Purpose: Mucinous cancers and signet ring carcinomas are distinct classes of colon cancers characterized by their production of copious quantities of intestinal goblet cell mucin, MUC2. Deletion of transcription factor HATH1 ablates the biogenesis of goblet cells in developing mouse intestine, and forced expression of HATH1 results in elevated expression of MUC2 in colon cancer cells. The aim of this study was to assess the possible role of HATH1 in the development of mucinous cancers and signet ring carcinomas.
Experimental Design: Immunohistochemistry and confocal microscopy was used to examine HATH1 expression and subcellular distribution in normal colon and small intestine, mucinous cancers, signet ring carcinomas, and nonmucinous cancers and in precursor lesions, including hyperplastic polyps, serrated adenomas, tubular adenomas, and villous adenomas. We also analyzed the transactivation of MUC2 promoter/reporter constructs by a HATH1 expression vector.
Results: HATH1 expression transactivated MUC2 promoter/reporter constructs, an activity that was significantly inhibited by mutation of putative HATH1-binding sites. HATH1 was expressed in the nuclei of goblet cells and in the cytoplasm and nuclei of enteroendocrine cells of the colon. In the small intestine, only cytoplasmic expression of HATH1 in enteroendocrine cells was detected. HATH1 was found to be strongly expressed in the nuclei of hyperplastic polyps, serrated adenomas, villous adenomas, mucinous cancers, and signet ring carcinomas but repressed in nonmucinous cancers and tubular adenomas.
Conclusions: This study confirms the importance of HATH1 for the development of intestinal secretory cells. The results further suggest that HATH1 is an important factor in the up-regulation of MUC2 expression that occurs in mucinous cancers and signet ring carcinomas. In addition, the expression of HATH1 in hyperplastic polyps, serrated adenomas, and villous adenomas lends support to the hypothesis that these neoplasms are frequent precursors in mucinous cancer and signet ring carcinoma development.
- MUC2 mucin
- hyperplastic polyps
- transcriptional regulation
- nuclear localization
HATH1 and MATH1 are basic helix-loop-helix transcription factors that are the human and mouse homologues of Drosophila atonal, respectively. MATH1 was initially identified as a transcriptional activator of the developing mouse nervous system that was also expressed in nonneuronal tissues of the adult gastrointestinal tract (1). Subsequent analysis has revealed the requirement of MATH1 for the development of several cell types, including granule neurons of the cerebellum, inner ear hair cells, and secretory cells (Paneth, goblet, and enteroendocrine cells) of the intestinal epithelium (2–6). Thus, MATH1 plays critical roles in regulating transcription and differentiation in diverse cell types.
Colorectal cancers are derived from the cells of the colonic epithelium. They develop as one of several distinct histologic subtypes. A major distinguishing characteristic of colon cancers is the degree to which they produce and secrete goblet cell mucin, MUC2. The majority of colon cancers are nonmucinous cancers; however, mucinous cancers and signet ring carcinomas make up 10% to 20% of all colorectal cancers, with a characteristic set of clinical/pathologic features and distinct molecular genetic attributes (7–12). Mucinous cancers and signet ring carcinomas may develop via different histogenic pathways as well (13–16).
Clearly, many factors influence the pathobiological and histologic features of colon cancers. Wnt signaling is attenuated in the nonproliferating compartment of normal colonic crypts but is activated in most colon cancers through either adenomatous polyposis or β-catenin mutations (17). Wnt signaling induces a pattern of gene expression that leads to uncontrolled cell growth and the suppression of apoptosis. HATH1 expression also influences the biological properties of colon cancer cells. This factor was found to be down-regulated in most colon cancers (18). Moreover, the forced expression of HATH1 in colon cancer cells diminished the proliferative potential of the cells and led to increased MUC2 expression. Linkage between Wnt signaling and HATH1 expression has been established by the demonstration that transfection with wild-type adenomatous polyposis or dominant-negative Lef1 vectors induced HATH1 expression in cells (18).
