAMs were associated with a negative prognostic value. Regarding infiltrating CD8 T, CD4 T, and Tregs, their role in the outcome of the disease remains controversial. Some soluble plasmatic factors such as LIF, IL-8, HGF, IL-2Ra, VEGF, MCP-3, TNF-RI, IL-6, and M-CSF were associated with a negative prognosis in STS. Nevertheless, only a few studies have tried to understand their role in this type of cancer. A favorable prognostic value was associated with the immune checkpoint E-Cadherin, and a negative prognostic value was associated with the presence of B7-H3, PD-1, PD-L1, NKp30, B7-H6, Sirpα, CD47, CD155, LAG3, and IDO. Likewise, immune-related genes such as IL-33, ST2, BIRC5, DUSP4, FOXP3, HLA-DQA1, HLA-DQB1, and LAG3 were associated with a better outcome, while C3, CD36, DOCK9, FCER2, FOS, HLA-DRB4, and NCAM1 were correlated with a worse outcome. In another study, an immune gene-related prognostic model using IFIH1, CTSG, STC2, SECTM1, and BIRC5 showed potential to predict the response of STS patients to immunotherapy. The immunological status of peripheral blood in STS is still largely unknown. Increased LMR and NLR ratios have been associated with a poor prognosis in some studies. Higher levels of M-MDSCs and Tim-3 positive CD8 T cells also appear to be negative prognostic markers. On the contrary, NKG2D-positive CD8 T cells were correlated with a better outcome.

The main limitations that concern the studies mentioned above are the small sample sizes, the short follow-up, and the use of restricted STS histology types. Taking this into account, the studies might not be representative of the whole. In addition, in most of these studies, the stage of STS and treatments were not considered and might have a significant impact on prognosis. For these reasons, a large-scale prospective study, investigation of each subtype, and studies that consider the STS stage and treatment are warranted to substantiate and validate the results discussed in this article.

**Author Contributions:** J.M.C. and P.R.-S.: manuscript outline. L.M.S., J.S.A., T.F.-A., M.S.-R., P.F.-T., J.M.C. and P.R.-S.: manuscript writing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational Program and through the COMPETE 2020—Operational Programme for Competitiveness and Internationalisation and Portuguese national funds via FCT— Fundação para a Ciência e a Tecnologia, under the projects POCI-01-0145-FEDER-007440, UIDB/04539 /2020 and UIDP/04539/2020 (to P.R.-S. and M.S.-R.).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** J.S.A. was supported by PhD Grant (SFRH/BD/148007/2019) from the Portuguese Science and Technology Foundation (FCT), through the European Social Fund from the European Union.

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

### **References**


## *Review* **Acid Microenvironment in Bone Sarcomas**

**Gemma Di Pompo <sup>1</sup> , Margherita Cortini <sup>1</sup> , Nicola Baldini 1,2 and Sofia Avnet 2,\***


**Simple Summary:** Although rare, malignant bone sarcomas have devastating clinical implications for the health and survival of young adults and children. To date, efforts to identify the molecular drivers and targets have focused on cancer cells or on the interplay between cancer cells and stromal cells in the tumour microenvironment. On the contrary, in the current literature, the role of the chemical-physical conditions of the tumour microenvironment that may be implicated in sarcoma aggressiveness and progression are poorly reported and discussed. Among these, extracellular acidosis is a well-recognized hallmark of bone sarcomas and promotes cancer growth and dissemination but data presented on this topic are fragmented. Hence, we intended to provide a general and comprehensive overview of the causes and implications of acidosis in bone sarcoma.

**Abstract:** In bone sarcomas, extracellular proton accumulation is an intrinsic driver of malignancy. Extracellular acidosis increases stemness, invasion, angiogenesis, metastasis, and resistance to therapy of cancer cells. It reprograms tumour-associated stroma into a protumour phenotype through the release of inflammatory cytokines. It affects bone homeostasis, as extracellular proton accumulation is perceived by acid-sensing ion channels located at the cell membrane of normal bone cells. In bone, acidosis results from the altered glycolytic metabolism of bone cancer cells and the resorption activity of tumour-induced osteoclasts that share the same ecosystem. Proton extrusion activity is mediated by extruders and transporters located at the cell membrane of normal and transformed cells, including vacuolar ATPase and carbonic anhydrase IX, or by the release of highly acidic lysosomes by exocytosis. To date, a number of investigations have focused on the effects of acidosis and its inhibition in bone sarcomas, including studies evaluating the use of photodynamic therapy. In this review, we will discuss the current status of all findings on extracellular acidosis in bone sarcomas, with a specific focus on the characteristics of the bone microenvironment and the acid-targeting therapeutic approaches that are currently being evaluated.

**Keywords:** bone sarcoma; extracellular acidosis; tumour microenvironment; tumour-associated stroma; acid-sensing ion channels; vacuolar-ATP-ase; carbonic anhydrase IX; acridine orange

### **1. Introduction**

### *1.1. Bone Sarcomas*

Bone sarcomas comprise highly heterogeneous subtypes of mesenchymal tumours originating from the bone. The most common types of bone sarcoma are osteosarcoma, Ewing's sarcoma, and chondrosarcoma. Bone sarcomas account for <0.2% of malignant neoplasms registered in the EUROCARE (European Cancer Registry-based study on survival and care of cancer patients) database [1] and their incidence varies according to the different histotype. Osteosarcoma is the first primary cancer of bone (incidence: 0.3 per 100,000 per year), with a higher incidence in adolescents (0.8–1.1 per 100,000 per year at age 15–19 years) [2,3]. Ewing's sarcoma is the second most common primary malignant bone tumour. It occurs most frequently in children and adolescents, but adults can also

**Citation:** Di Pompo, G.; Cortini, M.; Baldini, N.; Avnet, S. Acid Microenvironment in Bone Sarcomas. *Cancers* **2021**, *13*, 3848. https:// doi.org/10.3390/cancers13153848

Academic Editors: Joanna Szkandera and Dimosthenis Andreou

Received: 16 June 2021 Accepted: 28 July 2021 Published: 30 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/).

be affected. Chondrosarcoma is the most frequent bone sarcoma of adulthood. The incidence is around 0.2 per 100,000 per year, with a median age at diagnosis between 30 and 60 years [2,3]. The survival rate after 5 years for patients with localised primary tumour is 60–70% and 50–60% for osteosarcoma and Ewing's sarcoma, respectively, with a dramatic drop to 30% for the former and to only around 20% for the latter, in metastatic patients. The survival rate of chondrosarcoma is 50–60% at 10 years according to the histological grade [4]. Current treatments for osteosarcoma and Ewing's sarcoma combine surgery (preoperative or neoadjuvant), followed by chemotherapy (postoperative or neoadjuvant), and long-term polychemotherapy [5,6]. However, most conventional chemotherapy commonly fails, leading to the cogent need for the identification of novel therapeutic targets and the development of more effective approaches. Among them, the employment of tyrosine kinase and cyclin-dependent kinase inhibitors, DNA repair or chemoresistance targeting, and immunotherapies are currently the most attractive [7].

