CEACAM6 Gene Silencing Impairs Anoikis Resistance and in Vivo Metastatic Ability of Pancreatic Adenocarcinoma Cells

CEACAM6 gene silencing impairs anoikis resistance and in vivo metastatic ability of pancreatic adenocarcinoma cells

Mark S Duxbury1, Hiromichi Ito1, Michael J Zinner1, Stanley W Ashley1 and Edward E Whang*,1

1Department of Surgery, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA

Anoikis is the apoptotic response induced in normal cells maintain tissue order within multisystem organisms by by inadequate or inappropriate adhesion to substrate. It is preventingectopic cellular proliferation (Meredith et al., postulated that resistance to anoikis facilitates tumorigen- 1993; Frisch and Francis, 1994), resistance to anoikis, a esis and metastasis. Carcinoembryonic antigen-related property of tumor cells, is reported to contribute to cell adhesion molecule 6 (CEACAM6) is an immunoglo- primary tumorigenesis and metastasis in a number of bulin superfamily member overexpressed in a number of malignancies (Yawata et al., 1998; Streuli and Gilmore, human cancers and implicated in anoikis resistance. We 1999; Shanmugathasan and Jothy, 2000). Malignant tested the effect of CEACAM6 gene silencing on anoikis cells can be detected in compartments such as the blood in pancreatic adenocarcinoma cell lines. Anoikis was and peritoneal fluid, even in the early stages of disease, induced in PANC1, Capan2, MiaPaCa2 and MiaAR (a but metastasis is an inefficient process with only a small MiaPaCa2-derived anoikis-resistant subline) by culture in percentage of disseminated cells surviving to establish poly-2-hydroxyethylmethacrylate-coated wells. Anoikis metastases (Fidler et al., 1978). Inherent or acquired was quantified by YO-PRO-1/propidium iodide staining anoikis resistance is postulated to confer a selective and flow cytometry. The role of caspase activation was advantageon disseminatingtumor cells, promotingtheir determined using fluorometric profiling and the caspase survival in the absence of normal matrix attachment and inhibitor Z-Val-Ala-Asp-fluoromethyl ketone (Z-VAD- allowingthem to form metastatic colonies (Arvelo and fmk). CEACAM6 expression was suppressed by RNA Poupon, 2001; Zhu et al., 2001). Pancreatic adenocarci- interference. Using a nude mouse orthotopic xenograft noma is characterized by aggressive invasion and early model, we assessed the effect of this treatment on in vivo metastasis, such that 90% of patients have surgically metastatic ability. Anoikis resistance was associated with unresectable advanced disease at the time of diagnosis. increased CEACAM6 expression. CEACAM6-specific Even patients without overtly advanced disease often short interfering ribonucleic acid (siRNA), but not control subsequently develop metastases and tumor recurrence siRNA, increased susceptibility to caspase-mediated despite apparently curative resection (Farrell et al., anoikis, an effect abrogated by Z-VAD-fmk, and 1997). Anoikis resistance represents a potential target decreased Akt phosphorylation (Ser-473) under ancho- for novel antimetastatic strategies directed at preventing rage-independent conditions. CEACAM6 gene silencing or slowingprogression of this disease. reversed the acquired anoikis resistance of MiaAR and Carcinoembryonic antigen (CEA) and carcinoem- inhibited its in vivo metastatic ability. CEACAM6 bryonic antigen-related cell adhesion molecules (CEA– warrants further investigation as a novel therapeutic CAM), members of the immunoglobulin superfamily target for the treatment of pancreatic adenocarcinoma. (Thompson et al., 1991), function as intercellular Oncogene (2004) 23, 465–473. doi:10.1038/sj.onc.1207036 adhesion molecules (Kuroki et al., 2001). The 90 kDa glycosylphosphatidylinositol (GPI)-linked subtype Keywords: anoikis; pancreatic cancer; adenocarcinoma; CEACAM6, formerly known as nonspecific crossreact- CEACAM6; RNA interference; metastasis ingantigen90 or CD66c (Beauchemin et al., 1999), is found principally on neutrophils and some epithelia (Scholzel et al., 2000). Once thought merely to represent a ‘differentiation marker’, evidence is accumulatingthat CEACAM6, in particular, has functional importance in Introduction tumorigenesis, including disruption of tissue architec- ture, cell polarity, differentiation and anoikis (Ilantzis Epithelial and endothelial cells generally require cell– et al., 2002). Deregulated overexpression of CEACAM6 cell or cell–matrix contact to survive and proliferate. has been reported in numerous human cancers (Hase- The normal cellular response to inadequate or inap- gawa et al., 1993; Kodera et al., 1993; Scholzel et al., propriate intercellular or matrix attachment is a subset 2000) and may promote anoikis resistance (Ordonez of apoptosis termed anoikis. While anoikis is thought to et al., 2000; Ilantzis et al., 2002). Owingto these features of CEACAM6, we tested the *Correspondence: EE Whang; E-mail: [email protected] hypothesis that inhibition of its expression by short Received 29 May 2003; revised 21 July 2003; accepted 23 July 2003 interferingribonucleic acid (siRNA) would impair CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 466 anoikis resistance in pancreatic ductal adenocarcinoma was 1.2% compared to 14.1% for MiaPaCa2 (Po0.01. and determined the roles of Akt and caspases as Table 1). This increased anoikis resistance did not mediators of this process. We report for the first time equate to global resistance to apoptosis as MiaAR that suppression of CEACAM6 expression by siRNA remained sensitive to apoptosis induced by 100 nm enhances pancreatic adenocarcinoma cell anoikis, staurosporine. The staurosporine-induced apoptosis increases caspase activation in response to anchorage- fraction was 48% in MiaPaCa2 and 46% in MiaAR. independent conditions, impairs acquired anoikis resistance in a selected anoikis-resistant subline and inhibits metastasis in vivo. Anoikis resistance correlates with in vivo metastatic ability In vivo metastatic ability was assessed in a nude mouse xenograft model. At necropsy 4 weeks following Results orthotopic implantation of 1 Â 106 cells of each adenocarcinoma cell line, liver metastases were counted and Pancreatic cancer cells display variable anoikis resistance confirmed histologically. The in vivo liver metastatic Anoikis was induced by poly-2-hydroxyethylmethacry- ability of the untreated cell lines corresponded to late (polyHEMA) culture, a technique, originally relative anoikis resistance (Table 2). Mice implanted described by Folkman and Moscona (1978), which has with Capan2 did not develop metastases although been widely used to induce this subset of apoptosis by primary tumor growth was evident. In a separate deprivingcells of contact with substratum (Raz and Ben experiment, there was no evidence of Capan2 metastasis, Ze’ev, 1983; Kim et al., 1999). We observed considerable even 12 weeks followingimplantation. The ability of variation in the susceptibility of pancreatic adenocarci- MiaAR to form liver metastases was significantly greater noma cells to anoikis. The anoikis fraction (ratio of than parental MiaPaCa2, from which it was derived. Of apoptotic cells to total cells), quantified by flow the six mice implanted with MiaAR, all developed cytometric analysis of YO-PRO-1 and propidium iodide metastases by 4 weeks compared to 67% of mice stainingfollowing16 h polyHEMA culture, was greatest implanted with MiaPaCa2 (Po0.05). The median in the well-differentiated cell line Capan2 (anoikis number of metastases was greater in the MiaAR group fraction 27%). Resistance to anoikis was significantly (median 6 versus 2 in MiaPaCa2, Po0.05. Table 2). greater in the poorly differentiated cell lines PANC1 and MiaPaCa2 (anoikis fraction 15 and 14%, respectively, CEACAM6 suppression by siRNA suppresses in vitro Po0.01. Table 1). anoikis resistance and in vivo metastatic ability After eight selecting passages in which nonadherent MiaPaCa2 cells were serially subcultured, MiaAR, a cell Cell lines with greater anoikis resistance and metastatic line capable of proliferatingin suspension, was estab- ability demonstrated higher levels of CEACAM6 lished. The stability of this phenotype was confirmed transcript (Figure 1) and protein expression (Figure 2). after five passages under standard culture conditions. Furthermore, CEACAM6 was overexpressed at mRNA Parental MiaPaCa2 cells are adherent to culture vessel and protein levels in the anoikis-resistant cell line MiaAR surfaces, have spindle-like morphology and grow in a compared to the parental MiaPaCa2 line. Suppression discrete scattered fashion. In contrast, the majority of of CEACAM6 by specific siRNA (siCEACAM6) was MiaAR cells remain in suspension although, when confirmed at the mRNA level by Northern blot adherent to culture flask plastic, MiaAR cells have (Figure 3a), and at the protein level by Western blot rounder morphology and grow in clumped colonies. analysis 48 h post-treatment (Figure 3b). Control MiaAR showed marked anoikis resistance compared to siRNA did not affect CEACAM6 expression. Suppres- parental MiaPaCa2. Followingculture in polyHEMA- sion of CEACAM6 expression by siRNA increased coated wells for 16 h, the anoikis fraction of MiaAR cells anoikis in response to 16 h polyHEMA culture in all cell

