Elevated Choline Kinase −α Mediated Choline Metabolism Supports the Prolonged Survival of TRAF3-Deficient B Lymphocytes

This information is current as Samantha Gokhale, Wenyun Lu, Sining Zhu, Yingying Liu, of October 3, 2021. Ronald P. Hart, Joshua D. Rabinowitz and Ping Xie J Immunol published online 11 December 2019 http://www.jimmunol.org/content/early/2019/12/10/jimmun ol.1900658 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2019/12/10/jimmunol.190065 Material 8.DCSupplemental http://www.jimmunol.org/ Why The JI? Submit online.

• 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 by guest on October 3, 2021

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 © 2019 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published December 11, 2019, doi:10.4049/jimmunol.1900658 The Journal of Immunology

Elevated Choline Kinase a–Mediated Choline Metabolism Supports the Prolonged Survival of TRAF3-Deficient B Lymphocytes

Samantha Gokhale,*,† Wenyun Lu,‡,x,{ Sining Zhu,*,† Yingying Liu,* Ronald P. Hart,*,{,‖ Joshua D. Rabinowitz,‡,x,{ and Ping Xie*,{

Specific deletion of the tumor suppressor TRAF3 from B lymphocytes in mice leads to the prolonged survival of mature B cells and expanded B cell compartments in secondary lymphoid organs. In the current study, we investigated the metabolic basis of TRAF3- mediated regulation of B cell survival by employing metabolomic, lipidomic, and transcriptomic analyses. We compared the polar metabolites, lipids, and metabolic enzymes of resting splenic B cells purified from young adult B cell–specific Traf32/2 and littermate control mice. We found that multiple metabolites, lipids, and enzymes regulated by TRAF3 in B cells are clustered in the choline metabolic pathway. Using stable isotope labeling, we demonstrated that phosphocholine and phosphatidylcholine Downloaded from biosynthesis was markedly elevated in Traf32/2 mouse B cells and decreased in TRAF3-reconstituted human multiple myeloma cells. Furthermore, pharmacological inhibition of choline kinase a, an enzyme that catalyzes phosphocholine synthesis and was strikingly increased in Traf32/2 B cells, substantially reversed the survival phenotype of Traf32/2 B cells both in vitro and in vivo. Taken together, our results indicate that enhanced phosphocholine and phosphatidylcholine synthesis supports the prolonged survival of Traf32/2 B lymphocytes. Our findings suggest that TRAF3-regulated choline metabolism has diagnostic and thera- peutic value for B cell malignancies with TRAF3 deletions or relevant mutations. The Journal of Immunology, 2020, 204: 000– http://www.jimmunol.org/ 000.

berrant B cell survival is an important contributing factor TRAF3, a member of the TRAF family, regulates the signal to the pathogenesis of B cell malignancies, which transduction pathways of a diverse array of immune receptors, A comprise .50% of blood cancers and .80% of lym- including the TNFR superfamily, TLRs, NOD-like receptors, RIG- phomas (1–3). We and others have recently identified TNFR- I-like receptors, and cytokine receptors (10, 18, 19). Specifically in associated factor (TRAF) 3, a cytoplasmic adaptor protein, as a B lymphocytes, TRAF3 directly binds to two receptors pivotal

critical regulator of cell survival and tumor suppressor in mature for B cell physiology, the BAFFR and CD40, which are required by guest on October 3, 2021 B lymphocytes (4–9). Deletions and inactivating mutations of the for B cell survival and activation, respectively (20, 21). Specific TRAF3 gene are some of the most frequently occurring genetic deletion of the Traf3 gene in B lymphocytes results in severe alterations in a variety of human B cell malignancies (9, 10), in- peripheral B cell hyperplasia in mice due to the prolonged survival cluding multiple myeloma (MM, 17%) (6, 11), gastric marginal of mature B cells independent of the principle B cell survival zone lymphoma (MZL, 21%), splenic MZL (10%) (12, 13), dif- factor BAFF (4, 5). This effect of TRAF3 deficiency in B cells fuse large B cell lymphoma (DLBCL, 14%) (14), B cell chronic eventually leads to spontaneous development of splenic MZL and lymphocytic leukemia (13%) (15), Hodgkin lymphoma (HL, 15%) B1 lymphomas at high incidence by 18 mo of age (8). These (16), and Waldenstro¨m’s macroglobulinemia (5%) (17). in vivo findings are consistent with the frequent deletions and

*Department of Cell Biology and Neuroscience, Rutgers University, Piscataway, NJ 08854; Address correspondence and reprint requests to Dr. Ping Xie, Department †Graduate Program in Cellular and Molecular Pharmacology, Rutgers University, Piscat- of Cell Biology and Neuroscience, Rutgers University, 604 Allison Road, away, NJ 08854; ‡Department of Chemistry, Princeton University, Princeton, NJ 08544; Nelson Laboratory Room B336, Piscataway, NJ 08854. E-mail address: xLewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ 08544; [email protected] {Rutgers Cancer Institute of New Jersey, New Brunswick, NJ 08901; and ‖W.M. Keck The online version of this article contains supplemental material. Center for Collaborative Neuroscience, Rutgers University, Piscataway, NJ 08854 Abbreviations used in this article: AEC, adenylate energy charge; B-Traf32/2, ORCIDs: 0000-0003-0554-099X (S.G.); 0000-0003-4836-8712 (R.P.H.); 0000-0002- 2 2 B cell–specific Traf3 / ;Chka,Chokinasea; Cho, choline; DAG, diacylglycerol; 1247-4727 (J.D.R.). Dgka, diacylglycerol kinase a; DLBCL, diffuse large B cell lymphoma; Etn, Received for publication June 12, 2019. Accepted for publication November 13, ethanolamine; Faah, fatty acid amide hydrolase; Gdpd3, glycerophosphodiester 2019. phosphodiesterase domain containing 3; G6P, glucose-6-phosphate; HL, Hodgkin lymphoma; LC, liquid chromatography; LC-MS, liquid chromatography–mass This work was supported by National Institutes of Health (NIH) Grant R01 CA158402 spectrometry; LMC, littermate control; Lpcat, lysophosphatidylcholine acyltrans- (to P.X.), Department of Defense Grant W81XWH-13-1-0242 (to P.X.), a Pilot Award ferase; MAG, monoacylglycerol; MM, multiple myeloma; MZL, marginal zone of the Cancer Institute of New Jersey through Grant Number P30CA072720 from the lymphoma; NF-kB2, noncanonical NF-kB; NIK, NF-kB–inducing kinase; PC, National Cancer Institute (NCI) (to P.X. and J.D.R.), and a Busch Biomedical Grant (to phosphatidylcholine; P-Cho, phosphocholine; P-DMEtn, phosphodimethylethanolamine; P.X.); supported in part by grants from NCI R50 CA211437 (to W.L.) and NIH R01 PPP, pentose phosphate pathway; TRAF, TNFR-associated factor. ES026057 (to R.P.H.). The metabolomic and lipidomic analyses were supported by the Rutgers Cancer Institute of New Jersey Metabolomics Shared Resource and FACS Ó analyses were supported by the Flow Cytometry Core Facility with funding from Copyright 2019 by The American Association of Immunologists, Inc. 0022-1767/19/$37.50 NCI–Cancer Center Support Grant P30CA072720. The microarray data presented in this article have been submitted to the National Institutes of Health Gene Expression Omnibus database (https://www.ncbi.nlm.nih. gov/geo/) under the accession number GSE113920.

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1900658 2 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS inactivating mutations of the TRAF3 gene identified in human (Manassas, VA) (30). All mouse and human B lymphoma cell lines B cell neoplasms, demonstrating the tumor suppressive role of were cultured as previously described (30). TRAF3 in mature B lymphocytes. Reagents and Abs The signal transduction pathway underlying TRAF3-mediated regulation of B cell survival has been elucidated in previous Tissue culture supplements, including stock solutions of sodium pyruvate, L-glutamine, nonessential amino acids, and HEPES (pH 7.55), were from studies. It was found that, in the absence of stimulation, TRAF3 2 Invitrogen (Carlsbad, CA). Trimethyl-D9-Cho ([ H]9-Cho) was purchased constitutively binds to NF-kB–inducing kinase (NIK) (the up- from Cambridge Isotope Laboratories (Tewksbury, MA). MN58B and stream kinase of the noncanonical NF-kB [NF-kB2] pathway) and RSM932A were obtained from Aobious (Gloucester, MA). Abs against TRAF2, whereas TRAF2 also constitutively associates with Chka were from Abcam (Cambridge, MA). Polyclonal rabbit Abs to cIAP1/2 (18, 22). In this complex, cIAP1/2 induces K48-linked TRAF3 (H122) were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-actin Ab was from Chemicon (Temecula, CA). HRP- polyubiquitination of NIK, thereby targeting NIK for proteasomal labeled secondary Abs were from Jackson ImmunoResearch Labora- degradation and thus inhibiting NF-kB2 activation (23–26). Upon tories (West Grove, PA). BAFF or CD154 stimulation, trimerized BAFFR or CD40 recruits TRAF3, TRAF2, and cIAP1/2 to the plasma membrane, releasing Splenic B cell purification, culture, and stimulation NIK from the TRAF3–TRAF2–cIAP1/2 complex and allowing Mouse splenic B cells were purified using anti-mouse CD43–coated NIK to accumulate in the cytoplasm (26–28). Accumulated NIK magnetic beads and a MACS Separator (Miltenyi Biotec), following the protein subsequently induces the activation of IKKa and NF-kB2, manufacturer’s protocols as previously described (4). The purity of the and activated NF-kB2, in turn, promotes the expression of anti- isolated B cell population was monitored by FACS analysis, and cell preparations of .98% purity were used for metabolomic, lipidomic, and apoptotic proteins of the Bcl-2 family (such as Bcl-2, Bcl-xL, and transcriptomic analyses as well as protein preparation. An aliquot of pu- Downloaded from Mcl-1) to induce B cell survival (18, 22, 29). Specific deletion of rified splenic B cells were cultured ex vivo in mouse B cell medium (4) for TRAF3, TRAF2, or cIAP1/2 in B cells all results in similar 24 h before metabolomic and lipidomic analyses. phenotype in mice, with BAFF-independent constitutive NF-kB2 Water-soluble polar metabolite and lipid profiling using liquid activation and prolonged survival of mature B lymphocytes (4, 5, chromatography–mass spectrometry 28). However, it remains unclear whether this pathway regulates Traf32/2 and LMC splenic B cells were quickly washed twice with PBS. cellular metabolism to control B cell survival. 2/2