The observation that HATH1 is required for goblet cell biogenesis and its effects on MUC2 gene expression led us to hypothesize that HATH1 plays a role in the differentiation of colon cancers leading to the development of mucinous cancers and signet ring carcinomas. To examine this, we conducted an immunohistochemical examination of HATH1 expression levels in colonic neoplasms. The results indicate elevated HATH1 expression in both mucinous cancers and signet ring carcinomas. Moreover, the expression of HATH1 in hyperptastic polyps, serrated adenomas, and villous adenomas lends further support to the hypothesis that malignant progression of these neoplasms leads with high frequency to the development of mucinous cancers and signet ring carcinomas (13–16).
Materials and Methods
Tissue samples. Formalin-fixed, paraffin-embedded tissue blocks were obtained from the Department of Pathology, University of California-San Francisco and the San Francisco Veterans Affairs Medical Center. Eight samples of normal colon, 8 small intestines, 15 hyperplastic polyps, 15 serrated adenomas, 15 tubular adenomas, 7 villous adenomas, 15 nonmucinous cancers, 15 mucinous cancers, and 7 signet ring carcinomas were used in this study. The clinicopathologic features of these tumors are given in Table 1 . We excluded polyps from patients who had familial adenomatous polyposis, hereditary nonpolyposis colorectal cancer, or hyperplastic polyposis. This study was undertaken with approval of the Institutional Review Board for Protection of Human Subjects at the University of California-San Francisco.
Histologic evaluation. Serial sections (5 μm thick) were used for H&E staining and immunostaining. All H&E-stained sections were evaluated by two authors (E.T.P. and S.K.). Polyps with dysplasia and absence of serrated architecture were classified as tubular adenomas. Polyps with serrated architecture were classified as hyperplastic polyps. Polyps with dysplasia and serrated architecture were classified as serrated adenomas based on the definition of Longacre and Fenoglio-Preiser (19). Colorectal cancers were defined as mucinous if they contained mucin lakes representing >50% of the lesion.
Immunohistochemistry. Slides were deparaffinized with xylene and rehydrated using a graded series of ethanol. The sections were treated with 3% hydrogen peroxide in methanol and blocked with 10% nonimmune goat antiserum. After antigen retrieval in a pressure cooker (Biocare Medical, Walnut Creek, CA) for 15 minutes in 10 mmol/L sodium citrate (pH 6.0), the tissue sections were incubated sequentially with primary antibody (HATH1; 1:150; Chemicon, Temecula, CA), biotinylated secondary antibody (Zymed, South San Francisco, CA), streptavidin-peroxidase conjugate (Zymed), and 3,3′-diaminobenzidine substrate (Zymed). The sections were mounted with Permount following light counterstaining with Mayer's hematoxylin (Zymed). MUC2 immunohistochemistry was done as described (16).
Confocal microscopic analysis. For double-label immunofluorescence, formalin-fixed, paraffin-embedded sections were hydrated, treated for 10 minutes in 3% hydrogen peroxide in methanol, and subjected to antigen unmasking as described above. Following three rinses in PBS, the sections were treated for 1 hour in Histostain-Plus blocking solution (Zymed). Sections were incubated 1 hour at room temperature with rabbit anti-HATH1 diluted 1:150 in 1% bovine serum albumin in TBS with 0.1% Tween 20. Following PBS rinsing, the sections were treated with the Histostain-Plus biotinylated secondary antibody solution (Zymed). After rinsing, the sections were treated with a 1:500 dilution of streptavidin-Alexa 488 (Molecular Probes, Eugene, OR) conjugate for 30 minutes. Sections were again blocked and incubated with 1:1,000 rabbit anti–chromogranin A (CgA; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour at room temperature. Goat anti-rabbit Alexa 594 conjugate (Molecular Probes) was used at 1:500 dilution to detect the anti-CgA. In some experiments, a 1:1,000 dilution of propidium iodide (Sigma-Aldrich, Philadelphia, PA) for 30 minutes was used to visualize nuclei. The slides were mounted with Prolong Antifade mounting medium (Molecular Probes). Slides were examined with a Leica TCS-SP laser scanning confocal microscope (Leica, Heidelberg, Germany). All images were collected using a pinhole of 1 Airy. The PMT gain and offset were adjusted to collect the images below the level of saturation and with the maximum dynamic gray scale. Confocal images were further analyzed using Openlab software (Improvision Ltd., Coventry, United Kingdom).
