Host-related carcinoembryonic antigen cell adhesion ... - Nature

Oncogene (2013) 32, 849–860

& 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE

Host-related carcinoembryonic antigen cell adhesion molecule 1 promotes metastasis of colorectal cancer A Arabzadeh1, C Chan1,2, A-L Nouvion1, V Breton1, S Benlolo1, L DeMarte1, C Turbide1, P Brodt2,3, L Ferri2,3 and N Beauchemin1,3,4,5 1 Goodman Cancer Research Centre, McGill University, Montreal, Quebec, Canada; 2Department of Surgery, McGill University, Montreal, Quebec, Canada; 3Department of Oncology, McGill University, Montreal, Quebec, Canada; 4Department of Medicine, McGill University, Montreal, Quebec, Canada and 5Department of Biochemistry, McGill University, Montreal, Quebec, Canada

Liver metastasis is the predominant cause of colorectal cancer (CRC)-related mortality in developed countries. Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) is a cell adhesion molecule with reduced expression in early phases of CRC development and thus functions as a tumor growth inhibitor. However, CEACAM1 is upregulated in metastatic colon cancer, suggesting a bimodal role in CRC progression. To investigate the role of this protein in the host metastatic environment, Ceacam1/ mice were injected intrasplenically with metastatic MC38 mouse CRC cells. A significant reduction in metastatic burden was observed in Ceacam1/ compared with wild-type (WT) livers. Intravital microscopy showed decreased early survival of MC38 cells in Ceacam1/ endothelial environment. Metastatic cell proliferation within the Ceacam1/ livers was also diminished. Bone marrow-derived cell recruitment, attenuation of immune infiltrates and diminished CCL2, CCL3 and CCL5 chemokine production participated in the reduced Ceacam1/ metastatic phenotype. Transplantations of WT bone marrow (BM) into Ceacam1/ mice fully rescued metastatic development, whereas Ceacam1/ BM transfer into WT mice showed reduced metastatic burden. Chimeric immune cell profiling revealed diminished recruitment of CD11b þ Gr1 þ myeloid-derived suppressor cells (MDSCs) to Ceacam1/ metastatic livers and adoptive transfer of MDSCs confirmed the involvement of these immune cells in reduction of liver metastasis. CEACAM1 may represent a novel metastatic CRC target for treatment. Oncogene (2013) 32, 849–860; doi:10.1038/onc.2012.112; published online 2 April 2012 Keywords: CEACAM1; metastasis; microenvironment; intravital microscopy; MDSC; colorectal cancer

Correspondence: Dr N Beauchemin, Goodman Cancer Research Centre, McGill University, McIntyre Building, Laboratory 708, 3655 Promenade Sir-William-Osler, Montreal, Quebec H3G 1Y6, Canada. E-mail: [email protected] Received 15 July 2011; revised 27 January 2012; accepted 29 February 2012; published online 2 April 2012

Introduction Colorectal cancer (CRC) is often diagnosed at the metastatic stage. Approximately 40% of CRC patients develop liver metastases and the majority of patients with hepatic metastasis will succumb to the disease (Hanahan and Weinberg, 2011). The process of colon tumorigenesis and early progression from adenomas to carcinomas has been well described (Hanahan and Weinberg, 2011). The molecular and cell biology events that govern hepatic metastasis development include epithelial–mesenchymal transition, apoptosis, intravasation, endothelial-, leukocyte-, platelet-tumor cell contacts, extravasation and colonization as well as the inflammatory state of the pre-metastatic niche (Kaplan et al., 2006; Gout and Huot, 2008). Critical signaling pathways such as Wnt signaling, K-RAS, EGFR, PI3K and PTEN represent central nodes for intervention and clinical therapies (Hanahan and Weinberg, 2011). Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1, abbreviated CC1 herein) is involved in the development and progression of colon cancer. CC1 is a member of the carcinoembryonic antigen family within the immunoglobulin superfamily (Beauchemin et al., 1999) and is found in epithelial, endothelial, lymphoid and myeloid cells (Gray-Owen and Blumberg, 2006). Structurally, CC1 consists of many isoforms with either 2, 3 or 4 heavily glycosylated extracellular immunoglobulin domains, a transmembrane segment and either a long (71–73 amino acids) or short (10 amino acids) cytoplasmic tail, produced by alternative splicing (Gray-Owen and Blumberg, 2006). These are referred to as murine CC1-4L, CC1-4S, CC12L and CC1-2S (Beauchemin et al., 1999). CC1 functions as an adhesion molecule and contributes to the maintenance of normal tissue homeostasis (Gray-Owen and Blumberg, 2006). In addition, CC1 is an angiogenesis modulator (Horst et al., 2006), a regulator of insulin metabolism (Poy et al., 2002; Xu et al., 2009), and innate and adaptive immune responses (Gray-Owen and Blumberg, 2006) as well as a microbial and viral receptor (Hemmila et al., 2004; Gray-Owen and Blumberg, 2006). With respect to tumor development, CC1 functions as a growth inhibitor in a number of early solid neoplasms

Decreased CRC experimental metastasis in Ceacam1-null mice A Arabzadeh et al

850

including tumors in the colon (Neumaier et al., 1993), prostate (Busch et al., 2002), liver (Laurie et al., 2005), endometrium (Bamberger et al., 1998), bladder (OliveiraFerrer et al., 2004) and breast (Kirshner et al., 2003). Colonocytes in CC1-deficient mice (Cc1/) exhibit increased proliferation and decreased apoptosis (Leung et al., 2006). In addition, azoxymethane carcinogenic induction or crosses with an Apc1638N/ þ mouse model of intestinal cancer demonstrated that Cc1/ mice or compound mice (Apc1638N/ þ :Cc1/) exhibited a higher colonic or intestinal tumor burden than their wild-type (WT) or Apc1638N/ þ counterparts (Leung et al., 2008). Although downregulation of CC1 in a number of epithelial cell types including colonocytes correlates with tumor development, elevated CC1 expression in cancers such as non-small cell lung cancer (Dango et al., 2008), thyroid cancer (Liu et al., 2007), gastric carcinoma (Zhou et al., 2009), pancreatic tumors (Serra et al., 2009), malignant melanoma (Ebrahimnejad et al., 2004) and even metastatic colonic cancer cells (Yeatman et al., 1997; Ieda et al., 2011) is associated with increased invasiveness, metastatic spread and unfavorable prognosis. Whether CC1-expressing stromal and immune cells at the metastatic target site influence the tumor metastatic growth has not yet been addressed. In this study, we examined the effect of stromalderived CC1 on the establishment and growth of CRC metastases in the liver using experimental metastasis assays in Cc1/ mice. Intrasplenic injections bypasses the metastatic intravasation step but have been used as a model since 1986 and murine hepatic colonization of tumor cells resembles that observed in humans (Giavazzi et al., 1986). Our studies demonstrate a significant reduction in liver metastatic burden in Cc1/ mice upon intrasplenic injections of MC38 mouse colon carcinoma cells or in lung metastases after tail vein injection of B16F10 melanoma cells. Early survival of MC38 cells and metastatic cell proliferation within the Cc1/ liver was significantly decreased. We observed a reduction of bone marrow-derived cell (BMDC) recruitment and immune infiltrates coincident with downregulation of CCL2, CCL3 and CCL5 chemokine secretion in Cc1/ metastatic livers. Bone marrow transplantations (BMTs) and immune cell enumerations established that CC1 deletion impacted principally on recruitment of CD11b þ Gr1 þ myeloid-derived suppressor cells (MDSCs), which was further confirmed by adoptive transfer experiments and metastatic assays. Therefore, absence of CC1 from the stroma provides an inhibitory microenvironment for metastases that reduces the liver colonization by CRC cells.

