Involvement of Small Arfgap1 (SMAP1), a Novel Arf6-Specific

ORIGINAL ARTICLE Involvement of small ArfGAP1 (SMAP1), a novel Arf6-speciﬁc GTPase-activating protein, in microsatellite instability oncogenesis

F Sangar1,2, A-S Schreurs1,2, C Uman˜ a-Diaz1,2, A Clape´ ron1,2, C Desbois-Mouthon1,2, C Calmel1,2, O Mauger1,2, A Zaanan1,2, C Miquel3, J-F Fle´ jou1,2,4 and F Praz1,2

Small ArfGAP1 (stromal membrane-associated protein 1, SMAP1), a GTPase-activating protein specific for ADP-ribosylation factor 6 (Arf6), which is a small GTPase acting on membrane trafficking and actin remodeling, is frequently mutated in various tumors displaying microsatellite instability (MSI), notably in MSI colorectal cancers (CRC). Genotyping of 93 MSI CRCs (40 stage II, 32 stage III and 21 stage IV) allowed us to underscore that SMAP1 mutation frequency was inversely correlated with disease stage (P ¼ 0.01). Analysis of 46 cancer cell lines showed that SMAP1 mutations occurred only in MSI tumors, and consisted exclusively in short insertion or deletion in the coding 10-adenine repeat, generating a premature termination codon located downstream the ArfGAP domain. SMAP1 transcript levels were significant decreased (P ¼ 0.006), and truncated SMAP1 protein could not be detected in cells displaying biallelic SMAP1 mutations, owing to its sensitivity to proteasome degradation. To investigate the role of SMAP1 mutations, we used the SMAP1-null HCT116 cell line and we established three isogenic SMAP1-complemented clones. Cell proliferation was first assessed in vivo using subcutaneous xenografts into immunodeficient mice. Tumors developed in all animals regardless of the cell line injected, but tumor volumes were significantly smaller for both SMAP1-complemented clones compared with HCT116 (Po0.0001, at the time of killing). In vitro, SMAP1 mutations also increased cell clonogenicity (P ¼ 0.02–0.04), cell proliferation (P ¼ 0.008) by shortening the G2/M phase and decreased cell invasiveness (P ¼ 0.03–0.003). In keeping, SMAP1- complemented HCT116 gained several mesenchymal markers (Snail, Slug and vimentin) considered as a hallmark of epithelial-to- mesenchymal transition. These observations are reminiscent of some clinical characteristics of MSI CRCs, notably their larger size and lower rate of metastasis. Our observations suggest that SMAP1 loss-of-function mutations in MSI CRC may contribute to the emerging oncogenic pathway involving abnormal Arf6 regulation.

Oncogene (2014) 33, 2758–2767; doi:10.1038/onc.2013.211; published online 10 June 2013 Keywords: colorectal cancer; microsatellite instability; small ArfGAP1 (SMAP1); cell invasion; cell cycle; epithelial-to-mesenchymal transition

INTRODUCTION length,4 genes carrying a coding 10-adenine repeat (A10) may be Colorectal cancers (CRCs) displaying microsatellite instability (MSI) considered as real target genes, if mutations occur at a frequency represent 15% of all CRCs. MSI results from the inactivation of the exceeding 56% in MSI CRCs. mismatch repair system, which is dedicated to post replicative We have recently identified several novel putative MSI-driven correction of polymerase errors arising during replication.1,2 MSI target genes, including small ArfGAP1 (small ADP-ribosylation tumors accumulate numerous mutations especially in factor GTPase-activating protein, known as stromal membrane- microsatellites, because these simple sequence repeats are associated protein 1, SMAP1), which displayed mutations in 73% hotspots for indel mutations during replication.3,4 Coding cases of the 44 primary MSI CRCs studied,7 a mutation frequency microsatellites are most often mononucleotide repeats,5 in among the highest in MSI CRCs.4 The SMAP1 gene, located on which indel mutations cause a frameshift and a premature chromosome 6q13,8 encodes a 467-amino acid protein composed termination codon. Transcripts containing a premature of an ArfGAP domain, a C4 type zinc finger, a clathrin binding termination codon are generally degraded by the nonsense- domain and a CALM (clathrin assembly lymphoid myeloid mediated mRNA decay (NMD) system, but mRNA decay may be leukemia) protein binding domain.9 SMAP1 binds directly to the partially efficient, leading to the synthesis of a truncated protein.6 clathrin heavy chain, and regulates clathrin-dependent Among the numerous mutations arising in mismatch repair- endocytosis of transferrin receptor10 and E-cadherin.11 SMAP1 deficient tumors, driver mutations are positively selected during shows a high degree of homology with SMAP2, a recently MSI-driven oncogenesis, defining the so-called real target identified SMAP gene family member acting as an Arf1-specific genes.3,4 According to a statistical model based on regression ArfGAP in the clathrin-dependent early endosome to trans-Golgi analysis of mutation frequency in regard to the repeat tract network traffic,9,12 which harbors a coding A8 repeat. At the

