Autoantigen Can Promote Progression to a More Aggressive TCL1 Leukemia by Selecting Variants with Enhanced B-Cell Receptor Signaling

Autoantigen can promote progression to a more aggressive TCL1 leukemia by selecting variants with enhanced B-cell receptor signaling

Shih-Shih Chena, Franak Batliwallab, Nichol E. Holodickb, Xiao-Jie Yana, Sophia Yancopoulosa, Carlo M. Crocec,1, Thomas L. Rothsteinb,d, and Nicholas Chiorazzia,d,1

aKarches Center for Chronic Lymphocytic Leukemia Research and bCenter for Oncology and Cell Biology, The Feinstein Institute for Medical Research, Manhasset, NY 11030; cHuman Cancer Genetics Program and Department of Molecular Virology, Immunology, and Medical Genetics, Ohio State University School of Medicine, Columbus, OH 43210; and dDepartment of Medicine and Department of Molecular Medicine, Hofstra North Shore-LIJ Health System School of Medicine, Hempstead, NY 11549

Contributed by Carlo M. Croce, January 19, 2013 (sent for review December 4, 2012) (Auto)antigen engagement by the B-cell receptor (BCR) and pos- regions and HCDR3s of CLL BCRs permit cell-autonomous sibly the sites where this occurs influence the outcome of chronic interactions that lead to BCR signaling (11). lymphocytic leukemia (CLL). To test if selection for autoreactivity Leukemic B cells of Eμ-TCL1 transgenic (TCL1) mice, a murine leads to increased aggressiveness and if this selection plays out model of CLL, exhibit many features of CLL (12, 13). TCL1 mice equally in primary and secondary tissues, we used T-cell leukemia develop clonal CD5+ leukemias with stereotyped BCRs binding (TCL)1 cells reactive with the autoantigen phosphatidylcholine exo- and autoantigens such as DNA, cardiolipin, phospholipids, (PtC). After repeated transfers of splenic lymphocytes from a single apoptotic cells, or microbes, which can be targets of human CLL mouse with oligoclonal PtC-reactive cells, outgrowth of cells Igs (14–16). B-lymphocytes reactive with phosphatidylcholine expressing a single IGHV-D-J rearrangement and superior PtC-binding (PtC), a phospholipid component of biological membranes in every and disease virulence occurred. In secondary tissues, increased PtC- cell of the body, use predominantly IGHV11 or IGHV12 genes, are binding correlated with enhanced BCR signaling and cell prolifer- abundant in the normal mouse B-1 subset, and are enriched in the ation, whereas reduced signaling and division of cells from the peritoneal and pleural cavities (17). In addition, anti-PtC IgMs are same clone was documented in cells residing in the bone marrow, found in normal individuals (18), CLL patients (19), TCL1 mice blood, and peritoneum, even though cells from the last site had (14), and patients with systemic lupus erythematosus (20). highest surface membrane IgM density. Gene-expression analyses Here, we endeavored to understand (auto)antigen-promoted revealed reciprocal changes of genes involved in BCR-, CD40-, and leukemia progression by following the evolution of PtC-binding PI3K-signaling between splenic and peritoneal cells. Our results B cells from a single TCL1 transgenic mouse after serial transfers suggest autoantigen-stimulated BCR signaling in secondary tissues into SCID mice. We identified natural selection for a leukemic B- promotes selection, expansion, and disease progression by activat- cell clone (TCL1-192) binding this specific autoantigen and ing pro-oncogenic signaling pathways, and that—outside second- exhibiting a more virulent behavior with faster growth kinetics than ary lymphoid tissues—clonal evolution is retarded by diminished the standard TCL1 model. Notably, despite the clonal nature of the BCR-signaling. This transferrable, antigenic-specific murine B-cell cell line, it displayed different efficiencies and outcomes to BCR clone (TCL1-192) provides a platform to study the types and sites signaling based on the site of cell residence in vivo. of antigen-BCR interactions and genetic alterations that result and may have relevance to patients. Results Surface Membrane IgM of Normal and TCL1 B-1 Cells Bind PtC. Using B lymphocytes | cell signaling | hematology | oncology | cancer fluorescein-encapsulated liposomes made from distearoyl-phosphatidylcholine (21), we analyzed CD5+ B cells from WT and Eμ- alignancies develop and progress to more virulent stages by TCL1 animals to compare PtC binding. Splenic B-2 cells in WT or accumulating genomic abnormalities that are often pro- preleukemic TCL1 mice did not bind PtC, but polyclonal perito- M – moted by normal biologic functions, in a cell type-specificand neal B-1 cells isolated from WT mice at various ages (3 10 mo) stepwise manner. Several lines of evidence suggest antigen-binding site structural selection, mediated in part by B-cell antigen re- Significance ceptor (BCR)–(auto)antigen interaction, facilitates survival and expansion of precursor cells and leukemic cells in chronic lym- These studies indicate that autoantigen-reactivity plays a role phocytic leukemia (CLL) (1, 2). CLL cells often use restricted in the progression of a murine leukemia that models human IGHV genes that frequently associate with specific IGHD and chronic lymphocytic leukemia. This indication is consistent with IGHJ segments to code their BCRs (1–3), yielding Igs with char- the notion that chronic lymphocytic leukemia evolves by se- acteristic HCDR3 regions (stereotyped BCRs). Such stereotyped lection of normal B cells that bind autoantigen via the B-cell BCRs often pair with discrete IGLV and IGLJ segments (4, 5). antigen receptor. CLL clones can be subgrouped based on the presence (M-CLL) or absence (U-CLL) of IGHV mutations (6), with more U-CLLs Author contributions: S.-S.C., T.L.R., and N.C. designed research; S.-S.C., F.B., and N.E.H. performed research; C.M.C. contributed new reagents/analytic tools; S.-S.C., F.B., N.E.H., exhibiting stereotyped receptors than M-CLLs (4, 7). U-CLL BCRs X.-J.Y., S.Y., C.M.C., T.L.R., and N.C. analyzed data; and S.-S.C. and N.C. wrote the paper. are more often polyreactive, binding a diverse panel of antigens, The authors declare no conflict of interest. than M-CLL BCRs, which are more restricted in antigen reactivity Freely available online through the PNAS open access option. (8). Clinically, U-CLL patients often have worse clinical outcomes Data deposition: The data reported in this paper have been deposited in the Gene Ex- than M-CLL patients (9, 10), suggesting that degrees of BCR pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE45481). polyreactivity, and therefore (auto)antigen binding, affect CLL 1To whom correspondence may be addressed. E-mail: [email protected] or nchizzi@ disease progression (1, 2). Despite this evidence, it has been con- nshs.edu. fi jectured that rather than speci c antigens or classes of antigens This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. driving CLL, structural complementarities between framework 1073/pnas.1300616110/-/DCSupplemental.