### *1.2. Cancer-Associated Extracellular Acidosis*

Extracellular acidosis is a well-established hallmark of malignancy in solid tumours [8]. Similarly to hypoxia [9–11], it influences tumour cell behaviour and clinical outcome by modulating cancer stemness, invasion, invadopodia formation, metastasis, anticancer immune reaction, and response to therapy [8,12].

Solid tumours, including sarcomas [12–14], are characterised by an extracellular pH (pHe) that ranges from 6.4 to 7.3, whereas in normal tissues, the range is 7.2 to 7.5 [15]. Tumour formation and progression are strongly influenced by biophysical factors including extracellular acidosis. Understanding how sarcoma cells cope and adapt to the microenvironmental stress that is promoted by an excess of extracellular protons will contribute to a better knowledge of sarcoma pathophysiology and the identification of novel anticancer strategies.

In this review, we will discuss the current status of knowledge on interstitial acidosis in bone sarcomas, taking also into consideration the unique characteristics of sarcoma cells in the bone microenvironment and the acidtargeting therapeutic approaches that are under investigation.

### **2. Source of Acidosis in the Microenvironment of Bone Sarcomas**

Acidosis in bone sarcomas is mainly due to (1) the metabolic switch of cancer cells to glycolytic metabolism which, in turn, causes the efflux of lactic acid and protons in the extracellular space; (2) the active release of protons by normal bone cells, mainly osteoclasts, to resorb bone during the formation of osteolytic lesions that occurs with tumour expansion.

### *2.1. Altered Tumour Metabolism and Intratumoural Acidosis in Bone Sarcomas*

High glycolytic activity is a common feature of many cancer types, including sarcomas [10,16–19]. Cancer cells switch to a glycolytic phenotype in a poorly perfused environment. However, as elegantly described by Otto Warburg in 1927 [20,21], glycolysis in cancer cells also occurs under conditions of normal oxygen tension.

In bone cancers, hypoxia results from increased proliferation of cancer cells in association with a high rate of oxygen consumption, and from the intrinsic hypoxia of the bone microenvironment. Indeed, hypoxia greatly influences bone biology and physiology [22]. As a demonstration, in the medullary cavity of animal models, pO<sup>2</sup> values range from 11.7 to 31.7 mm Hg (1.5–4.2%), with a mean of 20.4 mm Hg (2.7%) [23].

The switch to glycolysis, both under normoxic and hypoxic conditions, follows the activation of hypoxia-inducible factor-1 (HIF-1), which drives the transcription of crucial enzymes of the glycolytic pathway [24]. As an end result, the increased glycolytic rate leads to intracellular accumulation of protons in the cytoplasm, but also the release of protons into the extracellular space as a waste product along with lactate. To survive this harsh microenvironment, cancer cells develop adaptive mechanisms, including transcriptional, posttranslational, and morphological alterations, which eventually lead to profound changes in their phenotype and the metabolic profile [12]. Cancer-associated acidosis is attracting increasing interest in the field of cancer research. New in vivo imaging tools are being developed to assess the association between cancer metabolism and the acidic microenvironment [25]. For example, in a near future, it will be possible to combine <sup>8</sup>F-FDG PET, currently used for staging bone sarcomas [26,27], together with chemical exchange saturation transfer magnetic resonance imaging (CEST-MRI), to detect acidic regions of the tumour in order to determine its metastatic potential.

Finally, an additional metabolic trigger of tumour interstitial acidosis, in addition to glycolysis exacerbation, could be the hydration of excessive CO<sup>2</sup> production in the more oxidative areas of the tumour [28]. However, this mechanism has not been explored in bone sarcoma.

In this context, although not thoroughly explored, it is noteworthy that acidosis, conversely, may lead to HIF-1 regulation. In order to maintain energy homeostasis, highly glycolytic cancer cells lead to glucose deprivation in the extracellular space by consuming large amounts of glucose (and glutamine). Low-glucose conditions in the tumour microenvironment, in turn, can cause a loss of stromal caveolin-1, yielding oxidative stress which mimics hypoxia ('pseudohypoxia') through activation of HIF-1 and NF-kB [29]. NF-kB has been shown to be a direct modulator of HIF-1 expression in inflammation and hypoxia [30,31], and in osteosarcoma, NF-kB upregulation has been demonstrated to be induced by acidosis [32], supporting the idea that acidosis and hypoxia can reciprocally modulate each other's behaviour.

### *2.2. Proton Extruders in Bone Sarcomas*

Cytosolic acidification is extremely toxic to both normal and cancer cells, eventually leading to apoptosis [33]. Sarcoma cells get rid of excessive intracellular proton accumulation through extruders and transporters located on the plasma membrane or lysosomal membrane, which strongly acidify the extracellular space via direct pumping/transport or by exocytosis, respectively [34].

Previous reports on extracellular acidification in bone sarcomas have made use of preclinical in vitro models and various techniques to measure pHe, such as the use of macro- or microelectrodes or the measurement of extracellular acidification rate (EACR) values by Seahorse technology. These techniques have shown that the activity of these extruders/transporters is responsible for strong acidification of the medium, both in the extracellular space and near the cell membrane [16]. Additionally, the enhanced acidification ability in stem cells derived from a soft tissue sarcoma has been demonstrated using acridine orange and lysosensor staining [16,34,35].

Among the most studied proton exchangers and transporters, sarcomas express certain subunits of vacuolar H<sup>+</sup> -ATPase (V-ATPase), such as V1B<sup>2</sup> and V0c V-ATPase subunits, the Na+/H<sup>+</sup> exchanger isoform 1 (NHEs, mainly NHE1), the monocarboxylate transporters (MCTs, mainly MCT1, also known as lactate–proton symporter), the Na<sup>+</sup> dependent Cl−/HCO3− exchanger, and carbonic anhydrases (CAs) isozymes, mainly CAII, CAIX and CAXII [36]. Studies describing the expression and the role of these molecules in the extracellular acidification and behaviour of bone sarcomas are reported in Tables 1–4.


**Table 1.** V-ATPase expression and targeting in bone sarcomas.



**Table 4.** NHE expression and targeting in bone sarcomas.

Several drugs have been tested to target these ion extruders/transporters as anticancer therapy. For a more extensive discussion, see Section 5.1.