Analysis of immunohistochemical data. Immunohistochemical results were evaluated by two investigators (E.T.P. and S.K.) independently. The results were evaluated for both percentage and intensity of stained cells. An intensity value of 0, 1, 2, and 3 correlated to no, weak, moderate, and strong staining of the cells, respectively. When <10% of positive cells stained and/or the intensity was 0 or 1, immunostaining was considered to be negative. Normal colonic and small intestinal tissues were used as positive controls.
Transient transfection and luciferase reporter assays. Promoter/reporter assays employed various regions of the MUC2 gene 5′-flanking sequence cloned into the pGL2 vector (Promega, Madison, WI; refs. 20, 21). The HATH1 expression vector described previously (18) was used in conjunction with its pcDNA3.1 empty vector control. LS174T cells at 50% to 80% confluency in 24-well plates were transfected with 1.5 μg MUC2 promoter/reporter plasmid, 0.2 μg pRLO (Renilla luciferase internal control; Promega), and 1.0 μg HATH1 expression or control plasmid with 5 μL SuperFect (Qiagen, Valencia, CA). LS174T cells were used as they express considerably less HATH1 than normal colon (18) but transfect well. Luciferase activities were measured 1 day after transfections using the Dual-Luciferase Reporter Assay System (Promega) using a Monolight 2010 Luminometer (Analytical Luminescent Laboratory, San Diego, CA). Firefly luciferase activity measurements were normalized with respect to pRLO Renilla luciferase activity to correct for variations in transfection efficiency.
Statistical analysis. Immunohistochemical data were analyzed using the χ2 test and Fisher's exact test using software from StataCorp (College Station, TX). Luciferase reporter assay data are presented graphically as the average of triplicates with statistical significance determined by two-tailed, paired Student's t test.
Transactivation of MUC2 promoter/reporter constructs by HATH1. A previous study has shown that forced expression of HATH1 in colon cancer cells results in the up-regulation of MUC2 expression (18). This, and the observation that mouse embryos lacking HATH1 also lack intestinal secretory cells, including goblet cells (4), suggests that this protein may up-regulate the transcription of intestinal goblet cell mucin, MUC2. To examine this hypothesis, we tested the ability of a HATH1 expression vector to transactivate MUC2 promoter/luciferase reporter constructs (Fig. 1A ). As shown in Fig. 1B, HATH1 expression was found to transactivate MUC2 promoter/reporter constructs of three different lengths compared with activity when the pcDNA3.1 control vector was used with the same MUC2 construct. As noted previously (20), in the absence of HATH1 expression vector, the largest MUC2 construct (2,864 bp) was considerably more active than the smallest (343 bp). This suggests that elements important for MUC2 transcription are located in upstream regions. Importantly however, the smallest construct (343 bp) was transactivated to the greatest degree. A feature of this construct is that it contains a cluster of three proximal E-box elements, potential sites for HATH1 action (1). To test the possibility that these sites are involved in mediating the effects of HATH1, we mutated them in the context of the 1,628-bp MUC2 promoter and did transactivation assays. As shown in Fig. 1C, the constructs containing mutated E-boxes were less efficiently transactivated by HATH1 expression. These results support the hypothesis that HATH1 facilitates mucin production by activating MUC2 transcription and further that these effects are mediated at least in part through a process involving the proximal E-box cluster.