Results Cc1-null mice exhibit reduced metastasis of cancer cells independent of tumor cell type and metastatic target organ In humans and in mouse models, CC1 is downregulated in the early phases of colon cancer development with Oncogene

significant reduction in hyperplastic lesions, microadenomas and adenomas (Nollau et al., 1997). Loss of CC1 disturbed colonic and intestinal homeostasis and azoxymethane-treated Cc1/ mice or compound Apc1638N/ þ : Cc1/ mice demonstrated a significant increase in the colon and intestinal tumor burden, respectively, relative to either WT or Apc1638N/ þ littermates (Leung et al., 2006, 2008). We thus questioned how systemic CC1 deletion influences host metastatic development by using an experimental metastasis model whereby MC38 metastatic CRC cells were injected intrasplenically into WT and Cc1/ mice. Animals were killed at 14 days postinjection, their livers dissected, metastatic nodules enumerated and tissues were processed for immunofluorescence. MC38 CC1-negative sorted cells (Figure 1a) remained negative in metastatic nodules in vivo, in comparison to WT parenchyma in close proximity with CD31- and CC1-positive sinusoids. A number of well-formed vessels inside the nodules showed colocalization of CD31 and CC1, indicative of expression of CC1 by endothelial cells in WT livers (Supplementary Figure S1). WT mice developed a fivefold increase in metastatic nodules (mean ¼ 74±10, Po0.0001; Figures 1c, d and i) relative to those of CC1-null livers (mean ¼ 13±6) (Figures 1e, f and i). Furthermore, WT livers exhibited nodules twice the size of those found in Cc1/ livers (0.40±0.04 versus 0.20±0.03 mm2, Po0.0001; Figures 1g, h and j) and thus systemic CC1 depletion impaired CRC metastatic development postintravasation. To evaluate whether this was a general phenomena in Cc1/ mice, we injected B16F10 melanoma cells, positive for cell surface CC1 (Figure 1b), into the tail vein of WT and Cc1/ mice. After 14 days of inoculation, WT lungs (Figures 1k and l) showed a threefold increase of metastatic foci relative to Cc1/ lungs (32±7 versus 12±2, Po0.01; Figure 1m). WT lung foci were larger than Cc1/ foci, with 11 WT nodules (n ¼ 8) versus 4 Cc1/ nodules (n ¼ 10) larger than 1.5 mm2. Thus, irrespective of the route of injection, type of metastatic epithelial cells, and target organ, systemic CC1 depletion leads to a reduced metastatic burden in experimental metastasis assays. Absence of CC1 in the liver compromises tumor cell arrest, early survival and proliferation of metastatic nodules but enhances vascular density To gauge whether lack of CC1 in the liver environment affected early survival of MC38 cells, CSFE-labeled cells were injected intrasplenically in WT or Cc1/ mice and enumerated using intravital microscopy at 0.5, 24 and 48 h postinoculation. This technique is performed on live anesthetized mice with their livers exposed over the objective of a fluorescence microscope. WT and CC1-null mice showed a significant difference in the average number of cells/field at 30 min (day 0) postinoculation with Cc1/ mice exhibiting a 2.2-fold decrease in tumor cell arrest in the hepatic sinusoids (Figure 2a). After 48 h, labeled MC38 cells in Cc1/


851

Figure 1 Metastasis development in WT and Cc1/ mice. Experimental metastasis was performed in WT and Cc1/ mice by injecting either MC38 CRC C57Bl/6 cells (a) negative for CC1 cell surface expression, or B16F10 melanoma cells (b) positive for CC1 as determined by fluorescent activated cell sorting analyses using the anti-CC1 polyclonal antibody # 3759. MC38 CRC cells were injected intrasplenically into either (c, d) WT, or (e, f) Cc1/ mice. (g, h) Mice were killed 14 days postinjection and livers were processed for immunofluorescence and hematoxylin and eosin staining (L, liver; T, tumor). Numbers and size of metastatic lesions were enumerated on the livers and then on tissue sections. Cc1/ mice exhibited a significant decrease in (i) the number (Po0.0001) and (j) size (Po0.0001) of metastatic nodules. WT mice: n ¼ 8, KO mice: n ¼ 6, sections/mouse: n ¼ 4. B16F10 melanoma cells were injected intravenously into the tail vein of either (k) WT or (l) Cc1/ mice. (m) Mouse lungs were retrieved 14 days postinoculation and metastatic nodules (darkly-stained melanin) were counted. Cc1/ mice exhibited a threefold decrease in the number of metastatic foci (Po0.05).

livers were reduced threefold compared with those in WT livers (6% versus 18%, respectively; obtained after normalization to data at 0.5 h time point). This indicated that perhaps unfavorable microenvironmental conditions prevented survival of metastatic cells in Cc1/ mice relative to their WT counterparts. In addition, once metastatic nodules formed in the liver, the percentage of proliferating Ki-67-stained MC38 cells in Cc1/ background was approximately twofold less than those in WT livers (36% versus 18%, respectively, Po0.001, Figures 2b–d). Overall, these results suggest that absence of CC1 in the liver significantly reduced MC38 tumor cell arrest and/or their proliferative capacity. We then questioned whether angiogenesis occurred efficiently in these Cc1/ metastatic lesions by quantifying CD31 immunostaining measuring blood vessel

density in metastatic WT and Cc1/ liver sections. CD31-positive staining was increased by 60% in Cc1/ versus WT metastatic lesions (Figures 2e–g). In contrast, however, two additional markers of angiogenesis, alphasmooth muscle actin and von Willebrand factor (VWF), were significantly reduced in Cc1/ metastatic nodules (Supplementary Figures S2-A and -B). Both VWF and alpha-smooth muscle actin are angiogenesis regulators and representative of mature and functional blood vessels (Skalli et al., 1989; Starke et al., 2011). These results suggest that increased vascular density of Cc1/ metastatic livers is likely uncoupled from vessel maturation. Overall, these results suggest that fewer metastatic cells reach the CC1-null liver and those that do display reduced proliferation. However, once established, the Oncogene