The most studied ion/proton extruders/transporter is the V-ATPase, followed by the CAIX enzyme. V-ATPases are ubiquitous proton pumps that are found either on the intracellular membranes, such as lysosomes, or, for specialised cells, at the plasma membrane. V-ATPases use the energy of ATP to transport protons from the cytosol to intracellular compartments or to the extracellular space. The V-ATPase consists of an ATP-hydrolytic domain (V1) and a proton-translocation domain (V0) [49]. Its energyconsuming activity requires the close association of all the components of the complex, which is provided by the C-loop [50]. Studies on V-ATPase expression and activity in bone sarcomas are mainly related to the analysis of preclinical models and, less frequently, of tissue samples.

CAIX is one of the 15 carbonic anhydrase isoforms present in humans, among which 12 are functional [51]. Carbonic anhydrases are a large family of dimeric zinc metalloenzymes with an extracellular active site that catalyses the reversible hydration of carbon dioxide to carbonic acid and are involved in respiration and acid–base balance, facilitating acid secretion in different cell types [52]. Evidence for CAIX expression in bone sarcomas has been largely based on the analysis of human tissue samples.

A less considered but important acid extruder is the voltage-gated proton channel (Hv1). This has been found to be expressed in the cells of origin of bone sarcoma, the mesenchymal stromal cells (MSC). Its pharmacological inhibition in MSC significantly decreases cell differentiation and mineral matrix deposition [53]. However, although Hv1 expression has been demonstrated in different cancers that frequently colonise bone, including breast and colorectal carcinomas [54,55], no data have been reported in bone sarcomas. In this context, it might be interesting to compare the ability of bone sarcomas to acidify the extracellular space with respect to other types of cancers that are able to expand in bone, such as bone metastases (BM) (see ref. [50]); carcinoma cells metastasizing to the bone share with bone sarcoma cells different mechanisms of proton extrusion, including the expression of V1B<sup>2</sup> and V0c V-ATPase subunits, CAIX, MCT1, and MCT4. As an example, we have recently found mRNA expression of CAIX in breast and renal carcinoma cell lines, with a significant increase under reduced oxygen conditions with respect to normoxia [50,56]. Additionally, different isoforms of V-ATPase, including the V1C<sup>1</sup> [57] and the V1B<sup>2</sup> and V1G<sup>1</sup> subunits [58], are expressed by breast carcinoma cells with a specific tropism for bone. Finally, it has been demonstrated that the expression of MCT4 in tumour cells is responsible for a metabolic coupling with bone-resorbing osteoclasts, thereby inducing a higher osteolytic activity in BM from breast carcinoma [18].

In summary, several lines of evidence suggest that an increased glycolytic rate and subsequent activation of several ion extruders and transporters in different cancer cells that grow in bone are the main causes of tumour interstitial acidosis.

### *2.3. Bone Resorption as a Source of Extracellular Acidification*

In the bone soil, to expand and invade the surrounding normal tissue, sarcoma cells degrade the hard extracellular matrix by directly or indirectly stimulating the activity of osteoclasts, the highly specialised bone-resorbing cells. The bone microenvironment is a fertile ground for tumour growth. Under physiological conditions, the process of bone

remodelling couples osteoclast-mediated bone resorption and osteoblast-promoted bone formation to maintain bone homeostasis. However, the development and progression of primary bone tumours, including osteosarcoma, severely disrupt this balance and induces a 'vicious cycle' between osteoclasts, osteoblasts, stromal cells, and cancer cells. In the bone soil, in order to expand and invade the surrounding normal tissue, sarcoma cells degrade the hard extracellular matrix by directly or indirectly stimulating the activity of osteoclasts that resorb bone, as well as directly eroding bone through the secretion of metalloproteinases (MMPs). The induction of osteoclast activity can be triggered by a plethora of growth factors that also commonly regulate physiological bone remodelling and can be secreted by cancer cells, or by tumour-stimulated osteoblasts. Of these, the most important is the receptor activator of nuclear factor-kappa B ligand (RANKL). Other factors, such as interleukin 1 (IL-1), interleukin 6 (IL-6), tumour necrosis factor-alpha (TNFa), parathyroid hormone-related protein (PTHrP), or transforming growth factorbeta (TGFb), mediate RANKL receptor (RANK) expression on the surface of osteoclasts, thereby favouring osteoclast maturation and activation [59]. Furthermore, we have recently demonstrated that a low pH further induces osteoclast activity, both directly and indirectly, by stimulating osteoblasts to secrete pro-osteoclastogenic paracrine mediators such as IL-8 and IL-6 [56].

Once stimulated, mature osteoclasts can resorb bone through a multistep dynamic process. First, osteoclasts migrate and attach to the bone surface that is to be degraded and removed, thus forming a tight 'sealing zone'. Then, the plasma membrane polarises to form the resorption organelle, the ruffled border, a unique folded highly permeable membrane facing to the bone surface to be resorbed [60]. Subsequently, to dissolve the mineralised component of bone, osteoclasts secrete hydrochloric acid into the resorption lacunae (Howship's lacunae) mainly via plasma membrane V-ATPase (a3 isoform) [61]. Proton pumping performed by osteoclasts during bone resorption activity is an energy-consuming intensive process that relies primarily on the glycolytic metabolism of osteoclasts [62]. It is noteworthy that the expression of a3 is 100-fold higher in osteoclasts than in other cell types [63]. The activity of V-ATPase is also coupled with the activity of the chloride ion–proton channel antiporter ClC-7 [64], and both proteins are clustered in the ruffled border domain.

Finally, as an additional player in the acidification activity of osteoclasts, it has been demonstrated the expression of Hv1 that helps proton release and bone mineral dissolution, thereby promoting bone resorption [65–67].

As a consequence of the proton extrusion activity, in Howship's lacunae, the pH reaches very low values, around 4.5 [60]. At the end of the resorption process, protons pumped into Howship's lacunae diffuse in the extracellular space, thus causing further acidification of the tumour microenvironment. Adversely, proteinaceous component of the matrix, mainly type I collagen, is degraded through the activity of the osteoclast-derived cysteine proteinase cathepsin K, which is responsible for the breakdown of collagen I, osteopontin, and osteonectin [68].