E1500–E1507 | PNAS | Published online April 2, 2013 www.pnas.org/cgi/doi/10.1073/pnas.1300616110 Downloaded by guest on September 23, 2021 contained 19–22% PtC-binders and peritoneal B-1 cells from expressed a series of IGHV1 and IGHV2 genes (Fig. 1B and SI PNAS PLUS young and old TCL1 mice varied in PtC binding ability (15–36%). Appendix,TableS1). Splenic CLL cells from moribund leukemic TCL1 mice (12–15 mo TCL1 clones propagate CLL in recipient mice after adoptive old) had even more diverse PtC-binding capacity, ranging from cell transfer (12, 22). To understand PtC-driven CLL progression, only 0.3–13% (SI Appendix,Fig.S1A–C). we retro-orbitally injected nonselected TCL1-192 spleen cells into PtC binding by CLL cells in TCL1 mice (SI Appendix,Fig.S1D) SCID mice. Recipient mice died within 6 mo, with lymphocytosis, ′ was blocked by preincubating cells with F(ab )2 fragments of lymphadenopathy, and hepatosplenomegaly, similar to TCL1 tu- polyclonal antibodies to IgM (anti-IgM pAbs), at concentrations mor-bearing mice. Most cells in the spleen and peritoneum of fi predetermined to signi cantly down-regulate surface membrane these mice were B220+CD5+, with 5.3% of spleen cells and 3.5% IgM (smIgM) expression by internalization. Thus, binding of PtC of peritoneal cells binding PtC. When sequenced, rearranged liposomes to the B-cell surface was mainly because of recognition IGHV-D-JsandIGLV-IGLJs from both PtC-binders (100%) and of PtC by smIgM. nonbinders (98%) mainly came from the original major clone, Selection for Accelerated Disease by Serial Adoptive Transfers of regardless of PtC binding capacity. The only different IGHV-D-J TCL1 CLL Cells Associates with BCR Binding of PtC. To investigate sequence was from the B-cell fraction that bound PtC minimally, if fi how autoantigen-binding promotes evolution of TCL1 cells to at all (Fig. 1, rst transfer). fi more aggressive disease, we selected a 15-mo-old mouse, TCL1- Nonselected splenic lymphocytes isolated from the rst mori- 192, whose spleen contained definable PtC-binding CLL cells bund recipients (∼6 mo posttransfer) were then injected into other (12.6%) (Fig. 1A). By IGHV-D-J sequence analyses, this mouse SCID mice; transfers were repeated for a total of four times. With contained two expanded B-cell clones; the major clone exhibited an each transfer, recipient mice had shortened survival. Recipients of IGHV11 rearrangement (58% of sequences) and the minor clone the first and second transfer died in 6 and 2 mo, respectively. By the an IGHV12 rearrangement (25% of sequences); the remaining cells third and fourth transfers, survival diminished to ∼6 wk. Impor-

A MEDICAL SCIENCES

Fig. 1. Adoptive transfer of TCL1-192 cells in SCID mice. (A)20× 106 nonselected splenic lymphocytes from TCL1-192 mouse were intravenously injected into two SCID mice. The recipient mice appeared moribund and were killed at 6 mo after cell engraftment; splenic lymphocytes were collected and injected into another three SCID mice (5 × 106 per mouse). At the time second transferred mice appeared moribund (2 mo posttransfer), the adoptive transfer process was repeated two more times; each time recipient mice survived for only 5–6 wk. Examples of PtC binders in donor and serial adoptive transferred mice were shown. Scatter graphs indicate mean ± SD. Statistical analysis was done by one-way ANOVA and unpaired t test using GraphPad Prism software. (B) Clones identiﬁed in donor TCL1-192 splenocytes and recipient splenic and peritoneal lymphocytes after each transfer. Results shown are sequences of IGHV-D-J and IGLVκ-Jκ cDNAs from unselected − splenic lymphocytes in the donor TCL1-192 mouse; and PtC+,PtC populations sorted from splenic and peritoneal lymphocytes in the transferred mice. *P < 0.05, **P < 0.01, ***P < 0.001.

