1078-0432.CCR-15-2157.Full.Pdf

Published OnlineFirst September 12, 2016; DOI: 10.1158/1078-0432.CCR-15-2157

Biology of Human Tumors Clinical Cancer Research Characterization of Biomarkers of Tumorigenic and Chemoresistant Cancer Stem Cells in Human Gastric Carcinoma Phu Hung Nguyen1,2, Julie Giraud1,2, Lucie Chambonnier1,2, Pierre Dubus2,3,4, Linda Wittkop2,5,6, Genevieve Belleannee 4, Denis Collet4, Isabelle Soubeyran7,8, Serge Evrard2,7,8, Benoit Rousseau2,9, Nathalie Senant-Dugot2,10, Francis Megraud 1,2,4, Fred eric Mazurier11, and Christine Varon1,2

Abstract

Purpose: Gastric carcinomas are heterogeneous, and the current activity presented the properties to generate new heterogeneous therapy remains essentially based on surgery with conventional tumorspheres in vitro and tumors in vivo. CD44 and CD166 were chemotherapy and radiotherapy. This study aimed to characterize coexpressed, representing 6.1% to 37.5% of the cells; ALDH biomarkers allowing the detection of cancer stem cells (CSC) activity was detected in 1.6% to 15.4% of the cells; and the þ þ þ in human gastric carcinoma of different histologic types. ALDH cells represented a core within the CD44 /CD166 Experimental Design: The primary tumors from 37 patients subpopulation that contained the highest frequency of tumori- þ with intestinal- or diffuse-type noncardia gastric carcinoma were genic CSCs in vivo. The ALDH cells possessed drug efflux prop- studied, and patient-derived tumor xenograft (PDX) models in erties and were more resistant to standard chemotherapy than the – immunodeficient mice were developed. The expressions of 10 ALDH cells, a process that was partially reversed by verapamil putative cell surface markers of CSCs, as well as aldehyde dehy- treatment. drogenase (ALDH) activity, were studied, and the tumorigenic Conclusions: CD44 and ALDH are the most specific properties of cells were evaluated by in vitro tumorsphere assays biomarkers to detect and isolate tumorigenic and chemore- and in vivo xenografts by limiting dilution assays. sistant gastric CSCs in noncardia gastric carcinomas inde- Results: We found that a subpopulation of gastric carcinoma pendently of the histologic classification of the tumor. Clin cells expressing EPCAM, CD133, CD166, CD44, and a high ALDH Cancer Res; 1–12. 2016 AACR.

Introduction pylori (H. pylori). Infection with H. pylori,classified as a class 1 carcinogen by the World Health Organization, induces a chron- Gastric cancer is the fourth most common cancer in frequen- ic inflammation evolving over decades from a chronic atrophic cy and the third leading cause of cancer mortality in the world. gastritis to intestinal metaplasia, dysplasia, and finally adeno- Ninety-five percent of gastric cancers are gastric carcinomas, carcinoma (1, 2). Some cases also include Lynch syndromes which are divided into two types depending on their localiza- (microsatellite instability, MSI) and Epstein–Barr Virus (EBV) tion in the stomach: adenocarcinomas of the cardia whose infection. The classification of gastric carcinomas is based etiology remains unclear, and noncardia gastric carcinomas for essentially on histologic criteria. The Lauren classification dis- which the main factor is a chronic infection by Helicobacter tinguishes two main subtypes, the intestinal type, which represents the majority of the cases, and the diffuse type (3). The 1INSERM, U853 Helicobacter Infection, Inflammation and Cancer, Bordeaux, intestinal type is composed of glands having more or less 2 3 France. University of Bordeaux, Bordeaux, France. EA 2406, University of preserved their organization and differentiation state, or having Bordeaux, Bordeaux, France. 4University Hospital Center of Bordeaux, Bor- 5 acquired intestinal characteristics; it is subclassified into tubu- deaux, France. INSERM, ISPED, Centre INSERM U1219 Bordeaux Population fi Health, Bordeaux, France. 6Pole^ de Sante Publique, Service d'information lar, mucinous, or papillary carcinoma in the WHO classi ca- medicale, University Hospital Center of Bordeaux, Bordeaux, France. 7Institut tion of gastric carcinoma (4). The diffuse type is poorly cohe- Bergonie, Bordeaux, France. 8INSERM, U1012 Actions for onCogenesis under- sive, composed of isolated cells (often signet ring cells) pro- standing and Target Identification in Oncology (ACTION), Bordeaux, France. ducing mucins. These classification systems have little clinical 9 10 Service Commun des Animaleries, Animalerie A2, Bordeaux, France. SFR utility, as they cannot orientate patient therapy. With the 11 TransBioMed, Bordeaux, France. CNRS UMR 7292, GICC LNOx, Tours, France. exception of Her2 positivity which orientates toward a specific Note: Supplementary data for this article are available at Clinical Cancer treatment, treatment is still based on surgery combined with Research Online (http://clincancerres.aacrjournals.org/). conventional chemotherapy and/or radiotherapy, and the Corresponding Author: Christine Varon, University of Bordeaux, INSERM, U853 5-year survival rates remain under 30% in most countries (5). Helicobacter Infection, Inflammation and Cancer, 146 rue Leo Saignat, Bordeaux Recently, the Cancer Genome Atlas Research Network and F33076, France. Phone: 33557579575; Fax: 33556796018; E-mail: Wang and colleagues published a molecular profiling of gastric [email protected] carcinomas based on two studies with 295 cases and 100 cases, doi: 10.1158/1078-0432.CCR-15-2157 respectively. Both studies led to a classification of gastric carci- 2016 American Association for Cancer Research. nomas into four main subtypes according to their molecular

www.aacrjournals.org OF1

Nguyen et al.