Immunohistochemical and immunofluorescence analysis of HATH1 expression in the normal human intestine. As a prelude to examining HATH1 expression in pathologic specimens, we used immunohistochemistry to determine expression of this factor in the normal human intestine. The rabbit anti-HATH1 antibody used in these experiments has been used previously for both immunohistochemistry and immunofluorescence (18). Immunostaining of normal human colon (n = 8) revealed two populations of cells detected using the HATH1 antibody. One population (Fig. 2A-C, short arrows ) was heavily stained in its basal, perinuclear region. Due to the close packing of cells in the colon, it was not possible to determine the cell type stained in this manner. These cells were relatively rare, representing only 2.1 ± 0.8% of colonic epithelial cells. Compartmental analysis revealed that 5.8 ± 2.8% of cells in the lower one fifth of the crypts were stained in this manner compared with 0.2 ± 0.5% of cells in the upper one fifth of the crypts. The second population of cells exhibited a fainter, predominately nuclear staining (Fig. 2A-C, longer arrows). In some cases, the nuclei stained could be seen to be those of goblet cells. We also examined HATH1 immunostaining of cells in the small intestine (n = 8), where the cells are not as densely packed as in the colon. Again, a population of intensely stained cells was detected (Fig. 2D-G, short arrows). Here, in favorable cases, where the cells were laterally well separated, intense staining could be seen to be located in the basal portion of granule-containing cells. These cells had apically displaced, rounded nuclei characteristic of enteroendocrine cells (Fig. 2F and G). The faint, predominately nuclear staining of goblet cells observed in the colon was not observed in the small intestine, indicating that HATH1 levels in small intestinal goblet cells were below the limits of detectability.
To further examine HATH1 expression in the normal colon and small intestine, we did double-label immunofluorescence staining using anti-HATH1 and anti-CgA, a marker of enteroendocrine cells. In these experiments, HATH1-expressing cells were imaged with green fluorescent dye (Alexa 488) and CgA-expressing cells were visualized with red fluorescent dye (Alexa 594; Fig. 3 ). In the colon, intense staining with anti-HATH1 was apparent in the cytoplasm of select cells with fainter staining detected in the nuclei (Fig. 3A). CgA staining in the cytoplasm of these cells was detected with the red fluorescent dye (Fig. 3B). The merged image shows overlapping of HATH1 and CgA staining in the cytoplasm but HATH1 staining alone in the nuclei of these cells (Fig. 3C). In the small intestine, HATH1 staining was detected in the cytoplasm of select cells where it colocalized with CgA (Fig. 3D-F). Nuclear immunofluorescence of HATH1 was not observed in these cells or in colonic goblet cells, perhaps because these cells had HATH1 levels below the limits of detection of this method. Taken together, these immunohistochemistry and immunofluorescence studies indicate that enteroendocrine cells of the normal adult intestine express high levels of HATH1, with lower levels being expressed in goblet cells.
Immunohistochemical staining of colorectal neoplasms for HATH1. We examined the expression of HATH1 in cancers and polyps of various histologic grades first using immunohistochemistry. In hyperplastic polyps, strong nuclear staining of HATH1 was observed in 15 of 15 samples examined (Fig. 4B ; Table 2 ). Strongly stained cells were found clustered near the bases of crypts in these lesions, 62 ± 8% of cells in the lower one fifth of the crypts were stained compared with only 4 ± 1% of cells in the upper one fifth of the crypts. Strong staining was also observed in 13 of 15 serrated adenomas and 5 of 7 villous adenomas (Fig. 4C and D; Table 2). Stained cells were dispersed more evenly throughout these lesions as opposed to being clustered near the bases of crypts. The incidence of HATH1 staining was much lower in tubular adenomas (1 of 15) and nonmucinous cancers (1 of 15; Fig. 4E and F; Table 2). When staining was observed in these lesions, it appeared as weak cytoplasmic staining. In contrast, all 15 mucinous cancer specimens examined stained strongly positive for HATH1 expression (Fig. 4G; Table 2). Cells stained included both the epithelial cells lining the mucin lakes and the cells suspended in the lakes. All 7 signet ring carcinomas examined also expressed HATH1 (Fig. 4H; Table 2). In mucinous cancers and signet ring carcinomas, both nuclear and cytoplasmic staining was observed. We also observed that the stromal cells in hyperplastic polyps, serrated adenomas, villous adenomas, and tubular adenomas were heavily immunostained for HATH1, whereas the stromal cells of the nonmucinous cancers, mucinous cancers, and signet ring carcinomas stained only faintly. This probably reflects different populations of stromal cells in the various lesions; for example, the cancer specimens may have increased fibrocytic infiltration. Another possibility is the presence of different populations of leukocytes in cancers versus adenomas. The isotype control using nonimmune IgG exhibited no staining.
