Abstract
Keywords
Although there is suggestive evidence from epidemiological studies that the incidence of human colorectal carcinoma is inversely related to calcium consumption (see, e.g., Garland et al. 1985), the mechanism by which inadequate dietary calcium levels could trigger neoplastic transformation and/or sustain tumor cell growth in the large intestinal epithelium has remained an enigma. It has been suggested that low calcium would reduce the amount of insoluble calcium salts formed from otherwise carcinogenic bile acids in the lumen of the intestine (Newmark and Lipkin 1992). However, extracellular calcium ([Ca++]o) apparently is also a direct modulator of colonocyte proliferation. In this respect we were able to show that the proliferative potential of intestinal Caco-2 cells is inversely related to [Ca++]o levels in the culture medium (Cross et al. 1991, 1992). Recently we obtained evidence that Caco-2 cells are able to express the same extracellular calcium-sensing receptor (CaR) (Kállay et al. 1997) that had been cloned from parathyroid, kidney, and other cells and was shown to belong to the family of G-protein-coupled membrane receptors (Brown et al. 1993; Garrett et al. 1995). The intestinal CaR might be responsible for the unique ability of the luminal plasma membrane of Caco-2 cells to sense variations in [Ca++]o and to transduce respective signals along the PKC pathway into c-myc-mediated stimulation of cell growth (Kállay et al. 1997).
Caco-2 cells, although originally derived from a human colon adenocarcinoma, are still able to spontaneously differentiate in culture in an enterocyte-like fashion (Pinto et al. 1983). Because they express the CaR in a mosaic pattern (Kállay et al. 1997), it is not clear whether the observed CaR positivity could be attributed to regaining of a differentiated cell function or whether it reflects a tumor-associated alteration. For this reason and to gain more insight into a possible role of the CaR in [Ca++]o-related colonocyte proliferation, we used reverse transcription polymerase chain reaction (PCR) and immunohistochemistry for identification and cellular localization of the CaR in normal and malignant human colon mucosa.
Materials and Methods
Immunohistochemical Procedures
Archival blocks of formalin-fixed, paraffin-embedded tissue from 20 patients who had undergone surgery for colon carcinoma were investigated in this study. All tumors were of the adenocarcinoma type and were graded according to WHO criteria (Jass and Sobin 1989): 11 were of low grade, i.e., well to moderately differentiated, and nine were of high grade, i.e., poorly differentiated or undifferentiated. Tumor stages were evaluated according to the TNM system: the number of tumors classified as pT1 was 1; pT2, 6; pT3, 4; and pT4, 9 (Hermanek and Sobin 1992).
For gross inspection of tumor morphology, 5-± sections were deparaffinized and stained with hematoxylin/eosin. Samples from the center and the periphery of a tumor as well as from adjacent histologically normal mucosa were chosen for subsequent immunostaining. In each case, sections from three to five blocks were used. For comparison, normal mucosa was obtained from three diverticulitis patients undergoing stoma re-operation.
For specific immunostaining, the following mouse monoclonal antibodies were used: an anti-CaR antibody directed against a synthetic peptide corresponding to amino acids 214-235 in the extracellular domain (generously provided by Drs. Allan Spiegel and Paul Goldsmith, NIDDK, NIH, Bethesda, MD, through NPS Pharmaceuticals, Salt Lake City, Utah; dilution 1:300); anti-chromogranin A (Biomeda, Foster City, CA; dilution 1:100); and anti-PCNA (DAKO, Carpinteria, CA; dilution 1:10).
For immunostaining of the CaR, an indirect horseradish peroxidase-labeled antibody technique was used. Five ± serial paraffin sections on poly-L-lysine-coated slides were incubated for 20 min at 60C, deparaffinized, and rehydrated. After washing three times in PBS, pH 7.2, endogenous peroxidase activity was blocked by incubation in H2O2 in methanol at room temperature (RT) for 10 min. Sections were stained according to the Histostain-Plus kit instructions (Zymed Laboratories; San Francisco, CA) with 3,3′-diaminobenzidine (DAB; Zymed) as substrate. Slides were counterstained with hematoxylin and then mounted with Histomount (Zymed).
Negative controls for CaR were carried out by incubation of the sections with the anti-CaR antibody preabsorbed with 6 μg/ml of the immunogenic peptide.