852

Figure 2 Ablation of CC1 from mouse tissues influences metastatic cell survival, proliferation and vascular density. (a) 2 105 CSFElabeled MC38 were injected intrasplenically in WT or Cc1/ mice and enumerated using intravital microscopy at times 0.5, 24 and 48 h postinoculation. A significant difference in the average number of cells/field between WT and CC1-null mice was computed 30 min postinoculation, with Cc1/ mice exhibiting a 2.2-fold decrease in hepatic sinusoidal arrest (Po0.05). After 48 h, only 6% of MC38 cells were present in the Cc1/ livers compared with 18% in the WT livers (Po0.05). Once implanted in the liver, the percentage of proliferating Ki-67-stained MC38 cells in Cc1/ livers was approximately two-fold less than those in WT livers (b, c, d, Po0.001). Liver sections of (e) WT and (f) Cc1/ mice were immunostained for the endothelial CD31 marker to compute blood vessel density in either small, medium or large size lesions. (g) Metastatic lesions in Cc1/ mice demonstrated a 60% increase of CD31-positive staining relative to WT mice (Po0.001).

Cc1/ metastatic lesions display enhanced vascular density, albeit with less vessel maturation. Immune cell infiltration and secretion of inflammatory cytokines/chemokines in the metastatic livers are deregulated in the absence of CC1 Microenvironmental differences between WT and Cc1/ mice rendered liver implantation of CRC cells less efficient in CC1-deficient mice. We then evaluated the impact of cell infiltrates in the liver metastatic niche, derived from either BMDCs or endothelial progenitor cells (Kaplan et al., 2006). Infiltration of macrophages (F4/80 þ ), granulocytic-derived cells (Gr1 þ ), dendritic and NK T myeloid-derived leukocytes (CD11b þ ), MDSC (CD11b þ Gr1 þ ), B (B220 þ ) and T lymphocytes (CD3 þ ) was tested 14 days postinoculation of MC38 cells. Myeloid or lymphoid-derived mature immune cells were significantly reduced in Cc1/ metastatic lesions (Figures 3b, d, f, h, j, l, m and n) compared with those of WT mice (Figures 3a, c, e, g, i, k, m and n). To evaluate the role of inflammation in metastatic livers, we analyzed a panel of 12 cytokines/chemokines (GM-CSF, IL-1a, IL-1b, IL-6, IL-10, IL-12p70, IL-17, IFNg, CCL2 (MCP-1), CCL3 (MIP-1a), CCL5 (RANTES) and TNF-a) in WT and Cc1/ liver lysates Oncogene

(n ¼ 6) before and 14 days postinoculation of MC38 CRC cells. No significant differences were observed in any of the tested cytokines/chemokines in WT and CC1null non-metastatic livers (data not shown). In contrast, only IL-6, IL-10 and TNF-a were significantly increased (Figure 3o) whereas CCL2, CCL3 and CCL5 were reduced in Cc1/ metastatic liver lysates compared with WT controls (Figure 3p). Therefore, macrophage and monocytic infiltration and BMDC recruitment normally driven by CCL2, CCL3 and CCL5 chemokines are significantly reduced in CC1-deficient animals, suggesting that CC1 expression in immune infiltrates specifically regulates chemokine secretion upon metastatic induction. Contribution of CC1 BMDCs to CRC metastasis We then performed BMT assays whereby Cc1/ or WT BM was transferred into irradiated converse recipient animals (Figures 4a and b). Chimeric mouse immune system underwent full reconstitution as demonstrated by fluorescent activated cell sorting analysis of peripheral blood for CD4 and B220 lymphocytes as well as CC1 expression 8 weeks posttransplantation (Supplementary Figure S3). Metastatic nodules of MC38 intrasplenically injected chimeric mice were scored 14 days postinjection


853

Figure 3 Immune cell infiltrates are decreased and cytokine secretion deregulated in the absence of CC1 expression. MC38-derived metastatic lesions obtained 14 days postinoculation in both WT and Cc1/ genotypes were tested for presence of macrophages (a, b, F4/80 þ ), of granulocytic-derived cells (c, d, Gr1 þ ), of dendritic and NK T myeloid-derived leukocytes (e, f, CD11b þ ), of MDSCs (g, h, CD11b þ Gr1 þ double-positive), of B (i, j, B220 þ ) and T lymphocytes (k, l, CD3 þ ). In every case, myeloid or lymphoid-derived mature immune cells were significantly reduced in Cc1/ metastatic lesions compared with those of WT mice (m, n). Metastatic livers were also evaluated for cytokine secretion. IL-6, IL-10 and TNF-a were significantly increased (o) whereas CCL2, CCL3 and CCL5 were significantly reduced in Cc1/ metastatic liver lysates as compared with their WT controls (p). *Po0.05; **Po0.001; ***Po0.0001.

(Figure 4a). Inhibition of metastatic burden was confirmed in a Cc1/ to Cc1/ transfer relative to WT to WT chimeras (Figure 4b), as originally demonstrated (Figures 1i and j). Transplanting WT BM to

Cc1/ mice rescued metastatic development (Figure 4b) indicating that BMDCs and their precursors regulate metastasis development in CC1-null mice. The reverse combination (Cc1/ BM transferred to WT mice) Oncogene


854

showed a significant reduction of metastatic burden relative to the converse transplant (Figure 4b) but the decreased metastasis scores were not as profound as those observed with the Cc1/ BM transferred to Cc1/ mice (Figure 4b). This suggests potential contribution of liver-resident cell types other than merely BMDCs during metastasis formation in the CC1 þ background. Enumeration of F4/80 þ , B220 þ , CD3 þ , Gr1 þ , CD11b þ and double positive CD11b þ Gr1 þ CFU-GM MDSCs on metastatic liver sections narrowed down the

Oncogene

immune cell types inhibiting CRC metastasis in Cc1/ mice (Figure 4c). Significant differences in all markers for Cc1/ to Cc1/ transfers were computed relative to WT to WT chimeras (black bars versus white bars in Figure 4c). However, transplanting Cc1/ BM to WT mice did not result in statistically significant differences in F4/80 þ or B220 þ and CD3 þ cells present in metastatic livers, but showed a trend toward diminished Gr1 þ cells, and an important decrease in CD11b þ and CD11b þ Gr1 þ cells compared with chimeric WT BM transfer into Cc1/ mice (Figure 4c).


