Background
Prostate carcinoma is the most common cancer and the second cause of death due to malignancy in men [
1]. It is clinically heterogeneous in aggressiveness, not with standing comparable clinicopathological features. Currently, only few biomarkers assist prostate carcinoma risk and aggressiveness prediction [
2].
During tumor growth, malignant cells become progressively distant from the vasculature, oxygen supply and nutrients, urging tumor cells to signal to the microenvironment their needs. The hypoxia inducible factor 1 alpha (HIF-1α) is a key factor by which tumors regulate the response to hypoxia, triggering cascades with effects in angiogenesis, energy metabolism, vasomotor function and on apoptosis and proliferation activity [
3‐
5]. In hypoxia, the HIF-1α/HIF-1β complex binds hypoxia response elements in promoters of many downstream target genes, notably vascular endothelial growth factor (
VEGF), carbonic anhydrase IX (
CAIX), and lysyl oxidase (
LOX) promoters. They have been demonstrated to be up-regulated by hypoxia, ensuing aggressive and treatment-resistant tumor phenotypes [
3,
5‐
9]. A large randomized study on radiotherapy and surgical cohorts described that markers of tumor hypoxia and angiogenesis were relevant for localized prostate carcinoma and outcome of radical treatment [
10]. However, further studies at the genetic and protein levels are required to confirm molecules in hypoxia pathway as useful markers in prostate carcinoma.
Genetic variants may predispose to prostate carcinoma and influence the clinical outcome [
2,
11,
12]. Single nucleotide polymorphisms (SNPs) in genes coding for molecules involved in the response to hypoxia, particularly a functional polymorphism in
HIF1A gene at locus +1772 C > T [
13‐
20], has been studied in association with prostate carcinoma with controversial results. Current knowledge suggests that we should consider a panel of genes in hypoxia pathway, in order to provide more accurate prediction of the response to tumor hypoxia [
21,
22]. Therefore, despite functional SNPs in genes of pathways downstream of HIF-1α, such as
KDR,
LOX and
CAIX, have not been studied so far in prostate carcinoma patients, they merit further research as they represent key molecules in hypoxia-generated stimulus in cancer.
Based on the role of hypoxia-associated molecules in cancer cell biological behaviour and clinical outcome, we assumed there might be an association, at the genetic and protein level, between HIF1A, LOX, CA9 and KDR genetic variants, the protein expression and prostate carcinoma. Hence, if these polymorphisms modulate protein expression in response to tumor hypoxia, then the knowledge of the genotype could aid identify patients at higher risk for prostate carcinoma and eventually more aggressive disease, thereby making it possible to undertake chemoprevention strategies adjusted to the individual characteristics of the patient.
Discussion
Tumor-associated hypoxia was found in over 70% of solid malignancies, including prostate carcinoma [
3]. It promotes tumor progression and resistance to therapies through an effect in reducing apoptosis, and increasing tumor cell proliferation and neoangiogenesis [
5]. However, the hypoxia-driven HIF-1α upregulation also activates downstream pathways involved in metabolism (e.g. CAIX), angiogenesis (e.g. VEGF/VEGFR2 pathway) and extracellular matrix activity (e.g. LOX), which can modulate cancer behavior [
28].
Experimental studies with prostate cancer cells demonstrated that HIF-1α overexpression was associated with higher proliferation and metastatic potential [
29]. Likewise, a greater expression of HIF-1α has been found in human prostate carcinomas compared to nodular prostate hyperplasia [
30,
31]. For prostate carcinoma and other oncologic models, besides the observed higher amount of HIF-1α in tumors, increased HIF-1α expression was also associated with prognosis [
10,
32‐
35]. In the current study, we found a trend for higher HIF-1α protein expression in prostate carcinomas compared to nodular prostate hyperplasia, which may be explained by the limited samples analysed. The use of cytoplasmic rather than nuclear staining, is unlikely to have influenced our results, since this method has been published before, reporting positive associations of HIF-1α with prostate carcinoma and prognosis [
10,
30].
