Mechanistic rationale for targeting the unfolded PNAS PLUS response in pre-B acute lymphoblastic

Behzad Kharabi Masouleha,c, Huimin Genga, Christian Hurtza, Lai N. Chana, Aaron C. Loganb, Mi Sook Changd, Chuanxin Huange,f, Srividya Swaminathana, Haibo Sung, Elisabeth Paiettah, Ari M. Melnicke,f, Phillip Koefflerg, and Markus Müschena,1

aDepartment of Laboratory Medicine and bDivision of Hematology–Oncology, University of California, San Francisco, CA 94143; cDepartment of Oncology, Hematology and Stem Cell Transplantation, Rheinisch-Westfaelische Technische Hochschule Aachen University Medical School, 52070 Aachen, Germany; dChildren’s Hospital Los Angeles, Los Angeles, CA 90027; Departments of eMedicine and fPharmacology, Weill Cornell Medical College, New York, NY 10065; gCedars Sinai Medical Center, Los Angeles, CA 90048; and hDepartment of Medicine, Albert Einstein College of Medicine, Bronx, NY 10466

Edited by Owen N. Witte, Howard Hughes Medical Institute, University of California, Los Angeles, CA, and approved April 18, 2014 (received for review January 20, 2014) The unfolded protein response (UPR) pathway, a stress-induced autophosphorylation and oligomerization and induces cleavage signaling cascade emanating from the (ER), and splicing of XBP1 by its endoribonuclease (RNase) domain, regulates the expression and activity of molecules including BiP resulting in the removal of a 26-nucleotide intron. This frame (HSPA5), IRE1 (ERN1), Blimp-1 (PRDM1), and X-box binding protein shift modification leads to expression of a longer, highly active 1 (XBP1). These molecules are required for terminal differentiation splice variant (XBP1-s) (9), responsible for the regulation of a of B cells into plasma cells and expressed at high levels in plasma variety of downstream targets to relieve ER stress. Activation cell-derived multiple myeloma. Although these molecules have of the UPR by ER stress has been linked to the transition of no known role at early stages of B-cell development, here we mature surface Ig-dependent B cells to Ig-secreting plasma cells show that their expression transiently peaks at the pre–B-cell re- that no longer express Ig on the surface. An important role in ceptor checkpoint. Inducible, Cre-mediated deletion of Hspa5, — this transition is played by Ig heavy chain-binding protein (BiP) MEDICAL SCIENCES Prdm1 Xbp1 , and consistently induces cellular stress and cell death also known as heat shock 70-kDa protein 5 (HSPA5) and Grp78— – in normal pre-B cells and in pre B-cell acute lymphoblastic leuke- which chaperones folding of Ig heavy, but not light, chain BCR-ABL1 NRASG12D mia (ALL) driven by - and oncogenes. Mecha- (10). A previous study also demonstrated that IRE1 (ERN1) is nistically, expression and activity of the UPR downstream effector – required during V(D)J recombination at the transition from pro- XBP1 is regulated positively by STAT5 and negatively by the B-cell to pre-B cells (11). specific transcriptional repressors BACH2 and BCL6. In two clinical XBP1 Here we provide genetic evidence for the emerging concept trials for children and adults with ALL, high mRNA levels at that the UPR molecules ERN1 and HSPA5, as well as their the time of diagnosis predicted poor outcome. A small molecule downstream effectors PRDM1 and XBP1, are not only critical inhibitor of ERN1-mediated XBP1 activation induced selective cell for the transition from surface Ig-dependent B cells to Ig- death of patient-derived pre-B ALL cells in vitro and significantly secreting plasma cells, but also regulate the pre–B-cell stage, when prolonged survival of transplant recipient mice in vivo. Collec- tively, these studies reveal that pre-B ALL cells are uniquely vul- nerable to ER stress and identify the UPR pathway and its Significance downstream effector XBP1 as novel therapeutic targets to over- come drug resistance in pre-B ALL. The unfolded protein response (UPR) mitigates endoplasmic reticulum (ER) stress. In this regard, ER stress-inducing agents erminal B-cell differentiation is regulated through two sets were found to be highly active in a clinical trial for children Tof antagonizing transcription factors: paired box 5 with relapsed acute lymphoblastic leukemia (ALL), a disease (PAX5), BTB and CNC homology 1, basic tran- derived from transformed pre-B cells. To understand the effi- scription factor 2 (BACH2), and BCL6 maintain B-cell identity cacy of ER stress-inducing agents in pre-B ALL, we studied the relevance of the UPR pathway in genetic and patient-derived of postgerminal center B cells (1), whereas the (xenograft) models of human pre-B ALL. Our studies revealed PR domain containing 1, with ZNF domain (PRDM1) (also known an unrecognized vulnerability of both normal pre-B cells and as B-lymphocyte-induced maturation protein 1; BLIMP1) and pre--derived ALL cells to genetic or pharmacological block- X-box binding protein 1 (XBP1) drive differentiation – ade of the UPR pathway. Our results establish a mechanistic ra- (2 4). The plasma cell transcription factor XBP1 and its upstream tionale for the treatment of children with pre-B ALL with agents regulator PRDM1 have been extensively studied in plasma cell that block the UPR pathway and induce ER stress. differentiation and the plasma cell malignancy multiple myeloma (5, 6), but not in early B-cell development or and lym- Author contributions: B.K.M. and M.M. designed research; B.K.M., C. Hurtz, L.N.C., M.S.C., phomas representing early stages of B-cell differentiation. Surpris- C. Huang, S.S., and H.S. performed research; A.C.L., E.P., A.M.M., and P.K. contributed new ingly, endoplasmic reticulum (ER) stress-inducing agents were reagents/analytic tools; B.K.M., H.G., L.N.C., M.S.C., C. Huang, S.S., H.S., E.P., and A.M.M. recently found to be highly active in a clinical trial for children analyzed data; and B.K.M. and M.M. wrote the paper. with relapsed acute lymphoblastic leukemia (ALL) (7), a disease The authors declare no conflict of interest. derived from transformed pre-B cells. This article is a PNAS Direct Submission. XBP1 is highly expressed in multiple myeloma and plasma Data deposition: Microarray data are available from the Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession nos. GSE5314, GSE34941, GSE28460, cells, where it mitigates ER stress through engagement of the GSE11877, GSE20987, GSE30883, GSE19599, GSM488979, GSE13411, GSE12453, GSE53684, unfolded protein response (UPR). The UPR network consists of and GSE53683). See also Supplemental S3h-k: www.stjuderesearch.org/data/ALL3. three major branches including Inositol-requiring enzyme 1a 1To whom correspondence should be addressed. E-mail: [email protected]. α (IRE1 , ERN1), PKR-like ER kinase, and activating transcrip- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. tion factor 6 (ATF6) (8). ERN1 is activated by ER stress through 1073/pnas.1400958111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1400958111 PNAS Early Edition | 1of10 Downloaded by guest on September 28, 2021 Ig heavy-chain variable region are rearranged and Ig heavy with strong up-regulation of the cell-cycle checkpoint regulators chains are expressed for the first time. Whereas recent studies Arf (Cdkn2a) and p21 (Cdkn1a; underlined in Fig. 1A). Micro- have identified XBP1 as a therapeutic target in multiple myeloma, array gene expression analysis was validated by single-gene we demonstrate that pre-B ALL cells are distinctly sensitive to quantitative RT (qRT)-PCR (SI Appendix, Fig. S2). Within 4 d of ER stress and pharmacological inhibition of XBP1. Pre-B ALL Hspa5 deletion, ∼50% of IL-7–dependent pre-B cells (Fig. 1C) arises from a transformed pre-B cell and represents the most and >70% of BCR-ABL1–transformed pre-B ALL cells (Fig. frequent type of in children (12). Frequent oncogenic 1D) underwent cell death. Deletion of Hspa5 in BCR-ABL1– lesions in pre-B ALL include BCR-ABL1 (present in ∼30% of transformed pre-B ALL cells in vivo resulted in a drastic re- adult ALL) (13) and hyperactivating mutations of RAS genes duction of leukemia burden, as measured by luciferase bio- (in ∼50% of childhood pre-B ALL) (14). Small-molecule in- imaging (Fig. 1E) and prolonged overall survival of transplant hibitors targeting the RNase domain of ERN1 prevent activation recipient mice (Fig. 1F). In these experiments, 1 million lucif- of XBP1 and have disease-modifying activity in multiple mye- erase-labeled Hspa5fl/fl BCR-ABL1 pre-B ALL cells carrying loma (15, 16). Here we show that small-molecule inhibition of either tamoxifen-inducible GFP-tagged Cre (Cre; n = 9) or a ERN1/XBP1 represents a promising novel approach to overcome GFP-tagged empty vector (EV) control (n = 9) were injected drug resistance in pre-B ALL. (i.v.) into each sublethally irradiated (2 Gy) NOD/SCID mouse. Although deletion of Hspa5 significantly extended overall sur- Results vival of recipient mice (P = 0.028; Fig. 1F), recipient mice in both Hspa5 (Ig Heavy-Chain Binding Protein) Is Required for the Survival groups eventually developed terminal disease and had to be killed. and Malignant Transformation of Pre-B Cells. To elucidate the In the analysis of bone marrows and spleens from transplant re- − + function of UPR-related molecules in both normal pre-B and cipient mice we distinguished infiltrating CD45.1 CD45.2 leu- + − transformed pre-B ALL cells, we first focused on the ER chap- kemia cells from CD45.1 CD45.2 host cells. Interestingly, + erone heat shock 70-kDa protein 5 (HSPA5) (17), which miti- virtually all CD45.2 ALL cells expressed the GFP-tagged EV gates ER stress upstream of the ER sensor ERN1 (IRE-1). In control, whereas expression of GFP-tagged Cre was silenced in a genetic experiment, we studied the consequences of acute ab- a fraction of leukemia cells (SI Appendix, Fig. S1B). Consistent lation of Hspa5 function in normal pre-B cells and BCR-ABL1– with these findings, ALL clones in all nine NOD/SCID mice re- transformed pre-B ALL cells that were generated from bone ceiving Cre-transduced Hspa5fl/fl leukemia cells retained Hspa5 marrow B-cell progenitor cells of Hspa5fl/fl mice (18) (Fig. 1). floxed alleles in contrast to complete deletion in vitro (SI Appendix, The 4-hydroxytamoxifen (4-OHT)-mediated activation of Cre Fig. S1C). These findings imply that Hspa5fl/fl ALL in this trans- for 2 d resulted in excision of Hspa5 exons 5–7 and compensatory plant model is strongly selected for clones that escaped Hspa5 up-regulation of Hspa5 mRNA (Fig. 1A), whereas Hspa5 protein deletion. Fatal disease developing in NOD/SCID mice receiving was destabilized and no longer detectable by Western blot (SI Cre-transduced Hspa5fl/fl leukemia cells might originate from clones Appendix,Fig.S1A). Interestingly, acute ablation of Hspa5 that had escaped complete Cre-mediated deletion. function results in strong up-regulation of the ER chaperones Hsp90b1, Dnajc3, and others as well as Prdm1 and Xbp1 (high- Expression of Plasma Cell-Specific UPR Molecules ERN1, HSPA5, lighted in Fig. 1A). Inducible deletion of Hspa5 has profound PRDM1, and XBP1 in Normal Pre-B Cells and Pre-B ALL. To study effects and rapidly induced cell cycle arrest (Fig. 1B) in parallel the function of the ER HSPA5 in the context of other

