The human tumor suppressor CEACAM1 modulates ... - Nature

Oncogene (2004) 23, 9306–9313

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

ORIGINAL PAPERS

The human tumor suppressor CEACAM1 modulates apoptosis and is implicated in early colorectal tumorigenesis Stefanie Nittka1, Juliane Gu¨nther1, Cornelia Ebisch2, Andreas Erbersdobler2 and Michael Neumaier*,1 1 Institute for Clinical Chemistry, Medical Faculty Mannheim of the University of Heidelberg; Mannheim, Germany; 2Department of Pathology, University Hospital Hamburg-Eppendorf, Hamburg, Germany

Defects in the adenomatous polyposis coli (APC) tumor suppressor pathway are sufficient for neoplastic transformation as the initiating step in colorectal carcinogenesis. In contrast, hyperplastic tumors possess normal APC function, and it is unclear whether they represent significant precursor lesion in cancer development. CEACAM1 is a tumor suppressor whose expression is known to be lost in the great majority of early adenomas and carcinomas. We found that loss of CEACAM1 expression is more common in neoplastic tumors than APC mutations. While APC function was normal in hyperplastic aberrant cypt foci and hyperplastic polyps, loss of CEACAM1 was observed as frequently as in the neoplasias. Moreover, the presence or absence of CEACAM1 expression in the hyperplastic tumors correlates with normal or reduced apoptosis, respectively. In vitro, CEACAM1 acts as a regulator of apoptosis in CEACAM1-transfected Jurkat cells. Finally, in human HT29 colon cancer cells, apoptosis can be specifically restored by induction of CEACAM1 expression. These data suggest an oncodevelopmental link between neoplasia and hyperplasia and demonstrate that CEACAM1 acts as a regulator of apoptosis in the colonic epithelium. Thus, failure of the maturing colon cell to express CEACAM1 is likely to contribute to the development of hyperplastic lesions, which may eventually pave the way to neoplastic transformation and colon cancer development. Oncogene (2004) 23, 9306–9313. doi:10.1038/sj.onc.1208259 Published online 29 November 2004 Keywords: CEACAM1; apoptosis; ACF; APC; b-catenin; carcinogenesis

Introduction The carcinoembryonic antigen (CEA) family of cell adhesion molecules (CEACAMs) comprises a group of *Correspondence: M Neumaier, Institute for Clinical Chemistry of the University of Heidelberg, University Hospital Mannheim, TheodorKutzer-Ufer 1-3, D-68167 Mannheim, Germany; E-mail: [email protected] Received 31 August 2004; revised 29 September 2004; accepted 30 September 2004; published online 29 November 2004

highly homologous glycoproteins that are expressed in a wide variety of tissues. In the colon, CEA and the CEACAMs 1, 6 and 7 are expressed by differentiating crypt cells representing a major constituent of the glycocalix in the brush border, but their physiological function is not clear (Hammarstrom, 1999). We and others have established that the expression of CEACAM1 is downregulated in more than 85% of early colorectal adenomas and carcinomas (Neumaier et al., 1993; Nollau et al., 1997b; Zhang et al., 1997). In addition, experimental data strongly indicate a tumor suppressor function of CEACAM1 in colon tumorigenesis (Kunath et al., 1995; Turbide et al., 1997; Izzi et al., 1999; Fournes et al., 2001). Colorectal carcinomas (CRC) develop by a successive and ordered accumulation of genetic alterations termed ‘multistep carcinogenesis’ (Fearon and Vogelstein, 1990). Around 80% of the colorectal cancers are characterized by defects in the ‘adenomatous polyposis coli (APC)/WNT signaling pathway’ leading to bcatenin accumulation. APC defects are very common and remarkably constant throughout all stages of colorectal neoplasia (Powell et al., 1992; Goss and Groden, 2000). Functional studies and experimental animal models implicate APC as the molecular gatekeeper, whose inactivation initiates neoplastic growth and a chromosomal instability (CIN) phenotype eventually leading to polypous CRC. Specifically, the dysplastic/neoplastic stage is initiated either by mutations in the APC tumor suppressor gene itself, gain of function mutations of the b-catenin gene or by methylation silencing of the APC gene promoter (Kinzler and Vogelstein, 1996; Esteller et al., 2000; Goss and Groden, 2000; Fearnhead et al., 2001). Finally, the epigenetic silencing of the WNT-inhibitory SFRP genes can lead to b-catenin accumulation and has been proposed to sensitize the affected cells for the downstream genetic defects of APC or CTNNB1 (Suzuki et al., 2004). In contrast to dysplasia, defects in the APC pathway are virtually absent from the hyperplastic tumors, that is, hyperplastic aberrant crypt foci (ACF) that represent the earliest mucosa lesions, and also from hyperplastic polyps (HP) (Jen et al., 1994; Smith et al., 1994). However, dysplastic subsets of ACF have been defined