Table 1 CEACAM6 gene silencing promotes anoikis and reverses acquired anoikis resistance Cell line Anoikis fraction (%)

PolyHEMA Control siRNA siCEACAM6 siCEACAM6+Z-VAD-fmk

MiaPaCa2 14.170.5 17.070.5 23.471.3* 3.071.0*** MiaAR 1.270.5** 2.570.5 21.171.1* 0.770.5*** PANC1 15.170.6** 16.170.9 33.672.2* 7.371.1*** Capan2 27.071.1** 29.271.1 35.272.4* 20.371.8***

Cells were treated with control siRNA, CEACAM6-suppressingsiRNA (siCEACAM6) or siCEACAM6 togetherwith the caspase inhibitor Z-VAD-fmk. Cells (1 Â 106) were cultured for 16 h in polyHEMA-coated wells to induce anoikis. The anoikis fraction was determined by YO-PRO- 1/propidium iodide stainingand ﬂow cytometry. siCEACAM6 attenuates inherent anoikis resistance and reverses acquired anoikis resistance. The role of caspase-mediated apoptosis is conﬁrmed by abrogation of the siCEACAM6-induced anoikis fraction following caspase inhibition. The data are means of three experiments7s.e.m. * ¼ Po0.01 compared to control; ** ¼ Po0.01 compared to MiaPaCa2; *** ¼ Po0.01 compared to siCEACAM6

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 467 Table 2 In vivo liver metastatic potential of pancreatic cancer cell lines Cell line Anoikis Median no. of % mice fraction in metastases with polyHEMA (range) metastasis culture (mean%7S.E.M.)