Fig. 1. Hspa5 (Ig heavy-chain binding fl/fl Hspa5fl/fl A B Hspa5fl/fl ALL CDHspa5 pre-B cells pre-B ALL protein) is required for the survival S-phase 300 Hspa5fl/fl pre-B ALL 180 P=0.0002 EV and malignant transformation of pre- EV Cre S: 64 250 EV 160 Crisp1 100 140 B cells. (A) Gene expression profiling Cebpb G2/M cells [%] fl/fl cells [%] 200 120 + Sdf2l1 + was assessed in Hspa5 pre-B ALL G2/M: 4 S phase Nucb2 EV 80 100 G 32 150 cells transduced with EV control and Hyou1 0/1 80 Ctla4 G0/1 – 60 100 60 4-OHT inducible retroviral Cre (Cre) Hspa5 Hsp90b1 S: 14 40 after 2 d of 4-OHT treatment. High- 40 50 Evi2a Cre Cre 20 Cre Fraction of GFP lighted are ER-related molecules such Dnajc3 Fraction of GFP 0 0 Pdia4 G /M: 7 20 as Hsp90b1, Dnajc3, Pdia4, and Ero1lb 2 0 2469 0 2469 13 18 24

Sel1l BrdU G 79 as well as Prdm1 and Xbp1 in red and Slc35b1 0/1 0 Time after transduction [days] Time after transduction [days] Ly6f EV Cre cell-cycle checkpoint regulators such Hspa5 7 AAD Prdm1 as Arf (Cdkn2a)andp21 (Cdkn1a)(GEO Cdkn1a Hspa5fl/fl pre-B ALL accession no. GSE53684). (B) Hspa5fl/fl Xbp1 E F Hspa5fl/fl pre-B ALL Cish EV Ero1lb P=0.03; n=5 per group ALL cells with EV and Cre were stained +4-OHT Cdkn2a 100 with BrdU and 7AAD and the percen- Cd21 tages of cells in the G ,S,andG/M cell Tmprss3 80 0/1 2 Mef2c cycles are indicated after 1 d of 4-OHT Alox5ap 60 Bach2 treatment and shown in bar graph Blk Cre (n = 3). (C) Hspa5fl/fl pre-B cells were Lgr5 + 4-OHT 40 Bnip3l transduced with MSCV-GFP (EV) and Spp1 20 EV Cre Lcp2 +4-OHT + 4-OHT Cre-GFP (Cre) and the relative fraction Fcrla 0 + = Eps8 of Survival NOD/.SCID mice [%] of GFP cells was measured up to 9 d (n Egr1 6 10 13 Days after injection 051015 20 3). (D) Similarly, the relative fraction of Ccng2 + fl/fl B3gnt5 GFP cells in Hspa5 pre-B ALL cells with Mycn = Cd22 EV and Cre was measured up to 24 d (n Igk-V1 3). (E) One million luciferase-labeled Cmah Cd22 Hspa5fl/fl pre-BALLcellscarryingeither tamoxifen-inducible GFP-tagged Cre -2 -1 0 +1 +2 Log2 fold change (CreGFP) or a GFP-tagged empty vector control (EVGFP) were injected (i.v.) into sublethally irradiated (2 Gy) NOD/SCID mice and leukemic expansion was tracked by luciferase bioimaging. (F)The overall survival of transplant recipient mice is shown in a Kaplan–Meier analysis (n = 5 per group, P = 0.03).

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1400958111 Kharabi Masouleh et al. Downloaded by guest on September 28, 2021 UPR-related genes [ERN1 (IRE1), HSPA5 (BIP), PRDM1 BLIMP, differentiation of pre-B cells past the pre-BCR checkpoint and PNAS PLUS and XBP1] we tracked expression of these genes throughout early induced reduction of Xbp1-s expression (SI Appendix,Fig.S3C). and late stages of B-cell development (SI Appendix,Fig.S3). As expected, plasma cells express these molecules at the highest The Plasma Cell Transcription Factor XBP1 Is Demethylated and Up- levels (SI Appendix, Fig. S3A). Consistent with our findings for Regulated in Pre-B ALL Cells. In patient-derived pre-B ALL sam- HSPA5, regions of ERN1, PRDM1, and XBP1 were ples including various ALL subsets, the promoter regions of ERN1, hypomethylated in human pre-B ALL (SI Appendix, Fig. S3 A, E, HSPA5, PRDM1,andXBP1 genes were consistently demethylated compared with mature B-cell non-Hodgkin lymphomas (SI Ap- and I) compared with normal pre-B cells or mature B-cell lym- pendix,Fig.S3D–G). Low levels of CpG methylation in promoter ERN1 PRDM1 XBP1 phomas. Similarly, mRNA levels of , , and regions correlated with high mRNA levels of ERN1, HSPA5, + were high in pre-B ALL and also peaked during normal early B- PRDM1 XBP1 – Ph SI – ,and ,andalsoforBCR-ABL1 driven ALL ( cell development at the pre B-cell (pre-BCR) check- Appendix, Fig. S3 H–K). High levels of promoter CpG methylation point in human bone marrow (SI Appendix, Fig. S3A). The XBP1 correlated with low expression levels in mature B-cell lymphoma. microarray probeset (200670_at) does not distinguish between Overall, relative mRNA levels of ERN1, HSPA5, PRDM1,andXBP1 the unspliced XBP1-u and the more active, spliced form XBP1-s were higher in ALL cells compared with normal pre-B cells but lower (9); therefore, specific up-regulation of Xbp1-s at the pre-BCR compared with plasma cells (SI Appendix,Fig.S3H–K). Similarly, checkpoint was confirmed by qRT-PCR (SI Appendix, Fig. S3B), WesternblotanalysisshowedthatbothPRDM1andXBP1-sare − + + + using sorted Lin Sca-1 c-Kit multilineage progenitors, bone expressedinCD19 magnetic activated cell sorting (MACS)- marrow B-cell precursor populations, as well as mature B cells enriched pre-B cells (SI Appendix,Fig.S4B); however, expression from spleens of C57BL/6 mice (Hardy fractions A–F; SI Ap- levels in patient-derived pre-B ALL samples are significantly higher pendix, Fig. S4A). Previous studies of ERN1, HSPA5, PRDM1, for XBP1-s whereas only a slight increase was observed in the case of and XBP1 largely focused on their critical role in plasma cell PRDM1 as verified by densitometry analysis (SI Appendix,Fig.S5A). differentiation (i.e., the transition from surface Ig-dependent B cells to Ig-secreting plasma cells) (19, 20). Specific up-regulation Prdm1 (Blimp1) Is Required for Proliferation and Survival of Pre-B ALL Cells. To study the role of Prdm1 in pre-B ALL cells using a at early pre–B-cell stages suggests that these molecules are also genetic experiment, we transformed bone marrow pre-B cells from functional at the pre-BCR checkpoint, when Ig heavy-chain Prdm1fl/fl mice (22) with the human BCR-ABL1 tyrosine kinase + variable genes are assembled and Ig heavy chains are expressed that drives Ph ALL. Inducible deletion of Prdm1 was achieved

“ MEDICAL SCIENCES for the first time. For instance, HSPA5 was first described as Ig by retroviral activation of 4-OHT–inducible Cre in Prdm1fl/fl ” heavy chain-binding protein (BiP), which binds to Ig heavy- BCR-ABL1 pre-B ALL cells and deletion was verified after 1, 2, chain proteins in pre-B cells (10). In mice lacking the trans- and 5 d (SI Appendix, Fig. S5 B and C), significantly reducing the −/− membrane exon of the Ig μ-heavy chain (Ighm ) (21), B-cell expression and activity of its target Xbp1, indicating that Prdm1 development is arrested at the pre-BCR checkpoint. Interestingly, is functionally active in BCR-ABL1 pre-B ALL cells (SI Ap- − − early pre-B cells from Ighm / mice express high levels of Xbp1-s; pendix, Fig. S5D). Interestingly, inducible activation of the however, reconstitution of Ig μ-heavy-chain expression reinstated oncogenic TEL-SYK tyrosine kinase was shown to result in