1INSERM, UMR_S 938, Saint-Antoine Research Center, Paris, France; 2UPMC Univ Paris 06, UMR_S 938, Saint-Antoine Research Center, Paris, France; 3Department of Pathology, Sainte-Anne Hospital, University Paris Descartes, Paris, France and 4Department of Pathology, Saint-Antoine Hospital, AP-HP, Paris, France. Correspondence: Dr F Praz, Centre de Recherche Saint-Antoine, INSERM-UPMC UMR_S 938, Baˆtiment Kourilsky, 184 rue du Faubourg Saint-Antoine, 75571 Paris cedex 12, France. E-mail: [email protected] Received 27 August 2012; revised 16 April 2013; accepted 3 May 2013; published online 10 June 2013 SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2759 protein level, SMAP1 and SMAP2 share 63% overall similarities, correction) or differentiation grade (Fisher’s exact test). When and 90% amino acid identity for the ArfGAP domains. stratifying tumors according to their stage, no clinical character- SMAP1 is an ArfGAP that acts on Arf6,10 a small GTPase involved istics differed according to the presence of SMAP1 mutation in various cellular functions, including clathrin-dependent (Supplementary Table S1). SMAP1 mutation frequency was highest endocytosis, actin remodeling and cytokinesis.13 Arf6 is involved among the 40 stage II MSI CRCs analyzed (70%), decreasing to in cell invasiveness by acting on several molecular processes, 56% for the group of 32 stage III MSI CRCs and to 38% for the notably E-cadherin endocytosis,14 the first step of epithelial-to- group of 21 stage IV MSI CRCs (Figure 1a, Table 1), indicating that mesenchymal transition (EMT). Furthermore, Arf6–GTP has a role disease extension and SMAP1 mutation frequency are significantly in actin remodeling by acting on Rac1, a small GTPase involved in inversely correlated (P ¼ 0.01, w2 test for trend). By contrast, no actin polymerization.14 The amount of Arf6–GTP is correlated to mutation could be found in the SMAP2 coding A8 repeat among tumor aggressiveness of breast cancer15 and melanomas16, further the 53 primary MSI CRCs analyzed (Table 1). As shown by underlining that Arf6 is a key molecule in tumor progression. immunohistochemistry, SMAP1 is expressed in the cytoplasm of In this study, we report that SMAP1 mutations arise frequently in adjacent normal colonic mucosa (Figure 1b). Among the tumors human MSI tumors of various tissue origins, both in solid tumors displaying SMAP1 mutations, some showed a homogeneous and in lymphomas, exclusively targeting the A10 repeat, while no negative pattern, whereas others showed a heterogeneous SMAP1 mutation could be detected in the SMAP2 A8 tract. In MSI CRCs, expression pattern (Figure 1b); the fact that some areas of the SMAP1 mutation frequency was inversely related to tumor tumor- tumor displayed normal expression, while others were negative, node-metastasis stage, being highest in stage II CRCs that are indicate that SMAP1 mutations did not arise at an early stage of devoid of local or distant metastasis, suggesting that SMAP1 tumor development, but rather appeared during tumor progres- mutations may counteract tumor metastatic potential. Using sion in limited tumor areas. various cell lines, we show that MSI-driven SMAP1 frameshift mutations are loss-of-function mutations, as they led to decreased Characterization of SMAP1 and SMAP2 mutations in cancer cell levels of abnormal transcripts, and to subsequent degradation of lines residual truncated protein by the proteasome. Using a cellular Indel mutations arising in the SMAP1 A10 and SMAP2 A8 coding experimental system consisting of the SMAP1-null HCT116 cell line repeats were screened in 46 cell lines established from 25 MSI and and three isogenic SMAP1-complemented clones, we show that 21 MSS (microsatellite stable) tumors from various tissues SMAP1 inactivation increased clonogenic capacities and cell (Figures 2a and b). Cell lines established from MSI CRCs displayed proliferation, owing to shorter G2/M phase of the cell cycle, but the highest mutation frequency, with 11 of 14 cell lines having reduced cell invasiveness by hampering EMT. In keeping, restoring SMAP1 mutations, most often biallelic (8 cell lines) (Figure 2b). wild-type SMAP1 expression in HCT116 was associated with the SMAP1 A10 repeat was also frequently mutated in cell lines expression of several mesenchymal markers (Snail, Slug and derived from gastric and endometrial MSI carcinomas, with 3 of 4 vimentin). In vivo, subcutaneous xenografts into immunodeficient cell lines displaying biallelic mutations (Figures 2a and b). Three of mice demonstrated that the tumors were significantly larger upon the 7 MSI lymphomas analyzed harbored mutations in SMAP1 A10 the injection of HCT116 cells compared with SMAP1-complemen- repeat, but only monoallelic mutations could be detected ted clones. These observations are in agreement with some of the (Figures 2a and b). Deletions were more frequent than insertion, characteristic clinical features of MSI tumors, which are known to with only 3 cell lines displaying an A11 repeat corresponding to a be large-sized and poorly invasive tumors. 1 bp insertion, whereas 25 mutated alleles had lost 1 bp (24 cases) or 2 bp (1 case). None of the cell lines established from the MSS tumors displayed mutations in the SMAP1 A10 repeat (Figure 2a). RESULTS No mutation could be found in the SMAP2 A8 coding repeat in any Clinical features of patients with MSI CRCs according to SMAP1 or of the 46 cell lines analyzed (Figure 2b). SMAP2 mutations To investigate whether SMAP1 mutations were restricted to the There were no difference in the age and sex of patients (t-test and A10 repeat or could occur elsewhere, we sequenced the entire Fisher’s exact test, respectively), tumor location (w2 Yates coding exons and the exon/intron junctions in cell lines devoid of

P=0.01 * 80 * 70

20 Mutation frequency (%) 10

0 II III IV Normal Heterogeneous Homogeneous TNM stage colic mucosa staining negative staining SMAP1mut MSI CRC Figure 1. SMAP1 mutations in human MSI CRCs. (a) SMAP1 mutation frequency according to tumor stage. Stage II: in situ tumors (n ¼ 40); stage III: tumors with lymph node metastasis (n ¼ 32); stage IV: tumors with distant synchronous or metachronous metastasis (n ¼ 21). (b) SMAP1 expression pattern by immunohistochemistry in normal colic mucosa, in MSI CRCs harboring a heterogeneous pattern, and in MSI CRC harboring a homogeneous pattern. Upper images: Â 20; lower images: Â 100. *Po0.05.