MiaPaCa2 (n ¼ 6) 14.170.5 2 (0–4) 67 MiaAR (n ¼ 6) 1.270.5** 6* (1–10) 100 PANC1 (n ¼ 6) 15.170.6** 1 (0–1) 50 Capan2 (n ¼ 6) 27.071.1** 0 (0) 0

Male nude mice, 5-week old, underwent orthotopic implantation of 1 Â 106 untreated pancreatic adenocarcinoma cells. After 4 weeks, necropsy was performed to identify the macroscopic metastases, which were excised and conﬁrmed histologically. In vitro anoikis resistance is shown for comparison. The anoikis-resistant cell line MiaAR demonstrated signiﬁcantly greater in vivo metastatic potential than the parental cell line MiaPaCa2. Capan 2 produced no metastases at 4 weeks. In a separate experiment, necropsy performed on mice 12 weeks followingimplantation with Capan2 revealed no metastases. * ¼ Po0.05, accordingto Mann–Whitney test. ** ¼ Po0.01 compared to MiaPaCa2

Figure 2 CEACAM6 is differentially expressed in pancreatic adenocarcinoma cells. Representative Western blot analysis of whole-cell lysates, containing30 mgtotal protein, extracted from pancreatic adenocarcinoma cells grown under standard culture conditions. Anti-CEACAM6 stainingdemonstrates differential expression of CEACAM6 protein, correspondingto mRNA levels. Antiactin staininghas been used as a loadingcontrol. CEACAM6 expression was quantiﬁed by densitometry and was normalized to b-actin. Results are the means of three independent determinations (7S.E.M.). *P ¼ o0.05

activities, particularly those of caspases 3 and 8, in both MiaPaCa2 and MiaAR. Mean caspase 3 activity follow- ingCEACAM6 suppression was 290% that of control siRNA-treated MiaAR (Po0.05), and 150% that of control in MiaPaCa2 (Po0.05). Mean caspase 8 activity followingCEACAM6 suppression was 200% that of control siRNA-treated MiaAR (Po0.05), and 160% that of control in MiaPaCa2 (Po0.05). Smaller but statistically significant changes in caspase 2 and 9 activities Figure 1 CEACAM6 mRNA levels differ in pancreatic adeno- were observed followingsuppression of CEACAM6 carcinoma cell lines. Northern blot analysis of total RNA extracted expression (Figure 5). The role of caspase activation in from cells cultured under standard conditions usinga CEACAM6 anoikis was confirmed by addition of the broad- sequence-specific oligonucleotide probe revealed a single band spectrum caspase inhibitor Z-VAD-fmk. Caspase in- (approximately 2.3 kb). MiaAR has the highest CEACAM6 mRNA levels. Densitometry was performed and CEACAM6 signal was hibition significantly decreased the anoikis fraction in all normalized to b-actin. *P ¼ o0.05 cell lines (Table 1). In order to determine the effects of CEACAM6 suppression on proliferation in monolayer culture, cell lines (Table 1). The magnitude of this effect was greatest countingwas performed. Treatment with neither control for MiaAR, the mean anoikis fraction increasingfrom 1.2 siRNA nor siCEACAM6 had significant effect on cell to 21% (Po0.01. Figure 4). In contrast, the anoikis proliferation at 48 h post-transfection compared to fraction was 2.5% followingcontrol siRNA treatment. untreated cells, suggesting relatively low toxicity from Increased anoikis in response to CEACAM6 gene the transfection procedure. There was also no statisti- silencingwas associated with increases in caspase cally significant difference in proliferation between cells

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 468 treated with control siRNA and those treated with Akt pathway impairs anchorage-independent growth in siCEACAM6. PANC1 pancreatic adenocarcinoma cells (Yao et al., Activation of the serine/threonine kinase Akt is 2002), and expression of constitutively active Akt has common in pancreatic adenocarcinoma (Semba et al., been shown to block epithelial cell anoikis (Khwaja 2003), and is an important determinant of pancreatic et al., 1997). We therefore assessed Akt activity in cells adenocarcinoma cellular proliferation (Perugini et al., harvested following16 h anoikis-inducingpolyHEMA 2000). Inhibition of the phosphoinositide-3-OH kinase/ culture, usingSer-473 phosphorylation as a marker of Akt activation (Downward, 1998). Cells were treated with siCEACAM6 or control siRNA 48 h before initiatingpolyHEMA culture. After 16 h of polyHEMA culture, Akt Ser-473 phosphorylation was signiﬁcantly inhibited in cells treated with siCEACAM6, compared to control siRNA (Figure 6). In contrast, there was no detectable effect on Akt phosphorylation status induced by siCEACAM6 treatment of cells in monolayer culture. At 4 weeks followingsurgical orthotopic implantation of MiaAR treated with siCEACAM6 (n ¼ 6) or control siRNA (n ¼ 6), necropsy was performed. All the control group mice developed histologically conﬁrmed liver metastases and two mice had developed gross ascites. In the siCEACAM6 group, no mouse developed metastases or ascites (Table 3, Po0.05). No lung metastases were detected in either group.

Discussion

This study was designed to investigate the role of CEACAM6 in anoikis. We have shown that native and acquired anoikis resistance in pancreatic adenocarcinoma cells is associated with increased CEACAM6 transcript levels and protein expression. We have also demonstrated that CEACAM6 gene silencing by speciﬁc siRNA increases susceptibility to caspase-mediated anoikis, reverses acquired anoikis resistance, down- regulates the Akt cell survival pathway under anchorage-independent conditions and inhibits metastasis of orthotopically xenografted pancreatic adenocarcinoma cells. These data suggest that, while overexpression of CEACAM6 may be a mechanism by which tumor cells develop greater anoikis resistance, it may also represent a therapeutic target. CEACAM6 is normally expressed by granulocytes (Skubitz et al., 2001), but is overexpressed in various human cancers (Cournoyer et al., 1988; Thompson et al., 1991; Scholzel et al., 2000; Gardner-Thorpe et al., 2002). In colorectal carcinoma, CEACAM6 levels correlate