For double immunostaining, an indirect alkaline phosphatase-labeled antibody technique was used. After washing three times in PBS, pH 7.2 (the last washing also contained 0.1% Tween-20), endogenous alkaline phosphatase activity was blocked by incubation in 0.2 N HCl at RT for 8 min. Sections were incubated in 2% normal horse serum in PBS for 30 min at RT to reduce unspecific staining and then with the primary antibody overnight at 4C in a moist chamber. Then, biotinylated anti-mouse IgG and Vectastain ABC-AP kit (both from Vector; Burlingame, CA) were used according to the manufacturer's recommendations.
Specifically bound alkaline phosphatase was visualized by incubation with either 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (BCIP/NBT; Vector) or with the Fast Red substrate system (DAKO). The development of the color reaction was followed by light microscopy. After visualization of one antigen, the sections were intensively washed three times in PBS for 10 min on a rocker platform, with the last washing also containing 0.1% Tween-20. Then the second antigen was detected using the same protocol, except that the procedure for blocking endogenous alkaline phosphatase activity was omitted. Controls for specificity of immunostaining for chromogranin A or PCNA, respectively, were performed by substituting the primary antibody with PBS or preimmune mouse serum, and were consistently negative.
Sections were counterstained with hematoxylin or, in some cases, with methyl green (Vector) as indicated, and then mounted with Histomount (Zymed) or Glycergel (DAKO).
The percentage of epithelial cells co-expressing the CaR and chromogranin A or PCNA was assessed by counting at least 250 epithelial cells at a magnification X200 on each section of normal colon mucosa (from three non-cancer, i.e., diverticulitis, patients), and of cancerous colon mucosa and of normal mucosa adjacent to the cancerous lesion from the same patient. Sections from the tumor center and, if necessary, also from the tumor periphery were inspected for the presence of representative differentiated areas exhibiting glandular-tubular structures. Results are presented as means ± SD from four tumors, of which two were of low grade (Stage pT2) and two of high grade (Stage pT3 and pT4) (cf. Table 1).
Semiquantitative Reverse Transcription-PCR
For analysis of CaR mRNA in relation to cytokeratin-8 mRNA expression by reverse transcription-PCR, snap-frozen surgical specimens of tumor tissue and adjacent normal mucosa from 16 patients were used. Tumor staging was done according to the TNM system: pT2, n = 4; pT3, n = 6; pT4, n = 6. Five samples of normal mucosa from diverticulitis patients undergoing stoma re-operation were included in the analysis.
Co-expression of CaR with PCNA or chromogranin A in human colon adenocarcinomas and adjacent normal mucosa a
a aMeans ± SD, n = 4 (see Materials and Methods).
Total RNA was extracted with the TRIzol reagent (Life Technology; Palo Alto, CA) according to the instructions of the manufacturer. RNA was quantitated by absorbance at 260 and 280 nm. Two μg was used for synthesis of single-stranded cDNA (SUPERSCRIPT II kit; Life Technology). PCR was performed at a final concentration of 1 × DyNAzyme buffer, 0.25 mM dNTP, 0.25 ± of forward and reverse primers, and 1 U DyNAzyme II DNA polymerase (Finnzymes OY; Riihitontuntie, Finland). Conditions were 33 cycles at 94C for 15 sec, 61C for 30 sec, 72C for 1 min, with a final extension at 72C for 10 min.
Cytokeratin 8, a common marker for intestinal epithelial cells, was used as an internal control for the RT-PCR reaction. Primers were designed to span an intron to ensure that the products were not the result of amplification of contaminating genomic DNA segments. To avoid differences among tubes, RT-PCR was performed in the same tube for amplification of the CaR mRNA and cytokeratin 8.
The amount of RNA and the number of amplification cycles was optimized to ensure that PCR products were in the exponential phase of amplification. Aliquots of the PCR products were resolved by electrophoresis on a 2% agarose gel. Validation of the PCR products was done by multiple digestion with restriction enzymes and size evaluation of the fragments obtained thereby. For quantitation of transcripts, photographs of ethidium bromide stained gels were analyzed by densitometry.
Primers used for CaR were 5′-CAGCAAGAGCAACAGCGAAG-3′, sense; 5′-GAAACCTCTCTGCATTCTCC-3′, antisense; for cytokeratin 8: 5′-AGTGGGCAGCAGCAACTTTCG-3′, sense; 5′-TTCAGCTTCTCCTGGCCCAGAG-3′, antisense.