855

We hypothesized that CC1-deficient MDSCs might be dysfunctional in Cc1/ liver metastatic growth and thus, WT and Cc1/ purified MDSCs were adoptively transferred to the Cc1/ mice 24 h postinjection of MC38 metastatic cells. This transfer led to enhanced metastasis in the Cc1/ mice receiving WT MDSCs as opposed to the significantly lower number of metastatic nodules in the Cc1/ mice accepting Cc1/ MDSCs, indistinguishable from that in Cc1/ mice without adoptive transfer (Figure 4d). This suggests that MDSCs (Gabrilovich and Nagaraj, 2009) are either less abundantly produced or sustain a migratory defect in CC1-deficient mice. In addition, the chimeric CC1positive hepatic environment (constituted of hepatocytes, resident macrophages (Kupffer cells) as well as stellate and endothelial cells) has a significant role in the recruitment and/or differentiation of the CD11b þ Gr1 þ cells during metastasis. BMDCs proliferation is controlled by CC1 during metastasis Horst et al. (2009) have shown that a BM CD11bhigh/ Ly-6Chigh subpopulation is reduced in naı¨ ve Cc1/ mice compared with their WT counterparts. In contrast, common myeloid precursors and megakaryocyte– erythrocyte precursors are CC1-negative whereas granulocyte–monocyte precursors expressed significant CC1 levels (Pan and Shively, 2010). To examine whether CC1 deletion affected the development, maturation and differentiation of progenitor cells, we performed colonyforming assays on both pre- and post-metastatic WT and Cc1/ BM and measured total CFCs and granulocyte–monocyte CFU precursor cells by morphological characteristics. No differences in numbers of total CFCs were noticed between the WT and CC1deficient mice, but significant increases of total CFCs and a trend towards increased CFU-GM numbers were quantified in metastatic Cc1/ mice versus those without liver metastases (Figures 5a and b). Given the reduced infiltration of MDSCs into the metastatic livers of Cc1/ mice (Figures 3g and h), we measured production of BM MDSCs and monocytic and granulocytic subsets by flow cytometry. We found a

significant reduction of total MDSCs (CD11b þ Gr1 þ ), monocytic (CD11b þ Ly6ChiLy6G) and granulocytic (CD11b þ Ly6CloLy6G þ ) lineages in naı¨ ve Cc1/ BM (Figures 5c, d and e), but an enhanced production of granulocytic MDSCs by Cc1/ BM under metastatic condition (Figure 5e), consistent with the fact that CC1 regulates basal and emergent granulopoiesis and neutrophilia (CFU-G) upon treatment with G-CSF (Pan and Shively, 2010). Thus, although CC1 deletion results in elevated levels of BM populations upon metastatic challenge, it appears that BMDC mobilization into metastatic liver is suppressed in absence of CC1 (Figure 3). Signaling in the Cc1/ liver metastatic environment To define which signaling pathways might be altered under metastatic conditions, WT or Cc1/ metastatic livers were retrieved 14 days postinjections and proteins immunoblotted with antibodies to activated phosphorylated proteins or unphosphorylated controls. Activated STAT1 and STAT3, but not STAT6, were significantly reduced in Cc1/ metastatic liver extracts relative to their WT counterparts (Figures 6a and b). No significant differences were noticed in NfkB activation (Figures 6a and b) and only a slight increase in pErk1/2 activity was revealed in Cc1/ versus WT metastatic liver extracts. Activated Akt in Cc1/ metastatic liver was significantly elevated relative to WT (Figures 6a and b), consistent with CC1’s role in mediating apoptosis (Kirshner et al., 2003; Nittka et al., 2004). We also examined signaling under BM transplant conditions by immunoblotting for STAT1, STAT3 and Akt in metastatic BM chimeric lysates. Changes in STAT3 and Akt activities (but not STAT1) in Cc1/ to Cc1/ transfer relative to WT to WT chimeras recapitulated those in non-chimeric Cc1/ versus WT metastatic livers (Supplementary Figure S4-A). No increased Akt activation and decreased STAT3 and STAT1 activities in Cc1/ to WT chimeras as opposed to the converse transfer were noticed (Supplementary Figure S4-A). In purified CC1-deficient MDSCs, no significant increased Akt activation was observed (Supplementary Figure S4B), and phospho-STAT levels were below detection levels

Figure 4 Contribution of CC1 BDMCs to CRC liver metastasis. (a) BMT assays were performed with WT or Cc1/ BM transferred to irradiated recipient animals of converse genotype. Transfer was also performed to mice of the same genotype serving as controls. Full reconstitution of the chimeric mouse immune system was monitored by fluorescent activated cell sorting analysis of peripheral blood samples for CD4 and B220 lymphocytes and CC1 expression 8 weeks posttransplant. Chimeric mice were then injected intrasplenically with MC38 cells and killed 14 days later. (b) Liver sections were processed for hematoxylin and eosin staining and the numbers of metastatic nodules scored and plotted as area fractions. Inhibition of metastatic burden was confirmed in a Cc1/ to Cc1/ transfer relative to WT to WT chimeras. Transplanting WT BM to Cc1/ mice rescued metastatic development. The reverse combination (Cc1/ BM transferred to WT mice) showed a significant reduction of metastatic burden relative to the converse transplant. (c) Macrophages (F4/80 þ ), B (B220 þ ) and T lymphocytes (CD3 þ ), granulocytic-derived cells (Gr1 þ ), dendritic and NK T myeloid-derived leukocytes (CD11b þ ) and double-positive CFU-GM (CD11b þ Gr1 þ ) cells on liver sections postmetastatic challenge were quantified. There were significant differences in all markers for Cc1/ to Cc1/ transfers relative to WT to WT chimeras (black bars versus white bars in c). Transplanting Cc1/ BM into WT mice did not result in statistically significant differences in mature macrophages (F4/80 þ ) or lymphocytes (B220 þ , CD3 þ ) present in metastatic livers, but showed a trend toward diminished Gr1 þ cells, and an important decrease in CD11b þ and CD11b þ Gr1 þ cells compared with those revealed on liver sections of chimeric WT BM transfer into Cc1/ mice. (d) CD11b þ Gr1 þ MDSCs were isolated and enriched from metastatic livers of WT and Cc1/ mice, then intravenously injected into two groups of recipient Cc1/ mice 24 h after intrasplenic injection of MC38 cells. Cc1/ mice that did not receive MDSCs but phosphate-buffered saline, served as ‘no transfer’ controls. Evaluation of metastatic livers 14 days after the transfers revealed an important increase in the number of nodules formed in Cc1/ mice receiving WT MDSCs. Metastasis formation in Cc1/ mice receiving Cc1/ MDSCs was indistinguishable from that in ‘no transfer’ Cc1/ mice. ns, non significant; *Po0.05; **Po0.001; ***Po0.0001. Oncogene


856

Figure 5 Bone marrow proliferation is regulated by CC1. (a) Pre- and postmetastasic WT and Cc1/ BM (n ¼ 4) was extracted, incubated for 14 days in a methylcellulose-based medium and total CFCs as well as granulocyte–monocyte CFU precursor cells enumerated using morphological characteristics. No differences in numbers of total pre-metastatic CFCs were noticed between the WT and CC1-deficient mice. However, a significant increase of total CFCs was demonstrated under metastatic conditions in Cc1/ mice relative to their naive Cc1/ mice. (b) A trend in increased Cc1/ CFU-GM numbers in metastatic conditions relative to nonmetastatic conditions was noticed. (c–e) Flow cytometry on pre- and postmetastatic WT and Cc1/ BM exhibited a significant decrease in the number of total MDSCs (CD11b þ Gr1 þ ), monocytic MDSCs (CD11b þ Ly6ChiLy6G) and granulocytic MDSCs (CD11b þ Ly6CloLy6G þ ) in Cc1/ BM. However this reduction was lost upon metastatic challenge because of significantly enhanced production of granulocytic MDSCs (CD11b þ Ly6CloLy6G þ ) by Cc1/ BM. ns, non significant; *Po0.05; **Po0.001.

