Hypoxia-Inducible Factor-2-Altered Urothelial Carcinoma: Clinical and Genomic Features

Background: Hypoxia is recognized as a key feature of cancer growth and is involved in various cellular processes, including proliferation, angiogenesis, and immune surveillance. Besides hypoxia-inducible factor 1-alpha (HIF-1α), which is the main mediator of hypoxia effects and can also be activated under normoxic conditions, little is known about its counterpart, HIF-2. This study focused on investigating the clinical and molecular landscape of HIF-2-altered urothelial carcinoma (UC). Methods: Publicly available next-generation sequencing (NGS) data from muscle-invasive UC cell lines and patient tumor samples from the MSK/TCGA 2020 cohort (n = 476) were interrogated for the level of expression (mRNA, protein) and presence of mutations, copy number variations, structural variants in the EPAS1 gene encoding HIF-2, and findings among various clinical (stage, grade, progression-free and overall survival) and molecular (tumor mutational burden, enriched gene expression) parameters were compared between altered and unaltered tumors. Results: 19% (7/37) of UC cell lines and 7% (27/380) of patients with muscle-invasive UC display high EPAS1 mRNA and protein expression or/and EPAS1 alterations. EPAS1-altered tumors are associated with higher stage, grade, and lymph node metastasis as well as with shorter PFS (14 vs. 51 months, q = 0.01) and OS (15 vs. 55 months, q = 0.01). EPAS1 mRNA expression is directly correlated with that of its target-genes, including VEGF, FLT1, KDR, DLL4, CDH5, ANGPT1 (q < 0.001). While there is a slightly higher tumor mutational burden in EPAS1-altered tumors (9.9 vs. 4.9 mut/Mb), they are enriched in and associated with genes promoting immune evasion, including ARID5B, SPINT1, AAK1, CLIC3, SORT1, SASH1, and FGFR3, respectively (q < 0.001). Conclusions: HIF-2-altered UC has an aggressive clinical and a distinct genomic and immunogenomic profile enriched in angiogenesis- and immune evasion-promoting genes.


Introduction
Urothelial carcinoma is a lethal disease, particularly at advanced stages, yet it has seen unprecedented advances over the last five years. Novel therapeutic strategies, including immune checkpoint inhibitors (ICIs), antibody-drug conjugates (ADCs), and targeted therapies, for example against fibroblast growth factor receptor (FGFR)-mediated signaling, have entered the treatment paradigm and are leading to better patient outcomes [1][2][3][4][5]. Understanding who should get what treatment at what time, as well as predictive and prognostic biomarkers, will be key. Dissecting the molecular landscape of the disease may help identify particular subsets that could benefit from tailored therapies.
A major aspect in the development of various tumors, including UC, is growth under hypoxic or/and normoxic conditions that is orchestrated by activated signaling through two major transcription factors, hypoxia-inducible factor 1-alpha (HIF-1α) and HIF-2. High immunohistochemical (IHC) expression of HIF-1α in primary UC tumors is associated with higher-grade disease, vascular endothelial growth factor-related angiogenesis, and worse prognosis with regard to disease-free and overall survival in both superficial and invasive disease [6,7]. Hypoxia-induced autophagy may also propagate chemoresistance to cisplatin via the HIF-1α pathway [8]. Less is known about the role of HIF-2 and whether it may act as an oncogenic driver in UC.
This study aimed at unraveling the clinical and molecular landscape of HIF-2-altered UC by investigating publicly available next-generation sequencing (NGS) data from muscleinvasive UC cell lines and patient tumor samples.

Materials and Methods
The Cancer Cell Line Encyclopedia (CCLE) database [9] was used to query various primary cell lines for EPAS1 mRNA and protein expression.

Patient Characteristics
A publicly available database, cBioportal for Cancer Genomics (www.cbioportal.org, accessed on 27 September 2022), was used to query DNA and RNA sequencing data for EPAS1 mutations, copy number alterations, structural variants, mRNA and protein expression in a prospective multicenter cohort of 476 patients with muscle-invasive bladder UC (https://www.cbioportal.org/study/summary?id=blca_msk_tcga_2020, accessed on 25 October 2022).
The cohort included 334 patients analyzed via whole-exome sequencing and 142 patients analyzed via Memorial Sloan Kettering (MSK)-IMPACT sequencing. Sequencing was performed on fresh frozen or formalin-fixed paraffin-embedded specimens obtained via transurethral resection or RC. All patients were chemotherapy-naïve.
All computational analyses for exploring multidimensional cancer genomics data after integrating clinical profiles were performed as previously described [10,11].