Thus, a significantly higher frequency of HATH1 expression was noted in hyperplastic polyps (100%; P < 0.001), serrated adenomas (87%; P < 0.001), and villous adenomas (71%; P < 0.01) compared with tubular adenomas (7%). Moreover, HATH1 was expressed in all mucinous cancers and signet ring carcinomas examined but only infrequently expressed in nonmucinous cancers (7%; P < 0.001; Table 2). Excellent concordance between HATH1 and MUC2 expression was noted in individual specimens of hyperplastic polyps, serrated adenomas, villous adenomas, mucinous cancers, and signet ring carcinomas as well (Table 3 ).
Subcellular distribution of HATH1 expression. Confocal microscopy was used to examine the subcellular distribution of HATH1 in normal colon and colon cancers. In these experiments, the nuclei were stained with propidium iodide to facilitate analysis. Similar to results shown in Fig. 3, occasional enteroendocrine cells exhibiting intense cytoplasmic expression of HATH1 with fainter nuclear expression were observed in normal colon (Fig. 5A-C ). Other nuclei were not stained using this method, indicating that HATH1 levels were below the limits of detection. In contrast, mucinous cancers and signet ring carcinomas exhibited abundant HATH1 expression in both nuclei and cytoplasm (Fig. 5D-I). Conversely, nonmucinous cancers expressed little, if any, HATH1 (Fig. 5J-L). These results confirm the expression of HATH1 in both nuclei and cytoplasm of mucinous cancers and signet ring carcinomas observed by immunohistochemistry as well as its absence or minimal expression in nonmucinous cancers.
Previous studies have shown that forced HATH1 expression in colon cancer cells results in increased MUC2 expression (18). Moreover, HATH1 rodent homologue MATH1 is required for development of secretory cells, including goblet cells, in the embryonic mouse intestine (4). This led us to hypothesize that HATH1 is involved in regulating MUC2 gene expression in colorectal cancers and possibly in regulating other biological properties of mucinous cancers and signet ring carcinomas, cancers that often exhibit a different set of clinicopathologic feature alterations than do nonmucinous cancers (7–12).
To examine these possibilities, we first used promoter/reporter constructs to test the effects of HATH1 on MUC2 gene transcription. We found that HATH1 expression vector markedly up-regulated the activity generated by three different MUC2 promoter segments (Fig. 1B). The effects were most pronounced on the shortest construct, which initiated at base −343 of the MUC2 promoter. This construct contains three proximal E-box motifs that have been identified as binding sites for HATH1 (1). Mutation of these sites diminished the stimulatory effect of HATH1, suggesting that HATH1 may activate MUC2 transcription via direct interaction with the MUC2 promoter (Fig. 1C). Other mechanisms are not ruled out by these experiments; however, several downstream effectors of HATH1 have been identified (18, 22, 23) and such factors may regulate MUC2 transcription. Additional experiments are required to clarify this point.
Immunohistochemical analysis of HATH1 expression in the normal colon revealed two populations of stained cells. One population had high-level HATH1 expression localized basal to the nuclei (Fig. 2A-C). Confocal microscopy revealed these cells to be enteroendocrine cells and further localized HATH1 expression to both the cytoplasm and the nuclei (Figs. 3A-C and 5A-C). The second population of cells exhibited a lower level of HATH1 expression that was predominately located in the nucleus (Fig. 2A-C). These cells could be identified as goblet cells in favorable instances in which nuclei could be seen to abut mucus granules. These data are consistent with the previous observation that MATH1 is required for the biogenesis of these cell types in developing mouse embryos (4). The expression of HATH1 in these cell types in the adult human colon suggests that this factor is involved in the expression of genes specific to these cell types. Along these lines, it may be important that enteroendocrine cells express higher levels of HATH1 than do goblet cells. Perhaps high levels of HATH1 expression are required for gene expression in enteroendocrine cells, whereas lower levels are sufficient in goblet cells. Alternatively, high-level HATH1 expression may be required for cell fate decisions that lead to the enteroendocrine cell lineage.