Chen et al. PNAS | Published online April 2, 2013 | E1501 Downloaded by guest on September 23, 2021 tantly, as disease became more aggressive at each transfer, the or spleen had enlarged spleens and more leukemic cells in PC and PtC-binding population expanded and eventually became the spleen (Fig. 3 B and C). Cell counts in blood were also elevated, major population in spleen and peritoneum (Fig. 1A)(P < 0.001), although not significantly (Fig. 3C). Hence, regardless of collection suggesting that BCR-autoantigen stimulation promoted CLL site, TCL1-192 autoreactive leukemic cells have a growth advan- progression. tage over nonautoreactive cells when residing within but not out- Although PtC-binders were the major population after the side of secondary lymphoid tissues. Notably, mice injected with − second transfer, 43–53% TCL1-192 cells did not react with PtC; PtC cells developed only ∼34% PtC-reactive cells, but mice re- all PtC-binders (100%) and almost all nonbinders (98%) used ceiving cells binding PtC extremely well generated ∼50% PtC identical IGHV11-IGHD1-IGHJ1/IGLVκ14-IGLJκ4 (Fig. 1B). binders, indicating the dynamic nature of smIgM expression and At the end of the third transfer and all subsequent transfers, only PtC recognition in CLL cells; again, PtC binding correlated with one IGHV-(D)-J sequence was found (IGHV11-IGHD1-IGHJ1/ smIgM intensity (SI Appendix,Fig.S1E and F). IGLVκ14-IGLJκ4), even though some still failed to bind PtC to an appreciable degree after four transfers. Notably, replacement TCL1 Cells from the BM and PC Resemble Hyporesponsive B-1 Cells mutations in the IGV genes of these rearrangements were not in Phenotype, Phosphorylation, and Responsiveness to smIgM found, consistent with their B-1 cell origin (23). We named this Stimulation. Because TCL1-192 cells from the BM and PC clonal B-cell line with enhanced PtC binding and more rapid responded poorly to BCR-mediated signaling in vitro and pro- kinetics “TCL1-192.” liferated minimally in vivo (Figs. 2 and 3), we checked if their BCR- To understand migratory and growth characteristics of TCL1- signaling phenotypes were similar to normal murine peritoneal B-1 192, we analyzed seeding and expansion sites of retro-orbitally in- cells that are less responsive to smIgM stimulation than splenic B-1 jected cells at weekly intervals. In fourth-transfer mice, B220+CD5+ cells (24, 26). Interestingly, compared with splenic cells, TCL1-192 cells were detected in the spleen (SP) at week 1, and by week 3 in the CLL cells from the PC had elevated surface IgM (Fig. 4A), elevated blood and peritoneal cavity (PC). Lymphocyte infiltration in the liver basal levels of phosphorylated phospholipase Cγ2 (PLCγ2) and was observed later (≥4 wk). By week 6, TCL1-192 cells were found in spleen tyrosine kinase (SYK) (Fig. 4B), and diminished re- spleen, occasional inguinal lymph nodes (LNs), bone marrow (BM), sponsiveness to anti-IgM pAbs and PtC liposome stimulation in and liver of all recipient mice (SI Appendix,Fig.S2A and B). vitro, as measured by calcium flux (Fig. 4 C and D). Similar to PC TCL1-192 cell proliferation in different locales was measured but different from splenic TCL1-192 cells, BM cells were also by injecting BrdU intraperitoneally 24 h before killing at week 6. nonresponsive to anti-IgM stimulation (Fig. 4D, Upper). These data Divided cells were found mainly in secondary lymphoid tissues; correlate with cell proliferation results shown in SI Appendix, cells from spleens and inguinal LNs contained ∼20% BrdU- Fig. S2C, and suggest the BCR-stimulated CLL cell proliferation labeled cells, whereas only 2–3% divided cells were found in the occurs mainly in secondary but not primary lymphoid tissues and peritoneum, BM, or peripheral blood (PB) (SP: 16.5% ± 0.02 and extratissue compartments (PC and blood). LN: 21.3% ± 0.03 vs. PC: 2.7% ± 0.04, BM: 2.3% ± 0.06, and PB: Importantly, the reduced Ca2+ mobilization in TCL1-192 cells 1.9% ± 0.02, P < 0.001) (SI Appendix, Fig. S2C), suggesting sec- from the PC could be induced by preincubating TCL1-192 sple- ondary lymphoid tissues, not central or peripheral sites, are pre- nocytes for 24 h with PtC liposomes or peritoneal lavage collected ferred areas for leukemic cell proliferation. from recipient mice (Fig. 4D, Lower), suggesting hyporesponsiveness is a result of factors in the PC or because of different con- PtC-Reactive TCL1-192 Cells from Spleens Have a Cell Growth and centrations of antigen in distinct compartments, as for normal Disease Progression Advantage. TCL1-192 cells differing in PtC- murine B-1 cells (24). binding capacity were studied using a gating strategy shown in Fig. Next, we compared the fate of TCL1-192 cells injected directly 2A. Because PtC binding was smIg-dependent, we analyzed smIgM into the PC or into the venous circulation. TCL1-192 cells injected − density on PtC+ and PtC binding cells. The PtC-binding fraction intraperitoneally or retro-orbitally into mice were killed 1 or 5 wk had higher mean smIgM density than the non/poor PtC-binding after transfer. At week 1, intraperitoneally injected TCL1-192 cells fraction (mean fluorescence intensity: 23,270 ± 3,377 vs. 13,760 ± mainly resided in the PC, whereas retro-orbitally injected TCL1- 1,524, P = 0.016) (Fig. 2B). Consistent with enhanced PtC binding 192 cells infiltrated spleen, peritoneum, and blood. By week 5, with serial adoptive transfers, smIgM density increased on TCL1- numbers of leukemic cells in all three sites for both injection 192 cells most markedly for cells in the PC (Fig. 2C). conditions were indistinguishable. Notably, TCL1-192 cells that − Differences in BCR stimulation in the PtC+ and PtC frac- homed to the spleen exhibited superior proliferation, regardless of tions were studied by measuring Ca2+ flux after stimulation route of administration (SI Appendix,Fig.S2D and E). Collec- with F(ab′)2 fragments of anti-IgM pAbs. Splenic PtC binders tively, the data suggest the splenic microenvironment supports responded better than nonbinders. However, only low percen- expansion of CLL cells, and the peritoneal milieu induces quies- − tages of PtC+ and PtC cells from the peritoneum responded to cence and hyporesponsiveness, in part because of soluble factors. anti-IgM stimulation, despite increased smIgM density (Fig. 2D). Because BCR-signaling promotes B-1 proliferation and sur- Genetic Signatures of PtC-Binding Cells Suggest Heightened BCR vival of PtC binders in vitro (24, 25), we determined if TCL1-192 Signaling in Splenic but Diminished BCR and NFκB Signaling in PtC binders had a growth advantage over the PtC poorly binding Peritoneal TCL1-192 Cells. Gene expression differences between − fraction. BrdU incorporation in spleen and PC cells was measured PtC+ and PtC populations in different locales were compared for in fourth-transfer mice 24 h after receiving BrdU. Results (Fig. WT, preleukemic TCL1, and TCL1-192 mice. When comparing − 3A) indicated: (i) a higher percentage of BrdU+ cells in spleen differences in PtC+ vs. PtC populations from peritoneum and than peritoneum (25.1% ± 0.01 vs. 3.3% ± 0.07, P = 0.0005); (ii) spleen, a 1.5-fold cutoff was applied to identify candidate genes for − a higher percentage of BrdU+ cells in PtC+ than in PtC fractions further study (27). Fifty-three genes differed between PtC+ and − in spleen (33.3% ± 0.01 vs. 23.3% ± 0.01, P = 0.0009); and (iii) PtC spleen cells; these included regulators of B-cell proliferation diminished cycling that was indistinguishable between the PtC+ (TNFRSF13B, −1.7×; TSC22D3, −1.96×; TK1, 1.5×) and genes − and PtC fractions (4.7% ± 0.02 vs. 3.1% ± 0.01, P = 0.34), related to mitotic progression (CNOT3, 1.5×; ESPL1, 1.75×; compared with cells taken from the spleen. KIFC1, 1.5×). Four hundred fifty-one genes differed in PtC+ − − Next, PtC+ and PtC TCL1-192 populations isolated from versus PtC PC cells; those down-regulated were involved in BCR, fourth-transfer animals were injected retro-orbitally into SCID PI3K, ERK/MAPK, and NFκB signaling pathways (e.g., NFATC1/ mice. Five weeks after injection, recipients were killed and com- 2, −2×;PLCγ2, −1.9×; PRKCB, −1.8×; POU2F2, −1.7×; MAPK9, pared for disease extent. Mice receiving PtC+ cells from either PC −1.6×; RelA −1.9×), and up-regulated were anti-apoptotic (APIP,

