In the present work and by using a set of well-characterized experimental EC tumor samples, and through a candidate gene approach, we characterized the frequently reported independent region of tumor suppressor activity in human tumors located telomeric to TP53. We subjected all 19 genes located in this chromosomal segment to gene expression analysis in a panel of 28 rat EC and seven control non-malignant endometrium (NME) samples. Nine genes showed significant reduced expression in EC compared to NME samples (Fig. 1, Table 1). Interestingly, two of the known tumor suppressor genes located in this region, Ovca1 (ovarian cancer-associated gene 1) and Ovca2, were not among these.
The question was then whether the observed reduced expression of these nine genes was due to the physical deletion of the genetic material or other regulatory mechanisms. To address this, we used earlier CGH, AI, and FISH data [12, 13, 15, 16] and divided the 28 tumors analyzed in the gene expression assays into two groups: ECs with deletion/AI and those without deletion/AI spanning the 700 kb candidate chromosomal segment. We subsequently used this new grouping of tumors and re-analyzed the real-time RT-PCR results to determine whether there existed a correlation between physical deletion and the observed lowered expression of the nine genes between the tumor groups (Fig. 2, Table 1). Our interest was to identify gene(s) showing no such correlation, indicating that the observed reduction in expression of these gene(s) was not only due to the physical deletion of the genetic material. The results revealed that lowered expression levels of five genes, Hic1, Rpa1, Inpp5k, Myo1c and Crk were independent of the observed physical deletions (Fig. 2, Table 1). Among these, Hic1, Inpp5k and Myo1c showed strong and highly significant down regulation in EC compared to the control samples (Table 1) and thus were selected as the best candidates. Rpa1 and Crk were not appealing as candidates: Rpa1 showed only a rather moderate down regulation in gene expression analysis (Table 1) and Crk (v-crk sarcoma virus CT10 oncogene homolog, avian) is mainly recognized as an oncogene in cancer studies [17–19].
In qPCR experiments, we found that Hic1, Innp5k and Myo1c were always expressed in the tumors, although at very low levels. According to Knudson’s theory of inactivation of tumor suppressor genes [20], if Hic1, Inpp5k and Myo1c are to behave as classical tumor suppressor genes, it is expected that both alleles of the genes be inactivated in the tumor material. We, therefore, hypothesized that the remaining allele of these genes might have become inactivated through mutation/s. However, in gene sequencing experiments, no mutation was found in coding sequences of these three genes, suggesting a potential haploinsufficient mode of function for these candidate tumor suppressor genes. In gene sequencings we identified ten SNPs in intronic, noncoding and/or coding sequences of the Hic1 and Myo1c genes (Additional file 2: Table S5). Identification of these SNPs offered the possibility to confirm our earlier AI/deletion data [11, 13] and also to make a detailed deletion map with information on the homozygote/hemizygote pattern of the observed deletions spanning the candidate region in the tumors.
Our interest was then to determine whether gene products of these three candidate genes were down regulated in the tumors. To this end, we examined expression of Hic1, Innp5k and Myo1c proteins in a selected panel of five ECs compared to three NME samples using Western blot. The result for Myo1c showed a good correlation between the qPCR results and protein expression level in the tumor samples, indicating that Myo1c protein was in fact down regulated in the majority of EC tumors tested compared to the control samples (Fig. 3a and b). However, such a correlation was not detected for Hic1 as it was shown that expression of Hic1 protein was not down regulated in the tumors, even in those showing a rather strong Hic1 down regulation at their mRNA level (NUT12, NUT51 and NUT98; Fig. 3a and b). No result was obtained for the Inpp5k protein, most likely due to the lack of specificity of the available human antibody for the rat samples.
There were a number of tumors that did not harbor AI/deletion in the candidate region, but nevertheless displayed a significant down-regulation of Hic1, Inpp5k and/or Myo1c (Table 1). The question was then whether other regulatory machineries, namely epigenetic regulation, were involved. To address this, we screened promoter regions of Hic1, Inpp5k and Myo1c for methylation at CpG islands. No methylation was detected in the Inpp5k and Myo1c promoter regions in any of the tumors tested, whereas partial DNA methylation was detected in Hic1 promoter in 64 % of tumors tested (Fig. 3c). No correlation, however, between the observed methylation status of Hic1 promoter and the level of Hic1 transcript was detected in tumor samples (Fig. 3c).
It is quite common that genes with critical functions are under restricting regulation by several promoters. Moreover, histone deacethyaltion is an alternative epigenetic regulation mechanism that may also result in the gene silencing. To examine whether other potential promoter(s) might have been involved in the expression regulation of Hic1, Inpp5k and Myo1c as well as to study the potential involvement of histone deacethylation in expression regulation of these genes, we treated cell lines with demethylating agent 5-Aza-dC and/or deacetylating inhibitor TSA. Surprisingly, Hic1 expression was not specifically restored after the treatments in the samples, especially not in those that showed partial methylation in the Hic1 promoter region (NUT12 and NUT51, Figs. 3c, 4a, d and e). This result suggested that the observed methylation pattern in the Hic1 promoter region in EC tumors, rather than being an epigenetic silencing mechanism, is likely to be a general structural feature for this gene. In agreement with this, dense hypermethylation of one of the HIC1 alleles has earlier been reported in a number of normal human tissues, including kidney [21], and histologically normal and benign hyperplastic prostate tissues [22].