Osteoclast differentiation and activity result in dysregulated bone lysis and release of bone matrix growth factors such as TGFb, insulin-like growth factor 1 (IGF1), fibroblast growth factor (FGF), or bone morphogenetic protein (BMP), which, in turn, can promote tumour cell proliferation and further bone destruction [69,70]. This 'vicious cycle' between cancer cells and the bone microenvironment was first described in bone metastasis, but in fact, there is evidence supporting the notion that osteosarcoma cells, for example, mediate bone destruction by stimulating osteoclast differentiation and activity as bone metastasis [71,72]. In addition to osteoclasts, acid-mediated resorption of the bone mineralised matrix can also be performed by osteocytes. Osteocytes are the final fully differentiated form of osteoblasts that are trapped in the hard matrix and directly remodel the bone walls of their lacunar–canalicular systems in a process known as perilacunar/canalicular osteocytic remodelling. As with osteoclasts, this process relies on the combined activity of MMPs, vacuolar acid-secreting H<sup>+</sup> -ATPases [73,74], and other enzymes, such as cathepsin

K and carbonic anhydrases [75]. However, the interaction between sarcoma cells and osteocytes is completely unexplored, and it is still unknown whether perilacunar remodelling can be induced by invading cancer cells.

Finally, in the context of the acid extracellular tumour microenvironment, it should be noted that an excess of extracellular protons may also modulate the activity of cation channels, including calcium receptors [76,77]. Calcium (Ca2+) signalling is crucial, both for bone physiology and sarcoma progression. Indeed, both osteoblasts and osteoclasts, as well as osteosarcoma cells, express calcium-sensing receptors on the cell membrane [78,79], and Ca2+ is an essential mediator for cell differentiation, bone resorption, and gene transcription in osteoclasts, and for the aggressiveness of tumour cells [80,81]. However, the interference of extracellular acidosis in Ca2+ signalling in bone sarcoma is an unexplored field of research. For a more detailed discussion of the potential effect of high H<sup>+</sup> extracellular concentration on cation channels that are expressed by bone sarcomas, see Section 4.1.

In conclusion, these findings demonstrate that, in addition to tumour cells, tumourinduced bone-resorbing cells of the bone microenvironment contribute to acidify the microenvironment of sarcomas.

### **3. Effect of Acidosis on Sarcoma Cells**

The extracellular acidification derived from cancer cells and from the tumour-associated stroma is responsible for the modulation of bone colonisation by sarcoma cells. Indeed, an acidic pHe promotes cancer invasion, survival, and angiogenesis, and alters the cell permeability to anticancer drugs by many different mechanisms, thereby preventing their effective targeting.

Extracellular acidosis has also been described to influence anticancer immune response and autophagy in a number of solid tumours. However, the role of autophagy in mediating survival to acidosis has not been confirmed in osteosarcoma, where the autophagic flux seems to be unchanged between pH conditions (7.4 and 6.8) [37,82], and the impact of acidosis on the infiltration of inflammatory or immune cells has not been explored thus far. Furthermore, the system by which tumour cells can sense extracellular acidosis has not been deeply investigated yet, but few specific sensors have been identified. Finally, it is important to bear in mind that preclinical studies that investigated on sarcoma acidic microenvironment and based on cell culture medium acidification have high heterogeneity of pH values: in most cases, the studies were carried out with a pH range between 6.5 and 6.8 [22,24–26], but in other cases, harsher experimental conditions were used (pH 5.8 in [9]). On the other hand, the development of a 3D model has led to the development of physiological pH culture values by using an unbuffered culture medium, thus allowing 3D spheroids to adapt the pH value to their own metabolism [54]. Thus, these different experimental systems and different pH values might have led to different outcomes.

### *3.1. pH Sensors in Sarcoma Cells*

In the TME, different ion channels behave not only similar to ion transporters but also similar to sensors and transducers of altered pH as they can be affected by both extracellular and intracellular pH [83]. Furthermore, they greatly contribute to cancer progression [84–87]. As an example, in osteosarcoma, the voltage-gated potassium channel Kv1.3, transient receptor potential cation channel subfamily M member 8 (TRPM8), and piezo type mechanosensitive ion channel component 1 (Piezo1) are among the most expressed pH-sensitive ion channels and correlate with tumour progression [88–90].

On the other hand, the high concentration of protons in TME may also strongly affect the biological functions of these pH-sensitive proteins and receptors, since it may induce Kv1.3 potassium channel inactivation, and the alteration of the signalling pathway mediated by the Ca2+-permeable channels, TRPM8 and Piezo1 [91–93], ultimately altering their proaptoptotic signalling. However, the acid-mediated effect on these ion channels and the downstream signalling has never been explored in sarcoma.

### *3.2. Effect of Extracellular Acidosis on Tumour Invasion, Survival, and Metabolism*

Invasion occurs through invadopodia, dynamic actin-rich membrane protrusions that penetrate within the extracellular matrix and degrade it through the spatial and temporal release of proteases and protons [94]. The protonation of the matrix metalloproteinases is dependent on the activation of the proteinases and requires the redistribution and activation of V-ATPases and NHE1 to the tip of the invadopodia. Thus, local invasion is strongly modulated by the acidification activity of these proton/ion transporters and by the presence of an acidic pHe.

In sarcomas, the acidic microenvironment activates survival pathways and increases migration and invasive potential [16,37]. However, further molecular mechanisms are responsible for the acidosis-mediated progression of sarcomas. In osteosarcoma, we demonstrated the pH-dependent activation (at a pH of 6.5) of a stress-regulated switch that promotes the recruitment of the TNF-receptor-associated factors/cellular inhibitor of apoptosis protein 1 (TRAF/cIAP) complexes, and nuclear factor kappa-light-chainenhancer of activated B cells (NF-κB) pathway [95]. This activation ultimately leads to an increase in cancer cell survival, suggesting a role for TRAF/cIAP proteins as promising targets for anticancer therapy. As an in vivo confirmation of the intimate association between acidosis and cancer cell survival, we found a significant correlation between V-ATPase and TRAF1 or NF-κB1 expression in tissues from osteosarcoma xenografts.

More recently, evidence has shown that extracellular acidosis, obtained in unbuffered conditions, is also responsible for prominent metabolic plasticity that leads to the accumulation of intracellular lipids, specifically sphingolipids and sphingosine 1-phosphate (S1P). Impairing S1P levels by means of Fingolimod, an FDA-approved drug for the treatment of multiple sclerosis, was of predominant importance to decrease the migration potential of acid-resistant cells, to increase apoptosis, and to impair xenograft growth [82]. This suggests, for the first time, the use of an anticancer drug that has the potential to specifically target the acid-resistant subpopulation in osteosarcoma.

Finally, by studying both standard monolayer cultures and cancer stem cells and by extensive metabolomic analysis, we demonstrated that extracellular acidosis completely remodels cancer cell metabolism by inducing glycolysis repression and by increasing the amino acid catabolism and the urea cycle [96].