Fig. 2. Regulation of XBP1 by STAT5, ERK, and A B C AKT (positive) and BACH2 and BCL6 (negative) in + 50 350 Ph ALL. (A) XBP1-s mRNA levels were measured 40 300 by qRT-PCR in sorted B-cell fractions from healthy 250 30 51 XBP1-s 200 human donors compared with patient-derived β-actin 20 150 pre-B ALL cells (ALL-Q5, ALL-O2, ALL-X2, BLQ11, 100 + 10 Human Ph+ ALL = 50 and LAX9; n 3) and (B) in two CD19 MACS- s mRNA levels [% COX6B] E 0 10 0 10 IM [μM] 0 0 enriched pre–B-cell samples and five patient-de- 51 XBP1-s

XBP1 Patient-derived Patient-derived Patient-derived – = Pre-B ALL XBP1s mRNA levels [% COX6B] Pre-B ALL Multiple Myeloma rived multiple myeloma samples (MM1 5; n 3). β-actin (C) Western blot analysis of spliced XBP1-s for + MACS-enriched normal human CD19 pre-B cells

D F fl/fl G + H + compared with patient-derived pre-B ALL cells (as XBP1s mRNA in human Ph+ ALL XBP1s mRNA in Stat5a/b ALL XBP1s mRNA in human Ph ALL XBP1s mRNA in human Ph ALL

in B) and one patient-derived multiple myeloma P=0.0017 P=0.0207 P=0.0001 ] P=0.01 P=0.002 P=0.02 P=0.004 P=0.0001

(MM6) sample using β-actin as loading control. (D) 4 12 6 Hprt 3 25 0.6 12 25 + 10 20 0.5 10 20 XBP1-s mRNA levels were measured in Ph ALL cell 3 2 0.4 8 lines (NALM1, TOM1, and BV173) treated with or 8 4 15 15 2 6 0.3 6 μ 10 10 without the TKI IM for 16 h (10 M IM) by qRT-PCR 4 2 1 0.2 4 1 5 5 (n = 3). (E) Protein levels of XBP1-s were measured 2 0.1 2 PD325901 AZD5363 + 0 0 0 0 0 0 0 0 Xbp1s mRNA levels [% fl/fl [nM] [μM] XBP1s mRNA levels [% COX6B] by Western blot analysis in patient-derived Ph EV Cre Stat5a/b XBP1s mRNA levels [% COX6B] 050 050 010 010

XBP1s mRNA levels [% COX6B] 010 010 0 10IM [μM] ALL cases (LAX9 and PDX59) treated with IM for 16 h (10 μM IM) using β-actin as loading control. I JLK Xbp1s (F) Xbp1-s mRNA levels were measured in Stat5a/ P=0.3 P=0.025 P=1.7e-6 P=0.4 P=5.7e-5 P=0.003 P=0.0027 ] P=0.009 P=0.02 Hprt 10 fl/fl 80 12 b ALL cells with EV control and 4-OHT–inducible IgG IgG] IgG BACH2 15 BCL6 10 8 Cre (Cre) after 24 h by qRT-PCR. (G and H) XBP1-s 60 8 6 10 mRNA levels were measured in patient-derived 6 10 4 4 pre-B ALL cases (PDX2 and LAX7R) treated with 5 2 5 2 Fold enrichment [% or without PD325901 or AZD5363 for 5 h (50 nM Fold enrichment [% IgG] Xbp1s mRNA levels [% 0 0 Xbp1s mRNA levels [% Hprt] 0 0 PD325901; 10 μM AZD5363) by qRT-PCR (n = 3). GAPDH XBP1 BACH2 GAPDH XBP1 BCL6 Bach2+/+ / BCL6 +/+ / +/+ / BCL6 0022IM [μM] (I) Quantitative ChIP validation of BACH2 bind- -20bp +1 -1kb +1 +10kb +1 -20bp +1 -1kb +1 +214bp + ingtotheXBP1 promoter in a human Ph ALL patient sample (ICN1), GAPDH as a negative and BACH2 as a positive control.(J) Quantitative ChIP validation of BCL6 binding to the XBP1 promoter in human pre-B ALL cells. (K) XBP1 expression was assessed in Bach2−/− ALL cells compared with WT controls (Bach2+/+) by qRT-PCR using specific primers for spliced Xbp1-s (n = 3). (L) Xbp1s mRNA levels were measured by qRT-PCR in Bcl6−/− ALL cells compared with WT controls (Bcl6+/+) treated either with or without the TKI IM for 16 h (2 μMIM).

Kharabi Masouleh et al. PNAS Early Edition | 3of10 Downloaded by guest on September 28, 2021 Prdm1-dependent terminal differentiation of B cells into plasma of Ras-MAPK and PI3K/AKT pathways, STAT5 represents cells (23). Nevertheless, activation of Prdm1 downstream of the a critical mediator of BCR-ABL1 downstream signaling (24). + BCR-ABL1 tyrosine kinase in human Ph ALL cells was not Inducible deletion of Stat5a and Stat5b (25) in BCR-ABL1– sufficient to induce terminal differentiation and did not affect transformed Stat5a/bfl/fl ALL cells reduced Xbp1-s mRNA levels expression levels of CD19 and CD138 as in plasma cells and to a degree similar to inhibition of BCR-ABL1 itself (Fig. 2F). multiple myeloma (2) (SI Appendix, Fig. S5E). Interestingly, Cre- Besides STAT5, oncogenic downstream signaling from BCR- mediated ablation of Prdm1 induced (day 2: P = 0.004; ABL1 is also involves the PI3K/AKT and RAS-MAPK pathways. day 5: P = 0.0002; SI Appendix, Fig. S5F) and cell-cycle arrest (SI For this reason, we treated patient-derived pre-B ALL cells with Appendix, Fig. S5G) in BCR-ABL1 pre-B ALL cells. We con- either an AKT (AZD5363) or MEK inhibitor (PD325901). Both clude that Prdm1 is important for survival and proliferation in inhibition of MEK (SI Appendix, Fig. S6A) and AKT (SI Ap- + a mouse model of human Ph ALL. pendix, Fig. S6B) significantly reduced XBP1s mRNA levels to a comparable degree as seen after genetic deletion of STAT5 + Oncogenic Tyrosine Kinase Activity Drives Expression of XBP1-s in Ph (Fig. 2 G and H). We conclude that BCR-ABL1 oncogenic ty- ALL Cells. As highlighted in our gene expression analysis (Fig. rosine kinase activity up-regulates expression and activity of 1A), increased ER stress owing to acute ablation of Hspa5 results XBP1 via STAT5, ERK, and AKT. in strong up-regulation of Xbp1 mRNA in BCR-ABL1 pre-B ALL cells, as also suggested by our microarray analyses (SI Ap- XBP1 Is a Target and Transcriptionally Repressed by Both BACH2 and + pendix,Fig.S3K), while not distinguishing XBP1-u and XBP1-s BCL6 in Ph ALL. Recent work from our laboratory had identified (9). Therefore, using qRT-PCR primers that specifically am- BCL6 (26) and BACH2 (27) as transcriptional repressors that plify XBP1-s, we confirmed that XBP1-s mRNA levels were 5- to oppose activation signals downstream of BCR-ABL1 and STAT5 + 41-fold higher in patient-derived pre-B ALL cells compared with in Ph ALL. Contrary to down-regulation of XBP1-s expression, normal bone marrow B-cell precursor populations (Fig. 2A; P = both BCL6 (26) and BACH2 (SI Appendix, Fig. S6C) are strongly + 0.001), albeit up to sevenfold lower than in some cases of mul- up-regulated by TKI (IM) treatment of Ph ALL cells. ChIP tiple myeloma (Fig. 2B). At the protein level, similar differences sequencing (ChIP-seq) and ChIP-on-chip analysis revealed were observed (Fig. 2C). Because the oncogenic BCR-ABL1 strong binding of both BACH2 and BCL6 to the XBP1 pro- tyrosine kinase accounts for ∼30% of cases of adult ALL (13), moter (SI Appendix,Fig.S6D), which was further enhanced by we tested whether oncogenic BCR-ABL1 tyrosine kinase activity TKI treatment (IM) (SI Appendix,Fig.S6E). Specific binding + + is responsible for increased levels of Xbp1-s in Ph ALL com- of BACH2 and BCL6 to XBP1 promoter regions in Ph ALL pared with normal pre-B cells and other ALL subsets lacking cells was confirmed by single- quantitative ChIP (Fig. 2 I − − an oncogenic tyrosine kinase (SI Appendix, Fig. S3K). Consistent and J). Genetic loss of function models for Bach2 / and − − − − with this hypothesis, tyrosine kinase inhibitor (TKI) (imatinib, BCL6 / demonstrated that Xbp1-s mRNA levels in Bach2 / + − − − − IM) treatment of human Ph ALL cells reduced mRNA (Fig. 2D) and BCL6 / BCR-ABL1 ALL cells were eightfold (Bach2 / ) − − and protein levels (Fig. 2E; 54 kDa) of XBP1-s. Besides activation and sixfold (BCL6 / ) higher compared with WT controls