Ten million Traf3 and LMC splenic B cells were centrifuged at 12,000 http://www.jimmunol.org/ In the current study, we aimed to investigate the metabolic rpm at 4˚C for 5 min to obtain a pellet. Separate pellets were used to basis of the tumor suppressor TRAF3-mediated regulation of B cell extract water-soluble polar metabolites and lipids, respectively. To extract survival. To address this, we first employed unbiased metabolome water-soluble metabolites, 184 ml of methanol/H2O (80:20) + 0.5% formic and lipidome screening approaches to compare the metabolism of acid at dry ice temperature was added to the pellet, vortexed for 5 s, and was allowed to sit on dry ice for 10 min. Then 16 ml of 15% NH4HCO3 resting splenic B cells purified from young adult B cell–specific 2 2/2 2/2 was added to neutralize the solution. The final solution was kept at 20˚C Traf3 (B-Traf3 ) and littermate control (LMC) mice. To for 15 min, and the resulting mixture was transferred into an Eppendorf understand how TRAF3 regulates B cell metabolism, we per- tube and centrifuged at 13,200 rpm at 4˚C for 15 min. The supernatant was formed transcriptome profiling to identify metabolic enzymes taken for liquid chromatography–mass spectrometry (LC-MS) analysis. downstream of TRAF3 signaling. Our results revealed that mul- Liquid chromatography (LC) separation was achieved using a Vanquish UHPLC System (Thermo Fisher Scientific, San Jose, CA) and an Xbridge by guest on October 3, 2021 tiple metabolites, lipids, and enzymes regulated by TRAF3 in BEH Amide Column (150 3 2 mm, 2.5-mm particle size; Waters, Milford, B cells are clustered in the choline (Cho) metabolic pathway. MA). Solvent A is water/acetonitrile (95:5) with 20 mM ammonium ac- Using stable isotope labeling, we demonstrated the increased etate and 20 mM ammonium hydroxide at a pH of 9.4, and solvent B is biosynthesis of phosphocholine (P-Cho) and phosphatidylcholine acetonitrile. The gradient is 0 min, 90% B; 2 min, 90% B; 3 min, 75% B; (PC) in Traf32/2 B cells. To assess the functional importance of 7 min, 75% B; 8 min, 70% B; 9 min, 70% B; 10 min, 50% B; 12 min, 50% B; 13 min, 25% B; 14 min, 25% B; 16 min, 0% B; 21 min, 0% B; 21.5 elevated Cho metabolism, we examined the effects of two Cho min, 90% B; and 25 min, 90% B (31). Total running time is 25 min at a kinase a (Chka) inhibitors, MN58B and RSM932A. We found that flow rate of 150 ml/min. For all experiments, 5 ml of extract was injected MN58B and RSM932A inhibited the survival of Traf32/2 Bcells with column temperature at 25˚C. The Q-Exactive Plus Mass Spectrometer both in vitro and in vivo. Taken together, we have identified P-Cho was operated in both negative and positive mode scanning m/z 70–1000 with a resolution of 140,000 at m/z 200. The mass spectrometry pa- and PC biosynthesis as a TRAF3-regulated metabolic pathway rameters were as follows: sheath gas flow rate, 28 (arbitrary units); critical for B cell survival. auxiliary gas flow rate, 10 (arbitrary units); sweep gas flow rate, 1 (arbitrary unit); spray voltage, 3.3 kV; capillary temperature, 320˚C; Materials and Methods S-lens radio frequency level, 65; automatic gain control target, 3E6; and Mice and cell lines maximum injection time, 500 ms. To extract lipids, 1 ml of 0.1 M HCl in methanol/H2O (50:50) was added Traf3flox/floxCD19+/Cre (B-Traf32/2) and Traf3flox/flox (LMC) mice were to the pellet and set in a 220˚C freezer for 30 min. Afterwards, 0.5 ml of generated as previously described (4). All experimental mice for this study chloroform was added to the mixture and vortexed, then set on ice for were produced by breeding of Traf3flox/flox mice with Traf3flox/floxCD19+/Cre 10 min. Samples were centrifuged at 13,200 rpm for 10 min, and the mice. All mice were kept in specific pathogen-free conditions in the Animal chloroform phase at the bottom was transferred to a glass vial as the first Facility at Rutgers University and were used in accordance with NIH extract, using a Hamilton syringe. Another 0.5 ml of chloroform was added guidelines and under an animal protocol (Protocol No. 08-048) approved by to the remaining material, and the extraction was repeated to get a second the Animal Care and Use Committee of Rutgers University. Equal numbers extract. The combined extract was dried under nitrogen flow and redis- of male and female mice were used in this study. solved in 200 ml of methanol/chloroform/2-propanol (1:1:1). Lipids were Traf32/2 mouse B lymphoma cell lines 27-9.5.3 (27-9) and 105-8.1B6 analyzed on a Q-Exactive Plus Mass Spectrometer coupled to a Vanquish (105-8) were generated and cultured as described previously (8, 30). UHPLC System (Thermo Fisher Scientific). Each sample was analyzed Human MM cell lines 8226 (contains biallelic TRAF3 deletions), twice using the same LC gradient but with a different ionization mode on the KMS11 (contains biallelic TRAF3 deletions), and LP1 (contains mass spectrometer to cover both positively and negatively charged species. inactivating TRAF3 frameshift mutations) were kindly provided by The LC separation was achieved on an Agilent Poroshell 120 EC-C18 Column Dr. L. Bergsagel (Mayo Clinic, Scottsdale, AZ) and cultured as pre- (150 3 2.1 mm, 2.7-mm particle size) at a flow rate of 150 ml/min. The viously described (30). TRAF3-sufficient mouse B lymphoma cell line gradient was 0 min, 25% B; 2 min, 25% B; 4 min, 65% B; 16 min, 100% A20.2J was generously provided by Dr. G. Bishop (University of Iowa, B; 20 min, 100% B; 21 min, 25% B; and 27 min, 25% B (32). Solvent A is Iowa City, IA) (30). TRAF3-sufficient mouse B lymphoma cell line 1 mM ammonium acetate + 0.2% acetic acid in methanol/H2O (10:90). Sol- m12.4.1 as well as human Burkitt’s B lymphoma cell lines Ramos vent B is 1 mM ammonium acetate + 0.2% acetic acid in methanol/2-propanol and Daudi were purchased from American Type Culture Collection (2:98). Other mass spectrometry parameters were the same as above. The Journal of Immunology 3

Data analyses were performed using MAVEN Software, which allows for expression levels of each gene were analyzed using the Sequence De- sample alignment, feature extraction, and peak picking (33). Extracted ion tection Software (Applied Biosystems) and the comparative cycle chromatograms for each metabolite were manually examined to obtain its threshold method, as previously described (36, 40). signal, using a custom-made metabolite library. Protein extraction and immunoblot analysis Stable isotope labeling analysis using LC-MS Total protein lysates were prepared as described (36). Proteins were sep- To analyze kinetic Cho flux into metabolic pathways, purified splenic arated by SDS-PAGE and immunoblotted with Abs to Chka, TRAF3, and B cells were cultured in RPMI medium containing 80 mg/ml of trimethyl- actin. Immunoblot analysis was performed as previously described (8, 30). 2 D9-Cho ([ H]9-Cho). At the indicated time points, cells were centrifuged at Images of immunoblots were acquired using a low-light imaging system 12,000 rpm at 4˚C for 5 min to obtain a pellet. To extract water-soluble (LAS-4000 mini; FUJIFILM Medical Systems USA, Stamford, CT) (8, 30). metabolites, 184 ml of methanol/H2O (80:20) + 0.5% formic acid at dry ice temperature was added to the pellet, vortexed for 5 s, and set on dry ice for MTT assay 10 min. Then 16 ml of 15% NH4HCO3 was added to neutralize the solu- 3 5 2 Mouse B lymphoma cell lines (1 10 cells per well) and human MM or B tion. The final solution was set at 20˚C for 15 min, and the resulting lymphoma cell lines (3 3 104 cells per well) were plated in 96-well plates in mixture was transferred into an Eppendorf tube and centrifuged at 13,200 the absence or presence of serial dilutions of Chka inhibitors. At the indicated rpm at 4˚C for 15 min. The supernatant was taken for LC-MS analysis of time point after treatment, cell viability and proliferation were measured using water-soluble metabolites and their labeled forms, as described above. In the MTT assay, as described (41). Possible influences caused by direct MTT- addition, P-Cho (P0378; Sigma-Aldrich) standards in 80% methanol at inhibitor interactions were excluded by studies in a cell-free system. concentrations of 0.5, 1.5, and 5 mM were run separately to estimate the concentrations of P-Cho in biological samples. Transduction of human MM cells with a lentiviral TRAF3 To extract lipids, 1 ml of chloroform was added to the remaining pellet expression vector and vortexed, then set on ice for 10 min. Samples were centrifuged at 13,200

rpm for 10 min, and the chloroform phase was transferred to a glass vial. The The full-length coding cDNA sequence of human TRAF3 was cloned from Downloaded from extract was dried under nitrogen flow and redissolved in 200 mlof the 293T cell line using reverse transcription PCR. Primers used for the methanol/chloroform/2-propanol (1:1:1) and taken for LC-MS analysis cloning of human TRAF3 are hTRAF3-F (59-CCT AAA ATG GAG TCG of lipids, as described above. In addition, 1,2-dipalmitoyl-rac-glycero- AGT AAA AAG-39) and hTRAF3-R (59-TTA TCA GGG ATC GGG CAG 3-P-Cho (P5911; Sigma-Aldrich) standards in the same solvent at con- A-39). The TRAF3 cDNA was subcloned into the lentiviral expression centrations of 0.5, 1.5, and 5 mM were run separately to estimate the vector pUB–eGFP–Thy1.1 (42) (generously provided by Dr. Z. Chen, concentrations of PC species in biological samples. University of Miami, Miami, FL) by replacing the eGFP coding sequence Data analyses were performed using MAVEN Software (33), as de- with the TRAF3 coding sequence. The resultant pUB–TRAF3–Thy1.1 scribed above, examining both the unlabeled and 2H-labeled forms of Cho- lentiviral vector was verified by DNA sequencing. Lentiviruses of pUB– http://www.jimmunol.org/ containing water-soluble metabolites and lipids (34). TRAF3–Thy1.1 and an empty vector pUB-Thy1.1 were packaged, and lentiviral titers were determined as previously described (30, 43). Human ATP measurement using an ATP Determination Kit MM 8226 cells were transduced with the packaged lentiviruses at a mul- To specifically measure ATP concentrations, fresh lysates of purified splenic tiplicity of infection of 1:5 (cell/virus) in the presence of 8 mg/ml poly- B cells were prepared using ice-cold 1.5% TCA in Mammalian Cell PE LB Lysis brene (30, 43). Transduction efficiency of cells was analyzed at day 3 Buffer (G-Biosciences, St. Louis, MO) and subsequently neutralized with 20 mM posttransduction using Thy1.1 immunofluorescence staining followed by Tris-HCl, pH 7.5 (35). ATP concentrations in lysates were determined using a flow cytometry. Transduced cells were subsequently analyzed for cell cycle luciferase-luciferin–based bioluminescence assay with an ATP Determination distribution, Chka expression, and Cho metabolism. For cell cycle anal- Kit (Molecular Probes, Eugene, OR), following the manufacturer’s protocol. ysis, transduced cells were fixed at day 4 or 7 posttransduction with ice-