### *3.3. Effect of Extracellular Acidosis on Tumour Sensitivity to Anticancer Drugs*

Tumour acidosis is also a major cause of drug resistance and therapeutic failure. First of all, a low pHe (pH 6.5) significantly decreases the growth rate of cancer cells, thereby affecting the IC50 values of drugs that target actively proliferating cells [37]. However, an acidic pHe may also impact the response to therapeutics through additional complex mechanisms. The pH gradient across cellular membranes is crucial for determining the passive diffusion of small molecules. 'Ion trapping' (or pH partitioning) is the physiological process regulating passive permeability through the cellular membrane of negatively or positively charged compounds, such as ionisable compounds containing weak bases or weak acids. The lysosomal and the cytoplasmic membranes can compartmentalise drugs and, as a consequence of the pH partitioning, drugs can be hindered from reaching their molecular target because they become trapped on the wrong side of cellular membranes. The extent of ionisation for a molecule depends both on its intrinsic pKa values(s) and the pH of the solution. In an acidic extracellular microenvironment, weak bases will be positively charged to a larger extent, thus influencing the diffusion of the drug inside the target cells [14].

The cellular membranes that can compartmentalise drugs are both the cytoplasmic membrane and the lysosomal membrane. Acidic lysosomes can sequester weakly basic molecules from the cytosol to an extent that is directly related to the level of lysosomal acidosis, thereby preventing the drug targeting [97]. In this context, it is noteworthy that a high extracellular concentration of protons increased both the number and the acidification of lysosomes in osteosarcoma cells [37].

Additionally, the cytoplasmic membrane contributes to the 'ion-trapping effect'. In the presence of an acidic extracellular microenvironment, weakly basic drugs are forced to stay outside the cell. We confirmed this mechanism for doxorubicin in osteosarcoma cells [37]. Conversely, the presence of a low extracellular pHe allows for the permeability of weakly acidic drugs. In such a case, the neutral form of a weakly acidic compound may be favoured, and the uncharged species can freely diffuse across the plasma membrane. Since the cytosolic pH is slightly alkaline, once the acidic drug has crossed the plasma membrane and entered the cell, it is ionised and trapped within the cell. In this case, the cytotoxic activity may be enhanced by extracellular acidosis. Known examples of anticancer drugs containing weak acids are 5-fluorouracil and cyclophosphamide [14]. However, preclinical studies on the comparison of cytotoxicity at different pH values of these drugs in inhibiting bone sarcomas have never been performed.

### *3.4. Effect of Extracellular Acidosis on Tumour Angiogenesis and Others*

The anarchic formation of new vessels that provide O<sup>2</sup> and nutrients needed by actively proliferating cells is induced by tumour cells through the release of pro-angiogenic factors, such as vascular endothelial growth factor (VEGF) and interleukin 8 (IL-8) [98], or through the stabilisation of HIF-1 that are promoted by extracellular acidosis [99,100]. Interestingly, in osteosarcoma cells under acidic conditions, we observed increased release of extracellular nanovesicles with proangiogenic activity, including urokinase-type plasminogen activator (uPA), angiopoietin-2 (Ang-2), and VEGF, as well as the presence of miRNAs related to angiogenesis, as demonstrated by the formation of tubule branches in the chorioallantoic membrane (CAM) [101], suggesting that local acidosis might be responsible for promoting neoangiogenesis.

### **4. Effect of Acidosis on Different Cells and Elements of Sarcoma Microenvironment**

Cancer cells are not solely responsible for the growth of cancer and the spread to distant organs. A complex structure, formed of cancer cells that directly interact with stromal cells under different microenvironmental conditions, constitutes the bulk of the tumour. Among stromal cells, the microenvironment of bone sarcomas includes MSC, osteoblasts and osteoclasts, cancer-associated fibroblasts, and immune cells: all these different cell types coexist and infiltrate the tumour [102,103]. In particular, similar to physiological wound healing, MSCs are recruited from the bloodstream to the site of the tumour lesion, where they contribute to the rapid tumour expansion [68]. MSCs are crucial for the initiation [104], as well as the progression of the lesion [105]. However, in the context of mesenchymal tumours, MSCs are hardly distinguishable from tumour cells.

Importantly, cancer cells are not the only population being affected by extracellular acidosis. The effects of a low pHe are observed also on stromal cells of the bone microenvironment, and these may, in turn, indirectly modulate the behaviour and the aggressiveness of tumour cells (Figure 1).

### *4.1. Bone Cells Sense and React to Extracellular Acidification*

It is widely recognised that local variations of pHe greatly impact osteoblast and osteoclast differentiation and activity. Thus, as in other pathological conditions (i.e., inflammation), in the altered tumour microenvironment, bone cells can perceive acidosis and react to such stress signals by modulating their activities, as well as through paracrine communication by stimulating cancer progression.

Cells of the osteogenic lineage react to a high extracellular concentration of protons by impairing their osteogenic activities, namely, osteoblast differentiation, matrix deposition, and mineralisation [106,107]. Adversely, in osteoclasts, a low pHe increases the formation of resorption pits (maximal stimulus at pH < 6.9 [108]) and upregulates the activity of cathepsin K, tartrate-resistant acid phosphatase (TRACP), and TNF-receptor-associated factor 6 (TRAF6) [109–111].

**Figure 1.** Graphical summary of the effects of extracellular acidosis on cells of the bone sarcoma microenvironment. **Figure 1.** Graphical summary of the effects of extracellular acidosis on cells of the bone sarcoma microenvironment.

*4.1. Bone Cells Sense and React to Extracellular Acidification*  It is widely recognised that local variations of pHe greatly impact osteoblast and osteoclast differentiation and activity. Thus, as in other pathological conditions (i.e., inflammation), in the altered tumour microenvironment, bone cells can perceive acidosis and react to such stress signals by modulating their activities, as well as through paracrine communication by stimulating cancer progression. Cells of the osteogenic lineage react to a high extracellular concentration of protons by impairing their osteogenic activities, namely, osteoblast differentiation, matrix deposition, and mineralisation [106,107]. Adversely, in osteoclasts, a low pHe increases the formation of resorption pits (maximal stimulus at pH < 6.9 [108]) and upregulates the activity of cathepsin K, tartrate-resistant acid phosphatase (TRACP), and TNF-receptor-Bone cells sense pH changes through specific proton sensors and channels that are typically expressed by sensory neurons. Among them, the acid-sensing ion channels ASIC2, also known as amiloride-sensitive cation channel 1, neuronal (ACCN1), and ASIC3/ACCN3 are mostly abundant in bone. Specifically, previous reports have shown the expression of ASIC1/ACCN2, ASIC2/ACCN1, and ASIC3/ACCN3 mRNAs [112] in human osteoblasts. Besides ASICs, metabotropic proton-sensing G protein-coupled receptors (GPCRs) have also been recently identified as proton-sensing machinery in osteoblasts [113–115]. Similarly, we recently found that MSC, osteoblasts, and CAF express ASIC4/ACCN4, ASIC3/ACCN3, G protein-coupled receptor (GPR)-65, and GPR4 at levels comparable to or even higher than those expressed by cells of neuronal origin and that in MSC, the incubation with an acidic medium increases the expression of ASIC4/ACCN4 and GPR65 [58].