Albeit mainly distributed in vascular endothelial cells, also epithelial cells express VEGFR2 that signals through signal transducer and activator of transcription 3 (STAT3), mitogen-activated protein kinase (MAPK) or phosphoinositide-3-kinase (PI3K) intracellular signalling cascades [
36‐
38]. Unambiguously, the VEGFR2 was shown to regulate protein kinase B (Akt)/mammalian target of rapamycin (mTOR)/ribosomal protein S6 kinase beta-1 (P70S6K) signalling pathway in PC-3 prostate cancer cell line [
39]. In the present study, VEGFR2 was more frequently expressed in epithelial tumor cells of organ confined or extra prostatic carcinomas than in nodular prostate hyperplasia, and to lower extent in endothelial cells. Hence, at least in prostate tissue, VEGFR2 expression is not specific of endothelial cells; it is mainly expressed in malignant epithelium where VEGF can act as a promoter of tumor cell proliferation. The expression of VEGFR2 in epithelial prostate carcinoma cells has been rarely reported, and its role in the occurrence and development of prostate cancer remains unclear. Previous immunohistochemistry studies reported VEGFR2 expression in high-grade prostate intra-epithelial neoplasia and carcinomas of the prostate [
40‐
42], whereas gene expression findings evidenced expression of
KDR mRNA in prostate cancer cell lines and a functional impact of using a
KDR antisense oligonucleotide in suppressing cell proliferation and promoting apoptosis [
43,
44].
The body of past evidences, taken together with present findings indicates that the distribution of VEGFR2 expression towards epithelial prostate carcinoma cells supports a function for VEGF that is not limited to angiogenesis. Thus, abrogation of VEGFR2 signalling in malignant epithelial cells may prove an effective therapeutic modality for the treatment of prostate cancer. At present, two anti-angiogenic drugs are being tested in the phase III setting for men with prostate cancer, carbozantinib (a dual VEGFR2/MET inhibitor) and tasquinimod (down-regulator of HIF-1α), which previously showed beneficial and encouraging results on phase II trials [
45].
Cancer-associated hypoxia switches cell metabolism towards increased production of acidic metabolites. However, tumor cells have to adapt to hypoxia and acidosis in order to survive. CAIX is a membrane-bound protein crucial to a wide variety of processes, including pH regulation in the highly metabolically active malignant cells. Expression of CAIX is associated with tumor cell hypoxia in a variety of human tumors, including urologic cancers [
46‐
49]. Carbonic anhydrase IX gene (
CA9) is a target of HIF-1α that is up-regulated in response to hypoxia [
50]. The expression of CAIX in prostate carcinoma has been rarely reported.
CA9 mRNA expression increases reliably following hypoxia incubation of PC-3 cells [
51], although no significant differences in
CA9 mRNA expression were found when comparing nodular prostate hyperplasia with prostate carcinomas [
7]. However, other studies reported lack of CAIX expression in primary prostate carcinoma and hypothesized that alternative pathway for maintaining pH balance (e.g. monocarboxylate transporters 2 and 4) [
26,
52,
53] may be more relevant than CAIX.
Our results disclosed increased frequency of cases with epithelial cell positivity for CAIX expressing in organ confined and extra prostatic carcinomas compared to BPH. Despite recent concern arisen for the specificity of the CAIX polyclonal antibody generated against a C-terminal peptide in detecting CAIX (except when used at high dilution, in prostate tissues) [
54], in this study we used the antibody at a dilution of 1:1000 and found membrane-bound staining for CAIX. Therefore, our findings are likely to reflect reliable expression of CAIX in epithelial prostate cells. Our findings taken together with reports of CAIX expression in malignant prostate epithelial cells [
7,
51,
55] sustains the need for reconsidering CAIX role in prostate carcinoma. CAIX may serve as one of the mechanisms by which prostate carcinoma cells regulate extracellular pH and induce cytoplasmic alkalization.
The lysyl oxidase gene (
LOX), one of the overexpressed genes among a tumor hypoxia signature [
56,
57], was shown to be directly regulated by HIF-1α transcription factor and is essential for hypoxia-induced metastasis and cancer cell proliferation [
58]. Hypoxia-driven cancer cell invasion is severely impaired when LOX expression or oxidase activity were inhibited [
59]. In prostate tissue we found that the LOX immunoreactivity score correlated with HIF-1α expression, thus supporting the regulatory nature of HIF-1α in LOX expression. Furthermore, although we have not observed an overrepresentation of cases with positive LOX expression in carcinomas compared to nodular prostate hyperplasia, the LOX immunoreactivity score was significantly higher in organ confined prostate carcinomas compared to nodular prostate hyperplasia. Interestingly, previous reports showed significantly increased expression of
LOX mRNA in prostate carcinomas compared to nodular prostate hyperplasia [
7], whereas stronger LOX expression was also observed in other solid malignancies [
27,
60,
61]. LOX is known to participate in critical biological functions that include cell migration, cell polarity, epithelial-to-mesenchymal transition (EMT) and angiogenesis [
58] (reviewed in Fraga et al., 2015) [
62], which fits with the increased LOX expression found in our carcinomas. Altogether, we suggest the possibility that a HIF-1α/LOX regulatory mechanism may act in synergy to foster tumor formation along with the adaptation of tumor cells to hypoxia.