ACfl/fl fl/fl Xbp1 ALL H2O Day 1 Day 2 Xbp1 ALL WT EV Cre EV +4-OHT Cre +4-OHT EV P=0.01 P=3.8e-5 bp 80 500 400 22 18 60 In vitro In vivo 40 B Xbp1fl/+ Hspa5fl/fl Cre Fig. 3. Inducible ablation of Xbp1 causes cell-cycle EV Cre EV Cre Ndn Cell cycle arrest 20 arrest and apoptosis in transformed pre-B cells. (A) Hexa Lysosomal beta-hexosaminidase Csnk1e Casein kinase epsilon 46 74 Hsp90b1 ER heat shock Genomic deletion of Xbp1 was verified using specific Annexin V Ctla4 Inhibitory receptor, CD80 is ligand of Number apoptotic cells [%] 0 fl/fl Ly6f CD59, activation/stress antigen EV Cre EV Cre primers to amplify floxed alleles in Xbp1 ALL cells Derl3 Derlin-3, Export from the ER Cd80 CD80, induced by CTLA4 – Rab32 Export from the ER Day 1 Day 2 transduced with EV control or 4-OHT inducible Cre Stxbp6 Export from the ER 7 AAD Mycl1 L- (Cre) after in vitro deletion or in vivo (three mice Rps6kc1 Ribosomal protein S6 kinase Rtn1 ER membrane synthesis Atf6 ER, Unfolded Protein Response after treatment with 4-OHT) WT and H2O controls. Rxra Xbp1fl/fl ALL Sdf2l1 ER, Unfolded Protein Response D Fbxo32 ligase S-phase (B) Gene expression profiling of Cre-mediated de- Il2ra CD25, activation/stress antigen EV P=0.0002 Pde4c cAMP phosphodiesterase S: 33 letion of Xbp1- or Hspa5-deletion in BCR-ABL1– Cbln1 Export from the ER + Pbx3 pre-B-cell leukemia 3 100 G /M fl/ fl/fl Evi2a Protooncogene, disrupts Nf1 2 transformed Xbp1 or Hspa5 pre-B ALL cells. Mgll Fatty acid saturase G2/M 6 S phase Cdc25c Cell cycle progression Highlighted are molecules related to the secretory Melk Maternal embryonic leucine kinase 80 Sirpa Secretory, Signal-regulatory protein G0/1 61 Lcp2 Lymphocyte cytosolic protein 2 and Golgi apparatus (Sirpa, B3gnt5, Cst7, Gldc, Fcrl1 Fc receptor-like 1, inhibitory receptor 60 G0/1 B3gnt5 Secretory, -1,3-N-acetylglucosaminyl-TF Cst7 Secretory, Cysteine-type endopeptidase Cre Hs3st2, and Slc16a3,underlined), Ig light chains and Dkk3 Dickkopf WNT signaling inhibitor 3 S: 6 Gldc Glycine dehydrogenase 40 the early B-cell antigens Enpep and Cd22 (high- Hs3st2 Secretory, O-sulfation of glucosamine Ptgs1 Cox1, prostaglandin-synthase Igk-C Immunoglobulin light chain G /M 9 20 lighted in blue) and ER chaperons (Hsp90b1, Derl3 Mgst1 Microsomal glutathione S-transferase 1 2 Enpep CD249, expressed on early B-cells BrdU Ak3l1 GTP:AMP and ITP:AMP transferase G0/1 85 and Atf6, highlighed in red; GEO accession nos. Cd22 CD22; B cell-specific receptor 0 Eps8 EGFR signaling EV Cre GSE53683 and GE53684). (C) Apoptosis was mea- Slc16a3 7 AAD Secretory, Warburg metabolic target fl/fl Log2 fold change sured by Annexin-V and 7AAD staining in Xbp1 -2 -1 0 +1 +2 ALL cells with EV or Cre after 1 and 2 d 4-OHT fl/fl EFXbp1fl/fl ALL, 24 hours after induction of Cre or EV Xbp1fl/fl ALL, 24 hours Xbp1fl/fl ALL, 24 hours treatment (n = 3). (D) Xbp1 ALL cells with EV or EV 8.3 ± 1.8 EV Cre were stained with BrdU and 7AAD and the per-

centages of cells in the G0/1, S, and G2/M cell cycles P= 0.0001 are indicated after 24 h of 4-OHT treatment and 70 = fl/fl 60 shown in bar graph (n 3). (E) Xbp1 ALL cells with 50 EV or Cre were stained with ER tracker green after Cre 63 ± 1.5 Cre fl/fl 40 24 h of 4-OHT treatment (n = 3). (F) Xbp1 ALL cells 30 with EV or Cre were stained in a senescence-associ- 20 ated β-galactosidase assay after 24 h of 4-OHT 10 treatment and percentages of positively senescent

Number of cells [%] Number senescent 0 EV Cre cells are indicated and shown in bar graph (n = 3).

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1400958111 Kharabi Masouleh et al. Downloaded by guest on September 28, 2021 (Fig. 2 K and L). These findings are consistent with our obser- tion of Hsp90b1, Derl3,andAtf6 (highlighted in red, Fig. 3B)to PNAS PLUS vation of down-regulation of Xbp1-s levels during normal pre–B- potentially mitigate increased ER stress following Xbp1 deletion. cell differentiation after the pre-BCR checkpoint (SI Appendix, This interpretation would be consistent with up-regulation of the Fig. S3 B and C). De novo expression of Ig μ-heavy chains results stress/activation markers Il2ra (Cd25), Cd80, and Ly6f (Cd59). in rapid down-regulation of Xbp1-s in parallel with up-regulation The ER-stress sensor ERN1 initiates splicing and activation of of its transcriptional repressors Bach2 (27) and BCL6 (28) at Xbp1 and is negatively regulated by the ER chaperone Hspa5 this stage. (BiP) (31). Given the importance of ERN1-mediated splicing for Xbp1 function, we hypothesized that inducible deletion of Hspa5 Inducible Ablation of Xbp1 Causes UPR-Related Gene Expression Change will result in a net gain of Xbp1 function. On this basis, we pre- in Transformed Pre-B Cells. To test the function of the plasma-cell dicted that observed gene expression changes upon Xbp1 deletion transcription factor XBP1 in normal pre-B cells and transformed (Fig. 3B, Left), would be reversed by Hspa5 deletion (Fig. 3B, pre-B ALL (BCR-ABL1 and NRASG12D) cells, we developed a Right). In fact, only two molecules (Pbx3 and Evi2a) were regu- genetic system for inducible deletion of Xbp1 in BCR-ABL1– (29) G12D lated in opposite direction by deletion of Xbp1 and Hspa5. Of the and NRAS - transformed pre-B ALL cells. To this end, we < isolated bone marrow B-cell precursors from Xbp1fl/fl mice (30) 41 most significant Xbp1-dependent gene expression changes (P and cultured the cells in the presence of IL-7. To study the role of 0.001), 27 were recapitulated in the same direction by inducible Xbp1 in B-cell precursor ALL, we transduced IL-7–dependent ablation of Hspa5, indicating that most of Xbp1- and Hspa5-de- Xbp1fl/fl pro-B and pre-BI cells with retroviral BCR-ABL1 to model pendent gene expression changes are not direct targets of Xbp1 + human Ph ALL (13) and NRASG12D to model RAS-driven but are affected indirectly through increased levels of ER stress. childhood ALL (14). We first examined gene expression changes in response to Cre- Inducible Ablation of Xbp1 Causes Cell Cycle Arrest and Apoptosis in mediated deletion of the Xbp1 transcription factor (Fig. 3A and Transformed Pre-B Cells. Inducible deletion of Xbp1 resulted in SI Appendix, Fig. S7) in BCR-ABL1–transformed pre-B ALL increased apoptosis (Fig. 3C) and cell-cycle arrest (Fig. 3D)and cells (Fig. 3B). Genetic deletion of Xbp1 function resulted in induced ER stress in the vast majority of BCR-ABL1 ALL cells as decreased expression of molecules related to the secretory and measured with a recently established ER tracker dye (32) (Fig. Golgi apparatus (Sirpa, B3gnt5, Cst7, Gldc, Hs3st2,andSlc16a3, 3E). Furthermore, Cre-mediated deletion of Xbp1 resulted in underlined in Fig. 3B) as well as Ig light chains and the early an approximately eightfold increased propensity of leukemia cells to B-cell antigens Enpep and Cd22 (highlighted in blue, Fig. 3B). cellular senescence (P = 0.0001; Fig. 3F) as measured by staining for Conversely, ablation of Xbp1 induced compensatory up-regula- senescence-associated β-galactosidase (26). MEDICAL SCIENCES

ABXbp1fl/fl ALL Xbp1fl/fl ALL C D EV Cre EV Cre P=0.0056 P<0.0001 25 40 p53 JNK pT183/Y185 20 p27 30 JNK p21 15 P38α 20 180/ 182 p19/ARF pT Y 10 10 CHOP p38α 5 β-actin β-actin Dnajc3 mRNA levels [% Hprt] Dnajc3 mRNA 0 [% Hprt] levels mRNA Dnajb9 0 Xbp1fl/fl EV Cre Xbp1fl/fl EV Cre E Xbp1fl/fl BCR-ABL1 ALL EV + 4-OHT

Fig. 4. Loss of Xbp1 function activates proa- poptotic pathways and prolongs survival of BCR- Cre ABL1 pre-B ALL transplant recipient mice. (A) Pro- + 4-OHT tein levels of Arf (Cdkn2a), , p21 (Cdkn1a), p27 (Cdkn1b), and Chop were studied by Western blot analysis in Xbp1fl/fl ALL cells with (Cre) or without (EV) deletion of Xbp1 using β-actin as loading 5 21 28 31 Days after injection control (n = 3). (B) Likewise, phosphorylation of MAP kinases p38α (T180/Y182) and JNK1/2 (T183/Y185) F Xbp1fl/fl BCR-ABL1 ALL P=0.007 G Xbp1fl/fl BCR-ABL1 ALL were assessed (n = 3). (C and D) mRNA levels of DnajC3 (p58IPK) and DnajB9 (Erdj4) were mea- 100 EV + 4-OHT Cre + 4-OHT sured in Xbp1fl/fl ALL cells with (Cre) or without 80 (EV) deletion of Xbp1 by qRT-PCR (n = 3). (E) One million luciferase-labeled Xbp1fl/fl ALL cells with EV 60 or Cre were injected (i.v.) into sublethally irradi- ated (2.5 Gy) NOD/SCID mice and leukemic expan- Cre 40 CD19 sion was tracked by luciferase bioimaging. (F) + 4-OHT Overall survival of transplant recipient mice is 20 shown in a Kaplan–Meier analysis (n = 7 per EV B220 group). (G) A representative FACS analysis for sur-

Survival of Survival NOD/SCID mice [%] + 4-OHT 0 face markers CD19 and B220 (n = 4) is shown for 20 40 60 Xbp1fl/fl ALL cells isolated from killed mice.