Results
CaR mRNA Expression in Normal and Malignant Mucosa
For analysis of CaR mRNA expression in human colon adenocarcinomas, in adjacent mucosa from the same cancer patients, and in normal mucosa from non-cancer patients, semiquantitative RT-PCR was used. Primers were selected to allow amplification of a 537-bp fragment within exon 7 of the CaR gene and of a 360-bp of the cytokeratin 8 gene. The latter is uniformly expressed in normal and malignant colon epithelial cells and can therefore be used as appropriate reference to exclude any contribution from stromal and other non-epithelial cells. Figure 1A shows that after 33 cycles PCR products appeared as single bands which, with respect to size and restriction fragment length, indicated the presence of mRNA specific for CaR and CK 8. Densitometric analysis (Figure 1B) revealed no significant difference in the CaR/CK 8 ratio among normal mucosa, adjacent mucosa, and cancerous tissue. It should be noted that CaR-specific mRNA fragments could be amplified from any of the 16 cancerous tissue samples inspected, which were derived from four tumors of stage pT2, six of pT3, and six of pT4. There was no correlation between tumor stages and CaR/CK 8 mRNA expression (not shown).
Immunohistochemistry of Normal Colon Mucosa
In sections of histologically normal mucosal tissue adjacent to a colon carcinoma, a monoclonal antibody specific for the amino acid sequence 214-235 in the extracellular domain of the CaR produced positive immunoreactivity exclusively in cells interspersed in the crypt epithelium, where they appeared to be located preferentially in or next to the crypt base (Figure 2A). No specific staining was detectable when the anti-CaR antibody had been preincubated with the immunogenic peptide (Figure 2B).
On perpendicular sections counterstained with methyl green (not shown), it became obvious that cells which stained positive with the anti-CaR antibody differed in appearance from most other cells because they had round nuclei and abundant cytoplasm. In contrast, CaR-negative epithelial crypt cells exhibited more elongated nuclei and rather scanty cytoplasm. Such morphological criteria suggested that CaR-positive cells might be endocrine cells. However, because of their position at the crypt base (cf. Ho et al. 1989), it was also possible that CaR-positive cells could belong to the fraction of undifferentiated highly proliferative cells that are known to reside at that location.
RT-PCR amplification of calcium-sensing receptor (CaR) and cytokeratin (CK) 8 mRNA from normal mucosa (NM) (n = 5),adjacent mucosa (AM), and tumor (TU) tissue at different stages (pT2-pT4) from a total of 16 patients. (
To address this question, sections were double-stained with an antibody against PCNA, a marker of proliferative cells, and with the anti-CaR antibody. Figure 2D shows that PCNA-positive cells, which can be easily recognized by intense nuclear staining, are almost uniformly distributed within epithelial cells at the crypt base, whereas CaR immunoreactivity is confined to cells at the very base of the crypt (Figure 2D). From a quantitative analysis of expression of PCNA and CaR, it became clear that of all PCNA-positive cells only a minute fraction co-express the CaR (Table 1). Conversely, almost one half of CaR-positive cells were devoid of PCNA (Table 1).
When sections of normal colon mucosa adjacent to the tumor were stained for CaR and chromogranin A (Figure 2C), CaR positivity was found exclusively in cells that also stained positively for the enteroendocrine cell marker chromogranin A (Table 1). However, a similarly large number of chromogranin A-positive cells showed no specific staining for the CaR (Table 1, see also Figure 2C).
Identical results were obtained when normal colon mucosa from three non-cancer patients was analyzed for co-expression of CaR with chromogranin A or PCNA (not shown).
It should be noted that, apart from the mucosal epithelium, the CaR could be detected by specific immunostaining in the neurons of the myenteric plexus but not in connective tissue, smooth muscle, or vascular endothelial cells (not shown).
Immunohistochemistry of Colon Adenocarcinoma
Immunoreactivity for the CaR was also detected in all samples (20/20) of carcinomatous lesions inspected. However, the pattern of immunostaining in each tumor was rather heterogeneous because the great majority of CaR-positive cells were confined to more differentiated areas exhibiting glandular-tubular structures, as illustrated in Figure 2E, whereas poorly or undifferentiated regions were either devoid of specific immunoreactivity (15/20 tumors) or contained only solitary CaR-positive cells (5/20 tumors) (not shown). Figure 2F shows that, in a pT2 carcinoma, distribution of CaR positive cells in glandular structures is similar to that seen in normal mucosa (cf. Figure 2A). In addition, the exclusive co-localization of the CaR to chromogranin A-positive cells, as noted in the normal adjacent mucosa (Table 1), is completely preserved in the more differentiated regions of colon carcinomas exhibiting glandular-tubular structures (Table 1).