CEACAM1 in colorectal tumorigenesis S Nittka et al

9307

(Bird, 1995) showing mutations of CTNNB1 and nuclear accumulation of b-catenin (reviewed in Mori et al., 2004). Hyperplastic ACF can develop focal losses of the DNA mismatch repair complex and chromosomal aberrations (Hawkins et al., 2000; Jass et al., 2000; Hawkins and Ward, 2001), and direct histological evidence for an early carcinomatous lesion has been recently demonstrated within HP (Hawkins and Ward, 2001), suggesting that hyperplasia may be a precursor lesions. While hyperplastic growth of the colon epithelium results from reduced apoptosis of unknown cause rather than from an increase in proliferation (Roncucci et al., 2000), the mechanism of how hyperplasia may develop is still unclear. We were interested in the role of CEACAM1 within the multistep tumorigenesis and have investigated (i) colorectal tissue specimens of CRC, adenomas, hyperplastic ACF and HP to study the relationship between defects in the APC pathway and a decrease or loss of CEACAM1 expression and (ii) differences in the apoptosis in hyperplastic and neoplastic colon crypts and their correlation with changes of CEACAM1 expression. Since the results suggested that CEACAM1 expression and apoptosis were functionally linked in colon tumor development, we have investigated the capability of CEACAM1 to act as a regulator of apoptosis in two different in vitro cell culture models.

Results APC mutations and CEACAM1 expression in hyperplastic and neoplastic lesions Since virtually all mutations of the APC gene in colon carcinogenesis lead to truncated peptides, a protein truncation test (PTT) (Powell et al., 1993; Laurent-Puig et al., 1999) can be used to efficiently screen for APC mutations. Truncated APC peptides were detected in 70% (14/20) of the small adenomas with previously confirmed downregulation of CEACAM1 mRNA expression (Nollau et al., 1997b) using a nonradioactive PTT assay (Kirchgesser et al., 1998). All APC mutations were detected between codons 653 and 1700. Of these, 65% (9/14) occurred within the mutational cluster region, consistent with data reported by others (Rowan et al., 2000; Fearnhead et al., 2001). Since APC is the key regulator of cytoplasmic bcatenin degradation, intracellular b-catenin accumulation can be used as a marker for APC defects (Herter et al., 1999). Expectedly, b-catenin accumulation with or without nuclear staining was present in 75% (12/16) of the adenomas and 80% (4/5) of the carcinomas while normal b-catenin expression was found in all hyperplastic ACF and in 88% (37/42) of the HP (Figure 1). In contrast, a marked decrease or complete loss of CEACAM1 protein expression was already observed in 72% (23/32) of the ACF and 86% (36/42) of the HP. Neoplastic tumors, that is, adenomas and carcinomas, were included in this study as positive controls. A total of 88% (14/16) of the adenomas and all carcinomas

Figure 1 Percentage of colon epithelium specimens displaying decrease/increase of antigen expression of CEACAM1 (shaded bars) and b-catenin (black bars) as assessed by immunohistochemistry. ACF: hyperplastic aberrant crypt focus (n ¼ 32); HP: hyperplastic polyp (n ¼ 42); AD: colorectal adenoma (n ¼ 16); CA: colorectal carcinoma (n ¼ 5). The numbers of tumors with altered staining patterns are indicated. The absolute numbers of tumors with altered staining patterns are indicated inside the respective columns

(5/5) were negative for CEACAM1 (Figure 1), confirming previous reports on the frequency of loss of CEACAM1 expression in benign and malignant colorectal neoplasias (Neumaier et al., 1993; Nollau et al., 1997b). Figure 2 demonstrates that, in both hyperplasia and neoplasia, the loss of CEACAM1 expression starts in the transition zone between normal mucosa and the tumor lesions. Typically, while CEACAM1 expression is present on the surface of normal colon epithelium within the differentiation zone (black arrowheads), it is absent in the affected tissues (white arrowheads). Consistent with the literature, CEACAM1 expression was not detected in the proliferative crypt compartment (Hammarstrom, 1999). Assessment of apoptosis in tissue sections We investigated apoptosis in normal colon epithelium, ACF, HP, adenomas and carcinomas. In longitudinal sections of colon crypts previously immunostained with mab M30, an average of 403–951 cells per tissue type (range: 136–2917 cells/sample) were microscopically evaluated in the tumor samples (Figure 3). Consistent with reports elsewhere (Kikuchi et al., 1997), normal colon epithelium showed a rate of 0.93% (71.25%) apoptotic cells. In detail, M30-staining apoptotic cells were identified within the differentiation zone mostly close to the luminal surfaces. In contrast, in the great majority of hyperplastic lesions, a marked reduction in apoptosis was found for hyperplastic ACF (0.17%70.37; Po0.0001) and HP (0.31%70.53; Po0.0012). We found a significant correlation between Oncogene


9308

Figure 2 Alteration of antigen expression in transition zones between normal mucosa and neoplastic alteration. Immunohistochemical staining of serial tissue sections for CEACAM1 (panels b, e, h) and b-catenin (panels c, f, i) in transition zones between normal colon epithelium and epithelial lesions: (a–c) ACF; (d–f) HP; (g–i) colorectal adenoma. Panels a, d and g are enlargements of tissue sections boxed in panels b, e and h, respectively. The presence or absence of CEACAM1 antigen expression in the transition zones is indicated by filled or open arrowheads, respectively. Vertical bars (panels b, e, h) show the border between normal and tumor tissue. Panels c, f and i show basolateral b-catenin staining in normal and hyperplastic lesions (panels c, f), while increased staining with diffuse intracellular distribution is observed in neoplastic epithelium (see arrows in panel i). Size bars are 400 mm