Figure 3 (a) Specific siRNA treatment suppresses CEACAM6 mRNA expression in MiaPaCa2 and MiaAR. Northern blot analysis of total RNA demonstratingsuppression of CEACAM6 mRNA at 48 h followingCEACAM6-specific siRNA (siCEACAM6) treatment compared to control. b-Actin serves as a loadingcontrol. ( b) Specific siRNA treatment reduces CEACAM6 protein expression in MiaPaCa2 and MiaAR. Western blot of whole-cell lysates was performed 48 h post siRNA treatment. Control siRNA did not affect CEACAM6, whereas CEACAM6-specific siRNA (siCEACAM6) suppressed protein levels. Antiactin staininghas been used as a loadingcontrol. CEACAM6 expression was quantified by densitometry and normalized to b-actin. Results are the means of three independent determinations (7S.E.M.). * ¼ Po0.05

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 469

Figure 5 Fluorometric caspase profilingin MiaPaCa2 and Mia AR. Anoikis induced by polyHEMA culture is associated with a significant increase in caspase 3 and 8 activities in both MiaPaCa2. In the case of MiaAR, only caspase 8 activity was significantly increased by polyHEMA culture. CEACAM6 gene silencing (siCEACAM6) substantially increased caspase 3 and 8 activities induced by polyHEMA culture in both MiaPaCa2 and MiaAR. Figure 4 MiaAR is an anoikis-resistant sub-line of MiaPaCa2. Changes in caspase 2 and 9 activities were of a lesser magnitude. Anoikis (apoptotic) fraction of MiaPaCa2 and MiaAR after 16 h of All values are shown relative to untreated cells in standard culture. (i) standard culture in 35 mm wells or anoikis induction by (ii) The data are means7S.E.M. of four independent determinations. polyHEMA culture, (iii) polyHEMA culture followingcontrol * ¼ Po0.05 between the polyHEMA þ siCEACAM6 and other siRNA treatment, (iv) polyHEMA culture followingCEACAM6- groups suppressingsiRNA treatment (siCEACAM6), or (v) polyHEMA culture followingsiCEACAM6 in the presence of the caspase inhibitor Z-VAD-fmk. DMEM with 10% FBS was used in all cases. In total, 105 cells were analysed by flow cytometry following fully characterized. Soeth et al. have reported that stainingwith Vybrant Apoptosis kit #4 (YO-PRO-1, FL1 and endogenous CEA, a closely-related GPI-linked molecule propidium iodide, FL2). MiaAR is significantly more anoikis that is often overexpressed in human malignancies resistant than MiaPaCa2 (Po0.01). CEACAM6 gene silencing (Chevinsky, 1991), serves an antiapoptotic function in markedly increased anoikis in MiaAR compared to control siRNA treatment (Po0.01). This response was abrogated by caspase colon cancer cells. The same group reported that inhibition (Po0.01). Experiments were repeated in triplicate, and ribozyme-mediated suppression of CEA expression values are means of three experiments inhibited colony formation and reduced metastasis in a nude mouse tail vein injection model (Wirth et al., 2002). Protection from apoptotic phenomena such as anoikis is inversely with cellular differentiation (Ilantzis et al., not universal to the CEACAM family. Functional 2002). Forced overexpression of CEACAM6 in L6 rat diversity within this group of structurally related myoblasts, MDCK epithelial cells, human SW1222 and molecules is illustrated by experiments in which Caco-2 colorectal cancer cells inhibits their anoikis in CEACAM1 overexpression is induced. While promot- vitro (Ordonez et al., 2000). CEACAM6 overexpression ingcell–cell adhesion, CEACAM1 overexpression does also promotes colorectal tumorigenesis in nude mice not protect cells from anoikis (Ordonez et al., 2000). (Raz and Ben Ze’ev, 1983; Gorelik et al., 1993; Ilantzis Interestingly, CEACAM1 which, unlike CEACAM6 has et al., 2002). These results, taken together with our transmembrane and cytoplasmic domains, is often findingthat the anoikis-resistant cell line Mia AR over- downregulated in human cancers and may function as expresses CEACAM6 and has greater in vivo metastatic a tumor suppressor in some cancer models (Neumaier ability compared to parental MiaPaCa2, support the et al., 1993; Kunath et al., 1995; Busch et al., 2002). premise that anoikis resistance and greater metastatic CEACAM6 interacts homophilically as well as ability are linked, and that CEACAM6 is functionally heterophilically with other members of the CEA family. important in these processes. CEACAM6 is reported to have a specific heterophilic The mechanisms by which the GPI-linked molecule interaction with CEACAM8 via the extracellular CEACAM6 allows cells to escape anoikis remain to be N-domain at its N-terminal end (Oikawa et al., 1991).

