Overexpression of SH2-Containing Inositol Phosphatase Contributes to Chronic Lymphocytic Leukemia Survival

This information is current as Simar Pal Singh, Marjolein J. W. de Bruijn, Catarina Velaso of September 29, 2021. Gago da Graça, Odilia B. J. Corneth, Jasper Rip, Ralph Stadhouders, Ruud W. J. Meijers, Stéphane Schurmans, William G. Kerr, Johanna ter Burg, Eric Eldering, Anton W. Langerak, Saravanan Y. Pillai and Rudi W. Hendriks

J Immunol 2020; 204:360-374; Prepublished online 13 Downloaded from December 2019; doi: 10.4049/jimmunol.1900153 http://www.jimmunol.org/content/204/2/360 http://www.jimmunol.org/ Supplementary http://www.jimmunol.org/content/suppl/2019/12/12/jimmunol.190015 Material 3.DCSupplemental References This article cites 50 articles, 17 of which you can access for free at: http://www.jimmunol.org/content/204/2/360.full#ref-list-1

Why The JI? Submit online. by guest on September 29, 2021

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2020 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Overexpression of SH2-Containing Inositol Phosphatase Contributes to Chronic Lymphocytic Leukemia Survival

Simar Pal Singh,*,†,‡ Marjolein J. W. de Bruijn,* Catarina Velaso Gago da Grac¸a,* Odilia B. J. Corneth,* Jasper Rip,* Ralph Stadhouders,*,x Ruud W. J. Meijers,† Ste´phane Schurmans,{ William G. Kerr,‖ Johanna ter Burg,# Eric Eldering,# Anton W. Langerak,† Saravanan Y. Pillai,*,1 and Rudi W. Hendriks*

Balanced activity of kinases and phosphatases downstream of the BCR is essential for B cell differentiation and function and is disturbed in chronic lymphocytic leukemia (CLL). In this study, we employed IgH.TEm mice, which spontaneously develop CLL, and stable EMC CLL cell lines derived from these mice to explore the role of phosphatases in CLL. Genome-wide expression m

profiling comparing IgH.TE CLL cells with wild-type splenic B cells identified 96 differentially expressed phosphatase genes, Downloaded from including SH2-containing inositol phosphatase (Ship2). We found that B cell–specific deletion of Ship2, but not of its close homolog Ship1, significantly reduced CLL formation in IgH.TEm mice. Treatment of EMC cell lines with Ship1/2 small molecule inhibitors resulted in the induction of caspase-dependent apoptosis. Using flow cytometry and Western blot analysis, we observed that blocking Ship1/2 abrogated EMC cell survival by exerting dual effects on the BCR signaling cascade. On one hand, specific Ship1 inhibition enhanced calcium signaling and thereby abrogated an anergic response to BCR stimulation in CLL cells. On the other

hand, concomitant Ship1/Ship2 inhibition or speciﬁc Ship2 inhibition reduced constitutive activation of the mTORC1/ribosomal http://www.jimmunol.org/ protein S6 pathway and downregulated constitutive expression of the antiapoptotic protein Mcl-1, in both EMC cell lines and primary IgH.TEm CLL cells. Importantly, also in human CLL, we found overexpression of many phosphatases including SHIP2. Inhibition of SHIP1/SHIP2 reduced cellular survival and S6 phosphorylation and enhanced basal calcium levels in human CLL cells. Taken together, we provide evidence that SHIP2 contributes to CLL pathogenesis in mouse and human CLL. The Journal of Immunology, 2020, 204: 360–374.

rotein phosphorylation downstream of the BCR represents crucial role in oncogenic signaling: Bruton’s tyrosine kinase the most common form of reversible posttranslational (BTK) and the serine/threonine kinase AKT (also known as pro- P modiﬁcation in B cells. This process is controlled by the tein kinase B) (1, 3). Their importance is demonstrated by the by guest on September 29, 2021 coordinated action of speciﬁc kinases and phosphatases, which exceptional clinical responses of small molecule inhibitors tar- add or remove phosphate groups, respectively (1, 2). The balanced geting either BTK (4–6) or PI3K, which is upstream of AKT (7, 8) activity of these enzymes determines the optimal BCR signaling in relapsed/refractory or treatment-naive CLL patients. Targeting threshold that is essential for B cell selection at various cellu- of PI3K/AKT/mammalian target of rapamycin (mTOR) signaling lar differentiation checkpoints (2). Conversely, aberrant kinase and its effector ribosomal protein S6 kinase is particularly at- activation is critical for survival of leukemic cells in various B cell tractive, because this pathway is activated downstream of both the malignancies, including chronic lymphocytic leukemia (CLL) (1). BCR and various chemokine receptors. Moreover, PI3K/AKT/ In CLL, which is characterized by the accumulation of mTOR/S6 signaling crucially controls protein synthesis, cell growth, CD5posIgMlow monoclonal B cells, two nonreceptor kinases play a proliferation, motility, and survival (9).

*Department of Pulmonary Medicine, Erasmus MC, NL 3000 CA Rotterdam, the performed experiments and analyzed the data. R.S. contributed to the analysis of Netherlands; †Department of Immunology, Erasmus MC, NL 3000 CA Rotterdam, bioinformatics data. E.E. and J.t.B. contributed to designing and performing Western the Netherlands; ‡Postgraduate School Molecular Medicine, Erasmus MC, NL blot experiments. A.W.L., S.S., and W.G.K. contributed to the writing of the manu- 3000 CA Rotterdam, the Netherlands; xDepartment of Cell Biology, Erasmus script. R.W.H. contributed to the research design and the writing of the manuscript MC, NL 3000 CA Rotterdam, the Netherlands; {LaboratoiredeGeńe´tique and supervised the study. Fonctionnelle, GIGA-Research Centre, Universite´ de Lie`ge, 4000 Lie`ge, Belgium; ‖ Address correspondence and reprint requests to Prof. Rudi W. Hendriks, Department of Department of Microbiology and Immunology, SUNY Upstate Medical University, Pulmonary Medicine, Erasmus MC Rotterdam, Room Ee2251a, Post Office Box 2040, Syracuse, NY 13210; and #Department of Experimental Immunology, Academic Med- NL 3000 CA Rotterdam, the Netherlands. E-mail address: [email protected] ical Center, University of Amsterdam, 1105 AZ Amsterdam, the Netherlands The online version of this article contains supplemental material. 1Current address: EpiExpressions, Rotterdam, the Netherlands. Abbreviations used in this article: 7-AAD, 7-aminoactinomycin D; BTK, Bruton’s ORCIDs: 0000-0002-2902-5899 (S.P.S.); 0000-0002-1060-5607 (R.S.); 0000-0001-8170- tyrosine kinase; CLL, chronic lymphocytic leukemia; DE, differentially expressed; 1447 (R.W.J.M.); 0000-0002-5579-3840 (S.S.); 0000-0002-4720-7135 (W.G.K.); 0000- GC, germinal center; Inpp, inositol polyphosphate phosphatase; IP4, inositol(1,3,4,5) 0002-2078-3220 (A.W.L.); 0000-0003-1871-918X (S.Y.P.). tetraphosphate; Mcl-1, myeloid leukemia cell differentiation protein 1; M-CLL, Received for publication February 6, 2019. Accepted for publication November 11, mutated CLL; mTOR, mammalian target of rapamycin; PI(3,4)P2, phosphatidylinositol 2019. (3,4)-bisphosphate; PI(3,4,5)P3, phosphatidylinositol (3,4,5)-trisphosphate; Ptpn, protein tyrosine phosphatase of the nonreceptor type; qRT-PCR, quantitative real-time This work was supported partly by the Dutch Cancer Society (KWF 2014-6564), the RT-PCR; RNA-seq, RNA sequencing; RPKM, reads per kb of a transcript per million Association for International Cancer Research (10-562) (to R.W.H.), a Netherlands mapped reads; U-CLL, unmutated CLL; Un-B, unstimulated naive splenic B; WT, Organisation for Scientific Research (NWO) Veni Fellowship (Grant 91617114) (to wild-type. R.S.), and an NWO Fellowship (to S.P.S.). S.P.S. designed the research studies, performed experiments, analyzed the data, and Copyright Ó 2020 by The American Association of Immunologists, Inc. 0022-1767/20/$37.50 wrote the manuscript. M.J.W.d.B., C.V.G.d.G., O.B.J.C., J.R., S.Y.P., and R.W.J.M. www.jimmunol.org/cgi/doi/10.4049/jimmunol.1900153 The Journal of Immunology 361

Despite the impressive clinical advances, it remains largely of the institutional review board MEC2015-741 (for CLL) and MEC2016- unexplored how the activity of these kinases is dysregulated in 202 (healthy controls) and in accordance with the declaration of Helsinki. CLL. Constitutive kinase signaling in CLL cells is accompanied by The BCR characteristics of CLL patients are included in Table I. PBMC fractions were isolated using Ficoll Hypaque (GE Healthcare, Little low surface IgM expression, elevated cell-autonomous calcium Chalfont, U.K.), according to the manufacturer’s instructions. Naive ma- signaling, and unresponsiveness to BCR stimulation (10–12). In ture B cells were isolated from healthy control PBMCs using FACS pu- these aspects, CLL cells resemble anergic B cells, but upon rification for CD19posCD27negIgDpos cells. The purity of naive mature . in vitro culture CLL cells readily upregulate surface IgM ex- healthy B cell samples was 95%, as determined by flow cytometry. pression and regain BCR responsiveness. Considering its transient Treatment of EMC cell lines, primary IgH.TEm cells, and nature, the maintenance of anergy in CLL seems to require con- human CLL cells with SHIP inhibitors tinuous BCR occupation by Ag next to dysregulation of intra- EMC cell lines were cultured in RPMI medium supplemented with 10% cellular signaling pathways. A key role for BCR signaling would FCS, 50 mg/ml gentamicin, 50 mM 2-ME (culture medium) at 37˚C, and be further supported by the finding that CLL with hypermutated 5% CO2 as previously described (11). IgH variable genes (mutated chronic lymphocytic leukemia The pan-SHIP1/2 inhibitor 3b-amino-5a-androstanane hydro- [M-CLL]) have a more favorable prognosis than unmutated CLL chloride (K118) has been described previously and was dissolved in (U-CLL), although it cannot formally be excluded that this dif- Milli-Q at 5 mM stock concentration (20). The SHIP2-specific inhibitor 3-[(4-chlorophenyl)methoxy]-N-[(1S)-1-phenylethyl]-2-thiophenecarboxamide ference simply reflects differences in the cell of origin. Moreover, (AS1949490) (21) was purchased from Echelon Biosciences (Salt Lake the BCR repertoire is highly restricted, because one-third of City, UT) and was dissolved in a dimethyl sulfoxide (Sigma-Aldrich) at patients can be classified as stereotypic CLL, in which BCRs are 10 mM stock concentration. highly similar across patients (1, 13). All inhibitor treatments were performed in culture medium at 37˚C, Downloaded from 5% CO2, except when otherwise indicated. For the analysis of apoptosis To investigate the role of phosphatases in CLL, we employed a and cell cycle, cells were incubated in the presence/absence of inhibitor or mouse model established in our laboratory based on sporadic vehicle for 6 and 24 h (EMC cells) or 6 h (human CLL). For Western blot expression of the oncogenic SV40 T Ag (IgH.TEm) (14). This CLL analysis, cells were incubated in the presence/absence of inhibitor for model parallels the commonly employed Em-Tcl-1 transgenic 4 and 24 h. For phospho-flow cytometry and calcium flux assays, cells mouse model in that both SV40 T Ag and Tcl-1 can inhibit apo- were cultured in the presence/absence of inhibitor for 3 and 24 h (EMC cell lines) or 1 h (human CLL). MACS-purified wild-type (WT) C57BL/6