Figure 3 Immunohistochemical detection of apoptosis in colonic tissues. Apoptosis of colonic crypt cells (%) in different tumor types was measured by mab M30. NM: normal mucosa; ACF: aberrant crypt focus; HP: hyperplastic polyp; AD: adenoma; CA: carcinoma. The median rate of apoptosis in normal mucosa is indicated as shaded area7standard error (s.e.). The apoptosis in tumors is depicted in brackets as median apoptosis7s.e. Differences of statistical significance between normal and timorous tissues are indicated by brackets. The numbers of cells evaluated per tissue and per sample are given below the figure

CEACAM1 expression and the rate of apoptosis in these hyperplastic tumors. Only 14% of the HP (n ¼ 6) investigated showed CEACAM1 expression. As shown in Figure 4, these polyps displayed a rate of apoptosis Oncogene

Figure 4 Rate of apoptosis in CEACAM1-positive HP (n ¼ 6) and CEACAM1-diminished/negative (n ¼ 36) HP is given as percentage of cells positive for mab M30. Double-sided Wilcoxon’s test was used for the evaluation of statistical significance (Po0.05)

that was significantly higher (0.36 vs 0.16%; Po0.04) than in the CEACAM1-negative HP (n ¼ 36) (Figure 4). In contrast to the hyperplastic lesions, an increased proliferation, as measured by Ki-67 staining intensity, was only noted in adenomas and carcinomas (data not shown).


9309

CEACAM1 triggers apoptosis in vitro We then investigated the potential functional basis for the histopathological correlation between loss of CEACAM1 and the loss of apoptosis previously observed in the hyperplastic tumors. As shown in Figure 5, crosslinking of CEACAM1 on the cell surface using the CEACAM1-specific mab 4D1C2 and mab T84.1, an antibody crossreacting with CEACAM1 (Nap et al., 1992), triggered apoptosis in CEACAM1-transfected Jurkat reporter cells, but not in the untransfected Jurkat cells (Po0.001). Importantly, the transfection did not change the susceptibility of the Jurkat cells for apoptotic stimuli as documented by the control CD95 antibody. Also, the CEACAM1 expression on the cell surface did not influence the base rate of apoptosis by itself. No increase in apoptosis was observed, when the antibodies were used with the untransfected Jurkat control cells. Thus, the increase in apoptosis was a specific function of CEACAM1 crosslinking and did not occur in cells treated with irrelevant antibody (Figure 5). Also, in CEACAM1-transfected Jurkat cells, we were able to assess the role of CEACAM1 as a mediator of apoptotic signaling without interference from other members of the CEACAM family. To study the apoptosis-mediating function in a more relevant in vitro system, the human colon cancer cell line HT29 was used. HT29 cells express low levels of CEACAM1 on the cell surface. Using IFNg as described by others (Takahashi et al., 1993) increases the CEACAM1 mRNA expression in HT29 cells between four- and sevenfold as measured by Light Cycler realtime RT–PCR (Roche Molecular Systems, Mannheim, Germany) (data not shown). Importantly, neither IFNg treatment nor unspecific antibodies influenced apoptosis in the HT29 cells under the conditions chosen by themselves (Figure 6). Also, the increased cell surface expression of CEACAM1 did not enhance, by itself, the rate of apoptosis in these cells. However, when

Figure 5 CEACAM1 crosslinking mediates apoptosis in cell culture models. Jurkat parental cells (shaded columns) and CEACAM1-transfected cells (black columns) were incubated with the following mabs: 7D11 (anti-b-catenin; negative control), 4D1C2 (anti-CEACAM1), T84.1 (anti-CEACAM), CH11 (antiCD95; positive control). The percentages of apoptotic cells were subsequently assessed after propidium iodide staining. Values are presented as mean7standard deviation

Figure 6 Sensitivity of HT29 cells to apoptosis by CEACAM1 crosslinking at high levels (dark columns) or low levels (light columns) of CEACAM1 gene expression. Crosslinking was performed using the following antibodies: mab 4D1C2 (antiCEACAM1), mab 7D11 (anti-b-catenin; negative control), polyclonal goat IgG (anti-mouse; secondary antibody). Apoptosis was analysed in the cell lysates as free DNA–histone complexes by ELISA, and is given after normalization to 106 cells/sample. Values are presented as mean7standard deviation

CEACAM1 was crosslinked using specific antibodies, a significant induction of apoptosis was noted in the stimulated compared to the unstimulated cells. This CEACAM1-mediated effect has to be appreciated considering the known relative apoptosis resistance of HT29 (Battu et al., 1998; Tan et al., 2002) and the fact that unstimulated HT29 cells programmed cell death could not be provoked in the CEACAM1 low expressors (Figure 6). These results clearly demonstrate in two independent and different models that CEACAM1 possesses a specific function to modulate apoptosis. All experiments have been carried out independently at least three times.