E1502 | www.pnas.org/cgi/doi/10.1073/pnas.1300616110 Chen et al. Downloaded by guest on September 23, 2021 A B PNAS PLUS PtC- PtC+ CD5 B220 B220 PtC + - + - + - MFI of surface IgM PtC PtC PtC PtC PtC PtC PC SP C 5000 PC 8000 4000 SP

6000 *** 3000 ***

4000 2000

2000 1000 PC SP MFI of surface IgM 0 0 2nd 3rd 4th st nd rd th Donor 1 2 3 4 transfer transfer transfer transfer transfer transfer transfer MFI of surface PtC binding D Anti-IgM stimulation SP PtC+ SP PtC- PC PtC+ PC PtC- Indo (FL1/FL3) 2+ Intracellular Ca

Time (sec) MEDICAL SCIENCES

Fig. 2. Sorted PtC+ and PtC− fractions differ in smIgM expression and responses to BCR stimulation. (A) Example of gating strategy used for sorting of PtC+ − − (purple) and PtC (blue) fractions from B-1 or CLL cells. (B) Average smIgM expression in unselected, selected PtC+ and PtC cells collected from peritoneal cavity (PC) and spleen (SP) in fourth-transferred mice (n = 3). Data presented as mean ± SD. Statistical analysis was done by student t test using GraphPad Prism software, *P < 0.05. (C) Intensity of smIgM (Left) and PtC surface binding (Right) in serial transfer mice. Data shown are average of three mice (mean ± SD). Two-way ANOVA repeated measurement by GraphPad Prism was used to compare mean ﬂuorescence intensity (MFI) changes between SP and PC cells during − serial transfers, ***P < 0.001. (D)Ca2+ mobilization in response to anti-mouse IgM (μ-chain–speciﬁc) in splenic PtC+ (purple), PtC (blue) cells, and peritoneal PtC+ (red), PtC− (light blue) cells. Data shown are representative of triplicate experiments. Results are expressed as the ratio of FL1/FL3 against time.