m http://www.jimmunol.org/ ptosis through activation of Akt (15, 16). In IgH.TE mice, the mice splenic B cells and PBMCs from healthy donors were included as a SV40 T gene is inserted in anti-sense direction into the IgH locus, control. between the D and J genes. Consequently, IgH.TEm mice H H Caspase apoptosis assay spontaneously develop a CLL-like disease with age, characterized by the accumulation of monoclonal CD5+CD43+IgM+IgDlowCD19+ To detect caspase activation, cells were stained with Caspase-Glo 3/7 B cells. We have generated EMC cell lines (EMC4 and EMC6) (G8091; Promega) assay reagent as described previously (22). Lumines- m cence of the stained cells was measured on an Infinite 200 spectropho- derived from splenocytes of IgH.TE mice that developed CLL tometer (Tecan). (11). These EMC cells resemble IgH.TEm primary tumors in phenotype and exhibit constitutively active BCR signaling, de- Flow cytometry procedures pendent on Btk, as well as full activation of Akt/S6 signaling. Apoptosis analysis. Cells were stained for annexin V (PE; BD Biosciences) by guest on September 29, 2021 Transcriptome analysis of primary tumors from IgH.TEm and 7-aminoactinomycin D (7-AAD) (PE-Cy5; BD Biosciences) in diluted mice and EMC cell lines revealed a unique phosphatase signature. binding buffer (BD Biosciences) at room temperature in the dark for 15 min. Human CLL samples were additionally stained with anti-CD3 (BV711, SH2-containing inositol phosphatase (Ship2) was among the most Clone UCHT1; BD Biosciences), anti-CD19 (PE-Cy7, Clone SJ25C1/HIB19; prominently increased genes in this signature. In this study, we BD Biosciences), and anti-CD5 (AF700, Clone UCHT-2; BD Biosciences). show by conditional in vivo deletion of the Ship2 gene in IgH.TEm Cell cycle analysis. For DNA content-based cell cycle analysis, cells were mice and by in vitro inhibition experiments with EMC cell lines fixed with 70% ethanol and stained overnight with 1 mg/ml propidium and primary IgH.TEm tumors that Ship2 promotes mouse CLL iodide (PI; Sigma Aldrich) in the presence of 1% Triton X-100 (Sigma Aldrich) and RNAse (10 mg/ml) (Roche) at 4˚C in the dark. survival, most likely via the Akt/mTOR/S6 pathway. Interest- Phospho-flow cytometry. Mouse EMC cells or human CLL samples were ingly, SHIP2 expression is also increased in human CLL and stained with a live/dead marker (Invitrogen Probes). Cells were then fixed in vitro SHIP inhibition reduced AKT/mTOR/S6-associated and permeabilized in Foxp3 Staining Kit Fix/Perm Solution (eBioscience) cellular survival. at 37˚C for 10 min as per manufacturer’s instructions. For the identification of B lymphocytes in the control WT splenocyte fractions, cell suspensions were additionally stained with anti-B220 (allophycocyanin, Clone Materials and Methods RAS-6B2; BD Biosciences) and anti-CD3 (APCef780, Clone 17-A2; Mice eBioscience) at 4˚C for 30 min in Foxp3 Staining Kit wash buffer (eBioscience) after fixation. Similarly, for the identification of fl/fl fl/fl cre/+ Ship1 mice (17), Ship2 mice (18), and Mb1 mice (19) were kept B lymphocytes in human CLL samples, the cells were additionally stained on a C57BL/6 genetic background, and IgH.TEm mice (14) were on a with anti-CD5 (BV421, Clone UCHT2; BD Biosciences), anti-CD3 mixed C57BL/6 3 129/Sv background. C57BL/6 mice and 129/Sv were (BV711, Clone UCHT; BD Biosciences), and anti-CD19 (FITC, Clone purchased from Charles River Laboratories. To achieve homozygous de- HIB19; BD Biosciences). Subsequently, cells were stained with anti– fl/fl cre/+ letion of the Ship1 or Ship2 gene specifically in B cells, Ship1 .Mb1 p-Btk(Y223) (PE, Clone N35-86; BD Phosflow), anti–p-Erk (PE, Clone fl/fl cre/+ or Ship2 .Mb1 mice were generated, which were crossed with 20A; BD Phosflow), or unconjugated anti–p-S6 (Clone D68F8; Cell fl/fl fl/fl IgH.TEm.Ship1 or IgH.TEm.Ship2 mice to obtain CLL panels, re- Signaling Technologies) and PE-conjugated anti-rabbit secondary Ab spectively. All mice were bred and maintained in the animal care facility at (Jackson ImmunoResearch). the Erasmus MC, Rotterdam, the Netherlands. All animal studies were Calcium flux assay. EMC cells, WT mouse splenocytes, or human CLL reviewed and approved by the Erasmus MC Committee of animal samples were stained with fluorogenic probes: 1 mM of Fluo3-AM and experiments. 1 mM of Fura Red-AM (Life Technologies) in loading buffer (HBSS CLL samples from human patients medium, supplemented with 10 mM HEPES and 5% FCS) in a water bath at 37˚C for 30 min. To gate for untouched B cells, mouse cells were Primary patient material was obtained from peripheral blood of CLL pa- additionally stained with biotinylated anti-NK1.1 (Clone PK136; BD tients, whereas peripheral blood from healthy controls (.50 y of age) was Biosciences), anti-CD4 (Clone GK1.5; BD Biosciences), anti-CD8a obtained via Erasmus MC and via Sanquin blood bank (Rotterdam, the (Clone 53-6.7; BD Biosciences), anti–Ter-119 (BD Biosciences), anti-CD11c Netherlands). Diagnostic and control samples were collected upon in- (Clone N418; eBioscience), anti–Gr-1(Clone RB6-8C5; eBioscience), formed consent and anonymized for further use, following the guidelines and anti-FcεR1(Clone MAR-1; eBioscience) in the final 10 min of the 362 SHIP CONTRIBUTES TO CLL SURVIVAL