Briefly, 10 ml of lysates was added into 90 ml of luciferase reagent and mea- cold 70% ethanol. Cell cycle distribution was subsequently determined by guest on October 3, 2021 sured immediately based on the maximal light intensity of the resulting biolu- by propidium iodide staining followed by flow cytometry, as previously minescence in a luminometer (GloMax 20/20; Promega, Fitchburg, WI) (35). A described (4, 44). calibration curve with serial dilutions of ATP standards was prepared in each In vivo administration of MN58B and RMS932A experiment. Data were normalized as nanomoles of ATP per 106 cells (35). Gender-matched, young adult LMC and B-Traf32/2 mice were injected i.p. Transcriptomic microarray analysis with a Chka inhibitor, MN58B or RSM932A, at 2 mg/kg per mouse or vehicle Total cellular RNA was extracted from splenic B cells purified from young control three times a week for 4 wk. Mouse spleen size was subsequently adult (8- to 12-wk-old) LMC and B-Traf32/2 mice using TRIzol reagent determined, and splenic B cell populations were analyzed by flow cytometry. (Invitrogen), according to the manufacturer’s protocol. RNA quality was Flow cytometry assessed on an RNA Nano Chip using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA) (36). The mRNA was amplified with a Single-cell suspensions were made from mouse spleens. Immunofluores- TotalPrep RNA Amplification Kit with a T7-oligo(dT) primer, following the cence staining and FACS analyses were performed as previously described manufacturer’s instructions (Ambion), and microarray analysis was carried (4, 8). Erythrocytes from spleens were depleted with ACK lysis buffer. out with the Illumina Sentrix Mouse Ref-8 24K Array at the Burnham In- Cells (1 3 106) were blocked with rat serum and Fc receptor–blocking Ab stitute (La Jolla, CA). Results were extracted with Illumina GenomeStudio (2.4G2) and incubated with various Abs conjugated to FITC, PE, PerCP, or v2011.1, background corrected, and variance stabilized in R/Bioconductor allophycocyanin for multiple color fluorescence surface staining. Analyses using the lumi package (37, 38) and modeled in the limma package (39). The of cell surface markers included Abs to CD45R (B220), CD3, CD21, and microarray analysis results have been deposited in the National Institutes of CD23 (BioLegend, San Diego, CA). Listmode data were acquired on a Health Gene Expression Omnibus database under the accession number FACSCalibur (Becton Dickinson, Mountain View, CA). The results were GSE113920 (https://www.ncbi.nlm.nih.gov/geo/). analyzed using the FlowJo Software (Tree Star, San Carlos, CA). TaqMan assays of the transcript expression of identified Statistics enzyme-encoding genes Statistical analyses were performed using Prism Software (GraphPad, La cDNA was prepared from RNA using High Capacity cDNA Reverse Jolla, CA). For direct comparison of the levels of polar metabolites, lipids, 2/2 Transcription Kit (Applied Biosystems, Carlsbad, CA). Quantitative or transcripts between LMC and Traf3 B cells, statistical significance was real-time PCR of specific genes was performed using corresponding determined with the unpaired t test for two-tailed data. The p values , 0.05 are TaqMan Gene Assay Kit (Applied Biosystems), as previously described considered significant, p values , 0.01 are considered very significant, and (36, 40). Briefly, real-time PCR was performed using TaqMan primers p values , 0.001 are considered highly significant. and probes (FAM-labeled) specific for mouse Chka, lysophosphati- dylcholine acyltransferase (Lpcat) 1, fatty acid amide hydrolase (Faah), glycerophosphodiester phosphodiesterase domain containing 3 Results (Gdpd3), Plcd3, or diacylglycerol kinase a (Dgka). Each reaction also TRAF3-mediated regulation of polar metabolites included the probe (VIC-labeled) and primers for mouse Actb,which 2/2 served as an endogenous control. Reactions were performed on a 7500 To explore the metabolic basis of the prolonged survival of Traf3 Fast Real-Time PCR System (Applied Biosystems). Relative mRNA B cells, we performed a metabolomic screening by LC-MS. We 4 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS compared the steady-state levels of water-soluble polar metabolites in 169 lipid species, among which 11 lipids were significantly elevated LMC and Traf32/2 splenic B cells purified from gender-matched and eight lipids were significantly decreased in Traf32/2 B cells as young adult (8- to 12-wk-old) mice by magnetic sorting, either di- compared with LMC B cells (Supplemental Fig. 2A, 2B). Interest- rectly ex vivo or after culture for 1 d. Comparison of direct ex vivo ingly, two PC species (32:2 and 34:3) and five phosphatidyletha- B cells revealed the metabolic differences between the two ge- nolamine species (32:1, 34:2, 36:3, 46:10, and O-36:1) were notypes in the presence of endogenous B cell survival factor markedly elevated in Traf32/2 B cells both directly ex vivo and at BAFF, whereas comparison of postculture B cells allowed us to day 1 after culture (Fig. 1B, 1C). Three phosphatidylinositol species identify the metabolic changes of the two genotypes in response to (36:4, 38:4, and 38:5) and one galactosylceramide (Gal-Cer 18:3) in vitro culture in the absence of BAFF. Our results showed that were increased in Traf32/2 B cells only at day 1 after culture but 16 metabolites were significantly upregulated in Traf32/2 B cells not directly ex vivo (Supplemental Fig. 2C). In contrast, four (Supplemental Fig. 1A, 1B). We have analyzed splenic B cells de- diacylglycerol (DAG) species (DAG_NH4 34:0, 36:0, 38:0, and 40:1) rived from three female and three male mice of each genotype and and two monoacylglycerol (MAG) species (MAG_Na 16:0 and did not detect significant gender difference in the effects of TRAF3 18:0) were significantly decreased in Traf32/2 B cells at day 0 but not ablation on polar metabolites. Both female and male Traf32/2 B cells at day 1 after culture, whereas two ceramides (d18:1/26:4 and d18:1/ exhibited consistent elevation of these 16 metabolites. 26:5) were moderately decreased in Traf32/2 B cells only at day 1 Interestingly, we found that P-Cho and phosphodimethylethanol- after culture (Supplemental Fig. 2C). Given the prolonged survival amine (P-DMEtn), two precursors of phospholipids, were significantly phenotype of Traf32/2 B cells both in vivo in mice and in vitro in increased in Traf32/2 B cells both directly ex vivo (day 0) and at culture (4, 5), the consistent elevation of the two PC species and five day 1 after culture (Fig. 1A). Five glucose metabolic intermediates phosphatidylethanolamine species in Traf32/2 B cells both directly were also significantly upregulated in Traf32/2 B cells, including ex vivo and at day 1 after culture highlights the potential importance Downloaded from glucose-6-phosphate (G6P), dihydroxyacetone phosphate (DHAP), of these phospholipids in Traf32/2 B cell survival. glyceraldehyde-3-phosphate (GA3P), ribose-5-phosphate (ribose-5- P), and sedoheptoluse-7-phosphate (S7P) (Supplemental Fig. 1C). Metabolic enzymes regulated by TRAF3 Surprisingly, although G6P is the convergence point of the glycolytic To understand how TRAF3 regulates cellular metabolism in and pentose phosphate pathways (PPP) (45), we only detected an B cells, we set out to identify metabolic enzymes regulated by increase in metabolites of the nonoxidative PPP. In contrast, metab- TRAF3. We performed a transcriptome profiling using LMC and http://www.jimmunol.org/ olites of aerobic glycolysis (such as 3-phosphoglyceric acid [3PG], Traf32/2 splenic B cells purified from gender-matched young pyruvate [PYR], and lactate), the TCA cycle (such as citrate, adult mice. We identified 101 genes differentially expressed be- a-ketoglutarate, and malate), and the oxidative PPP (such as 6- tween LMC and Traf32/2 splenic B cells, including 65 upregulated phosphogluconolactone [6PGL], 6-phosphogluconic acid [6PG], and genes and 36 downregulated genes in Traf32/2 B cells (the cutoff NADPH) were not significantly different between Traf32/2 and LMC fold of changes are as follows: 2-fold up or down, p , 0.05) B cells (Supplemental Fig. 1C). It is possible that some changes in (National Center of Biotechnology Information Gene Expression metabolites may have been missed in our metabolomic study, espe- Omnibus accession number: GSE113920) (https://www.ncbi.nlm. cially given the evidence that some metabolites are unstable during nih.gov/geo/). Functional clustering and signaling pathway analy- by guest on October 3, 2021 the LC-MS run time (46, 47). However, it is notable that the primary ses by Ingenuity (http://www.ingenuity.com) revealed that 17 meta- role of the nonoxidative PPP is to generate ribose-5-P, the molecular bolic enzymes were up- or downregulated at least 2-fold in Traf32/2 backbone of ribonucleotide biosynthesis (45, 48–50). Consistent B cells (Supplemental Fig. 3A). These include eight enzymes in- with this notion, we detected an increase of nine ribonucleotides in volved in Cho and phospholipid metabolism, four enzymes in nu- Traf32/2 B cells, including UMP, CMP, GMP, AMP, CDP, UDP, cleotide and DNA metabolism, three enzymes in glycolysis and GDP, ADP, and inosine (Supplemental Fig. 1C). It is intriguing that glycosylation, and two enzymes in amino acid metabolism no significant differences were observed in the concentrations of the (Supplemental Fig. 3B). Moreover, we detected an additional 29 ribonucleotide triphosphates (Supplemental Fig. 1C). Based on the metabolic enzymes that were up- or downregulated 1.5- to 2-fold adenylate energy charge (AEC) index (51), LMC B cells had an in Traf32/2 B cells (p , 0.05; Supplemental Fig. 3C, 3D). increased AEC at day 1 after culture because of drastically de- By integrating information across our metabolomic, lipidomic, and creased AMP concentrations (Supplemental Fig. 1D). This in- transcriptomic datasets, we identified a major clustering of TRAF3- creased AEC might predispose LMC B cells to apoptosis, as a mediated regulation of B cell metabolism in the interconnected Cho similar increase of AEC was observed in cerebellar granule cells en and ethanolamine (Etn) metabolic pathways (Fig. 1E). route to apoptosis (52). We also directly measured ATP concen- We prioritized the metabolic enzymes identified by the micro- trations in splenic B cells using an ATP Determination Kit. Our array analysis according to their potential importance in the Cho results showed that ATP concentrations were slightly lower in LMC and Etn metabolic pathways. We selected six enzyme-encoding B cells than in Traf32/2 B cells at day 1 after culture, but this genes altered in Traf32/2 B cells for further verification by quanti- difference is not statistically significant (Supplemental Fig. 1D). tative real-time PCR. Our results verified the transcript changes of Taken together, our results of polar metabolites indicate that TRAF3 the examined enzyme genes in Traf32/2 B cells (Fig. 1D). Most regulates the Cho metabolism, nonoxidative PPP, and ribonucleo- relevant to the upregulation of PC and phosphatidylethanolamine tide metabolic pathways in B cells. was increased Chka and Lpcat1, as well as decreased Gdpd3.The decreased DAG and MAG may relate to lower Dgka and upregulated Regulation of lipid metabolism by TRAF3 Faah (Fig. 1D, 1E). These results suggest that TRAF3 regulates the In light of the observed upregulation of two precursors of phos- expression of metabolic enzymes in B cells to control PC and pholipids and the known pivotal importance of phospholipids in phosphatidylethanolamine metabolism. maintaining cell survival (53, 54), we sought to identify lipids and phospholipids that are regulated by TRAF3 in B cells. We per- TRAF3-mediated regulation of the Kennedy pathway of the formed an LC-MS–based lipidomic screening to compare the P-Cho–PC synthesis steady-state levels of lipids between LMC and Traf32/2 splenic The two major pathways of the phospholipid PC synthesis are de B cells purified from gender-matched young adult mice. We detected novo biosynthesis (also termed the Kennedy pathway) and sequential The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 1. TRAF3 inhibited PC and phosphatidylethanolamine metabolic pathways in B cells. Splenic B cells were purified from gender-matched, young adult (8- to 12-wk-old) LMC or B-Traf32/2 mice. Water-soluble polar metabolites (A) or lipids (B and C) were extracted from cells directly ex vivo (day 0) or after culture for 1 d and then analyzed by LC-MS as described in Materials and Methods.(A) Increased levels of P-Cho and P-DMEtn in Traf32/2 Bcells.(B and C) Elevated levels of PC and phosphatidylethanolamine species in Traf32/2 Bcells.(D) Verification of transcript regulation of selected metabolic enzymes identified by the microarray analysis. Total cellular RNA was prepared from purified splenic B cells, and cDNA was synthesized by reverse transcription. Quantitative real-time PCR was performed using TaqMan assay kits specific for mouse Chka, Lpcat1, Faah, Gdpd3, Plcd3,orDgka. Relative mRNA levels were analyzed using the ΔΔcycle threshold method and normalized using b-actin mRNA as an endogenous control. Results shown in (A)–(D) are mean 6 SD (n = 6, including three female and three male samples for each genotype). (E) Pathway schematics showing TRAF3-mediated regulation of the interconnected PC and phosphatidylethanolamine metabolism in B cells. Enzymes are denoted in italic font in the schematics. Metabolites, lipids, and metabolic enzymes that were regulated by TRAF3 are indicated in red (for those upregulated in Traf32/2 B cells) or blue (for those downregulated in Traf32/2 B cells). *p , 0.05 (significantly different; t test), **p , 0.01 (very significantly different; t test), ***p , 0.001 (highly significantly different; t test) between LMC and Traf32/2 Bcells. methylation of phosphatidylethanolamine (Fig. 1E) (54–56). Chka, by Western blot analysis (Fig. 2A). Chka overexpression is thought also called Etn kinase, is the first enzyme of the Kennedy pathway. It to be responsible for the elevated P-Cho and PC phenotype in most directly phosphorylates both Cho and Etn with higher activity on Cho cases of human breast, lung, and prostate cancers (55, 57). We (Fig. 1E) (54–56). In this study, we demonstrated that Chka was likewise detected significantly elevated levels of P-Cho and PC in remarkably enhanced at the protein level in Traf32/2 splenic B cells premalignant Traf32/2 B cells (Fig. 1A, 1B). In this context, we next 6 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS

FIGURE 2. Elevated Chka and Cho metabolism in Traf32/2 B cells. Splenic B cells were purified from gender-matched, young adult LMC and B-Traf32/2 mice. (A) Upregulation of Chka protein in Traf32/2 B cells analyzed by Western blot analysis. Total cellular proteins were prepared from purified splenic B cells, separated by SDS- PAGE, and immunoblotted for Chka, fol- lowed by TRAF3 and b-actin. Results shown are representative of three indepen- dent experiments. (B and C) Analysis of

Cho metabolism by stable isotope labeling. Downloaded from Purified splenic B cells were cultured in mouse B cell medium containing 80 mg/ml 2 of trimethyl-D9-Cho ([ H]9-Cho) for the indicated time periods. The D9-labeled and unlabeled water-soluble metabolites (B) and lipids (C) of cultured splenic B cells were subsequently analyzed using LC-MS. (B) http://www.jimmunol.org/

Increased incorporation of D9-Cho into P-Cho in Traf32/2 B cells. (C) Elevated incorporation of D9-Cho into PC species in Traf32/2 B cells. The composition of the

D9-labeled and unlabeled P-Cho or PC species of each sample is shown in the bar graphs (left panel), and the kinetic in- crease of the D9-labeled P-Cho or PC species of each genotype of cells is by guest on October 3, 2021 shown in the curves (right panel). *p , 0.05 (significantly different; t test), **p , 0.01 (very significantly different; t test) between LMC and Traf32/2 B cells.

performed stable isotope labeling to determine if the in- indicate that the biosynthesis of P-Cho–PC is elevated in creased P-Cho is generated from de novo synthesis or from Traf32/2 Bcells. catabolism of Cho-containing metabolites and if the Kennedy a pathway contributes to the elevation of PC species. We cul- Chk inhibitors suppressed the survival of TRAF3-deficient tured LMC and Traf32/2 splenic B cells purified from gender- B cells in culture matched young adult mice in mouse B cell medium containing We next asked if Chka expression is consistently upregulated in 2 the stable isotope tracer trimethyl-D9-Cho ([ H]9-Cho) (58). TRAF3-deficient malignant B cells and if elevated Chka-mediated We subsequently analyzed the labeled and unlabeled polar Cho metabolism is important for oncogenic B cell survival. To ad- metabolites of cultured cells at different time points using dress this, we analyzed Chka expression levels in two Traf32/2 LC-MS. We found that Traf32/2 splenic B cells exhibited mouse B lymphoma cell lines, 27-9 and 105-8, that were previously significantly increased incorporation of the stable isotope derived from aging B-Traf32/2 mice (8, 30). Our results showed tracer D9-Cho into P-Cho (Fig. 2B). We also compared the that upregulation of Chka expression was maintained in these two labeled and unlabeled lipids of the cells at different time Traf32/2 mouse B lymphoma cell lines at a similarly high level as points using LC-MS (58). We detected a significantly higher that detected in premalignant Traf32/2 splenic B cells (Fig. 3A). levels of D9-Cho–labeled PC species (32:2 and 34:3) in To investigate the functional importance of elevated Cho meta- Traf32/2 B cells than in LMC B cells at 24 h after culture bolism, we tested the effects of two pharmacological inhibitors (Fig. 2C). Collectively, these stable isotope labeling data of Chka, MN58B (IC50: 1.4 mM) and RSM932A (IC50: 1 mM) The Journal of Immunology 7