gastric carcinoma cell lines proposed CD44 as a gastric CSC Translational Relevance marker, but this marker was expressed in three out of six cell We report the screening of the expression of 10 cell surface lines studied, and confirmation in primary tumors was lacking markers and aldehyde dehydrogenase (ALDH) activity on cells (21). Then, the study performed by Rocco and colleagues on from primary gastric carcinoma. We found that a subpopula- 12 human primary gastric carcinomas failed to demonstrate þ þ tion of tumor cells expressing CD133, CD166, CD44, and tumor-initiating properties of CD133 -andCD44 -sorted cells ALDH presented cancer stem cell (CSC) tumorigenic proper- after xenograft in both NOD/SCID and nude immunodeficient þ ties in vitro and in vivo. Among them, ALDH cells represented mice (22). 1.6% to 15.4% of the tumor cells and contained the highest Another important point concerns the origin of the CSCs. As þ frequency of tumorigenic CSCs before CD44 cells. In addi- Houghton and colleagues and our group reported in mouse þ þ tion, the tumorigenic CD44 ALDH cells possessed drug models of Helicobacter-induced gastric carcinogenesis, gastric efflux and chemoresistance properties, constituting the cells dysplasia and carcinoma may originate from the transformation to target in the development of new therapy. Results also of a local epithelial stem cell or of a bone marrow (BM)–derived showed that CD44, which is poorly expressed or absent in stem cell (23–25). In this model, dysplastic lesions were com- þ healthy gastric epithelium, is overexpressed in gastric carcino- posed of CD44 cells, regardless of their BM or local origin (24). ma and may constitute a good biomarker for the detection of In addition, the heterogeneity of gastric carcinomas suggests that CSCs by standard immunohistochemistry on patient tissue gastric CSC markers, if they indeed exist, may be different accord- samples, whereas detection of CSCs possessing a high ALDH ing to the origin and/or the histologic type of gastric carcinoma. activity is not yet possible by standard immunohistochemistry In this study, we performed an extensive screening of the and involves many ALDH isozymes. expression of putative cell surface markers of CSCs as well as ALDH activity in order to identify biomarkers allowing the detection and isolation of tumorigenic and chemoresistant CSCs in human primary intestinal- and diffuse-type noncardia gastric carcinoma. þ profiles: (1) EBV tumors (frequent PIK3CA mutations, extreme DNA hypermethylation), (2) MSI tumors (elevated mutation rates, hypermethylation), (3) genomically stable tumors Materials and Methods (enriched for the diffuse type; driver mutations include CDH1, Human samples and mouse xenografts RHOA, cytoskeleton, and cell junction regulators), and (4) chro- Fresh tumors samples were collected from gastric surgical mosomal instability tumors (marked aneuploidy, focal amplifi- wastes from patients who underwent gastrectomy for noncardia cation of tyrosine kinase receptors; refs. 6, 7). These studies were gastric carcinoma and for whose informed consent was performed without distinguishing between cardia and noncardia obtained. Fresh samples of tumor and paired nontumor tissues gastric carcinomas, whose etiology is different. were transported in DMEM medium with 20% FCS, 50 IU/mL Tumors are heterogeneous, composed of cells which are more penicillin, 50 mg/mL streptomycin, 50 mg/mL vancomycin, and or less differentiated, and not all proliferative. Over the last 15 mg/mL amphotericin-B. Samples were minced in small decade, extensive research has focused on the discovery and the pieces of 2 mm 2 mm size and were subcutaneously trans- characterization of cancer stem cells (CSC) at the origin of cancers planted into the right dorsal flank of 7-week-old male NSG mice in numerous organs. Tumors are hierarchically organized with under 2.5% isoflurane anesthesia (Belamont). Alternatively, CSCs at the top of this pyramid and at the origin of tumor after mechanical mincing, cells were dissociated by incubation initiation, heterogeneity, and propagation (8–10). The CSCs in a solution of 1 mg/mL collagenase IV and 0.2 mg/mL correspond to a subpopulation of cells within the tumor defined hyaluronidase in DMEM (Sigma) for 1 hour at 37Cwith by self-renewal, asymmetrical division, and differentiation pro- shaking (15), then suspended in 100 mLof7mg/mLice-cold perties, giving rise to the more or less differentiated cells com- Matrigel (BD Biosciences) for subcutaneous injection. Xeno- posing the tumor mass. CSCs can stay in quiescence under grafts were carried out within 3 to 5 hours following gastrec- some conditions, resist conventional therapies, and be at the tomy. The tumor size was monitored with callipers once a week, origin of tumor relapse and metastasis. The definition of CSCs and tumor volume was estimated as (D2 x d)/2, where D is the remains largely operational and based on functional assays that large diameter and d is the small diameter (26). At the end of register their self-renewal and tumorigenic properties, assessed the experiments (until 10 months after engraftment for the by the formation of new heterogeneous tumors after xenograft primary xenograft) and when tumor reached approximately in vivo, and of tumorspheres in particular culture conditions 500 mm3,miceweresacrificed by cervical dislocation and in vitro (8–10). tumors were immediately harvested and processed for analyses. Indeed CSC may display both genetic and phenotypic hetero- Secondary tumors were amplified subcutaneously in mice geneity, markers allowing their identification have been charac- by serial transplantation of pieces of tumor bulk, or by injection terized in tumors of different organs, including CD133, CD44, of tumor cells in Matrigel. For xenograft experiments in extreme and CD24 among those studied (8, 10–16). More recently, the limiting dilution assay (ELDA), 10,000 to 30 FACS-sorted cells activity of aldehyde dehydrogenases (ALDH), intracellular were subcutaneously injected with Matrigel; tumor size was enzymes involved in oxidation of aldehydes and retinoic acid recorded twice a week. signaling, also led to the identification of CSCs in tumors of the breast (17), lung (18), colon (19), and other organs (20). Gastric carcinoma cell lines In the stomach, the existence of CSCs has been subject to Gastric carcinoma cell lines were cultured in DMEM/F12 debate. The first study performed by Takaishi and colleagues on media for AGS (ATCC CRL1739) and in RPMI1640 media

OF2 Clin Cancer Res; 2017 Clinical Cancer Research

Biomarkers of Tumorigenic and Chemoresistant Gastric CSCs

for NCI87 (ATCC CRL-5822), MKN45, MKN74, MKN7, and tumorsphere by trypsin/EDTA procedure and seeded in new MKN28 (all from RIKEN) cells, supplemented with 10% heat- nonadherent 96-well plates (8 < n < 10 per condition) was inactivated FCS, 50 IU/mL penicillin and 50 mg/mL strepto- evaluatedafter6to9days. mycin (all from Invitrogen) at 37 Cina5%CO2 atmosphere Ethic statements, histology, immunohistochemistry, immuno- (26, 27). All cell lines were routinely verified mycoplasma fluorescence procedures, and statistical analysis are described in free (by PCR) and were tested and authenticated by short Supplementary Materials. tandem repeat (STR) profiling within 6 months preceding the experiments (last STR profiling report, October 2015; LGC Standards). Cell viability was assessed by the trypan blue Results exclusion method. Establishment of a mouse PDX model of diffuse- and intestinal-type noncardia gastric carcinoma Flow cytometry analysis Fresh gastric carcinoma and nontumor tissue samples were Cells were dissociated by collagenase/hyaluronidase proce- collected by pathologists upon surgical resection from consenting dure from fresh patient-derived tumor xenografts (PDX), patients who underwent gastrectomy for noncardia gastric carci- passed through a 70-mmmeshfilter (BD), and red blood cells noma at the University Hospital Center and the Bergonie Institute were removed by incubating in a solution containing 2 mmol/L in Bordeaux. Among the 37 cases studied, the median age was 72 years, 59.5% were males and 54.1% did not receive preoperative KHCO3, 0.1 mmol/L EDTA, and 170 mmol/L NH4Cl for 8 minutes at 4C. Then, 100,000 cells in 100 mL PBS-0.5% BSA-2 chemotherapy (Supplementary Table S1). Approximately half of mmol/L EDTA (Sigma) were stained with 3 to 5 mLoffluores- the tumors were tubular and one third were poorly cohesive fi cent-labeled primary antibodies including EPCAM-FITC (Stem according to the WHO and Lauren classi cations of gastric car- Cell Technologies) or EPCAM-VioBlue (MACS-Miltenyi Bio- cinomas. More than half were high grade (51.4%), highly pen- ¼ – ¼ tec), CD10-PE, CD24-PE, CD73-PE, CD49f-PE, CD105-PE, etrant (T4 51.3%) with lymph node invasion (N1 4 67.6%), fi CD166-PE, CD90-PECy5, CD44-PE, CD44-APC, CD338-APC and at stages 3 to 4 (51.4%). When tissue size was suf cient, pieces (all from BD), and CD133-PE (MACS-Miltenyi Biotec) for 20 were dissociated by enzymatic procedures, and single cells were fl minutes at 4 C. Cells were rinsed twice with PBS-0.5% BSA-2 analyzed by ow cytometry for the expression of EPCAM, an mmol/L EDTA containing 50 mg/mL 7-aminoactinomycin-D epithelial marker, in combination with CD24, CD133, and CD44 (7-AAD;BD)beforebeinganalyzedusingaFACSCantoII as putative CSC markers, and 7-AAD to exclude dead cells instrument and DIVA software (BD; refs. 26, 27). The ALDE- (Fig. 1A). To overcome the problem of the small size of the FLUOR Kit (Stem Cell Technologies) was used to detect ALDH human samples which was a limiting factor in the study, we fi activity according to the manufacturer's instructions. Dead cells developed PDX models in NSG immunode cient mice. For this, were excluded based on light scatter characteristics and 7-AAD small pieces of fresh tissue from the patient's biopsy were sub- < n < positivity. For SP cells analysis, cells were incubated with 10 mg/ cutaneously xenografted in mice (3 10 mice per case). mL Hoechst-33342 in HBSS-2% FCS for 60 minutes at 37Cor, Nontumor tissues xenografted as controls never led to tumor when indicated, in ALDEFLUOR buffer for 30 minutes at room growth over the 10 months. Among the 37 tumors xenografted, 8 temperature, with or without 100 mmol/L verapamil or 50 led to the growth of secondary tumors serially transplantable in mmol/L reserpine (Sigma), and then washed with ice-cold mice; 7 were intestinal type (GC04, GC07, GC10, GC35, GC40, HBSS-2% FCS. The Hoechst-33342 dye was excited at 375 nm, and GC44), and 1 was diffuse type (GC06; Supplementary Table fi and its fluorescence was dual wavelength analyzed (blue, 402– S1). There was no signi cant association between PDX growth 446 nm; red, 650–670 nm; ref. 28). Cell sorting was performed and the following characteristics: patient's gender, age, preoper- fi on 5 to 10 million cells stained with primary fluorescent- ative treatment, WHO and Lauren classi cations of gastric tumor fi labeled antibodies or ALDEFLUOR reagent and 7-AAD, on histologic type, grade, tumor-node-metastasis classi cation, and 3 7-AAD–negative cells using a FACSAria (BD). stage (Supplementary Table S1). PDXs reached a 500 mm size between 2 to 6 months (mean 16.7 3.4 weeks) after the first Tumorsphere assay passage (P) (P1) in mice, and earlier following successive A total of 1,000 FACS-sorted cells were plated in nonadher- passages (after 10.9 5.9 weeks at P2 and 7.2 0.8 weeks at ent 24-well plates (or alternatively 200 cells in 96-well plates) P5; Fig. 1B). Case GC42, a mucinous type according to the previously coated with a 10% poly(2-hydroxyethyl methacry- WHO classification, was excluded because tumors developed late) solution in 95% ethanol (v/v; Sigma), in DMEM-F12 slowly and were mostly composed of mucus, rendering its study media supplemented with 20 ng/mL human-epidermal growth impossible. Histopathologic analyses confirmed that the PDXs factor, 20 ng/mL basic-fibroblast growth factor, 5 mg/mL insu- obtained between P1 and P5 remained similar to the respective lin, 0.3% glucose, 50 IU/mL penicillin, and 50 mg/mL strepto- patients' primary tumor for all cases (Fig. 1C) except GC07, which mycin (Sigma; ref. 26). For PDX cells, the media were supple- appeared to dedifferentiate after P2 and therefore was excluded mented with 5% FCS for the first 2 days of culture and were (data not shown). then replaced by serum-free media. After 7 days, the number of spheroids/well was counted under light microscopy using a Evaluation of CD24, CD133, and CD44 cell surface marker 20 objective. For drug treatment experiments, 5-day tumor- expression on patients' gastric tissues and PDXs spheres grown in nonadherent 96-well plates (8 < n < 10 per The expression of CD24, CD133, and CD44 was evaluated by þ condition) were treated with 10 to 20 mmol/L of verapamil with flow cytometry in live (7-AAD negative) EPCAM epithelial 5-fluorouracil, doxorubicin, and cisplatin (all from Sigma). cells dissociated from freshly collected paired nontumor and After 48 hours, the number of tumorspheres was recorded. The tumor gastric tissue samples from 7 cases. In nontumor gastric self-renewal ability of residual viable cells dissociated from tissues, CD24 and CD133 were expressed in about half of the