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 470 Table 3 CEACAM6 gene silencing by RNA interference suppresses the metastatic ability of pancreatic adenocarcinoma cells in vivo siCEACAM6 Control siRNA (n ¼ 6) (n ¼ 6)

Median no. of metastases (range) 0 3 (1–8)* Mice with metastases (%) 0 100*

At 4 weeks followingorthotopic implantation of 1 Â 106 MiaAR cells, necropsy was performed. Liver metastases were counted and conﬁrmed histologically. siCEACAM6 treatment signiﬁcantly inhibited metastasis in this nude mouse model compared to control siRNA. *Po0.05

CEACAM6 expression can influence important cancer cell survival pathways. Our understandingof how GPI-linked membrane proteins influence the intracellular milieu is far from complete. Despite absence of a transmembrane domain, GPI-linked proteins such as CEACAM6 are capable of transducingtransmembrane signals.In the neutrophil, specific antibody-mediated crosslinkingof CEACAM6 induces neutrophil activation (Robinson and Hederer, 1994; Skubitz et al., 2001). Although lacking a proteinaceous intracellular domain, colocalization experiments indicate that GPI-linked proteins such as CEACAM6 may be able to influence intracellular events by modulatingsecond messengersand src-family tyr- osine kinases in lipid rafts via their GPI anchor (Brown, 1993; Malek et al., 1994). These molecular interactions Figure 6 The effect of siCEACAM6 treatment on Akt activity. At suggest potential pathways through which CEACAM6 48 h followingtreatment with control siRNA or siCEACAM6, cells can influence susceptibility to anoikis. were harvested and cell lysates analysed for serine-473-phosphory- RNA interference is emerging as a powerful technique lated Akt (p-Akt) and total Akt by Western blot. Densitometry was for the dissection of individual gene function. However, performed as p-Akt normalized to total Akt. siCEACAM6 treatment produced a significant decrease in p-Akt in both MiaAR this technique is not without limitations. Although and MiaPaCa2. *¼ Po0.05 siCEACAM6 treatment produces a marked decrease in CEACAM6 expression, silencingis not absolute. This may be due to a number of factors includingincomplete CEACAM6 also appears to interact with non-CEA suppression of CEACAM6 mRNA and the persistence cell surface molecules. Overexpression of CEACAM6 of pre-existingCEACAM6 protein. The time to in L6 myoblasts induces functional changes in the degradation of CEACAM6 may mean that some protein a5b1 integrin receptor (Stanners and Fuks, 1998), and persists despite siCEACAM6 treatment. In addition, it has been proposed that CEACAM6 interaction CEACAM6 gene suppression by this method of siRNA with integrin receptors represents a step in a putative delivery is transient, largely due to siRNA dilution as CEACAM6-mediated cell survival pathway (Ordonez cells divide (McManus and Sharp, 2002). Despite these et al., 2000). Potentially relevant to any CEACAM6- limitations, transient suppression of CEACAM6 expres- integrin cell survival pathway are observations that sion is sufficient to inhibit in vivo metastasis in this overexpression of integrin-linked kinase (ILK) induces model. This suggests that following implantation, early anchorage-independent epithelial cell cycle progression tumor cell escape and anoikis resistance, which can be (Radeva et al., 1997). ILK, a serine/threonine kinase, modulated by CEACAM6 gene silencing, contribute interacts with integrin b1, b2 and b3 cytoplasmic substantially to metastasis in the nude mouse. However, domains and phosphorylates Akt at Ser-473 (Delcomm- the suppressive effect of siCEACAM6 on in vitro anoikis enne et al., 1998). Increased resistance to caspase- resistance and on metastasis in vivo can only be mediated anoikis followingILK overexpression in considered correlative. rodent epithelial cells has been associated with In conclusion, this study provides evidence that increased Ser-473 phosphorylation of Akt (Attwell CEACAM6 plays a significant role in anoikis resistance. et al., 2000). Our observation that CEACAM6 gene The postulated link between anoikis resistance and silencingsuppresses Akt Ser-473 phosphorylation metastasis is also supported. Future experiments will under anoikis-inducingconditions, but not in mono- need to elucidate the mechanisms linkingCEACAM6 to layer culture, demonstrates that CEACAM6 assumes cell survival further. CEACAM6 appears to represent a greater functional importance under anchorage- promisingbasis on which to develop novel antimeta- independent conditions, and suggests that levels of static therapies targeting anoikis resistance.

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 471 Materials and methods transferred electrophoretically to PVDF membranes (Invitro- gen, Carlsbad, CA, USA). After blocking with PBS containing Cell lines 0.2% casein for 1 h at room temperature, membranes were incubated with 10 mg/ml mouse monoclonal anti-CEACAM6 Poorly differentiated human pancreatic cancer cell lines antibody (Mayne et al., 1993) (InnoGenex, San Ramon, CA, PANC1 and MiaPaCa2 and the well-differentiated line USA) in PBS containing0.1% Tween 20 overnightat 4 1C. Akt Capan2 were obtained from ATCC (Rockville, MD, USA). activity was determined in cells harvested from polyHEMA PANC1 and MiaPaCa2 were maintained in Dulbecco’s culture by detection of serine-473 phosphorylated Akt, using modified Eagle’s medium (DMEM) containing 10% fetal anti-phospho-Akt antibody (Cell signaling Technology, Bev- bovine serum (FBS), and Capan2 was maintained in McCoy’s erly, MA, USA) diluted 1 : 1000. The phospho-Akt signal was 5A medium containing10% FBS in a humidified (37 1C, 5% normalized to the Akt produced by 0.5 mg/ml anti-Akt (Santa CO ) incubator. All cell lines were subcultured before 2 Cruz. The Vecstaint ABC kit and DAB liquid substrate were confluence and seeded at a density of 1 Â 104 cells/cm2. All cell used for chromogenic detection, according to the manufac- lines are able to form tumors in nude mice (Miknyoczki et al., turer’s instructions. Equal protein loadingwas confirmed by 1999). probingwith 3 mg/ml antiactin monoclonal antibody (Lab Vision, Freemont, CA, USA). CEACAM6 expression was Selection of MiaAR quantified by densitometry and normalized to actin; Ser-473 We developed MiaAR, an anoikis-resistant subline of the parent phospho-Akt was normalized to total Akt, usingImagePro- line MiaPaCa2, by a method previously described (Zhu et al., Plus software. 2001). MiaPaCa2 was selected because it is a well-characterized poorly differentiated cell line with substantial inherent in Cellular proliferation assays vivo metastatic potential (Bouvet et al., 2002). Cells found suspended in the medium overlyingconfluent MiaPaCa2 cells, In total, 105 cells were seeded into wells of a six-well culture initially seeded in 75 cm2 culture flasks, were collected. These dish (Corning, Acton, MA, USA) in 2 ml of appropriate cells were centrifuged at 400 r.p.m. for 5 min, resuspended in medium. After 48 h of incubation (371C, 5% CO2), cells were fresh DMEM with 10% FBS and transferred to a 25 cm2 flask. harvested with trypsin–EDTA solution (Life Technologies, When the flask became confluent, floatingcells were again Gaithersburg, MD, USA) and counted using a hemocytometer harvested and transferred to a new 25 cm2 culture flask. After (Hausser Scientific, Horsham, PA, USA) and trypan blue eight identical serial passages, the stable subline MiaAR was exclusion. Each experiment was repeated three times and the established and transferred as nonadherent cells in suspension. mean values were used for comparison.