(59–61), on the survival of these Traf32/2 mouse B lymphoma reconstitution of TRAF3 expression markedly decreased the cells. Our results obtained by MTT assay demonstrated that both protein levels of Chka in the transduced 8226 cells at day 3 MN58B and RSM932A potently inhibited the survival and pro- posttransduction, as shown by Western blot analysis (Fig. 4C). liferation of TRAF3-deficient mouse B lymphoma cell lines 27-9 We next performed stable isotope labeling to assess if recon- and 105-8 cells in a dose-dependent manner (Fig. 3C). These stitution of TRAF3 expression inhibits the Kennedy pathway of results suggest that TRAF3 inactivation–induced Chka upregu- P-Cho–PC biosynthesis. We cultured the transduced 8226 cells in lation is important for the survival of cultured TRAF3-deficient human B cell medium containing the stable isotope tracer trimethyl- mouse B lymphoma cells. D9-Cho at day 3 posttransduction. We subsequently analyzed the To strengthen the clinical relevance of our findings obtained from labeled and unlabeled polar metabolites as well as phospholipids at TRAF3-deficient mouse B cells, we next examined the expression different time points using LC-MS. Similar to that observed in of CHKA in three patient-derived MM cell lines that contain mouse splenic B cells, reconstitution of TRAF3 expression signifi- biallelic deletions or inactivating mutations of TRAF3, including cantly decreased the pool size of P‐Cho and two downstream 8226, KMS11, and LP1 (6). Our results revealed that the expression phospholipid PC species (32:2 and 34:3). This was associated with of CHKA was upregulated in all three TRAF3-deficient human MM slower production of labeled P-Cho and PC lipids (32:2 and 34:3) cell lines to a varied extent (3- to 7-fold as compared with normal from the D9‐labeled Cho tracer (Fig. 4D, 4E). We noticed that the B cells) (Fig. 3B), suggesting that TRAF3 inactivation also up- labeling fractions of both P-Cho and the two PC species did not regulates CHKA expression in human B cells. Of particular inter- significantly change between TRAF3-transduced and the control est, the two Chka inhibitors MN58B and RSM932A also suppressed vector–transduced cells, suggesting that the same metabolic path- the survival and proliferation of all these TRAF3-deficient human ways are being used in both cases. Therefore, the differences that MM cell lines in a dose-dependent manner (Fig. 3D). Thus, elevated we detected in the pool size of P-Cho and the two PC species were Downloaded from Chka-mediated Cho metabolism appears to play an indispensable due to different production rates of these molecules between and causal role in the survival phenotype of both mouse and human TRAF3-transduced and the control vector–transduced cells. Together, TRAF3-deficient malignant B cells in culture. our results demonstrate that reconstitution of TRAF3 inhibited To determine if elevation of Chka-mediated Cho metabolism is Chka expression and the Kennedy pathway of P-Cho–PC biosyn- ubiquitously critical for malignant B cell survival, we also ex- thesis as well as survival in transduced human MM 8226 cells. amined several TRAF3-sufficient B cell lines for comparison. We http://www.jimmunol.org/ ∼ Pharmacological inhibitors of Chka reversed the phenotype of found that Chka expression was ubiquitously upregulated ( 2- to 2/2 9-fold increase as compared with normal B cells) in all the the B cell compartment in B-Traf3 mice TRAF3-sufficient B cell lines examined, including the mouse B Our in vitro data regarding the roles of TRAF3 in Chka-mediated lymphoma cell lines A20.2J and m12.4.1 as well as human Bur- Cho metabolism in human MM cells (Fig. 4) and the effects of kitt’s lymphoma cell lines Ramos and Daudi (Fig. 3A, 3B). Thus, Chka inhibitors on the survival of TRAF3-deficient mouse B Chka expression may be upregulated in malignant B cells via lymphoma and human MM cells (Fig. 3) prompted us to further either TRAF3 inactivation–dependent or TRAF3-independent evaluate the in vivo effects of these inhibitors on the B cell hy- pathways. Interestingly, however, inhibition of Chka by MN58B perplasia phenotype in B-Traf32/2 mice. We administered by guest on October 3, 2021 was not effective, and RSM932A had minimal effects on sup- MN58B or RSM932A to gender-matched, young adult LMC and pressing the survival and proliferation of TRAF3-sufficient mouse B-Traf32/2 mice via i.p. injection at 2 mg/kg per mouse three B lymphoma cell lines A20.2J and m12.4.1 cells (Fig. 3C). In times a week for 4 wk and then analyzed the spleen size and contrast, both MN58B and RSM932A were effective in inhibiting splenic B cell compartment of the treated mice. Our results the survival and proliferation of TRAF3-sufficient human Bur- demonstrated that in vivo administration of either MN58B or kitt’s lymphoma cell line Ramos cells. Furthermore, it is per- RSM932A substantially decreased the spleen size as measured by plexing that MN58B failed to inhibit the survival and proliferation spleen weight in B-Traf32/2 mice but not in LMC mice (Fig. 5A). of another Burkitt’s lymphoma cell line, Daudi, although this cell Flow cytometric analysis revealed that treatment with MN58B or line was susceptible to RSM932A treatment (Fig. 3D). Taken RSM932A partially reduced the percentage and drastically de- together, the responses of different TRAF3-sufficient malignant creased the numbers of splenic B cells in B-Traf32/2 mice but not B cell lines to the Chka inhibitors varied considerably from cell in LMC mice (Fig. 5B, 5C). Furthermore, treatment with MN58B line to cell line, which likely reflects the differences in their on- or RSM932A remarkably decreased the percentage and numbers cogenic mutation landscapes and the presence or absence of re- of the splenic marginal zone B cell subset and also dramatically dundant Chka-independent Cho metabolic pathways. reduced the numbers of the follicular B cell subset in B-Traf32/2 mice, which was not observed in LMC mice (Fig. 5B, 5C). In Reconstitution of TRAF3 inhibits the Kennedy pathway of the contrast, MN58B or RSM932A did not significantly affect P-Cho–PC biosynthesis in human MM cells splenic T cell numbers in B-Traf32/2 or LMC mice (Fig. 5C). To further verify that TRAF3 regulates Chka-mediated Cho In summary, administration of pharmacological inhibitors of metabolism in human B cells, we selected a patient-derived MM Chka substantially reversed the B cell hyperplasia pheno- cell line 8226 that contains biallelic deletions of the TRAF3 gene type observed in B-Traf32/2 mice, suggesting that elevated to conduct TRAF3 reconstitution experiments (6). We transduced Chka-mediated P-Cho and PC biosynthesis contributes to the 8226 cells using a lentiviral expression vector, pUB–TRAF3– prolonged survival of Traf32/2 B cells in vivo. Thy1.1, to reconstitute the expression of TRAF3 proteins. Cells transduced with an empty lentiviral vector, pUB-Thy1.1, were Discussion used as a negative control in these experiments. Transduction Targeted therapies are currently the focus of anticancer research efficiency of the lentiviral vectors was .99% in human MM and are the core concept of precision medicine. Interestingly, 8226 cells, as demonstrated by FACS analysis (Fig. 4A). Re- targeting cancer metabolism has recently emerged as a promising constitution of TRAF3 expression induced cellular apoptosis in strategy for the development of selective and effective anticancer ∼50% of the transduced 8226 cells at day 7 posttransduction, drugs (62). Defining metabolic pathways regulated by critical on- as revealed by cell cycle analysis (Fig. 4B). Interestingly, cogenes and tumor suppressor genes is required for such therapeutic 8 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 3. Chka inhibitors MN58B and RSM932A suppressed the survival of TRAF3-deficient mouse and human malignant B cells in culture. (A) Upregulation of Chka expression in TRAF3-deficient mouse B lymphoma cell lines 27-9 and 105-8 as well as TRAF3-sufficient mouse B lymphoma cell lines A20.2J and m12.4.1. (B) Upregulation of Chka expression in TRAF3-deficient human MM cell lines 8226, KMS11, and LP1 as well as TRAF3- sufficient human Burkitt’s lymphoma cell lines Ramos and Daudi. Chka expression was determined by quantitative real-time PCR using TaqMan assays specific for mouse (A) and human (B) Chka, respectively, and results shown are mean 6 SD (n = 4). (C) Mouse B lymphoma cell lines 27-9, 105-8, A20.2J, and m12.4.1 cells were treated with various concentrations (1:2 serial dilutions) of MN58B or RSM932A for 24 h. (D) Human MM or Burkitt’s lymphoma cell lines 8226, KMS11, LP1, Ramos, and Daudi cells were treated with various concentrations (1:2 serial dilutions) of MN58B or RSM932A for 72 h. Total viable cell numbers were subsequently determined by MTT assay. Viable cell curves shown (C and D) are the results of at least three independent ex- periments with duplicate samples in each experiment (mean 6 SEM). The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 4. Reconstitution of TRAF3 expression inhibited the P-Cho–PC biosynthesis in human MM cells. The human MM cell line 8226 cells were transduced with a lentiviral expression vector of TRAF3 (pUB–TRAF3–Thy1.1) or an empty lentiviral expression vector (pUB-Thy1.1). (A) Transduction efficiency. Transduced 8226 cells were analyzed by Thy1.1 staining and flow cytometry at day 3 posttransduction. Dashed profiles show the untransduced cells (a negative control of the Thy1.1 staining), and gated populations (Thy1.1+) indicate cells that were successfully transduced with the lentiviral expression vector. (B) Representative FACS histograms of cell cycle analysis. Cell cycle distribution of transduced 8226 cells was analyzed by propidium iodide staining and flow cytometry at day 4 and 7 posttransduction. Gated populations indicate apoptotic cells (sub-G0: DNA content , 2n) and pro- liferating cells (S/G2/M phase: 2n , DNA content # 4n). (C) Expression of Chka and TRAF3 determined by Western blot analysis. Total cellular proteins were prepared at day 3 posttransduction and then immunoblotted for Chka, TRAF3, and actin. (D and E) Analysis of Cho metabolism by stable isotope labeling. At day 3 posttransduction, transduced 8226 cells were cultured in human B cell medium containing 80 mg/ml of trimethyl-D9-Cho. D9-labeled and unlabeled P-Cho (D) and PC species (E) were analyzed at the indicated time points by LC-MS. (D) Decreased incorporation of D9-Cho into P-Cho in pUB– TRAF3–Thy1.1 (TRAF3)–transduced 8226 cells. (E) Reduced incorporation of D9-Cho into two PC species in TRAF3-transduced 8226 cells. The composition of the D9-labeled and unlabeled P-Cho or PC species of each sample is shown in the bar graphs (top panel), and the kinetic increase of the D9-labeled P-Cho or PC species of each transduce cell type is shown in the curves (bottom panel). Results shown are mean 6 SEM (n = 3). *p , 0.05 (significantly different; t test), **p , 0.01 (very significantly different; t test) between TRAF3-transduced and the control vector pUB-transduced cells. exploitation. In the current study, we elucidated the metabolic well as responsiveness to therapy in human cancers (55, 57). The pathways regulated by a new tumor suppressor, TRAF3, in metabolic pathways of PC and phosphatidylethanolamine are B lymphocytes using metabolomic, lipidomic, and transcriptomic closely interconnected, and moderate perturbations of PC and analyses. We found that multiple polar metabolites, lipids, and phosphatidylethanolamine metabolism have profound effects on enzymes regulated by TRAF3 in B cells are clustered in the P-Cho, cell viability and apoptosis (63). Consistent with these previous PC, and phosphatidylethanolamine biosynthesis pathways (Fig. 1E). findings, our results showed that premalignant Traf32/2 B cells Aberrant Cho metabolism has been recognized as a cancer contained markedly elevated levels of P-Cho, P-DMEtn, PC, and hallmark associated with oncogenesis, invasion, and metastasis as phosphatidylethanolamine as well as Chka, the first enzyme of de 10 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 5. Chka inhibitors MN58B and RSM932A substantially reversed the B cell hyperplasia phenotype in B-Traf32/2 mice. Gender-matched, young adult LMC and B-Traf32/2 mice were administered i.p. with a Chka inhibitor, either MN58B or RSM932A, at 2 mg/kg per mouse or vehicle control three times a week for 4 wk. Spleen size and B cell compartment were analyzed at 2 d after the last injection. (A) Chka inhibitors reduced the spleen weights of B-Traf32/2 mice. (B) Representative FACS profiles of mouse splenocytes. B cells and T cells were identified by B220 and CD3 staining, respectively. Splenic B cell subsets were further differentiated by the markers CD21 and CD23. Marginal zone (MZ) B cell subset was identified as B220+CD21+CD23low, and follicular (FO) B cell subset was identified as B220+CD21intCD23+.(C) The percentages and numbers of splenic B cells, B cell subsets, and T cells analyzed by FACS. Effects of MN58B or RSM932A treatment were compared with the vehicle control group for each genotype. Results shown are mean 6 SD (n = 6, including three female and three male samples for each genotype). *p , 0.05 (significantly different; t test), **p , 0.01 (very significantly different; t test), ***p , 0.001 (highly significantly different; t test), ns, not significantly different (t test, p $ 0.05) between MN58B or RSM932A treated and the vehicle control group. novo PC and phosphatidylethanolamine biosynthesis (the Kennedy breast, lung, and prostate cancers as well as T cell lymphoma and pathway). Interestingly, Chka expression was ubiquitously up- leukemia (55, 57, 67, 68). In this study, our stable isotope labeling regulated in a variety of TRAF3-deficient as well as TRAF3- experiments demonstrated increased biosynthesis of P-Cho and sufficient malignant B cell lines examined in the current study, PC in Traf32/2 mouse B cells as well as decreased biosynthesis of including mouse B lymphoma cell lines, human MM, and Bur- P-Cho and PC in TRAF3-reconstituted human MM cells. Inter- kitt’s lymphoma cell lines. We also surveyed the public gene estingly, pharmacological inhibition of Chka reversed the survival expression databases of human cancers at Oncomine (http://www. phenotype of TRAF3-deficient B cells both in vitro and in vivo. oncomine.org) and learned that the expression of CHKA is sig- Taken together, our findings indicate that elevated Chka- nificantly increased in human DLBCLs (64–66). Overexpression mediated Cho metabolism supports the aberrantly prolonged of Chka has been shown to cause an elevated P-Cho and PC survival of Traf32/2 B cells. Therefore, our study established, phenotype and thus has been identified as a therapeutic target in to our knowledge, a novel connection between the tumor The Journal of Immunology 11 suppressor TRAF3 and the known oncogenic Chka–P-Cho–PC support that the Chka–P-Cho–PC metabolic pathway has diag- metabolic pathway in premalignant and malignant B cells. nostic and therapeutic value for B cell malignancies. For example, PC and phosphatidylethanolamine are the most abundant changes in P-Cho cellular levels can be noninvasively monitored phospholipids in cell membranes and play a dual role as structural in patients by in vivo imaging using Cho analogue tracers [18F]- components and substrates for lipid second messengers, such as Cho or [11C]-Cho in positron emission tomography/computed phosphatidic acid and DAG (69). We detected significantly in- tomography (82). Our study supports the use of such a Cho creased levels of multiple species of PC and phosphatidyletha- metabolism-based approach for early detection of B cell malig- nolamine but decreased levels of their catabolic products, such as nancies and for the assessment of malignant B cell responses to DAG and MAG in Traf32/2 B cells. In addition to Chka, we also therapy. Furthermore, inhibition of Cho metabolism by Chka in- discovered TRAF3-mediated regulation of several other enzymes hibitors or other drugs has the potential to improve the efficacy of that catalyze the anabolic or catabolic pathways of phospholipids standard chemotherapy, radiation therapy, and immunotherapy in in B cells, including Lpcat1, Faah, Gdpd3, Plcd3, Dgka, Lacc1, B cell malignancies, especially in patients with TRAF3 deletion or and Pip5k1b. Among these, upregulation of LPCAT1 and FAAH, relevant mutations. as well as downregulation of DGKA and PIP5K1B, has been documented in human cancers (70–75). In particular, the key Acknowledgments enzyme of an alternative pathway of PC synthesis, LPCAT1 of the We thank Dr. Jianjun Feng, Peter Zhang, Jemmie Tsai, Eris Spirollari, and Lands Cycle that converts lysophosphatidylcholine (LPC) to PC, Kyell Schwartz for technical assistance of this study. has been shown to play important roles in cancer pathogenesis and progression (70–72). Interestingly, Anxa4, a gene robustly up- 2 2 Disclosures Downloaded from regulated in Traf3 / B cells, has been recognized as an inhibitor The authors have no financial conflicts of interest. of phospholipase A2 (PLA2) (76), the catabolic enzyme of the Lands Cycle that mediates the breakdown of PC and phosphati- dylethanolamine (Fig. 1E). Thus, simultaneous upregulation of References both Lpcat1 and Anxa4 would likely promote the synthesis of PC 1. Morton, L. M., S. S. Wang, S. S. Devesa, P. Hartge, D. D. Weisenburger, and and inhibit the breakdown of PC and phosphatidylethanolamine, M. S. Linet. 2006. Lymphoma incidence patterns by WHO subtype in the United Blood States, 1992-2001. 107: 265–276. http://www.jimmunol.org/ which may also contribute to the increased PC and phosphati- 2. Ruddon, R. W. 2007. The epidemiology of human cancer. In Cancer Biology, 2 2 dylethanolamine levels detected in Traf3 / B cells. Collectively, 4th Ed. Oxford University Press, New York, p. 62–116. our results revealed the complex interactions between TRAF3 and 3. Horner, M. J., L. A. G. Ries, M. Krapcho, N. Neyman, R. Aminou, N. Howlader, S. F. Altekruse, E. J. Feuer, L. Huang, A. Mariotto, et al. 2008. SEER cancer phospholipid metabolic pathways associated with survival regu- statistics review, 1975-2006. Bethesda, MD: National Cancer Institute. Available lation in B cells. at: www.seer.cancer.gov/archive/csr/1975_2006/. 4. Xie, P., L. L. Stunz, K. D. Larison, B. Yang, and G. A. Bishop. 2007. Tumor We noticed that another major effect of TRAF3 on B cell necrosis factor receptor-associated factor 3 is a critical regulator of B cell ho- metabolism was clustered in the nonoxidative PPP and ribonu- meostasis in secondary lymphoid organs. Immunity 27: 253–267. cleotide metabolic pathways (Supplemental Fig. 4). In line with 5. Gardam, S., F. Sierro, A. Basten, F. Mackay, and R. Brink. 2008. TRAF2 and TRAF3 signal adapters act cooperatively to control the maturation and survival the increased levels of G6P, nonoxidative PPP metabolites, and