Restoration of gene expression was detected for Myo1c and Inpp5k genes in the tumor samples at RNA and/or protein levels (Fig. 4b-e). Notably, the observed gene expression restoration for Myo1c protein was irrespective of AI/deletion in the candidate region, as NUT12 with AI/deletion and NUT98 without AI/deletion showed strong restoration of Myo1c expression after either or both treatments (Fig. 4e). These data suggest that epigenetic gene silencing might have an important role in inactivation of Myo1c and Inpp5k tumor suppressor candidates.
To summarize, qPCR analysis of 19 genes located within the commonly deleted region distal to Tp53 in experimental ECs, suggested Hic1, Inpp5k and Myo1c as the best candidate tumor suppressor genes in this region. No mutation was detected in the coding sequences of the retained alleles, suggesting a potential haploinsufficient mode of function for these candidate tumor suppressor genes. Hic1, Inpp5k, and Myo1c showed reduced expression in tumor samples irrespective of the presence or absence of physical deletion in the candidate region, suggesting the potential involvement of alternative gene silencing regulatory mechanism(s). We found a rather normal expression of Hic1 protein in the tumors, even in those that showed down regulation of Hic1 at the mRNA levels, indicating lack of correlation between expression of Hic1 at mRNA and protein levels. Promoter methylation analysis revealed partial methylation of Hic1 promoter in a number of tumors. Nevertheless, there was again no correlation between the presence or absence of Hic1 promoter methylations and gene expression levels in the tumors. Moreover, Hic1 expression could not specifically be restored after treatments with demethylating and histone deacetylase inhibitor agents. In contrast, while promoter methylation was not detected in Inpp5k and Myo1c promoters, down regulation of Inpp5k and Myo1c were detected in EC tumors, and this could be rescued after the 5-Aza-dC and TSA treatments, even in tumors without AI/deletion in the candidate tumor suppressor region. Taken together, results from the present work exclude Hic1 as a fitting candidate and provide evidence for Inpp5k and Myo1c as two attractive candidates for the observed independent tumor suppressor activity at the neighborhood of Tp53.
INPP5K (inositol polyphosphate-5-phosphatase K, also known as SKIP, skeletal muscle and kidney enriced inositol phosphatase) is a member of the inositol polyphosphate 5-phosphatases family [23] with a poorly characterized function in vivo. The inositol polyphosphate 5-phosphatases family is known as negative regulators of PI 3-kinase signaling [24]. Analysis of the role of INPP5K in insulin-stimulated cells indicated that endogenous INPP5K might be one of the key regulators of insulin signaling in skeletal muscle and adipocytes for glucose homeostasis [25]. INPP5K was identified as a 5'-inositol phosphatase that hydrolyzes phosphatidylinositol 3,4,5-triphosphate (PI-3,4,5-P3) and phosphatidylinositol 4,5-bisphosphate (PI-4,5-P2) to negatively regulate intracellular phosphatidylinositol 3-kinase signaling. It is thus suggested that INPP5K exerts its functions through direct binding to PIP3 or forming a complex with molecules located downstream of PI 3-kinase. Activated PI 3-kinase generates PIP3 that in turn activates the downstream target AKT, which then positively regulates a range of cellular functions, including actin rearrangement, protein synthesis, cell metabolism, cell cycle (G1-S transition) and cell survival [24]. So far, there is no report on potential involvement of INPP5K in cancer progression. However, another member of the phosphoinositide phosphatase family, PTEN, has already been identified as a haploinsufficient tumor suppressor gene [26] and its inactivation has been implicated in a variety of human cancers, including endometrial carcinoma. Therefore, it is tempting to speculate that INPP5K might likewise be potentially involved in carcinogenesis.
The gene adjacent to Inpp5k, and the second candidate tumor suppressor gene identified in the present work, is the molecular motor myosin 1c (Myo1c). Myo1c exerts overlapping functions in phosphoinositide (PI) 3-kinase/AKT (PI3K/AKT) signaling as those of Inpp5k. Myosin 1C is a widely expressed vertebrate unconventional myosin-I isoform that concentrates in perinuclear regions, on ruffling cell membranes, and within stereocilia of hair cells. Increasing evidence points to the role of myosin 1c in many signaling cascades, from the integrin-dependent signaling involved in cell migration to the signaling events underlying insulin resistance (reviewed in [27]). Myo1c is a lipid raft-associated motor protein that is specifically involved in the recycling of lipid raft membrane and proteins that regulate plasma membrane plasticity, cell motility and pathogen entry [28, 29]. MYO1C binds tightly and specifically to PIP2 [30], an important second messenger involved in a variety of crucial cellular functions, including regulation of the actin cytoskeleton and signal transduction in insulin and AKT signaling pathways. This protein is additionally involved in glucose uptake in muscle and adipocytes through controlling of movement of intracellular GLUT4–containing vesicles to the plasma membrane [31]. It has been shown that insulin-dependent phosphorylation of Myo1c is required for GLUT4 translocation and transport of glucose through phosphoinositide (PI) 3-kinase/AKT pathway [32, 33]. There is no earlier report on potential tumor suppressor activity of this gene, but several other members of the myosin-I gene family have been reported as cancer-related genes, including tumor suppressor gene MYO18B in lung, ovarian and colorectal cancer [34] and involvement of MYO1F in chromosomal translocation and gene fusion in infant acute monocytic leukemia [35].