1.5×; SGK3, 1.6×; TRIAP1, 1.8×) (Fig. 5, Left, and SI Appendix, splenic and peritoneal PtC+ cells contained some relating to cell Figs. S3, S4A, and Tables S2 and S3). These features are similar to survival and proliferation (AuRKA, 1.5×; BIRC5, 1.9×; CDC25C, anergic circulating CLL cells that are unresponsive to sIgM–cross- 1.4×). Thus, murine TCL1-192 PtC reactive cells exhibit prosurvival linking, constitutively expressing phosphorylated ERK, MEK, and and proliferation gene signatures. transactivated NF-AT (28). Results from the comparison of gene-expression differences Among the 53 and 451 genes identiﬁed in splenic and peritoneal between splenic and peritoneal PtC+ cells support active BCR, PtC binders versus nonbinders, only TK1 (1.5×) and SPIB (−1.7×) NFκB, and PI3K signaling in splenic but not PC PtC+ cells. In were found in both compartments. Therefore, a less stringent total, 272 genes were altered >twofold in splenic versus peritoneal − cutoff (1.3-fold, P < 0.05) was applied, identifying 101 genes of PtC+ but not PtC cells (SI Appendix,TableS5); those up-regulated which 33 and 68 genes changed concordantly and discordantly in genes in splenic PtC+ cells were involved in BCR and PI3K path- spleen and peritoneum (SI Appendix, Table S4 and Fig. S4B). The ways: for example, POU2F2 (3.1×), PLCγ2(2.9×), SHIP1/2 (2×), 68 discordant genes that were highly expressed in splenic PtC+ and PIK3CD(2×) (Fig. 5, Right). To validate the microarray cells but lower in peritoneal PtC+ cells included ones involved in analyses, genes involved in these pathways were tested by real- BCR, PI3K, and NFκB signaling. The 33 genes found in both time PCR and had the same pattern (SI Appendix, Fig. S3C).

Chen et al. PNAS | Published online April 2, 2013 | E1503 Downloaded by guest on September 23, 2021 A ** B 40 *** *** n.s. + 30 *** PtC 20

10 - PtC intensity SCID SCID

0 PtC SP PC SP PC SP PC + + % BrdU+ CLL cells % BrdU+ CLL PtC+ PtC- PtC+ PtC+ PtC- PtC- C Cells harvested from SP Cells harvested from PC Cells harvested from PB 1.5x109 3x107

1x109 2x107

5x108 1x107

SP PC SP PC SP PC SP PC SP PC SP PC PtC+ PtC- PtC+ PtC- PtC+ PtC-

Number of CLL cells Number of CLL Fractions transferred Fractions transferred Fractions transferred

Fig. 3. More aggressive CLL in mice adoptively transferred with enriched PtC+ cells. (A) In vivo analysis of proliferating cells in PtC+ and PtC− fractions in − fourth transferred mice (n = 5). CLL cells were collected from spleen (SP) and peritoneal cavity (PC) and then sorted for PtC+ and PtC fractions to analyze the − percentage of BrdU incorporation. (B)1× 106 PtC+ and PtC populations sorted from fourth transferred mice were injected into six mice, and mice were killed 5 wk later. Cells from peritoneal washout, spleen and blood were harvested and analyzed for the total number of CLL cells (shown in C). Bar graphs indicate mean values and SD. Statistical analysis was performed with unpaired t test using GraphPad Prism software. **P < 0.01, ***P < 0.001.