Kharabi Masouleh et al. PNAS Early Edition | 5of10 Downloaded by guest on September 28, 2021 Loss of Xbp1 Function Activates Proapoptotic Pathways and Prolongs Xbp1fl/fl NRASG12D ALL Xbp1fl/fl NRASG12D ALL ABXbp1s mRNA Survival of BCR-ABL1 Pre-B ALL Transplant Recipient Mice. Consistent P=0.0001 120 with induction of cellular senescence upon loss of Xbp1 function, ] 8 100 EV we observed that Cre-mediated deletion of Xbp1 resulted in up- 6 6 cells [%] regulation of the checkpoint molecules Arf (Cdkn2a), p53, p21 + 80 4 4 (Cdkn1a), and p27 (Cdkn1b; Fig. 4A). Activation of C/EBP ho- 60

2 40 mologous protein (Chop) is the result of prolonged ER stress, [% HPRT levels mRNA 2 Fraction of GFP 20 Cre

ultimately leading to apoptosis (33). Cre-mediated deletion of Xbp1s 0 0 EV Cre 0812346 10 Xbp1 led to increased expression of Chop (Fig. 4A)andal- Time after transduction [days]

so caused increased phosphorylation of the proapoptotic MAP Xbp1fl/fl NRASG12D ALL Xbp1fl/fl NRASG12D ALL 183 185 C Day 1 Day 4 Day 7 D EV S-phase kinase Jnk1/2 (T /Y ), whereas increased phosphorylation of -5 180 182 4 10 6 S: 70 P=5.16 x 10 p38α (T /Y )maybeduetoanincreaseofglobalp38α protein 100

levels (Fig. 4B). Cre-mediated deletion of Xbp1 demonstrated that G2/M: 1 80

G0/1 29 Xbp1 is also required to relieve ER stress through activation of 60 12 21 28 Cre DnajC3 (p58IPK)andDnjaB9 (Erdj4; Fig. 4 C and D). DnajC3 is S: 33 40

known to reduce the overall substrate influx (34) and DnjaB9 is exin V nn BrdU 20 G2/M: 2 important in the stabilization of Hspa5 and thereby prevents the A G0/1 65 0 EV Cre accumulation of unfolded proteins (35). We next tested whether 7 AAD Cre-mediated deletion of Xbp1 affects the course of pre-B ALL 7 AAD fl/fl Xbp1 Pre-B cells Xbp1fl/fl Pre-B cells disease in an in vivo transplant model. To this end, 1 million BCR- E Day 1 Day 2 Day 5 F fl/fl – – S-phase ABL1 transformed Xbp1 pre-B ALL cells carrying 4-OHT 8.0 19 35 EV EV S: 54 T2 P=0.0001 inducible Cre (Cre-ER ) or an EV control were injected into 100 G2/M G2/M: 1 sublethally irradiated NOD/SCID mice. Pre-B ALL cells were 80 S phase G 44 labeled with firefly luciferase and leukemic expansion was moni- 0/1 fl/fl 18 27 73 Cre Cre 60 tored by luciferase bioimaging (Fig. 4E). Xbp1 leukemia cells S: 29 40 G0/1 transduced with EV rapidly expanded and caused lethal disease in xin V ne G /M: 2 2 20 An all recipient mice within 45 d. By contrast, inducible Cre-mediated BrdU G 69 0/1 0 deletion of Xbp1 delayed the onset of disease and substantially 7 AAD EV Cre prolonged overall survival of recipient mice (Fig. 4 E and F; P = 7 AAD 0.007). Analysis of infiltrating BCR-ABL1 pre-B ALL cells in mice Fig. 5. Xbp1 function is critical for the survival and proliferation of normal that succumbed to disease revealed that deletion of Xbp1 did not and NRASG12D transformed pre-B cells. (A) Xbp1 mRNA levels were measured cause a significant phenotypic change of the ALL cells (Fig. 4G). in Xbp1fl/fl NRASG12D ALL cells transduced with EV control or 4-OHT– inducible retroviral Cre (Cre) after 48 h of 4-OHT treatment by qRT-PCR (n = 3). (B) Xbp1fl/fl NRASG12D ALL were transduced with GFP-tagged EV control or Xbp1 Function Is Critical for the Survival and Proliferation of Normal + G12D 4-OHT–inducible GFP-tagged Cre (Cre) and the relative fraction of GFP cells and NRAS Transformed Pre-B Cells. Oncogenic lesions that re- = sult in hyperactivation of the RAS pathway occur in ∼50% of was measured (n 3). (C) Apoptosis was assessed by Annexin-V and 7AAD staining in Xbp1fl/fl NRASG12D ALL cells with EV and Cre after 1, 4, and 7 d of childhood ALL cases (14) and are particularly frequent in MLL- treatment with 4-OHT (n = 3). (D) Xbp1fl/fl NRASG12D ALL cells with EV and

rearranged and low hypodiploid ALL (36). To investigate Cre were stained with BrdU and 7AAD and the percentages of cells in the G0/1, whether our observations in BCR-ABL1–transformed pre-B S, and G /M cell cycles are indicated after 5 d of 4-OHT treatment and + 2 ALL and Ph ALL (∼30%ofALLcasesinadults)alsopertain shown in a bar graph (n = 3). (E) Apoptosis was assessed by Annexin-V and to RAS-driven ALL, we used as described above a NRASG12D - 7AAD FACS staining in IL-7–dependent Xbp1fl/fl pre-B cells with (Cre) or driven ALL model and induced deletion of Xbp1 using 4-OHT– without (EV) deletion of Xbp1 after 1, 2, and 5 d of treatment with 4-OHT fl/fl inducible Cre (Fig. 5A). As observed for BCR-ABL1–trans- (n = 3). (F) IL-7–dependent Xbp1 pre-B cells with EV and Cre were stained formed ALL, NRASG12D-driven ALL cells also underwent cell with BrdU and 7AAD and the percentages of cells in the G0/1, S, and G2/M cell cycles are indicated after 5 d of 4-OHT treatment and shown in bar graph (n = 3). cycle arrest and apoptosis (Fig. 5 B–D and SI Appendix, Fig. S7E), although the kinetics of cell death induction following ablation of Xbp1 were slower compared with BCR-ABL1 ALL targets in patients with pre-B ALL. For this reason, we tested the cells (Fig. 5B). We conclude that Xbp1 function is critical for ability of a recently developed small-molecule inhibitor of the survival and proliferation in two major subsets of high-risk ALL, + ERN1 endoribonuclease activity (STF-083010) (15) required for namely Ph ALL and ALL driven by hyperactive mutant RAS. XBP1 activation. First, we tested the specificity of STF-083010 Compared with mature B-cell subsets, normal pre-B cells express after treatment with the ER-stress inducer thapsigargin (TN) in XBP1-s at increased levels, albeit 5- to 41-fold lower than in pre-B patient-derived pre-B ALL cells (Fig. 6A). TN treatment alone ALL cells (Fig. 2A and SI Appendix, Fig. S3G). We, therefore, induced up-regulation of XBP1-s, whereas STF-083010 (ERN1i) tested how Cre-mediated inducible deletion of Xbp1 affects reduced XBP1-s in a dose-dependent manner (Fig. 6A and SI normal IL-7–dependent pro- and pre-BI cells that were trans- Appendix, Fig. S8A). A recent study showed that spontaneous duced with a 4-OHT–inducible Cre and EV controls. Although effects of Xbp1 deletion were less profoundthaninpre-BALLcells, hydrolysis of STF-083010 results in two fragments, termed A106 and Cre-mediated deletion of Xbp1 caused increased apoptosis and A107 (37) (Fig. 6B), and the A106 moiety retained full inhibitory activity on ERN1 endoribonuclease activity (37). Indeed, dose– G0/G1 cell-cycle arrest in normal pro- and pre-BI cells (Fig. 5 E and F and SI Appendix,Fig.S7F). These findings indicate that high ex- response curves for A106 and STF-083010 were virtually undis- pression levels of Xbp1 at the pro-B and pre-BI stages of B-cell tinguishable (SI Appendix,Fig.S8B and C). Removal of the A107 development (SI Appendix,Fig.S3B) reflect an important function of moiety results in a significantly smaller molecule and hence increases Xbp1 in regulating pre–B-cell survival. Therapeutic targeting of Xbp1 inhibitory specificity of the remaining A106 compound (37). in human ALL will most likely also affect normal B lymphopoiesis. Both A106 and STF-083010 also affected proliferation and survival (A106, Fig. 6C and SI Appendix, Fig. S8D; STF-083010, Effects of Pharmacological ERN1/XBP1 Inhibition in Patient-Derived Fig. 6D and SI Appendix, Fig. S9A) of patient-derived pre-B ALL Pre-B ALL Cells in Vitro and in Vivo. These findings collectively cells in a dose-dependent manner. Interestingly, mature B-cell suggest that ERN1 and XBP1 represent potential therapeutic lymphoma and multiple myeloma cells were significantly less

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1400958111 Kharabi Masouleh et al. Downloaded by guest on September 28, 2021 PNAS PLUS A E Patient-derived Pre-B ALL Patient-derived Pre-B ALL P 100 S-phase =0.0001 DMSO S: 26 100 G /M 80 2 S-phase G /M: 80 60 2 11 60 G0/1 63 40 G0/1 levels [% levels COX6B] 40

NA NA 20 STF 20

mR S: 9 30 M 0 0 TN G2/M: 8 DMSO STF; BrdU XBP1s 510153060 510153060 ERN1i 30 M mol/l] G0/1 82 Fig. 6. Efficacy of pharmacological inhibition of