signals delivered to B cells by the BAFF receptor. Immunity 28: 391–401. by guest on October 3, 2021 ribonucleotides, we detected upregulation of the key enzyme of 6. Keats, J. J., R. Fonseca, M. Chesi, R. Schop, A. Baker, W. J. Chng, S. Van Wier, glycogen breakdown Pygl and two enzymes of ribonucleotide R. Tiedemann, C. X. Shi, M. Sebag, et al. 2007. Promiscuous mutations activate the noncanonical NF-kappaB pathway in multiple myeloma. Cancer Cell 12: biosynthesis (Mthfd1 and Adssl1) as well as downregulation of 131–144. two enzymes involved in ribonucleotide catabolism (Upb1 and 7. Annunziata, C. M., R. E. Davis, Y. Demchenko, W. Bellamy, A. Gabrea, F. Zhan, 2/2 G. Lenz, I. Hanamura, G. Wright, W. Xiao, et al. 2007. Frequent engagement of Pde2a)inTraf3 B cells. Relevant to our observation, Mam- the classical and alternative NF-kappaB pathways by diverse genetic abnor- betsariev et al. (77) reported that a glucose transporter (Glut1) and malities in multiple myeloma. Cancer Cell 12: 115–130. an enzyme hexokinase II (HKII) that converts glucose to G6P are 8. Moore, C. R., Y. Liu, C. Shao, L. R. Covey, H. C. Morse, III, and P. Xie. 2012. 2/2 Specific deletion of TRAF3 in B lymphocytes leads to B-lymphoma develop- upregulated in Traf3 B cells, although not detected in our ment in mice. Leukemia 26: 1122–1127. microarray analysis. Therefore, both elevated glycogen break- 9. Moore, C. R., S. K. Edwards, and P. Xie. 2015. Targeting TRAF3 downstream down and increased glucose uptake and conversion may lead to signaling pathways in B cell neoplasms. J. Cancer Sci. Ther. 7: 67–74. 2/2 10. Zhu, S., J. Jin, S. Gokhale, A. M. Lu, H. Shan, J. Feng, and P. Xie. 2018. Genetic increased nonoxidative PPP metabolism in Traf3 B cells. alterations of TRAF proteins in human cancers. Front. Immunol. 9: 2111. Corroborating our finding, MTHFD1, an enzyme of one-carbon 11. Walker, B. A., E. M. Boyle, C. P. Wardell, A. Murison, D. B. Begum, N. M. Dahir, P. Z. Proszek, D. C. Johnson, M. F. Kaiser, L. Melchor, et al. 2015. Mutational metabolism that is essential for de novo purine biosynthesis, is spectrum, copy number changes, and outcome: results of a sequencing study of upregulated in a variety of human B cell malignancies, including patients with newly diagnosed myeloma. J. Clin. Oncol. 33: 3911–3920. DLBCL, follicular lymphoma, B cell acute lymphoblastic leuke- 12. Hyeon, J., B. Lee, S. H. Shin, H. Y. Yoo, S. J. Kim, W. S. Kim, W. Y. Park, and Y. H. Ko. 2018. Targeted deep sequencing of gastric marginal zone lymphoma mia, B cell chronic lymphocytic leukemia, Burkitt’s lymphoma, identified alterations of TRAF3 and TNFAIP3 that were mutually exclusive for HL, and MM (Oncomine) (65, 66, 78, 79). A common polymor- MALT1 rearrangement. Mod. Pathol. 31: 1418–1428. phism of MTHFD1 R653Q (c.1958 G . A) in the synthetase 13. Rossi, D., S. Deaglio, D. Dominguez-Sola, S. Rasi, T. Vaisitti, C. Agostinelli, V. Spina, A. Bruscaggin, S. Monti, M. Cerri, et al. 2011. Alteration of BIRC3 domain impairs purine synthesis, and the AA genotype of MTHFD1 and multiple other NF-kB pathway genes in splenic marginal zone lymphoma. G1958A is associated with a decreased risk of B cell acute Blood 118: 4930–4934. 14. Zhang, B., D. P. Calado, Z. Wang, S. Fro¨hler, K. Ko¨chert, Y. Qian, S. B. Koralov, lymphoblastic leukemia and non-HL (80, 81). It is thus likely that M. Schmidt-Supprian, Y. Sasaki, C. Unitt, et al. 2015. An oncogenic role for the increased nonoxidative PPP metabolism and ribonucleotide alternative NF-kB signaling in DLBCL revealed upon deregulated BCL6 biosynthesis may also contribute to the prolonged survival of expression. Cell Rep. 11: 715–726. 2/2 15. Nagel, I., S. Bug, H. To¨nnies, O. Ammerpohl, J. Richter, I. Vater, E. Callet- Traf3 B cells. Bauchu, M. J. Calasanz, J. A. Martinez-Climent, C. Bastard, et al. 2009. Biallelic In summary, our study has contributed to a better understanding inactivation of TRAF3 in a subset of B-cell lymphomas with interstitial of the metabolic mechanisms underlying aberrant survival of del(14)(q24.1q32.33). Leukemia 23: 2153–2155. 16. Otto, C., M. Giefing, A. Massow, I. Vater, S. Gesk, M. Schlesner, J. Richter, B lymphocytes induced by inactivation of the tumor suppressor W. Klapper, M. L. Hansmann, R. Siebert, and R. Ku¨ppers. 2012. Genetic lesions TRAF3. We demonstrated that elevated Cho metabolism via the of the TRAF3 and MAP3K14 genes in classical Hodgkin lymphoma. Br. J. Haematol. 157: 702–708. Chka-driven de novo Kennedy pathway plays an important causal 17. Braggio, E., J. J. Keats, X. Leleu, S. Van Wier, V. H. Jimenez-Zepeda, R. Valdez, 2/2 role in the survival phenotype of Traf3 B cells. Our findings R. F. Schop, T. Price-Troska, K. Henderson, A. Sacco, et al. 2009. Identification 12 TRAF3 REGULATES CHOLINE METABOLISM IN B CELLS