& 2014 Macmillan Publishers Limited Oncogene (2014) 2758 – 2767 SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2760 devoid of SMAP1 mutations, despite these two cell lines carried Table 1. Clinical features of patients with an MSI CRC according to biallelic SMAP1 mutations, suggesting that the NMD system was SMAP1 or SMAP2 mutations inefficient (Figure 2c). SMAP1wt SMAP1mut SMAP2wt (n ¼ 39) (n ¼ 54) (n ¼ 53) SMAP1 mutations are loss-of-function mutations In order to determine whether residual levels of mutated No. of No. of No. of transcripts were translated into a truncated protein that might patients (%) patients (%) P-value patients (%) act as a dominant negative, we used a polyclonal antibody Age recognizing amino acid 101–115 of human SMAP1, within the wt mut Median 52.48 59.05 0.31 49.3 ArfGAP domain and present in SMAP1 and SMAP1 forms Range 31.1–88.2 24.2–90.6 26.7–82.1 (Figure 2e). In HeLa cells with SMAP1wt genotype, SMAP1 migrated o65 year 20 (51.3) 30 (55.6) 0.84 43 (81.1) at the expected size (55 kDa) (Figure 2f, left lane), whereas no mut Sex band could be detected at the size expected for the A9 SMAP1 Male 23 (60) 26 (48.1) 0.28 28 (52.8) (21 kDa) in the HCT116 cells (Figure 2f, right lane). In HeLa cells Female 13 (33.3) 25 (46.3) 26 (49.1) transfected with an expression vector of the truncated SMAP1-A9 Unknown 3 3 1 form, an additional band migrating at the expected size for the mut Tumor location truncated A9 SMAP1 was revealed, confirming that the Proximal 27 (69.2) 34 (63) 0.29 41 (7.4) antibody could recognize both the wild-type and truncated Distal 8 (20.5) 15 (27.8) 11 (20.8) SMAP1 forms (Figure 2f, middle lane). To determine whether Unknown 4 5 3 truncated SMAP1 was sensitive to proteasome degradation, we Differentiation grade transfected HCT116 cells with either pcDNA3.1–SMAP1wt-HA or Well/moderate 19 (48.7) 12 (22.2) 0.58 33 (62.3) pcDNA3.1–SMAP1-A9-HA, and treated with the proteasome Poor 12 (30.8) 11 (20.4) 21 (39.6) inhibitor MG132 or vehicle (Figure 2g). Expression of p53 Unknown 8 31 1 increased upon MG132 treatment, showing that proteasome Stage was efficiently inhibited (Figure 2g). MG132 treatment also II 12 28 17 increased SMAP1-A9-HA expression, but not that of wild-type III 14 18 19 SMAP1, indicating that truncated SMAP1 was sensitive to IV 13 8 17 proteasome degradation (Figure 2g). Yet, proteasome inhibition Abbreviation: SMAP1, stromal membrane-associated protein 1. Main in HCT116 did not allow cells to express endogenous truncated clinical features were compared between patients whose tumor carried SMAP1 at a detectable level (Figure 2g). Thus, because no SMAP1 mutations to those with wild-type SMAP1. No tumor displayed truncated SMAP1 protein could be detected, SMAP1 mutations are SMAP2 gene mutation in the A8 repeat among the 53 tumors analyzed. loss-of-function mutations. Differences in age distribution were assessed using Student’s t-test. Patients under 65 years, sex, differentiation were compared with the Fisher’s exact test. Differences in tumor location and stage were assessed SMAP1 expression controls in vivo tumor growth in a xenograft using the w2. Analyses of SMAP1 mutations frequency according to the tumor model tumor-node-metastasis stage of the tumor were performed using w2 test In order to investigate the consequences of SMAP1 inactivation on for trend comparing all data and a two-tailed t-test for pairwise oncogenesis, we established an experimental model based on the comparisons. SMAP1-null HCT116 cell line and HCT116–SMAP1-complemented isogenic clones obtained by stably transfecting the pcDNA3.1– SMAP1wt-HA expression vector; three clones that expressed SMAP1 at a level similar to that of HeLa cells (SMAP1wt) were biallelic mutations in the A10 repeat. Four cell lines, two MSS selected (Figure 3a). Evidence that SMAP1 was functionally active (HGT1, MKN28) and two MSI (SNU1, TC71), carried a previously was obtained by showing that the transferrin receptor endocy- described single-nucleotide polymorphism in exon 7 (c.635C4T, tosis, a well-known SMAP1-dependent activity, was altered in the rs2273566), resulting in Ala212Val change (Figure 2a). We found three HCT116–SMAP1 isogenic clones compared with HCT116 (see another single-nucleotide polymorphism located in the A10 Supplementary Figure S1 and data). repeat (c.515A4G, rs138685504), which replaces Lys by Arg at In order to investigate the impact of SMAP1 expression on codon 172 in three cell lines, one MSS (MKN28) and two MSI tumor progression, we performed a series of subcutaneous (LIM2405 and RKO) (Figure 2a). No SMAP1 mutation could be xenograft experiments in nude mice, using HCT116 and two found in the 21 MSS cell lines analyzed. isogenic HCT116–SMAP1 clones (C7 and C12). Tumors developed SMAP1 mRNA levels have been quantified in 13 MSI and 8 MSS in all the animals regardless of the cell line injected, indicating that cell lines, and expressed as a ratio using HeLa cells as a reference. SMAP1 did not influence the frequency of tumor takes. As shown As shown in Figure 2c, cell lines lacking SMAP1 mutations in Figures 3b and c, tumor volumes were significantly smaller for expressed SMAP1 mRNA at similar levels regardless of the MSI mice implanted with either C7 or C12 SMAP1-positive clone status, with mean levels of 128±15 vs 105±19 (mean±s.e.m.) for compared with mice injected with parental HCT116 cells, with MSS and MSI cell lines, respectively (P ¼ 0.50, Mann–Whitney two- highly significant P-values (Po0.0001 at the time of killing for both tailed test). As expected, SMAP1 transcript levels were significantly C7 and C12, two-way analysis of variance, ANOVA) (Figure 3b). lower in the MSI cell lines harboring SMAP1 mutations (40±10) These observations further underline that SMAP1 likely exerts a compared with MSS and MSI cell lines lacking SMAP1 mutation growth-inhibiting action on xenografted tumors in vivo. (P ¼ 0.006 and P ¼ 0.02, respectively, Mann–Whitney one-tailed test); the decrease in SMAP1 transcript levels was more SMAP1 mutations increase clonogenic capacity and cell pronounced for MSI cell lines displaying biallelic mutations proliferation by shortening the G2/M phase (31±13; Po0.001 and P ¼ 0.02, compared with MSS and MSI cell To better characterize the effects of SMAP1 on cell growth control, lines) than for those with monoallelic mutations (mean 63; we performed in vitro experiments (Figure 4). Restoring SMAP1 P ¼ 0.02 and P ¼ 0.10, compared with MSS and MSI cell lines, expression in HCT116 decreased cell proliferation of all the three Mann–Whitney one-tailed test) (Figure 2d). However, SMAP1 HCT116–SMAP1 clones to the same extent; the numbers of cells transcripts levels in LoVo and SW48 were comparable to cell lines recovered 48 h after seeding were significantly lower compared

Oncogene (2014) 2758 – 2767 & 2014 Macmillan Publishers Limited SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2761 MSS MSI Wild-type alleles Wild-type alleles A10 monoallelic mutation A10 biallelic mutation

ALA CRC HCT8 CRC LIM2405a CRC A9/wt CO115 CRC A9 CBS CRC HCT15 CRC RKOa CRC A9/wt HCT116 CRC A9 COLO320 CRC LS411 CRC TC71b CRC wt/A11 KM12 CRC A9 EB CRC SNU1b STO Jurkat LYMP A9/wt LIM1215 CRC A9 FET CRC CEM LYMP KCL22 LYMP A9/wt LoVo CRC A9/A11 FRI CRC EHRB LYMP REH LYMP A9/wt LS174T CRC A9 GLY CRC MOLT-4 LYMP SW48 CRC A8/A9 IS1 CRC MOLT-16 LYMP TC7 CRC A9 IS2 CRC HEC1A ENDO A9/A11 IS3 CRC ISHI ENDO A9/A9 LS1034 CRC RL95-2 ENDO A9/A9 LS513 CRC SW1116 CRC SMAP1 SMAP2 mutation mutation SW480 CRC GTL16 STO None Monoallelic Biallelic None HGT1b STO Human MSI cancer cell lines n (%) n (%) n (%) n (%) MKN1 STO MKN28a,b STO Colon, rectum 3 (21.4) 3 (21.4) 8 (57.2) 14 (100)

MKN7 STO Lymphoma4 (57.1) 3 (42.9) - 7 (100) TMK1 STO aExon 6: c.515A>G HeLa UT bExon 7: c.635C>T Stomach, endometrium 1 (25.0) - 3 (75.0) 4 (100)

213 200 P=0.006 P=0.02 200 150

100

149 50 144 150 139 140 0 wt wt mono bi 114 MSS MSI 100 98 100 93 77 85 83 74 73 58 45

Relative SMAP1 transcript expression (%) 50

24 11 11 7 5

HeLa ALA FET RKO LoVo TC7 HCT8 TC71 KM12 SW48 HCT15 LS411 SW480 LIM2405 HCT116 LS174T LLBCD14LLBCD16LLBCD37 CoLo320 LIM1215

MSS MSI

Wild-type SMAP1 (55k Da) pcDNA3.1-SMAP1 A10 wt A9 Ctrl C CALM HeLa ArfGAP MG132 - + - + - + B BD wt A9 HCT116 p53 55kDa antibody epitope SMAP1 55kDa (aa 101-115) 21kDa SMAP1 A9* 21kDa

