Overexpression of carbonic anhydrase and HIF-1α in Wilms tumours

Wilms tumour (WT) is the most common paediatric renal tumour, with an incidence of 1:10,000 children. Although the prognosis of children with favourable clinical and pathological factors is good, children with high-risk, drug-resistant or metastatic disease have a mortality rate of 10-15%. Alternative therapeutic strategies are therefore essential to improve the outcomes of the children in this group.

Tissue hypoxia contributes to the progression and chemoresistance of many human cancers by inducing the up-regulation of genes associated with angiogenesis, metabolism, cell survival and apoptosis [1]. The cellular response to hypoxia is mediated by a key gene, hypoxia-inducible transcription factor-1 α-subunit (HIF-1α), via hypoxia response elements in the promoter regions of target genes, which include carbonic anhydrase IX (CA9) and vascular endothelial growth factor (VEGF)[1]. CA IX, a membrane glycoprotein, plays an important role in the growth and survival of tumour cells under both normoxic and hypoxic conditions [2, 3]. Although hypoxia plays a major role in HIF-1α stabilisation and activation, the amplitude of the response can be modulated in normoxia by growth factors, oncogenes and cytokines [4].

Up-regulation of HIF-1α and CA IX is observed in a variety of epithelial cancers and correlates with poor survival [1, 5]. Epithelial cancers typically occur in adults and only rarely occur in children. In contrast, solid tumours (such as WTs) that occur in children are developmental disorders presenting different biological characteristics [6]. In a study consisting of 18 WT samples, HIF-1α and VEGF expression was detected in all WTs using immunohistochemistry [7]. In a recent study, up-regulation of CA9 in WTs was associated with high-risk histology and was significantly correlated with metastases [8]. The existence of two variants of CA9, full length (FL) and alternate spliced variant (ASV), has been described in both hypoxic and non-hypoxic conditions [9, 10]. The ASV mRNA, which lacks the exons 8-9, competes with FL in regulating extracellular pH in mildly hypoxic conditions. ASV is expressed in both normal and diseased tissues, while FL mRNA is upregulated in non-small cell lung cancer and its expression associated with poor prognosis [10]. Wittmann et al used primers that did not distinguish between the two isoforms of CA9 [8].

As CA IX is an attractive target for anti-cancer therapy for several reasons, overexpression of CA9 [8] is an important finding that is likely to be clinically relevant [11, 12]. Although changes in protein expression are more likely to have a direct impact on biological behaviour than changes in transcriptional regulation or mRNA expression, there is no information about CA IX expression in WTs. The aims of this study were as follows: 1) to perform a comprehensive expression analysis of carbonic anhydrase at mRNA (FL and ASV CA9) and protein levels and HIF-1α protein in matched-paired WTs and non-neoplastic kidneys (NKs), precursor lesions (nephrogenic rests) [13], and fetal kidneys; (2) to look for associations between the expression of either CA IX or HIF-1α and clinicopathological features; and (3) to compare the expression of CA IX with other hypoxia markers and vascularity.

This study is the first to assess the cellular distribution of CA IX protein and quantify it in Wilms tumours (WTs). The main strengths of our study include the use of a wide range of techniques for expression analysis and the comparison between WTs and matched-paired NKs to reduce the confounding variation inherent in biological samples. Our study clearly demonstrates that a) CA IX is over-expressed in up to 63% of WTs at the cellular level and b) CA9 FL and CA IX levels are significantly increased in WTs compared with NKs. We also demonstrate that HIF-1α protein is up-regulated in WTs compared to NKs. The present study also demonstrates the co-expression of CA IX with other hypoxia markers such as VEGF and GLUT1 in addition to HIF-1 α. The presence of CA IX in well vascularised areas, as shown by double staining of CD31 with CA IX suggests that CA IX is regulated by other factors other than hypoxia.

By using the M75 antibody, which recognises the extracellular proteoglycan-like domain of CA IX and is considered the gold standard in immunohistochemical CA IX detection, CA IX was detected in all three WT components. In our samples, the stained tumour fraction was 5%, an observation in agreement with the low frequencies previously described by others [19, 21]. Generally, the method used to establish the prognostic significance of HIF-1α and CA IX expression in other tumours is immunohistochemistry [1, 5]. Intratumoral heterogeneity, however, is a feature of CA IX cellular detection that is emphasised by this study and has been reported for other tumours [21, 22]. In a seminal study of 110 previously treated cervical cancers, measurement of CA IX in multiple biopsies indicated that intratumoral heterogeneity accounted for 41% of the total variance in the data set [21]. We used conventional, whole tissue sections to reduce the inherent sampling error resulting from distributional heterogeneity of CA IX [22] and in small biopsies. We were unable to detect a significant association with clinicopathological variables, including metastatic status. A possible explanation for the lack of association is that there were too few samples in some subgroups to effectively evaluate the relationship between CA IX expression and these variables.