7 AAD XBP1 activation on patient-derived pre-B ALL cells in vitro and in vivo. (A) Patient-derived pre-B ALL ERN1 endoribonuclease inhibitors B F Patient-derived Pre-B ALL ALL cells (ALL-Q5) were treated with either the ER-stress

0 30 60 STF083010 [ mol/l] μ O ALL-Q5 inducer TN (10 g) to activate XBP1 splicing (XBP1-s) H 3 alone, increasing concentrations of the ERN1 in- A-106 A-107 15 72 93 hibitor STF-083010 (0–60 μM; ERN1i), or in combi- STF-083010 nation for 24 h and mRNA levels of XBP1-s were = BLQ11 measured by qRT-PCR (n 3). (B) Chemical struc- BCR-ABL1T315I tures of the STF-083010 or its two hydrolysis prod- 9 24 76 C Patient-derived Pre-B ALL, ERN1i A106 ucts, A106 and A107, are shown. (C)Patient- derived pre-B ALL (ALL-Q5) were treated with A106 100 ERN1i 0 (0–60 μM) and TN (10 μg) and viability was assessed ERN1i 5 ALL-X2 80 ERN1i 10 by DAPI staining up to 4 d and relative cell viability ERN1i 15 60 12 38 88 is shown (n = 3). (D) Patient-derived pre-B ALL (ALL- Annexin V ERN1i 30 O2, ALL-Q5, and ALL-X2), Ph+ ALL (ICN1), B-cell 40 y [% DAPI cells] non-Hodgkin lymphoma (B-NHL; JEKO1 and TO- 7AAD

iabilit 20 LEDO) or multiple myeloma (JJN3) cells were trea- V Patient-derived pre-B ALL cells TN 10 ted with STF-083010 (0–60 μM) and the relative 0 ERN1i 60 G DMSO A106 100 1,000,000 viability was assessed by CCK-8 assay. (E)Patient- 012 3 4 Days 80 60 derived pre-B ALL cells (ALL-Q5) were stained with 40 BrdU and 7AAD and the percentages of cells in G0/1, 20 P=0.317 MEDICAL SCIENCES Ph+ ALL, B-NHL, Multiple Myeloma, ERN1i STF-083010 p=NS D 0 S, and G2/M cell cycle phases are indicated after 24 100 0204060 h of treatment with STF-083010 (30 μM) and shown 100 B-NHL 100,000 = 80 80 in bar graph (n 3). (F) Apoptosis was assessed by B-NHL 60 Multiple Annexin-V and 7AAD staining in patient-derived 40 60 Myeloma 20 P=0.272 pre-B ALL cells (ALL-Q5, BLQ11, and ALL-X2) after 3 0

ival of ival cells [%] d of treatment with STF-083010 at two concen- 40 02040 60

urv 100 trations (30 and 60 μM; n = 3). (G) Patient-derived S Patient- 50,000 80 20 derived pre-B ALL cells (ALL-Q5) were treated for 24 h with pre-B ALL 60 Ph+ ALL 40 A106 (30 μM) or DMSO control, labeled with firefly 0 20 P

Survival of Survival mice [%] =0.025 02040 60 ERN1i [ mmol/l] luciferase, and injected (i.v.) into sublethally irra- 0 0 40 60 diated (2.5 Gy) NOD/SCID mice at different dose 10,000 100 levels (106,105, 50,000, and 10,000 cells per mouse) 80 60 and leukemic expansion was tracked by luciferase 40 bioimaging. The overall survival of transplant re- 20 P=0.0017 0 cipient mice is shown in a Kaplan–Meier analysis 0 20 80 Day 44 after injection (n = 5 per group).

+ sensitive than patient-derived pre-B ALL and Ph ALL cells treatment with A106 was sufficient to significantly prolong survival (Fig. 6D and SI Appendix, Fig. S9A). Cell cycle analyses revealed of xenotransplant recipients (P = 0.0017; Fig. 6G). that treatment of patient-derived pre-B ALL cells with STF- These findings suggest that pharmacological inhibition of 083010 caused G0/G1 arrest (Fig. 6E). Interestingly, one case ERN1/XBP1 may be useful to overcome drug resistance and to + of Ph ALL carrying mutant BCR-ABL1T315I, which confers re- treat pre-B ALL to prevent relapse arising from a small number sistance to most currently available TKIs, was equally as sensitive of drug-resistant leukemia-initiating cells. as other pre-B ALL cases to STF-083010 treatment at concen- High XBP1 Expression at the Time of Diagnosis Predicts Poor Outcome trations previously achieved in vivo (15) (Fig. 6F). In a manner in Patients with Pre-B ALL. Based on genetic studies in mouse comparable to Cre-mediated deletion of Xbp1 (Fig. 3E), treat- + models for Ph and RAS-driven ALL, we demonstrated that ex- ment with A106 caused significant increased uptake of ER pression and activity of the plasma cell transcription factor tracker dye in a time- and dose-dependent manner (SI Appendix, XBP1 is critical for survival and proliferation of ALL cells in Fig. S9B). these two subsets. For this reason, we analyzed mRNA levels of To validate potential usefulness of pharmacological inhibition XBP1 in ALL samples at the time of diagnosis from two clinical + of ERN1/XBP1 in vivo, we performed a limiting dilution ex- trials, ECOG E2993 for Ph ALL in adults (n = 55) and COG periment to determine the impact of A106 on leukemia-initiating P9906 for high-risk childhood ALL (n = 207). One caveat of this cells in a xenotransplantation setting. We injected patient- analysis is the fact that currently available XBP1 microarray 6 5 derived pre-B ALL cells at different dose levels (10 ,10, probesets used in these analyses bind to the 3′UTR of the XBP1 50,000, and 10,000 cells) into sublethally irradiated NOD/SCID mRNA and do not distinguish between XBP1-u and XBP1-s mice (n = 5 per group). Single-agent inhibition of ERN1/XBP1 (SI Appendix, Fig. S10 A and B). qRT-PCR and Western blot had no significant effect on established leukemia when pre-B ALL (Fig. 2 A–C and SI Appendix, Fig. S5A) show, however, that cells were injected at high cell numbers (106 and 105;Fig.6G). At patient-derived pre-B ALL cells, like multiple myeloma cells, lower cell numbers (50,000 and 10,000), however, single-agent express higher levels of XBP-s compared with XBP-u.

Kharabi Masouleh et al. PNAS Early Edition | 7of10 Downloaded by guest on September 28, 2021 A B 100 ECOG E2993 n=55 Ph+ ALL 100 COG P9906 n=207 pre-B ALL P=0.017 80 80 XBP1Low WBCLow l [%] XBP1Low WBCHigh 60 60 XBP1High WBCLow

40 40 XBP1Low High High

rall surviva XBP1 WBC 20 20 Ove XBP1High P=8.95e-06 0

0 survival [%] Relapse-free 0 2 4 6 8 0 1 2 3 4 5 6 Time after diagnosis [years] Time after diagnosis [years] C D 100 COG P9906 n=207 pre-B ALL 100 COG P9906 n=207 pre-B-ALL

80 XBP1Low BACH2High 80 XBP1Low MRD XBP1High BACH2High XBP1High MRD XBP1Low BACH2Low 60 60 e survival e [%]

fre 40 High Low 40 XBP1Low MRD+ XBP1 BACH2 00000 20 P=0.00219 P=5.85e-05 XBP1High MRD+

elapse- 20 00000 R Relapse-free survival survival [%] Relapse-free 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 Time after diagnosis [years] Time after diagnosis [years]

E Oncogenes in pre-B ALL BCR-ABL1, NRASG12D Fig. 7. XBP1 expression and activity is a predictor of poor clinical outcome and a therapeutic target in + pre-B ALL. (A) In an analysis, Ph ALL patients UPR effectors Unfolded proteins PRDM1 (Fig. S5) (ECOG E2993, n = 55, log rank test P = 0.017) were segregated into two groups based on high or low mRNA levels in respect to the median mRNA value HSPA5 (Fig. 1) PAX5 of the XBP1 probeset and the overall survival was assessed. (B) In a multivariate analysis, pre-B ALL patients (COG clinical trial P9906, n = 207) were ERN1 (Fig. 6; S3) BACH2 (Fig. 2, S6) segregated into four groups based on high or low mRNA levels of XBP1 (higher or lower than median mRNA levels of XBP1) and WBC counts, (C) mRNA XBP1s (Fig. 2-5) levels of BACH2 (27), or (D) minimal residual disease (MRD) status. (E) A schematic overview summariz- ing the key findings of this study.