Because analyses of expression profiles of patient CLL cells of the BCR signaling pathway (PI3K and NFκB) (Figs. 4 and 5, from LN, BM, and blood suggest that cells residing in LNs are and SI Appendix, Fig. S4B). more proliferative than those in the other sites and the level of Normal murine B-1 cells use germ-line versions of IGHV11 or proliferation correlates with clinical course (29), we compared the IGHV12 to react with the autoantigen PtC (24). We found that human CLL and murine TCL1-192 genetic profiles. Interestingly, young, preleukemic TCL1 mice exhibited a polyclonal anti-PtC B-1 of the 128 genes changed in human CLL LN versus periphery (PB cell repertoire and that TCL1 mice with advanced disease had or BM), a total of 24 and 4 were changed concordantly and dis- oligo- or monoclonal expansions of such B cells. PtC binding cordantly in TCL1-192 SP versus PC genes, suggesting at least depended on smIgM (SI Appendix,Fig.S1D), and cells with higher a 20% genetic signature overlap in lymphoid tissues versus pe- levels of smIgM bound PtC better (Fig. 2). As with TCL1-192 (Figs. riphery (SI Appendix, Table S6). 3– 5), smIgM-antigen engagement leads to activation of signaling Taken together, the genetic signatures support that PtC+ pop- molecules affecting tumor cell behavior, and smIgM expression ulations residing in spleen and peritoneum exhibit divergent levels influence BCR signaling (34). Of note, the smIgM of U-CLL phenotypic, functional, and signaling properties: the former rep- patients, who generally have poor outcomes, is less down-modu- resenting cells with active BCR-, PI3K-, and NFκB-signaling lated upon antigen binding and is more effective at signal trans- pathways, and the latter cells hyporesponsive to BCR engagement, duction than for M-CLL patients (35, 36). similar to normal B-1 lymphocytes (26, 30, 31). These biologic and Increased antigen binding can result from the development molecular observations may apply to CLL patients whose disease and selection of replacement mutations in IGVs, as can occur for progression associates with unique BCR structures, antigen-re- M-CLL patients whose clones use a stereotyped BCR (37). These activity, and signaling. cases can exhibit intraclonal diversification and selection of specific changes in IGLVκ (37), suggesting clonal evolution in human Discussion CLL may be partly driven by ongoing interactions with (auto)an- Antigen binding and selection may be involved in the clonal evo- tigen (38). However, we did not observe a change in BCR structure lution that leads to divergent clinical outcomes of patients with of TCL1-192 cells over time. This paradox may be because of their leukemic cells differing in BCR structure (32, 33). Here, we used deriving from B-1 cells that rarely develop IGV mutations (23). TCL1 mice to address this possibility in vivo, hoping to identify Moreover, TCL1 leukemias resemble U-CLL patient clones (14) therapeutic targets to attenuate or cure the disease. By serially that rarely accumulate IGHV mutations (6), which emerge as transferring into SCID mice splenocytes from a single TCL1 major components of the clone (39). Thus, the observed increased mouse that spontaneously developed oligoclonal B-cells express- reactivity with PtC over time suggests antigen selection and drive ing IGHV genes involved in PtC binding, we observed natural can lead to a more hostile leukemia by mechanisms other than selection for PtC-binding leukemic cells (Fig. 1). This process IGV diversification and selection. In addition, it is important to established an aggressive, transferrable, antigen-specific cell line highlight that not all tumor cells of the TCL1-192 clone bound PtC, (TCL1-192) that permitted comparisons of cells not binding PtC even after several transfers (Fig. 1). Although this finding could be with those binding PtC and also cells residing in different anatomic because of differences in cell-cycle stage among clonal members, it sites (in or outside of primary and secondary lymphoid tissues). By could also reflect signaling mediated by other BCR–cross-reactive comparing trafficking, triggering, and proliferation, we demon- antigens or selection for an alternative type of BCR-signaling strated a direct relationship between increased PtC binding and mediated by autoantigens in adjacent smIgs, as was recently more robust BCR signaling, in part because of higher smIgM identified in human CLL (11). density, with disease virulence (Figs. 2 and 3, and SI Appendix,Fig. TCL1-192 cells proliferated robustly in spleen and LNs but not S2). Notably, TCL1-192 cells residing in secondary lymphoid tissue in the PC, BM, and blood (SI Appendix,Fig.S2), consistent with (spleen) differed from those at other sites in surface phenotype, normal splenic B-1 cells having a higher turnover rate (24, 26). BCR-initiated responses, activated signaling molecules and effectors Furthermore, despite their increased smIgM density and PtC

E1504 | www.pnas.org/cgi/doi/10.1073/pnas.1300616110 Chen et al. Downloaded by guest on September 23, 2021 A PNAS PLUS *P=0.02 30000 SP PC 20000

10000 MFI IgM

0 Surface IgM Peritoneum Spleen B

SP PC 2 300 PC 250 PV 2min 200 SP PV 1min 150 H2O2 2min 100 Untreated 50 MFI p-PLC 0 Alone 1min 2min p-PLC 2 250 200 PC 150 SP 100 50 MFI p-SYK 0 Alone 1min 2min p-SYK

C PtC liposome stimulation D Anti-IgM stimulation SP SP PC Indo BM (FL1/FL3) (FL1/FL3)

2+ PC 2+ Indo SP SP in medium SP in PtC PC SP in PL Intracellular Ca Indo Intracellular Ca Indo

Fig. 4. Peritoneal and splenic TCL1-192 CLL cells differ in smIgM density and response to smIgM cross-linking. (A) Intensity of smIgM in TCL1-192 cells collected from spleen (SP, black) and peritoneal cavity (PC, red). Mean values and SD are indicated with ﬁlled circles and error bars. Statistical analysis was done by unpaired t test using GraphPad Prism software. (B) Constitutively phosphorylated PLCγ2 and SYK were investigated in splenic and peritoneal TCL1-192 cells

by exposing cells to pervanadate (in H2O2) for 1 or 2 min. Untreated cells or cells incubated in H2O2 alone for 2 min were used as control. Fixed and per- meablized cells were then stained and analyzed by flow cytometry. (C)Ca2+ mobilization in CLL cells stimulated with 1:80,000 (Upper) or 1:40,000 (Lower) diluted PtC-bearing liposomes. Results are expressed as the ratio of FL1/FL3 against time. (D, Upper)Ca2+ mobilization in response to anti-mouse IgM (μ-chain– fi speci c) was assayed in CLL SP, BM, and PC cells. (Lower) TCL1-192 CLL cells collected from spleen were precultured in control medium (green), medium with PtC MEDICAL SCIENCES liposome (blue), or the supernatant of peritoneal lavage (PL) harvested from TCL1-192 mice (brown). Twenty-four hours later cells were labeled with Indo-1 for Ca2+ mobilization in response to anti-mouse IgM (μ-chain–specific) stimulation. The data shown in B–D represent examples of triplicate experiments.