Figure 6 Decreases in activated STAT1 and STAT3 in CC1-deficient metastatic livers. MC38 metastatic livers from either WT or Cc1/ mice were retrieved 14 days postmetastatic induction and protein lysates were prepared. (a) Activities of STAT1, 3 and 6, NfkB, Akt and Erk1/2 proteins were evaluated by immunoblotting with anti-phospho antibodies. (b) Positive signals were evaluated relative to total of those obtained with antibodies specific to the same proteins and plotted. Activities of STAT1 and STAT3 were significantly reduced whereas that of Akt was increased in Cc1/ metastatic livers compared with their WT counterparts. There were no differences in STAT6 activity and a slight increase in Erk1/2 activities in the same samples. ns, non significant; *Po0.05; **Po0.001. Oncogene


857

(data not shown). Overall, these data suggest contribution of other Cc1/ liver cell populations to the observed signaling alterations upon metastatic formation.

Discussion CC1 conveys a tumor suppressive role in the early phases of many different epithelially derived cancers. However, in the last few years, controversies regarding CC1’s role in tumor development have arisen in the literature (Obrink, 2008). In human CRC progression, tumor CC1 expression is generally downregulated in the early phases of CRC development (Neumaier et al., 1993; Rosenberg et al., 1993). Specific anti-human CC1 antibodies have now revealed low cell surfacetethered CC1 levels in low-grade dysplastic adenomas, with elevated expression in high-grade adenomas and adenocarcinomas with most patients with metastatic disease expressing significant CC1 levels (Kang et al., 2007; Song et al., 2011). In addition, multivariate analysis has highlighted the CC1-L dominance relative to other independent risk factors for CRC lymph node metastasis, hematogenous metastasis and short survival time, thereby confirming the role of CC1-L in CRC migration and invasion (Ieda et al., 2011). The roles of CC1 as an anti-tumorigenic molecule and its present demonstrated role as pro-metastatic in CRC metastasis depend on a number of CC1 functions such as its implication in proliferation, differentiation, angiogenesis, migration and invasion, the balance of its two major splice isoforms (CC1-S and CC1-L), its dimerization state as well as its interacting partners in a given cell type (Gray-Owen and Blumberg, 2006; Obrink, 2008). We first showed that systemic CC1 abrogation significantly reduced liver CRC metastatic burden and melanoma lung metastasis using experimental models bypassing the intravasation step, thus examining later metastatic steps. This phenotype argues for a significant contribution of the host metastatic microenvironment. Second, a more complicated network of cells (e.g., immune and endothelial cells) is at play during the metastatic cascade other than merely tumor cells of epithelial origin. We have examined how endothelial, liver and immune cell CC1 expression modulates CRC metastasis and shown that, in this context, CC1 exhibits pro-metastatic functions rather than anti-metastatic ones. We showed that endothelial CC1 expression favors tumor cell arrest and early survival within the tumor vasculature, thereby contributing to enhanced metastasis. Once tumor cells have implanted into the liver, CC1-positive liver cells (whether hepatocytes, Kupffer cells or endothelial and immune cells) contribute to the enhanced tumor proliferation. Counterintuitively, however, these liver metastatic nodules show increased CD31 positivity and vascular density but diminished alpha-smooth muscle actin and VWF staining and vessel maturation, as previously noted in VWFdeficient mice (Starke et al., 2011) and CC1-null mice (Nouvion et al., 2010). Finally, the metastatic stromal

compartment and in particular, CC1-proficient immune MDSC cells enhance the metastatic process. This effect is compounded by the acute reduction in immune infiltration within the liver and the consequent deregulated CCL2, CCL3 and CCL5 chemokine secretion, all crucial for myeloid-derived BMDC mobilization to the periphery (Niwa et al., 2001; Yoshidome et al., 2009). Importantly, CCR2-deficient mice lacking the CCL2 receptor or WT mice treated with CCL2 antagonists showed a lowered tumor burden and less macrophage infiltration in a colitis-associated CRC model (Popivanova et al., 2009). CCR2-expressing Gr1 þ inflammatory monocytes and breast metastasis-associated macrophage mobilization are also dependent upon tumor and stromal CCL2 expression (Qian et al., 2011). As well, human CCL2 blockade with specific neutralizing antibodies in combination with docetaxel reduced prostate tumor development and metastatic burden to initial levels (Loberg et al., 2007). Immune infiltrate reduction is even more striking considering that metastatic Cc1/ mice produced elevated total BM CFCs, total MDSCs (CD11b þ Gr1 þ ) and elevated granulocytic (CD11b þ Ly6CloLy6G þ ) subpopulation. Either these uncontrolled Cc1/ progenitor cells cannot undergo selective migration to the periphery or fail to fully differentiate into monocytes and macrophages within the liver. Further experiments to clarify these possibilities are underway. CD11b þ Gr1 þ MDSCs are one of the main metastatic regulators in Cc1/ mice. Indeed, WT BM transfer into CC1-deficient recipients completely rescued CRC metastatic development, whereas Cc1/ BM transplanted into WT recipients showed reduced metastatic expansion. Two potential mechanisms might dictate CEACAM1-mediated metastatic development by MDSCs and these will need to be further addressed. Ablation of STAT3 or use of STAT3 inhibitors contribute to reduced expansion of MDSCs and increased T-cell responses in mouse tumor models (Gabrilovich and Nagaraj, 2009) and activated STAT1 and inducible nitric oxide synthase are responsible for monocyte differentiation to macrophages (Movahedi et al., 2008). Another inferred mechanism is that failure to recruit SHP-1 to PIR-B, a close cousin of CC1, caused activation of STAT1 and NfkB in MDSCs, thus polarizing them towards the M1 macrophages versus the M2 phenotype upon tumor development and metastasis (Ma et al., 2011). If also demonstrated in Cc1/ mice, this would implicate M1 macrophage accumulation in the periphery with decreased CRC metastatic load. CC1-L acts as an inhibitory coreceptor in immune cells of both lymphoid and myeloid origins (Chen et al., 2001, 2011; Nagaishi et al., 2006; Pan and Shively, 2010). The data herein connects the very high expression of CC1 in host CD11b þ Gr1 þ myeloid cells to its newly assigned role as a pro-metastatic molecule. In fact, CC1 regulates MDSC immune-suppressive and pro-metastatic functions, thereby favoring metastatic growth of colon tumor cells in the hepatic niche. Further elucidation of CC1-mediated mechanisms in CRC metastasis may thus reveal novel therapeutic avenues targeting CC1-expressing MDSCs. Oncogene