Statistical Analyses
The Kaplan-Meier method was used to assess the association between altered and unaltered EPAS1 with progression-free survival (PFS) and overall survival (OS), using a threshold z-score of ≥2.0. The Chi-squared and Kruskal-Wallis tests were used to compare clinical and pathological characteristics with altered vs. unaltered EPAS1 tumors/patients. The student's t-test was used to assess correlations between the expression of EPAS1 and highly expressed genes in EPAS1 altered vs. unaltered tumors. The Spearman's correlation test was used to measure the degree of association between the expression of selected genes and EPAS1. Multiple hypothesis test correction was applied using the Benjamini-Hochberg method. p and q values of <0.05 were considered significant for all analyses.

EPAS1 Expression in UC Cancer Cell Lines
To assess the expression of HIF-2 in UC relative to various cancer types, the Cancer Cell Line Encyclopedia (CCLE) [9] was interrogated for mRNA and protein levels of EPAS1, the gene encoding HIF-2 protein. Seven out of 37 bladder UC cases (19%) displayed high EPAS1 expression, which was the highest proportion among all different cell types ( Figure 1). Thus, HIF-2 is overexpressed in up to one-fifth of UC cancer cells and could represent another mechanism of UC proliferation and growth.

EPAS1 Genomic Alterations in UC Patients
To dissect the molecular landscape of EPAS1 molecular alterations in UC, publicly available data from 476 patients from MSK/TCGA 2020 were analyzed through cbioportal.org. EPAS1 was altered in 27 (7%) of patients-samples ( Figure 2).
There were six mutations in total, of which five were missense and one splice (Table 1, Figure 4).  Thus, HIF-2 is overexpressed in up to one-fifth of UC cancer cells and could represent another mechanism of UC proliferation and growth.

EPAS1 Genomic Alterations in UC Patients
To dissect the molecular landscape of EPAS1 molecular alterations in UC, publicly available data from 476 patients from MSK/TCGA 2020 were analyzed through cbioportal.org. EPAS1 was altered in 27 (7%) of patients-samples ( Figure 2). Thus, HIF-2 is overexpressed in up to one-fifth of UC cancer cells and could represent another mechanism of UC proliferation and growth.

EPAS1 Genomic Alterations in UC Patients
To dissect the molecular landscape of EPAS1 molecular alterations in UC, publicly available data from 476 patients from MSK/TCGA 2020 were analyzed through cbioportal.org. EPAS1 was altered in 27 (7%) of patients-samples ( Figure 2).
There were six mutations in total, of which five were missense and one splice (Table 1, Figure 4).

Prognostic Value of EPAS1 Alterations in UC Patients
This study then sought to examine whether there is a prognostic significance of EPAS1-altered UC in patients with muscle-invasive bladder UC. Kaplan-Meier analysis revealed that patients whose tumors displayed EPAS1 overexpression had a significantly shorter progression-free survival (PFS) (14 months) compared to those without alterations (51 months, q-value = 0.0123) (Figure 5a). Likewise, high EPAS1 expression predicted a shorter overall survival (OS, 15 vs. 55 months, q-value = 0.0123) (Figure 5b).

Prognostic Value of EPAS1 Alterations in UC Patients
This study then sought to examine whether there is a prognostic significance of EPAS1-altered UC in patients with muscle-invasive bladder UC. Kaplan-Meier analysis revealed that patients whose tumors displayed EPAS1 overexpression had a significantly shorter progression-free survival (PFS) (14 months) compared to those without alterations (51 months, q-value = 0.0123) (Figure 5a). Likewise, high EPAS1 expression predicted a shorter overall survival (OS, 15 vs. 55 months, q-value = 0.0123) (Figure 5b).

Prognostic Value of EPAS1 Alterations in UC Patients
This study then sought to examine whether there is a prognostic significance of EPAS1-altered UC in patients with muscle-invasive bladder UC. Kaplan-Meier analysis revealed that patients whose tumors displayed EPAS1 overexpression had a significantly shorter progression-free survival (PFS) (14 months) compared to those without alterations (51 months, q-value = 0.0123) (Figure 5a). Likewise, high EPAS1 expression predicted a shorter overall survival (OS, 15 vs. 55 months, q-value = 0.0123) (Figure 5b).