ArfGAP GAPDH GAPDH

1bp del mutant SMAP1 (21 kDa) Figure 2. Characterization of SMAP1 and SMAP2 mutations arising in tumor-derived cell lines, and SMAP1 transcript and protein expression according to tumor SMAP1 mutations. (a) SMAP1 gene sequencing in cell lines derived from MSS or MSI tumors of various tissue origins. (b) Distribution of MSI tumor-derived cell lines from various tissue origins according to SMAP1 or SMAP2 mutation status, wild-type (no mutation), monoallelic or biallelic mutations. (c) Relative SMAP1 transcript expression in MSS and MSI tumor-derived cell lines. Hatched box: MSS cell lines derived from noncolorectal tumors; black box: MSS CRC-derived cell lines; closely punctuated box: MSI CRC-derived cell lines with wild-type SMAP1 allele only; largely punctuated box: MSI CRC-derived cell lines harboring a monoallelic SMAP1 mutation; white box: MSI CRC-derived cell lines harboring biallelic SMAP1 mutations. (d) Relationship between SMAP1 transcript level in tumor-derived cell lines according to SMAP1 mutation status using the Mann–Whitney test. wt: wild-type; mono: monoallelic mutation; bi: biallelic mutations. (e) Representation of wild-type and mutant SMAP1 proteins. The 1 bp deletion leads to the loss of both the clathrin box (CB) and the clathrin assembly lymphoid myeloid leukemia (CALM) binding domain (CALM BD). The epitope recognized by the antibody used in f and g is located within the ArfGAP domain (101–115 aa, amino acid). (f) SMAP1 expression was analyzed in HeLa cells (SMAP1wt) (left lane), in HeLa cells transfected with the expression vector of the HA-tagged truncated SMAP1 form corresponding to the 1 bp deletion mutation (SMAP1-A9-HA) (middle lane) and in the HCT116 cell line carrying a homozygous 1 bp deletion SMAP1 mutation (SMAP1-A9) (right lane). The membrane was revealed with an antibody directed against the SMAP1 peptide 101–115 aa (Sigma); the 55 and 21 kDa bands correspond to the endogeneous wild-type SMAP1 protein and the SMAP1-A9 truncated protein, respectively. Loading controls were obtained using an horseradish peroxidase (HRP)-conjugated anti-GAPDH antibody. (g) HCT116 cells were transfected with pcDNA3.1–SMAP1wt-HA (left), pcDNA3.1–SMAP1-A9-HA (middle) or pcDNA3.1 control vector (right lane) and treated during 8 h with MG132 (5 mM), a proteasome inhibitor or vehicle. The p53 protein used as a control for proteasome inhibition was detected with the DO-1 monoclonal mouse antibody. *Po0.05.

& 2014 Macmillan Publishers Limited Oncogene (2014) 2758 – 2767 SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2762 HCT116 C7 HCT116-SMAP1

HeLa HCT116C7 C12 C16 SMAP1

GAPDH

HCT116 HCT116 )

3 1500 C7 C12 C7 1000 **** *** 500 **** **** **** C12 Tumor Volume (mm * 0 10 15 20 25 30 Days Figure 3. Effect of SMAP1 expression on tumor growth. (a) Western blot analysis of SMAP1 expression in HeLa (SMAP1wt), parental HCT116 (SMAP1-null) and three clones derived from HCT116 cells by stably transfecting SMAP1; immunoﬂuorescence detection of SMAP1 is shown for HCT116 cells and a representative HCT116–SMAP1 clone. (b) After cell injection in nude mice, tumor volume was monitored twice a week to follow tumor growth according to SMAP1 expression. Results represent the mean tumor volumes (mm3±s.e.m.) from day 11 to day 29 postinjection. (c) Pictures of tumors after the mice were killed. Four tumors per group are represented. ***Po0.001; ****Po0.0001.

with the HCT116 parental cells (P ¼ 0.010, one-way ANOVA, b-catenin in HCT116–SMAP1 C12 cell cytoplasm (Figure 5d, Dunnett’s test) (Figure 4a). bottom panel) in contrast to parental HCT116 cells (Figure 5d, By performing a clonogenic assay, we observed that SMAP1 top panel). Furthermore, upon SMAP1 expression, levels of complementation not only decreased the proliferation of HCT116 E-cadherin and b-catenin were decreased in the HCT116–SMAP1 cells, but also altered their ability to grow when seeded at a low clones compared with HCT116 (Figure 5e). concentration (Figures 4b and c). The number of clones The gain of mesenchymal markers such as Snail, Slug and established from all the three HCT116–SMAP1 clones varied from vimentin, is a hallmark of EMT. A significant increase of Snail, Slug 28 to 41, which was significantly lower compared with the HCT116 and vimentin mRNA levels was observed in HCT116–SMAP1 parental cells that gave rise to 70 clones in average (P ¼ 0.02–0.04, clones as compared with HCT116 (Po0.001, except for vimentin t-test). Moreover, the size of colonies derived from HCT116– expression in C12, one-way ANOVA, Dunnett’s test) (Figure 5f). The SMAP1 clones was smaller (Figure 4c) in agreement with our ultimate step of the cells that undergo EMT is to acquire the ability results showing that restoring SMAP1 expression decreased cell to invade. HCT116–SMAP1 clones were able to invade signifi- proliferation (Figure 4a). cantly, more efficiently through a Matrigel layer than the parental Cell cycle analysis of cells synchronized with nocodazole HCT116 cell line (Figure 5g; P ¼ 0.03–0.003, one-way ANOVA, showed that the transition from G2/M to G1 phase was faster Dunnett’s test). for HCT116 cells compared with the HCT116–SMAP1 clones. After 4 h of nocodazole chase, 50% of SMAP1-complemented cells remained in the G2/M phase compared with 20% for HCT116 DISCUSSION (P ¼ 0.008) (Figure 4d), a difference that may explain the In MSI CRCs, only few mutated genes have been clearly proliferation defect observed for the HCT116–SMAP1 clones. demonstrated as having a role in oncogenesis, apart from the transforming growth factor receptor 217 and the activin receptor 218 that are both involved in the transforming growth factor b SMAP1 mutations alter adherent junctions, induce mesenchymal pathway. Recently, SMAP1 was found to be among the most markers and decrease cell invasiveness frequently mutated genes in MSI CRCs.7 In this study, we report HCT116, in which the expression of SMAP1 was restored, that loss-of-function SMAP1 mutations occur in MSI tumors of displayed a fibroblast-like phenotype and scattered as compared various tissue origins, including CRCs, endometrium and gastric with HCT116 parental cells that grew in cluster (Figure 5a and data carcinomas, as well as lymphomas. In a series of primary MSI CRCs, not shown for clone 7 and 16). Immunofluorescence experiments immunohistochemical analyses of SMAP1 allowed us to visualize showed that in HCT116, E-cadherin was concentrated at the that, in some tumors, SMAP1 protein expression was plasma membrane (Figure 5b, top panel). In contrast, in HCT116– heterogeneous, with some areas totally lacking detectable SMAP1 C12, E-cadherin was mainly located in the cell cytoplasm SMAP1, while others were normally labeled, indicating that (Figure 5b, bottom panel). Similar results were obtained for the SMAP1 mutations are likely to arise during tumor progression two other clones C7 and C16 (data not shown). Accordingly, rather than at the initiation step. In our series of CRC patients, surface expression of E-cadherin was significantly decreased SMAP1 mutation frequency was inversely related to the stage of (ranging from 30 to 70%) in HCT116–SMAP1 clones as attested the disease, suggesting that SMAP1 mutations may prevent tumor by flow cytometry analysis using an antibody that recognized the progression. Thus, it would be interesting to investigate in a larger extracellular domain of E-cadherin (Figure 5c). As for E-cadherin, series of patients with MSI CRC whether the risk for tumor immunofluorescence analyses indicated an internalization of recurrence is lower when tumors homogeneously lack SMAP1