858 Materials and methods Cell lines and cell culture Metastatic MC38 mouse colon cancer cells were a gift from Dr Shoshana Yakar (Wu et al., 2010). A CC1-negative population of MC38 cells was sorted using fluorescent activated cell sorting. MC38 cells and B16F10 melanoma cells (ATCC, Manassas, VA, USA) were grown in D-MEM medium (InVitrogen, Carlsbad, CA, USA) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone, Logan, UT, USA), 100 U/ml penicillin and 10 mg/ml streptomycin (Multicell, Woonsocket, RI, USA). Mouse experiments and metastasis induction The generation of Cc1/ mice has been described (Leung et al., 2006). Experimental procedures were conducted in compliance with Canadian Committee on Animal Care guidelines. All mice were used at 8–12 weeks of age. Liver metastases were produced by intrasplenic injection of 2 105 viable MC38 cells in 50 ml of phosphate-buffered saline, followed by splenectomy 3 min after injections. Mice were killed 14–20 days post cell injection and liver tissue excised and processed. phosphate-buffered saline-injected mice served as controls. lung metastases were produced by tail vein injection of 100 ml (1 105) viable B16F10 melanoma cells. Mice were euthanized and lungs removed 14 days postinjection for metastatic quantification by counting darkly stained melanincontaining nodules. Histology and immunofluorescence A 4-mm thick formalin-fixed liver sections (four step sections/ mouse, separated by 200 mm) were hematoxylin- and eosinstained and scanned using a ScanScope XT digital scanner (Aperio Technologies, Vista, CA, USA). Numbers and average size of metastatic lesions were computed with ImageScope software and reported as either absolute number or area fraction (ratio of surface nodule area/total surface area). For immunofluorescent detection, frozen sections of the same liver tissues were stained with primary antibodies (Ki67 (1:50, Abcam, Cambridge, MA, USA), CD31 (1:400, BD Pharmingen, San Diego, CA, USA), alpha-smooth muscle actin (1:100, Millipore, Billerica, MA, USA), VWF (1:100, Sigma, Montreal, QC, Canada), F4/80 (1:100, eBioscience, San Diego, CA, USA), Gr1 (1:100, R&D Systems, Minneapolis, MN, USA), CD11b (1:100, eBioscience), B220 (1:100, eBioscience), CD3 (1:100, Abcam) and CD4 (1:100, eBioscience), followed by incubation with FITC- or PE-conjugated antibodies, and with DAPI-containing mounting medium (DAKO, Glostrup, Denmark). Images were captured using a 20 objective of a Zeiss AxioSkop fluorescence microscope. Enumeration of five microscopic fields/liver sections is reported (percentage positive cells/DAPI-stained cells). CD31 positivity (four mice/ genotype) was quantified using Image J software (NIH, Bethesda, MD, USA). Fluorescence intravital microscopy WT and Cc1/ mice were injected intrasplenically with 50 ml of 2 105 of carboxyfluorescein succinimidyl ester (CSFE, InVitrogen)-stained MC38 cells for hepatic survival assays. The animals were anesthetized after 0.5, 24 or 48 h and the abdomen opened with midline and subcostal incisions. Mice were placed in a left supine position on a Plexiglas stage and the liver left lobe was positioned on a glass cover slip over a 20 inverted fluorescent microscope objective (Nikon TE300, Nikon, Montreal, QC, Canada). Blood flow to the liver sinusoids was monitored. CSFE-labeled cells were visualized Oncogene

along the edges of the exposed liver using epi-fluorescence and five microscope fields/animal were counted. Cytokine/chemokine immunoassays Liver samples were lysed in a 100 mM TrisCl pH 8.0 buffer containing 0.1 mM EDTA, 150 mM NaCl, 1% NP-40, 20 mM NaF, 1 mM Na3VO4, 1 mM NaH2PO4 and complete EDTA-free Protease Inhibitor cocktail (Roche, Montreal, QC, Canada). Protein concentrations were determined (Bio-Rad Laboratories, Hercules, CA, USA). Equal amounts of liver lysate proteins (three independent experiments, four mice/genotype/ experiment) were subjected to Quansys 12 cytokine/chemokine multiplex ELISA arrays (Quansys Biosciences, Logan, UT, USA). The cytokines/chemokines tested were: IL-1a, IL-1b, IL-6, IL-10, IL-12p70, IL-17, CCL2 (MCP-1), IFNg, TNFa, CCL3 (MIP-1a), GM-CSF and CCL5 (RANTES). Western blot analyses Liver extracts and cell pellets from isolated and enriched MDSCs were lysed in the same buffer described above. Proteins were separated on SDS–PAGE gels, transferred to PVDF membranes and immunoblotted with the following antibodies: phospho and total STAT1, STAT3, STAT6, NFkB p65, ERK, AKT (each at 1:1000; ERK at 1:2000; phosphoAKT at 1:500; total AKT at 1:2000, Cell Signaling, Danvers, MA, USA or Millipore), CC1 (1:1000), and a-Actin (1:5000, BD Biosciences, San Diego, CA, USA). The secondary HRP anti-rabbit or -mouse antibodies (1:3000, GE Healthcare, Waukesh, WI, USA) were detected using the Western Lightning Plus-ECL kit (Perkin-Elmer, Waltham, MA, USA). Colony forming assays and flow cytometry 8–10 week-old mice were euthanized and their femurs and tibias were harvested and crushed to extract BM. BM single cell suspensions were prepared and quantified. 1 105 BM cells (in 35-mm dishes) were cultured in a methylcellulose-based medium (MethoCult, M3434) according to manufacturer’s instructions (StemCell Technologies, Vancouver, BC, Canada) for 14 days in a humidified incubator at 37 1C with 5% CO2. Total number of colonies (total CFCs) and colony-forming unit-granulocyte macrophage (CFU-GM) were evaluated by morphology from four mice/genotype. In all, 106 isolated BMDCs were stained with fluorescently-conjugated antibodies to CD11b (BD Pharmingen), Gr1, Ly6C and Ly6G (BioLegend, San Diego, CA, USA) with data acquisition performed using the BD LSRII flow cytometer and the FACSDiva (version 6.0) software (San Jose, CA, USA). Data analysis was performed using the FlowJo software (Ashland, OR, USA). BMT assays BM was extracted from donor mice as above. Recipient mice (6–8 weeks old) were lethally irradiated (1000 rads) and reconstituted the following day by tail vein injection of 1 107 BM cells (in 0.2 ml aliquots) isolated from donor mice. Recipient mice received antibiotic treatment (Sulfamethoxazole and Trimethoprim, Roche) for 2 weeks after BMT. Full reconstitution was verified by fluorescent activated cell sorting analysis of peripheral blood using B220 and CD4 Abs 8 weeks after BMT. Correct status of chimeras was ascertained using an anti-CC1 antibody. Adoptive transfer of MDSCs from metastatic livers MDSCs were purified from metastatic livers of WT or Cc1/ mice. Briefly, livers were cut into small pieces and digested in a solution containing RPMI 1640 (InVitrogen), Liberase (1:250;