Clinical and Molecular Characteristics of HIF-2-Altered UC
To investigate any specific phenotypic/genotypic pattern in the subset of patients with HIF-2-altered UC, we compared the most common clinical, pathological, and molecular characteristics with those that did not demonstrate any HIF-2 alterations. The two groups were equally distributed with respect to age, sex, and race ( Figure S1). There were 325 patients in total who were first diagnosed with primary muscle-invasive bladder cancer (MIBC), whereas 55 had progressed to T2 status (secondary MIBC) after an initial diagnosis of non-MIBC ( Figure S1). The median number of mutations (298 vs. 144), as well as the total mutational burden (TMB) (9.9 vs. 4.9 mut/Mb) and nodal involvement (52% vs. 29%), were proportionately, but not significantly, more frequent in patients with tumors harboring high HIF-2 expression compared to those without HIF-2 alterations ( Figure 6).

Clinical and Molecular Characteristics of HIF-2-Altered UC
To investigate any specific phenotypic/genotypic pattern in the subset of patients with HIF-2-altered UC, we compared the most common clinical, pathological, and molecular characteristics with those that did not demonstrate any HIF-2 alterations. The two groups were equally distributed with respect to age, sex, and race ( Figure S1). There were 325 patients in total who were first diagnosed with primary muscle-invasive bladder cancer (MIBC), whereas 55 had progressed to T2 status (secondary MIBC) after an initial diagnosis of non-MIBC ( Figure S1). The median number of mutations (298 vs. 144), as well as the total mutational burden (TMB) (9.9 vs. 4.9 mut/Mb) and nodal involvement (52% vs. 29%), were proportionately, but not significantly, more frequent in patients with tumors harboring high HIF-2 expression compared to those without HIF-2 alterations (Figure 6).

EPAS1 Is Associated with Expression of Immune Suppression Genes
Since UC is known to be immunogenic and responsive to immunotherapy including intravesical BCG and ICIs, this study sought to investigate whether this "aggressive" and poor prognosis subset of HIF-2-altered tumors could be associated with aberrant expression of key genes involved in regulating immune responses in UC. Based on previous works from others and ours [15][16][17], FGFR3 and IFNG genes were chosen, and their mRNA expression was assessed in association with EPAS1.

EPAS1 Is Associated with Expression of Immune Suppression Genes
Since UC is known to be immunogenic and responsive to immunotherapy including intravesical BCG and ICIs, this study sought to investigate whether this "aggressive" and poor prognosis subset of HIF-2-altered tumors could be associated with aberrant expression of key genes involved in regulating immune responses in UC. Based on previous works from others and ours [15][16][17], FGFR3 and IFNG genes were chosen, and their mRNA expression was assessed in association with EPAS1.