Discussion It has been known for some years that the expression of the human tumor suppressor CEACAM1 is lost in the great majority of both colon cancers and early colon adenomas (Neumaier et al., 1993; Nollau et al., 1997a, b; Zhang et al., 1997). In this report, we have investigated the role of CEACAM1 in context with the genetic multistep model of carcinogenesis (Fearon and Vogelstein, 1990) and its temporal relationship to the defects in the APC pathway. In early adenomas with loss of CEACAM1, we detected APC defects in 70 and 75% of the cases by PTT and immunohistochemistry, respectively. This represents the overall frequency of APC mutations reported for colorectal neoplastic tumors (Powell et al., 1993; Jen et al., 1994; Suzuki et al., 2004) and demonstrates that loss of CEACAM1 is not an epiphenomenon of APC pathway defects, but occurs as an independent event with even higher frequency, thus raising the possibility that loss of CEACAM1 may precede APC pathway defects during tumorigenesis. We hypothesized that CEACAM1 expression occurs in tumor lesions prior to neoplastic transformation and Oncogene


9310

first investigated microscopically small hyperplastic ACF and HP as the earliest candidate lesions. In the past, hyperplastic lesions have been considered harmless, in part because they do not carry APC mutations (Jen et al., 1994; Otori et al., 1998). However, they are now being recognized as potential precursors within the multistep carcinogenesis (Otori et al., 1995; Jass et al., 2000; Jass, 2001). Indeed, recent evidence suggests that carcinomas can develop within a hyperplastic tumor (Hawkins et al., 2000). Similarly, it has been demonstrated in experimental animals on a carcinogenic diet that, while most hyperplastic ACF will regress, there was a 17-fold increased risk to develop a neoplastic tumor progress from a hyperplastic ACF compared to its development from normal epithelium (Shpitz et al., 1996). Considering the well-documented variety of genetic alterations in hyperplastic tumors and the very recent reports of complex and extensive DNA hypermethylation, these lesions represent a heterogeneous continuum, in which subsets are likely to have a significant malignant potential (Suzuki et al., 2004; Wynter et al., 2004). It is important to note that, in addition to the concept of a tumorigenesis through gene mutations, altered expressions of genes involved in proliferation and/or apoptosis are now being recognized to predispose colon crypt cells for mutational events and subsequent colon tumor development (Suzuki et al., 2004; Wynter et al., 2004). For example, epigenetic silencing of the SFRP genes leads to constitutive activation of the WNT pathway in an in vitro transfection model (Suzuki et al., 2004), and SFRP genes are methylated in a limited number of ACF investigated. Our data confirm that intracellular b-catenin accumulation, as a marker for APC pathway defects, is very uncommon in hyperplastic tumors (n ¼ 1/59, see Figure 1), while accumulation is observed very often in the neoplastic tumors (n ¼ 16/21, see Figure 1), mainly in the cytoplasm (Figure 2i). In contrast, decreased or complete loss of CEACAM1 expression is observed in up to 86% (Figure 1), a percentage close to identical to the frequencies reported for neoplastic colon tumors (Neumaier et al., 1993; Nollau et al., 1997b). Altogether, the loss of CEACAM1 expression is (i) independent of defects of the APC pathway and (ii) equally common in both hyperplastic and neoplastic tumors. These results suggest a pathophysiological link between hyperplastic lesions and the multistep carcinogenesis rather than assuming that these tumors are completely unrelated entities. One attractive concept of tumorigenesis assumes the failure of crypt cells to differentiate. Very recently, Baas et al. (2004) have shown in vitro that the polarization and differentiation of colon cancer cells can be achieved through an activation of the LKB1 tumor suppressor. However, cells in hyperplastic lesions still show a polarized morphology, and to the best of our knowledge, LKB1 has not been reported to be altered in these early lesions. Also, the role of LKB1 in sporadic colorectal tumors is controversial (Avizienyte et al., 1999; Launonen et al., 2000). Oncogene

Alternatively, the emergence of hyperplastic lesions in the colon is explained by an imbalance of crypt homeostasis due to decreased apoptosis rather than increased proliferation (Otori et al., 1995; Roncucci et al., 2000). This is consistent with our finding of markedly decreased apoptosis in the CEACAM1negative hyperplastic ACF (Po0.0001) and HP (Po0.0012). Since no increase of either Ki-67 staining or b-catenin accumulation was observed, we believe that defects in the WNT signaling pathway were not involved in the origin of these very early lesions. Interestingly however, only few HP showed normal CEACAM1 expression, and these tumors had a significantly higher rate of apoptosis than CEACAM1-positive tumors (Po0.05; Figure 4) supporting the concept of a functional role of CEACAM1 in apoptosis in the colon epithelium. We tested this hypothesis in vitro using two standard models: firstly, CEACAM1-transfected Jurkat cells express the antigen on their cell surface (gift by Dr J Shively) and provide a favorable read-out system compared to many cancer cell lines, as they are sensitive to apoptotic stimuli. Secondly, we used human HT29 colon cells, in which the low levels of CEACAM1 expression can be upregulated by IFNg (Takahashi et al., 1993). These cells display a relative resistance to various apoptotic stimuli (Battu et al., 1998; Tan et al., 2002). Both model systems independently demonstrate that CEACAM1 acts as a specific trigger for apoptosis, when crosslinked on the cell surface. For example, Figure 5 shows that the CEACAM1 transfection does neither change the susceptibility of the Jurkat cells for apoptotic stimuli nor does CEACAM1 expression influence the base rate of apoptosis. In contrast, crosslinking with CEACAM1 antibodies leads to increased apoptosis in these CEACAM1-expressing cells, but not in the untransfected Jurkat control cells. Significant CEACAM1mediated apoptosis can also be demonstrated in the HT29 cells after induction of antigen expression (Figure 6). As in the Jurkat experiments, apoptosis was independent of the level of CEACAM1 expression, but required specific crosslinking of the antigen on the cell surface. Similarly, crosslinking of CEACAM1 also appears to be required for signal transduction to ERK1/ 2 (Singer et al., 2002). Taken together, these experiments clearly demonstrate a new function of CEACAM1 as a trigger of apoptosis in the colon and suggest for the first time a function for the maintenance of tissue homeostasis and maturation in the colon epithelium. While this work was in progress, Shively and colleagues had shown that CEACAM1 can induce lumen formation during glandular differentiation and morphogenesis of mammary epithelia by apoptosis in vitro (Kirshner et al., 2003). Accordingly, the regulation of morphogenesis by apoptosis may be a major function of this molecule in different tissues. CEACAM1 is known to play an important role in tumor suppression and differentiation, although its mechanism is unclear (for review see Kunath et al., 1995; Turbide et al., 1997; Beauchemin et al., 1999; Hammarstrom, 1999; Fournes et al., 2001). We believe


