Hypoxia-inducible factor 1 alpha expression increases during colorectal carcinogenesis and tumor progression

As a consequence of increased cellularity and proliferation, as well as enhanced metabolism within a tumor, the oxygen concentration in solid neoplasms is generally lower than in the adjacent non-neoplastic tissue [1, 2]. In fact, histopathological examination of carcinomas frequently reveals hypoxic areas within the tumor mass, mostly in the form of necrotic regions. As a reaction to hypoxia tumor cells can alter their metabolism and activate survival mechanisms. Hypoxia inducible factor 1 (HIF-1) is a transcription complex that plays a crucial role in coordinating the cellular response to oxygen stress conditions [3]. As a result of hypoxic stress, HIF-1 activates the transcription of a variety of genes regulating cell survival. HIF-1 is a heterodimer composed of one of the three alpha subunits (HIF-1α, HIF-2α or HIF-3α) and one HIF-1β subunit. Although HIF-1β is constitutively expressed, hypoxia-mediated responses are determined by HIF-α subunits. In normoxia, degradation of HIF-1α occurs. Hypoxia leads to HIF-1α stabilization, resulting in a translocation of HIF-1α to the nucleus and binding to HIF-1β forming the active HIF-1 complex. Activation of HIF-1α has been reported in many solid tumors including carcinomas of the gastrointestinal tract [4‐12]. However, the role of HIF-1α in tumor progression of colorectal carcinomas is still unclear and the published data so far are controversial. Thus, the expression of HIF-1α in some studies was correlated with an increased tumor aggressiveness, whereas in other studies a direct contribution to tumor progression was not found.

HIF-1α is one of the key factors promoting carcinogenesis independently of histogenetical origin. An enhanced expression of HIF-1α from normal tissue through premalignant lesions to carcinomas has already been observed in prostate [13], gastric [14], breast [15], oral cavity [16], cervical [17] and endometrial [18] carcinogenesis. So far, detailed studies about the role of HIF-1α in the development of colorectal cancer do not exist. In the case of colorectal carcinogenesis two major pathways with different morphological features of the precursor lesions has been described [19, 20]. One is the convensional "adenoma-carcinoma pathway" and the other is the alternative "serrated pathway", in which serrated polyps replace the traditional adenoma as the precursor lesion. The incidence of colorectal cancer did not differ significantly between serrated and traditional adenoma.

In addition to hypoxia, more recent evidence suggests that non-hypoxic pro-inflammatory stimuli, including cytokines and growth factors, can also activate HIF-1α under normoxic conditions and modulate the transcription of hypoxia-associated genes [21]. This phenomenon is well documented in inflammatory cells such as macrophages and monocytes [22‐24] but similar observations have also been made in tumor cell lines [25, 26]. In this context, the very potent pro-inflammatory bacterial lipopolysaccharide (LPS) may be of particular importance as a possible stimulator of HIF-1α in the gut under normoxic conditions. Because LPS is a component of the cell wall of gram-negative bacteria ubiquitously present in the colon, normal and neoplastic intestinal cells are continuously exposed to this factor [27]. Our previous studies suggest that LPS could influence the progression of colorectal cancer through its receptor TLR4 (Toll-like receptor 4) [28] and the downstream transcription factor NFκB (nuclear factor-κB) [29], leading to release of factors from colon carcinoma cells capable of up-regulating endothelial cell adhesion molecules [30]. Interestingly, a study investigating gut ischemia, has found that LPS can induce HIF-1α expression in enterocytes under normoxic conditions [31]. However, a potential activation of HIF-1α in colorectal carcinoma cells after exposure to LPS has not been reported.