binding, TCL1-192 cells residing in the peritoneum and BM, peritoneal B-1 cells (24, 31, 43) and some CLLs (28, 36), con- exhibited signaling properties similar to hyporesponsive B-1 cells tinuous BCR occupancy may lead to impaired smIgM signaling, that survive longer in vitro (31, 40). These findings are reminiscent elevated intracellular Ca2+ levels, sustained phosphorylation of of human CL cells that proliferate more actively in LNs and less so PLCγ2, SYK, ERK1/2, MEK1/2, activated NFAT proteins, and in BM and blood, as demonstrated by Ki-67 expression (29), weakened BCR and NFkB signaling. Here we found peritoneal deuterium incorporation into newly synthesized DNA in vivo (41, and BM CLL cells, especially PtC-reactive ones, were hypores- 42), and differential expression of activation-associated genes in ponsive to smIg stimulation and PC cells expressed molecular LNs, BM, and blood (29). Thus, both mouse and human CLL cells signatures similar to hyporesponsive B-1 cells (Figs. 4 and 5, and divide more actively in secondary lymphoid tissues [spleen and SI Appendix, Fig. S4B). LNs for TCL1-192 (SI Appendix,Fig.S2), and LNs for human CLL The importance of local microenvironmental factors, either (29, 42)] and survive better in primary lymphoid areas (PC and BM soluble or cellular, was highlighted by finding that unresponsive for TCL1-192, and BM for human CLL), consistent with the idea peritoneal TCL1-192 cells regained antigen responsiveness when that BCR signaling is more supportive of the proliferation of CLL transferred to new recipients, if they migrated to secondary lym- cells located in secondary LNs than elsewhere (BM, blood). phoid tissue (Fig. 3). Conversely, active splenic TCL1-192 CLL Our findings also suggest local influences controlling site- cells were unable to respond to antigenic stimulation if precultured specific functional differences are at least in part soluble (Fig. 4), with peritoneal washout or after migration to the PC (Fig. 4), possibly antigen or cytokines differing in amount or type at the consistent with gene-expression findings of suppressed BCR- − two locations. In this regard, it is noteworthy that human CLL signaling in peritoneal but not splenic PtC+ vs. PtC cells (Fig. 5 and cells may be tolerized by an apparently constant antigen receptor SI Appendix,Figs.S3andS4). For example, genes involved in BCR occupancy that is negated after incubation in vitro (43). In mouse signaling, such as PLCγ2, a downstream target regulating B-cell

Chen et al. PNAS | Published online April 2, 2013 | E1505 Downloaded by guest on September 23, 2021 Fig. 5. Networks of BCR-regulated genes in PtC+ and PtC− cells from spleens and peritoneal cavity. Ingenuity pathways analysis (IPA) software was used to analyze identiﬁed genes involved in the BCR signaling pathway that were differently expressed in peritoneal PtC+ and PtC− cells (Left, n = 451, >1.5×), and PtC+ cells isolated from spleen and peritoneum (Right, n = 272, >2×). Genes labeled in red and green were those identiﬁed as up- and down-regulated and other genes were those related to the regulated genes on the basis of the network analysis.

migration and growth, was more highly expressed in young TCL1 Flow Cytometric Analyses. Cellular PtC binding was determined by incubation B-1 and TCL1-192 splenic CLL cells (1.4–1.6×) than WT B-1 cells, with PE anti-CD45R/B220, APC anti-CD5 (BD Biosciences), and FITC-conjugated − PtCliposomes, a kind gift from StephenH. Clarke(DukeUniversity, Durham, NC) but was not altered in TCL1-192 peritoneal PtC cells and even fl + − × (21). For intracellular phospho- ow cytometry analysis, cells were exposed or suppressed in peritoneal PtC cells ( 1.8 ). Proliferation of CLL μ fi + not to 25 M pervanadate, xed with 16% (wt/vol) paraformaldehyde, per- cells in LNs appears to be enhanced by interactions with CD40L meabilized with ice-cold methanol, and exposed to Alexa647-conjugated + CD4 T cells (44, 45) and the LN microenvironment appears to p-SYK or pY-PLCγ2 antibodies (BD Biosciences). favor BCR and NFκB signaling in CLL (29). It is noteworthy that the majority of genes differing between WT or young TCL1 B-1 Ca2+ Flux Analysis. Cells were incubated in dye-free RPMI1640 supplemented PtC+ and PtC+ TCL1-192 cells suggest antigen-dependent BCR with 5% (vol/vol) FBS and 1 μM fluorescent Ca2+ indicator Indo-1, acetox- × 6 × 6 stimulation in CLL development (SI Appendix,Fig.S3). In addi- ymethyl ester (Invitrogen) at the density of 5 10 cells/mL. Cells (1 10 cells per tube) were run for 30 s on LSRII to obtain a baseline, then stimu- tion, there is considerable similarity between human CLL and μ ′ fi lated for 5 min by 1 gF(ab)2 fragments of anti-IgM pAbs, PtC-bearing TCL1-192 genetic pro les in lymphoid tissue versus periphery (SI liposomes, or ionomycin. Kinetic data were analyzed with FlowJo or Cell Appendix,TableS6). Quest (BD Biosciences). Finally, because of its enhanced aggressiveness, this murine CLL-like cell line provides a shorter time-frame platform to study IGHV-D-J DNA Sequencing, Gene-Expression Profiling, and Data Analysis. cDNA clonal evolution of genetic alterations in vivo. In addition, because sequencing was performed as described previously (6). Sequences were of the defined BCR antigenic specificity, the influences of antigen compared with germ-line genes in the IMGT and GenBank databases. Nu- binding and signaling on clonal progression are evaluable. Our cleotide and amino acid sequences of V region segments from all major and minor clones were compared with each other and to those contained in the findings that autoantigen–BCR interactions lead to clinical pro- above databases using Ig-BLAST and IMGT/V-QUEST. gression support the use of BCR-pathway inhibitors in human CLL For microarray analysis, combined spleen and PC B-1 cells from 15 3-mo-old – (22, 46 48). WT mice, five 4-mo-old TCL1 mice, and CLL cells from two individual second- − transfer TCL1-192 mice were sorted into PtC+ and PtC populations. RNA Materials and Methods extracted from these samples was hybridized to Sentrix Mouse WG-6 Bead- Mice. (C57BL6×C3H)F1 WT mice, TCL1 transgenic mice (12) and CB17 SCID Chip (Illumina). Data were normalized using quantile normalization by mice were housed under conventional barrier protection. Animal procedures BeadStudio software (Illumina). Ingenuity Pathway Analysis and GeneCards were performed in accordance with The Feinstein Institute’s Institutional Batch Queries were applied to analyze the significant changed genes. Heat- Animal Care and Use Committee requirements. Mechanistic studies were maps of gene-expression values in each comparison were generated using carried out on cells frozen after the fourth transfer. heatmap.2 in the g-plots package of R (www.Rproject.org).