For a clinical trial (ECOG E2993) including 55 patients with by oncogenic tyrosine kinase activity via downstream targets such + Ph ALL, we segregated patients into two groups based on as Stat5, AKT, and ERK signaling (schematic in Fig. 7E). In + whether XBP1 mRNA levels in Ph ALL cells were higher mature B cells and B-cell lymphoma, XBP1 levels are low, owing (XBP1High)orlower(XBP1Low) than the median expression value to transcriptional repression by BACH2 and BCL6. Whereas + in this trial. Comparing these two groups, we found that patients Ph ALL is particularly frequent in adults (∼30%), RAS-driven + with Ph ALL in the XBP1High group had significantly worse leukemias account for ∼50% of childhood ALL (14). Modeling overall survival (P = 0.017; Fig. 7A). Likewise, multivariate analyses both diseases with BCR-ABL1 and NRASG12D oncogenes, ge- for patients in a pediatric trial with high-risk childhood ALL (COG netic ablation experiments established that Xbp1 is critical to 9906) revealed that high mRNA levels of XBP1 at the time of di- enable oncogenic signaling in both major subtypes of pre-B ALL. agnosis was a strong and independent predictor of poor clinical SI Appendix C P = Plasma Cell-Specific Gene Expression As a Result of Genetic Lesions outcome ( ,Fig.S10 ) relative to WBC counts ( + 8.95e-06; Fig. 7B), BACH2 mRNA levels (P = 0.0022; Fig. 7C and Targeting B-Cell Lineage Transcription Factors. Both adult Ph ALL SI Appendix,Fig.S10D), and minimal residual disease status 29 d and childhood pre-B ALL cells express high levels of PRDM1, after initiation of chemotherapy (P = 5.85e-05; Fig. 7D). Similarly, a key transcription factor that orchestrates terminal differentia- higher than median mRNA levels of the ER-stress sensor and ac- tion into plasma cells by extinguishing B-cell–specific gene ex- tivator of XBP1-s, ERN1, were predictive of failure to achieve pression (2), for instance through repression of PAX5, IKZF1, + complete remission in patients with Ph ALL (P = 0.01; n = 53, EBF1, and BACH2 (http://lymphochip.nih.gov/blimp/). Con- ECOG 2993; SI Appendix,Fig.S10E and F). versely, PAX5 (38) and, as we show here, BACH2, are tran- scriptional repressors of XBP1 (schematic in Fig. 7E). In this + Discussion context, it seems relevant that adult Ph ALL (39) and childhood Here we show that molecules that typically mitigate ER stress in pre-B ALL (40) are characterized by frequent genetic lesions in plasma cells (ERN1, HSPA5, PRDM1, and XBP1) are expressed B-cell lineage transcriptions factors (IKZF1, PAX5, EBF1, and at surprisingly high levels specifically at the pre-BCR checkpoint BACH2) (27) that oppose plasma cell differentiation, including + during early B-cell development. Genetic studies of inducible expression of XBP1. Although Ph ALL and childhood pre-B deletion of Hspa5, Prdm1, and Xbp1 demonstrated that these ALL cells clearly do not exhibit a plasma cell-like phenotype, molecules are not only critical for survival and proliferation of genetic lesions targeting IKZF1, PAX5, EBF1, and BACH2 may early pre-B cells but in particular in pre-B ALL cells that are facilitate expression of PRDM1 and XBP1 and, hence, enable typically arrested at developmental stages before or at the pre- pre-B ALL cells to mitigate ERstress resulting from strong + BCR checkpoint. Ph ALL cells carrying the BCR-ABL tyrosine oncogenic signaling. Here we studied the dependency of pre-B kinase exhibit particularly high levels of XBP1, which is induced ALL cells transformed by BCR-ABL1 and NRASG12D oncogenes

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1400958111 Kharabi Masouleh et al. Downloaded by guest on September 28, 2021 on XBP1 expression and activity. A previous study showed that using a BD FACSAriaII (BD Biosciences). Gating strategies and analysis of the PNAS PLUS overexpression of TCL1 results in increased ER stress (37). In population purity are shown in SI Appendix, Fig. S1A. addition, results by us and others demonstrate that oncogenic tyrosine kinase activity results in ectopic expression of Prdm1 Flow Cytometry. About 1 million cells per sample were resuspended in PBS blocked using Fc blocker for 10 min on ice, followed by staining with the (23) and XBP1 (this study; Fig. 7E). appropriate dilution of the for 15 min on ice. Cells were washed μ – and resuspended in PBS with propidium iodide (0.2 g/mL) as a dead cell B-cell Specific Transcription Factors Limit the Ability of Pre-B ALL marker. The antibodies used for flow cytometry are listed in SI Appendix, Cells to Relieve Oncogene-Induced ER Stress. Further studies will Table S4. For Annexin V stainings, Annexin V binding buffer (BD Biosciences) was be needed to show whether genetic lesions targeting IKZF1, used instead of PBS and 7AAD (BD Biosciences) instead of propidium iodide. PAX5, EBF1, and BACH2 are required to enable surprisingly high levels of PRDM1 and XBP1 expression observed in pre-B Cell-Cycle Analysis. For cell-cycle analysis in BCR-ABL1 ALL cells, the BrdU flow ALL cells compared with normal pre-B cells. However, the un- cytometry kit for cell-cycle analysis (BD Biosciences) was used according to expected finding of high expression levels of ERN1, HSPA5, the manufacturer’s instructions. BrdU incorporation (FITC-labeled anti-BrdU PRDM1, and XBP1 in the context of known deletions of IKZF1, antibodies) was measured along with DNA content (7-amino-actinomycin-D) PAX5, EBF1, and BACH2 in pre-B ALL support the concept in fixed and permeabilized cells. The analysis was gated on viable cells that were identified based on scatter morphology. that pre-B ALL cells are distinctly vulnerable to ER stress. Our proof-of-principle studies using pharmacological inhibitors of ER Tracker Green. ER tracker green was used according to the manufacturer’s ERN1/XBP1 in patient-derived pre-B ALL cells support this instructions (Life Technology). For cytospin experiments, cells were incubated for scenario and indicate that combinations between conventional 15 min with ER tracker green and washed and cytospin preparations performed. chemotherapy and novel agents targeting ER stress and UPR may improve clinical outcomes for patients with drug-resistant MTT/CCK8 Proliferation Assays. Cell proliferation was determined by CCK8 pre-B ALL. proliferation kit (Dojindo) according to the manufacturer’s instructions.

Materials and Methods Inhibitor Studies. STF-083010 and A106 were previously described (15, 37) and Primary Human Samples and Human Cell Lines. We obtained primary cases (SI obtained from Axon Medchem and VWR. AZD5363 and PD325901 were − Appendix, Table S1) in compliance with the institutional review board of the obtained from Axon Medchem and dissolved in DMSO and stored at 20 °C University of California, San Francisco. The cells were cultured as described in for further experiments. The cells were treated with the inhibitors for spe- SI Appendix, Supplemental Material and Methods. cific time points as mentioned in discussion of the experiments. MEDICAL SCIENCES β β Extraction of Bone Marrow Cells from Mice. Bone marrow cells were extracted Senescence-Associated -Galactosidase Assay. Senescence-associated -galacto- from young age-matched mice with different genetic backgrounds (SI Ap- sidase activity was performed on cytospin preparations as described (26). For pendix, Table S2) and further processed as described in SI Appendix, Sup- a detailed description see SI Appendix, Supplemental Material and Methods. plemental Material and Methods. All mouse experiments were subject to approval by the Children’s Hospital Los Angeles Institutional Animal Care Single-Locus Quantitative ChIP, ChIP-on-Chip, and ChIP-Seq Analysis. Single- and Use Committee. locus quantitative ChIP, ChIP-on-chip, and ChIP-seq analysis were performed as previously described (26, 41). A detailed protocol and primer sequences + Mouse Model of Human Ph ALL. We collected bone marrow cells from the mice are available in SI Appendix, Supplemental Material and Methods. and retrovirally transformed them using BCR-ABL1 or NRASG12V in the presence of 10 ng·mL−1 IL-7 (Peprotech) in RetroNectin- (Takara) coated Petri dishes as Gene Expression Microarray and RNA Sequencing Analysis. Total RNA from cells described below. The cells were maintained as previously described (26). A de- used for microarray was isolated by RNeasy purification. Biotinylated cRNA tailed description is given in SI Appendix, Supplemental Material and Methods. was generated and fragmented according to the Affymetrix protocol and hybridized to mouse 430A 2.0 Array (Affymetrix). A detailed description is Retroviral Transduction. We performed transfections of retroviral constructs given in SI Appendix, Supplemental Material and Methods. and their corresponding empty vector controls (SI Appendix, Table S5)as previously described (26, 27). A detailed description is given in SI Appendix, Patient Outcome and Gene Expression Microarray Data. Gene expression Supplemental Material and Methods. microarray and patient outcome data were obtained from the GEO database [accession nos. GSE5314 (42) and GSE34941 (41)] of the Eastern Cooperative qRT-PCR and Genomic PCR. Total RNA from cells was extracted using RNeasy Oncology Group Clinical Trial E2993 for adult pre-B ALL, from GSE11877 and ’ isolation kit from Qiagen. cDNA was generated using a poly(dT) oligonu- GSE28460 (43) of the Children sOncologyGroupClinicalTrialP9906forpe- cleotide and the SuperScript III Reverse Transcriptase (Invitrogen). Quanti- diatric pre-B ALL (the National Cancer Institute TARGET Data Matrix, http:// tative real-time PCR was performed with the SYBRGreenER mix (Invitrogen) targetnci.nih.gov/dataMatrix/TARGET_DataMatrix.html), and from the St. Jude and the ABI7900HT real-time PCR system (Applied Biosystems) according to Research Hospital pediatric ALL clinical trial (raw data can be downloaded standard PCR conditions. Primers for qRT-PCR or genomic PCR are listed in SI from www.stjuderesearch.org/site/data/ALL3/). Appendix, Table S3. Array-Based Methylation Analysis Using HELP. The HELP (HpaII tiny fragment Western Blotting. After harvest, cells were washed twice with PBS and lysed enrichment by ligation mediated PCR) assay was performed as previously in CelLytic MT buffer (Sigma-Aldrich) supplemented with Mini Complete pro- published (41). A detailed description is given in SI Appendix, Supplemental tease inhibitor (Roche), 1% phosphatase inhibitor mixture (Calbiochem), and Material and Methods. 1 mM PMSF. After 10 min of incubation on ice and centrifugation at 11,000 × g – for 10 min at 4 °C, the protein concentration was determined by Survival Analysis. Kaplan Meier survival analysis was used to estimate overall Coomassie blue assay (Thermo Scientific). Protein samples were loaded survival. Basically, patients were segregated into two groups according on 4–20% Bis-Tris gradient gels and transferred on nitrocellulose membranes whether they had above or below the median expression level of the XBP1 or PVDF (BioRad). The primary antibodies used are listed in SI Appendix,Table probeset. Log rank test was used to compare survival differences between “ ” S4. For protein detection, the WesternBreeze immunodetection system (Invi- patient groups. R package survival version 2.35-8 was used for the sur- trogen) was used and light emission was either detected by film exposure or by vival analysis. the BioSpectum imaging system (UPV). ACKNOWLEDGMENTS. We thank Laurie H. Glimcher (Weill Cornell College of Medicine) for Xbp1fl/fl mice, Lothar Hennighausen (National Institute of Cell Sorting. For magnetic bead sorting, peripheral blood or bone marrow Diabetes and Digestive and Kidney Diseases) for Stat5a/bfl/fl mice, and Amy mononuclear cells were purchased from AllCells LLC and CD19-enrichment S. Lee (University of Southern California) for Hspa5fl/fl mice. This work is sup- was performed by magnetic bead cell sorting according to the manufacturer’s ported by National Institutes of Health/National Cancer Institute Grants instructions (Miltenyi Biotech). Cell sorting by flow cytometry was performed R01CA137060, R01CA139032, R01CA157644, R01CA169458, and R01CA172558

Kharabi Masouleh et al. PNAS Early Edition | 9of10 Downloaded by guest on September 28, 2021 (to M.M.), the William Lawrence and Blanche Hughes Foundation, Stand Up To TR2-01816 (to M.M.) and Leukaemia and Lymphoma Research (M.M.). M.M. Cancer American Association for Cancer Research Innovative Research Grant is a Scholar of the Leukemia and Lymphoma Society and a Senior Investigator IRG00909 (to M.M.), California Institute for Regenerative Medicine Grant of the Wellcome Trust.