To check the proliferative activity of CaR-positive carcinoma cells, tissue sections were double-stained with an anti-CaR and an anti-PCNA antibody. In glandular-tubular structures of cancerous lesions (see Figure 2G), the expression pattern of the CaR and PCNA was similar to that observed in colon mucosa outside the tumor border (Table 1). Whereas a large portion of tumor cells exhibited immunostaining of PCNA but not of the CaR, the majority of CaR-positive cells also stained positively for the proliferation marker.
Discussion
The CaR, which is likely to transduce changes in [Ca++]o into specific cell functions in parathyroid gland, kidney, thyroid C-cells, and neuronal cells in various regions of the brain, is apparently also expressed by gastrointestinal epithelial cells. Apart from human colon adenocarcinoma-derived cell lines such as Caco-2 (Kállay et al. 1997) or T84 and HT-29 (Gama et al. 1997), the presence of the CaR has been shown in normal rat small and large intestine (Gama et al. 1997; Chattopadhyay et al. 1998). In this study we were able to show that human colon mucosal cells also express the CaR at the message and protein level (cf. Figures 1 and 2). Our immunohistochemical data reveal that in normal mucosa the CaR is present exclusively in cells at the crypt base that could be characterized as chromogranin A-positive enteroendocrine cells. This is consistent with a report from Ray et al. (1997), who were able to localize the CaR in human antral endocrine cell-enriched cultures to gastrin- but not to somatostatin-producing cells. It should be noted that the frequency of chromogranin A-reactive endocrine cells in the human gastrointestinal tract is highest, apart from the proximal duodenum, in the distal colon and rectum (Facer et al. 1989). Chromogranin A reactivity is displayed by various subtypes of enteroendocrine cells with the exception of somatostatin-, gastric inhibitory peptide-, or motilin-positive cells (Buffa et al. 1988). The occurrence of the CaR in enteroendocrine cells and also in the neurons of the myenteric plexus (not shown) is not unexpected in view of the fact that expression of the CaR is intimately related to specialized functions of other cell types in the neuroendocrine system, such as parathyroid gland and thyroid parafollicular cells (Brown et al. 1993; Freichel et al. 1996).
Immunocytochemical detection of the CaR on sections from human colon mucosa. (
The notion that the CaR is present in enteroendocrine cells of human colon mucosa is substantiated by the observation of Chattopadhyay et al. (1998) that in the rat colon the CaR is also mainly found in cells at the crypt base, which is known to harbor the majority of enteroendocrine cells.
It should be noted that another product of neuroendocrine cells, i.e., parathyroid hormone-related protein which, similar to the CaR, is rarely expressed in normal human colon mucosa, was detected by specific immunostaining in more than 90% of colon carcinomas (Malakouti et al. 1996).
A number of human colon adenocarcinoma-derived cell lines, e.g., Caco-2 and HT-29, display the CaR at the mRNA and protein level (Gama et al. 1997; Kállay et al. 1997). Studies on the effect of CaR agonists in HT-29 cells suggest that human colon cancer cells contain the CaR in a functional state (Gama et al. 1997). Moreover, by showing that the well-known CaR agonist Gd3+ elicits a dose-dependent anti-mitogenic effect in human colon adenocarcinoma-derived Caco-2 cells (Kállay et al. in press), we were able to link activation of the CaR to the growth-inhibitory effect of extracellular Ca2+ on human colon cancer cell lines (cf. Kállay et al. 1997). However, because the majority of colonocytes, either in normal mucosa or in cancerous lesions, do not show any immunoreactivity for the CaR, this of course raises the question of how realistic it is to assume that also in the human large intestinal variations in luminal Ca2+ concentrations can be transduced by the CaR into appropriate modulation of colon epithelial cell growth. This could, however, be the case if one assumes that the CaR is part of a paracrine mechanism of growth control. Thus, activation of the CaR by extracellular Ca2+ could either inhibit the release of a mitogen or growth factor such as EGF or TGF-β, as already suggested by Whitfield (1991), or even stimulate secretion of chromogranin A, whose proteolytic cleavage product pancreastatin (Cohn et al. 1990) not only plays a role in glucose metabolism but has profound suppressive effects on growth of gastrointestinal cancer cells (Fisher et al. 1998).
Footnotes
Acknowledgements
Acknowledgments
Supported by a grant from ASTA Medica GmbH (Vienna, Austria) and by grant no. 6152 from the Jubiläumsstiftung der Österreichischen Nationalbank (Vienna, Austria).