Fluo3-AM/FuraRed-AM staining, as well as biotinylated anti-CD3 (BD differentiation protein 1 (Mcl-1) (ab32087; Abcam). Binding was visual- Biosciences) and anti-CD33 (BD Biosciences). For the identification of ized using IRDye 680– or 800–labeled secondary Abs and an Odyssey biotinylated Abs, WT cells were washed and resuspended in flux buffer Imager (Li-Cor). Quantification of signal was performed using Odyssey (loading buffer supplemented with 1 mM CaCl2) with fluorochrome- 3.0 software. conjugated streptavidin in a water bath at 37˚C. Subsequently, cells were washed, resuspended in flux buffer, filtered, and left at room temperature Statistical analysis for at least 30 min. Before measurement, cells were placed at 37˚C for Statistical analysis was performed using GraphPad Prism software 5 min in a water bath. Basal calcium levels were measured in the first m 9 (San Diego, CA). The log-rank test was used to assess statistical significance minute. Cells were then stimulated with 20 g/ml goat anti-mouse F(ab )2 in survival between mouse groups. To evaluate differences in the size of anti-IgM (Jackson ImmunoResearch) or anti-human F(ab9) anti-IgM 2 various B cell subsets between two groups we used a Mann–Whitney U test. (Southern Biotech), and the calcium flux was measured for another 4 min. For evaluation of differences between SHIP inhibitor or vehicle-treated Measurements were performed on an LSR II flow cytometer (BD and untreated groups in in vitro assays, we used Kruskal–Wallis one- Biosciences) and results were analyzed using FlowJo-V10 software way ANOVA test corrected with Dunn multiple comparison test for (TreeStar). comparing multiple pairs. To evaluate differences between human CLL MACS purification and RNA isolation and mature naive B cells (after normalization to mature naive B cells), and between inhibitor-treated and untreated CLL samples from patients, we Splenic single-cell suspensions were prepared in MACS buffer (PBS/2 mM used Wilcoxon signed rank test. EDTA/0.5% BSA). Primary tumors from IgH.TEm mice were purified using MACS purification for CD19pos cells. Naive splenic B cells from 8- to 12 wk-old WT C57BL/6 mice were purified by MACS, as previously Results described (23). Purity of MACS-sorted naive B cells was confirmed by CLL from IgH.TEm mice express a unique flow cytometry (typically .99% CD19pos cells). To obtain fractions phosphatase signature Downloaded from of activated B cells, purified naive B cells were cultured in RPMI/FCS To identify genome-wide differentially expressed (DE) phosphatase culture medium in the presence of 10 mg/ml F(ab9)2 anti-IgM (Jackson ImmunoResearch) for 12 h. RNA was extracted with the RNeasy Micro kit genes in CLL we performed RNA-seq on primary CLL samples (Qiagen) according to manufacturer’s instructions. (tumor load . 95%, n =6)fromIgH.TEm mice. For comparison, RNA sequencing unstimulated naive splenic B (Un-B, n = 4) and anti-IgM stimulated (anti–IgM-B stimulated, n = 4) naive splenic B cells from WT mice The TruSeq RNA Library Prep kit (Illumina) was used to construct mRNA were included. We used a complete list of 221 unique phosphatase http://www.jimmunol.org/ sequencing libraries that were sequenced on an Illumina HiSeq 2500 (single entries (https://www.genenames.org/cgi-bin/genefamilies/set/1076) read mode, 36 bp read length). Raw reads were aligned using Bowtie to murine transcripts (RefSeq database) from the University of California at belonging to 31 different families to identify key phosphatases Santa Cruz mouse genome annotation (NCBI37/mm9) (24). Normalized that were DE between un-B, anti–IgM-B, and primary IgH.TEm gene expression levels quantified as reads per kb of a transcript per million CLL samples. mapped reads (RPKMs) were used for various clustering approaches. By applying a threshold of RPKM . 1, we first identified 162 of Unsupervised hierarchical clustering and principal component analysis ∼ were performed using R software (R studio version 1.1.383). For visual- these 221 phosphatase genes ( 73%) that were expressed in at ization of gene clustering analysis, heatmaps depicting row Z-scores of least one of the three experimental groups. Next, by analyzing RPKM values were generated using R. differential gene expression (.2-fold change in IgH.TEm CLL RNA sequencing (RNA-seq) data used in this study have been deposited compared with Un-B and/or anti–IgM-B cells; false discovery rate by guest on September 29, 2021 in the Gene Expression Omnibus database (accession number GSE117713; q , 0.05), we identified 96 DE genes (Fig. 1A, Supplemental https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE117713) (25). Gene expression data for Ship1 and Ship2 in various immunological Table I). Hierarchical clustering of expression values revealed cell types were obtained from the Immunological genome project that the six CLL samples clustered together, separately from the ultra-low input RNA-seq data (accession number GSE109125; https:// Un-B and anti–IgM-B cells, which we confirmed by principal www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109125 (26). component analysis (Fig. 1B). Interestingly, protein tyrosine Quantitative real-time RT-PCR analysis phosphatases of the nonreceptor type (Ptpn) 22 and Ptpn11 (SHP2), which have already been described to play a role in CLL, Quantitative real-time RT-PCR (qRT-PCR) analysis was performed with m TaqMan probes. Primers and probes were designed using the Ensembl were significantly higher in IgH.TE CLL than in WT B cells genome browser and Universal Probe Library (Roche Applied Science). (28, 29). We did not find differential expression of Inpp5d (Ship1), cDNA was generated using the RevertAid H-Minus First Strand cDNA but the expression of the Ship1 close homolog Inppl1 (Ship2) was Synthesis Kit (Thermo Fisher Scientific) in accordance with the manu- ∼6-fold higher in IgH.TEm CLL cells compared with un-B and facturer’s instructions. Briefly, each 15 ml of RT-PCR consisted of 7.5 ml m anti–IgM-B cells (p = 0.004). These 96 DE genes were also DE of TaqMan Universal Mastermix II (Thermo Fisher Scientific), 4.15 lof m nuclease-free water, 0.6 ml of forward and reverse primers (10 pmol/ml), in stable CLL-like cell lines generated from IgH.TE mice (11), and 0.15 ml of probe. qRT-PCR was performed on the 7300 Real-Time whereby the expression in the cell lines and primary CLL PCR system (Applied Biosciences). Gene expression was analyzed with an cells were highly correlated (Spearman correlation r2, r = 0.836; ABI Prism 7300 Sequence Detector and ABI Prism Sequence Detection p , 0.0001) (Fig. 1C). Software version 1.4 (Applied Biosystems). The housekeeping genes Gapdh and ubiquitin were used for normalization of expression of all To further strengthen the existence of a unique phosphatase quantified genes for mouse and human, respectively. Primers and probes signature in IgH.TEm CLL, we validated DE genes by qRT-PCR in used are available on request. an extended cohort of primary CLL samples (tumor load . 95%, n = 30) from IgH.TEm mice (Fig. 1D). Hereby, unstimulated naive Western blot splenic B cells from WT mice (n = 4) served as controls. We For Western blot experiments, EMC cells and WT naive splenic B cells selected robustly expressed genes belonging to two families of (purified by MACS as described) (23) were lysed in RIPA lysis buffer phosphatases, inositol polyphosphate phosphatases (Inpp) and (50 mM TrisCl [pH 7.4], 150 mM NaCl, 1 mM EDTA [pH 8], 0.1% SDS, 1% Nonidet P-40) containing protease and phosphatase inhibitors (Roche). Ptpn, as members of these families play a role in B cell biology or After sonication, cellular protein content was measured using the CLL (28–32). Validation by qRT-PCR showed that all selected BCA protein assay (Pierce, Thermo Fisher Scientific). Cellular proteins genes were significantly higher expressed in IgH.TEm CLL than in (30–50 mg) were loaded on an SDS–polyacrylamide gel and separated WT B cells (p , 0.05). Although Ship1 was not a DE gene in the using a Bio-Rad Mini-PROTEAN electrophoresis system (27). Proteins were then transferred onto Immobilon-P polyvinylidene difluoride mem- RNA-seq analysis, qRT-PCR analysis revealed increased expression branes (MilliporeSigma, Bedford, MA), which were probed with primary (∼9-fold) in the larger IgH.TEm CLL cohort, when compared with Abs against b-Actin (AB0145-200; SICGEN) and myeloid leukemia cell WT splenic B cells. Comparison of average expression fold-changes The Journal of Immunology 363 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 1. CLL cells from IgH.TEm mice express a unique phosphatase signature. (A) Hierarchical clustering analysis (top) and accompanying heat map showing differences in expression levels (RPKM, depicted as row Z-scores) of the 96 DE genes between unstimulated (Un-B, n = 4), anti-IgM– stimulated (aIgM-B, n = 4) WT splenic B cells, and CLL cells (n = 6) from IgH.TEm mice. (B) Principle component analysis using the 96 DE genes in Un-B (n = 4, gray), anti–IgM-B (n = 4, green) WT splenic B cells, and CLL cells (n = 6, red) from IgH.TEm mice. (Figure legend continues) 364 SHIP CONTRIBUTES TO CLL SURVIVAL as determined by RNA-seq (n = 6) and qRT-PCR (n = 30) for primary CLL samples revealed signiﬁcantly correlated trends for these 11 genes (Spearman correlation r2, r = 0.490; p = 0.017), validating our RNA-seq analysis when extrapolated to a larger cohort of IgH.TEm CLL (Fig. 1D). In summary, these data identiﬁed a unique phosphatase signature of 96 DE genes in IgH.TEm CLL. Because particularly Ship1 has been implicated in regulating BCR signaling, maintaining anergy in autoreactive B cells (33), and survival of multiple myeloma in vivo (34), we next focused on Ship1 and its close homolog Ship2 for further studies. Different effects of conditional Ship1 or Ship2 deletion on B cell development To explore expression of Ship1 and Ship2 in various hematopoietic cell types, we analyzed previously reported gene expression data (26) (Supplemental Fig. 1A). This revealed that Ship1 is stably expressed throughout B and T cell development. Expression of

Ship2 is substantially lower than Ship1 yet detectable across all Downloaded from B cell developmental stages and, at a lower level, also in T cells. Likewise, Ship1 expression was higher than Ship2 in various innate immune cell types. To study the role of Ship1 and Ship2 in B cell development in the context of CLL, we ﬁrst analyzed a cohort of ∼40 wk-old mice in

which the Ship1 and Ship2 genes were targeted specifically in the http://www.jimmunol.org/ B cell lineage using Mb1cre/+ transgenic mice (n $ 5 each group). In these experiments, mice were on a mixed 129/sv 3 C57BL/6 background and Mb1+/+ littermates served as controls. Deletion of Ship1 and Ship2 had no effect on early B cell development in the bone marrow, although Ship1 deficiency resulted in a significant reduction of recirculating IgMlow/posIgDpos B cells, as reported previously for young mice (35) (Fig. 2A, 2B, see Supplemental Fig. 1B for flow cytometry gating strategy). The absolute numbers of follicular B cells in the spleen were significantly reduced in by guest on September 29, 2021 Ship1-deficient—but not in Ship2-deficient—mice compared with control littermates (Fig. 2C). In the spleen, Ship1 or Ship2 deficiency had no effect on transitional or marginal zone B cells. Ship1-deficient mice showed higher numbers of germinal center (GC) B cells, compared with control littermates (Supplemental Fig. 1C, 1D). This did not seem to be the case for Ship2-deficient FIGURE 2. Differential effects of conditional Ship1 or Ship2 deletion on mice, which showed slightly reduced numbers of GC B cells, B cell development. (A) FACS plots depicting differences in recirculating pos pos memory cells, and nonswitched IgM CD138 plasma cells (IgMlow/posIgDhigh) B cell subsets in bone marrow of Ship1fl/fl.Mb1cre/+ mice (Supplemental Fig. 1D). Absolute numbers of splenic B-1 cells (n =5),Ship2fl/fl.Mb1cre/+mice (n = 6), and control mice (Ship1fl/flor Ship2fl/fl on were not significantly different, although Ship1-deficient mice had Mb1+/+background, n = 6). Shown are dot plots on gated CD19posB220high cells, higher proportions of B-1 cells compared with control littermates as detailed in Supplemental Fig. 1B. (B and C) Quantification of various B cell (Supplemental Fig. 1D, 1E). subsets in the indicated mouse groups. (B) In the bone marrow, gated CD19pos low neg neg neg In conclusion, these data show that at ∼40 wk of age Ship1 and cells were defined as pro-B cells (B220 sIgk/l CD2 intracellular Igm ), low k lneg neg mpos Ship2 have differential effects on mouse B cell development. large pre-B cells (B220 sIg / CD2 intracellular Ig ), small pre-B low k lneg pos mpos Deficiency for Ship1, but not for Ship2, was associated with cells (B220 sIg / CD2 intracellular Ig ), immature B cells (B220highIgMposIgDlow/neg), and recirculating B cells (B220highIgMlow/posIgDpos). reduced numbers of follicular B cells and spontaneous GC (C) In the spleen, gated CD19posB220high cells were defined as follicular B formation. cells (CD21negCD23pos), marginal zone (MZ) B cells (CD21posCD23neg), neg neg B cell–specific Ship2 deletion reduces tumor formation in or transitional B cells (CD21 CD23 ). *p , 0.05, **p , 0.01 (Mann– IgH.TEm mice Whitney U test). We generated IgH.TEm mice with B cell–specific homozygous deletion of Ship1 or Ship2, referred to as IgH.TEm.Ship12/2 and to define CLL incidence as an accumulation of .70% 2 2 IgH.TEm.Ship2 / mice, respectively. To monitor CLL onset, IgMbposCD5posCD43posCD19pos B cells in peripheral blood, as we collected blood every 3–6 wk. Their mixed 129/sv 3 described previously (14, 25). CLL onset was observed in control C57BL/6backgroundenabledustouseIgMa/IgMballotypes IgH.TEm mice and IgH.TEm.Ship12/2 mice at median ages of

(C) Correlation plot comparing gene expression (log2 RPKM + 1) between primary tumor from IgH.TEm mice (mean values of n = 6) and EMC cell lines (mean of n = 3) for the 96 unique phosphatase genes detected by RNA-seq (Spearman r2, r = 0.836, p , 0.0001). (D) Correlation plot comparing relative expression of 11 phosphatases to WT splenic B cells in primary tumors as measured by RNA-seq (RPKM, n = 6) and primary tumors (tumor load . 95%, n = 30) (Spearman r2, r = 0.490, p = 0.017). The Journal of Immunology 365