9311

that the results presented here give an explanation for some of the tumor suppressor properties. It is firmly established that the cell surface expression of CEACAM1 and other members of the CEACAM family commences in the lower half of the colonic crypt and becomes more prominent in the glycocalix as these cells differentiate and migrate toward the gut lumen. Within the glycocalix coat, the CEACAM molecules form a dense network via homo- and heterophilic adhesions (Frangsmyr et al., 1995; Hammarstrom, 1999), a process simulated by our in vitro crosslinking models. We believe that the maturing and migrating crypt cells receive proapoptotic signal through the increasing number of CEACAM1 molecules crosslinked by molecular adhesion between the CEACAMs on the cellular surface. It is noteworthy that CEACAM1 is the only member in this gene family equipped with a transmembrane and a cytoplasmic domain. Indeed, signal transduction through CEACAM1 has been shown by us and others (Brummer et al., 1995; Hammarstrom, 1999). As we have shown in HT29 cells expressing different levels of CEACAM1, the failure to express CEACAM1 is therefore likely to reduce apoptosis and may directly contribute to hyperplastic tumor formation. At this point, the reasons for the frequent loss of CEACAM1 expression are unclear. Strategies to assess this question are currently restricted to in vitro models. Murine animal models have proven to be of limited value, since mice – in contrast to man – possess two CEACAM1 homologs with different functions (Robitaille et al., 1999) and knockouts have only been partially successful for the murine homolog ceacam1 (Blau et al., 2001).

How can the loss of CEACAM1 contribute to our current models of colon carcinogenesis? As we propose in Figure 7, the failure to express the proapoptotic CEACAM1 contributes to the generation of hyperplastic lesions, which may accumulate various genetic or epigenetic alterations. However, prior to defects of a gatekeeper like the APC pathway, these are not permissive for tumor progression and will lead to spontaneous regression of the lesion (Powell et al., 1992; Jen et al., 1994; Otori et al., 1995; Hawkins et al., 2000). If cells within the hyperplastic lesion suffer a gatekeeper defect, they would take the ‘Vogelstein exit’ to neoplastic tumor growth. The proposition that hyperplastic lesions may represent the initial, although very inefficient precursors of neoplastic tumor development also allows for a molecular explanation of the high frequency of APC defects in neoplastic tumors. Specifically, somatic gene defects must occur independently on both APC alleles within few cell divisions, that is, before the differentiating crypt cells lose their ability to replicate (reviewed in Goss and Groden, 2000; Michor et al., 2004). Although the first hit on APC should be expected within the crypt stem cell compartment to perpetuate the first defective allele for the progeny, recent data demonstrate that APC mutations are first identified outside the stem cell compartment (Shih et al., 2001). The broadened proliferation zones and decreased apoptosis in the CEACAM1negative hyperplastic lesions are consistent with these data and the view that hyperplasia represents a predisposing event preceding APC defects and neoplasia in these crypts. Support for this model of comes from a very recent study by Michor et al. (2004). Using a mathematical model for the cellular dynamics in the colon crypt and colon cancer initiation, these authors conclude that chromosomal changes are very likely to precede APC mutations in colon carcinogenesis. Further studies are now needed to unravel the gene regulatory pathways governing CEACAM1 expression to detect the cause for the failure of crypt cells in starting the expression program and the molecular mechanisms by which CEACAM1-expressing cells are susceptible to apoptotic stimuli.

Materials and methods General study design

Figure 7 Extended model of colorectal tumorigenesis based on the loss of CEACAM1 antigen expression that is common to both the hyperplastic and neoplastic pathways of tumor development. Loss of CEACAM1 expression results in reduced apoptosis of crypt cells and contributes to hyperplastic growth (ACF and polyps). Hyperplastic tumors subsequently contract genetic and/or epigenetic defects of different qualities: nonpermissive alterations, for example, non-gatekeeper mutations in genes like k-ras or p53 lead to regression of the lesion and subsequently restoring normal mucosa. Permissive alterations (asterisk) involve mutations of gatekeeper genes, for example, APC or b-catenin, subsequently leading to initiation of the adenoma/carcinoma sequence (referred to as ‘Vogelstein exit’)