On the basis of the above considerations, in the present study we set out to investigate the expression and role of HIF-1α in colorectal cancer development and progression. To determine the potential role of HIF-1α in colorectal carcinogenesis we evaluated its expression in classical and serrated premalignant lesions and in normal mucosa. To examine whether HIF-1α can influence the progressive behavior of colorectal carcinomas, we compared its expression in non-metastatic and metastatic malignant cases. Additionally, we verified its stimulation in colorectal carcinomas under normoxic inflammatory conditions by determining its inducibility after LPS stimulation in various colorectal carcinoma cell lines.

The present study represents a systematic morphological investigation of HIF-1α in colorectal cancer development. So far, only one study reported detection of HIF-1α m-RNA in four of nine investigated adenomas [7]. Our analysis demonstrated a significant induction of HIF-1α in half of the cases of the premalignant lesions SSA and TA-HGD. In contast, in normal mucosa, in benign HP and in TA-LGD, which is known to have only a low risk of malignant transformation HIF-1α expression was absent. Since the investigated preneoplastic lesions did not show any necrosis that expression is independent of hypoxia. Due to the fact that in benign tissue HIF-1α is not detected, the expression in tumor tissue could arise from genetic alterations. In fact, HIF-1α is involved in tumor progression by promoting genomic instability [32]. In the most common traditional pathway of colorectal carcinogenesis the chromosomal instability by stepwise mutation of Ki-ras, APC, and p53 genes generally leads to development of approximately 70%–85% of colorectal carcinomas [33]. Whether an interactive correlation exists between Ki-ras or APC and HIF-1α activity status still remains to be determined. However, induction of HIF-1 activity by loss-of-function mutations of the tumor suppressor gene p53 has already been observed [34]. Sugai et al reported about progressive increase in mutations of the p53 gene in colorectal adenomas from low-grade to high-grade [35]. In this context, we suggest that this observation could be associated with our observations of HIF-1α upregulation in the same stage of colorectal carcinogenesis. The serrated pathway shown in about 15% of the colorectal carcinomas demonstrates BRAF mutation and methylation/silence of the MLH-1 mismatch repair complex in the context of microsatellite instability [34]. Interestingly, BRAF mutations are rare in HP (2%) but very frequent in SSA (75%) [36‐38]. Notably, a recent study with melanoma cells has shown for the first time that BRAF mutation increases HIF-1α expression [39]. MLH-1 methylation has been detected in 36% of HP and in 70% of SSA [40, 41]. MLH-1 gene was downregulated in hypoxia but a direct involvement of HIF-1α has not been found [42]. We conclude that overexpression of HIF-1α can occur early in colorectal carcinogenesis in both the traditional (adenoma-carcinoma seqence) and serrated pathway. We suppose that this HIF-1α activation is associated with the stepwise progressive genetic instability during colorectal cancer development. Further investigations are necessary to elucidate the precise mechanisms of this interaction.

In a previous in vitro study with the human colon carcinoma cell line HCT-116 it was demonstrated that HIF-1α regulates the expression of genes encoding proteins involved in the pathophysiology of tumor invasion [43]. In our present in situ study, in comparison with the adenomas in colorectal adenocarcinomas a conspicuous increase in density and intensity of HIF-1α expression was found, reflecting a gradual increase in expression level throughout the sequence of intestinal carcinogenesis with a maximum in the stage of the invasive phenotype. The distinctly elevated HIF-1α expression in malignant tissue allows the assumption of a relevant contribution of the characteristic hypoxic microenvironment. In all investigated cases a positive HIF-1α expression was seen with cytosolic and nuclear localization, consistent with synthesis in the cytoplasm and subsequent translocation to the nuclei. Although the cytoplasmic staining distribution was homogenous, nuclear staining distribution was not uniform and showed an accumulated presence in the vicinity of tumor necrosis and the neighborhood of severe peritumoral inflammation. This topological pattern of the transcriptionally active form provides evidence that HIF-1α could play a crucial role in conditions characteristic of tumoral microenvironment, such as necrosis and inflammation, giving low levels of oxygen and high levels of inflammatory cytokines. Interestingly, in the perinecrotic regions nuclear positivity was seen in cohesive tumor epithelia of the atypical glands surrounding areas of necrosis but also in dissociated cells within necrosis. According to our view, these histopathological findings underline the dual, apparently paradoxical role of HIF-1α in regulating both pro-survival and pro-death processes [44]. HIF-1α activation in the cohesive tumor cells could be explained as transduction to an adequate adaptative intracellular response to prevent from cell death. Simultaneously, HIF-1α activation in the dissociated tumor cells could be explained as initiation of a "one way" cascade leading to cell death.