231 and 201 d, respectively (p = 0.55) (Fig. 3A). In contrast, the two IgH.TEm.Ship22/2 mice with circulating CLL cells did CLL formation was significantly reduced in IgH.TEm.Ship22/2 not develop splenomegaly, although .95% of the splenic mice (p , 0.001): only two of the 18 mice showed .70% CD19pos cells had the IgMbposCD5posCD43pos CLL phenotype IgMbposCD5posCD43posCD19pos cells in peripheral blood (Fig. 3B). (Fig. 3E, 3F). In all three groups, mice in which CLL cells were Accordingly, IgH.TEm and IgH.TEm.Ship12/2 mice showed not detected in peripheral blood at ∼42 wk of age consistently similar overall survival (Fig. 3C). The reduced tumor incidence in had close to equal proportions of IgMa- and IgMb-expressing IgH.TEm.Ship22/2 mice was associated with significantly im- CD19pos B cells in the spleen (Fig. 3F). In a single IgH.TEm. proved survival, compared with control IgH.TEm mice (p =0.0031) Ship22/2 mouse (ES2KO-04), however, we found that (Fig. 3D). IgMbposCD5posCD43posCD19pos B cells were abundantly present We analyzed various lymphoid organs in CLL-positive mice in the spleen (.80% of total live cells), whereas tumor load in within the three groups, either when mice reached the humane end blood was ,9% (Supplemental Table II). point or at the age of ∼42 wk (see Supplemental Table II for tumor The CLL phenotype in IgH.TEm.Ship22/2 mice did not appear load in various lymphoid organs). Both control IgH.TEm mice and to differ from that in the other two groups, regarding phospha- IgH.TEm.Ship12/2 mice developed significant splenomegaly and tidylcholine specificity of their BCR, S6 phosphorylation, or enlarged lymph nodes, compared with age-matched WT control basal and anti–IgM-induced calcium flux (data not shown), al- mice (Fig. 3E). When we compared surface marker profiles though statistical analysis was precluded by the low numbers of in splenic B cells from CLL-positive mice, the IgH.TEm and IgH.TEm.Ship22/2 leukemias. IgH.TEm.Ship12/2 mice exhibited massive accumulation of a In conclusion, whereas B cell–specific deletion of Ship1 did not

IgMbCD19posCD5pos B cell population with exclusive (∼97%) signiﬁcantly affect CLL formation, conditional deletion of Ship2 Downloaded from expression of the nontargeted IgMb allele (Fig. 3F). Interestingly, resulted in signiﬁcantly reduced CLL development and absence of http://www.jimmunol.org/

FIGURE 3. B cell–specific Ship2 deletion reduces tumor incidence in IgH.TEm mice. (A and B) Kaplan–Meier CLL incidence curves of (A) IgH.TEm (dotted line, n = 30) versus 2/2 IgH.TEm.Ship1 (solid line, n = 18) mice or by guest on September 29, 2021 (B) IgH.TEm (dotted line, n = 30) versus IgH.TEm.Ship22/2 (solid line, n = 18) mice. (C and D) Kaplan–Meier tumor-free survival curves of (C) IgH.TEm (dotted line, n = 30) versus IgH.TEm.Ship12/2 (solid line, n = 18) mice or (D) IgH.TEm (dotted line, n = 30) versus IgH.TEm.Ship22/2 (solid line, n = 18) mice. (E) Comparison of spleen weight between IgH.TEm and IgH.TEm.Ship12/2 mice at end-stage disease (.95% IgMbposCD5posCD43posCD19pos CLL-like cells in peripheral blood) or ∼42-wk- old mice with detectable CLL development in spleen (IgH.TEm Ship22/2). Dotted line represents average spleen weight of 42 wk-old WT (C57BL/6) control mice. *p , 0.05 (Mann– Whitney U test). (F) Flow cytometric profiles for IgMa versus IgMb and IgM versus CD5 on gated CD19posCD3negCD11bneg alive single lymphocyte cell fractions from spleen of non- tumor bearing (non-CLL, upper row) and tumor-bearing (CLL, lower row) mice of the indicated phenotypes. Numbers (percentages) indicate the proportions of cells within the specified gates. Data are representative for CLL samples described in Supplemental Table II. 366 SHIP CONTRIBUTES TO CLL SURVIVAL splenomegaly at ∼42 wk. From these findings we conclude that B cells (Fig. 5A, Supplemental Fig. 2A) (11). Both K118 and Ship2-mediated signaling in B cells plays a crucial role in the AS1949490 had minimal effects on constitutive active levels of induction or aggressiveness of CLL in IgH.TEm mice. p-Btk(Y223) and p-Erk in EMC cells (Fig. 5B, 5C, Supplemental Fig.2B,2C).However,3hofK118treatmentresultedin Ship inhibition decreases survival of EMC cells and primary substantial downregulation of p-S6 in both EMC cell lines m IgH.TE tumor cells in vitro (Fig. 5B, 5C, Supplemental Fig. 2B, 2C). Similar downregu- To investigate the role of Ship-mediated signaling in CLL, we lation of p-S6 levels was seen with AS1949490, but only after employed the IgH.TEm-derived CLL cell lines EMC4 and EMC6 24 h of treatment. Additionally, K118 treatment also resulted in (11). Cells were treated either with the pan-Ship1/Ship2 inhibitor downregulation of p-Akt(S473) in EMC cells (data not shown). K118 (3b-amino-5a-androstanane, hydrochloride) (20) or with We also investigated the effect of Ship inhibition in vitro on the selective Ship2 inhibitor AS1949490 (3-[(4-chlorophenyl)- leukemic cells from primary IgH.TEm-tumors (Fig. 5D). In- methoxy]-N-[(1S)-1-phenylethyl]-2-thiophene-carboxamide) (21). cubation with either 5 mM of K118 or 15 mM of AS1949490 As a control, vehicle-treated cells were included. resultedindownregulationofp-S6after3and24h,respec- To determine the effect of Ship inhibition on cell survival, tively, paralleling our findings in the EMC cell lines. K118-, AS1949490-, and vehicle-treated EMC cell lines were Because Ship1 is a negative regulator of calcium signaling in stainedwithannexinVand7-AADandanalyzedbyflow B lymphocytes (36), we measured anti–IgM-induced calcium cytometry (Fig. 4A). Both inhibitors significantly decreased the mobilization in untreated or K118 or AS1949490-treated EMC proportions of alive cells in a dose-dependent manner. However, cells. In contrast to WT B cells, which display a strong calcium the kinetics differed between the two inhibitors: whereas the flux upon anti-IgM stimulation, EMC cells exhibited low and Downloaded from IC50 value for K118 was reached after ∼6 h of incubation at unsustained calcium flux (Fig. 5E, Supplemental Fig. 2D). This 5 mM, the IC50 for AS1949490 was only reached following is in agreement with the anergic phenotype of these cell lines ∼24 h of incubation at 15 mM. (11). Treatment with K118 (5 mM) enhanced basal calcium levels Next, we performed propidium iodide DNA content analysis and rendered EMC cells fully unresponsive to BCR stimulation after ∼24 h of inhibitor treatment to investigate their effects on (Fig. 5F, Supplemental Fig. 2E). AS1949490 (15 mM) had limited

EMC cell proliferation and apoptosis. Both K118 and AS1949490 effects on basal and anti–IgM-induced calcium mobilization in http://www.jimmunol.org/ induced a significant increase ($4-fold) in the proportions of EMC cells, even after 24 h of exposure (Fig. 5G, Supplemental apoptotic (sub-G1) cells, validating our annexin V/7-AAD data Fig. 2F). (Fig. 4B). AS1949490 induced a significant dose-dependent de- In summary, EMC cell lines and primary IgH.TEm-tumors crease in the proportions of cycling cells (S/G2/M), whereas K118 exhibited high S6 phosphorylation, indicative of constitutive had a more moderate effect on the fraction of proliferating EMC Akt/mTORC1 signaling, which was supported (independently of cells (Fig. 4B). Btk) by the activity of both Ship1 and Ship2. In addition, Ship1- Next, we confirmed that the reduced cell viability seen upon mediated inhibitory signaling limits the high level of basal calcium incubation with Ship inhibitors was due to the induction of apo- signaling in EMC cells and contributed to their anergic phenotype. ptotic pathways, by measuring the activity of executioner caspase-3 by guest on September 29, 2021 and caspase-7 following treatment with K118 or AS1949490. We Ship inhibition decreases Mcl-1 expression in EMC cells and m observed up to ∼2-fold induction of caspase-3/7 after 6 h of primary IgH.TE -tumor cells in vitro treatment with K118 in a dose-dependent manner (Fig. 4C). Because the Akt/mTORC1 pathway was shown to be an important Treatment with AS1949490 resulted in moderate caspase-3/7 in- regulator of the translation of antiapoptotic Mcl-1 in CLL (37), we duction after 6 h (data not shown) and .2-fold induction after investigated the effects of inhibitor treatment on Mcl-1 protein 24 h (Fig. 4C). levels in EMC cells by Western blot analysis. We found that Mcl-1 Next, we further validated the findings obtained in EMC cell was constitutively expressed in EMC cells, as Mcl-1 protein levels lines in primary CLL cells from spleens of tumor-bearing IgH.TEm were higher in EMC cells than in anti–IgMposLPS-stimulated mice (n = 6). Primary CLL cells were treated with K118 or WT B cells (Fig. 6A). Both K118 and AS1949490 treatment AS1949490, stained with annexin V and 7-AAD and analyzed by ledtosubstantialdownregulationofMcl-1proteininEMC4and flow cytometry (Fig. 4D). Both inhibitors significantly decreased EMC6 cells (Fig. 6A). Interestingly, such downregulation was the proportions of alive primary IgH.TEm-tumors cells, paralleling also seen in EMC cells in the presence of the caspase inhibitor our findings in EMC cells. Quinoline-Val-Asp-Difluorophenoxymethyl Ketone (QVD). This Taken together, inhibition of Ship1 and Ship2 induced apo- suggests that Mcl-1 downregulation is independent of caspase ptosis of both EMC cell lines and primary IgH.TEm tumor cells activity, which is also induced upon Ship1/2 inhibition in in vitro. The two inhibitors differed in their potency (IC50)and EMC cells (Fig. 4C). Of note, we detected only limited effects kinetics, and Ship2 inhibition in particular resulted in decreased of Ship1/2 inhibitors on the protein levels of the proapoptotic proliferation. Bcl-2–like protein 11 (BIM) (data not shown), which interacts with Mcl-1 (27). Ship exerts a dual effect on the BCR signaling cascade in Next, we performed intracellular flow cytometry experiments EMC cells to evaluate Mcl-1 expression. Using this technique, induction of To investigate if Ship1 and Ship2 play a role in regulating con- Mcl-1 upon anti-IgM/LPS stimulation in WT B cells was readily stitutive BCR signaling in EMC cells, we measured the effects of detectable (Fig. 6B). Likewise, we could confirm downregulation Ship inhibition on phosphorylated Btk(Y223), Erk(T202/Y204), of Mcl-1 by Ship inhibition in EMC6 cells (Fig. 6C) and found and S6(S240/244; as a sensitive read-out of Akt/mTORC1 sig- parallel effects of K118 (5 mM; 4 h) and AS1949490 (15 mM; naling), as well as BCR-mediated calcium flux. We used 5 mMof 24 h) in a panel of six primary IgH.TEm-tumors (Fig. 6D). K118 and 15 mM of AS1949490, corresponding to the IC50 value Taken together, these findings show that EMC cells and pri- of the respective inhibitor. mary IgH.TEm-tumor cells exhibit constitutive Mcl-1 induction As reported previously, EMC cells exhibited high basal phos- downstream of Ship1 and Ship2, at least partly via the mTORC1 phorylation of Btk(Y223), Erk, and S6 in comparison with WT pathway. The Journal of Immunology 367 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 4. Ship inhibition decreases survival of EMC cells and IgH.TEm tumors in vitro (A) Viable (annexinVneg7-AADneg) cells were determined after 6 h (left) or 24 h (right) of treatment with the indicated concentrations of the pan-Ship1/2 inhibitor K118 (top) or the Ship2-specific inhibitor AS1949490 (bottom). Bar graphs represent the proportions of viable cells, normalized to untreated control cells (set to 1.0). Dot plot show representative flow cytometric analyses of EMC6 cells. (B) Cell cycle profiling in EMC cells after 24 h of treatment with the indicated concentrations of K118 (top) or AS1949490 (bottom). Histogram shows the gating strategy for DNA content (propidium iodide) analysis in the absence (gray) or presence (black) of SHIP inhibitor. Bar graphs represent the proportions of apoptotic (subG1, left) and cycling (S-G2-M, right) cells normalized to untreated control cells.(C) Analysis of caspase activity using the Caspase-glo assay in EMC cells treated with the indicated concentrations of K118 for 6 h (top) or AS1949490 for 24 h (bottom). (A–C) Each set of data consists of at least three independent experiments. Statistical analysis was performed by comparing the effect of the SHIP inhibitor to the respective vehicle using Kruskal-Wallis test and corrected with Dunn’s multiple comparison test. *p , 0.05, **p , 0.01, ***p , 0.001. (D) CLL cells (n = 6) from IgH.TEm tumors (.85% IgMbposCD5posCD43posCD19pos) were treated with the indicated concentrations of K118 or AS1949490 to evaluate cellular survival. Bar graph represents mean 6 SEM proportion of viable (annexinVneg7-AADneg) cells normalized to untreated control. Statistical analyses were performed using a Wilcoxon signed rank test. *p , 0.05, **p , 0.01. 368 SHIP CONTRIBUTES TO CLL SURVIVAL Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 5. Ship inhibition decreases constitutive Akt/p-S6 signaling in EMC cells and IgH.TEm tumors. (A–C) Phospho-flow analysis of the indicated phosphoproteins on (A) unstimulated B220posCD3neg WT splenic B cells and EMC6 cells; (B) EMC6 cells either untreated or treated with the pan-Ship1/2 inhibitor K118 (5 mM) for 3 h or (C) the Ship2-specific inhibitor AS1949490 (15 mM) for 3 h (in case of p-Btk Y223 and p-Erk) or 24 h (for p-S6). Each histogram shows a representative flow cytometric analysis from at least three independent experiments. (D) Phospho-flow analysis of the S6 on CLL cells from unstimulated IgH.TEm tumors (n = 6) (as in Fig. 4D), either untreated or treated with K118 (5 mM) or AS1949490 (15 mM) for the indicated time points. Each histogram (left) shows a representative analysis. Bar graphs (right) depict quantification (mean 6 SEM) of mean fluorescence intensity (MFI) of p-S6. Statistical analyses were performed using the Wilcoxon signed rank test. *p , 0.05. (E–G) Calcium flux assay in WT splenic B cells and EMC6 cells either untreated (E) or treated with the pan-Ship1/2 inhibitor K118 (5 mM) for 3 h (F) or the Ship2-specific inhibitor AS1949490 (15 mM) for 24 h (G). Each histogram shows a representative analysis from three independent experiments.