of copy number abnormalities and inactivating mutations in two negative reg- 44. Xie, P., Z. J. Kraus, L. L. Stunz, Y. Liu, and G. A. Bishop. 2011. TNF receptor- ulators of nuclear factor-kappaB signaling pathways in Waldenstrom’s macro- associated factor 3 is required for T cell-mediated immunity and TCR/CD28 globulinemia. Cancer Res. 69: 3579–3588. signaling. J. Immunol. 186: 143–155. 18. Xie, P. 2013. TRAF molecules in cell signaling and in human diseases. J. Mol. 45. Hay, N. 2016. Reprogramming glucose metabolism in cancer: can it be exploited Signal. 8: 7. for cancer therapy? Nat. Rev. Cancer 16: 635–649. 19. Lalani, A. I., S. Zhu, S. Gokhale, J. Jin, and P. Xie. 2018. TRAF molecules in 46. Siegel, D., H. Permentier, D. J. Reijngoud, and R. Bischoff. 2014. Chemical and inflammation and inflammatory diseases. Curr. Pharmacol. Rep. 4: 64–90. technical challenges in the analysis of central carbon metabolites by liquid- 20. Bishop, G. A., and P. Xie. 2007. Multiple roles of TRAF3 signaling chromatography mass spectrometry. J. Chromatogr. B Analyt. Technol. in lymphocyte function. Immunol. Res. 39: 22–32. Biomed. Life Sci. 966: 21–33. 21. Xie, P., Z. J. Kraus, L. L. Stunz, and G. A. Bishop. 2008. Roles of TRAF 47. Al Kadhi, O., A. Melchini, R. Mithen, and S. Saha. 2017. Development of a LC- molecules in B lymphocyte function. Cytokine Growth Factor Rev. 19: 199–207. MS/MS method for the simultaneous detection of tricarboxylic acid cycle in- 22. Ha¨cker, H., P. H. Tseng, and M. Karin. 2011. Expanding TRAF function: TRAF3 termediates in a range of biological matrices. J. Anal. Methods Chem. 2017: as a tri-faced immune regulator. Nat. Rev. Immunol. 11: 457–468. 5391832. 23. Vince, J. E., W. W. Wong, N. Khan, R. Feltham, D. Chau, A. U. Ahmed, 48. Stincone, A., A. Prigione, T. Cramer, M. M. Wamelink, K. Campbell, E. Cheung, C. A. Benetatos, S. K. Chunduru, S. M. Condon, M. McKinlay, et al. 2007. IAP V. Olin-Sandoval, N. M. Gru¨ning, A. Kru¨ger, M. Tauqeer Alam, et al. 2015. The antagonists target cIAP1 to induce TNFalpha-dependent apoptosis. Cell 131: return of metabolism: biochemistry and physiology of the pentose phosphate 682–693. pathway. Biol. Rev. Camb. Philos. Soc. 90: 927–963. 24. Varfolomeev, E., J. W. Blankenship, S. M. Wayson, A. V. Fedorova, 49. Cairns, R. A., I. S. Harris, and T. W. Mak. 2011. Regulation of cancer cell N. Kayagaki, P. Garg, K. Zobel, J. N. Dynek, L. O. Elliott, H. J. Wallweber, et al. metabolism. Nat. Rev. Cancer 11: 85–95. 2007. IAP antagonists induce autoubiquitination of c-IAPs, NF-kappaB activa- 50. Jiang, P., W. Du, and M. Wu. 2014. Regulation of the pentose phosphate pathway tion, and TNFalpha-dependent apoptosis. Cell 131: 669–681. in cancer. Protein Cell 5: 592–602. 25. Zarnegar, B. J., Y. Wang, D. J. Mahoney, P. W. Dempsey, H. H. Cheung, J. He, 51. Teo, Z., M. K. Sng, J. S. K. Chan, M. M. K. Lim, Y. Li, L. Li, T. Phua, J. Y. T. Shiba, X. Yang, W. C. Yeh, T. W. Mak, et al. 2008. Noncanonical NF-kappaB H. Lee, Z. W. Tan, P. Zhu, and N. S. Tan. 2017. Elevation of adenylate energy activation requires coordinated assembly of a regulatory complex of the adaptors charge by angiopoietin-like 4 enhances epithelial-mesenchymal transition by cIAP1, cIAP2, TRAF2 and TRAF3 and the kinase NIK. Nat. Immunol. 9: 1371– inducing 14-3-3g expression. Oncogene 36: 6408–6419. 1378. 52. Atlante, A., S. Giannattasio, A. Bobba, S. Gagliardi, V. Petragallo, P. Calissano, 26. Vallabhapurapu, S., A. Matsuzawa, W. Zhang, P. H. Tseng, J. J. Keats, H. Wang, E. Marra, and S. Passarella. 2005. An increase in the ATP levels occurs in Downloaded from D. A. Vignali, P. L. Bergsagel, and M. Karin. 2008. Nonredundant and com- cerebellar granule cells en route to apoptosis in which ATP derives from both plementary functions of TRAF2 and TRAF3 in a ubiquitination cascade that oxidative phosphorylation and anaerobic glycolysis. Biochim. Biophys. Acta activates NIK-dependent alternative NF-kappaB signaling. Nat. Immunol. 9: 1708: 50–62. 1364–1370. 53. Mayr, J. A. 2015. Lipid metabolism in mitochondrial membranes. J. Inherit. 27. Matsuzawa, A., P. H. Tseng, S. Vallabhapurapu, J. L. Luo, W. Zhang, H. Wang, Metab. Dis. 38: 137–144. D. A. Vignali, E. Gallagher, and M. Karin. 2008. Essential cytoplasmic trans- 54. Mejia, E. M., and G. M. Hatch. 2016. Mitochondrial phospholipids: role location of a cytokine receptor-assembled signaling complex. Science 321: 663– in mitochondrial function. [Published erratum appears in 2015 J. Bioenerg.

668. Biomembr. 47: 279–280.] J. Bioenerg. Biomembr. 48: 99–112. http://www.jimmunol.org/ 28. Gardam, S., V. M. Turner, H. Anderton, S. Limaye, A. Basten, F. Koentgen, 55. Glunde, K., Z. M. Bhujwalla, and S. M. Ronen. 2011. Choline metabolism in D. L. Vaux, J. Silke, and R. Brink. 2011. Deletion of cIAP1 and cIAP2 in murine malignant transformation. Nat. Rev. Cancer 11: 835–848. B lymphocytes constitutively activates cell survival pathways and inactivates the 56. Ridgway, N. D. 2013. The role of phosphatidylcholine and choline metabolites germinal center response. Blood 117: 4041–4051. to cell proliferation and survival. Crit. Rev. Biochem. Mol. Biol. 48: 20–38. 29. Rickert, R. C., J. Jellusova, and A. V. Miletic. 2011. Signaling by the tumor 57. Arlauckas, S. P., A. V. Popov, and E. J. Delikatny. 2016. Choline kinase alpha- necrosis factor receptor superfamily in B-cell biology and disease. Immunol. Rev. putting the ChoK-hold on tumor metabolism. Prog. Lipid Res. 63: 28–40. 244: 115–133. 58. Ecker, J., and G. Liebisch. 2014. Application of stable isotopes to investigate the 30. Edwards, S. K., C. R. Moore, Y. Liu, S. Grewal, L. R. Covey, and P. Xie. 2013. metabolism of fatty acids, glycerophospholipid and sphingolipid species. Prog. N-benzyladriamycin-14-valerate (AD 198) exhibits potent anti-tumor activity on Lipid Res. 54: 14–31. TRAF3-deficient mouse B lymphoma and human multiple myeloma. BMC 59. Lacal, J. C., and J. M. Campos. 2015. Preclinical characterization of RSM-932A, Cancer 13: 481. a novel anticancer drug targeting the human choline kinase alpha, an enzyme

31. Wang, L., X. Xing, L. Chen, L. Yang, X. Su, H. Rabitz, W. Lu, and involved in increased lipid metabolism of cancer cells. Mol. Cancer Ther. 14: by guest on October 3, 2021 J. D. Rabinowitz. 2019. Peak annotation and verification engine for untargeted 31–39. LC-MS metabolomics. Anal. Chem. 91: 1838–1846. 60. Rodrı´guez-Gonza´lez, A., A. Ramı´rez de Molina, F. Ferna´ndez, M. A. Ramos, 32. Papazyan, R., Z. Sun, Y. H. Kim, P. M. Titchenell, D. A. Hill, W. Lu, M. Damle, M. del Carmen Nu´n˜ez, J. Campos, and J. C. Lacal. 2003. Inhibition of choline M. Wan, Y. Zhang, E. R. Briggs, et al. 2016. Physiological suppression of lip- kinase as a specific cytotoxic strategy in oncogene-transformed cells. Oncogene otoxic liver damage by complementary actions of HDAC3 and SCAP/SREBP. 22: 8803–8812. Cell Metab. 24: 863–874. 61. Sanchez-Lopez, E., T. Zimmerman, T. Gomez del Pulgar, M. P. Moyer, 33. Melamud, E., L. Vastag, and J. D. Rabinowitz. 2010. Metabolomic analysis and J. C. Lacal Sanjuan, and A. Cebrian. 2013. Choline kinase inhibition induces visualization engine for LC-MS data. Anal. Chem. 82: 9818–9826. exacerbated endoplasmic reticulum stress and triggers apoptosis via CHOP in 34. Kamphorst, J. J., J. Fan, W. Lu, E. White, and J. D. Rabinowitz. 2011. Liquid cancer cells. Cell Death Dis. 4: e933. chromatography-high resolution mass spectrometry analysis of fatty acid 62. Rahman, M., and M. R. Hasan. 2015. Cancer metabolism and drug resistance. metabolism. Anal. Chem. 83: 9114–9122. Metabolites 5: 571–600. 35. Chida, J., K. Yamane, T. Takei, and H. Kido. 2012. An efficient extraction 63. van der Veen, J. N., J. P. Kennelly, S. Wan, J. E. Vance, D. E. Vance, and method for quantitation of adenosine triphosphate in mammalian tissues and R. L. Jacobs. 2017. The critical role of phosphatidylcholine and phosphatidyl- cells. Anal. Chim. Acta 727: 8–12. ethanolamine metabolism in health and disease. Biochim. Biophys. Acta Bio- 36. Edwards, S. K., J. Baron, C. R. Moore, Y. Liu, D. H. Perlman, R. P. Hart, and membr. 1859(9 Pt B): 1558–1572. P. Xie. 2014. Mutated in colorectal cancer (MCC) is a novel oncogene in 64. Rosenwald, A., G. Wright, W. C. Chan, J. M. Connors, E. Campo, R. I. Fisher, B lymphocytes. J. Hematol. Oncol. 7: 56. R. D. Gascoyne, H. K. Muller-Hermelink, E. B. Smeland, J. M. Giltnane, et al; 37. Du, P., W. A. Kibbe, and S. M. Lin. 2007. nuID: a universal naming scheme Lymphoma/Leukemia Molecular Profiling Project. 2002. The use of molecular of oligonucleotides for illumina, affymetrix, and other microarrays. Biol. profiling to predict survival after chemotherapy for diffuse large-B-cell lym- Direct 2: 16. phoma. N. Engl. J. Med. 346: 1937–1947. 38. Du, P., W. A. Kibbe, and S. M. Lin. 2008. Lumi: a pipeline for processing 65. Alizadeh, A. A., M. B. Eisen, R. E. Davis, C. Ma, I. S. Lossos, A. Rosenwald, Illumina microarray. Bioinformatics 24: 1547–1548. J. C. Boldrick, H. Sabet, T. Tran, X. Yu, et al. 2000. Distinct types of diffuse 39. Smyth, G. K. 2005. Limma: linear models for microarray data. In Bioinformatics large B-cell lymphoma identified by gene expression profiling. Nature 403: 503– and Computational Biology Solutions using R and Bioconductor. R. Gentleman, 511. V. Carey, S. Dudoit, R. Irizarry, and W. Huber, eds. Springer, New York, p. 397– 66. Rosenwald, A., A. A. Alizadeh, G. Widhopf, R. Simon, R. E. Davis, X. Yu, 420. L. Yang, O. K. Pickeral, L. Z. Rassenti, J. Powell, et al. 2001. Relation of gene 40. Lalani, A. I., C. R. Moore, C. Luo, B. Z. Kreider, Y. Liu, H. C. Morse, III, expression phenotype to immunoglobulin mutation genotype in B cell chronic and P. Xie. 2015. Myeloid cell TRAF3 regulates immune responses and lymphocytic leukemia. J. Exp. Med. 194: 1639–1647. inhibits inflammation and tumor development in mice. J. Immunol. 194: 67. Xiong, J., J. Bian, L. Wang, J. Y. Zhou, Y. Wang, Y. Zhao, L. L. Wu, J. J. Hu, 334–348. B. Li, S. J. Chen, et al. 2015. Dysregulated choline metabolism in T-cell lym- 41. Edwards, S. K., Y. Han, Y. Liu, B. Z. Kreider, Y. Liu, S. Grewal, A. Desai, phoma: role of choline kinase-a and therapeutic targeting. Blood Cancer J. J. Baron, C. R. Moore, C. Luo, and P. Xie. 2016. Signaling mechanisms of 5: 287. bortezomib in TRAF3-deficient mouse B lymphoma and human multiple mye- 68. Mariotto, E., R. Bortolozzi, I. Volpin, D. Carta, V. Serafin, B. Accordi, G. Basso, loma cells. Leuk. Res. 41: 85–95. P. L. Navarro, L. C. Lo´pez-Cara, and G. Viola. 2018. EB-3D a novel choline 42. Zhou, S., E. A. Kurt-Jones, A. M. Cerny, M. Chan, R. T. Bronson, and kinase inhibitor induces deregulation of the AMPK-mTOR pathway and apo- R. W. Finberg. 2009. MyD88 intrinsically regulates CD4 T-cell responses. ptosis in leukemia T-cells. Biochem. Pharmacol. 155: 213–223. J. Virol. 83: 1625–1634. 69. Mori, N., F. Wildes, S. Kakkad, D. Jacob, M. Solaiyappan, K. Glunde, and 43. Edwards, S. K., A. Desai, Y. Liu, C. R. Moore, and P. Xie. 2014. Expression Z. M. Bhujwalla. 2015. Choline kinase-a protein and phosphatidylcholine but and function of a novel isoform of Sox5 in malignant B cells. Leuk. Res. 38: not phosphocholine are required for breast cancer cell survival. NMR Biomed. 393–401. 28: 1697–1706. The Journal of Immunology 13