Our study adds new data to support the concept of HIF-1α involvement in normoxic inflammatory processes. HIF-1α activation in colorectal cancer specimens was seen in the vicinity of severe inflammation located in the superficial tumor ulceration and along the invasion pathway. Due to the fact that intestinal mucosal injury may permit translocation of bacteria and endotoxin, this topological accumulation of the active form of HIF-1α could be of great importance for tumor progression [45]. Strong direct evidence suggests that cancer-associated inflammation promotes tumor growth and progression [46]. In the past, we identified LPS, an endotoxin of ubiquitously existing colonic bacteria, as a pivotal pro-inflammatory stimulus increasing the aggressive behavior of colorectal carcinomas [28‐30]. We report here that LPS stimulation of diverse colorectal carcinoma cell lines induced HIF-1α expression and translocation to the nucleus. These findings demonstrate that LPS is a non-hypoxic stimulus for HIF-1α in colorectal carcinomas. Based on our results we hypothesize that HIF-1α could function as a linking transcription factor for LPS-mediated processes along the tumor invasion from the luminal site to the site of deepest penetration.

In an in vitro study we recently reported that HIF-1α upregulation in response to hypoxia was accompanied by a significantly increased production of vascular endothelial growth factor (VEGF) in two of five colorectal carcinoma cell lines [47]. Kuwai et al reported an hypoxia-induced increase of HIF-1α and VEGF in three of the four examined colorectal carcinoma cell lines [6]. These researchers and others described a strong association between HIF-1α and VEGF expression in colorectal carcinoma specimens [6, 7, 10]. Since VEGF promotes lymphangiogenesis and angiogenesis in the primary tumor, thereby paving the way for metastasis, these findings suggest that HIF-1α plays a determinative direct role in the processs of colorectal cancer metastasis. However, in our current in situ investigations there were no significant differences in the HIF-1α expression levels between non-metastatic tumors and carcinomas with lymph node and distant metastasis. The published data so far concerning a correlation between HIF-1α expression and metastatic disease status of colorectal carcinomas are controversial. In accordance with our results Joshimura et al [8] and Wincewicz et al [12] found that HIF-1α expression did not differ in relation to metastatic stage. In contrast, in two other studies a significantly higher expression of HIF-1α was observed in Dukes stages C or D, characterized by a positive status for nodal or distant metastases [7, 11]. Lu et al [10] showed that HIF-1α expression was strongly associated with lymph node metastasis, whereas Kuwai et al [6] report a significant correlation with liver but not with lymph node metastasis. Because of the chosen design in this study with comparison of non-metastatic and metastatic colorectal carcinomas with the same grading of differentiation and infiltrating depth, the influence of both essential tumor features on the HIF-1α expression can be excluded. In previous in vitro experiments we demonstrated the relevance of tumor differentiation for HIF-1α inducibility [47]. Well-differentiated colorectal carcinoma cell lines were hypoxia-resistent, showing unchanged levels of HIF-1α and VEGF in response to reduced oxygen stress conditions.

We conclude that HIF-1α expression occurs in early stages in the classical and serrated pathway of colorectal carcinogenesis and achieves a maximum in the invasive stage, independent of the metastatic status of the primary tumor (Figure 7). Since the acquisition of metastatic phenotype is a multistep event, the HIF-1α associated perinecrotic processes of both tumor cell survival and death and LPS-mediated inflammatory response could indirectly contribute to this endpoint.