SHIP1/2 inhibition decreases survival of human CLL cells PTPN22 were significantly higher in U-CLL than in M-CLL (data To translate our findings in IgH.TEm mice to human CLL, we first not shown), these phosphatases may have a unique role in the analyzed by qRT-PCR the expression of several phosphatases in more aggressive U-CLL subgroup. We investigated SHIP1/2 inhibition in vitro and analyzed sur- leukemic cells obtained from a panel of 10 U-CLL and 10 M-CLL vival of human CLL samples by incubation with K118 (2.5 or patients (details of IGHV sequence in these CLL are provided in 5 mM) for 6 h and subsequent staining with annexin V and Table I). Expression levels in naive mature B cells purified from 7-AAD. This treatment resulted in a significant decrease in the n PBMCs of healthy volunteers ( = 3) were included as a reference. proportions of alive cells and a ∼3-fold increase in the proportions Expression of six out of eight selected phosphatase genes was of apoptotic cells (Fig. 7B). These findings support an important significantly higher in CLL than in naive resting B cells from role of SHIP1/2 in cellular survival of human CLL cells. Similarly, , healthy individuals (p 0.05), suggesting a role for these genes in incubation of CLL cells from an independent cohort of 20 patients ∼ CLL (Fig. 7A). Hereby, SHIP1 and SHIP2 showed 3-fold and with either K118 (5 mM) or AS1949490 (15 mM) for 6 h ∼8-fold higher expression in CLL than in control naive B cells, resulted in a significant reduction of the proportions of alive respectively. Although INPP5F was downregulated in CLL from cells (Supplemental Fig. 2G). IgH.TEm mice, it was expressed ∼10-fold higher in our panel Finally, we determined the effect of SHIP1/2 in the regulation of human CLL compared with naive B cells, consistent with of the BCR downstream signaling cascade in human CLL cells. published findings (30). Because expression levels of INPP5F and Unlike our findings in EMC cells (Fig. 5), SHIP inhibition in The Journal of Immunology 369 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 6. Ship inhibition decreases constitutive Mcl-1 signaling in EMC cells and IgH.TEm tumors. (A) Western blot analysis for Mcl-1 in the indicated cells (top), either untreated or treated with the pan-Ship1/2 inhibitor K118 (5 mM) for 4 h or the Ship2-specific inhibitor AS1949490 (15 mM) for 24 h. To exclude caspase-mediated Mcl-1 breakdown, 5 mM Quinoline-Val-Asp-Difluorophenoxymethyl Ketone (QVD) was added to untreated or SHIP inhibitor-treated EMC6 cells. To show constitutive Mcl-1 expression in EMC cells, MACS-purified untouched WT splenic B cells were included as either uncultured (0 h), cultured in the absence of stimulation (24 h unstim) or stimulated in the presence (24 h anti-IgM and LPS) for 24 h. Bar graphs depict quantification of signals normalized to actin in each condition from two independent experiments. (B–D) Flow cytometry analysis for intracellular Mcl-1 protein expression of WT splenic B cells either unstimulated or stimulated with anti-IgM and LPS for 24 h (B), EMC6 cells either untreated or treated with K118 (5 mM) or AS1949490 (15 mM) for the indicated time points (C), and CLL cells from unstimulated IgH.TEm tumors (n = 6) (as in Fig. 4D), either untreated or treated with K118 (5 mM) or AS1949490 (15 mM) for the indicated time points (D). Each histogram (left) shows a representative analysis and bar graph (right) depicts a quantification (mean 6 SEM) of mean fluorescence intensity (MFI) of intracellular Mcl-1. Statistical analyses were performed using a Wilcoxon signed rank test. *p , 0.05. human CLL cells by K118 incubation for 1 h resulted in lim- and human CLL. First, we show that several phosphatases, in- ited downregulation of p-BTK(Y223) and p-ERK and moderate cluding SHIP1 and SHIP2, are overexpressed in both mouse and downregulation of p-S6 expression (Fig. 7C). Incubation with human CLL B cells. Second, in our CLL mouse model we ob- K118 resulted in an upregulation of basal and anti-IgM–induced served that conditional deletion of Ship2 in the B cell lineage calcium levels, essentially in two CLL cases (Fig. 7D). significantly decreased CLL formation. Third, reducing SHIP1/2 In summary, various phosphatases are also highly expressed in activity by the small molecule inhibitor K118 decreased the human CLL and SHIP1/2 inhibition resulted in reduced survival of in vitro survival of human and mouse CLL B cells. Our data CLL cells. Thus, these data indicate parallel roles of SHIP1/2 in further indicate that SHIP1/2 promotes CLL survival by exert- human CLL and in our IgH.TEm CLL mouse model (Fig. 8). ing dual effects on the BCR signaling cascade (Fig. 8). On one hand, SHIP1/2 increases phosphatidylinositol (3,4)-bisphosphate Discussion [PI(3,4)P2] levels and thereby enhances the AKT/mTORC1/S6 In this study, we provide compelling evidence that specific pathway, resulting in increased Mcl-1 protein expression, which phosphatases contribute to malignant B cell survival in both mouse mediates survival of CLL B cells. On the other hand, SHIP1 370

Table I. B cell receptor characteristics of CLL patients

Sample % BCR Germline VH CDR3 Mutation CLL ID Inclusion Identity IGHV Gene IGHD Gene IGHJ Gene Length VH CDR3 Sequence AA Gender Status Age KL2015-240 a,b 89.73 IGHV3-30*03 IGHD6-13*01 IGHJ6*02 19 AKVGRPAAFEEYYYYGMDV Female M-CLL 54 KL2015-134 a,b 90.6 IGHV4-61*02 IGHD5-18*01 IGHJ4*02 12 ARDPDTYGYVDC Male M-CLL 58 KL2017-070 a,b 91.03 IGHV3-30*03 IGHD3-9*01 IGHJ4*02 18 AKPGSVFRYFDWISGLWYW Male M-CLL 71 KL2014-413 a,b 91.9 IGHV3-30*03 IGHD5-12*01 IGHJ3*02 14 ANELVTSSYDGIDIW Male M-CLL 66 KL2015-241 a,b 92.02 IGHV3-48*03 IGHD2-15*01 IGHJ1*01 9 ARDGGSYPL Male M-CLL 74 KL2010-250 a,b 94.05 IGHV3-72*01 IGHD2-8*01 IGHJ6*02 20 GRIYCTLSRCSIDQYYGMDV Male M-CLL 52 KL2010-510 a 94.6 IGHV3-74*01 IGHD2-15*01 IGHJ5*02 18 AREVCIGDNCYSRGWFDP Male M-CLL 82 KL2015-555 a 95 IGHV3-53*01 IGHD5-18*01 IGHJ4*02 16 ARDRGGGYSYGGGFDY Male M-CLL 61 KL2013-007 a,b 96.02 IGHV4-30*01 IGHD2-2*01 IGHJ6*03 16 ARDAGVVPVHYYYMDV Male M-CLL 66 KL2016-263 b 96.85 IGHV5-10-1*01 IGHD2-15*01 IGHJ6*02 28 ATGEGGLGWNPRYCSGGSCYEVGYGMDV Male M-CLL 67 KL2010-168 a 97.31 IGHV3-11*03 IGHD2-15*01 IGHJ4*02 13 ARGGEVMVSPLDR Male M-CLL 66 KL2011-589 a,b 100.00 IGHV1-69*06 IGHD3-3*01 IGHJ6*03 21 ASGSIFGVVIGSYYYYYYMDVW Male U-CLL 57 KL2014-260 a,b 100.00 IGHV3-20*01 IGHD3-3*01 IGHJ4*02 21 ARGTGITIFGVVHTTEYYFDYW Male U-CLL 85 KL2014-372 a,b 100.00 IGHV1-69*01 IGHD3-16*02 IGHJ5*02 22 ARDPPFDYIWGSYRYRANWFDPW Female U-CLL 61 KL2015-035 a 100.00 IGHV1-69*01 IGHD2-2*01 IGHJ6*02 24 ARDSPHKQDIVVVPAAMVFYSMDV Male U-CLL 68 KL2013-006 a,b 100.00 IGHV4-39*01 IGHD3-3*01 IGHJ6*02 26 ARHASPRDFWSGYPELIYYYYYGMDVW Male U-CLL 48 KL2014-420 a,b 100.00 IGHV1-69*01 IGHD2-2*01 IGHJ6*02 21 ASLTIVVVPAAMSYYYYGMDVW Male U-CLL 61 KL2011-399 a,b 100.00 IGHV4-4*05 IGHD3-16*02 IGHJ6*02 26 ARGRRDDYIWGSYRYTDLGYYYGMDV Female U-CLL 73 KL2011-447 a,b 100.00 IGHV1-69*01 IGHD3-3*01 IGHJ4*02 22 ARAAVPYYDFWSGYSLDSGFDY Male U-CLL 49 KL2016-025 a 100.00 IGHV4*31*01 IGHD3-22*01 IGHJ6*02 22 ARDSSPRLYYDSSGYYGMGLDV Male U-CLL 85 SURVIVAL CLL TO CONTRIBUTES SHIP KL2010-451 a 100.00 IGHV4-59*01 IGHD3-22*01 IGHJ6*03 23 ARGNYYDSSGYYYVGYYYYYMDV Male U-CLL 57 KL2014-264 b 100.00 IGHV1-24*01 IGHD4-23*01 IGHJ4*02 16 ATLGAARQLGWYYFDYW Male U-CLL 68 aSamples included for expression analysis by qRT-PCR. bSamples included for phospho-ﬂow analysis.