1. Hetz C, Glimcher LH (2009) Fine-tuning of the unfolded protein response: Assembling 22. Shapiro-Shelef M, et al. (2003) Blimp-1 is required for the formation of immuno- the IRE1alpha interactome. Mol Cell 35(5):551–561. globulin secreting plasma cells and pre-plasma memory B cells. Immunity 19(4):607–620. 2. Shaffer AL, et al. (2002) Blimp-1 orchestrates plasma cell differentiation by extinguishing 23. Hug E, Hobeika E, Reth M, Jumaa H (2013) Inducible expression of hyperactive Syk in the mature B cell gene expression program. Immunity 17(1):51–62. B cells activates Blimp-1-dependent terminal differentiation. Oncogene, 10.1038/ 3. Shaffer AL, et al. (2004) XBP1, downstream of Blimp-1, expands the secretory appa- onc.2013.326. ratus and other organelles, and increases protein synthesis in plasma cell differenti- 24. Hoelbl A, et al. (2010) Stat5 is indispensable for the maintenance of bcr/abl-positive ation. Immunity 21(1):81–93. leukaemia. EMBO Mol Med 2(3):98–110. 4. Hetz C (2012) The unfolded protein response: Controlling cell fate decisions under ER 25. Cui Y, et al. (2004) Inactivation of Stat5 in mouse mammary epithelium during stress and beyond. Nat Rev Mol Cell Biol 13(2):89–102. pregnancy reveals distinct functions in cell proliferation, survival, and differentiation. – 5. Harding HP, et al. (2003) An integrated stress response regulates amino acid me- Mol Cell Biol 24(18):8037 8047. + tabolism and resistance to oxidative stress. Mol Cell 11(3):619–633. 26. Duy C, et al. (2011) BCL6 enables Ph acute lymphoblastic leukaemia cells to survive – 6. Leung-Hagesteijn C, et al. (2013) Xbp1s-negative tumor B cells and pre-plasmablasts BCR-ABL1 kinase inhibition. Nature 473(7347):384 388. mediate therapeutic proteasome inhibitor resistance in multiple myeloma. Cancer 27. Swaminathan S, et al. (2013) BACH2 mediates negative selection and p53-dependent – Cell 24(3):289–304. tumor suppression at the pre-B cell receptor checkpoint. Nat Med 19(8):1014 1022. 7. Messinger YH, et al.; Therapeutic Advances in Childhood Leukemia & Lymphoma 28. Nahar R, et al. (2011) Pre-B cell receptor-mediated activation of BCL6 induces pre-B – (TACL) Consortium (2012) Bortezomib with chemotherapy is highly active in advanced cell quiescence through transcriptional repression of MYC. Blood 118(15):4174 4178. 29. Li S, Ilaria RL, Jr., Million RP, Daley GQ, Van Etten RA (1999) The P190, P210, and P230 B-precursor acute lymphoblastic leukemia: Therapeutic Advances in Childhood Leu- forms of the BCR/ABL oncogene induce a similar chronic myeloid leukemia-like syn- kemia & Lymphoma (TACL) Study. Blood 120(2):285–290. drome in mice but have different lymphoid leukemogenic activity. J Exp Med 189(9): 8. Scheuner D, et al. (2001) Translational control is required for the unfolded protein 1399–1412. response and in vivo glucose homeostasis. Mol Cell 7(6):1165–1176. 30. Todd DJ, et al. (2009) XBP1 governs late events in plasma cell differentiation and 9. Lee K, et al. (2002) IRE1-mediated unconventional mRNA splicing and S2P-mediated is not required for antigen-specific memory B cell development. J Exp Med 206(10): ATF6 cleavage merge to regulate XBP1 in signaling the unfolded protein response. 2151–2159. Genes Dev 16(4):452–466. 31. Bertolotti A, Zhang Y, Hendershot LM, Harding HP, Ron D (2000) Dynamic interaction 10. Haas IG, Wabl M (1983) Immunoglobulin heavy chain binding protein. Nature 306(5941): of BiP and ER stress transducers in the unfolded-protein response. Nat Cell Biol 2(6): 387–389. 326–332. 11. Zhang K, et al. (2005) The unfolded protein response sensor IRE1alpha is required at 2 32. Dengler MA, et al. (2011) Oncogenic stress induced by acute hyper-activation of Bcr- distinct steps in B cell lymphopoiesis. J Clin Invest 115(2):268–281. Abl leads to cell death upon induction of excessive aerobic glycolysis. PLoS ONE 6(9): 12. Pui C-H, et al.; Total Therapy Study XIIIB at St Jude Children’s Research Hospital (2004) e25139. Improved outcome for children with acute lymphoblastic leukemia: Results of Total 33. Oyadomari S, Mori M (2004) Roles of CHOP/GADD153 in endoplasmic reticulum stress. Therapy Study XIIIB at St Jude Children’s Research Hospital. Blood 104(9):2690–2696. Cell Death Differ 11(4):381–389. 13. Westbrook CA, et al. (1992) Clinical significance of the BCR-ABL fusion gene in adult 34. Yan W, et al. (2002) Control of PERK eIF2alpha kinase activity by the endoplasmic acute lymphoblastic leukemia: A Cancer and Leukemia Group B Study (8762). Blood reticulum stress-induced molecular chaperone P58IPK. Proc Natl Acad Sci USA 99(25): – 80(12):2983 2990. 15920–15925. 14. Zhang J, et al. (2011) Key pathways are frequently mutated in high-risk childhood 35. Kurisu J, et al. (2003) MDG1/ERdj4, an ER-resident DnaJ family member, suppresses ’ acute lymphoblastic leukemia: A report from the Children s Oncology Group. Blood cell death induced by ER stress. Genes Cells 8(2):189–202. – 118(11):3080 3087. 36. Holmfeldt L, et al. (2013) The genomic landscape of hypodiploid acute lymphoblastic 15. Papandreou I, et al. (2011) Identification of an Ire1alpha endonuclease specific in- leukemia. Nat Genet 45(3):242–252. – hibitor with cytotoxic activity against human multiple myeloma. Blood 117(4):1311 1314. 37. Kriss CL, et al. (2012) Overexpression of TCL1 activates the endoplasmic reticulum α 16. Mimura N, et al. (2012) Blockade of XBP1 splicing by inhibition of IRE1 is a promising stress response: A novel mechanism of leukemic progression in mice. Blood 120(5): – therapeutic option in multiple myeloma. Blood 119(24):5772 5781. 1027–1038. 17. Brewer JW, Cleveland JL, Hendershot LM (1997) A pathway distinct from the mam- 38. Nera K-P, et al. (2006) Loss of Pax5 promotes plasma cell differentiation. Immunity malian unfolded protein response regulates expression of endoplasmic reticulum 24(3):283–293. chaperones in non-stressed cells. EMBO J 16(23):7207–7216. 39. Mullighan CG, et al. (2008) BCR-ABL1 lymphoblastic leukaemia is characterized by the 18. Luo S, Mao C, Lee B, Lee AS (2006) GRP78/BiP is required for cell proliferation and deletion of Ikaros. Nature 453(7191):110–114. protecting the inner cell mass from apoptosis during early mouse embryonic de- 40. Mullighan CG, et al. (2007) Genome-wide analysis of genetic alterations in acute velopment. Mol Cell Biol 26(15):5688–5697. lymphoblastic leukaemia. Nature 446(7137):758–764. 19. Reimold AM, et al. (2001) Plasma cell differentiation requires the transcription factor 41. Geng H, et al. (2012) Integrative epigenomic analysis identifies biomarkers and XBP-1. Nature 412(6844):300–307. therapeutic targets in adult B-acute lymphoblastic leukemia. Cancer Discov 2(11): 20. Turner CA, Jr., Mack DH, Davis MM (1994) Blimp-1, a novel -containing 1004–1023. protein that can drive the maturation of B lymphocytes into immunoglobulin- 42. Juric D, et al. (2007) Differential gene expression patterns and interaction networks in secreting cells. Cell 77(2):297–306. BCR-ABL-positive and -negative adult acute lymphoblastic leukemias. J Clin Oncol 21. Kitamura D, Roes J, Kühn R, Rajewsky K (1991) A B cell-deficient mouse by targeted 25(11):1341–1349. disruption of the membrane exon of the immunoglobulin mu chain gene. Nature 43. Hogan LE, et al. (2011) Integrated genomic analysis of relapsed childhood acute 350(6317):423–426. lymphoblastic leukemia reveals therapeutic strategies. Blood 118(19):5218–5226.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1400958111 Kharabi Masouleh et al. Downloaded by guest on September 28, 2021