Oncogene (2014) 2758 – 2767 & 2014 Macmillan Publishers Limited SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2763 800 90 HCT116 80 HCT116 ** C7 70 C7 600 C12 60 C12 50 C16 * C16 40 * * 400 30 20 10 200

Number of clones per well 0

Number of cells per well HCT116 C7 C12 C16 0 02448 Hours

HCT116 C7 C12 C16

HCT116 C7 100% 100%

75% 75% * * 50% 50% **

25% 25% * * 0% ** 0% * 01234 01234

C12 C16 100% 100%

75% 75% ** **

50% 50% **

25% 25% ** *** 0% 0% * ** 01234 01234 Time of chase (h) Figure 4. Cell proliferation and cell cycle analysis. (a) To evaluate proliferation, cells were seeded at 12 Â 103 per well and counted 24 or 48 h after seeding. Results represent the mean of three independent experiments±s.e.m. (b, c) For clonogenic assays, cells were seeded at 50 per well, and stained with toluidine blue at day 6 to count clones. Results represent the mean of three independent experiments±s.e.m. each performed in triplicate. (d) For cell cycle analysis, cells were treated with nocodazole during 16 h at 37 1C, then washed and incubated and ﬁxed at several time points. Cells were labeled with propidium iodide and analyzed by ﬂow cytometry. Results are means of three independent experiments and are expressed as percentage±s.e.m. (except for C12, for which only one experiment was performed). Black box: G0/G1 phase; gray box: S phase; white box: G2/M phase. *Po0.05; **Po0.01.

expression and whether SMAP1 is less frequently mutated in the mutation frequency for A8 repeats is around 10% in MSI CRCs and recurrences. SMAP2,aSMAP1 homolog that codes an ArfGAP over 30% to be considered as possibly selected in tumors, our speciﬁc for Arf1, acting in the retrograde, early endosome-to- observations suggest that SMAP2 mutations are likely to be neutral trans-Golgi network pathway in a clathrin- and AP-1-dependent in these tumors and may even be counterselected. Alternatively, manner,12 contains an A8 coding repeat that was devoid of SMAP1 and SMAP2 mutations may be functionally redundant, as mutations in all the tumors and cell lines analyzed. As the average they both target intracellular trafﬁc. On the basis of this

& 2014 Macmillan Publishers Limited Oncogene (2014) 2758 – 2767 SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2764 E-cadherin DAPI 120

100 ** 80 *** HCT116 HCT116 60

40 ***

20 Surface E-cadherin (%)

0 HCT116 C7 C12 C16 C12 7μm C12 18μm

β-catenin DAPI HCT116-SMAP1 HCT116 C7 C12 C16

E-cadherin HCT116 β-catenin

SMAP1

C12 18μm GAPDH

4 *** HCT116 *** C7 P = 0.003 *** 4000 C12 C16 3 P = 0.01 *** 3000

P = 0.03 *** 2000 2 *** ns *** Number of cells

Fold Increase 1000 *** 1 0 HCT116 C7 C12 C16

0 HCT116-SMAP1 SnailSlug Vimentin Figure 5. Effects of SMAP1 mutations on cell invasion. (a) HCT116 cells and the HCT116–SMAP1 C12 clone stably expressing SMAP1 were seeded on dish, and pictures were taken 48 h after seeding. Scale bar: 7 mm. (b, c) Cells were stained for E-cadherin and b-catenin. Cells were analyzed by confocal microscopy (b) and flow cytometry (c). Scale bar: 18 mm. Experiments were performed three times and results are represented as means±s.e.m. (d) Western blot of HCT116 and the three HCT116–SMAP1 clones (C7, C12, C16). SMAP1, b-catenin and E-cadherin were detected with specific antibodies and GAPDH protein was used as loading control, as described in Materials and methods. (e) Transcript expression of some EMT target genes in the HCT116 parental cell line and the three isogenic SMAP1-complemented clones. Expression of Snail and Slug transcription factors, and the mesenchymal marker, vimentin, are shown relative to their levels in HCT116. (f) Invasion assay. Cells were seeded on Matrigel, and 24 h after seeding invaded cells were counted. Results are represented as mean of at least five independent experiments±s.e.m. **Po0.01; ***Po0.001.

hypothesis, it is not surprising that mutations target deﬁcient in these cell lines. Indeed, several genes coding NMD predominantly SMAP1 as its repeat is longer, thus being more components harbor coding mononucleotide repeats that are prone to polymerase slippage during replication. prone to MSI-driven mutations, notably SMG7 has one (A)9 repeat Several cellular mechanisms exist to avoid the expression of and UPF3A has two such repeats. In view of this, it would be abnormal truncated proteins that may act in a dominant-negative interesting to evaluate the contribution of NMD defects in MSI manner. One is the NMD system, which degrades abnormal oncogenesis. transcripts harboring a premature termination codon.6,19 As In order to get further insights in the molecular mechanisms expected, SMAP1 mutations caused a signiﬁcant decrease in through which SMAP1 inactivation may drive colorectal SMAP1 transcript levels in cell lines that was more marked for oncogenesis, we compared the SMAP1-null HCT116 cell line to biallelic than for monoallelic mutations, with the exception of three isogenic clones expressing wild-type SMAP1. All HCT116– LoVo and SW48. The fact that SMAP1 transcript levels remained SMAP1 clones behaved similarly, indicating that the differences within the normal range in LoVo and SW48 cells despite the observed when compared with the parental HCT116 cell line were presence of biallelic mutations suggests that NMD may be most likely due to SMAP1. In vitro cell cultures and in vivo