ID, identiﬁer.

Downloaded from from Downloaded http://www.jimmunol.org/ by guest on September 29, 2021 29, September on guest by The Journal of Immunology 371 Downloaded from http://www.jimmunol.org/ by guest on September 29, 2021

FIGURE 7. SHIP mediates AKT/mTORC1/S6-associated survival of human CLL cells. (A) mRNA expression of the indicated phosphatase genes as measured by qRT-PCR in tumor cells isolated from PBMCs from CLL patients (n = 20, 10 M-CLL and 10 U-CLL). mRNA expression was normalized to naive mature B cells (FACS-purified CD19posCD27negIgDpos) obtained from healthy donors PBMCs (dashed line, n = 3). (B) CLL cells (n = 20) were treated with the indicated concentrations of the pan-Ship1/2 inhibitor K118 for 6 h to evaluate cellular survival. Bar graph represent mean values and SEM for the proportions of viable (annexinVneg7-AADneg, left) or dead (7-AADpos, right) cells normalized to untreated control. (C) Histograms (top) depicting phospho-flow analysis of indicated protein in CLL cells (n = 14) either left untreated (gray) or treated (black) with the pan-Ship1/2 inhibitor K118 (5 mM) for 1 h. Quantification (bottom) of the MFI values of p-BTK, p-ERK, and p-S6, whereby each line represents a paired analysis for individual patients. (D) Histograms (left) depicting basal and anti–IgM-stimulated calcium influx analysis in CLL cells (n = 14) either left untreated (Figure legend continues) 372 SHIP CONTRIBUTES TO CLL SURVIVAL

it is conceivable that Ship2 may compensate for the loss of Ship1 function in CLL formation in IgH.TEm.Ship12/2 mice. By contrast, Ship1 did not compensate for Ship2 deficiency (Fig. 3B), revealing differential roles for Ship1 and Ship2 in CLL pathogenesis. Taken together, these findings indicate the importance of inhibiting both Ship1 and Ship2 enzymes for optimal decline of CLL cell survival. Nevertheless, it remains puzzling why the effects of AS1949490 were only reached at a higher inhibitor concentration and at a later time point, compared with K118. This might be due to a limited cell penetrance or potency of AS1949490. However, it cannot be excluded that Ship2 activity is important for CLL induction, but once established the survival of tumor cells is less sensitive to inhibition of Ship2 activity. Possibly in line with this, we found that primary Ship2-deficient CLL cells showed similar S6 phosphorylation as control IgH.TEm primary tumors (S. Pal Singh and R.W. Hendriks, unpublished observations). Furthermore, it remains unclear why, in contrast to control IgH.TEm mice, CLL formation in Ship2-deficient IgH.TEm mice FIGURE 8. SHIP promotes CLL survival by exerting dual effects on the Downloaded from BCR signaling cascade. Schematic model for SHIP1/2-mediated survival was not associated with splenomegaly, despite their high num- of CLL cells. SHIP1/2 can increase PI(3,4)P2 levels and thereby enhance bers of circulating CLL cells. Further experiments should show the AKT/mTORC1/S6 pathway, associated with increased Mcl-1 protein whether this is due to slow CLL tumor growth in Ship2-deficient expression, which promotes survival of CLL B cells. At the same time, IgH.TEm mice or whether Ship2 activity is involved in migration, SHIP1/2 can maintain optimal calcium levels in the presence of consti- homing or adhesion of CLL cells. tutive kinase signaling, thereby engaging an anergic response to chronic Measurement of the phosphorylation status of BCR downstream

BCR stimulation in CLL B cells. AKT, serine/threonine kinase or protein proteins indicated that apoptosis due to SHIP1/2 inhibition likely http://www.jimmunol.org/ kinase B; PI3K, phosphatidylinositol-3-kinase; PTEN, phosphatase and relies on downregulation of AKT/mTORC1/S6 signaling in the tensin homolog; S6, ribosomal protein S6. EMC cell lines, IgH.TEm primary tumors, and human CLL. In parallel, inhibition of BTK and PI3K following treatment with maintains optimal calcium levels in the presence of constitutive ibrutinib and idelalisib, respectively, resulted in downregulation of active kinase signaling, thereby engaging an anergic response to AKT/S6 in mouse and human CLL cells (7, 11, 40). This dual BCR stimulation in CLL B cells. Because in many tumors PI3K is regulation of the AKT/mTORC1/S6 signaling is in agreement constitutively active, SHIP1/2 have long been thought to act as with the proposed “two PIP hypothesis,” whereby a malignant tumor suppressors by hydrolyzing phosphatidylinositol (3,4,5)- state results from a balance between the levels of PI(3,4,5)P3 trisphosphate [PI(3,4,5)P3]. This phospholipid functions to acti- and PI(3,4)P2 (41) (Fig. 8). Our results indicate that SHIP1/ by guest on September 29, 2021 vate downstream signaling pathways, particularly AKT/mTORC1/ 2-mediated generation of PI(3,4)P2 contributes to CLL cell sur- S6, and thereby driving cell growth and survival. However, sub- vival by promoting AKT/mTORC1/S6 signaling and Mcl-1 ex- sequent studies of SHIP inhibitory molecules showed that both pression at the expense of PI(3,4,5)P3-dependent signals. Therefore, SHIP1 and SHIP2 can have oncogenic roles in hematologic can- it would be interesting to evaluate the efficacy of combination cers and epithelial cancers through their ability to produce PI(3,4) therapy with a SHIP1/2 inhibitor and either ibrutinib or idelalisib P2 and activate AKT (34, 38, 39). Our data provide additional to achieve complete inhibition of AKT/mTORC1/S6-mediated evidence for a role of SHIP1/2 enzymes as proto-oncogenes that survival pathways in CLL. Given that the mTOR pathway is an activate AKT through the PI(3,4,5)P3 hydrolysis product PI(3,4) important regulator of Mcl-1, pyrimidine biosynthesis, and mito- P2 (Fig. 8) and indicate for the first time, to our knowledge, that chondrial respiration, it constitutes a potential treatment target in SHIP1/2 can be oncogenic in CLL. CLL (37, 42, 43). In this context, it has been shown that mTORC1 Our in vitro studies targeting survival of cultured EMC cells with inhibition in vitro induces cell cycle arrest in CLL cells with- small molecule inhibitors showed that dual inhibition of Ship1/2 out inducing apoptosis (43). Evidence was provided that dual with K118 led to cellular apoptosis at a lower dose, compared mTORC1/2 inhibition exhibited an improved response in a CLL with the Ship2-specific inhibitor AS19494940. These findings mouse model compared with mTORC1 inhibition with rapalogs suggest that in CLL, Ship1 and Ship2 have redundant roles: and thus shows promise as future CLL therapy, particularly in both are able to maintain PI(3,4)P2-mediated activation of the combination with ibrutinib (44). AKT/mTORC1/S6 pathway in the absence of one another, Optimal BCR signaling is essential for B cell selection, survival, albeit suboptimally. In agreement with this, pan-SHIP inhibition of and proliferation. Too weak (nonfunctional BCR) or too high multiple myeloma cell lines induced cellular death more efficiently (autoreactive BCR) signaling strength results in cell death (negative than specific inhibition of SHIP1 by the small molecule 3AC (34). selection) of B cells (2). Such immune checkpoints were suggested Interestingly, it was also shown that 3AC-resistant tumor cells to be fully functional in some B cell malignancies. In autoreactive expressed increased levels of SHIP2, indicating that absence of B cell acute lymphoblastic leukemia clones, hyperactivation of SHIP1 might lead to increased SHIP2 expression, supporting a BCR signaling has been shown to trigger clonal deletion (45). In redundant function of SHIP1 and SHIP2. Given this redundancy, addition, knockdown of the PTPN22 phosphatase reduced survival