Fresh colorectal tissues (normal mucosa, n ¼ 55; ACF, n ¼ 32; HP, n ¼ 42; adenomas, n ¼ 36; carcinomas, n ¼ 5) were obtained from surgical specimens and processed using histopathological examination with dissection of normal and tumor tissue. Of the 36 adenomas, 20 were subjected to nucleic acids preparation, while the remaining samples were used for immunohistochemical analyses. Patients with adenomatosis or a history of chronic inflammatory bowel disease were excluded. Functional studies were performed in in vitro models. Protein truncation test for APC Nucleic acids were isolated from 20 adenomas as described (Nollau et al., 1997b; Sambrook et al., 1989). From 2 mg total Oncogene


9312 RNA, RT–PCR amplification of exons 1–14 of the APC gene was performed as described (Bala et al., 1996) using 10 pmol of primer RT15 50 ACGAGAACTATCTA AGCTTCC30 and 300 U Superscript II (Gibco BRL, Karlsruhe, FRG). For PCR, the High Fidelity PCR system (ROCHE Molecular Systems, Mannheim, FRG) was used with 20 pmol of the primers APC1T7 50 GGATCCTAATACGACTCACTATAG GAACAG ACCACCATGGCTGCAGCTTCATATGATC30 and APC15.3 50 GTAGACAGTTGTTCTCTC TTAGGATTT 30 . PCR conditions were 941C (3 min) followed by 35 cycles of 941C (1 min), 641C (1.5 min) and 721C (2 min) with a final extension of 721C for 10 min. Exon 15 sequences were amplified from genomic DNA as described (Kirchgesser et al., 1998) initially at 941C (5 min) followed by 35 cycles of 941C (30 s), 601C (30 s) and 721C (2 min) with a final extension of 721C for 10 min. In vitro mRNA transcription and translation reactions were performed using commercially available PTT kit reagents (ROCHE Molecular Systems). The translated products were separated by 10% SDS–PAGE, transferred to PVDF membranes with a semidry blotter (Biometra, Go¨ttingen, FRG) for 45 min at 200 mA and stained with the Biotin/Streptavidin Chemiluminescence’s Kit (ROCHE Molecular Systems) (Kirchgesser et al., 1998). Chemiluminescent bands were detected on the PVDF membranes after exposure of X-OMAT AR film (Kodak GmbH, Stuttgart, FRG) for up to 30 s.

Immunochemistry of tissue samples For the identification of ACF, areas of 10–18 cm2 were cut from normal colon mucosa tissues, fixed on glass slides and immersed in 4% formaldehyde/PBS at 41C for 1–4 h. The samples were stained with 0.2% methylene blue (Sigma, Taufkirchen, FRG) for 3 min. ACF were identified at a low power magnification (Bird, 1987), marked with permanent ink, dissected and subsequently embedded in paraffin. Tissues from HP, adenomas and carcinomas also were formalin-fixed and embedded in paraffin. Prior to H&E staining and immunohistochemistry, 5 mm tissue sections were deparaffinized and washed in PBS. Antigen retrieval was carried out in a microwave oven for 10 min at 650 W in 10 mM citric buffer (pH 6.0) or for 3 min at 650 W in 1 mM EDTA buffer (pH 8.0) followed by 7 min at 160 W. The following monoclonal antibodies were used for immunohistochemistry: M30 against a cytokeratin 18 neo-epitope generated by caspase activity (ROCHE Biochemicals, Mannheim, FRG), 29H2 against human CEACAM1 (Novocastra, Newcastle, UK) (dilution 1/50; v/v), T84.1 and T84.66 against CEA (5 mg/ml) (Nap et al., 1992), CH11 to human CD95 (Coulter Immunotech, Heidelberg, FRG) (100 ng/ml) and 7D11 against b-catenin (Alexis, Gru¨nberg, FRG) (5 mg/ml). Antibody binding was detected using the HRP-based immunohistochemistry Elite kit with diaminobenzidine according to the manufacturer’s standard protocol (Vector Laboratories, Burlingame, CA, USA). Immunohistochemical staining was examined by light microscopy (Leica Microsystems, Wetzlar, Germany) and images were captured by an attached DCC camera (Canon, Krefeld, Germany). Expression of antigens was scored semiquantitatively as lost (), diminished ( þ ), normal þ or elevated þ þ in comparison to adjacent normal tissue of the same section. The CEACAM1 staining in hyperplastic and neoplastic lesions was assessed by comparing the intensity to the staining in the adjacent normal mucosa of the same slide to exclude interassay variance. With respect to the normal mucosa section, a Oncogene