Previous work has shown HATH1 gene repression in most colorectal cancers (18). Our study confirms this finding, as we found HATH1 expression, as detected by immunohistochemistry, in only 1 of 15 nonmucinous cancers (Fig. 4; Table 2). The present study does, however, detect elevated HATH1 expression in 15 of 15 specimens of mucinous cancers examined. HATH1 was also expressed in signet ring carcinomas, which are characterized by cells with large stores of MUC2 retained within their cytoplasm. Thus, these results confirm and extend the previous observation that the single colorectal cancer specimen that expressed HATH1 of 12 examined was a mucinous cancer (18). These results are also consistent with in vitro experimentation, indicating up-regulation of MUC2 gene expression by HATH1.
Also clear from this study is the expression of HATH1 in hyperplastic polyps, serrated adenomas, and villous adenomas but not tubular adenomas (Fig. 4). This pattern of expression correlates well with the expression of MUC2 detected in these lesions (Tables 2 and 3). This provides further evidence for a causal link between HATH1 expression and MUC2 induction. Moreover, the coexpression of HATH1 and MUC2 in these lesions and mucinous cancers provides added support for the hypothesis that mucinous cancers arise in significant proportion from the hyperplastic polyp-serrated adenoma-mucinous cancer or villous adenoma-mucinous cancer pathway (13–16). Mucinous cancers as a group contain a stochastic pattern of genetic and epigenetic alterations that differ significantly from the pattern found in nonmucinous cancers (8–10, 24, 25). The alterations found in individual cancers are varied, however, and there is considerable overlap in the patterns found in mucinous cancers and nonmucinous cancers. For example, mucinous cancers as a group have higher levels of BRAF mutations, CpG island methylator phenotype, and microsatellite instability than do nonmucinous cancers, but individual mucinous cancers can be found that mimic nonmucinous cancers in terms of these variables (25). Thus, although disproportionate numbers of mucinous cancers may arise from the hyperplastic polyp-serrated adenoma-mucinous cancer or villous adenoma-mucinous cancer pathway, it seems likely that some mucinous cancers share oncogenic pathways with nonmucinous cancers. Additional factors that lead to mucinous cancer development in multiple genetic and epigenetic settings remain unclear and in need of further analysis.
As noted above, abrogation of Wnt signaling was found to increase HATH1 and MUC2 expression (18). In a previous study, we determined that mucinous cancer retained adenomatous polyposis gene expression with higher frequency than nonmucinous cancers (16), suggesting possible differences in the Wnt pathways of the two cancer types. These differences in Wnt signaling may play a role in HATH1 and therefore MUC2 up-regulation in mucinous cancer. Mucinous cancers as a group have been shown to have slower proliferation rates and higher apoptotic indices than nonmucinous cancers (24). Either diminished Wnt signaling and/or HATH1 expression may account for this difference.
Finally, the use of immunohistochemistry and immunofluorescence in this study has revealed strong expression of HATH1 in the nuclei of hyperplastic polyps, serrated adenomas, villous adenomas, mucinous cancers, and signet ring carcinomas (Figs. 4 and 5). In contrast, only weak nuclear expression of HATH1 was observed in normal colon goblet and enteroendocrine cells and enteroendocrine cells had abundant levels of cytoplasmic HATH1 (Figs. 2, 3, and 5). Along these lines, it is interesting to note that translocation to the nucleus is an important regulatory mechanism for many transcription factors (26). High levels of this factor in the nuclei of neoplasms may have important biological consequences in addition to its effects on MUC2 gene expression. HATH1 expression diminishes several variables of growth measured in colon cancer cells, including bromodeoxyuridine incorporation, growth in soft agar, and xenograft formation (18). The presence of abundant nuclear HATH1 may be at least partially responsible for a slower growth rate in mucinous cancers compared with nonmucinous cancers (7–12, 24).
We thank Dr. Rina Wu (Zymed) for aid with immunohistochemistry.
Grant support: Department of Veterans Affairs Medical Research Service and Theodora Betz Foundation Grant.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received March 8, 2006.
- Revision received June 20, 2006.
- Accepted July 11, 2006.