(gray) or treated (black) with the pan-Ship1/2 inhibitor K118 (5 mM) for 1 h. Basal and anti–IgM-stimulated calcium signaling in representative healthy control (gray shaded) is also depicted. Quantification (right) of the MFI values basal calcium signals, whereby each line represents a paired analysis for individual patients. Statistical comparisons were performed using a Wilcoxon signed rank test. *p , 0.05, **p , 0.01, ***p , 0.001. The Journal of Immunology 373 of anti-IgM–stimulated CLL cells by abrogating AKT-mediated 4. Byrd, J. C., B. Harrington, S. O’Brien, J. A. Jones, A. Schuh, S. Devereux, J. Chaves, W. G. Wierda, F. T. Awan, J. R. Brown, et al. 2016. Acalabrutinib signaling (28). These observations indicate that malignant B cells (ACP-196) in relapsed chronic lymphocytic leukemia. N. Engl. J. Med. 374: are susceptible to negative selection. In concordance, we observed 323–332. increased basal calcium levels in SHIP1/2-inhibited EMC and 5. Byrd, J. C., R. R. Furman, S. E. Coutre, I. W. Flinn, J. A. Burger, K. A. Blum, B. Grant, J. P. Sharman, M. Coleman, W. G. Wierda, et al. 2013. Targeting BTK human CLL cells, which are likely to contribute to enhanced with ibrutinib in relapsed chronic lymphocytic leukemia. N. Engl. J. Med. 369: cellular apoptosis. 32–42. Two independent mechanisms via which SHIP1/2 can regulate 6. Burger, J. A., A. Tedeschi, P. M. Barr, T. Robak, C. Owen, P. Ghia, O. Bairey, P. Hillmen, N. L. Bartlett, J. Li, et al; RESONATE-2 Investigators. 2015. intracellular calcium flux have been reported. First, dephosphory- Ibrutinib as initial therapy for patients with chronic lymphocytic leukemia. lation of PI(3,4,5)P3 by SHIP1/2 inhibits the recruitment of BTK N. Engl. J. Med. 373: 2425–2437. 7. Brown, J. R., J. C. Byrd, S. E. Coutre, D. M. Benson, I. W. Flinn, N. D. Wagner- to the plasma membrane and consequently downregulates calcium Johnston, S. E. Spurgeon, B. S. Kahl, C. Bello, H. K. Webb, et al. 2014. signaling (46). Second, SHIP was reported to downregulate the Idelalisib, an inhibitor of phosphatidylinositol 3-kinase p110d, for relapsed/ levels of inositol(1,3,4,5) tetraphosphate (IP4) at the endoplasmic refractory chronic lymphocytic leukemia. Blood 123: 3390–3397. 8. Furman, R. R., J. P. Sharman, S. E. Coutre, B. D. Cheson, J. M. Pagel, reticulum, which in turn diminishes intracellular calcium levels P. Hillmen, J. C. Barrientos, A. D. Zelenetz, T. J. Kipps, I. Flinn, et al. 2014. (47, 48). Because we did not observe increased Btk or Erk phos- Idelalisib and rituximab in relapsed chronic lymphocytic leukemia. N. Engl. phorylation in the EMC cell lines, increased IP4 levels due to inhi- J. Med. 370: 997–1007. 9. Manning, B. D., and A. Toker. 2017. AKT/PKB signaling: navigating the bition of Ship1 might be the mechanism responsible for the observed network. Cell 169: 381–405. increase in calcium flux, as reported previously in chicken DT40 10. Mockridge, C. I., K. N. Potter, I. Wheatley, L. A. Neville, G. Packham, and F. K. Stevenson. 2007. Reversible anergy of sIgM-mediated signaling in the two B cells (49). Although Ship2 has been reported to dephosphorylate subsets of CLL defined by VH-gene mutational status. Blood 109: 4424–4431. IP4 (50), we saw limited changes in basal calcium flux upon Ship2 11. Singh, S. P., S. Y. Pillai, M. J. W. de Bruijn, R. Stadhouders, O. B. J. Corneth, Downloaded from inhibition in the EMC cell lines, indicating that calcium signaling in H. J. van den Ham, A. Muggen, W. van IJcken, E. Slinger, A. Kuil, et al. 2017. Cell lines generated from a chronic lymphocytic leukemia mouse model exhibit EMC cells is exclusively regulated by Ship1. Additional experiments constitutive Btk and Akt signaling. Oncotarget 8: 71981–71995. are required to further elucidate the relationship between SHIP1/2, 12. Du¨hren-von Minden, M., R. Ubelhart,€ D. Schneider, T. Wossning, M. P. Bach, IP4levels,andcalciumfluxinhumanandmouseCLL. M. Buchner, D. Hofmann, E. Surova, M. Follo, F. Ko¨hler, et al. 2012. Chronic lymphocytic leukaemia is driven by antigen-independent cell-autonomous sig- Optimal BCR signaling requires a proper balance between nalling. Nature 489: 309–312. kinase and phosphatase activity of several BCR downstream 13.Stamatopoulos,K.,A.Agathangelidis,R.Rosenquist,andP.Ghia.2017. http://www.jimmunol.org/ Antigen receptor stereotypy in chronic lymphocytic leukemia. Leukemia 31: proteins (2). Current therapeutic regimens in CLL focus on 282–291. blocking the positive kinase signal, which successfully decreases 14. ter Brugge, P. J., V. B. Ta, M. J. de Bruijn, G. Keijzers, A. Maas, D. C. van Gent, cell survival and proliferation (51). However, phosphatase inhi- and R. W. Hendriks. 2009. A mouse model for chronic lymphocytic leukemia based on expression of the SV40 large T antigen. Blood 114: 119–127. bition may be a valid approach in B cell malignancies, as in 15. Yu, Y., and J. C. Alwine. 2002. Human cytomegalovirus major immediate-early vivo efficacy of SHIP inhibition was found in a human multiple proteins and simian virus 40 large T antigen can inhibit apoptosis through myeloma SCID xenograft model (34). Increased understanding activation of the phosphatidylinositide 39-OH kinase pathway and the cellular kinase Akt. J. Virol. 76: 3731–3738. of the role of phosphatases in regulating the increased kinase 16. Laine, J., G. Ku¨nstle, T. Obata, M. Sha, and M. Noguchi. 2000. The proto- signaling will help open new therapeutic avenues for these ma- oncogene TCL1 is an Akt kinase coactivator. Mol. Cell 6: 395–407.

17. Wang, J. W., J. M. Howson, T. Ghansah, C. Desponts, J. M. Ninos, S. L. May, by guest on September 29, 2021 lignancies. To the best of our knowledge, this is the first study K. H. Nguyen, N. Toyama-Sorimachi, and W. G. Kerr. 2002. Influence of focusing on inhibition of SHIP2 phosphatase in CLL as a thera- SHIP on the NK repertoire and allogeneic bone marrow transplantation. Science peutic strategy. Our data indicate that SHIP1/2 activity promotes 295: 2094–2097. 18. Dubois, E., M. Jacoby, M. Blockmans, E. Pernot, S. N. Schiffmann, AKT/mTORC1/S6 signaling and sustains calcium levels at an L. C. Foukas, J. C. Henquin, B. Vanhaesebroeck, C. Erneux, and S. Schurmans. optimal threshold, thereby contributing to CLL survival. As 2012. Developmental defects and rescue from glucose intolerance of a such regulation is susceptible to the pharmacologic inhibition of catalytically-inactive novel Ship2 mutant mouse. Cell. Signal. 24: 1971–1980. 19. Hobeika, E., S. Thiemann, B. Storch, H. Jumaa, P. J. Nielsen, R. Pelanda, and SHIP1/2, this may provide a novel therapeutic strategy to target M. Reth. 2006. Testing gene function early in the B cell lineage in mb1-cre mice. CLL cells. Inhibition of SHIP1/2 may be combined with ibrutinib Proc. Natl. Acad. Sci. USA 103: 13789–13794. 20. Gumbleton, M., R. Sudan, S. Fernandes, R. W. Engelman, C. M. Russo, treatment to target both BTK and AKT-mediated survival signals J. D. Chisholm, and W. G. Kerr. 2017. Dual enhancement of T and NK cell in CLL. Further experiments should explore whether SHIP1/2 function by pulsatile inhibition of SHIP1 improves antitumor immunity and inhibition may have clinical potential in CLL or other tumors survival. Sci. Signal. 10. 21. Suwa, A., T. Yamamoto, A. Sawada, K. Minoura, N. Hosogai, A. Tahara, (52), because it is currently unknown how well SHIP1/2 inhibition T. Kurama, T. Shimokawa, and I. Aramori. 2009. Discovery and functional is tolerated in human. characterization of a novel small molecule inhibitor of the intracellular phosphatase, SHIP2. Br. J. Pharmacol. 158: 879–887. 22. Ho, T. T., M. R. Warr, E. R. Adelman, O. M. Lansinger, J. Flach, Acknowledgments E. V. Verovskaya, M. E. Figueroa, and E. Passegue´. 2017. Autophagy maintains the metabolism and function of young and old stem cells. Nature 543: 205–210. We thank Otto M. Dungan and John D. Chisholm (SUNY Upstate Medical 23. Kil, L. P., M. J. de Bruijn, M. van Nimwegen, O. B. Corneth, J. P. van Hamburg, University, Syracuse, NY) for generating pan-SHIP1/2 inhibitor (K118). G. M. Dingjan, F. Thaiss, G. F. Rimmelzwaan, D. Elewaut, D. Delsing, et al. We thank Melanie Lukkes, Mariana P. de Almeida (Erasmus MC), Erik 2012. Btk levels set the threshold for B-cell activation and negative selection of Slinger, and Arnon P. Kater (AMC Amsterdam, the Netherlands) for as- autoreactive B cells in mice. Blood 119: 3744–3756. 24. Robinson, M. D., D. J. McCarthy, and G. K. Smyth. 2010. edgeR: a bioconductor sistance and advice. package for differential expression analysis of digital gene expression data. Bioinformatics 26: 139–140. 25. Pal Singh, S., M. J. W. de Bruijn, M. P. de Almeida, R. W. J. Meijers, Disclosures L. Nitschke, A. W. Langerak, S. Y. Pillai, R. Stadhouders, and R. W. Hendriks. The authors have no financial conflicts of interest. 2018. Identification of distinct unmutated chronic lymphocytic leukemia subsets in mice based on their T cell dependency. Front. Immunol. 9: 1996. 26. Heng, T. S., and M. W. Painter; Immunological Genome Project Consortium. 2008. The Immunological Genome Project: networks of gene expression in References immune cells. Nat. Immunol. 9: 1091–1094. 1. Pal Singh, S., F. Dammeijer, and R. W. Hendriks. 2018. Role of Bruton’s ty- 27. Hallaert, D. Y., R. Spijker, M. Jak, I. A. Derks, N. L. Alves, F. M. Wensveen, rosine kinase in B cells and malignancies. [Published erratum appears in 2019 J. P. de Boer, D. de Jong, S. R. Green, M. H. van Oers, and E. Eldering. 2007. Mol. Cancer 18: 79.] Mol. Cancer 17: 57. Crosstalk among Bcl-2 family members in B-CLL: seliciclib acts via the Mcl-1/ 2. Mu¨schen, M. 2018. Autoimmunity checkpoints as therapeutic targets in B cell Noxa axis and gradual exhaustion of Bcl-2 protection. Cell Death Differ. 14: malignancies. Nat. Rev. Cancer 18: 103–116. 1958–1967. 3. Hendriks, R. W., S. Yuvaraj, and L. P. Kil. 2014. Targeting Bruton’s tyrosine 28. Negro, R., S. Gobessi, P. G. Longo, Y. He, Z. Y. Zhang, L. Laurenti, and kinase in B cell malignancies. Nat. Rev. Cancer 14: 219–232. D. G. Efremov. 2012. Overexpression of the autoimmunity-associated phosphatase 374 SHIP CONTRIBUTES TO CLL SURVIVAL