www.aacrjournals.org Clin Cancer Res; 2017 OF3

Nguyen et al.

AB First 35 Primary xenograft Secondary tumors n 16.7 ± 3.4 tumor from (P1, = 8) 30 10.6 ± 6.9 patient (n = 37) 2-6 months GC04 25 10.9 ± 5.9 GC06 Second xenograft GC07 5-15 weeks 20 GC10

Serial tumors (P2 to P5) 15 GC35 Histologic analyses 7.9 ± 1.0 GC40 Flow cytometry analyses 10 7.2 ± 0.8 Cell sorting by FACS GC42 Number of weeks for tumor development after xenograft 5 GC44 EPCAM+/CD133+ EPCAM+/CD133- EPCAM+/CD44+ EPCAM+/CD44- 0 EPCAM+/ALDH+ EPCAM+/ALDH- P1 P2 P3 P4 P5

Xenografts/ELDA Tumorsphere assay

C GC04 GC06 GC10 GC35 GC40 GC44

Patient tumor

Tumor xenograft

Figure 1. Establishment of mouse xenograft models using primary tumors from patients with noncardia gastric carcinomas (GC). A, schematic representation of the strategy used to collect, analyze, and perform serial tumor xenografts in NSG immunodeﬁcient mice from primary noncardia gastric carcinoma freshly collected from patients who underwent gastrectomy. At each passage (P) in mice, histology and ﬂow cytometry analyses were performed respectively on tissue samples and on cells freshly dissociated by enzymatic procedures. In vitro tumorsphere assays and in vivo CSC frequency determination were monitored after cell sorting by FACS of cell subpopulations based on the expression of EPCAM, CD133, CD44, and ALDH activity following the second passage in mice (P2). B, number of weeks when a tumor size reached 500 mm3 after serial transplantation in mice (from P1 to P5). Histograms represent the mean SD for each case, and numbers represent the global mean SD of all cases (1 < n < 12). C, representative images of hematoxylin–eosin saffron staining of primary tumors from patients and corresponding P2–P3 tumor xenografts. Scale bars, 50 mm.

cells, whereas CD44 was expressed only in 10% 9% of previously reported (26). In patients' tumors, CD44 was cells. In paired tumors, CD24, CD133, and CD44 expression expressed in some tumor cells but not all, mainly at the was significantly higher compared with nontumor cells, CD24 periphery of the tumor islets for the intestinal-type tumors. being expressed in most of the tumor cells (90 8%), CD133 A similar pattern of CD44 expression was observed in PDXs, in 71% 17%, and CD44 in 27% 17% of the tumor cells confirming that the cellular heterogeneity of the primary tumor (Fig. 2A). The percentage of tumor cells expressing these mar- was reproduced in serial PDXs. kers were then evaluated on cells dissociated from freshly collectedserialxenograftsfromP1toP4ofthe6PDXscases Gastric carcinoma cells expressing CD133, CD166, CD44, and (Fig. 2B). The percentage of cells expressing the different an ALDH activity have tumorigenic CSC properties markers remained stable in serial PDXs from P1 to P4, EPCAM We then analyzed the expression of seven additional cell being expressed in most of the cells (>80%) followed by CD24 surface markers as putative markers of gastric CSCs, i.e., CD10, and CD133 expressed at a relatively high proportion of cells, CD49f, and CD166 described in CSCs of other organs (29–31), except in GC35. In all cases, CD44 was expressed in less than a CD73, CD90, and CD105 as the main markers of mesenchymal third of total cells. These results were confirmed by immuno- stemcells(27,32),aswellasALDHactivityincellsfromfive histochemistry analyses of the expression of CD44 on patient's PDXs (freshly collected at P2–P4) and five gastric carcinoma nontumor and tumor tissues and on the corresponding P2–3 cell lines (seven intestinal type: GC04, GC10, GC35, GC44, PDXs for the 6 cases studied (Fig. 2C). In the nontumor area, MKN74, MKN7, NCI87; and three diffuse type: GC06, AGS, CD44 was expressed at a low level, preferentially in cells in the MKN45; Fig. 3A). CD10 was negative except for 2 of the 10 isthmus region of gastric glands in the area of gastritis, as cases studied. CD49f was expressed in more than 80% of the

OF4 Clin Cancer Res; 2017 Clinical Cancer Research

Biomarkers of Tumorigenic and Chemoresistant Gastric CSCs

A B EPCAM CD24 CD133 CD44 300 150 * * 250

100 200 * 150

50 100

of positive cells 50

0 Cumulated percentages 0 Percentage of positive cells Healthy Tumor Healthy Tumor Healthy Tumor P1 P2 P3 P4 P1 P2 P3 P4 P1 P2 P3 P4 P1 P2 P1 P2 P3 P1 P2 P3 GC0 GC06 GC10 GC3 GC4 GC4 CD24 CD133 CD44 4 5 0 4

C GC04 GC06 GC10 GC35 GC40 GC44

Patient’s distant mucosa

Patient’s tumor

Tumor xenograft

Figure 2. Expression of CD24, CD133, and CD44 on gastric epithelial cells from patients and gastric tumor xenografts. The percentage of cells expressing CD24, CD133, and CD44 was determined by flow cytometry analyses on cells dissociated from freshly collected specimens. A, analyses were performed on 7-AAD-EPCAMþ gastric epithelial cells dissociated from paired nontumor (healthy) and tumor gastric mucosa samples from patients (n ¼ 7). Bars, median. , P < 0.05. B, histograms represent the cumulated percentages of positive cells determined for each marker in serial xenografts. Note: CD24 and CD44 expressions were not evaluated on GC06 P1. n, number of tumors analyzed per passage and per case. P1, 1 < n < 3; P2 to P4, 3 < n < 17. C, representative images of CD44 detection by immunohistochemistry on paired tumor and nontumor distant mucosa from patients and corresponding P2–P3 tumor xenografts in mice. Scale bars, 50 mm. cells in all cases, similar to EPCAM. For the other markers, the or negative for these markers were sorted by FACS and submitted expression pattern was more homogeneous in PDXs than in cell to the tumorsphere assay. Similar experiments were performed on lines which exhibited more heterogeneity, being either highly 7-AAD–FACS-sorted cells based on ALDH activity. In all of the þ þ þ þ þ þ positive or negative. CD133 was expressed only in PDXs cells cases, EPCAM CD133 , EPCAM CD44 , EPCAM CD73 , þ þ þ and not in gastric carcinoma cell lines. In PDXs, CD49f and EPCAM CD166 , and ALDH cells formed significantly more CD24 were highly expressed, followed by CD133 and CD90 tumorspheres after 10 days of in vitro culture than their respective expressed in nearly half of the cells, then by CD73 expressed in negative counterparts (Fig. 3C). CSCs forming tumorspheres were þ þ þ more than a third of the cells. At a level similar to CD44, CD166 essentially present in CD44 , CD166 , and ALDH subpopula- þ þ was expressed in 21% 13% of the cells. CD105 expression tions, and to a lesser extent in CD133 and CD73 subpopula- and ALDH activity were detected in only 9% 6% and 8% tions; they were essentially CD90 and CD105 . The high tumor- þ 5% of the cells, respectively (Fig. 3A). Results from flow sphere capacity of ALDH cells was confirmed on both MKN45 cytometry costaining analyses revealed that CD166 and CD44 and MKN74 cell lines (Fig. 3C). were coexpressed and detected the same cell subpopulation To confirm these results in vivo, xenografts were performed in þ þ (Fig. 3B). The majority of CD44 cells were positive for CD24, mice with 7-AAD-EPCAM FACS-sorted cells based on the þ CD133, and CD73. Less than 50% of CD44 cells were positive expression of CD133 and CD44 on GC04, GC06, and GC10 for ALDH, CD105, and CD90. Interestingly, ALDH activity but PDXs. The same experiments were performed based on ALDH not CD90 and CD105 expression was recorded mainly in activity on GC06, GC10, MKN45, and MKN74 cells (Table 1). In þ þ CD44 cells, showing that ALDH cells representing a core all cases, tumors developed at a significant higher frequency in þ þ þ within the CD44 subpopulation of cells (Fig. 3B). EPCAM CD44 cells (1/29 to 1/1,020) than in their respective þ þ In order to evaluate the tumorigenic properties of the cells EPCAM CD44 cells (1/568 to 1/28,963), and in EPCAM þ expressing or not CD44, CD133, CD73, CD166, CD90, and CD133 cells (1/105 to 1/1,911) than in their respective þ CD105, P2–P3 tumors of 3 PDX cases, GC04, GC06, and GC10, EPCAM CD133 cells (1/781 to 1/66,876; Table 1). CSC fre- þ þ þ þ were freshly dissociated and 7-AAD-EPCAM cells either positive quency was higher in EPCAM CD44 cells than in EPCAM

www.aacrjournals.org Clin Cancer Res; 2017 OF5

Nguyen et al.

A CLs (Cancer cell lines), intestinal 120 PTs (Primary tumors), intestinal 100 CLs – diffuse PTs – diffuse 80 60 40 20 0 Percentage of positive cells CLs CLs CLs CLs CLs CLs CLs CLs CLs PTs CLs PTs PTs CLs PTs PTs PTs PTs PTs PTs PTs PTs

EPCAM CD24 CD133 CD44 CD10 CD49f CD73 CD166 CD90 CD105 ALDH B 44.7 23.0 39.5 25.7 13.3 15.7 2.2 17.5 34.4 8.1 6.0 1.7 2.3 10.8 ALDH CD73-PE CD24-PE CD105-PE CD166-PE 29.9 2.4 CD133-PE 33.5 1.3 57.9 9.1 77.3 3.0 42.7 14.8 67.6 24.7 73.3 13.6 CD90-PECy5 CD44-APC

PTs - intestinal 100 PTs - diffuse

20 Percentage of cells

0 + + + + + + + + + + + + + + + + + + + +

+ CD44 + CD44 + CD44 + CD44 + CD44 + CD44 + CD44 into CD24 into CD44 into CD44into CD73 into CD44 into CD44into CD90 into CD44into CD105 into CD44into ALDH + into CD44+ + into CD133 + + + into CD166 + + + + + + CD24 + CD73 + CD133 CD166 CD90 CD105 ALDH CD73 CD44 CD90 CD44 CD44 ALDH CD24 CD44 CD133 CD44 CD166 CD44 CD105 CD44

C 78.2 + DEAB – DEAB 0.4 10.1 2.0

CD44-APC 0.0 CD44-APC EPCAM-FITC 86.7 ALDH Activity

EPCAM-FITC GC04 MKN45 GC06 100 MKN74 150 GC10 * * * * 80 100 * # 60 # 40 50

(per 500 cells) 20 (per 1,000 cells)

Tumorsphere number 0

Tumorsphere number 0 – + – – – – – – – + + + + + + + ALDH ALDH CD44 CD73 CD90 CD44 CD73 CD90 ALDH ALDH CD166 CD105 CD133 CD166 CD105 CD133

OF6 Clin Cancer Res; 2017 Clinical Cancer Research

Biomarkers of Tumorigenic and Chemoresistant Gastric CSCs

Table 1. Gastric cancer–initiating cell frequencies determined on FACS-sorted cells according to the expression of CD133, CD44, and ALDH after tumor xenografts in limiting dilutions in NSG mice Number of tumors/number of transplanted mice Test for difference in Number of transplanted cells Gastric cancer–initiating cell stem cell frequencies frequencies (95% conﬁdence between positive and CASE Marker 10,000 3,000 1,000 300 100 30 interval) negative cells 1 EPCAMþCD133þ 5/5 4/5 1/5 2/5 0/5 1/1,911 (1/4,019–1/908) P < 104 1 EPCAMþCD133– 1/5 0/5 0/5 0/5 0/5 1/66,876 (1/467,214–1/9,573) 2a EPCAMþCD133þ 5/5 5/5 4/10 1/9 0/10 1/1,658 (1/3,071–1/895) P ¼ 0.0001 2a EPCAMþCD133– 2/2 0/5 0/10 0/10 0/10 1/19,065 (1/70,548–1/2,153) 3 EPCAMþCD133þ 4/4 3/5 1/6 1/105 (1/229–1/48) P ¼ 0.0057 3 EPCAMþCD133– 2/4 0/5 0/6 1/781 (1/3,030–1/202)

1 EPCAMþCD44þ 5/5 5/5 8/15 4/15 2/10 1/1,020 (1/1,670–1/623) P < 104 1 EPCAMþCD44– 2/5 0/4 0/5 0/5 1/28,963 (1/113,477–1/7,392) 2 EPCAMþCD44þ 8/8 9/10 6/10 2/5 1/352 (1/625–1/198) P ¼ 0.0012 2 EPCAMþCD44– 4/4 4/8 1/8 0/5 1/1,688 (1/3,913–1/728) 3a EPCAMþCD44þ 15/15 15/15 19/20 7/10 1/29 (1/50–1/17) P < 104 3a EPCAMþCD44– 5/5 2/5 0/5 1/568 (1/1,202–1/268)

2a ALDHþ 2/2 5/5 2/5 1/38 (1/89–1/16) P ¼ 0.0002 2a ALDH– 2/3 0/5 0/5 1/613 (1/2,348–1/160) 3 ALDHþ 5/5 5/5 3/5 2/5 1/81 (1/175–1/37) P ¼ 0.0042 3 ALDH– 4/5 3/5 2/5 1/5 1/372 (1/788–1/176) 4a ALDHþ 6/6 5/6 1/168 (1/455–1/62) P ¼ 0.2145 4a ALDH– 6/6 3/6 1/428 (1/1,285–1/143) 5a ALDHþ 6/6 5/6 5/6 1/746 (1/2,057–1/271) P < 104 5a ALDH– 5/6 1/6 0/6 1/8,024 (1/18,066–1/3,564) NOTE: Case 1, GC04; case 2, GC06; case 3, GC10; case 4, MKN45; case 5, MKN74. aThere was complete data separation, thus estimates may not be reliable. Furthermore, the single-hit assumption that one cell is sufﬁcient for a positive response may not be true.

þ CD133 cells, confirming that CD44 is a better marker of gastric were able to reproduce heterogeneous tumorspheres in vitro, þ CSCs than CD133, as also determined by the in vitro tumor- composed of a similar proportion of cells with ALDH activity þ sphere assays. ALDH cells led to the development of tumors at compared with the initial situation and expressing CD44, and of þ a significant higher frequency than the respective ALDH cells CD44 ALDH and CD44 ALDH cells incorporating the þ þ þ (1/38 to 1/746 for ALDH cells vs. 1/372 to 1/8,024 for ALDH Hoechst-33342 stain (Fig. 4B). In addition, the CD44 ALDH þ cells) for all cases studied, with the exception of MKN45 which FACS-sorted cells generated more tumorspheres than the CD44 was highly tumorigenic at the doses studied (Table 1). For both ALDH cells which generated less but still a significant number PDXs cases GC10 (intestinal type) and GC06 (diffuse type), the of tumorspheres and which therefore could correspond to þ CSC frequencies were high in ALDH cells (1/81 and 1/38, progenitor/transit amplifying cells. The CD44 ALDH cells may þ þ respectively) and in EPCAM CD44 cells (1/29 and 1/352, correspond to more differentiated cells with very limited pro- þ þ respectively), and both were higher than in EPCAM CD133 liferation capacities as they formed significantly less or no þ þ cells (1/105 and 1/1,658, respectively), suggesting that ALDH tumorspheres compared with the CD44 ALDH and CD44 þ and CD44 are more specific markers of gastric CSCs than ALDH cells (Fig. 4C). CD133. Finally, we showed that the CSCs contained in þ þ þ þ EPCAM CD44 and ALDH FACS-sorted cells generated ALDH CSCs are more resistant to conventional chemotherapy – tumors that recapitulated the phenotypic heterogeneity of the than ALDH cells in gastric carcinoma þ initial tumors, giving rise both to CD44 cells containing CSCs The combined analysis of CD44 expression, ALDH activity, and and to more differentiated CD44 cells (Fig. 4A). The asym- Hoechst-33342 incorporation was assessed on live MKN45 metric division and differentiation properties of these CSCs was tumorsphere (as in Fig. 4B) during their development. In young, þ confirmed in vitro with MKN45 cells; ALDH FACS-sorted cells small tumorspheres (after 5 days), most cells were positive

Figure 3. Expression of putative markers of CSCs and tumorigenic properties of corresponding cell subpopulations of PDXs and gastric carcinoma (GC) cell lines. A, the percentages of cells expressing EPCAM, CD24, CD133, CD44, CD10, CD49f,CD73,CD166,CD90,CD105,andALDH activity were determined by flow cytometry analyses on 7-AAD cells from 5 PDXs after P2–P4 (PTs: GC04, GC06, GC10, GC35, and GC44) and 5 gastric carcinoma cell lines (CLs: AGS, MKN28, MKN45, MKN74, and NCI87) of diffuse and intestinal types. Bars, median. B, representative dot-plot analyses of CD24, CD133, CD73, CD166, CD90, and CD105 stained with PE and PECy5-labeled antibodies and ALDH activity determined by ALDEFLUOR assay in combination with anti-CD44/APC antibodies on GC10 cells (top). Quantification of the percentage of positive cells for the different markers in the PDXs (5 < n < 6). Bars, median. C, 7-AAD tumor cells were sorted by FACS based on the expression of the indicated markers and on ALDH activity; and the positive and negative subpopulations of sorted cells were submitted to in vitro tumorsphere assays. Representative images of cell sorting by FACS and phase contrast microscopy of tumorspheres formed by GC10 cells after 10 days of culture in nonadherent conditions in vitro (top). Quantification of the number of tumorspheres formed by FACS-sorted cells after 5 to 10 days of culture (bottom). Results represent the mean SD (n ¼ 8 per condition). , P < 0.05 for all cases studied. #, P < 0.05 for only 2 of 3 cases studied. Scale bars, 50 mm.

www.aacrjournals.org Clin Cancer Res; 2017 OF7

Nguyen et al.

A GC04 GC06GC10 MKN45 MKN74

Tumors obtained with CD44+/high sorted cells

Tumors obtained with ALDH+ sorted cells

B + DEAB – DEAB Merge ALDH Activity (green) CD44 (red) 84.0 13.8

0.0% 11.4% ALDH+

SSC-A sorted cells CD44-APC 2.1 0.1

ALDH Activity ALDH Activity MKN45 NCI-N87 GC07 250 20 * 10 * C 80.0 19.2 * * * 8 200 15

APC Sorted cells * - 150 6 10 * 4 200 cells 100 200 cells 200 cells CD44 5 50 2 Tumorsphere number/ Tumorsphere number/

0.8 0.0 Tumorsphere number/ 0 0 0 - - ALDH Activity CD44 + + + + + + ALDH Activity - + --+ --+

Figure 4. CD44þ and ALDHþ cells generate heterogeneous tumors in vivo and tumorspheres in vitro. A, representative images of CD44 detection by immunohistochemistry on tumors obtained after xenograft of EPCAMþCD44þ and ALDHþ FACS-sorted cells of the indicated PDXs and gastric carcinoma (GC) cell lines. Scale bars, 50 mm. B, MKN45 ALDHþ FACS-sorted cells were submitted to the tumorsphere assay for 8 days, and then analyzed for CD44 expression and ALDH activity. Representative images of ﬂuorescent imaging of CD44 stained with anti-CD44/PE antibodies (in red), ALDH activity detected by ALDEFLUOR reagent (in green), and nuclei staining with Hoechst-33342 (in blue), and of ﬂow cytometry analysis (right) of CD44 stained with anti-CD44/APC antibodies and ALDH activity detected by ALDEFLUOR reagent. Scale bar, 25 mm. C, cells from MKN45, NCI-87, and GC07 PDX were stained as in B and with 7-AAD (to exclude 7-AADþ dead cells) and anti-EPCAM/VioBlue antibodies for GC07 cells dissociated from a fresh PDX (to select EPCAMþ carcinoma cells). The 7-AAD-(EPCAMþ) cells were sorted by FACS on the expression of CD44 and ALDH activity and submitted to the tumorsphere assay. Plots (min to max) represent the number of tumorspheres formed per 200 cells seeded per well after 5to8daysofculture(n ¼ 10 per condition). , P < 0.05. B and C, numbers indicate the percentage of cells in each quarter.

for both CD44 and ALDH activity and were negative for Hoechst- Hoechst-33342 stain efflux properties in specific experimental 33342 (Supplementary Fig. S1). When tumorspheres became conditions. In these particular experimental conditions, we found bigger (after 10–15 days), only a fraction of cells remained a consistent proportion of Hoechst-SP cells in MKN45 cell line þ þ CD44 ALDH and Hoechst-33342, which may correspond to (3.9% 0.8 % of total cells; Fig. 5C) but not in MKN74 cell line þ þ CSCs, with the appearance of CD44 ALDH Hoechst-33342 (<0.2%, data not shown) as previously reported by others (28). cells which may correspond to progenitor/transit amplifying The percentage of MKN45 SP cells was significantly decreased with cells (Supplementary Fig. S1). verapamil and reserpine compared with untreated cells (0.6% An important property of CSCs is to be resistant to conven- 0.1% and 1.2% 0.2%, respectively, vs. 3.9% 0.8%; Fig. 5C). tional therapies, leading to tumor recurrence and metastasis after The determination of ALDH activity within SP cells was then treatment (8, 9). Verapamil treatment, known to inhibit drug assessed in the experimental conditions of the ALDEFLUOR assay, þ efflux systems, restored Hoechst-33342 incorporation in ALDH those of the SP cells assay being incompatible and leading to cells in MKN45 and GC10 tumorspheres in vitro (Fig. 5A). This ALDEFLUOR substrate clearing (Supplementary Fig. S2; Supple- þ effect was confirmed by flow cytometry analyses on MKN45 mentary Methods). Results showed that most of the ALDH cells and confirmed at a lesser extent with reserpine, another cells were present in the Hoechst-subpopulation of SP-like cells þ inhibitor of drug efflux systems (Fig. 5B). The drug efflux prop- (Fig. 5D). Combined together, these results suggest that ALDH erties of CSCs are usually assessed by the functional analysis of the cells could possess drug efflux properties and may be more Side Population (SP), a minor subpopulation of cells defined by resistant to treatments.

OF8 Clin Cancer Res; 2017 Clinical Cancer Research

Biomarkers of Tumorigenic and Chemoresistant Gastric CSCs

Figure 5. ALDHþ cells have drug efflux properties, and verapamil treatment sensitizes them to chemotherapeutic drugs. A, representative images of fluorescent imaging of CD44 stained with anti–CD44-PE antibodies (in red), ALDH activity detected by ALDEFLUOR reagent (in green), and nuclei staining with Hoechst-33342 (in blue) on 12-day MKN45 tumorspheres and on 5-day GC10 tumorspheres. Tumorspheres were treated or not (control) with verapamil for 10 minutes before staining. Dotted circles point out ALDHþ cells without or with Hoechst-33342 stained nuclei. Bars, 10 mm. B, histogram represents the percentage of Hoechst-33342 cells analyzed by flow cytometry on cells dissociated from tumorspheres (A) treated or not with verapamil 100 mmol/L and reserpine 50 mmol/L (n ¼ 3). C, analysis of the SP and main population (MP) in MKN45 cells treated or not with verapamil 100 mmol/L and reserpine 50 mmol/L and incubated with Hoechst-33342 in HBSS-2% FCS for 60 minutes (n ¼ 3). D, dot-plot analyses of ALDH activity in Hoechst– (SP-like) and Hoechstþ (MP-like) cells, detected after 30-minute incubation in ALDEFLUOR buffer at 37C, then 30-minute incubation with Hoechst-33342 at room temperature (n ¼ 3). E, percentage of viable ALDHþ (black bars) and ALDH (white bars) FACS-sorted MKN45 and MKN74 cells after 48 hours of adherent culture and treatment without (control, CT) or with 10 mmol/L verapamil with or without 50 mmol/L 5-fluorouracil (5-FU), 1 mmol/L doxorubicin (DOXO), or 50 mmol/L cisplatin. B–E, results represent the mean SD. F and G, plots (min to max) represent the number of MKN45 and MKN74 tumorspheres formed: F, after a 48-hour treatment of 5-day tumorspheres without (control, CT) or with verapamil (dotted bars), 5-FU, DOXO, and cisplatin as in E; G, after 5 days by cells dissociated from residual-treated tumorspheres (from experiment described in F). E–G, 8 < n < 10 per condition. , P < 0.05.

To explore this hypothesis, first, the sensitivity of MKN45 and PDXs, given that these cells are not cultivable in adherent cul- þ MKN74 cells to drugs commonly used in gastric carcinoma ture conditions and that only ALDH cells and not ALDH cells treatment was determined on cell viability in adherent culture can form tumorspheres in vitro (Fig. 3C). In both cell lines, þ conditions. Both cell lines showed dose-dependent antiproli- ALDH cells were more resistant than ALDH cells to both ferative responses to 5-fluorouracil, doxorubicin, and cisplatin 5-fluorouracil and doxorubicin treatments but not to cisplatin treatments, which were increased in the presence of verapamil at the dose studied (Fig. 5E). Verapamil treatment sensitized þ (Supplementary Fig. S3). Second, MKN45 and MKN74 cells ALDH cells to these chemotherapies (Fig. 5E). This effect was were sorted by FACS based on ALDH activity, and the viability confirmed on the formation of tumorsphere by both cell lines, þ of ALDH and ALDH cells was evaluated in the presence of in which verapamil treatment potentiated significantly the 5-fluorouracil, doxorubicin, and cisplatin in combination with reduction of tumorsphere number in response to 5-fluorouracil, verapamil in vitro. This assay cannot be performed on cells from doxorubicin, and cisplatin treatments (Fig. 5F). The same result

www.aacrjournals.org Clin Cancer Res; 2017 OF9

Nguyen et al.

þ was obtained on tumorspheres of GC10 PDX (Supple- CD44 cells (see Fig. 3B) that possess the CSC properties, i.e., mentary Fig. S4). Self-renewal assays with residual cells from to generate a new heterogeneous tumor in vivo and tumorsphere MKN45- and MKN74-treated tumorspheres confirmed that in vitro (Figs. 3C and 4). the combination of verapamil with these conventional chemo- The ALDEFLUOR assay used to isolate CSCs in liquid and therapeutic drugs significantly reduced the number of tumori- solid tumors detects the activity of several isoforms of ALDH genic CSCs (Fig. 5G). Altogether, these results indicate that, (40). Among them, the main isoforms expressed in tumors are þ within the CD44 subpopulation of tumor cells, the determi- retinaldehyde dehydrogenases, ALDH1A1 and ALDH1A3, nation of ALDH activity allows the detection and isolation of responsible for the oxidation of retinal to retinoic acid and, CSCs with tumorigenic and chemoresistant properties in gastric to a lesser extent, ALDH3A1 (20, 40). They can metabolize and carcinoma. detoxify chemotherapeutic agents such as cyclophosphamide in hematopoietic stem cells, and their level of expression was shown to be predictive of response to treatment in breast cancer þ Discussion (41, 42). In this study, we show that ALDH cells were more In this study, we characterized the expression of cell surface resistant to treatment with conventional chemotherapeutic biomarkers and ALDH activity of gastric CSCs in intestinal- and drugs than ALDH cells. We adapted the detection of ALDH diffuse-type noncardia gastric carcinomas. activity with the ALDEFLUOR reagent to fluorescent microsco- þ Contrary to the situation in colon cancer in which CD133 py on live cells. Using both methods, we showed that these þ cells containing colon CSCs represent a small and rare sub- ALDH cells did not incorporate the vital DNA dye Hoechst- population within tumors (13, 14), we demonstrated that 33342 instead the ALDH cells incorporated it, confirming that þ þ CD133 cells (detected by similar experimental procedures) ALDH cells may correspond to the SP of cells with CSCs were frequent in gastric carcinoma. We also showed that CD133 properties as previously described by Fukuda and colleagues þ was a less specific marker for the enrichment of gastric CSCs in gastric carcinoma cell lines (28). The ability of ALDH cells than CD44 and ALDH, as demonstrated by their lower capacity to efflux Hoechst-33342 and to resist conventional chemother- to form tumorspheres in vitro and a new tumor after xenograft apy was reversed by verapamil or reserpine treatment, two in vivo. CD44 was expressed in all patient-derived primary inhibitors of efflux pumps such as the ATP-binding cassette gastric carcinomas but not in the healthy gastric mucosa or at (ABC) transporters family members, confirming that these cells averylowlevelintheisthmusofthecorpusgastricglands are associated with chemotherapy resistance as proposed in where stem cells reside. We and others previously reported that other cancers (43). In this study, we did not find a noticeable þ these CD44 stem/progenitor cells expand from the isthmus coexpression of BCRP (ABCG2) and MDR-1 (not expressed), þ toward the base of the unit in metaplastic and dysplastic areas the two leaders of the ABC transporters family, in ALDH induced in response to chronic H. pylori infection (24, 26, 33). gastric CSCs (flow cytometry and qRT-PCR analyses, data not This occurs via an epithelial–mesenchymal-like transition, con- shown), suggesting that the Hoechst-33342 and drug efflux þ ferring CSC-like properties to CD44 cells (25, 26). CD44 may result from the activity of other members of the ABC expression has been reported in gastric carcinoma (34–37). transporters family. This family includes at least 49 genes A recent study demonstrated that CD44 inhibition by peptide groupedinto7families,andatleast 16 of these proteins have inhibitors prevented the development of cellular hyperproli- been implicated in cancer drug resistance (44). feration and chronic atrophic gastritis in animal models of A limit to the use of ALDH as a biomarker of chemoresistant H. pylori–induced gastric carcinogenesis (33, 38). gastric CSC is that ALDH activity can be detected only by the Interestingly, we showed that CD166 was coexpressed with ALDEFLUOR assay on live cells by flow cytometry or fluorescent þ CD44, and, as a consequence, CD166 cells presented the same microscopy analyses. So, its detection as a biomarker of CSCs in þ tumorigenic properties as CD44 cells in vitro. Similar analyses of current practice on patients' specimens may be possible for in vitro coexpression with CD44 and tumorigenic properties led to circulating cancer cells and liquid cancers such as leukemia but the conclusion that the gastric CSC phenotype corresponds to remain elusive for the analysis of solid tumors such as gastric þ þ þ þ þ EPCAM , CD24 , CD133 , CD73 , CD90 , CD105 , CD166 , carcinoma. These findings also imply that ALDH isozymes can be þ and CD44 , associated with ALDH activity. Finally, in all PDXs considered not only as biomarkers of CSCs but also as putative studied, cells with ALDH activity represented the smallest sub- targets to inhibit tumor growth and to overcome resistance to population of cells compared with all other markers studied, with cancer therapy. high tumorigenic properties both in vitro and in vivo, and with It is of importance to note that gastric carcinoma PDXs asymmetric division and differentiation properties reproducing always remained heterogeneous and composed of tumor cell the heterogeneity of the initial tumors. As for other cancers, we subpopulations expressing EPCAM, CD24, CD133, and CD44, must consider that the gastric CSC phenotype may be plastic, similar to the patients' situation, whereas gastric carcinoma cell subjected to regulation by the surrounding tumor microenviron- lines were found to be negative for CD133 and either positive ment in vivo. Recent work has demonstrated that breast CSCs or negative for CD44 and others markers including ALDH. coexist between two different phenotypic states: a more quiescent These results strengthen the importance and the necessity to and invasive, mesenchymal-like state characterized by a CD24 study CSC on models as close as possible to the patients' þ CD44 phenotype and located mainly at the tumor periphery situation, as is the case in this study, and not only on cancer and invasive front, and a more proliferative epithelial-like cell lines. state, characterized by ALDH activity and located more centrally Currently, PDXs represent the most pertinent preclinical (9, 39). However, unlike the situation in breast cancer, we have model to study the capacity of CSCs to give rise to tumor growth, shown that gastric CSCs express both CD44 and ALDH activity, heterogeneity, and sensitivity/resistance to new treatment strate- and that ALDH activity reveals a subpopulation within the gies. However, PDXs models present some limitations,

OF10 Clin Cancer Res; 2017 Clinical Cancer Research

Biomarkers of Tumorigenic and Chemoresistant Gastric CSCs

particularly because the contribution of the patient's tumor Analysis and interpretation of data (e.g., statistical analysis, biostatistics, microenvironment—including inflammation, CSCs niche within computational analysis): P.H. Nguyen, J. Giraud, P. Dubus, L. Wittkop, the organ of origin, and cross-talk with immune and stromal C. Varon — Writing, review, and/or revision of the manuscript: J. Giraud, P. Dubus, cells cannot be taken into account on CSCs plasticity, tumor L. Wittkop, F. Megraud, C. Varon progression, and metastasis. These limits are partly illustrated here Administrative, technical, or material support (i.e., reporting or organizing by the low tumor engraftment success of patient's gastric carci- data, constructing databases): L. Chambonnier, N. Senant-Dugot noma samples, being only approximately 20% of engraftment Study supervision: S. Evrard, F. Megraud, C. Varon success, as described by others for other type of cancers, unveiling the contribution of uncontrolled microenvironment parameters Acknowledgments for cancer propagation in the patient. Nevertheless, there is an We thank Marie-Edith Lafon (CNRS UMR 5234, University of Bordeaux) and fi technicians from the Department of Tumor Pathology (Haut-Leveque Hospital, urgent unmet need of new, more ef cient and better tolerated University Hospital Center of Bordeaux) for molecular analyses on tumor therapeutic strategies for gastric carcinomas, which could focus on tissues, Pierre Costet (animal facilities, University of Bordeaux), Vincent Pitard gastric CSCs. and Santiago Gonzalez (Flow Cytometry and FACS Platform, University of In this study, the development of original PDXs models Bordeaux), Philippe Brunet de la Grange (CNRS UMR5164 CIRID, University allowed us to demonstrate that tumorigenic and chemoresistant of Bordeaux) for assistance on SP cells analyses, and Alban Giese (Experimental gastric CSCs coexpress EPCAM, CD133, CD166, CD44, and Pathology Platform of the Canceropole GSO and SIRIC BRIO, University of fi Bordeaux), Elodie Siffre, and Lucie Benejat (INSERM U853) for technical ALDH, ALDH activity being the most speci c biomarker of CSC assistance. enrichment before CD44 in both diffuse- and intestinal-type fi noncardia gastric carcinomas. This nding led to the hypothesis Grant Support that treatment strategy for noncardia gastric carcinomas can focus fi þ þ This study was nancially supported by the French "Association pour la on CD44 ALDH CSCs, independently of the histologic classi- Recherche contre le Cancer" (grant number 8412), the "Institut National fication of the tumor. du Cancer" (grant 07/3D1616/IABC-23-12/NC-NG and grant 2014-152), the "Conseil Regional d'Aquitaine" (grant numbers 20071301017 and 20081302203), the French National Society for Gastroenterology, and the Disclosure of Potential Conflicts of Interest Canceropole Grand Sud-Ouest (grant 2010-08-canceropole GSO-Universite L. Wittkop is a consultant/advisory board member for Bristol-Myers Squibb. Bordeaux 2). This project was also supported by SIRIC BRIO (Site de No potential conflicts of interest were disclosed by the other authors. Recherche Integree sur le Cancer – Bordeaux Recherche Integree Oncologie; grant INCa-DGOS-Inserm 6046). The costs of publication of this article were defrayed in part by the Authors' Contributions payment of page charges. This article must therefore be hereby marked advertisement Conception and design: P. Dubus, F. Mazurier, C. Varon in accordance with 18 U.S.C. Section 1734 solely to indicate Development of methodology: P.H. Nguyen, P. Dubus, C. Varon this fact. Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Giraud, P. Dubus, G. Belleannee, D. Collet, Received September 9, 2015; revised July 19, 2016; accepted July 31, 2016; I. Soubeyran, S. Evrard, B. Rousseau, F. Mazurier, C. Varon published OnlineFirst September 12, 2016.

References 1. World Health Organization, International Agency for Research on Cancer. 12. Li C, Heidt DG, Dalerba P, Burant CF, Zhang L, Adsay V, et al. Identification Schistosomes, liver flukes, and Helicobacter pylori. IARC monographs on the of pancreatic cancer stem cells. Cancer Res 2007;67:1030–7. evaluation of carcinogenic risks to humans. Lyon, France: IARC; 1994. Vol. 13. Collins AT, Berry PA, Hyde C, Stower MJ, Maitland NJ. Prospective 61. p. 177–240. identification of tumorigenic prostate cancer stem cells. Cancer Res 2. Megraud F, Bessede E, Varon C. Helicobacter pylori infection and gastric 2005;65:10946–51. carcinoma. Clin Microbiol Infect 2015;21:984–90. 14. Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, et al. 3. Lauren P. The two histological main types of gastric carcinoma: Diffuse and Identification of human brain tumour initiating cells. Nature 2004;432: so-called intestinal-type carcinoma. An attempt at a histo-clinical classifi- 396–401. cation. Acta Pathol Microbiol Scand 1965;64:31–49. 15. O'Brien CA, Pollett A, Gallinger S, Dick JE. A human colon cancer cell 4. Flejou JF. [WHO Classification of digestive tumors: The fourth edition]. capable of initiating tumour growth in immunodeficient mice. Nature Ann Pathol 2011;31:S27–31. 2007;445:106–10. 5. Luis M, Tavares A, Carvalho LS, Lara-Santos L, Araujo A, de Mello RA. 16. Ricci-Vitiani L, Lombardi DG, Pilozzi E, Biffoni M, Todaro M, Peschle C, Personalizing therapies for gastric cancer: Molecular mechanisms and et al. Identification and expansion of human colon-cancer-initiating novel targeted therapies. World J Gastroenterol 2013;19:6383–97. cells. Nature 2007;445:111–5. 6. TCGARN. Comprehensive molecular characterization of gastric adenocar- 17. Ginestier C, Hur MH, Charafe-Jauffret E, Monville F, Dutcher J, Brown cinoma. Nature 2014;513:202–9. M, et al. ALDH1 is a marker of normal and malignant human mammary 7. Wang K, Yuen ST, Xu J, Lee SP, Yan HH, Shi ST, et al. Whole-genome stem cells and a predictor of poor clinical outcome. Cell Stem Cell sequencing and comprehensive molecular profiling identify new driver 2007;1:555–67. mutations in gastric cancer. Nat Genet 2014;46:573–82. 18. Jiang F, Qiu Q, Khanna A, Todd NW, Deepak J, Xing L, et al. Aldehyde 8. Clevers H. The cancer stem cell: Premises, promises and challenges. dehydrogenase 1 is a tumor stem cell-associated marker in lung cancer. Nat Med 2011;17:313–9. Mol Cancer Res 2009;7:330–8. 9. Brooks MD, Burness ML, Wicha MS. Therapeutic implications of cellular 19. Huang EH, Hynes MJ, Zhang T, Ginestier C, Dontu G, Appel- heterogeneity and plasticity in breast cancer. Cell Stem Cell 2015;17:260–71. man H, et al. Aldehyde dehydrogenase 1 is a marker for normal 10. Zhu Y, Luo M, Brooks M, Clouthier SG, Wicha MS. Biological and clinical and malignant human colonic stem cells (SC) and tracks SC significance of cancer stem cell plasticity. Clin Transl Med 2014;3:32. overpopulation during colon tumorigenesis. Cancer Res 2009;69: 11. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. 3382–9. Prospective identification of tumorigenic breast cancer cells. Proc Natl 20. Ma I, Allan AL. The role of human aldehyde dehydrogenase in normal and Acad Sci U S A 2003;100:3983–8. cancer stem cells. Stem Cell Rev 2011;7:292–306.

www.aacrjournals.org Clin Cancer Res; 2017 OF11

Nguyen et al.

21. Takaishi S, Okumura T, Tu S, Wang SS, Shibata W, Vigneshwaran R, et al. metaplastic gastric epithelial progenitor cell proliferation. J Biol Chem Identification of gastric cancer stem cells using the cell surface marker 2013;288:16085–97. CD44. Stem Cells 2009;27:1006–20. 34. Yasui W, Kudo Y, Naka K, Fujimoto J, Ue T, Yokozaki H, et al. Expression of 22. Rocco A, Liguori E, Pirozzi G, Tirino V, Compare D, Franco R, et al. CD133 CD44 containing variant exon 9 (CD44v9) in gastric adenomas and and CD44 cell surface markers do not identify cancer stem cells in primary adenocarcinomas: Relation to the proliferation and progression. Int J human gastric tumors. J Cell Physiol 2012;227:2686–93. Oncol 1998;12:1253–8. 23. Houghton J, Stoicov C, Nomura S, Rogers AB, Carlson J, Li H, et al. Gastric 35. Chen JQ, Zhan WH, He YL, Peng JS, Wang JP, Cai SR, et al. Expression of cancer originating from bone marrow-derived cells. Science 2004;306: heparanase gene, CD44v6, MMP-7 and nm23 protein and their relation- 1568–71. ship with the invasion and metastasis of gastric carcinomas. World 24. Varon C, Dubus P, Mazurier F, Asencio C, Chambonnier L, Ferrand J, et al. J Gastroenterol 2004;10:776–82. Helicobacter pylori infection recruits bone marrow-derived cells that 36. da Cunha CB, Oliveira C, Wen X, Gomes B, Sousa S, Suriano G, et al. De participate in gastric preneoplasia in mice. Gastroenterology 2012;142: novo expression of CD44 variants in sporadic and hereditary gastric cancer. 281–91. Lab Invest 2010;90:1604–14. 25. Bessede E, Dubus P, Megraud F, Varon C. Helicobacter pylori infection and 37. Lau WM, Teng E, Chong HS, Lopez KA, Tay AY, Salto-Tellez M, et al. stem cells at the origin of gastric cancer. Oncogene 2015;34:2547–55. CD44v8–10 is a cancer-specific marker for gastric cancer stem cells. Cancer 26. Bessede E, Staedel C, Acuna Amador LA, Nguyen PH, Chambonnier L, Res 2014;74:2630–41. Hatakeyama M, et al. Helicobacter pylori generates cells with cancer 38. Bertaux-Skeirik N, Feng R, Schumacher MA, Li J, Mahe MM, Engevik AC, stem cell properties via epithelial-mesenchymal transition-like changes. et al. CD44 plays a functional role in helicobacter pylori-induced epithelial Oncogene 2014;33:4123–31. cell proliferation. PLoS Pathog 2015;11:e1004663. 27. Ferrand J, Noel D, Lehours P, Prochazkova-Carlotti M, Chambonnier L, 39. Liu S, Cong Y, Wang D, Sun Y, Deng L, Liu Y, et al. Breast cancer stem cells Menard A, et al. Human bone marrow-derived stem cells acquire transition between epithelial and mesenchymal states reflective of their epithelial characteristics through fusion with gastrointestinal epithelial normal counterparts. Stem Cell Reports 2014;2:78–91. cells. PLoS One 2011;6:e19569. 40. Marcato P, Dean CA, Giacomantonio CA, Lee PW. Aldehyde dehydroge- 28. Fukuda K, Saikawa Y, Ohashi M, Kumagai K, Kitajima M, Okano H, et al. nase: Its role as a cancer stem cell marker comes down to the specific Tumor initiating potential of side population cells in human gastric cancer. isoform. Cell Cycle 2011;10:1378–84. Int J Oncol 2009;34:1201–7. 41. Magni M, Shammah S, Schiro R, Mellado W, Dalla-Favera R, Gianni AM. 29. Bachelard-Cascales E, Chapellier M, Delay E, Pochon G, Voeltzel T, Induction of cyclophosphamide-resistance by aldehyde-dehydrogenase Puisieux A, et al. The CD10 enzyme is a key player to identify and regulate gene transfer. Blood 1996;87:1097–103. human mammary stem cells. Stem Cells 2010;28:1081–8. 42. Sladek NE, Kollander R, Sreerama L, Kiang DT. Cellular levels of aldehyde 30. Dalerba P, Dylla SJ, Park IK, Liu R, Wang X, Cho RW, et al. Phenotypic dehydrogenases (ALDH1A1 and ALDH3A1) as predictors of therapeutic characterization of human colorectal cancer stem cells. Proc Natl Acad Sci responses to cyclophosphamide-based chemotherapy of breast cancer: a U S A 2007;104:10158–63. retrospective study. Rational individualization of oxazaphosphorine- 31. Shackleton M, Vaillant F, Simpson KJ, Stingl J, Smyth GK, Asselin-Labat based cancer chemotherapeutic regimens. Cancer Chemother Pharmacol ML, et al. Generation of a functional mammary gland from a single 2002;49:309–21. stem cell. Nature 2006;439:84–8. 43. Raha D, Wilson TR, Peng J, Peterson D, Yue P, Evangelista M, et al. 32. Dominici M, Le Blanc K, Mueller I, Slaper-Cortenbach I, Marini F, Krause D, Thecancerstemcellmarkeraldehyde dehydrogenase is required to et al. Minimal criteria for defining multipotent mesenchymal stromal cells. maintain a drug-tolerant tumor cell subpopulation. Cancer Res The International Society for Cellular Therapy position statement. 2014;74:3579–90. Cytotherapy 2006;8:315–7. 44. Di C, Zhao Y. Multiple drug resistance due to resistance to stem cells 33. Khurana SS, Riehl TE, Moore BD, Fassan M, Rugge M, Romero-Gallo J, and stem cell treatment progress in cancer (Review). Exp Ther Med et al. The hyaluronic acid receptor CD44 coordinates normal and 2015;9:289–93.

OF12 Clin Cancer Res; 2017 Clinical Cancer Research

Characterization of Biomarkers of Tumorigenic and Chemoresistant Cancer Stem Cells in Human Gastric Carcinoma

Phu Hung Nguyen, Julie Giraud, Lucie Chambonnier, et al.

Clin Cancer Res Published OnlineFirst September 12, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-15-2157

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2016/09/10/1078-0432.CCR-15-2157.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2017/01/11/1078-0432.CCR-15-2157. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.