859 Roche) and DNaseI (0.05%; Roche) at 37 1C for 30 min. Cells were strained through a 70-mM cell strainer, spun at 1200 r.p.m. (4 1C) for 10 min and red blood cells were lysed using hypotonic NaCl solution. Liver leukocytes were collected from a 35% Percoll (Sigma) gradient centrifugation and resuspended in phosphate-buffered saline/2% fetal bovine serum. MDSCs were enriched using Miltenyi beads (Miltenyi Biotec, Auburn, CA, USA) on an autoMACS separator, according to the manufacturer’s instructions. In all, 5 105 enriched MDSCs in 100 ml phosphate-buffered saline were injected intravenously into two groups of Cc1/ mice (Cc1/ mice receiving either WT or Cc1/ MDSCs) that had undergone intrasplenic injection of MC38 cells 24 h before MDSC injection. Mice were euthanized and livers removed 16 days postinjection for metastasis quantification. Statistical analysis Data were expressed as mean±s.e. Statistical analysis was performed using GraphPad Prism 5 statistical software (La Jolla, CA, USA) for Microsoft Windows. The Student t test was used to determine significance and P-values of o0.05 were considered significant.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements We wish to thank Drs Janusz Rak, Peter Siegel and Julie St-Pierre (McGill University) for their helpful corrections and comments on the manuscript. We are also indebted to Dr Russell Jones (McGill University) for help with bone marrow transplantations and Dr Kathryn V Holmes (University of Colorado) for the monoclonal CC1 antibody. This work was supported by grants from the Canadian Institutes for Health Research (NB, LF and PB) and the Canadian Cancer Society Research Institute (LF). AA and ALN were recipients of post-doctoral fellowships from the Fonds de la Recherche en Sante´ du Que´bec (FRSQ). AA and CC received post-doctoral awards from the McGill Integrated Cancer Research Training Program funded by the Canadian Institutes for Health Research and the Fonds de la Recherche en Sante´ du Que´bec (FRSQ).

References Bamberger AM, Riethdorf L, Nollau P, Naumann M, Erdmann I, Gotze J et al. (1998). Dysregulated expression of CD66a (BGP, C-CAM), an adhesion molecule of the CEA family, in endometrial cancer. Am J Pathol 152: 1401–1406. Beauchemin N, Draber P, Dveksler G, Gold P, Gray-Owen S, Grunert F et al. (1999). Redefined nomenclature for members of the carcinoembryonic antigen family. Exp Cell Res 252: 243–249. Busch C, Hanssen TA, Wagener C, Obrink B. (2002). Downregulation of CEACAM1 in human prostate cancer: correlation with loss of cell polarity, increased proliferation rate, and Gleason grade 3 to 4 transition. Hum Pathol 33: 290–298. Chen T, Zimmermann W, Parker J, Chen I, Maeda A, Bolland S. (2001). Biliary glycoprotein (BGPa, CD66a, CEACAM1) mediates inhibitory signals. J Leukoc Biol 70: 335–340. Chen Z, Chen L, Baker K, Olszak T, Zeissig S, Huang YH et al. (2011). CEACAM1 dampens anti-tumor immunity by downregulation of NKG2D ligand expression on tumor cells and NKG2D function on NK cells. J Exp Med 208: 2633–2640. Dango S, Sienel W, Schreiber M, Stremmel C, Kirschbaum A, Pantel K et al. (2008). Elevated expression of carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM-1) is associated with increased angiogenic potential in non-small-cell lung cancer. Lung Cancer 60: 426–433. Ebrahimnejad A, Streichert T, Nollau P, Horst AK, Wagener C, Bamberger AM et al. (2004). CEACAM1 enhances invasion and migration of melanocytic and melanoma cells. Am J Pathol 165: 1781–1787. Gabrilovich DI, Nagaraj S. (2009). Myeloid-derived suppressor cells as regulators of the immune system. Nat Rev Immunol 9: 162–174. Giavazzi R, Campbell DE, Jessup JM, Cleary K, Fidler IJ. (1986). Metastatic behavior of tumor cells isolated from primary and metastatic human colorectal carcinomas implanted into different sites in nude mice. Cancer Res 46: 1928–1933. Gout S, Huot J. (2008). Role of cancer microenvironment in metastasis: focus on colon cancer. Cancer Microenviron 1: 69–83. Gray-Owen SD, Blumberg RS. (2006). CEACAM1: contact-dependent control of immunity. Nat Rev Immunol 6: 433–446. Hanahan D, Weinberg RA. (2011). Hallmarks of cancer: the next generation. Cell 144: 646–674.

Hemmila E, Turbide C, Olson M, Jothy S, Holmes KV, Beauchemin N. (2004). Ceacam1a-/- mice are completely resistant to infection by murine coronavirus MHV-A59. J Virol 78: 10156–10165. Horst A, Bickert T, Brewig N, Schumacher U, Beauchemin N, Fleischer B et al. (2009). CEACAM1 expression on macrophage precursors affects early lymphangiogenesis in vivo. Blood 113: 6726–6736. Horst AK, Wulf DI, Dabelstein J, Schumacher U, Sander H, Turbide C et al. (2006). Carcinoembryonic antigen-related cell adhesion molecule 1 modulates vascular remodeling in vitro and in vivo. J Clin Investig 116: 1596–1605. Ieda J, Yokoyama S, Tamura K, Takifuji K, Hotta T, Matsuda K et al. (2011). Re-expression of CEACAM1 long cytoplasmic domain isoform is associated with invasion and migration of colorectal cancer. Int J Cancer 129: 1351–1361. Kang WY, Chen WT, Wu MT, Chai CY. (2007). Expression of CD66a and possible roles in colorectal adenoma and adenocarcinoma. Int J Colorectal Dis 22: 869–874. Kaplan RN, Rafii S, Lyden D. (2006). Preparing the ‘‘soil’’: the premetastatic niche. Cancer Res 66: 11089–11093. Kirshner J, Chen CJ, Liu P, Huang J, Shively JE. (2003). CEACAM14S, a cell-cell adhesion molecule, mediates apoptosis and reverts mammary carcinoma cells to a normal morphogenic phenotype in a 3D culture. Proc Natl Acad Sci USA 100: 521–526. Laurie NA, Comegys MM, Carreiro MP, Brown JF, Flanagan DL, Brilliant KE et al. (2005). Carcinoembryonic antigen-related cell adhesion molecule 1a-4L suppression of rat hepatocellular carcinomas. Cancer Res 65: 11010–11017. Leung N, Turbide C, Balachandra B, Marcus V, Beauchemin N. (2008). Intestinal tumor progression is promoted by decreased apoptosis and dysregulated Wnt signaling in Ceacam1(-/-) mice. Oncogene 27: 4943–4953. Leung N, Turbide C, Marcus V, Jothy S, Beauchemin N. (2006). Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) contributes to progression of colon tumors. Oncogene 25: 5527–5536. Liu W, Wei W, Winer D, Bamberger AM, Bamberger C, Wagener C et al. (2007). CEACAM1 impedes thyroid cancer growth but promotes invasiveness: a putative mechanism for early metastases. Oncogene 26: 2747–2758. Oncogene