70. Abdelzaher, E., and M. F. Mostafa. 2015. Lysophosphatidylcholine acyl- 77. Mambetsariev, N., W. W. Lin, A. M. Wallis, L. L. Stunz, and G. A. Bishop. 2016. transferase 1 (LPCAT1) upregulation in breast carcinoma contributes to tumor TRAF3 deficiency promotes metabolic reprogramming in B cells. Sci. Rep. progression and predicts early tumor recurrence. Tumour Biol. 36: 5473–5483. 6: 35349. 71. Grupp, K., S. Sanader, H. Sirma, R. Simon, C. Koop, K. Prien, C. Hube-Magg, 78. Andersson, A., C. Ritz, D. Lindgren, P. Ede´n, C. Lassen, J. Heldrup, T. Olofsson, G. Salomon, M. Graefen, H. Heinzer, et al. 2013. High lysophosphatidylcholine J. Rade,˚ M. Fontes, A. Porwit-Macdonald, et al. 2007. Microarray-based clas- acyltransferase 1 expression independently predicts high risk for biochemical sification of a consecutive series of 121 childhood acute leukemias: prediction of recurrence in prostate cancers. Mol. Oncol. 7: 1001–1011. leukemic and genetic subtype as well as of minimal residual disease status. 72. Uehara, T., H. Kikuchi, S. Miyazaki, I. Iino, T. Setoguchi, Y. Hiramatsu, Leukemia 21: 1198–1203. M. Ohta, K. Kamiya, Y. Morita, H. Tanaka, et al. 2016. Overexpression of 79.Zhan,F.,B.Barlogie,V.Arzoumanian,Y.Huang,D.R.Williams, lysophosphatidylcholine acyltransferase 1 and concomitant lipid alterations in K. Hollmig, M. Pineda-Roman, G. Tricot, F. van Rhee, M. Zangari, et al. gastric cancer. Ann. Surg. Oncol. 23(Suppl. 2): S206–S213. 2007. Gene-expression signature of benign monoclonal gammopathy ev- 73. Endsley, M. P., R. Thill, I. Choudhry, C. L. Williams, A. Kajdacsy-Balla, ident in multiple myeloma is linked to good prognosis. Blood 109: 1692– W. B. Campbell, and K. Nithipatikom. 2008. Expression and function of fatty 1700. acid amide hydrolase in prostate cancer. Int. J. Cancer 123: 1318–1326. 80. Lautner-Csorba, O., A. Ge´zsi, D. J. Erde´lyi, G. Hulla´m, P. Antal, 74. Kong, Y., Y. Zheng, Y. Jia, P. Li, and Y. Wang. 2016. Decreased LIPF expression A. F. Semsei, N. Kutszegi, G. Kova´cs, A. Falus, and C. Szalai. 2013. Roles of is correlated with DGKA and predicts poor outcome of gastric cancer. Oncol. genetic polymorphisms in the folate pathway in childhood acute lympho- Rep. 36: 1852–1860. blastic leukemia evaluated by Bayesian relevance and effect size analysis. 75. Caprini, E., C. Cristofoletti, D. Arcelli, P. Fadda, M. H. Citterich, F. Sampogna, PLoS One 8: e69843. A. Magrelli, F. Censi, P. Torreri, M. Frontani, et al. 2009. Identification of key 81. Weiner, A. S., O. V. Beresina, E. N. Voronina, E. N. Voropaeva, U. A. Boyarskih, regions and genes important in the pathogenesis of sezary syndrome by com- T. I. Pospelova, and M. L. Filipenko. 2011. Polymorphisms in folate- bining genomic and expression microarrays. Cancer Res. 69: 8438–8446. metabolizing genes and risk of non-Hodgkin’s lymphoma. Leuk. Res. 35: 508– 76. Zhang, F., J. Sha, T. G. Wood, C. L. Galindo, H. R. Garner, M. F. Burkart, 515. G. Suarez, J. C. Sierra, S. L. Agar, J. W. Peterson, and A. K. Chopra. 2008. 82. Cuccurullo, V., G. D. Di Stasio, L. Evangelista, G. Castoria, and L. Mansi. 2017. Alteration in the activation state of new inflammation-associated targets by Biochemical and pathophysiological premises to positron emission tomography phospholipase A2-activating protein (PLAA). Cell. Signal. 20: 844–861. with choline radiotracers. J. Cell. Physiol. 232: 270–275. Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021 Supplementary Figure 1. Metabolomic screening of splenic B cells by LC-MS. Splenic B cells were purified from gender-matched, young adult (8-12-week-old) LMC and B-Traf3-/- mice. Water-soluble polar metabolites were extracted from cells directly ex vivo (day 0) or after cultured in vitro in mouse B cell medium for 1 day, and then analyzed by LC-MS. Totally, 107 metabolites were detected in resting splenic B cells. (A) Numbers of metabolites that were not changed or up-regulated in Traf3-/- B cells. (B) Distribution of the metabolites that were significantly increased in Traf3-/- B cells. (C) The graphic results of glucose metabolic intermediates and ribonucleotides that were significantly increased in Traf3-/- B cells. The 4 NTP graphs are also included for comparison to the NMP and NDP data. (D) AEC index calculated based on the LC-MS data of ATP, ADP and AMP using the formula: AEC = ([ATP]+0.5[ADP])/([ATP]+[ADP]+[AMP]). (E) Cellular ATP concentration measured by a Luciferase-Luciferin-based ATP Determination Kit (Molecular Probes). Results shown are mean ± SD (n=6, including 3 female and 3 male samples for each genotype). *, significantly different (t test, p < 0.05); **, very significantly different (t test, p < 0.01); ***, highly significantly different (t test, p < 0.001); ns, not significantly different between LMC and Traf3-/- B cells (t test, p ≥ 0.05). Supplementary Figure 2. Lipidomic screening of splenic B cells by LC-MS. Splenic B cells were purified from gender-matched, young adult (8-12-week-old) LMC or B-Traf3-/- mice. Lipids were extracted from cells directly ex vivo (day 0) or after cultured in vitro for 1 day, and then analyzed by LC-MS. Totally, 169 lipid molecules were detected in resting splenic B cells. (A) Numbers of lipids that were not changed, up-regulated or down-regulated in Traf3-/- B cells. (B) Distribution of lipid molecules that were elevated (red) or decreased (blue) in Traf3-/- B cells. (C) The graphic results of PI, DAG, MAG and ceramide species that were significantly altered in Traf3-/- B cells. DAGs were detected as NH4+ adducts and MAGs were detected as Na+ adducts. Results shown are mean ± SD (n=6, including 3 female and 3 male samples for each genotype). *, significantly different (t test, p < 0.05); **, very significantly different (t test, p < 0.01); ***, highly significantly different (t test, p < 0.001); ns, not significantly different between LMC and Traf3-/- B cells (t test, p ≥ 0.05). Supplementary Figure 3. Transcriptomic screening of splenic B cells by microarray analysis. Total cellular RNA was prepared from splenic B cells purified from gender-matched, young adult LMC and B-Traf3-/- mice, and processed for the microarray analysis to identify genes differentially expressed between the two genotypes of B cells. (A and B) Heatmap representation (A) and functional clustering (B) of the 17 metabolic enzymes that were up- or down-regulated at least 2-fold in Traf3-/- B cells. (C and D) Heatmap representation (C) and functional clustering (D) of additional 29 metabolic enzymes identified by the microarray analysis that were up- or down-regulated by 1.5 to 2-fold in Traf3-/- B cells. Supplementary Figure 4. Pathway schematics depicting TRAF3-mediated regulation of glucose and ribonucleotide metabolism in B cells. Enzymes are denoted in Italic font in the schematics. Glucose metabolic intermediates, ribonucleotides and metabolic enzymes that were regulated by TRAF3 are indicated in red (for those up-regulated in Traf3-/- B cells) or blue (for those down-regulated in Traf3-/- B cells). Abbreviations: G1P, glucose-1-phosphate; G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; F1,6P, fructose-1,6-biphosphate; DHAP, dihydroxyacetone phosphate; GA3P, glyceraldehyde 3-phosphate, 3PG, 3-phosphoglycerate; PYR, pyruvate; TCA cycle, the tricarboxylic acid cycle or the Krebs cycle; NADPH, dihydronicotinamide-adenine dinucleotide phosphate; 6PGL, 6-phospho-D-glucono-1,5-lactone; 6PG, 6-phosphogluconate; Ribulose-5-P, ribulose-5-phosphate; Ribose-5-P, ribose-5-phosphate; X5P, xylulose-5-phosphate; E4P, erythrose-4-phosphate; S7P, sedoheptoluse-7-phosphate; PRPP, 5-phospho- alpha-D-ribose 1-diphosphate; PPP, pentose phosphate pathway; Formyl-THF, formyltetrahydrofolate; THF, tetrahydrofolate; S-AMP, adenylosuccinate; Gly, glycine; Gln, glutamine; Asp, aspartic acid; IMP, inosine monophosphate; AMP, adenosine monophosphate; XMP, xanthosine monophosphate; GMP, guanosine monophosphate; UMP, uridine monophosphate; CMP, cytidine monophosphate; ADP, adenosine diphosphate; GDP, guanosine diphosphate; UDP, uridine diphosphate; CDP, cytidine diphosphate; ATP, adenosine triphosphate; GTP, guanosine triphosphate; UTP, uridine triphosphate; CTP, cytidine triphosphate; β-UPA, β-ureidopropionate; cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; Pygl, glycogen phosphorylase, liver form; Pgm2, phosphoglucomutase 2; GYS, glycogen synthase; GPI, glucose-6-phosphate isomerase; HK, hexokinase; HKII, hexokinase II; TKT, transketolase; TAL, transaldolase; RPE, D-ribulose-5-phosphate 3-epimerase; Rpia, ribose-5-phosphate isomerase A; Mthfd1, methylenetetrahydrofolate dehydrogenase 1; ADSSL1, adenylosuccinate synthase like 1; Upb1, β-ureidopropionase 1; Pde2a, cGMP-dependent 3',5'-cyclic phosphodiesterase 2A