associated factor 6 (TRAF6) [109–111]. Bone cells sense pH changes through specific proton sensors and channels that are typically expressed by sensory neurons. Among them, the acid-sensing ion channels ASIC2, also known as amiloride-sensitive cation channel 1, neuronal (ACCN1), and ASIC3/ACCN3 are mostly abundant in bone. Specifically, previous reports have shown the expression of ASIC1/ACCN2, ASIC2/ACCN1, and ASIC3/ACCN3 mRNAs [112] in human osteoblasts. Besides ASICs, metabotropic proton-sensing G protein-coupled Regarding the osteoclastic lineage, human monocytic osteoclast precursors express ASIC1/ACCN2, ASIC2/ACCN1, and ASIC3/ACCN3. This expression persists also after the induction of osteoclast differentiation, albeit at a lower level. Likewise, transient receptor potential vanilloid (TRPV) channels, which are typically expressed by sensory neurons, and the ovarian cancer G protein-coupled receptor 1 (OGR1) which belongs to the GPCR family are proton sensors and have been involved in osteoclast differentiation and survival [50]. In particular, TRPV1, TRPV2, and TRPV4 channels are crucial for osteoclast

biology [80,116,117]. Notably, TRPVs are activated also by severe acidosis (pH 5.4) [118] and TRPV4 seems to be the major mediator of the acidosis-induced osteoclast formation as its antagonist, RN1734, partially inhibited the pH-dependent osteoclastogenesis, while its agonist 4-α PDD enhanced osteoclast formation under mild acidosis [117].

Under this context, it is noteworthy that both ASICs and TRPVs are also permeable to cations other than H<sup>+</sup> , like Ca2+ and the reciprocal interactions between H<sup>+</sup> and Ca2+, and the competition of H<sup>+</sup> for the same binding-site of Ca2+ may modulate the activity of these pH sensors and, thus, the downstream biological effects. Specifically, for ASICs, due to the binding competition, Ca2+ binding favours the closed state, and H<sup>+</sup> binding leads to the open state [119]. Furthermore, increased extracellular Ca2+ concentration can significantly decrease the pH sensitivity of ASIC1 and ASIC3 [120]. Thus, this strong interplay between H<sup>+</sup> and Ca2+ may occurs also in normal bone cells in the sarcoma microenvironment: it is already well known that TRPVs mediate Ca2+ signalling and are produced in mature osteoclast differentiation to sustain the intracellular Ca2+ level for the maintenance of active NFATc1 that regulates terminal cell differentiation [121], and the presence of an excess of protons in the sarcoma TME may interfere with Ca2+ signalling mediated by TRPVs in osteoclasts and may directly alter the osteoclast physiology and activity.

Overall, these data indicate that normal bone cells perceive and react to the acidification of the bone sarcoma microenvironment. The ultimate result is an unbalance of bone remodelling. In conclusion, a low pHe appears to be an essential requirement for the initiation of the osteolytic process, but it may also be involved in altered bone formation as it occurs in osteogenic sarcomas such as osteosarcoma.

### *4.2. The Acid-Stimulated Secretome*

Tumour-derived acidosis may favour tumour expansion by reprogramming stromal cells to the secretion of proinflammatory cytokines. Decrease of local pH is per se an inflammatory stimulus that causes the release of various enzymes during phagocytosis, the damage of vasculature and other surrounding tissues, and the prolonging of the healing process by stimulating new inflammatory reactions [29]. We have recently shown that extracellular acidosis directly activates the NF-kB inflammatory family of transcription factors and thus the secretion of NF-kB-related cytokines, chemokines, and growth factors by the osteosarcoma-associated stromal compartment formed by osteoblasts, MSC, and CAF [32,58]. Regardless of the source of acidosis, after a few hours, incubation with a pHe 6.8 activates RelA, RelB, or NF-kB that, in turn, induce the expression of the inflammatory cytokines IL-8 and IL-6 and enhances cancer stemness (formation of spheroids and expression of the stemness-related markers oct4 or Nanog [32]). In the same study, by using a blocking antibody against the IL-6 receptor, we demonstrated that acid-induced release of IL-6 by normal mesenchymal cells was directly responsible for bone cancer migration and invasion [32]. Intriguingly, IL-6 secretion seems to be directly dependent on the acid-stimulated MSC, whereas the tumour cells contribute little to the release of paracrine tumour-stimulating factors under acidic pH conditions. This is of note because it highlights the importance of the stromal subpopulation in enhancing cancer progression. Furthermore, the exposure of osteosarcoma cells to the secretome of acid-stimulated MSC reduced the toxicity of doxorubicin and thus promoted the development of a chemoresistant phenotype [32].

Altogether, these observations warrant the role of local acidosis in promoting a protumourigenic phenotype in bone sarcomas also by inducing a proinflammatory and a pro-osteolytic secretome by cells of the osteogenic lineage.

### *4.3. Matrix Remodelling/Degradation*

Another important feature of the sarcoma microenvironment is the composition and organisation of the ECM, whose mechanical properties affect cancer cell behaviour and that may be, in turn, influenced by tumour-derived extracellular acidosis. ECM is mainly secreted by stromal cells, and it is composed of various macromolecules, including collagens,

glycoproteins (fibronectin and laminins), proteoglycans, and polysaccharides [122]. ECM is also secreted from tumour cells, especially from osteogenic sarcoma. However, how low extracellular pH affects the synthesis and secretion of proteins of ECM is an almost uncovered field of investigation. During cancer progression, an excessive ECM remodelling occurs by proteinase activity, such as MMP-2 and MMP-9, and small ECM fragments are released into the circulation [123]. In melanoma cells, it has been reported that acidic culture conditions induce the increase of 103-kDa gelatinase/type IV collagenase secretion [124]. Furthermore, the membrane-bound MMP-14 has an acidic pH optimum and has been observed to be in close association with CAIX in invadopodia [125]. Additionally, in sarcoma, with particular regard to Ewing sarcoma, we previously found an increase of MMPs activity when tumour cells were cultured at low pH, as evaluated by gelatine-quenching assay and an increase of the ability to degrade type I collagen [16]. However, with the exception of the mentioned report, no other data have been published about the correlation between acidosis and ECM remodelling/degradation in bone sarcomas.