Northern blot analysis Anoikis induction Total RNA samples were prepared usingTrizol reagent(Gibco BRI, Rockville MD, USA) accordingto the manufacturer’s Anoikis was induced by polyHEMA (Sigma, St Louis, MI, instructions. Total RNA (20 mg), quantified by UV spectro- USA) culture. A solution of 120 mg/ml polyHEMA in 100% ethanol was made and diluted 1 : 10 in 95% ethanol; 0.95 ml/ photometry, was electrophoresed through a denaturing gel 2 (1.2% agarose, 2.2 m formaldehyde) and blotted onto a mm of this solution was pipetted into 35 mm wells and left to dry for 48 h at room temperature. Prior to use, wells were positively charged nylon membrane. After crosslinking with 6 120 mJ/cm2 UV irradiation, the filter was prehybridized for 4 h washed twice with PBS and once with DMEM. In all, 1 Â 10 at 421Cin5Â SSPE, 5 Â Denhardt’s solution, 0.1% SDS and cells of each line, suspended in 2 ml DMEM with 10% FBS, 100 mg/ml denatured salmon sperm DNA. Oligonucleotide were incubated in the polyHEMA-coated wells for 16 h in a 1 probes were custom synthesized (CEACAM6: 50-TGTAGC- humidified (37 C, 5% CO2) incubator. 0 Anoikis resistance was differentiated from general resistance CATTTTAACCAAGCAGCACATTTGTTAATT-3 , b-actin: m 50-ATTTCCCGCTCGGCCGTGGTGGTGAAGCTGTAGC- to apoptosis by exposingcells to 100n Staurosporine (Sigma) 30, Sigma-Genosys, TX, USA) and labeled with [a-32P]ATP in DMEM with 10% FBS for 16 h. Staurosporine is a broad- usingthe StripEZ kit (Ambion Inc., Austin, TX, USA). spectrum inhibitor of protein kinases and is a widely used Hybridization was carried out with 50 ngprobe at 42 1C for promoter of intracellular stress-induced apoptosis in cell 16 h. The filter was washed four times in 5 Â SSPE, 0.1% SDS culture models employingmalignantcell lines (Qiao et al., for 10 min and exposed to XAR film (Eastman Kodak Co., 1996). Rochester, NY, USA). After detection, the filter was stripped and reprobed for b-actin. The signal was quantified using Detection of anoikis ImagePro Plus software version 4.0 (Media Cybernetic, Silver Spring, MD, USA) and normalized to that of b-actin. Followinginduction of anoikis, cells were washed, resuspended in 0.5 ml of PBS, and 1ml/ml YO-PRO-1 and propidium iodide were added (Vybrants Apoptosis Assay Western blot analysis Kit #4, Molecular Probes, Eugene, OR, USA). Cells were Cells (2 Â 106) were harvested and rinsed twice with phosphate- incubated for 30 min on ice, then analysed by flow cytometry buffered saline (PBS), pH 7.4. Cell extracts were prepared with (FACScan, Becton Dickinson), measuringfluorescence emis- lysis buffer (20 mm Tris, pH 7.5, 0.1% Triton X, 0.5% sion at 530 and 575 nm. Cells stained with the green fluorescent deoxycholate, 1 mm phenylmethylsulfonyl fluoride, 10 mg/ml dye YO-PRO-1 were counted as havingundergoneanoikis. aprotinin, 10 mg/ml leupeptin) and cleared by centrifugation at Necrotic cells were stained with propidium iodide. The anoikis 12 000 g,41C. Total protein concentration was measured using index (% cells undergoing anoikis) was calculated by scoring the BCA assay kit (Sigma, St Louis, MO, USA) with bovine no less than 10 000 cells. The same method was used for serum albumin as a standard, accordingto the manufacturer’s detection of apoptosis (apoptotic index) of cells trypsinized instructions. Cell extracts containing30 mgtotal protein were from standard culture. All observations were repeated in subjected to 10% SDS/PAGE, and the resolved proteins triplicate and mean values reported.