The analysis of protein expression in distinct pathological groups (by stage, differentiation score and PSA serum levels at diagnosis), which are predictive of prostate cancer aggressiveness, showed at most only trends for increased expression of VEGFR2 in carcinomas with Gleason >7 or patients with PSA > 10 ng/mL, and of CAIX in patients with PSA > 10 ng/mL. These findings indicate relevant clues but require further studies.
The genotypic distributions for the putative functional target SNPs in HIF1A, LOX, CA9 and KDR were similar between nodular prostate hyperplasia and prostate carcinomas. We might have hypothesized that carriers of variant alleles are prone to be more susceptible to have cancer, but the underpowered sample size limits conclusions regarding genetic association for these SNPs. Nevertheless, it is expected that only the combination of several SNPs within pathways or mechanisms may have significant impact in the association with complex diseases as prostate carcinoma. Further studies are warranted to evaluate the predictive/prognostic value of these genetic polymorphisms in prostate cancer.
In this study, evaluation of protein expression according to SNPs in the respective coding genes disclosed a genotype-phenotype effect for the
LOX and
KDR SNPs, but no functional validation at the protein level was observed for the studied
HIF1A and
CA9 SNPs. In the
HIF1A gene, a C-to-T substitution at locus +1772 (rs11549465) results in non-synonymous proline-by-serine aminoacid substitution at codon 582. Association studies of this SNP with prostate carcinoma risk and with microvessel density, yielded conflicting results [
13,
16,
19,
20,
63‐
65]. This SNP localizes in the oxygen-dependent domain of the gene where the variant allele was shown to stabilize
HIF1A mRNA and enhance
HIF1A transcriptional activity [
64]. In our study there were no differences in HIF-1α protein expression according to the
HIF1A +1772 C > T genotypes as reported previously in localised prostatic carcinomas [
16]. As we measured HIF-1α protein levels and it is known that
HIF1A is subjected to post-transcriptional and post-translational regulation [
66], this SNP may indeed influence mRNA transcription that is not reflected in protein expression. The low frequency of T homozygous genotype in our sample (only 2 cases carried TT genotype) may have influenced statistical power, since the HIF-1α protein and mRNA overexpression have been associated with the
HIF1A +1772 TT [
14,
67,
68].
A functional genetic variant on
KDR gene that codifies for VEGFR2 is located in the promoter region (−604, rs2071559), where a T-to-C substitution occurs. Preceding in vitro luciferase assays showed that the C-allele was associated with lower transcription activity than T-allele, whereas serum VEGFR2 levels were significantly lower in CC versus TT carriers [
69]. Interestingly, we found that CT and TT carriers had significantly increased VEGFR2 expression in prostate epithelial cells. We postulate that this SNP might prove useful for predictive and/or prognostic evaluations in prostate carcinoma. Studies in colorectal cancer reported association of this SNP in
KDR with susceptibility and recurrence [
70,
71], whereas, to the best of our knowledge, no studies using this SNP were conducted in prostate carcinoma patients. Likewise, it is expected that this SNP might increase susceptibility to prostate cancer by upregulating the number of available VEGFR2 proteins in malignant cells.