### *4.4. Effect of Extracellular Acidosis on Immune Reactivity to Cancer Cells*

In the sarcoma microenvironment, different cells and cytokines of the immune system may be included, such as tumour infiltrating lymphocytes (TILs) and associated macrophages, expression of immune checkpoint inhibitors such as cytotoxic T-lymphocyteassociated protein 4 (CTLA-4), programmed cell death-1 (PD-1), and programmed deathligand 1 (PD-L1), and major histocompatibility complex (MHC) antigen expression [126]. All of these components may be important for prognosis and responses of tumours to immunologically targeted therapies and are potential therapeutics or therapeutic targets. However, although significant progress in the field of immunotherapy, particularly as regards the clinical use of immune checkpoint inhibitors, has been made [127], durable response rates remain low [128], and current sarcoma immunotherapies still fails to induce an antitumoural response [129] implying that other immunosuppressive activities or effects are possibly present. Among these, tumour-derived extracellular acidosis may have an unexplored role. Indeed, in other types of cancer, the formation of an acidic microenvironment represents an efficient tumour strategy and forms such as an immune sanctuary to overcome immune surveillance since it profoundly alters the functions of cells of the immune system, including T cells, neutrophils, macrophages and dendritic cells (DCs) [130,131]. In particular, both cancer and immune cells are highly dependent on the glycolytic pathway for survival, proliferation, and activity. An increased rate of glycolysis, as it occurs in cancer environment, leads to a significant decrease in glucose availability and, although cancer cells can enter quiescence in the absence of glucose, activated T cells are not able to survive without glucose when attempting to expand into an acidic environment [132]. Notably, a high extracellular concentration of protons impairs glycolysis per se [96,133]. Furthermore, an acidic pH blocks the activation and antitumour functions of T cells in vitro through sequestration of interferon-gamma (IFN-γ) [134]. Tumour acidity also promotes tumour progression by negatively affecting the maturation and function of Th1 lymphocytes while stimulating the progression of tumour-promoting Th2 lymphocytes by inactivation of IFN-γ and suppression of tumour necrosis factor-α [132]. Finally, several lines of evidence have suggested that the contribution of extracellular acidosis to cancer growth is related to both the suppression of T cell function and to modulatory effects on additional cells of the immune system. In particular, Husain et al. demonstrated that tumour-derived lactate inhibits natural killer (NK) cell function, both directly and indirectly, i.e., by increasing the numbers of myeloid-derived suppressor cells (MDSCs) that, in turn, inhibit NK cytotoxicity [135]. A low pH also reprogrammes tumour-associated macrophages (TAMs) into a proangiogenic phenotype [136], activates neutrophils [137,138], and improves the antigen-presenting capacity of DCs derived from murine bone marrow [139]. However, no investigations have been performed thus far on sarcomas in this regard. Future studies will help find possible novel approaches to improve the outcomes of immunotherapy in sarcoma patients.

### **5. Targeting Acidosis in Bone Sarcomas**

The use of preclinical models that can mimic the extracellular acidic sarcoma microenvironment and the selection of assays that are not technically affected by the presence of an acidic pH have been fundamental for the identification of novel targets and the development of effective therapeutic strategies against acidosis in cancer. In vitro, monolayer models have to face the caveat of acidic pH being adjusted by a buffer solution that cannot be regulated throughout the experiment. In 3D experiments, spheroids or organoids grown in unbuffered media can, instead, adjust the pH to their own metabolic features and to the intrinsic acidification processes. The latter method has the advantage of resembling the physiological pH regulations seen in vivo. Additionally, the expression and the activity of reporter or housekeeping genes that are commonly used to study the induction or the inhibition of specific targets or proteins, such as b-actin or the green fluorescent protein (GFP), can be strongly affected by an acidic microenvironment. A recent paper has highlighted that among the most commonly used housekeeping genes, only YWHAZ, GAPDH, GUSB, and 18S rRNA are stable throughout pH modifications [140]. Furthermore, scientists working in this field of research should be aware that the fluorescence of wildtype GFP is stable from pH 6 to 10 but decreases at pH < 6 and increases from pH 10 to 12 [141]. The pH stability of GFP can also be exploited for specific purposes: for example, the superecliptic pHluorin (SEP) is a mutant GFP widely used in vitro as a pH reporter, as it is nearly nonfluorescent at pH 6 but brightly green at pH 7.4 [142].

Additionally, in vivo, the assessment of pH imaging methods is invasive, costly, or requires long acquisition times, and in some cases may not be suitable for high-throughput preclinical animal studies. Imaging methods include CEST-MRI, a quantitative method that accurately recapitulates tumour pH maps [143], or pH-sensitive ratiometric reporters such as pHLuc [142]. Despite the limitations, these imaging methods are of crucial importance in the assessment of therapeutics based on targeting cancer acidosis.

The therapeutic strategies that have been developed to target cancer acidosis are based on several approaches, namely, (1) hampering of proton extruders/ion transporters; (2) targeting cancer cell lysosomes through the use of photodynamic therapy; (3) use of inhibitors of acid-sensing ion channels that can possibly hinder the activation of the tumourassociated stroma (see Section 4.1). However, the last class of drugs has been extensively studied only as analgesic and anxiolytic drugs, and as drugs for the treatment of ischemic stroke [144], but has never been considered thus far for the treatment of sarcomas. Finally, recent evidence has highlighted molecular pathways that are selectively activated in acidictreated cells. These pathways can regulate oncogenes or oncometabolites or be involved in the generation of bioactive lipids. In the former case, the RAB39A-RXRB axis has been shown to have a prominent role in the development of osteosarcoma stemness and aggressiveness at a pH of 6.5 [145], while in the latter case, the pH-dependent accumulation of S1P seems to be of paramount importance in the survival and growth potential of osteosarcoma xenografts [82].

### *5.1. Hampering Proton Extruders/Ion Transporters*

Several drugs have been developed to target ion extruders/transporters as anticancer therapy. Inhibitors of the V-ATPase and CAIX have been the most explored for treating sarcomas. Studies considering these two approaches are listed in Tables 1 and 2.