We showed CA9 ASV expression using qualitative PCR in all NKs and WTs in keeping with previously published data, [9, 10]. We also showed an increase in CA9 FL in WTs compared to NKs using quantitative PCR. Our results are in line with increased CA9 FL expression shown in non-small cell lung cancer samples [10]. As CA9 FL was not expressed or expressed in very low amounts in normal kidneys, our study also indicates that it is a tumour-associated gene, which is an important finding because the other genes detected in WTs and nephrogenic rests are usually expressed during organogenesis [13]. The precise role of CA IX in WTs, however, remains to be elucidated. A recent study demonstrated that hypoxia down-regulates several genes involved in maintaining cell-cell interactions, including E-cadherin, which promotes epithelial to mesenchymal transition and invasion [23]. The proteoglycan-like region at the N-terminus of CA IX is implicated in cell adhesion [24, 25], and CA IX's distribution overlaps with that of E-cadherin in canine kidney epithelial cells in vitro [26]. Furthermore, CA IX co-precipitates with β-catenin, reduces the binding of E-cadherin to β-catenin and thereby destabilises intercellular adhesion [26]. Importantly, E-cadherin expression is not only decreased in WT [27‐29] but is also associated with advanced-stage disease [30]. Thus, it is possible that CA IX destabilises E-cadherin's linkage to the cytoskeleton, resulting in decreased cell-cell adhesion and contributing to WT invasion.

The cellular overlap between HIF-1α and CA IX in perinecrotic regions indicates that both are regulated by hypoxia in WTs. The localisation of CA IX in areas of tumour necrosis is of particular interest in terms of hypoxia regulation [15‐19, 31]. HIF-1α is the key transcriptional regulator of CA IX, and both co-localise in many embryonic and fetal tissues. A lack of co-expression in other tissues, however, indicates that CA IX is only partially regulated by HIF-1α expression. For example, HIF-1α, but not CA IX, is expressed in the human fetal and adult kidney as detected by immunohistochemistry [32, 33]. Although a predominant perinecrotic pattern was observed in our study, CA IX was also present in areas distant from necrosis. This partial correlation is likely due to differences in the kinetics of HIF-1α and CA IX. HIF-1α is rapidly induced and degraded within minutes of re-oxygenation, whereas CA IX, induced hours later, is relatively stable and has a half-life of 2-3 days [34].

HIF-1α and CA IX expression in perivascular areas is in agreement with their regulation by hypoxia-independent-pathways in WT and in line with previous studies of epithelial cancers [15‐19, 31]. Mild hypoxia generated in high-density cultures can also increase HIF-1α synthesis and CA IX expression via the phosphatidylinositol 3-kinase (PI-3K) pathway [35]. Extracellular acidosis, a common feature in tumours, increases HIF-1α protein accumulation and CA9 transcriptional activity under normoxic conditions [36]. Additionally, growth factors, such as transforming growth factor-α (TGF-α), can also increase CA IX expression by increasing HIF-1α synthesis [35] via both the PI-3K and mitogen-activated protein kinase pathways [4]. In contrast to regulation under hypoxic conditions, which affects all cells by stabilising HIF-1α protein and increasing HIF-1 transactivation function, this regulation is cell-type specific and works mainly by increasing HIF-1α synthesis [4]. Although neither activating mutations [37] nor the amplification of EGFR [38] have been detected, increased TGF-α and EGFR expression is observed in WTs at both the mRNA [8] and histological levels [39].

Our expression data clearly demonstrate that carbonic anhydrase (mRNA and protein) is up-regulated in untreated and treated WTs as compared with normal kidneys and WT precursor lesions (nephrogenic rests). We did not detect significant associations between either CA IX or HIF-1α protein and clinicopathological variables, including metastatic status in post-chemotherapy-treated WTs. Cellular localisation studies in untreated WTs suggest that CA IX and HIF-1α are regulated by hypoxia and non-hypoxia mechanisms.