decreased CEACAM1 expression was defined as uniformly diminished staining intensity across the entire lesion or the complete CEACAM1 loss in single crypts of a lesion. Complete absence of staining for CEACAM1 was scored as loss of expression. Digital image analysis For the evaluation of b-catenin accumulation in tissue sections, microscopic image files displaying both tumor and adjacent normal tissues were scanned and analysed using a pixel-based graphics software (Photo Paint 7.0, Corel Corp.). Briefly, comparably sized tissue areas representing the luminal half of cells were converted to monochromatic images and subsequently corrected for background staining. A histogram function was generated from Pixel density and mean pixel shades and the b-catenin accumulation was expressed as the ratio of mean pixel shade {lesion}/mean pixel shade {normal tissue}. Apoptosis in CEACAM1 transfectants and human HT29 colon cancer cells The cell culture media of CEACAM1-expressing Jurkat cells (kindly provided by Dr J Shively) were supplemented for 4 or 16 h with the different monoclonal antibodies CH11 (100 ng/ml), 4D1C2 (4 mg/ml), T84.1 (8 mg/ml) or 7D11 (10 mg/ml). Subsequently, the media were replaced with media containing 15 mg/ml goat anti-mouse immunoglobulin (Dianova, Hamburg, FRG). At 80 or 92 h later, the apoptotic cells were evaluated using propidium iodide or the cell death detection ELISA (Roche, Mannheim, FRG) according to the manufacturer’s protocol. HT-29 cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated FCS, 4 mM L-glutamine and penicillin/ streptomycin. Cells from confluent cultures were trypsinized, seeded at approximately 60 000 cells/well overnight and stimulated with IFNg as described (Takahashi et al., 1993). Antibodies were used under the conditions described above for the Jurkat cell experiments. Statistical analysis Apoptosis in the colon tissue sections was evaluated as mean percentage of M30-positive mucosa cells in the respective tissue sections. Statistical analysis was performed using the Welch’s method of the Student’s t-test for unequal standard deviations. All cell culture experiments were carried out in triplicate and were independently repeated twice. Evaluations were carried out in a blinded fashion. For statistical analysis, the Wilcoxon’s rank test was used.

Abbreviations ACF, aberrant crypt foci; APC, gene for adenomatous polyposis coli; CEACAM, CEA-like cell adhesion molecule; HP, hyperplastic polyp; PTT, protein truncation test. Acknowledgements We are grateful to Bert Vogelstein for critical reading of the manuscript and his helpful advice. We thank John E Shively for the generous gift of CEACAM1-transfected Jurkat cells and Christoph Wagener for supplying mab 4D1C2. MN was supported in part by the German Research Council Grant Ne677/2.


9313 References Avizienyte E, Loukola A, Roth S, Hemminki A, Tarkkanen M, Salovaara R, Arola J, Butzow R, Husgafvel-Pursiainen K, Kokkola A, Jarvinen H and Aaltonen LA. (1999). Am. J. Pathol., 154, 677–681. Baas AF, Kuipers J, van der Wel NN, Batlle E, Koerten HK, Peters PJ and Clevers HC. (2004). Cell, 116, 457–466. Bala S, Kraus C, Wijnen J, Meera Khan P and Ballhausen WG. (1996). Hum. Genet., 98, 528–533. Battu S, Rigaud M and Beneytout JL. (1998). Anticancer Res., 18, 3579–3583. Beauchemin N, Draber P, Dveksler G, Gold P, Gray-Owen S, Grunert F, Hammarstrom S, Holmes KV, Karlsson A, Kuroki M, Lin SH, Lucka L, Najjar SM, Neumaier M, Obrink B, Shively JE, Skubitz KM, Stanners CP, Thomas P, Thompson JA, Virji M, von Kleist S, Wagener C, Watt S and Zimmermann W. (1999). Exp. Cell Res., 252, 243–249. Bird RP. (1987). Cancer Lett., 37, 147–151. Bird RP. (1995). Cancer Lett., 93, 55–71. Blau DM, Turbide C, Tremblay M, Olson M, Letourneau S, Michaliszyn E, Jothy S, Holmes KV and Beauchemin N. (2001). J. Virol., 75, 8173–8186. Brummer J, Neumaier M, Gopfert C and Wagener C. (1995). Oncogene, 11, 1649–1655. Esteller M, Sparks A, Toyota M, Sanchez-Cespedes M, Capella G, Peinado MA, Gonzalez S, Tarafa G, Sidransky D, Meltzer SJ, Baylin SB and Herman JG. (2000). Cancer Res., 60, 4366–4371. Fearnhead NS, Britton MP and Bodmer WF. (2001). Hum. Mol. Genet., 10, 721–733. Fearon ER and Vogelstein B. (1990). Cell, 61, 759–769. Fournes B, Sadekova S, Turbide C, Letourneau S and Beauchemin N. (2001). Oncogene, 20, 219–230. Frangsmyr L, Baranov V, Prall F, Yeung MM, Wagener C and Hammarstrom S. (1995). Cancer Res., 55, 2963–2967. Goss KH and Groden J. (2000). J. Clin. Oncol., 18, 1967–1979. Hammarstrom S. (1999). Semin. Cancer Biol., 9, 67–81. Hawkins NJ, Gorman P, Tomlinson IP, Bullpitt P and Ward RL. (2000). Am. J. Pathol., 157, 385–392. Hawkins NJ and Ward RL. (2001). J. Natl. Cancer Inst., 93, 1307–1313. Herter P, Kuhnen C, Muller KM, Wittinghofer A and Muller O. (1999). J. Cancer Res. Clin. Oncol., 125, 297–304. Izzi L, Turbide C, Houde C, Kunath T and Beauchemin N. (1999). Oncogene, 18, 5563–5572. Jass JR. (2001). Dis. Colon Rectum, 44, 163–166. Jass JR, Iino H, Ruszkiewicz A, Painter D, Solomon MJ, Koorey DJ, Cohn D, Furlong KL, Walsh MD, Palazzo J, Edmonston TB, Fishel R, Young J and Leggett BA. (2000). Gut, 47, 43–49. Jen J, Powell SM, Papadopoulos N, Smith KJ, Hamilton SR, Vogelstein B and Kinzler KW. (1994). Cancer Res., 54, 5523–5526. Kikuchi Y, Dinjens WN and Bosman FT. (1997). Virchows Arch., 431, 111–117. Kinzler KW and Vogelstein B. (1996). Cell, 87, 159–170. Kirchgesser M, Albers A, Vossen R, den Dunnen J, van Ommen GJ, Gebert J, Dupont C, Herfarth C, von KnebelDoeberitz M and Schmitz-Agheguian G. (1998). Clin. Chem. Lab. Med., 36, 567–570. Kirshner J, Chen CJ, Liu P, Huang J and Shively JE. (2003). Proc. Natl. Acad. Sci. USA, 100, 521–526. Kunath T, Ordonez-Garcia C, Turbide C and Beauchemin N. (1995). Oncogene, 11, 2375–2382. Launonen V, Avizienyte E, Loukola A, Laiho P, Salovaara R, Jarvinen H, Mecklin JP, Oku A, Shimane M, Kim HC, Kim