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 472 Caspase profiling Laboratories (Wilmington, MA, USA). Mice were housed in microisolator cages with autoclaved bedding in a specific Whole-cell lysates were assayed for caspase 2, 3, 8 and 9 pathogen-free facility with 12 h light–dark cycles. They activities usingthe BD ApoAlert t fluorometric Caspase Assay received water and food ad libitum. Animals were observed Plate (BD Biosciences Clontech, Palo Alto, CA, USA) for signs of tumor growth, activity, feeding and pain in accordingto the manufacturer’s instructions. Plates were read accordance with the guidelines of the Harvard Medical Area (excitation 360 nm; emission 480 nm) usinga CytoFluor s 4000 StandingCommittee on Animals. (Applied Biosystems, Foster City, CA, USA) multiwell fluores- Mice were implanted with each pancreatic adenocarci- cence plate reader. The role of caspase activation in anoikis was noma cell line to allow comparison of metastatic ability with confirmed by performingpolyHEMA culture in the presence of in vitro anoikis resistance. To determine the effect of 100 mM benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone CEACAM6 gene silencing, mice were also implanted with (Z-VAD-fmk, Clontech). Z-VAD-fmk is a broad-spectrum anoikis-resistant cell line MiaAR treated ex vivo with caspase inhibitor shown to inhibit caspase-mediated apoptosis either control siRNA or siCEACAM6 48 h previously. Mice in pancreatic adenocarcinoma cells (Wang et al., 2001). were anesthetized with intraperitoneal ketamine (200 mg/kg), xylazine (10 mg/kg) and inoculated with tumor cells by siRNA treatment surgical orthotopic implantation (Tan and Chu, 1985). In a siRNAs were synthesized and purified (Qiagen-Xeragon, laminar flow hood, the abdomen was cleaned with isopropyl Germantown, MD, USA). CEACAM6 sense (50-CCGGACA- alcohol and a left upper transverse incision was made to GUUCCAUGUAUA-dTT-30) and antisense (50-UAUA- and includingthe peritoneum. The pancreas was exposed CAUGGAACUGUCCGG-dTT-30) siRNAs and control and 1 Â 106 cells suspended in 50 ml of PBS were slowly siRNAs (sense 50-AGGAGAUAUUUCGAGGCUU-dTT-30, injected into the body of the pancreas. The pancreas was antisense 50-AAGCCUCGAAAUAUCUCCU-dTT-30), which returned and the abdomen closed with 5-0 Vicryl (Somerville, bears no homology with relevant human genes, were dissolved NJ, USA). Six mice for each cell line were observed over 4 in buffer (100 mm potassium acetate, 30 mm HEPES-potassium weeks and killed by overdose of ketamine (400 mg/kg) and hydroxide, 2 mM magnesium acetate, pH 7.4) to a final xylazine (50 mg/kg). The liver of each animal was harvested, concentration of 20 mm, heated to 901C for 60 s and incubated metastatic foci were counted under a dissectingmicroscope at 371C for 60 min prior to use to disrupt any higher aggregates (Gorelik et al., 1993) and confirmed by hematoxylin and eosin formed duringsynthesis. Cells (2 Â 106) were plated into 35 mm staining. six-well trays and allowed to adhere for 24 h. In all, 8 ml siPORT Amine transfection reagent (Ambion Inc, Austin, TX, USA) per well was added to serum-free DMEM for a final complexing Statistical analysis volume of 200 ml, vortexed and incubated at room temperature Differences between groups were analysed using Student’s for 20 min. Then, 2 ml of siRNA solution was added to the t-test, ANOVA and Mann–Whitney U-test, for nonparametric diluted siPORT Amine transfection reagent, gently mixed and data, as appropriate, usingStatistica 5.5 software (StatSoft, incubated at room temperature for 15 min. The transfection Inc, Tulsa, OK, USA). P 0.05 was considered as statistically agent/siRNA complex was added into the wells containing o significant. 800 ml DMEM with 10% FBS and incubated in normal cell culture conditions for 6 h after which 1 ml DMEM with 10% FBS was added. All assays were performed 48 h post-treatment. Acknowledgements We acknowledge the technical assistance provided by Jan Nude mouse orthotopic xenograft model Rounds. This work was supported by The National Pancreas Male athymic nu/nu mice 5 weeks of age, weighing 20–22 g and Foundation and by departmental funds from the Department specific pathogen-free were obtained from Charles River of Surgery, Brigham and Women’s Hospital.

References

Arvelo F and Poupon MF. (2001). Acta Cient. Venez., 52, Delcommenne M, Tan C, Gray V, Rue L, Woodgett J and 304–312. Dedhar S. (1998). Proc. Natl. Acad. Sci. USA, 95, Attwell S, Roskelley C and Dedhar S. (2000). Oncogene, 19, 11211–11216. 3811–3815. Downward J. (1998). Curr. Opin. Cell Biol., 10, 262–267. Beauchemin N, Draber P, Dveksler G, Gold P, Gray-Owen S, Farrell TJ, Barbot DJ and Rosato FE. (1997). Ann. Surg., 226, Grunert F, Hammarstrom S, Holmes KV, Karlsson A, 66–69. Kuroki M, Lin SH, Lucka L, Najjar SM, Neumaier M, Fidler IJ, Gersten DM and Hart IR. (1978). Adv. Cancer Res., OBrink B, Shively JE, Skubitz KM, Stanners CP, Thomas P, 28, 149–250. Thompson JA, Virji M, von Kleist S, Wagener C, Watt S Folkman J and Moscona A. (1978). Nature, 273, 345–349. and Zimmermann W. (1999). Exp. Cell Res., 252, 243–249. Frisch SM and Francis H. (1994). J. Cell Biol., 124, 619–626. Bouvet M, WangJ, Nardin SR, Nassirpour R, YangM, Gardner-Thorpe J, Ito H, Ashley SW and WhangEE. (2002). Baranov E, JiangP, Moossa AR and Hoffman RM. (2002). Biochem. Biophys. Res. Commun., 293, 391–395. Cancer Res., 62, 1534–1540. Gorelik E, Kim M, Duty L, Henion T and Galili U. (1993). Brown D. (1993). Curr. Opin. Immunol., 5, 349–354. Clin. Exp. Metast., 11, 439–452. Busch C, Hanssen TA, Wagener C and OBrink B. (2002). Hasegawa T, Isobe K, Tsuchiya Y, Oikawa S, Nakazato H, Hum. Pathol., 33, 290–298. Nakashima I and Shimokata K. (1993). Br. J. Cancer, 67, Chevinsky AH. (1991). Semin. Surg. Oncol., 7, 162–166. 58–65. Cournoyer D, Beauchemin N, Boucher D, Benchimol S, Fuks Ilantzis C, DeMarte L, Screaton RA and Stanners CP. (2002). A and Stanners CP. (1988). Cancer Res., 48, 3153–3157. Neoplasia, 4, 151–163.