Oncogene (2014) 2758 – 2767 & 2014 Macmillan Publishers Limited SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2765 subcutaneous xenograft experiments allowed us to show that the MATERIALS AND METHODS SMAP1-null HCT116 cell line grew significantly faster than its Patients isogenic clones, in which SMAP1 expression was restored. These This study included 93 patients who underwent surgical resection of observations are in line with of one of the characteristic clinical histologically proven MSI CRC and for whom clinical data and tumor tissue features of MSI CRCs, as they tend to be larger-sized tumors could be retrieved. The procedures followed in this study were in compared with MSS CRCs. Increased cell proliferation associated accordance with the ethical standards of the Institutional Ethics with SMAP1 mutations was due, as least in part, to a shorter G2/M Committee. Tumor type and grade of differentiation were determined phase. Interestingly, it was recently reported that the expression of according to the World Health Organization criteria. Forty patients had a an Arf6 mutant unable to undergo GTP hydrolysis led to tumor devoid of lymph node and distant metastasis (stage II), 32 had positive lymph nodes (stage III) and 21 had distant metastasis (stage IV). cytokinesis defects, suggesting that the hydrolysis of the Arf6– GTP is essential for completion of cytokinesis.20 It would be interesting to define whether SMAP1 is one of the Arf6GAPs Immunohistochemical analyses involved in cytokinesis, as its inactivation would increase the Arf6– Blocks of formalin-fixed paraffin-embedded CRC tissue comprising an area GTP amount at the plasma membrane, subsequently shortening of normal colonic mucosa adjacent to the tumor were selected, as 22 the G2/M phase, as seen in our experiments. described. Antigen retrieval was performed on 4-mm sections in citrate SMAP1 has been previously described as being involved in buffer (pH 6.0) using a microwave technique (15 min at 95 1C), sections were then incubated for 2 h at room temperature with a rabbit antibody clathrin-mediated endocytosis of E-cadherin, the main component 11 against SMAP1 (1/75 dilution, HPA030574, Atlas Antibodies, Stockholm, of the adherent junctions. HeLa cells that normally express Sweden) and revealed using the Biogenex Detection kit (Biogenex, SMAP1 internalized E-cadherin efficiently, while those Fremont, CA, USA). SMAP1 expression was considered negative when overexpressing wild-type SMAP1 or a GAP-negative mutant there was a complete absence of staining of neoplastic cells, provided that failed to internalize E-cadherin, suggesting that Arf6–GTP hydro- adjacent lymphocytes, normal epithelial cells, endothelial cells or lysis is necessary for membrane proteins recycling.11 Clathrin- fibroblasts showed positive staining. dependent E-cadherin endocytosis represents a major pathway controlling the distribution of E-cadherin molecules in adherens Cell line cultures, proliferation and clonogenic assays junctions of epithelial cells. After its endocytosis, E-cadherin is Cell lines were cultured in Dulbecco’s Modified Eagle Medium containing either recycled back to the membrane or degraded by the stable glutamine (PAA Laboratories GmbH, Velizy, France) supplemented with proteasome. Why restoring SMAP1 in HCT116 cells significantly 10% heat-inactivated fetal calf serum and antibiotics at 37 1C. Cell lines were decreased the expression of E-cadherin not only at the cell authenticated using five highly polymorphic (CA)n markers (D2S123, D5S107, surface, but also its overall expression, without repressing its D17S1791, D17S250 and D18S1127). For proliferation assays, cells seeded in transcription (data not shown) remains to be elucidated. One six-well plates at 12 000 cells per well were harvested 24 or 48 h and counted possibility could be that SMAP1 regulates the equilibrium using the Scepter counter (Millipore, Billerica, MA, USA). For clonogenic assays, between endocytosis, recycling and/or degradation of cells were seeded at 50 cells per well in 24-well plates, grown for 6 days, then 1 E-cadherin, a hypothesis that remains to be investigated in CRC fixed in ice-cold 100% methanol for 10 min at À 20 C and stained with 1% toluidine blue in 1% borax. Clones were counted using the G:BOX chem- cell lines. iluminescence image analyzer and the GeneTools software (Syngene, Ozyme, Restoring wild-type SMAP1 expression in HCT116 induced Cambridge, UK). several cellular modifications characteristic of EMT: loss of E-cadherin expression at the cell membrane, b-catenin transloca- Cell cycle analysis tion to the nucleus, expression of mesenchymal markers (Snail, 5 Slug and vimentin), cell spreading and increased invasiveness. Cells were seeded in six-well plates at 3 Â 10 cells per well, and treated The fact that SMAP1 inactivation counteracts cell invasion seems with 50 ng/ml nocodazole during 16 h at day 2. Cells were harvested and counted at various time points; 106 cells were fixed in 2 ml ice-cold ethanol not to match the dogma stating that selected mutations are those 70%, permeabilized with 1% NP40, treated with 1 mg/ml RNase A and giving an advantage to the tumor cells. It is important to recall labeled with 50 mg/ml propidium iodide. Cells were analyzed by flow that SMAP1 mutations occur exclusively in mismatch repair- cytometry using LSRII cytometer (BD Bioscience, Le Pont de Claix, France). deficient tumors, which continuously accumulate mutations Cell cycle data were calculated using ModFit software (Verity Software during replication, giving rise to highly heterogeneous popula- House, Topsham, ME, USA). tions of cells carrying various combinations of mutations. Some mutations are believed to behave as driver mutations, conferring SMAP1 and SMAP2 genotyping, SMAP1 sequencing growth advantage to tumor cells (generally shared by most MSI Indel mutations occurring in the SMAP1 A10 and SMAP2 A8 repeats were CRC), while others are considered as passenger mutations, not screened on genomic DNA, as described.7 The sequencing reactions essential for tumor progression (being present in a minority of MSI corresponding to each exon and the flanking sequences of SMAP1 were CRC cases). Nevertheless, passenger mutations may have a role performed with the Big Dye Terminator v1.1 kit (Applied Biosystems, Grand when arising in combination with additional acquired genetic Island, NY, USA) and one of the primers used for PCR. Primer sequences are alterations or in response to specific stimuli, by hampering available as Supplementary Table S2. Electrophoreses were run on an ABI adequate cellular responses. Thus, we propose that SMAP1 PRISM 3100 Genetic Analyzer using 36-cm capillaries and GS Performance mutations counteract tumor progression only when arising in Optimized Polymer 7 (Applied Biosystems). tumors carrying a particular pattern of alterations that remains to be identified. Using a collection of MSI CRC cell lines displaying Analysis of SMAP1 transcript expression by quantitative RT–PCR various patterns of mutations might give some clues to unravel RNA extraction was performed on cells homogenized with the Mixer Mill the signaling pathways involved in response to SMAP1 in the MM300 (Qiagen) using the RNeasy Mini kit with a DNA digestion step by future. DNase I before eluting RNA (Qiagen). Reverse transcription into In conclusion, we report that loss-of-function SMAP1 mutations complementary DNA was performed with the High Capacity cDNA RT kit frequently occur in the MSI tumors of various tissue origins, and utilizing random primers (Applied Biosystems). SMAP1 transcript expres- we provide evidence for their implications in diverse biological sion was determined with the SMAP1 assay on demand gene expression assay (Hs 00222567) and normalized using 18S (Hs 99999901). Reverse functions, unraveling their role in oncogenesis. Owing to the transcription–PCR reactions were performed using an ABI Prism 7900HT recently reported role for several ArfGAP proteins, including Arf6, 15,16,21 (PE Applied Biosystems) and the TaqMan PCR master mix (Applied in regulating adhesion, migration and cancer cell invasion, Biosystems). Amplification plots and predicted threshold cycle (Ct) values we propose to assign a novel role for SMAP1 in this emerging were obtained with the Sequence Detector Software (SDS 2.1, Applied oncogenic pathway. Biosystems); ratios between SMAP1 and 18S were calculated as the mean of

& 2014 Macmillan Publishers Limited Oncogene (2014) 2758 – 2767 SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2766 the ratios of independent triplicate experiments and expressed relative to antibodies coupled to Alexa Fluor 488 (Molecular Probes, 1/500); nuclei the HeLa cell line. were stained with 4’,6-diamidino-2-phenylindole 300 nM during 10 min.