860 Loberg RD, Ying C, Craig M, Day LL, Sargent E, Neeley C et al. (2007). Targeting CCL2 with systemic delivery of neutralizing antibodies induces prostate cancer tumor regression in vivo. Cancer Res 67: 9417–9424. Ma G, Pan PY, Eisenstein S, Divino CM, Lowell CA, Takai T et al. (2011). Paired-immunoglobulin-like receptor-B regulates the suppressive function and fate of myeloid-derived suppressor cells. Immunity 34: 385–395. Movahedi K, Guilliams M, Van den Bossche J, Van den Bergh R, Gysemans C, Beschin A et al. (2008). Identification of discrete tumor-induced myeloid-derived suppressor cell subpopulations with distinct T cell-suppressive activity. Blood 111: 4233–4244. Nagaishi T, Pao L, Lin SH, Iijima H, Kaser A, Qiao SW et al. (2006). SHP1 phosphatase-dependent T cell inhibition by CEACAM1 adhesion molecule isoforms. Immunity 25: 769–781. Neumaier M, Paululat S, Chan A, Matthaes P, Wagener C. (1993). Biliary glycoprotein, a potential human cell adhesion molecule, is down-regulated in colorectal carcinomas. Proc Natl Acad Sci USA 90: 10744–10748. Nittka S, Gunther J, Ebisch C, Erbersdobler A, Neumaier M. (2004). The human tumor suppressor CEACAM1 modulates apoptosis and is implicated in early colorectal tumorigenesis. Oncogene 23: 9306–9313. Niwa Y, Akamatsu H, Niwa H, Sumi H, Ozaki Y, Abe A. (2001). Correlation of tissue and plasma RANTES levels with disease course in patients with breast or cervical cancer. Clin Cancer Res 7: 285–289. Nollau P, Prall F, Helmchen U, Wagener C, Neumaier M. (1997). Dysregulation of carcinoembryonic antigen group members CGM2, CD66a (biliary glycoprotein), and nonspecific cross-reacting antigen in colorectal carcinomas. Comparative analysis by northern blot and in situ hybridization. Am J Pathol 151: 521–530. Nouvion AL, Oubaha M, LeBlanc S, Davis EC, Jastrow H, Kammerer R et al. (2010). CEACAM1: a key regulator of vascular permeability. J Cell Sci 123: 4221–4230. Obrink B. (2008). On the role of CEACAM1 in cancer. Lung Cancer 60: 309–312. Oliveira-Ferrer L, Tilki D, Ziegeler G, Hauschild J, Loges S, Irmak S et al. (2004). Dual role of carcinoembryonic antigen-related cell adhesion molecule 1 in angiogenesis and invasion of human urinary bladder cancer. Cancer Res 64: 8932–8938. Pan H, Shively JE. (2010). Carcinoembryonic antigen-related cell adhesion molecule-1 regulates granulopoiesis by inhibition of granulocyte colony-stimulating factor receptor. Immunity 33: 620–631.

Popivanova BK, Kostadinova FI, Furuichi K, Shamekh MM, Kondo T, Wada T et al. (2009). Blockade of a chemokine, CCL2, reduces chronic colitis-associated carcinogenesis in mice. Cancer Res 69: 7884–7892. Poy MN, Yang Y, Rezaei K, Fernstrom MA, Lee AD, Kido Y et al. (2002). CEACAM1 regulates insulin clearance in liver. Nat Genet 30: 270–276. Qian BZ, Li J, Zhang H, Kitamura T, Zhang J, Campion LR et al. (2011). CCL2 recruits inflammatory monocytes to facilitate breasttumour metastasis. Nature 475: 222–225. Rosenberg M, Ne´dellec P, Jothy S, Fleiszer D, Turbide C, Beauchemin N. (1993). The expression of mouse biliary glycoprotein, a carcinoembryonic antigen-related gene, is down-regulated in malignant mouse tissues. Cancer Res 53: 4938–4945. Serra S, Asa SL, Bamberger AM, Wagener C, Chetty R. (2009). CEACAM1 expression in pancreatic endocrine tumors. Appl Immunohistochem Mol Morphol 17: 286–293. Skalli O, Pelte MF, Peclet MC, Gabbiani G, Gugliotta P, Bussolati G et al. (1989). Alpha-smooth muscle actin, a differentiation marker of smooth muscle cells, is present in microfilamentous bundles of pericytes. J Histochem Cytochem 37: 315–321. Song JH, Cao Z, Yoon JH, Nam SW, Kim SY, Lee JY et al. (2011). Genetic alterations and expression pattern of CEACAM1 in colorectal adenomas and cancers. Pathol Oncol Res 17: 67–74. Starke RD, Ferraro F, Paschalaki KE, Dryden NH, McKinnon TA, Sutton RE et al. (2011). Endothelial von Willebrand factor regulates angiogenesis. Blood 117: 1071–1080. Wu Y, Brodt P, Sun H, Mejia W, Novosyadlyy R, Nunez N et al. (2010). Insulin-like growth factor-I regulates the liver microenvironment in obese mice and promotes liver metastasis. Cancer Res 70: 57–67. Xu E, Leung N, Dubois MJ, Charbonneau A, Streichert T, Elcheby M et al. (2009). Targeted disruption of Ceacam1 promotes dietinduced hepatic steatosis and insulin resistance. Endocrinol 150: 3503–3512. Yeatman TJ, Mao W, Karl RC. (1997). Biliary glycoprotein is overexpressed in human colon cancer cells with high metastatic potential. J Gastrointest Surg 1: 292–298. Yoshidome H, Kohno H, Shida T, Kimura F, Shimizu H, Ohtsuka M et al. (2009). Significance of monocyte chemoattractant protein-1 in angiogenesis and survival in colorectal liver metastases. Int J Oncol 34: 923–930. Zhou CJ, Liu B, Zhu KK, Zhang QH, Zhang TG, Xu WH et al. (2009). The different expression of carcinoembryonic antigenrelated cell adhesion molecule 1 (CEACAM1) and possible roles in gastric carcinomas. Pathol Res Pract 205: 483–489.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene