Activating Transcription Factor 4 Modulates Tgfβ-Induced Aggressiveness in Triple-Negative Breast Cancer Via SMAD2/3/4 and Mtorc2 Signaling

Published OnlineFirst July 16, 2018; DOI: 10.1158/1078-0432.CCR-17-3125

Biology of Human Tumors Clinical Cancer Research Activating Transcription Factor 4 Modulates TGFb-Induced Aggressiveness in Triple-Negative Breast Cancer via SMAD2/3/4 and mTORC2 Signaling Adrian Gonzalez-Gonz alez 1,2, Esperanza Munoz-Muela~ 1,2, Juan A. Marchal3,4, Francisca E. Cara1,2, Maria P. Molina1,2, Marina Cruz-Lozano1,2, Gema Jimenez 3,4, Akanksha Verma5, Alberto Ramírez6, Wei Qian5, Wen Chen6, Anthony J. Kozielski6, Olivier Elemento5, María D. Martín-Salvago7, Rafael J. Luque7, Carmen Rosa-Garrido8, David Landeira2,9, María Quintana-Romero2,9, Roberto R. Rosato6, Maria A. García10, Cesar L. Ramirez-Tortosa7, Hanna Kim11, Cristian Rodriguez-Aguayo12, Gabriel Lopez-Berestein12, Anil K. Sood12, Jose A. Lorente2, Pedro Sanchez-Rovira 1,2, Jenny C. Chang6, and Sergio Granados-Principal1,2,8

Abstract

Purpose: On the basis of the identified stress-independent talk between ATF4 and TGFb was identified. ATF4 expres- cellular functions of activating transcription factor 4 (ATF4), sion inhibition reduced migration, invasiveness, mam- we reported enhanced ATF4 levels in MCF10A cells treated mosphere-forming efficiency, proliferation, epithelial– with TGFb1. ATF4 is overexpressed in patients with triple- mesenchymal transition, and antiapoptotic and stemness negative breast cancer (TNBC), but its impact on patient marker levels. In PDX models, ATF4 silencing decreased survival and the underlying mechanisms remain unknown. metastases, tumor growth, and relapse after chemother- We aimed to determine ATF4 effects on patients with breast apy. ATF4 was shown to be active downstream of cancer survival and TNBC aggressiveness, and the relationships SMAD2/3/4 and mTORC2, regulating TGFb/SMAD and between TGFb and ATF4. Defining the signaling pathways may mTOR/RAC1–RHOA pathways independently of stress. help us identify a cell signaling–tailored gene signature. We defined an eight-gene signature with prognostic Experimental Design: Patient survival data were deter- potential, altered in 45% of 2,509 patients with breast mined by Kaplan–Meier analysis. Relationship between TGFb cancer. and ATF4, their effects on aggressiveness (tumor proliferation, Conclusions: ATF4 may represent a valuable prognostic metastasis, and stemness), and the underlying pathways were biomarker and therapeutic target in patients with TNBC, and analyzed in three TNBC cell lines and in vivo using patient- we identified a cell signaling pathway–based gene signature derived xenografts (PDX). that may contribute to the development of combinatorial Results: ATF4 overexpression correlated with TNBC targeted therapies for breast cancer. Clin Cancer Res; 1–13. patient survival decrease and a SMAD-dependent cross- 2018 AACR.

Introduction of breast cancer, with a poor survival rate. TNBC is characterized Breast cancer is the most commonly diagnosed type of cancer in by high proliferation, heterogeneity, metastases, drug resistance, women and it is associated with high incidence and death rates and incidence of relapse, and enriched in aggressiveness-related (1, 2). Triple-negative breast cancer (TNBC) is an estrogen (ER), signaling pathways such as TGFb or mTOR. Currently, no progesterone, and HER2 receptor-negative, very aggressive form approved targeted therapies exist for its treatment (2, 3). New

1UGC de Oncología Medica, Complejo Hospitalario de Jaen, Jaen, Spain. Interference and Non-Coding RNA, The University of Texas MD Anderson 2GENYO, Centre for Genomics and Oncological Research, Granada, Spain. Cancer Center, Houston, Texas. 3Instituto de Investigacion Biosanitaria de Granada (ibs.GRANADA), Uni- versidad de Granada, Granada, Spain. 4Biopathology and Regenerative Note: Supplementary data for this article are available at Clinical Cancer Medicine Institute (IBIMER), Centre for Biomedical Research (CIBM), Uni- Research Online (http://clincancerres.aacrjournals.org/). versity of Granada, Granada, Spain. 5Institute for Computational Biomedi- ~ cine, Weill Cornell Medical College, New York, New York. 6Houston Methodist A. Gonzalez-Gonzalez and E. Munoz-Muela contributed equally to this article. Cancer Center, Houston Methodist Hospital, Houston, Texas. 7UGC de Ana- Corresponding Author: Sergio Granados-Principal, Complejo Hospitalario de tomía Patologica, Complejo Hospitalario de Jaen, Jaen, Spain. 8FIBAO. Jaen, Jaen 23007, Spain. Phone: 346-5155-7921; Fax: þ34 958 637 071. Complejo Hospitalario de Jaen, Servicio Andaluz de Salud, Jaen, Spain. E-mail: [email protected] 9Department of Biochemistry and Molecular Biology II, Faculty of Pharmacy, 10 University of Granada, Granada, Spain. Department of Oncology, Virgen de doi: 10.1158/1078-0432.CCR-17-3125 las Nieves University Hospital, Granada, Spain. 11Texas Tech University Health Sciences Center, School of Pharmacy, Amarillo, Texas. 12Center for RNA 2018 American Association for Cancer Research.

www.aacrjournals.org OF1

Gonzalez-Gonz alez et al.

Translational Relevance Materials and Methods Tumor heterogeneity, metastases, and drug resistance define Supplementary Information includes Supplementary Materials the aggressiveness and poor survival rates of triple-negative and Methods, Supplementary Figure Legends, and Supplemen- breast cancer (TNBC). ATF4 is overexpressed in breast cancer tary Tables. and TNBC, but its impact on patient survival remains unclear. We demonstrated that ATF4 expression correlates with lower Bioinformatic analysis – overall and relapse-free survival rates in patients with breast Using the Kaplan Meier plotter (www.kmplot.com/analysis), cancer and TNBC. ATF4 has growth factor–dependent func- the effects of query genes on survival were assessed using tions, which remain unclear in breast cancer. We showed in 5,143 samples from patients with breast cancer. Gene ¼ vitro and in vivo that ATF4 depletion leads to the metastasis rate, expression, relapse-free (RFS; n 3,951) and overall survival ¼ cancer stemness, and tumor cell survival reduction through the (OS) data (n 1402) were obtained from Gene Expression modulation of TGFb/SMAD and PI3K/mTOR pathways and Omnibus, European Genome-phenome Archive, and The identified a pathway-guided gene signature with prognostic Cancer Genome Atlas (24). Correlations, genomic and tran- potential. Differential outcomes of patients of the same cancer scriptomic alterations, and their impact on patient survival – subtype, treated with the same therapies, demonstrate that were studied using OncoPrint and Kaplan Meier analyses of novel biomarkers and therapeutic targets are required for 2,509 patients with breast cancer (25) by using cBioPortal the personalized treatment approach. Our findings suggest database (26, 27). that ATF4 may serve as a prognostic biomarker and therapeutic target in patients with TNBC. Human tissue samples Paraffin-embedded tissue from patients with TNBC (n ¼ 35), with pathologic information and follow-up, no previous chemo or radiotherapy, and with previous written or improved targeted therapies, patient stratification into respon- informed consents signed by all patients, were obtained sive-to-treatment subgroups using novel prognostic biomarkers, from the Jaen's node of the Biobank of the Public Health and the identification of new therapeutic targets are required to System of Andalusia (Complejo Hospitalario de Jaen, ensure an effective personalized therapy (1). Spain). All samples and procedures were approved by the Under stress conditions, including hypoxia, nutrient depri- Ethical Committee for the Research of Jaen and were con- vation, or endoplasmic reticulum stress (ERS), the integrated ducted in accordance with the Declaration of Helsinki and stress response (ISR) is activated in cells to preserve homeo- International Ethical Guidelines for Biomedical Research stasis. The activation of the ISR leads to the global protein Involving Human Subjects (CIOMS). synthesis reduction through the eukaryotic translation initiation factor 2 alpha (eIF2a, EIF2S1) phosphorylation, driving ATF4 IHC and scoring in TNBC patients' tumor tissue the translation-regulated activation of activating transcription Tumor tissue was stained for ATF4 (Abcam; ab28830) at 1:50 factor 4 (ATF4) that regulates cell fate. eIF2a phosphorylation dilution as reported (28). ATF4 was assessed blindly by three is initiated by protein kinase-like endoplasmic reticulum kinase different pathologists. Both staining intensity and extent in neo- (PERK, EIF2AK3), general control nonderepressible 2 (GCN2, plastic cells were considered by using semiquantitative scores: (i) EIF2AK4), protein kinase double stranded RNA-dependent staining intensity (granular, cytoplasmic) was graded as 0: no (PKR, EIF2AK2),andheme-regulatedinhibitor(HRI,EIF2AK1) staining; 1þ: weak, 2þ: moderate, 3þ: intense (Fig. 1B). (ii) in response to the ERS, amino acid deprivation, viral infection, staining extent was assigned with a value of 0 to 3 by the and heme-deficiency, respectively (4, 5). following criteria based on % of stained tumor cells: 0–25% ¼ ATF4 is a transcription factor belonging to the ATF/cyclic 0; 26–50% ¼ 1; 51–75% ¼ 2; 76–100% ¼ 3. Finally, an integrated adenosine monophosphate response element binding protein score was obtained by ponderation of the results as follows: values (ATF/CREB) family, overexpressed in tumors, including breast of staining extent were multiplied by the value of its correspond- cancer and TNBC (6–8). ATF4 regulates tumor growth, autop- ing intensity score. Therefore, score 0 was multiplied by 0, score hagy, drug resistance, and metastasis during ISR through PERK 1þ was multiplied by 1, score 2þ was multiplied by 2, and score and GCN2 pathways (9–17). Independent of the cellular stress, 3þ was multiplied by 3. The sum of these values (from 0 to 7) was ATF4 regulates cell metabolism (8, 18, 19), osteoblast differen- the final score. Example: negative ¼ 10%, 1þ¼50%, 2þ¼30%, tiation (20), drug resistance (21), invasion, and metastasis in 3þ¼10% are assigned with the values 0, 1, 1, 0, respectively. ATF4 esophageal squamous cell carcinoma (22). In the absence of score: (00)þ(11)þ(12)þ(03) ¼ 3. stress, high ATF4 levels correlate with poorer cancer patient survival rate (22). We previously reported increased ATF4 expres- Cell culture sion in the unstressed MCF10A cells treated with TGFb1 (23), TNBC cell lines, MDA-MB-231 and BT549, were purchased indicating a potential TGFb-mediated stress-independent control from the ATCC, whereas SUM159PT cells were obtained from of ATF4 activity. Asterand Bioscience. SBE (SMAD binding element) reporter- Because of these reports, we investigated whether ATF4 can HEK293 (SBE-HEK293) cell line was purchased from BPS regulate the TGFb-induced aggressiveness of TNBC and affect Bioscience. All cells were maintained in DMEM (Gibco) sup- patient survival. The identification of the relevant signaling path- plemented with 10% FBS (Thermo Fisher Scientific) and 1% way may facilitate the design of combinatorial targeted therapies antibiotic–antimycotic (Gibco). SBE-HEK293 cells were cul- and provide a gene signature that may improve personalized tured under Geneticin selection (Sigma), following the manu- medicine in breast cancer. facturer's instructions.

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

siRNA-mediated knockdown (intensity 1þ; Fig. 1B). Our results showed that patients with a The cells were transiently transfected with siRNAs targeting score 1 (determined by ROC curve analysis and considered as ATF4 (25 nmol/L), SMAD2/3, SMAD4, PERK, PKR, GCN2, HRI, positive cases for the Kaplan–Meier analysis; Supplementary eIF2a (EIF2S1), RPTOR, RICTOR, TAK1 (MAP3K7), and RAS (50 Fig. S1B) had less OS after diagnosis (37 month) than patients nmol/L) using Lipofectamine RNAiMAX (Invitrogen). TGFb1 (10 with score <1 (considered as negative cases; 46 months) starting at ng/mL) was added 48-hour posttransfection, and the samples a 24-months follow-up (Fig. 1C); however, it was not significant were incubated for 24 or 72 hours, depending on the experiment. (P ¼ 0.125). We previously reported increased ATF4 levels in TGFb1-treated Animal experiments MCF10A cells (23). Because TNBC microenvironment is often Patient-derived xenografts. All animal procedures were approved enriched in TGFb ligands (31), we analyzed TGFb activation by the Methodist Hospital Research Institute Animal Care and Use effects on ATF4 expression, and demonstrated that it increases Review Office. Experiments were conducted using two human in BT549 and SUM159PT cells treated with TGFb1, which was TNBC-derived patient-derived xenografts (PDX), BCM-4664 and abrogated by the TGFbR1 kinase inhibitor LY2157299 treatment BCM-3887 (basal intrinsic subtype; ref. 29). PDXs were trans- (P < 0.001; Fig. 1D), suggesting that ATF4 represents its down- planted into the cleared mammary fat pad of 4- to 5-week-old stream target. ATF4 expression induced by TGFb1 and thapsi- NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) mice maintained in the gargin was similar in SUM159PT and BT549, and lower in MDA- standard conditions (28). When tumors reached 150 to 200 mm3 MB-231 (Supplementary Fig. S1C). Knockdown experiments in size, the mice were randomly assigned to four treatment groups demonstrated that ATF4 is regulated by SMAD2/3/4 (Fig. 1E). (n ¼ 8/group): (i) noncoding siRNA (SCR), (ii) ATF4-siRNA To analyze whether SMAD2/3 directly regulates ATF4, we ana- (siRNA#2), (iii) SCR plus docetaxel (ChemoþSCR, 20 mg/kg), lyzed the human ATF4 promoter region for SBEs and found the and (iv) siRNA#2 plus docetaxel (ChemoþsiRNA#2, 20 mg/kg) conserved CAGAC, CAGA, GTCT, GGCGC, GGCCG motifs (Sup- groups. siRNAs were injected twice weekly for 6 weeks at plementary Fig. S1D; ref. 32). Careful inspection of ChIP-Seq data 5 mg/mouse, and docetaxel was administered once per week on of SMAD2/3 in BT549 cells treated with TGFb1 for 1.5 hours (33) days 1, 14, and 28. Tumor volumes and body weights were showed specific binding of SMAD2/3 to SERPINE1 and MMP2 recorded every 2 days. Tumors were calipered and volume was (positive controls), ATF4, and the TGFb1 responsive genes ID1, calculated as previously described (30). Mice were euthanized JUN,andCDKN1A promoters, but not to HBB and HPRT1 (neg- 24 hours after the last injection and tumors were collected ative controls; Supplementary Fig. S1E). Further, we carried out for further analyses. For tumor relapse, docetaxel was given at ChIP-qPCR analysis in BT549 cells treated with TGFb1for1.5hours 33 mg/kg dose to BCM-4664-bearing mice, and tumor volume and found that SMAD2/3 bind to the ATF4 promoter, comparable was recorded until the appearance of morbidity, loss of 20% of to the SERPINE1 and MMP2 promoters (Fig. 1F), what suggests that body weight, or when tumors reached 2 cm3 in size. The metastatic SMAD2/3 can bind and regulate the ATF4 gene transcription. PDX model of TNBC is detailed in Supplementary Materials To ascertain the importance of ATF4 effects on the TGFb and Methods. pathway, we inhibited ATF4 expression and SBE activity was tested. The most effective siRNA sequence, siRNA#2 (Supplemen- Statistical analysis tary Fig. S2A), decreased SBE activity in HEK-293 cells (Fig. 1G), Differences between two groups were analyzed by two-tailed and phosphorylated (p)-SMAD2/3, SMAD2/3, and SMAD4 levels Student t test. Correlation between ATF4 staining and OS after in BT549 and SUM159PT cells (Fig. 1H), indicating a positive diagnosis in tumor tissue of patients with TNBC was analyzed by TGFb feedback. In patients with breast cancer, coexpression of the Kaplan–Meier method and further log-rank test with SPSS ATF4 and the canonical TGFb pathway members correlated with 21.0. Patient samples were stratified by computing all ATF4 scores poorer OS (P ¼ 0.0038; Fig. 1I), with a shift from positive to by a ROC curve analysis, and the best performing threshold was negative effects on survival when coexpressed with SMAD4 or used as a cutoff of positive staining in the analysis. Tumor volume SMAD3 in All_BC group (Fig. 1J). LOOCV results showed that was assessed by two-way ANOVA and Bonferroni's post hoc test. ATF4 overexpression induces a significant decrease in OS (Sup- Median survival posttreatment in mice was analyzed using log- plementary Table S1). rank (Mantel–Cox) test and the hazard ratio with 95% confidence intervals were calculated. Each dead animal was assigned with a ATF4 inhibition suppresses the aggressiveness of TNBC cells number 1, and each surviving mouse with a 0. The last day of ATF4 depletion in the TNBC cells decreased their wound- treatment (day 42) was considered as day 0 for the survival healing ability independently of the treatment with TGFb1 analysis. A P-value <0.05 was considered significant. (Fig. 2A; Supplementary Fig. S2B). According to its capacity to silence ATF4 (Supplementary Fig. S2A), siRNA#2 was more effi- cient to reduce the tumor cell migration. The migration index was Results reduced in BT549, SUM159PT, and MDA-MB-231 cells, with High ATF4 expression, downstream of SMAD2/3/4, correlates TGFb1 (41%, 50%, and 45%, respectively) and without it with lower patient survival (42%, 61%, and 65%, respectively; Fig. 2A). ATF4 knockdown High ATF4 expression was shown to correlate with poorer OS with siRNA#2 in BT549, SUM159PT, and MDA-MB-231 cell (n ¼ 1,402, P ¼ 0.0095) and RFS (n ¼ 3,951, P ¼ 8.4e6) in all lines reduced the number of invading cells with (67%, 50%, and – breast cancer cases (All_BC), and RFS in ER (n ¼ 801, P ¼ 46%, respectively) and without TGFb1 (38%, 23%, and 54%, þ 0.0058), ER (n ¼ 2,061, P ¼ 0.0011), and TNBC (n ¼ 255, respectively; Fig. 2B). In absence of chemoattractant, less number P ¼ 0.016) patients (Fig. 1A; Supplementary Fig. S1A). We next of invading cells was seen upon TGFb1 treatment in BT549 investigated the ATF4 expression in 35 patients with TNBC by IHC and SUM159PT cell lines (50% and 42%, respectively). Such a staining. The frequency of ATF4-positive staining was of 66% decrease was seen in MDA-MB-231 regardless the absence

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Gonzalez-Gonz alez et al.

Figure 1. ATF4 expression correlates with poor patient survival and SMAD-dependent TGFb signaling. A, Kaplan–Meier showing that high ATF4 expression correlates with poorer OS (n ¼ 1,402) and RFS in all breast cancer (All_BC, n ¼ 3,951), estrogen receptor negative (ER–, n ¼ 801) and TNBC patients (n ¼ 255). (Continued on the following page.)

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

(44% decrease) or presence (55% decrease) of TGFb1 in the (P < 0.05; Fig. 5A), whereas the ALDFþ subpopulation number medium. These changes were accompanied with the downregula- decreased in BCM-3887 model (Fig. 5B). In the BCM-4664 model, tion of epithelial–mesenchymal transition (EMT)-related ATF4 inhibition restrained the tumor growth (Fig. 5C) and the transcription factors (ZEB1, TWIST1, SNAIL, and SLUG) in all ALDFþ subpopulation number (Fig. 5D) compared with those in cells after TGFb1 treatment, and TWIST1 and SNAIL without the controls. To investigate tumor relapse after chemotherapy, we TGFb1. N-cadherin levels were decreased in BT549 and co-administered ATF4-siRNA and docetaxel (33 mg/kg) twice SUM159PT cells, but they were not detected in MDA-MB-231 per week for 6 weeks to mice bearing BCM-4664 tumors. Che- cells (Fig. 2C). Cell proliferation diminished after ATF4 knock- moþSCR tumors reached the minimum volume (124 mm3)at down (Fig. 2D), which was followed by a reduction in BCL2 day 24, whereas the regrowth was initiated at day 28 (128 mm3), and MCL1 in these cells (Fig. 2E). showing a 2.4-fold increase at day 38. In contrast, tumor volume Cancer stem cells (CSC) contribute to metastasis, tumor in ChemoþsiRNA#2 mice was 63 mm3 at day 24, and they started growth, and treatment resistance. To determine whether the role to regrow at day 28 (78 mm3), reaching a 1.4-fold increase at day of ATF4 in the TNBC aggressiveness is affected by CSC alterations, 38. At day 56, tumor volume in ChemoþSCR was 2083 mm3, we assessed ATF4 expression in mammospheres as a surrogate whereas it was shown to reach 548 mm3 in ChemoþsiRNA#2 marker of CSCs versus the attached cells. Protein levels were mice (P < 0.001; Fig. 5C). Median survival posttreatment was shown to increase with time and mammosphere generation stage 28 days in ChemoþsiRNA#2 and 15 days in ChemoþSCR (P < (Fig. 3A; Supplementary Fig. S2C). We investigated the effects of 0.0001; Fig. 5E). To confirm that ATF4 targeting was successful, ATF4 depletion on mammosphere-forming efficiency (MSFE), we determined ATF4 expression in BCM-3887 (Fig. 5F) and which was reduced after ATF4 knockdown in all cells (Fig. 3B). BCM-4664 tumors (Fig. 5G). Because ATF4 expression was induced by oxidative stress in suspension cultures (13), we measured the expression levels of ATF4 is a downstream mTORC2 target and is involved in the stemness markers after ATF4 inhibition in the attached cells, to regulation of mTOR/RAC1-RHOA in a stress-independent determine whether our results were due to the modulation of manner stem-like properties or they represent a consequence of detach- To analyze whether TGFb activates ISR-dependent ATF4 expres- ment. NANOG, SOX2, OCT4, and CXCL10 levels were decreased sion, we knocked down PERK, PKR, GCN2, HRI, and eIF2a in BT549 and SUM159PT cells (Fig. 3C). These results were (EIF2S1) in the presence of TGFb1, and demonstrated that their confirmed at protein levels as well, except for CXCL10, which inhibition did not downregulate ATF4 expression consistently was not detected (Fig. 3D). With TGFb1, cleaved NOTCH1, OCT4, across the cell lines (Fig. 6A). PERK, PKR, and GCN2 depletion in and CD44 expression levels were consistently decreased in all SUM159PT cells, and GCN2 in MDA-MB-231, inhibited ATF4, cells. Without TGFb1, NANOG, NOTCH1, and CD44 proteins indicating a cell line-dependent relevance of these pathways on were inhibited by ATF4-siRNA (Fig. 3D). ATF4 expression. Unexpectedly, ATF4 levels were increased upon eIF2a (EIF2S1) knockdown. Interestingly, after PERK knock- ATF4 inhibition reduces metastases, tumor growth, and relapse down, p-eIF2a was inhibited only in BT549 and MDA-MB-231 in the PDX models cells, what did not correlate with a decrease of ATF4. We selected the BCM-3887 and BCM-4664 PDX models for our We inhibited the noncanonical TGFb pathways MEK/ERK, analyses, with high and medium ATF4 expression, respectively, by PI3K, TAK1, and P38-MAPK (34) using the pharmacologic inhi- RNA-Seq and IHC (Fig. 4A and B). To determine the effects of bitors and TGFb1. ATF4 expression was decreased after PI3K and ATF4 silencing on metastases, we used a highly metastatic PDX TAK1 inhibition (Supplementary Fig. S3A). PI3K, mTOR, and model (3887-LM) in mice. After the primary tumor removal, mice SGK1/2 were shown to represent the upstream ATF4 regulators, were treated with DOPC-conjugated ATF4-siRNA#2 and SCR independent of AKT and PDK1 (Supplementary Fig. S3B). A twice weekly for 6 weeks. siRNA#2-treated animals had less second PI3K inhibitor treatment excluded possible inhibitor- metastatic nodules in liver and lungs (P < 0.05; Fig. 4C and dependent off-target effects on ATF4 expression (Supplementary D). Metastatic lesions were confirmed microscopically by Ki67 Fig. S3C). To test whether the crosstalk between TGFb and RAS, staining (Fig. 4E) and they were positive for ATF4 (Fig. 4F). upstream of PI3K, represents the leading signal, we transfected We assessed tumor growth, ALDFþ subpopulation number, BT549 and SUM159PT cells with RAS-siRNA, accompanied or and recurrence following the mouse treatment with ATF4-siRNA not by the treatment with TGFb1. RAS inhibition failed to and/or docetaxel. In vivo ATF4 silencing significantly reduced the decrease ATF4 levels, independent of p-AKT levels (Supplemen- tumor growth alone (P < 0.01) or in combination with docetaxel tary Fig. S3D).

(Continued.) Follow-up threshold was set at 10 years. B, Representative images of negative, 1þ,2þ,and3þ ATF4 staining intensity in TNBC patients' tumor tissue (original optical objective: 20). C, Kaplan–Meier analysis showing the impact of ATF4 staining on the OS after diagnosis of TNBC patients' tumor tissue (n ¼ 35). D, RT-PCR and Western blot analysis of ATF4 in BT549 and SUM159PT cells treated with TGFb1 (10 ng/mL), LY2157299 (5 mmol/L), and combination for 72 hours. Inhibitor was added 1 hour before TGFb1. E, Western blot analysis of ATF4 in BT549 and SUM159PT cells transfected with SMAD2/3 and SMAD4-siRNAs. TGFb1 was added 48 hours after transfection for 24 hours. F, Binding of SMAD2/3 to the ATF4 promoter region in BT549 cells upon TGFb1 treatment for 1.5 hours was assayed by ChIP-qPCR. Values are expressed relative to input for the promoter regions of SMAD2/3 bound-genes (SERPINE1, MMP2), negative controls (LAMB3, HPRT), and ATF4. IgG was used as a nonspeciﬁc binding control. G, SBE reporter assay in SBE-HEK293 cells after ATF4 knockdown with/without TGFb1 for 24 hours. RLU, relative light units. H, Effect of ATF4 knockdown following treatment with TGFb1 for 24 and 72 hours on SMAD2/3 and SMAD4, respectively. Two targeted ATF4-siRNAs (siRNA#1 and siRNA#2) were used in BT549. siRNA#2 was the most efﬁcient and further used in SUM159PT cells. I, Changes in OS of patients with breast cancer when ATF4 (n ¼ 1,402), SMAD2 (n ¼ 626), SMAD3 (n ¼ 1,402), SMAD4 (n ¼ 626), and TGFBR1 (n ¼ 626) are expressed alone or coexpressed. Survival fold change was tested by multiple testing correction; , P < 0.0038. J, Kaplan–Meier showing patient with breast cancer OS when ATF4 is coexpressed with SMAD2, SMAD3,orSMAD4; , P < 0.001.

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Gonzalez-Gonz alez et al.

Figure 2. ATF4 silencing inhibits the metastatic and proliferative properties of tumor cells and correlates with less expression of EMT and prosurvival markers. A, Migration and (B) invasion of BT549, SUM159PT, and MDA-MB-231 cells after ATF4 knockdown treated with/without TGFb1 for 24 hours (MDA-MB-231 for 72 hours). C, Changes in protein expression of EMT markers (N-cadherin, ZEB1, SNAIL, SLUG, TWIST1) after ATF4 silencing in BT549, SUM159PT, and MDA-MB-231 cells with and without TGFb1 for 24 hours. D, Proliferation after ATF4 knockdown with/without TGFb1 for 24 hours (MDA-MB-231 for 72 hours). E, Western blot analysis of prosurvival proteins (BCL2 and MCL1) after transfection with ATF4-siRNA#2 and treatment with TGFb1 for 72 hours. , P < 0.05; , P < 0.01; , P < 0.001.

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

Figure 3. Mammosphere formation is decreased after ATF4 knockdown and correlates with lower stemness markers expression. A, Increased ATF4 protein expression in primary and secondary mammosphere generations (1MS and 2MS, respectively) compared with attached (Att.) cells. B, Mammosphere-forming efﬁciency (MSFE) in three TNBC cell lines after ATF4 inhibition and treatment with TGFb1 for 24 hours. C, mRNA expression of NANOG, SOX2, OCT4, NOTCH1,andCXCL10 after ATF4 knockdown and treatment with TGFb1 for 72 hours in BT549 and SUM159PT cells in adherent conditions. D, Western blot analysis of stemness markers after ATF4 silencing and treatment with TGFb1 for 24 hours (BT549, MDA-MB-231) and 72 hours (SUM159PT). , P < 0.05; , P < 0.01; , P < 0.001.

Rapamycin inhibits mTORC1 and mTORC2 in a dose- and Collectively, our data suggest that ATF4 is involved in and time-dependent manner, together with SGK1 expression, which regulates both the canonical, SMAD2/3/4, and noncanonical, is activated by mTORC2 (35). To determine whether ATF4 is a PI3K/mTORC2/RHOA-RAC1, TGFb signaling pathways to mod- downstream target of mTORC1 and/or mTORC2 with active ulate metastasis, stemness, and tumor cell survival (Fig. 6F). TGFb, TNBC cells were transfected with RPTOR or RICTOR- siRNAs and treated with TGFb1. We observed decreased ATF4 Prognostic potential of a mechanism-based gene signature in levels only following the RICTOR inhibition in all analyzed patients with breast cancer cells (Fig. 6B). Because we showed that SNAIL expression To help improve the prognosis and treatment decision-making considerably decreases after ATF4 knockdown, it was used as in patients with breast cancer, using Kaplan–Meier plotter data- a surrogate for ATF4 inhibition. SNAIL levels were decreased base, we studied the impact of different members of the TGFb/ after RICTOR silencing and TGFb1 treatment in all analyzed SMAD/ATF4 and PI3K/mTOR/ATF4 pathways on the breast can- cells (Fig. 6C). cer patient RFS (Supplementary Table S2) using multivariate mTOR signaling activity is modulated by several feedback analysis and LOOCV. Here, we identified an eight-gene signature, loops (35), and therefore, we analyzed a potential feedback loop including ATF4, TGFBR1, SMAD4, PIK3CA, RPTOR, EIF4EBP1, between ATF4 and mTORC1/2. In 2,509 patients with breast RICTOR, and NDRG1 genes, that predicts a poorer RFS in the cancer, ATF4 expression was shown to correlate with the expres- high-expression cohort of All_BC (61-fold-change decrease; – sion of mTORC1 (EIF4E, R ¼ 0.463; RPS6, R ¼ 0.380) and n ¼ 1,764, P < 0.005), ER (81-fold-change decrease, n ¼ 347, mTORC2 targets (NDRG1, R ¼ 0.213; RHOA, R ¼ 0.320; P < P < 0.005; Fig. 6G), and basal intrinsic subtype (n ¼ 618, P < 0.005; 0.0001; Fig. 6D). The positive feedback between ATF4 and Supplementary Table S3). In All_BC group, this signature predicts mTORC1 and mTORC2 activity was further confirmed by dem- a 27-time poorer RFS compared with that predicted by using ATF4 onstrating that ATF4-siRNA treatment inhibited the downstream expression alone, as the single gene associated with the highest – targets of mTORC2 (p-NDRG1, RHOA, RAC1) and mTORC1 significant decrease in RFS of this group. In the ER group, pathway (p-AKT, p-P70S6K) in SUM159PT and BT549 cells multigene signature predicts a 53-time poorer RFS compared (Fig. 6E). Interestingly, RHOA and RAC1 levels were consistently with that predicted using NDRG1 expression alone, which was reduced after TGFb1 treatment, and RAC1 expression inhibition the gene associated with the highest significant decrease in the RFS was maintained at different time points in all cell lines (Fig. 6E; of this group (P < 0.005; Fig. 6G). LOOCV results demonstrated þ Supplementary Fig. S3E). that the lower RFS in ER and patients with TNBC depended on

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Gonzalez-Gonz alez et al.

Figure 4. ATF4 targeting reduces liver and lung metastases in the PDX model 3887-LM. A, ATF4 mRNA levels in 20 different TNBC PDX models by RNA-sequencing. B, Representative images of ATF4 IHC staining of BCM-3887 and BCM-4664 PDX tumor tissues (original optical objective: 20). C, Representative images and percentage of mice (n ¼ 5/group) with liver and (D) lung metastases, after treatment with ATF4-siRNA#2 and SCR (control) for 6 weeks. E, IHC assessment of liver and lung metastases by Ki67 staining (original optical objectives: 4 and 20). F, Representative images of ATF4 IHC staining of liver and lung metastases (original optical objectives: 4 and 20). , P < 0.05.

the NDRG1 expression, although its impact on RFS could not be that of the patients without alterations (n ¼ 925, 173 months, determined in patients with TNBC (P < 0.008; Supplementary P ¼ 0.00005; Fig. 6H). Table S2). OncoPrint in 2,509 patients with breast cancer showed that the Discussion gene-signature expression is altered in 1,138 patients (45; Sup- plementary Fig. S3F). The percentages of alterations ranged from ATF4 has been proposed as a potential contributor to the 4% to 25% for individual genes (ATF4, 4%; TGFBR1, 4%; SMAD4, pathogenesis and development of breast cancer; however, the 7%; PIK3CA, 7%; RPTOR, 8%; EIF4EBP1, 16%; RICTOR, 6%; underlying mechanisms and the impact on patient survival NDRG1, 25%). Kaplan–Meier analysis and LOOCV results remain unclear. In patients with breast cancer, inﬁltrating carci- showed that patients with the altered expressions (n ¼ 1,055) of noma had higher p-ATF4 than normal breast tissue, what was these genes have poorer survival (143 months) compared with associated with lymph node metastases (7). Recently, a gene

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

Figure 5. ATF4 inhibition delays PDX tumor growth, cancer stem cell population number, tumor relapse, and widens posttreatment survival. A, Volume of BCM-3887 tumors (n ¼ 8/group) treated with siRNA#2 and SCR with and without docetaxel (20 mg/kg). B, Flow cytometric analysis of Aldeﬂuor-positive (ALDFþ) subpopulation after ATF4 knockdown and treatment with/without docetaxel in BCM-3387 tumor tissue. C, Volume of BCM-4664 tumors (n ¼ 8/group) treated with siRNA#2 and SCR. Cotreatment of siRNAs with docetaxel (33 mg /kg) for 6 weeks was used to study tumor relapse after treatment. The arrow indicates the end of treatment. D, ALDFþ subpopulation after ATF4 knockdown in BCM-4664 tumors. E, Kaplan–Meier curve of median survival posttreatment in BCM-4664-bearing mice after ATF4 knockdown in combination with docetaxel (33 mg/kg), P ¼ 0.0001. F, Western blot analysis and densitometric analysis showing ATF4 knockdown efﬁciency in BCM-3387 and (G) BCM-4664 tumor tissues (n ¼ 5/PDX). , P < 0.05; , P < 0.01; , P < 0.001.

www.aacrjournals.org Clin Cancer Res; 2018 OF9

Gonzalez-Gonz alez et al.

Figure 6. The TGFb-induced mTORC2 is the upstream regulator of ATF4 complementary to TGFb/SMAD signaling. Prognostic potential of a mechanism-based gene signature in breast cancer patients. A, Western blot analysis of ATF4 in cells transfected with siRNA for typical ISR mediators and treated with TGFb1 for 72 hours. B, ATF4 protein levels after knockdown of RPTOR and RICTOR treated with TGFb1 for 72 hours (SUM159PT, BT549) or 24 hours (MDA-MB-231). TAK1-siRNA was also tested in SUM159PT. C, Change in SNAIL expression by RPTOR and RICTOR-siRNAs in three cell lines treated with TGFb1 for 24 hours. D, Pearson's correlation of ATF4 mRNA expression with components of mTORC2 (NDRG1, RHOA) and mTORC1 (RPS6, EIF4E) signaling in a cohort of 2,509 patients with breast cancer. E, Western blot analysis of mTORC2 and mTORC1 components in cells transfected with ATF4-siRNA and treated with TGFb1 for 72 hours (SUM159PT cells were treated for 24 hours to test for mTORC1 signaling). F, Schematic showing the upstream regulators, the positive feedbacks detected, the downstream targets of ATF4, and its corresponding biological effects that modulate TNBC aggressiveness upon activation of TGFb which are conserved in the three TNBC cell lines tested. G, Prognostic value (RFS fold change) of the eight-gene signature versus each-single-gene in all (All_BC) and ER– breast cancer patients. Survival fold change was tested by multiple testing correction (, P < 0.005) and leave-one-out cross-validation. H, Impact of the eight-gene signature on the survival of patients with breast cancer with alterations in DNA (ampliﬁcation, deletion) and RNA expression (up- and downregulation) of each gene. Every gene was tested by leave-one-out cross-validation.

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

expression analysis revealed that ATF4 is overexpressed in TNBC increased mouse survival. Accordingly, independent of the patient tissues (8). Here, we investigated the potential of ATF4 as a ISR-induced ATF4 expression, the effects of ATF4 on the cell prognostic marker and therapeutic target in breast cancer and functions regulated by growth-factors were shown to be impor- showed that high ATF4 RNA expression correlates with poorer tant (8, 18–22). Taken together, our findings indicate that ATF4 þ – survival in patients with All_BC, ER ,ER , and TNBC. In a cohort modulates the aggressiveness of TNBC through the regulation of of 35 patients with TNBC, we found a trend showing that ATF4 ISR-independent key signaling pathways, suggesting a potential protein expression correlates with a poorer OS. Our results dem- usefulness of this gene as a therapeutic target. onstrate that ATF4 positiveness starts to have a negative impact on ATF4 is regulated at both transcriptional and translational survival of patients with TNBC at 24 months of follow-up. A levels by different signals (6). Our results show that ATF4 expres- higher follow-up period and a bigger cohort would be necessary to sion depends on the canonical SMAD-dependent TGFb pathway; show statistically significant results. however, there are not evidences in literature showing that SMADs During tumor invasion and metastases, active pathways like are responsible to modulate protein translation. Numerous stress TGFb or NOTCH induce EMT, a shift from the epithelial into types induce the ISR-regulated ATF4 activation mediated by mesenchymal phenotype, induced by transcription factors such as p-eIF2a (4). The ISR controlled by PERK-GCN2/eIF2a/ATF4 SNAIL, SLUG, TWIST1, and ZEB1 (36). TGFb-induced EMT leads mediates EMT and metastasis (10, 11, 13), tumorigenesis to the generation of CSCs with increased self-renewal and tumor- (9, 14), and chemoresistance (16, 17). In nonstressing conditions initiating capabilities, resistance to apoptosis and chemotherapy, and presence of TGFb1, we found that the ISR did not drive ATF4 decreased proliferation, and enhancing tumor recurrence (31). expression for all the cell lines tested herein; however, it was TNBC samples exhibit gene expression profiles observed in CSCs important in SUM159PT and MDA-MB-231 cells. Contrary to and during EMT, such as increased TGFb and mTOR expression previous reports (9–11, 13, 15, 16), eIF2a depletion induced (3), together with a more frequent expression of CSC markers, ATF4 expression and, noteworthy, when PERK was inhibited, which is associated with poorer patient outcomes (37). Similar to reduced eIF2a phosphorylation was only observed in BT549 and earlier reports in osteoblasts and pancreatic adenocarcinoma cells MDA-MB-231, what did not reduce ATF4 levels. These results (11, 20), we reported previously enhanced ATF4 levels in suggest that neither PERK nor eIF2a are responsible for the ATF4 MCF10A cells treated with TGFb1 (23), which suggests that ATF4 translation in absence of stress and presence of TGFb1. Therefore, expression is regulated by TGFb. This pathway is commonly we sought to investigate the eIF2a-independent regulator mech- upregulated and necessary in tumor progression and EMT in anism of ATF4 activation that could be shared by the three TNBC patients with TNBC (3, 31, 38). Our results showed that ATF4 cell lines. TGFb activates noncanonical pathways such as PI3K, is expressed after TGFb1 treatment in TNBC cells, and its expres- MAPK, and TAK1 (34). Pharmacologic inhibitor treatment sion is inhibited by the treatment with LY2157299, suggesting a showed that ATF4 expression in the presence of TGFb1 is also direct effect of TGFb on ATF4 expression. We showed that regulated by the PI3K/mTOR pathway independent of AKT activ- SMAD2/3/4 were at least partially responsible for the regulation ity. In colorectal cancer, ATF4 stabilized by mutant PI3K was of ATF4 expression after TGFb1 treatment. Further analysis of found downstream of PDK1/RSK2, and shown to reprogram previously published ChIP-Seq data (33) and subsequent ChIP- glutamine metabolism independently of AKT (19). We showed qPCR in TGFb1-treated BT549 cells demonstrated for the first time that ATF4 expression does not depend on the presence of AKT, that SMAD2/3 bind and regulate ATF4 transcription. Previous PDK1, RSK2, or PIK3CA mutations, as only SUM159PT cells reports show that ATF4 was dependent of SMAD3 in mouse harbored a mutation in PIK3CA (3). In contrast to previous adipocytes (39) but independent of SMAD4 in osteoblasts reports (18, 40), our findings revealed mTORC2 to be the leading (40). As previously described for ATF3 (41), ATF4 depletion upstream regulator of ATF4 upon TGFb1 treatment (although reduced TGFb activity and SMAD2/3/4 expression, indicating mTORC1 was also important in SUM159PT and BT549 cells). the presence of a feedback loop between ATF4 and TGFb pathway. mTORC2 has been reported as a necessary mediator in the In patients with breast cancer, coexpression of ATF4/TGFBR1, TGFb-induced EMT through AKT phosphorylation (Ser473; ATF4/SMAD2, ATF4/SMAD4, and ATF4/SMAD3 resulted in ref. 42), a well-known feedback loop that activates mTORC1 poorer OS, shown to depend on ATF4 overexpression. Together, (35), as well as in protein translation by direct interaction with these results demonstrate that TGFb/SMAD2/3/4 are upstream ribosomal proteins (43). Similarly, mTORC1 modulates not only of ATF4, which regulate the signaling through a positive ATF4 transcription but also translation independently of eIF2a feedback with the TGFb pathway, and it may be involved in the (5). Whether mTORC2 can directly regulate ATF4 translation and TGFb-associated aggressiveness of TNBC. transcription remains elusive, but it would explain why ATF4 Here, we report a more important role of ATF4 in the consti- expression is independent of AKT, PDK1, RSK2, or PIK3CA tutive (average of 56% decrease) than in the TGFb1-induced mutations. According to previous studies (5, 15), we observed tumor cell migration (average of 45% decrease). However, ATF4 that ATF4 also regulates mTOR signaling by a feedback loop on was more relevant in the TGFb1-induced than in the basal tumor mTORC1, what may regulate cell survival and drug resistance cell invasiveness (average of 54% and 38% decrease, respectively). induced by MCL1 and BCL2 (21, 44), and mTORC2. We hypoth- In addition, different EMT transcription factors and stemness esize that this dual regulation on mTOR may be attributed to the markers were inhibited by ATF4 silencing when TGFb1 was added ATF4-mediated regulation of RAC1 that further affects mTORC1 or not to the medium. Our results suggest that, under nonstressing and mTORC2 activity in response to growth-factor stimulation conditions, ATF4 is involved in the aggressiveness of TNBC cells (45), which may potentiate the mTORC2/AKT feedback loop on mediated not only by TGFb, but also by other signaling pathways. mTORC1. Because TGFb signaling activates both mTORC1 (46) In TNBC PDX mouse models, ATF4 depletion resulted in a and mTORC2 (47) in a SMAD-dependent way by inhibition reduced lung and liver metastasis rate, tumor growth, ALDFþ of DEPTOR (47), we suggest that TGFb could activate mTORC2 CSC-like population numbers, delayed tumor relapse, and (and mTORC1 in some cell lines) through a SMAD-dependent

www.aacrjournals.org Clin Cancer Res; 2018 OF11

Gonzalez-Gonz alez et al.