Constructions of the wild-type and mutant SMAP1 HA-tagged Flow cytometry analysis of E-cadherin expression expression vectors Cultured cells were isolated by scraping and resuspended in 1% The SMAP1 complementary DNA originated from pENTR221-SMAP1 vector paraformaldehyde for 15 min. After a 15 min incubation in phosphate- (clone IOH27289, Gateway, Life Technologies, Grand Island, NY, USA) was buffered saline containing 1% bovine serum albumin, cells were incubated cloned in pcDNA3.1-Hygro þ expression vector (Life Technologies), after for 1 h with a mouse monoclonal antibody raised against the extracellular site-directed mutagenesis to correct the C1181A mutation (Pro394His). The domain of E-cadherin (HECD-1 clone, Abcam, 1/400), washed twice and SMAP1-A9 mutation, corresponding to a 1 bp deletion in the A10 repeat, further incubated for 1 h with a goat anti-mouse immunoglobulin G was obtained by site-directed mutagenesis using the 50-GACAAA antibody coupled to Alexa Fluor 388 (A-11001, Molecular Probes, 1/1000). AATAAATTGGAGAAAGAAAAGGAAAAAAAAAGGAAGAGAAAAAGAGAGAAA The level of ﬂuorescence was measured using LSRII ﬂow cytometer (BD AG-30 oligonucleotide and its complementary sequence, as primers. Wild- Biosciences). type SMAP1wt-HA and mutant SMAP1-A9-HA complementary DNAs were cloned in the pcDNA3.1-Hygro þ expression vector, and further genotyped Analysis of Snail, Slug and vimentin transcript expression by and sequenced, as previously described. quantitative reverse transcription–PCR Total RNA (1 mg) extracted from cell cultures was reverse transcribed using Transfection of HCT116 and HeLa cell lines with wild-type random hexamers and 200 U Moloney murine leukemia virus reverse SMAP1wt-HA and mutant SMAP1-A9-HA expression vectors transcriptase (Life Technologies) for 1 h at 37 1C. Quantitative reverse Exponentially growing HeLa cells were transfected with pcDNA3.1–SMAP1- transcription–PCR of target and reference (hypoxanthine–guanine phos- phoribosyltransferase) genes was performed with the SYBR Green PCR A9-HA and Lipofectamine 2000 (Life Technologies). The HCT116 cell line, 23 which carries a biallelic SMAP1-A9 mutation, was used to establish clones Master Mix (Roche) on a LightCycler 480 (Roche), as previously described; wt results are expressed relative to HCT116. The primers used in each reaction that stably re-expressed a SMAP1 allele by transfecting pcDNA3.1– were as follows: SNAIL 50-CCTAACTACAGCGAGCTGCAG-30 and 50-ATC SMAP1wt-HA, using Nucleofector kit (Amaxa, Lonza, Basel, Switzerland). 0 0 0 Cells were plated at 300 to 10 000 cells per well in 96-well plates, and TCCGGAGGTGGGATG-3 , SLUG 5 -AAGAAGCTGGCTGACATGTACGGA-3 and 50-CCACCAGCAACGTGATTTCTGCAT-30, and vimentin, 50-ATCTGAGCCTGC selected with 100 mg/ml hygromycin B; resistant clones were screened for 0 0 0 the expression of SMAP1wt-HA by western blotting, as described below. AGCTCCT-3 and 5 -CCCTGCAATCTTTCAGACAG-3 .

Invasion assays Proteasome inhibition Experiments were done in a modified Boyden chamber consisting in an insert MG132 (Sigma-Aldrich, St Louis, MO, USA) was added to HCT116 cells with a membrane perforated by 8-mm pores (BD Biosciences) coated with immediately after transfection with pcDNA3.1–SMAP1wt-HA or pcDNA3.1– 1 mg/ml Matrigel (BD Biosciences) and placed in 24-wells plates containing SMAP1-A9-HA, incubated for 8 h at 37 1C, and lysed to perform western blot complete medium. Cells were incubated on Matrigel in Dulbecco’s Modified analysis. Eagle Medium for 24 h. Invading cells were fixed in ice-cold methanol 100% during 10 min at À 20 1C and stained with 4’,6-diamidino-2-phenylindole Western blot 300 nM for 10 min. The membranes were then placed on a cover slip to take pictures that were analyzed with the ImageJ software using the plugging Total protein extracts (25 mg) obtained by directly lysing cells in Laemmli ‘Boyden’(INSERMUMR_S903,Reims,France). sample buffer (Bio-Rad, Marnes-la-Coquette, France) were separated on NuPAGE Novex 4–12% Bis-Tris Gels (Life Technologies), transferred onto 0.2 mm nitrocellulose membranes (Bio-Rad), saturated with phosphate- Statistics buffered saline containing 0.1% v/v Tween-20 and 5% w/v nonfat dry milk, Differences in distributions between the variables examined were assessed and analyzed using the following primary antibodies: MaxPab mouse with the Student’s t-test (age distribution), the w2 (tumor location, stage) or polyclonal antibody anti-SMAP1 (H00060682-B01P, Abnova, Taipei City, the Fisher’s exact test (patients under 65 years, sex, differentiation). Taiwan, 1/1000), mouse monoclonal anti-E-cadherin (HECD-1 clone, Transcript expression levels were compared using a one-tailed Mann– Abcam, Cambridge, UK), rabbit whole-serum anti-b-catenin (C2206 clone, Whitney test. To analyze SMAP1 mutations frequency according to the Sigma-Aldrich, 1/400), or mouse monoclonal anti-p53 (DO-1 clone, tumor-node-metastasis stage of the tumor, we used a w2 test for trend Calbiochem, Darmstadt, Germany, 1/1000) antibodies. Loading controls comparing all data and a two-tailed t-test for pairwise comparisons. For cell were performed with a mouse monoclonal anti-GAPDH conjugated with proliferation assay, xenograft tumor growth and quantitative reverse HRP (ab9482, Abcam). Membrane signals were revealed with the Super- transcription–PCR assays, we used a one-way ANOVA, Dunnett’s test Signal West Femto Substrate (Pierce, Rockford, IL, USA), Thermo Fisher comparing the clones to the HCT116 parental cell line. For clonogenic and Scientific (Illkirch, France), detected and quantified using the G:BOX invasion assays, we performed two-tailed t-tests. Analyses were performed chemiluminescence image analyzer and the GeneTools software using the Prism 5 software (GraphPad Software Inc, La Jolla, CA, USA). (Syngene).