Discussion
This study examined the significance of HIF-2 in UC. First, by exploring HIF-2 mRNA and protein expression among several different cell types, it provides evidence that bladder UC is the most commonly affected tumor type with the highest frequency of HIF-2/EPAS1 overexpression. Second, by using the largest-to-date cohort of muscle-invasive UC this study revealed a distinct molecular and clinical profile in a subset of patients carrying somatic alterations in HIF-2/EPAS1.While these occur in nearly one-tenth of patients/tumors, they are associated with more aggressive histopathological features including higher T stage, presence of adenopathy, and high-grade disease. The presence of EPAS1 alterations, the majority of which involves high EPAS1 expression and/or amplification, is predictive of shorter PFS and OS. At the molecular level, HIF-2-altered UC tumors display a direct association between EPAS1 mRNA and expression of its target-genes, including VEGF, FLT1, KDR, DLL4, CDH5, and ANGPT1. While there is a slightly higher overall mutation burden in EPAS1-altered tumors, they are enriched in and associated with genes promoting immune evasion, including ARID5B, SPINT1, AAK1, CLIC3, SORT1, SASH1, and FGFR3, respectively.
Expression of HIF-2/EPAS1 in bladder UC cell lines and a small number of paraffinembedded samples from patients has been previously reported to be more abundant in muscle-invasive compared to superficial disease [18]. Interestingly, HIF-2 protein was not found in cancer cells or in normal tissues but rather in stroma around cancer cells, particularly in tumor-associated macrophages (TAMs) within perinecrotic regions, whereby it correlated with higher pathological stage, grade, and VEGF-related tumor angiogenesis [19]. Moreover, patients with tumors harboring EPAS1-expressing TAMs were characterized by shorter cancer-specific survival [20].
This study, by providing a more comprehensive assessment of HIF-2 genomic alterations at the DNA, RNA, and protein level, confirms and complements previous preliminary evidence, supporting a negative prognostic role of HIF-2/EPAS1 expression on both PFS and OS. More importantly, this study describes for the first time the molecular landscape of HIF-2-altered muscle-invasive UC. It shows that HIF-2-overexpressing or/and amplified UC tumors are characterized by active angiogenic signaling, evidenced by the direct association of expression of EPAS1 with HIF-2-regulated genes involved in the initiation of angiogenesis (VEGF, FLT1, KDR, CDH5, DLL4), neo-vessel formation (VEGF, FLT1, KDR, CDH5, DLL4) and maturation (VEGF, ANGPT1) [11]. Further, these HIF-2-altered tumors demonstrate a slightly higher but not significantly different mutational load compared to UC lacking HIF-2 alterations. Thus, while an overall TMB of above 10 is usually predictive of response to ICIs in various primaries [21], HIF-2-overexpressing tumors behave rather in the opposite direction due to overexpression of ARID5B, SPINT1, AAK1, CLIC3, SORT1, and SASH1.
The AT-rich interaction domain (ARID) family is a superfamily belonging to switch/ sucrose nonfermenting (SWI/SNF) chromatin remodeling complexes and the presence of inactivating mutations in any of their members, including ARID5B, has been associated with greater benefit from ICI therapy in pan-cancer analyses [22]. Conversely, the enrichment of ARID5B expression in HIF-2-altered tumors could suggest innate resistance to ICIs. SPINT1, also known as hepatocyte growth factor activator inhibitor 1 (HAI-1) is an endogenous protease inhibitor of HGF that is found at increased levels in tumors and urine from patients with muscle-invasive UC [23] and may be involved in migration and metastasis [24]. The finding of enhanced SPINT1 expression in HIF-2-altered tumors could also have an impact on these tumors' resistance to ICIs in view of recent data including SPINT1 in an 8-gene prognostic signature that stratifies patients with bladder UC into two risk groups with distinct immune profile and responsiveness to immunotherapy [25]. AAK1 is involved in clathrin-mediated endocytosis of chemokine receptors and AAK1-inhibition in mice resulted in increased intratumoral infiltration, supporting an immune-suppressive role for this gene product, as well [26]. CLIC3 is a chloride intracellular channel protein the gene expression of which has been correlated with low immune infiltration of myeloid dendritic cells (DCs) and poor prognosis in hepatocellular carcinoma [27]. Therefore, it is possible that a similar immune suppressive role could be exerted in HIF-2-altered UC. SORT1 is another gene enriched in HIF-2 amplified UC tumors that are likely to be involved in allowing cancer cells to evade the immune system, given previously reported negative correlation with the infiltration levels of DCs, cytotoxic T cells, NK CD56dim cells, Tgd, and pDCs, as well as a positive correlation with immune evasion checkpoints including PDCD1, CD274, and CTLA-4 [28]. SASH1 encodes a scaffold molecule involved in NF-kappa-B activation and promotes immune escape via subsequent upregulation of PD-L1 expression [29,30].
In further support of this immune-suppressive environment forged by HIF-2 overexpression in UC is the presence of a direct association between EPAS1 and FGFR3 mRNA levels while the opposite was the case between EPAS1 and IFNG expression. FGFR3, as previously described by many groups, including ours, is a major player causing immune resistance in both lower and upper tract UC which is often associated with attenuated IFNG signaling [17]. In presence of an approved therapy for advanced UC, erdafitinib [5], FGFR3-inhibition could become a valid strategy to overcome the "immune-cold" nature of HIF-2-altered UC. Besides that, specific HIF-2 inhibitors, such as belzutifan, are already designed and in an advanced phase of testing in renal cell carcinoma, with promising activity [31]. Stromal HIF-2 exerts an immune suppressive role in other tumor types as well, for example, pancreatic adenocarcinoma [32]. By revealing a network of genes-players that are involved in this process in direct association with HIF-2 in UC, this study provides several insights for deepening our understanding of the mechanistic processes that are potentially involved.
This study was limited by its computational design and single cohort evaluation, which is however the largest reported-to-date with both molecular and survival data available. This work represents the first comprehensive effort to study the molecular and clinical significance of HIF-2-altered UC. This study's findings place UC within a particular group of cancers, including renal cell carcinoma, HER2-positive breast cancer, hepatocellular carcinoma, and head and neck cancers, in which HIF-2/EPAS1 expression has prognostic value [42][43][44][45][46]. Additional studies are needed to further elucidate the genomic and immunogenomic profile of HIF-2-altered UC in order to design targeted therapies for this small but important subset of patients.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/curroncol29110681/s1, Figure S1: Clinico-pathological characteristics without significant differences between EPAS-altered and unaltered tumors. Informed Consent Statement: Informed consent was obtained from all subjects involved in this study.
Data Availability Statement: All data used for this analysis are available at https://www.cbioportal. org/study/summary?id=blca_msk_tcga_2020 (accessed on 25 October 2022).

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