signaling, what would induce ATF4 expression to mediate EMT, Authors' Contributions motility, metastasis, pluripotency, and self-renewal. Active Conception and design: A. Gonzalez-Gonzalez, J.A. Marchal, D. Landeira, mTORC2 could also feedback on AKT to enhance mTORC1- P. Sanchez-Rovira, J.C. Chang, S. Granados-Principal ~ dependent ATF4 translation and transcription. This circuit would Development of methodology: A. Gonzalez-Gonzalez, E. Munoz-Muela, F.E. Cara, M.P. Molina, M. Cruz-Lozano, G. Jimenez, A. Ramírez, M.D. Martín- be maintained by the feedback of ATF4 on mTOR and TGFb Salvago, R.J. Luque, C.L. Ramirez-Tortosa, C. Rodriguez-Aguayo, A.K. Sood, signaling through the regulation of SMAD2/3/4 and mTORC1/ P. Sanchez-Rovira, J.C. Chang, S. Granados-Principal 2-RHOA-RAC1 pathways (48, 49). Acquisition of data (provided animals, acquired and managed patients, Identifying patterns that predict signaling pathway activation, provided facilities, etc.): E. Munoz-Muela,~ W. Qian, W. Chen, A.J. Kozielski, by using gene signatures and considering the target interactions, M.D. Martín-Salvago, R.J. Luque, M. Quintana-Romero, R.R. Rosato, has been demonstrated to be a viable approach to the personal- M.A. García, H. Kim, C. Rodriguez-Aguayo, G. Lopez-Berestein, P. Sanchez- Rovira, J.C. Chang, S. Granados-Principal ized TNBC treatment (1). Because ATF4 is involved in TGFb/ b fi Analysis and interpretation of data (e.g., statistical analysis, biostatistics, SMAD and TGF /PI3K/mTOR pathways, we identi ed an computational analysis): A. Gonzalez-Gonzalez, E. Munoz-Muela,~ F.E. Cara, eight-gene prognostic signature, including ATF4, TGFBR1, M.P. Molina, M. Cruz-Lozano, G. Jimenez, A. Verma, A. Ramírez, O. Elemento, SMAD4, PIK3CA, RPTOR, EIF4EBP1, RICTOR, and NDRG1 genes M.D. Martín-Salvago, R.J. Luque, C. Rosa-Garrido, D. Landeira, M. Quintana- that can be used for the prediction of patient survival in all breast Romero, R.R. Rosato, C.L. Ramirez-Tortosa, H. Kim, A.K. Sood, P. Sanchez- – cancer, ER , and the basal subtype groups. The expression of these Rovira, J.C. Chang, S. Granados-Principal Writing, review, and/or revision of the manuscript: A. Gonzalez-Gonzalez, signature genes was shown to be altered in 45% of 2,509 patients E. Munoz-Muela,~ F.E. Cara, G. Jimenez, D. Landeira, R.R. Rosato, H. Kim, with breast cancer, with lower survival rates observed in these – G. Lopez-Berestein, A.K. Sood, J.A. Lorente, P. Sanchez-Rovira, S. Granados- patients. Breast cancer patient stratification, especially ER Principal patients, according to this gene signature may provide a useful Administrative, technical, or material support (i.e., reporting or organizing strategy for designing effective signaling pathway-guided combi- data, constructing databases): W. Qian, C. Rosa-Garrido, P. Sanchez-Rovira, natorial targeted therapies aimed at the reduction of tumor J.C. Chang, S. Granados-Principal Study supervision: R.R. Rosato, J.A. Lorente, P. Sanchez-Rovira, S. Granados- growth, metastases, and relapse risk, and may allow the identi- Principal fication of potentially responsive patients. In conclusion, we demonstrate here for the first time the Acknowledgments potential of ATF4 as a prognostic biomarker and a therapeutic This work was funded by Instituto de Salud Carlos III (to S. Granados- target in patients with TNBC. Furthermore, we showed that Principal: CP14/00197, PI15/00336, MV16/00005; to J.A. Marchal: PIE16/ ATF4 is involved in the regulation of signaling pathways asso- 00045), European Regional Development Fund (to S. Granados-Principal), ciated with tumor metastasis, proliferation, and drug resistance, the Chair "Doctors Galera-Requena in Cancer Stem Cell Research" (to which induce the aggressiveness of TNBC. In contrast to the J.A. Marchal), Spanish Ministry of Economy and Competitiveness and Andalusian Regional Government (to D. Landeira: RYC-2012-10019; previous reports, ATF4 activity was shown to be independent of BFU2016-75233-P; PC-0246-2017). M. Quintana-Romero is funded by the the ISR, integrating and modulating TGFb/SMAD2/3/4 and "Garantia Juvenil" program (REF-2813; Andalusian Regional Government and TGFb/PI3K/mTORC1/2 pathways. We identified a signaling University of Granada). We thank Dr. Pedro Carmona-Sanchez for the assis- pathway-guided prognostic gene signature in patients with tance on statistics of the meta-analysis and Editage (www.editage.com) for breast cancer that may help design combinatorial targeted English language editing. therapies, identify potential responsive patients, and predict and overcome drug resistance. Precision medicine may benefit The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked from our approach by improving the treatment decision-mak- advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate ing in breast cancer patients. this fact.

Disclosure of Potential Conflicts of Interest Received November 6, 2017; revised April 30, 2018; accepted July 11, 2018; No potential conflicts of interest were disclosed. published first July 16, 2018.

References 1. Jhan J-R, Andrechek ER. Effective personalized therapy for breast cancer 8. van Geldermalsen M, Wang Q, Nagarajah R, Marshall AD, Thoeng A, based on predictions of cell signaling pathway activation from gene Gao D, et al. ASCT2/SLC1A5 controls glutamine uptake and tumour expression analysis. Oncogene 2017;36:3553–61. growth in triple-negative basal-like breast cancer. Oncogene 2016;35: 2. Jitariu A, C^mpean AM, Ribatti D, Raica M. Triple negative breast cancer: the 3201–8. kiss of death. Oncotarget 2017;8:46652–62. 9. Bobrovnikova-Marjon E, Grigoriadou C, Pytel D, Zhang F, Ye J, 3. Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y, et al. Koumenis C, et al. PERK promotes cancer cell proliferation and Identiﬁcation of human triple-negative breast cancer subtypes and pre- tumor growth by limiting oxidative DNA damage. Oncogene 2010;29: clinical models for selection of targeted therapies. J Clin Invest 3881–95. 2011;121:2750–67. 10. Nagelkerke A, Bussink J, Mujcic H, Wouters BG, Lehmann S, Sweep FCGJ, 4. Pakos-Zebrucka K, Koryga I, Mnich K, Ljujic M, Samali A, Gorman AM. The et al. Hypoxia stimulates migration of breast cancer cells via the PERK/ integrated stress response. EMBO Rep 2016;17:1374–95. ATF4/LAMP3-arm of the unfolded protein response. Breast Cancer Res 5. Park Y, Reyna-Neyra A, Philippe L, Thoreen CC. mTORC1 balances cellular 2013;15:R2. amino acid supply with demand for protein synthesis through post- 11. Feng Y-X, Sokol ES, Del Vecchio CA, Sanduja S, Claessen JHL, Proia TA, et al. transcriptional control of ATF4. Cell Rep 2017;19:1083–90. Epithelial-to-mesenchymal transition activates PERK-eIF2a and sensitizes 6. Singleton DC, Harris AL. Targeting the ATF4 pathway in cancer therapy. cells to endoplasmic reticulum stress. Cancer Discov 2014;4:702–15. Expert Opin Ther Targets 2012;16:1189–202. 12. Demay Y, Perochon J, Szuplewski S, Mignotte B, Gaumer S. The PERK 7. Fan C-F, Mao X-Y, Wang E-H. Elevated p-CREB-2 (ser 245) expression is pathway independently triggers apoptosis and a Rac1/Slpr/JNK/Dilp8 potentially associated with carcinogenesis and development of breast signaling favoring tissue homeostasis in a chronic ER stress Drosophila carcinoma. Mol Med Rep 2012;5:357–62. model. Cell Death Dis 2014;5:e1452.