PTPN22 promotes survival of antigen-stimulated CLL cells by selectively acti- Lipid phosphatase SHIP2 functions as oncogene in colorectal cancer by regu- vating AKT. Blood 119: 6278–6287. lating PKB activation. Oncotarget 7: 73525–73540. 29. Schliffke, S., S. Buhs, S. Bolz, H. Gerull, L. von Wenserski, K. Riecken, 40. Cheng, S., J. Ma, A. Guo, P. Lu, J. P. Leonard, M. Coleman, M. Liu, J. J. Buggy, B. Fehse, P. Nollau, and M. Binder. 2018. The phosphotyrosine phospha- R. R. Furman, and Y. L. Wang. 2014. BTK inhibition targets in vivo CLL tase SHP2 promotes anergy in chronic lymphocytic leukemia. Blood 131: proliferation through its effects on B-cell receptor signaling activity. Leukemia 1755–1758. 28: 649–657. 30. Johnston, H. E., M. J. Carter, M. Larrayoz, J. Clarke, S. D. Garbis, D. Oscier, 41. Kerr, W. G. 2011. Inhibitor and activator: dual functions for SHIP in immunity J. C. Strefford, A. J. Steele, R. Walewska, and M. S. Cragg. 2018. Proteomics and cancer. Ann. N. Y. Acad. Sci. 1217: 1–17. profiling of CLL versus healthy B-cells identifies putative therapeutic targets 42. Sharma, A., A. J. Janocha, B. T. Hill, M. R. Smith, S. C. Erzurum, and A. Almasan. and a subtype-independent signature of spliceosome dysregulation. Mol. Cell. 2014. Targeting mTORC1-mediated metabolic addiction overcomes fludarabine Proteomics 17: 776–791. resistance in malignant B cells. Mol. Cancer Res. 12: 1205–1215. 31. Li, H., X. Wu, S. Hou, M. Malek, A. Kielkowska, E. Noh, K. J. Makondo, Q. Du, 43. Davids, M. S., and J. R. Brown. 2012. Targeting the B cell receptor pathway in J. A. Wilkins, J. B. Johnston, et al. 2016. Phosphatidylinositol-3,4-bisphosphate chronic lymphocytic leukemia. Leuk. Lymphoma 53: 2362–2370. and its binding protein lamellipodin regulate chemotaxis of malignant 44. Cosimo, E., A. Tarafdar, M. W. Moles, A. K. Holroyd, N. Malik, M. A. Catherwood, B lymphocytes. J. Immunol. 196: 586–595. J. Hay, K. M. Dunn, A. M. Macdonald, S. M. Guichard, et al. 2019. AKT/mTORC2 32. Gabelloni, M. L., M. Borge, J. Galletti, C. Canõnes, P. Fernańdez Calotti, inhibition activates FOXO1 function in CLL cells reducing B-cell receptor-mediated R. F. Bezares, J. S. Avalos, M. Giordano, and R. Gamberale. 2008. SHIP-1 survival. Clin. Cancer Res. 25: 1574–1587. protein level and phosphorylation status differs between CLL cells segregated 45. Chen, Z., S. Shojaee, M. Buchner, H. Geng, J. W. Lee, L. Klemm, B. Titz, by ZAP-70 expression. Br. J. Haematol. 140: 117–119. T. G. Graeber, E. Park, Y. X. Tan, et al. 2015. Signalling thresholds and negative 33. Akerlund, J., A. Getahun, and J. C. Cambier. 2015. B cell expression of the B-cell selection in acute lymphoblastic leukaemia. [Published erratum appears in SH2-containing inositol 5-phosphatase (SHIP-1) is required to establish anergy e2016 Natur 534: 138.] Nature 521: 357–361. to high affinity, proteinacious autoantigens. J. Autoimmun. 62: 45–54. 46.Bolland,S.,R.N.Pearse,T.Kurosaki,andJ.V.Ravetch.1998.SHIP 34. Fuhler, G. M., R. Brooks, B. Toms, S. Iyer, E. A. Gengo, M. Y. Park, modulates immune receptor responses by regulating membrane association of M. Gumbleton, D. R. Viernes, J. D. Chisholm, and W. G. Kerr. 2012. Thera- Btk. Immunity 8: 509–516. peutic potential of SH2 domain-containing inositol-59-phosphatase 1 (SHIP1) 47. Damen, J. E., L. Liu, P. Rosten, R. K. Humphries, A. B. Jefferson, P. W. Majerus, and SHIP2 inhibition in cancer. Mol. Med. 18: 65–75. and G. Krystal. 1996. The 145-kDa protein induced to associate with Shc Downloaded from 35. Leung, W. H., T. Tarasenko, Z. Biesova, H. Kole, E. R. Walsh, and S. Bolland. by multiple cytokines is an inositol tetraphosphate and phosphatidylinositol 2013. Aberrant antibody affinity selection in SHIP-deficient B cells. Eur. J. 3,4,5-triphosphate 5-phosphatase. Proc. Natl. Acad. Sci. USA 93: 1689–1693. Immunol. 43: 371–381. 48. Lu¨ckhoff, A., and D. E. Clapham. 1992. Inositol 1,3,4,5-tetrakisphosphate ac- 36. Liu, Q., A. J. Oliveira-Dos-Santos, S. Mariathasan, D. Bouchard, J. Jones, tivates an endothelial Ca(2+)-permeable channel. Nature 355: 356–358. R. Sarao, I. Kozieradzki, P. S. Ohashi, J. M. Penninger, and D. J. Dumont. 1998. 49. Okada, H., S. Bolland, A. Hashimoto, M. Kurosaki, Y. Kabuyama, M. Iino, The inositol polyphosphate 5-phosphatase ship is a crucial negative regulator of J. V. Ravetch, and T. Kurosaki. 1998. Role of the inositol phosphatase SHIP in B cell antigen receptor signaling. J. Exp. Med. 188: 1333–1342. B cell receptor-induced Ca2+ oscillatory response. J. Immunol. 161: 5129–5132.

37. van Attekum, M. H. A., S. Terpstra, E. Slinger, M. von Lindern, P. D. Moerland, 50. Pesesse, X., C. Moreau, A. L. Drayer, R. Woscholski, P. Parker, and C. Erneux. http://www.jimmunol.org/ A. Jongejan, A. P. Kater, and E. Eldering. 2017. Macrophages confer survival 1998. The SH2 domain containing inositol 5-phosphatase SHIP2 displays signals via CCR1-dependent translational MCL-1 induction in chronic lym- phosphatidylinositol 3,4,5-trisphosphate and inositol 1,3,4,5-tetrakisphosphate phocytic leukemia. Oncogene 36: 3651–3660. 5-phosphatase activity. FEBS Lett. 437: 301–303. 38. Brooks, R., G. M. Fuhler, S. Iyer, M. J. Smith, M. Y. Park, K. H. Paraiso, 51. Jerkeman, M., M. Hallek, M. Dreyling, C. Thieblemont, E. Kimby, and R. W. Engelman, and W. G. Kerr. 2010. SHIP1 inhibition increases immu- L. Staudt. 2017. Targeting of B-cell receptor signalling in B-cell malignancies. noregulatory capacity and triggers apoptosis of hematopoietic cancer cells. J. Intern. Med. 282: 415–428. J. Immunol. 184: 3582–3589. 52. Kerr, W. G., R. Brooks, M.-Y. Park, M. Gumbleton, S. Iyer, D. Viernes, 39. Hoekstra, E., A. M. Das, M. Willemsen, M. Swets, P. J. Kuppen, C. J. van der J. Chisholm, and G. M. Fuhler. 2011. Abstract C81: therapeutic potential of Woude, M. J. Bruno, J. P. Shah, T. L. Ten Hagen, J. D. Chisholm, et al. 2016. SHIP1 and SHIP2 inhibitors in cancer. Mol. Cancer Ther. 10. by guest on September 29, 2021 Supplemenatry Figure 1

A Ship2 Ship1 Ship2 Ship1 Ship2 Ship1

proB_CLP_BM preT_DN1_Th GN_Sp proB_FrA_BM preT_DN2a_Th GN_BM proB_FrBC_BM preT_DN3_Th MF_RP_Sp B_FrE_BM T_DN4_Th MF_Fem_PC B_T1_Sp T_ISP_Th MF_PC B_Sp T_DP_Th B_Fem_Sp T_4_Th pDC_Sp B_Fo_Sp T_4_Nve_Sp CD8+DC_Sp B_MZ_Sp T_4_Fem_Sp CD4+DC_Sp B_mem_Sp T_8_Th ILC3_SI B_GC_CC_Sp T_8_Nve_Sp ILC2_SI B_GC_CB_Sp Treg_4_25hi_Sp NK_BM B_PB_Sp NKT_Sp NK_Sp B_PC_Sp Tgd_Th B_PC_BM Tgd_Sp LTHSC_BM 4 3 2 1 0 0 1 2 3 4 4 3 2 1 0 0 1 2 3 4 4 3 2 1 0 0 1 2 3 4 Log 10 normalised DESeq2 Log 10 normalised DESeq2 Log 10 normalised DESeq2 Log 10 normalised DESeq2 Log 10 normalised DESeq2 Log 10 normalised DESeq2

B 105 250K 250K 105

200K 200K 4 10 104 150K 150K BM 103 100K 100K 103 0 50K 50K 0 -103 -103 0 50K 100K 150K 200K 250K 0 0 0 50K 100K 150K 200K 250K 0 50K 100K 150K 200K 250K -103 0103 104 105

5 250K 250K 10 105

200K 200K 104 104 150K 150K 103

Spleen 100K 100K 3 0 10

-103 50K 50K 0

-104 0 0 -103 0 50K 100K 150K 200K 250K 0 50K 100K 150K 200K 250K 0 50K 100K 150K 200K 250K

0103 104 105

B D FSC-A FSC-A FSC-W CD19

C D E fl/fl +/+ ) 6 5

6 Ship Mb1 * Ship1fl/flMb1cre/+ Shipfl/fl Ship1fl/fl Ship2fl/fl 4 Mb1+/+ Mb1cre/+ Mb1cre/+ Ship2fl/flMb1cre/+

105 105 105 4 * 3 4 10 104 104

2.67 0.85 103 2.18 103 5.71 103 2 0 0 0

IgD 2 -103 -103 -103 % cells in Spleen -103 0103 104 105 -103 0103 104 105 -103 0103 104 105 1 CD95 * * *

Number of cells in 10 Spleen (X 0 0 B-1 GC Memory IgM+ CD138+ CD19+CD5+ B cells B cells B cells B cells B cells

Suppl. Fig. S1: Ship1 and Ship2 expression in various hematopoieƟc cell populaƟon and diīerenƟal eīects of condiƟonal Ship1 and Ship2 deleƟon on B-cell development. (A) Normalized DE-seq expression values for Ship1 (black) and Ship2 (grey) in various hematopoieƟc cell populaƟons from C57BL/6J mice (obtained from GSE109125). Bar graphs depict log10 transformed expression values from DE-seq soŌware across all developmental stages in B-cells (top), T cells (middle) and innate immune cells, including granulocytes (GN), macrophages (MF), dendriƟc cells (DC), innate lymphocytes (ILC), NK cells and long-term haematopoieƟc stem cells (LTHSC) from various organs. BM, Bone marrow; Sp, Spleen; Th, Thymus; PC, Peritoneal cavity. (B) FACS plots depicƟng representaƟve gaƟng strategy for alive single B220posCD19pos B lymphocytes from total cell suspensions obtained from bone marrow (BM) (top) and spleen (below) of mice depicted in Fig. 2. (C) FACS plots depicƟng diīerences in germinal center cells (CD19posB220posIgDnegCD95pos) in spleen of the indicated mice. Data are shown as dot plots on gated CD19posB220high cells. (D) QuanƟĮcaƟon of B-1 cells (CD19posB220intCD5pos) germinal center (GC) B cells (CD19posB220posIgDnegCD95pos), memory B cells (CD19posB220posCD80posPDL2pos) and plasma cells (CD19pos/negIgMposCD138pos) in spleen of indicated mice. (E) QuanƟĮcaƟon of proporƟon of B-1 cells (CD19pos B220intCD5pos) of total B-cells in spleens of indicated mice. Data are from 5-6 mice per group. *p < 0.05, (Mann-Whitney U test).