JC, Nezu J and Aaltonen LA. (2000). Cancer Res., 60, 546–548. Laurent-Puig P, Blons H and Cugnenc PH. (1999). Eur. J. Cancer Prev., 8, S39–S47. Michor F, Iwasa Y, Rajagopalan H, Lengauer C and Nowak MA. (2004). Cell Cycle, 3, 358–362, Epub 2004 March. Mori H, Yamada Y, Kuno T and Hirose Y. (2004). Mutat. Res., 566, 191–208. Nap M, Hammarstrom ML, Bormer O, Hammarstrom S, Wagener C, Handt S, Schreyer M, Mach JP, Buchegger F, von Kleist S, Grunert F, Seguin P, Fuks A, Holm R and Lamerz R. (1992). Cancer Res., 52, 2329–2339. Neumaier M, Paululat S, Chan A, Matthaes P and Wagener C. (1993). Proc. Natl. Acad. Sci. USA, 90, 10744–10748. Nollau P, Prall F, Helmchen U, Wagener C and Neumaier M. (1997a). Am. J. Pathol., 151, 521–530. Nollau P, Scheller H, Kona-Horstmann M, Rohde S, Hagenmuller F, Wagener C and Neumaier M. (1997b). Cancer Res., 57, 2354–2357. Otori K, Konishi M, Sugiyama K, Hasebe T, Shimoda T, Kikuchi-Yanoshita R, Mukai K, Fukushima S, Miyaki M and Esumi H. (1998). Cancer, 83, 896–900. Otori K, Sugiyama K, Hasebe T, Fukushima S and Esumi H. (1995). Cancer Res., 55, 4743–4746. Powell SM, Petersen GM, Krush AJ, Booker S, Jen J, Giardiello FM, Hamilton SR, Vogelstein B and Kinzler KW. (1993). N. Engl. J. Med., 329, 1982–1987. Powell SM, Zilz N, Beazer-Barclay Y, Bryan TM, Hamilton SR, Thibodeau SN, Vogelstein B and Kinzler KW. (1992). Nature, 359, 235–237. Robitaille J, Izzi L, Daniels E, Zelus B, Holmes KV and Beauchemin N. (1999). Eur. J. Biochem., 264, 534–544. Roncucci L, Pedroni M, Vaccina F, Benatti P, Marzona L and De Pol A. (2000). Cell Prolif., 33, 1–18. Rowan AJ, Lamlum H, Ilyas M, Wheeler J, Straub J, Papadopoulou A, Bicknell D, Bodmer WF and Tomlinson IP. (2000). Proc. Natl. Acad. Sci. USA, 97, 3352–3357. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Cloning: A Laboratory Manual 2nd edn. Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY. Shih IM, Wang TL, Traverso G, Romans K, Hamilton SR, Ben-Sasson S, Kinzler KW and Vogelstein B. (2001). Proc. Natl. Acad. Sci. USA, 98, 2640–2645. Shpitz B, Hay K, Medline A, Bruce WR, Bull SB, Gallinger S and Stern H. (1996). Dis. Colon Rectum, 39, 763–767. Singer BB, Scheffrahn I, Heymann R, Sigmundsson K, Kammerer R and Obrink B. (2002). J. Immunol., 168, 5139–5146. Smith AJ, Stern HS, Penner M, Hay K, Mitri A, Bapat BV and Gallinger S. (1994). Cancer Res., 54, 5527–5530. Suzuki H, Watkins DN, Jair KW, Schuebel KE, Markowitz SD, Dong Chen W, Pretlow TP, Yang B, Akiyama Y, Van Engeland M, Toyota M, Tokino T, Hinoda Y, Imai K, Herman JG and Baylin SB. (2004). Nat. Genet., 36, 417–422. Takahashi H, Okai Y, Paxton RJ, Hefta LJ and Shively JE. (1993). Cancer Res., 53, 1612–1619. Tan S, Seow TK, Liang RC, Koh S, Lee CP, Chung MC and Hooi SC. (2002). Int. J. Cancer, 98, 523–531. Turbide C, Kunath T, Daniels E and Beauchemin N. (1997). Cancer Res., 57, 2781–2788. Wynter CV, Walsh MD, Higuchi T, Leggett BA, Young J and Jass JR. (2004). Gut, 53, 573–580. Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, Hamilton SR, Vogelstein B and Kinzler KW. (1997). Science, 276, 1268–1272. Oncogene