Mouse xenograft tumor models CONFLICT OF INTEREST Animal experiments were done in accordance with the French Animal The authors declare no conflict of interest. Research Committee guidelines. Cells (5 Â 106 cells) from HCT116 and two SMAP1-complemented clones (C7 and C12) were implanted subcutaneously into the flank of 5-week-old immunodeficient female mice (NMRI-nu nu/nu; ACKNOWLEDGEMENTS Janvier, Le Genest Saint-Isle, France) (eight mice per group). Measurement of We thank Drs Mathieu Boissan, Philippe Chavrier, Marie-Lise Lacombe and Guillaume tumor size was performed twice a week using a caliper, and tumor volume Montagnac for their constant support and helpful discussions, as well as Dr Peggy (V) was calculated using the formula: V ¼ ab2p/6, where a is the longest and b Cuilliere-Dartigues for sharing some data. We are grateful to Martine Mergey, Anne- is the shortest of two perpendicular diameters. After 30 days, animals were Marie Faussat and Philippe Fontanges for their invaluable help in RT-PCRq analyses, killed and tumors were measured and weighted. cytometry and confocal microscopy, respectively. We further acknowledge the expert technical assistance of Fatiha Me´rabtene in performing immunohistochemical analyses of mouse and human tumor samples. We are indebted to the members Immunofluorescence of INSERM UMR_S 903 for sharing the ImageJ plugin. Fatiha Sangar received a 3-year Cells grown on cover slips were fixed in paraformaldehyde 2% for 15 min, fellowship from ‘Ministe`re de l’Enseignement Supe´rieur et de la Recherche’ followed permeabilized in acetone/methanol (1:1, v/v) 15 min, incubated for 1 h with by a 6-month funding from ‘Fondation ARC pour la Recherche sur le Cancer’. primary antibodies diluted in phosphate-buffered saline–bovine serum This work has been supported by grants from the Association pour la Recherche albumin 1% (mouse anti-E-cadherin, HECD-1 clone, at 1/400 or rabbit anti- sur le Cancer (#5091) and from the Ligue Nationale Contre le Cancer (RS10/75-72 b-catenin, C2206, at 1/400), followed by a 1 h incubation with secondary and RS11/75-89).

Oncogene (2014) 2758 – 2767 & 2014 Macmillan Publishers Limited SMAP1 mutations in MSI-driven oncogenic pathway F Sangar et al 2767 REFERENCES protein and functions on the AP-1-positive early endosome/trans-Golgi network. 1 Peltomaki P. Lynch syndrome genes. Fam Cancer 2005; 4: 227–232. Mol Biol Cell 2006; 17: 2592–2603. 2 Zaanan A, Meunier K, Sangar F, Flejou JF, Praz F. Microsatellite instability in col- 13 D’Souza-Schorey C, Chavrier P. ARF proteins: roles in membrane traffic and orectal cancer: from molecular oncogenic mechanisms to clinical implications. Cell beyond. Nat Rev Mol Cell Biol 2006; 7: 347–358. Oncol (Dordr) 2011; 34: 155–176. 14 Palacios F, Price L, Schweitzer J, Collard JG, D’Souza-Schorey C. An essential role 3 Duval A, Hamelin R. Mutations at coding repeat sequences in mismatch repair- for ARF6-regulated membrane traffic in adherens junction turnover and epithelial deficient human cancers: toward a new concept of target genes for instability. cell migration. Embo J 2001; 20: 4973–4986. Cancer Res 2002; 62: 2447–2454. 15 Hashimoto S, Onodera Y, Hashimoto A, Tanaka M, Hamaguchi M, Yamada A et al. 4 Woerner SM, Yuan YP, Benner A, Korff S, von Knebel Doeberitz M, Bork P. Requirement for Arf6 in breast cancer invasive activities. Proc Natl Acad Sci USA SelTarbase, a database of human mononucleotide-microsatellite mutations and 2004; 101: 6647–6652. their potential impact to tumorigenesis and immunology. Nucleic Acids Res 2010; 16 Muralidharan-Chari V, Hoover H, Clancy J, Schweitzer J, Suckow MA, Schroeder V 38: D682–D689. et al. ADP-ribosylation factor 6 regulates tumorigenic and invasive properties 5 Loire E, Praz F, Higuet D, Netter P, Achaz G. Hypermutability of genes in Homo in vivo. Cancer Res 2009; 69: 2201–2209. sapiens due to the hosting of long mono-SSR. Mol Biol Evol 2009; 26: 111–121. 17 Wang J, Sun L, Myeroff L, Wang X, Gentry LE, Yang J et al. Demonstration that 6 Isken O, Maquat LE. Quality control of eukaryotic mRNA: safeguarding cells from mutation of the type II transforming growth factor beta receptor inactivates its abnormal mRNA function. Genes Dev 2007; 21: 1833–1856. tumor suppressor activity in replication error-positive colon carcinoma cells. J Biol 7 El-Bchiri J, Guilloux A, Dartigues P, Loire E, Mercier D, Buhard O et al. Nonsense- Chem 1995; 270: 22044–22049. mediated mRNA decay impacts MSI-driven carcinogenesis and anti-tumor 18 Hempen PM, Zhang L, Bansal RK, Iacobuzio-Donahue CA, Murphy KM, immunity in colorectal cancers. PLoS One 2008; 3: e2583. Maitra A et al. Evidence of selection for clones having genetic inactivation of the 8 Marcos I, Borrego S, Rodriguez de Cordoba S, Galan JJ, Antinolo G. Cloning, activin A type II receptor (ACVR2) gene in gastrointestinal cancers. Cancer Res characterization and chromosome mapping of the human SMAP1 gene. Gene 2003; 63: 994–999. 2002; 292: 167–171. 19 Gardner LB, Nonsense-mediated RNA. decay regulation by cellular stress: 9 Tanabe K, Kon S, Natsume W, Torii T, Watanabe T, Satake M. Involvement implications for tumorigenesis. Mol Cancer Res 2010; 8: 295–308. of a novel ADP-ribosylation factor GTPase-activating protein, SMAP, in 20 Montagnac G, Sibarita JB, Loubery S, Daviet L, Romao M, Raposo G et al. ARF6 membrane trafficking: implications in cancer cell biology. Cancer Sci 2006; 97: interacts with JIP4 to control a motor switch mechanism regulating endosome 801–806. traffic in cytokinesis. Curr Biol 2009; 19: 184–195. 10 Tanabe K, Torii T, Natsume W, Braesch-Andersen S, Watanabe T, Satake M. A novel 21 Sabe H. Requirement for Arf6 in cell adhesion, migration, and cancer cell invasion. GTPase-activating protein for ARF6 directly interacts with clathrin and regulates J Biochem 2003; 134: 485–489. clathrin-dependent endocytosis. Mol Biol Cell 2005; 16: 1617–1628. 22 Miquel C, Jacob S, Grandjouan S, Aime A, Viguier J, Sabourin JC et al. Frequent 11 Kon S, Tanabe K, Watanabe T, Sabe H, Satake M. Clathrin dependent endocytosis alteration of DNA damage signalling and repair pathways in human colorectal of E-cadherin is regulated by the Arf6GAP isoform SMAP1. Exp Cell Res 2008; 314: cancers with microsatellite instability. Oncogene 2007; 26: 5919–5926. 1415–1428. 23 Claperon A, Guedj N, Mergey M, Vignjevic D, Desbois-Mouthon C, Boissan M et al. 12 Natsume W, Tanabe K, Kon S, Yoshida N, Watanabe T, Torii T et al. SMAP2, a novel Loss of EBP50 stimulates EGFR activity to induce EMT phenotypic features in ARF GTPase-activating protein, interacts with clathrin and clathrin assembly biliary cancer cells. Oncogene 2012; 31: 1376–1388.

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

& 2014 Macmillan Publishers Limited Oncogene (2014) 2758 – 2767