OF12 Clin Cancer Res; 2018 Clinical Cancer Research

ATF4, Biomarker, and Target in Triple-Negative Breast Cancer

13. Dey S, Sayers CM, Verginadis II, Lehman SL, Cheng Y, Cerniglia GJ, et al. 31. Bhola NE, Balko JM, Dugger TC, Kuba MG, Sanchez V, Sanders M, et al. ATF4-dependent induction of heme oxygenase 1 prevents anoikis and TGF-b inhibition enhances chemotherapy action against triple-negative promotes metastasis. J Clin Invest 2015;125:2592–608. breast cancer. J Clin Invest 2013;123:1348–58. 14. Wu H, Wei L, Fan F, Ji S, Zhang S, Geng J, et al. Integration of Hippo 32. Martin-Malpartida P, Batet M, Kaczmarska Z, Freier R, Gomes T, Aragon E, signalling and the unfolded protein response to restrain liver overgrowth et al. Structural basis for genome wide recognition of 5-bp GC motifs by and tumorigenesis. Nat Commun 2015;6:6239–53. SMAD transcription factors. Nat Commun 2017;8:2070. 15. Liu X, Lv Z, Zou J, Liu X, Ma J, Wang J, et al. Afatinib down-regulates MCL-1 33. Sundqvist A, Morikawa M, Ren J, Vasilaki E, Kawasaki N, Kobayashi M, et al. expression through the PERK-eIF2a-ATF4 axis and leads to apoptosis JUNB governs a feed-forward network of TGFb signaling that aggravates in head and neck squamous cell carcinoma. Am J Cancer Res 2016;6: breast cancer invasion. Nucleic Acids Res 2018;46:1180–95. 1708–19. 34. Bierie B, Moses HL. Tumour microenvironment: TGFbeta: the molecular 16. Dekervel J, Bulle A, Windmolders P, Lambrechts D, Van Cutsem E, Verslype Jekyll and Hyde of cancer. Nat Rev Cancer 2006;6:506–20. C, et al. Acriflavine inhibits acquired drug resistance by blocking the 35. Efeyan A, Sabatini DM. mTOR and cancer: many loops in one pathway. epithelial-to-mesenchymal transition and the unfolded protein response. Curr Opin Cell Biol 2010;22:169–76. Transl Oncol 2017;10:59–69. 36. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. 17. Nagasawa I, Kunimasa K, Tsukahara S, Tomida A. BRAF-mutated cells J Clin Invest 2009;119:1420–8. activate GCN2-mediated integrated stress response as a cytoprotective 37. Idowu MO, Kmieciak M, Dumur C, Burton RS, Grimes MM, Powers CN, mechanism in response to vemurafenib. Biochem Biophys Res Commun et al. CD44(þ)/CD24(-/low) cancer stem/progenitor cells are more abun- 2017;482:1491–7. dant in triple-negative invasive breast carcinoma phenotype and are 18. Ben-Sahra I, Hoxhaj G, Ricoult SJH, Asara JM, Manning BD. mTORC1 associated with poor outcome. Hum Pathol 2012;43:364–73. induces purine synthesis through control of the mitochondrial tetrahy- 38. Jovanovic B, Beeler JS, Pickup MW, Chytil A, Gorska AE, Ashby WJ, et al. drofolate cycle. Science 2016;351:728–33. Transforming growth factor beta receptor type III is a tumor promoter in 19. Hao Y, Samuels Y, Li Q, Krokowski D, Guan B-J, Wang C, et al. Oncogenic mesenchymal-stem like triple negative breast cancer. Breast Cancer Res PIK3CA mutations reprogram glutamine metabolism in colorectal cancer. 2014;16:R69. Nat Commun 2016;7:11971. 39. Liu Z, Gu H, Gan L, Xu Y, Feng F, Saeed M, et al. Reducing Smad3/ATF4 was 20. Lian N, Lin T, Liu W, Wang W, Li L, Sun S, et al. Transforming growth factor essential for Sirt1 inhibiting ER stress-induced apoptosis in mice brown b suppresses osteoblast differentiation via the vimentin activating tran- adipose tissue. Oncotarget 2017;8:9267–79. scription factor 4 (ATF4) axis. J Biol Chem 2012;287:35975–84. 40. Karner CM, Lee S-Y, Long F. Bmp Induces Osteoblast Differentiation 21. Zhu H, Xia L, Zhang Y, Wang H, Xu W, Hu H, et al. Activating transcription through both Smad4 and mTORC1 Signaling. Mol Cell Biol 2017;37: factor 4 confers a multidrug resistance phenotype to gastric cancer cells 1–12. through transactivation of SIRT1 expression. PLoS One 2012;7:e31431. 41. Yin X, Wolford CC, Chang Y-S, McConoughey SJ, Ramsey S a, Aderem A, 22. Zhu H, Chen X, Chen B, Chen B, Song W, Sun D, et al. Activating et al. ATF3, an adaptive-response gene, enhances TGF{beta} signaling and transcription factor 4 promotes esophageal squamous cell carcinoma cancer-initiating cell features in breast cancer cells. J Cell Sci 2010;123: invasion and metastasis in mice and is associated with poor prognosis in 3558–65. human patients. PLoS One 2014;9:e103882. 42. Lamouille S, Connolly E, Smyth JW, Akhurst RJ, Derynck R. TGF-b-induced 23. Granados-Principal S, Liu Y, Guevara ML, Blanco E, Choi DS, Qian W, et al. activation of mTOR complex 2 drives epithelial-mesenchymal transition Inhibition of iNOS as a novel effective targeted therapy against triple- and cell invasion. J Cell Sci 2012;125:1259–73. negative breast cancer. Breast Cancer Res 2015;17:25. 43. Oh WJ, Wu C, Kim SJ, Facchinetti V, Julien L-A, Finlan M, et al. mTORC2 24. Gyorffy€ B, Lanczky A, Eklund AC, Denkert C, Budczies J, Li Q, et al. An can associate with ribosomes to promote cotranslational phosphorylation online survival analysis tool to rapidly assess the effect of 22,277 genes on and stability of nascent Akt polypeptide. EMBO J 2010;29:3939–51. breast cancer prognosis using microarray data of 1,809 patients. Breast 44. Pugazhenthi S, Nesterova A, Sable C, Heidenreich KA, Boxer LM, Cancer Res Treat 2010;123:725–31. Heasley LE, et al. Akt/protein kinase B up-regulates Bcl-2 expression 25. Pereira B, Chin S-F, Rueda OM, Vollan H-KM, Provenzano E, Bardwell HA, through cAMP-response element-binding protein. J Biol Chem 2000; et al. The somatic mutation profiles of 2,433 breast cancers refines their 275:10761–6. genomic and transcriptomic landscapes. Nat Commun 2016;7:11479. 45. Saci A, Cantley LC, Carpenter CL. Rac1 regulates the activity of mTORC1 26. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio and mTORC2 and controls cellular size. Mol Cell 2011;42:50–61. cancer genomics portal: an open platform for exploring multidimensional 46.DasR,XuS,NguyenTT,QuanX,ChoiS-K,KimS-J,etal. cancer genomics data. Cancer Discov 2012;2:401–4. Transforming growth factor b1-induced apoptosis in podocytes via 27. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. the extracellular signal-regulated kinase-mammalian target of rapa- Integrative analysis of complex cancer genomics and clinical profiles using mycin complex 1-NADPH oxidase 4 axis. J Biol Chem 2015;290: the cBioPortal. Sci Signal 2013;6:pl1. 30830–42. 28. Dave B, Granados-Principal S, Zhu R, Benz S, Rabizadeh S, Soon-Shiong P, 47. Das F, Ghosh-Choudhury N, Bera A, Dey N, Abboud HE, Kasinath BS, et al. et al. Targeting RPL39 and MLF2 reduces tumor initiation and metastasis in Transforming growth factor b integrates Smad 3 to mechanistic target of breast cancer by inhibiting nitric oxide synthase signaling. Proc Natl Acad rapamycin complexes to arrest deptor abundance for glomerular mesangial Sci U S A 2014;111:8838–43. cell hypertrophy. J Biol Chem 2013;288:7756–68. 29. Zhang X, Claerhout S, Prat A, Dobrolecki LE, Petrovic I, Lai Q, et al. 48. Gulhati P, Bowen KA, Liu J, Stevens PD, Rychahou PG, Chen M, et al. A renewable tissue resource of phenotypically stable, biologically and mTORC1 and mTORC2 regulate EMT, motility, and metastasis of colo- ethnically diverse, patient-derived human breast cancer xenograft models. rectal cancer via RhoA and Rac1 signaling pathways. Cancer Res Cancer Res 2013;73:4885–97. 2011;71:3246–56. 30. Choi DS, Blanco E, Kim Y-S, Rodriguez AA, Zhao H, Huang TH-M, et al. 49. Yu JSL, Cui W. Proliferation, survival and metabolism: the role of PI3K/ Chloroquine eliminates cancer stem cells through deregulation of Jak2 and AKT/mTOR signalling in pluripotency and cell fate determination. DNMT1. Stem Cells 2014;32:2309–23. Development 2016;143:3050–60.

www.aacrjournals.org Clin Cancer Res; 2018 OF13

Activating Transcription Factor 4 Modulates TGFβ-Induced Aggressiveness in Triple-Negative Breast Cancer via SMAD2/3/4 and mTORC2 Signaling

Adrián González-González, Esperanza Muñoz-Muela, Juan A. Marchal, et al.

Clin Cancer Res Published OnlineFirst July 16, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-17-3125

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/07/14/1078-0432.CCR-17-3125.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/2018/09/06/1078-0432.CCR-17-3125. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.