To specifically target the V-ATPase, siRNA or Bafilomycin have been taken into account; nonetheless, their use can be hardly translated to the clinic for their instability or high toxicity, respectively. On the contrary, the use of proton pump inhibitors (PPIs), such as omeprazole or esomeprazole, has been extensively investigated. PPIs are acidactivated pro-drugs that reduce gastric acid production by inhibiting the H+/K<sup>+</sup> -ATPase pump and have been successfully used for the treatment of peptic disease [146]. Intriguingly, when used at high concentrations, PPIs can also effectively inhibit the activity of V-ATPase [147,148]. In preclinical models of bone sarcomas, although tumour growth was unaffected, treatment with a high concentration of PPI significantly increases the sensitivity

to doxorubicin [16,37,95] Finally, in a multicentre trial on human patients, pretreatment with omeprazole increased the local cytotoxicity of standard chemotherapy, as expressed by the increased percentage of tumour necrosis. This was particularly evident in chondroblastic osteosarcoma, an histological subtype that normally shows poor histological response [149].

CAIX targeting has shown successful results with the use of sulphonamide-derived inhibitors. Among them, a compound obtained starting from benzenesulphonamide derivatives (covered by patent) has been successfully used to inhibit tumour growth in a xenograft model of osteosarcoma. Although not well investigated yet, the use of this compound is quite promising, since, among the different CA isoforms, CAIX appears to be highly and selectively expressed in cancer cells, concomitantly implying less toxicity and an increased selective anticancer effect [40]. In a recent paper, Tauro et al. have developed a dual CA/matrix metalloproteinase inhibitor incorporating a bisphosphonic acid, which increases selective anticancer targeting [150]; this drug possibly and directly targets tumourinduced osteolysis by combining a cargo molecule of bisphosphonate that delivers a blocker of MMP-mediated invasion and an inhibitor of CAIX-mediated acidification to the site of osteolysis.

Regarding the use of MCT1 and NHEs inhibitors for the treatment of bone cancers, very few in vitro data have been reported (see Tables 3 and 4), with the exception of the use of [alpha]-Cyano-4-hydroxycinnamate (CHC) that, in an orthotopic model of osteosarcoma, strongly impaired both chemoresistance and tumour growth [46].

### *5.2. Targeting of Cancer Cell Lysosomes by Photodynamic Therapy*

Photodynamic therapy (PDT) is defined as the photo-induced irreversible destruction of abnormal cells and is based on the uptake of a photosensitiser molecule which, upon being excited by visible or near-infrared light, reacts with oxygen and generates reactive oxygen species (ROS) in target tissues, leading to cell death. PDT is therefore a minimally invasive anticancer modality with low-power light energy. ROS comprise singlet oxygen, superoxide anion, and radicals that generate from the conversion of molecular oxygen that reacts with the triplet state of the photosensitiser that is formed via photoexcitation. The generated free ROS oxidise biological substances, including nucleic acids, lipids, and proteins, leading to severe alterations in cell signalling cascades or in gene expression regulation and to activation of death-promoting physiological responses.

As discussed in Section 2.2, to avoid intracellular acidification, the excess of protons in the cytosol of tumour cells may be pumped into the lumen of the lysosomes, thereby decreasing the intra-organelle pH [16,151,152]. Acridine orange is a fluorescent cationic dye originally known as a detector of bacteria and parasites and an antimalarial drug. More recently, it has been described as an anticancer agent [153]. Since it has a low molecular weight, acridine orange easily diffuses into interstitial tissues and the cytoplasm and, due to protonation, accumulates into intracellular acid vesicles, leading to the formation of membrane-impermeable monomeric, dimeric, or oligomeric aggregates [16,151,152]. Acridine orange has thus a strong and selective tropism for tumour cells, as tumour cells have more acidic vesicles than normal cells because of their specific ability to effectively reduce the excess of protons in the cytoplasm by active transport across the plasma membrane and storage within the lysosomal compartment [154]. Furthermore, when photo-activated by blue light (466.5 nm) [155], or exposed to low-dose (1–5 Gy) X-ray irradiation [156], it generates singlet oxygen (1O2) thereby acting as an acid-targeting photosensitiser. The formed reactive species oxidise the fatty acids of the lysosomal membrane, causing the leakage of lysosomal enzymes and protons, followed by cell death [157].

To date, several data have demonstrated that acridine orange exerts selective cytocidal effects on tumour cells, showing no toxicity on normal cells. Furthermore, in the last 20 years, a combined technique of PDT and radioactivation (RDT) of acridine orange has been successfully developed and applied to clinical cases, demonstrating excellent outcomes in terms of inhibition of local recurrence and preservation of limb function after intra- or marginal tumour resection. These studies include humans affected by bone sarcomas, although the same type of approach has been tested in companion animals with spontaneous fibrosarcoma [158–162]. Specifically, following marginal or even intralesional gross removal of the tumour, it was possible to selectively target residual sarcoma and spare the surrounding normal tissues, with a satisfactory functional result. The procedure is safe without local or systemic complications. This technique proved to be particularly advantageous in sarcomas arising around the forearm and a valid alternative to wide surgical resection followed by limb reconstruction, without increasing the local recurrence rate [163]. Systemic administration of acridine orange with low-dose radiation therapy is also under evaluation for nonresectable bone sarcomas. The procedure appears to be safe and preliminary results are encouraging.

Talaporfin, also known as aspartyl chlorin, mono-L-aspartyl chlorin e6, NPe6, or LS11, is another photosensitiser that can target lysosomes and has been proposed for the treatment of bone sarcoma in addition to acridine orange [164–166]. Talaporfirin is uptaken by sarcoma cells through a KRAS-dependent endocytotic process. However, the correlation between its selective targeting and the degree of lysosomal acidification has not been unveiled yet.

### **6. Conclusions**

After over 10 years of research, the crucial role of acidosis in bone sarcoma growth and progression has been clearly established. However, the development of acid-targeted drugs for the treatment of bone sarcomas is still in its infancy. To date, most of the drugs targeting ion/proton extruders and transporters have failed to be translated to clinical trials. One possible explanation is the redundancy of cellular systems controlling pHe. Thus, their targeting is quite challenging: it can easily turn to be ineffective, or when it works, extremely toxic. Nevertheless, given the relevance of intratumoural acidosis in bone cancers, the use of CAIX inhibitors, acid-targeted PDT strategies, or novel drugs that can safely and selectively impair the protumourigenic pathways that are selectively induced by extracellular acidosis may hold, for the future, helpful results to improve patient survival.

**Author Contributions:** Conceptualisation and writing—original draft preparation, S.A.; writing review and editing, G.D.P., M.C., N.B. and S.A.; funding acquisition, N.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** Funded by AIRC under IG 2018, grant number 21403 (to. N.B.), and by the financial support for Scientific Research 5xMille (to N.B.).

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

### **References**