Oncogene CEACAM6 and pancreatic adenocarcinoma anoikis MS Duxbury et al 473 Khwaja A, Rodriguez-Viciana P, Wennstrom S, Warne PH Raz A and Ben Ze’ev A. (1983). Science, 221, 1307–1310. and Downward J. (1997). EMBO J., 16, 2783–2793. Robinson P and Hederer R. (1994). Braz. J. Med. Biol. Res., Kim HR, Lin HM, Biliran H and Raz A. (1999). Cancer Res., 27, 263–267. 59, 4148–4154. Scholzel S, Zimmermann W, Schwarzkopf G, Grunert F, Kodera Y, Isobe K, Yamauchi M, Satta T, Hasegawa T, Rogaczewski B and Thompson J. (2000). Am. J. Pathol., Oikawa S, Kondoh K, Akiyama S, Itoh K and Nakashima I. 156, 595–605. (1993). Br. J. Cancer, 68, 130–136. Semba S, Moriya T, Kimura W and Yamakawa M. (2003). Kunath T, Ordonez-Garcia C, Turbide C and Beauchemin N. Pancreas, 26, 250–257. (1995). Oncogene, 11, 2375–2382. Shanmugathasan M and Jothy S. (2000). Pathol. Int., 50, 273– Kuroki M, Abe H, Imakiirei T, Liao S, Uchida H, Yamauchi 279. Y, Oikawa S and Kuroki M. (2001). J. Leukoc. Biol., 70, Skubitz KM, Campbell KD and Skubitz AP. (2001). J. Pept. 543–550. Res., 58, 515–526. Malek TR, FlemingTJ and Codias EK. (1994). Semin. Soeth E, Wirth T, List HJ, Kumbhani S, Petersen A, Neumaier Immunol., 6, 105–113. M, Czubayko F and Juhl H. (2001). Clin. Cancer Res., 7, Mayne KM, Pulford K, Jones M, Micklem K, Nagel G, van 2022–2030. der Schoot CE and Mason DY. (1993). Br. J. Haematol., 83, Stanners CP and Fuks A. (1998). Cell Adhesion and Commu- 30–38. nication Mediated by the CEA family: Basic and Clinical McManus MT and Sharp PA. (2002). Nat. Rev. Genet., 3, 737– Perspective. Stanners CP (ed), 5, 57–71. 747. Streuli CH and Gilmore AP. (1999). J. Mammary. Gland. Biol. Meredith Jr JE, Fazeli B and Schwartz MA. (1993). Mol. Biol. Neoplasia, 4, 183–191. Cell, 4, 953–961. Tan MH and Chu TM. (1985). Tumour Biol., 6, 89–98. Miknyoczki SJ, ChangH, Klein-Szanto A, Dionne CA and Thompson JA, Grunert F and Zimmermann W. (1991). Ruggeri BA. (1999). Clin. Cancer Res., 5, 2205–2212. J. Clin. Lab. Anal., 5, 344–366. Neumaier M, Paululat S, Chan A, Matthaes P and Wagener C. WangW, Waters SJ, MacDonald JR, Von Hoff DD, Strodel (1993). Proc. Natl. Acad. Sci. USA, 90, 10744–10748. WE and Miller AR. (2001). Anticancer Res., 21, 1789–1794. Oikawa S, Inuzuka C, Kuroki M, Arakawa F, Matsuoka Y, Wirth T, Soeth E, Czubayko F and Juhl H. (2002). Clin. Exp. Kosaki G and Nakazato H. (1991). J. Biol. Chem., 266, Metast., 19, 155–160. 7995–8001. Yao Z, Okabayashi Y, Yutsudo Y, Kitamura T, Ogawa W and Ordonez C, Screaton RA, Ilantzis C and Stanners CP. (2000). Kasuga M. (2002). Pancreas, 24, 42–46. Cancer Res., 60, 3419–3424. Yawata A, Adachi M, Okuda H, Naishiro Y, Takamura T, Perugini RA, McDade TP, Vittimberga Jr FJ and Callery MP. Hareyama M, Takayama S, Reed JC and Imai K. (1998). (2000). J. Surg. Res., 90, 39–44. Oncogene, 16, 2681–2686. Qiao L, Koutsos M, Tsai LL, Kozoni V, Guzman J, Shiff SJ Zhu Z, Sanchez-Sweatman O, HuangX, Wiltrout R, Khokha and Rigas B. (1996). Cancer Lett., 107, 83–89. R, Zhao Q and Gorelik E. (2001). Cancer Res., 61, Radeva G, Petrocelli T, Behrend E, Leung-Hagesteijn C, 1707–1716. Filmus J, Slingerland J and Dedhar S. (1997). J. Biol. Chem., 272, 13937–13944.

Oncogene