A SNP in exon 1 of
CA9 gene is located at locus +201 (rs2071676), where an A-to-G substitution leads to a change of valine-by-methionine in codon 33. Although we observed an overrepresentation of CAIX positive immunoreactivity in prostate carcinoma compared to BPH, the nonsynonymous SNP in
CA9 + 201 were unable to explain variations in the levels of CAIX protein expression in the prostatic tissue. Likewise, a recent report described lack of association between the
CA9 + 201 SNP with CAIX protein expression in renal cell carcinoma [
72]. These findings may suggest that lack of influence of this SNP in protein expression, even though the potential molecular structure modifications of this nonsynonymous substitution (valine to methionine) in CAIX protein activity remains to be confirmed. In fact, genetic association studies that included the
CA9 + 201 A > G polymorphism showed neither risk for renal cell carcinoma [
72] nor for oral squamous cell carcinoma [
73]. Noteworthy, the G-allele was associated with lymph node metastasis in oral cancer and represented increased risk for cancer when combined into a haplotype with other two SNPs in this gene [
73]. Furthermore, another SNP in
CA9 (rs12553173) was independently associated with improved overall survival and greater likelihood of response to therapy in renal cell carcinoma [
72], thus warranting further functional analysis. In our study, although we are aware that haplotype analyses can be expedite over analysis of individual SNPs for detecting an association between alleles and a disease phenotype, the small size sample prevented the consideration of such evaluation.
The
LOX gene is translated and secreted as a proenzyme (Pro-LOX), and then processed to a functional enzyme (LOX) and a propeptide (LOX-PP) [
74,
75]. While LOX-PP was described as a Ras tumor suppressor, reversing mesenchymal tumor cells to a more epithelial phenotype [
76‐
78], the LOX enzyme was found to facilitate a more migratory and invasive phenotype during breast cancer progression [
58,
79]. We studied a SNP in
LOX gene that has been identified at locus +473 (rs1800449), presenting a G-to-A substitution that cause an aminoacid substitution arginine-by-glutamine in codon 158. This SNP located in a highly conserved region within LOX-PP has been associated with attenuated ability of LOX-PP to oppose the effects of LOX, resulting in tumor cell invasive phenotype. Functional studies revealed that the A-allele decreases the protective capacity of LOX-PP, while increasing the Pro-LOX-associated invasive ability of tumor cells [
78]. When evaluating LOX immunoreactivity and expression intensity by immunohistochemistry in prostate tissues, we found it significantly lower in carriers of the
LOX +473 A-allele. Indeed,
LOX A-carriers disclosed decreased LOX protein expression in the nucleus of prostate epithelial cells.
The complex nature of LOX protein domain structure and biological functions makes noticeable that it can act as both a tumor suppressor and a metastasis promoter gene in cancer [
80]. Under hypoxic conditions, the increased expression of LOX enzyme correlates with tumor invasiveness [
81,
82]. In the present study, we found that lysyl oxidase was present primarily intracellular in the nucleus of epithelial cells, which fits with other reports asserting that this enzyme may have important functions in secretory cells, either as catalyser of histones in the nucleus or in association with cytoskeletal proteins at the cytoplasm [
83,
84]. Thus, our findings seem to suggest a wider variety of functions for LOX in prostate epithelial cells, beyond those related to cross-link formation in collagen and elastin, which merit further research. We hypothesize that the trafficking of LOX towards inside the cell or a specific cell compartment may be subordinated to the structural molecular characteristics and folding of the protein, which could be determined by
LOX +473 G > A polymorphism. Further studies should clarify the meaning of increased nuclear LOX intensity for PCa development.
Our endeavour to study the genotype-phenotype correlation in key hypoxia markers and its association with prostate cancer yielded novel and interesting findings, nevertheless our results should be interpreted in the context of several potential limitations. Sample size was a major issue as conclusions were impracticable for genetic association analysis and limited for genotype-phenotype inferences. Nevertheless, considering the hypothesis-generating nature of this study, we report findings that provide important clues to further work in larger samples. The use of tissue microarrays for immunohistochemical evaluation has been subject of concern mainly due to limited sample of diagnostic tissue, although in our series the representative tumor sections were adequately selected by an experienced pathologist. The comparison of hypoxia markers between patients with benign and malignant prostate disease might attenuate differences since it is known that hypoxia is altered in cancer but also in benign hyperproliferative diseases. The group of benign prostate disease seemed adequate for several order of reasons: 1) the diagnosis was contemporary with that of cancers; 2) their advanced age at diagnosis allowed matching with elderly prostate cancer patients; 3) all patients underwent digital rectal examination, PSA testing and prostate needle biopsy, making the possibility of crossover remote, and 4) most men develop nodular prostate hyperplasia or chronic prostatitis by the 7th–8th decades of life, making it normal in men of that age to carry benign prostatic disease.