E1506 | www.pnas.org/cgi/doi/10.1073/pnas.1300616110 Chen et al. Downloaded by guest on September 23, 2021 Real-time quantitative PCR assay using SYBR Green/ROX was performed in ACKNOWLEDGMENTS. We thank Dr. Adrian Wiestner, National Heart, Lung, PNAS PLUS triplicates. Data were analyzed with the comparative CT (ΔΔCT, where CT is and Blood Institute/National Institutes of Health, for sharing his human ﬁ threshold cycle) method. Differences in CT were used to quantify the relative gene-expression pro ling data. This study was supported by the National Cancer Institute (R01 CA81554) and by The Karches Foundation, The Prince amount of PCR target contained within each well. Data were expressed as rel- Family Foundation, The Nash Family Foundation, The Mildred and Frank ative mRNA expression in reference to control, normalized to quantity of RNA Feinberg Foundation, The Marks Foundation, The Jerome Levy Foundation, input by performing measurements on an endogenous reference gene, GAPDH. The Leon Levy Foundation, and the Joseph Eletto Leukemia Research Fund.

1. Chiorazzi N, Ferrarini M (2003) B cell chronic lymphocytic leukemia: Lessons learned 25. Chumley MJ, et al. (2000) A VH11V kappa 9 B cell antigen receptor drives generation from studies of the B cell antigen receptor. Annu Rev Immunol 21:841–894. of CD5+ B cells both in vivo and in vitro. J Immunol 164(9):4586–4593. 2. Chiorazzi N, Rai KR, Ferrarini M (2005) Chronic lymphocytic leukemia. N Engl J Med 26. Morris DL, Rothstein TL (1993) Abnormal transcription factor induction through the 352(8):804–815. surface immunoglobulin M receptor of B-1 lymphocytes. J Exp Med 177(3):857–861. 3. Stevenson FK, Caligaris-Cappio F (2004) Chronic lymphocytic leukemia: Revelations 27. Canales RD, et al. (2006) Evaluation of DNA microarray results with quantitative gene from the B-cell receptor. Blood 103(12):4389–4395. expression platforms. Nat Biotechnol 24(9):1115–1122. 4. Messmer BT, et al. (2004) Multiple distinct sets of stereotyped antigen receptors 28. Muzio M, et al. (2008) Constitutive activation of distinct BCR-signaling pathways in indicate a role for antigen in promoting chronic lymphocytic leukemia. JExpMed a subset of CLL patients: A molecular signature of anergy. Blood 112(1):188–195. 200(4):519–525. 29. Herishanu Y, et al. (2011) The lymph node microenvironment promotes B-cell receptor 5. Tobin G, et al. (2004) Subsets with restricted immunoglobulin gene rearrangement signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic features indicate a role for antigen selection in the development of chronic leukemia. Blood 117(2):563–574. lymphocytic leukemia. Blood 104(9):2879–2885. 30. Morris DL, Rothstein TL (1994) Decreased surface IgM receptor-mediated activation of 6. Fais F, et al. (1998) Chronic lymphocytic leukemia B cells express restricted sets of phospholipase C gamma 2 in B-1 lymphocytes. Int Immunol 6(7):1011–1016. mutated and unmutated antigen receptors. J Clin Invest 102(8):1515–1525. 31. Wong SC, et al. (2002) Peritoneal CD5+ B-1 cells have signaling properties similar to 7. Stamatopoulos K, et al. (2007) Over 20% of patients with chronic lymphocytic leukemia tolerant B cells. J Biol Chem 277(34):30707–30715. carry stereotyped receptors: Pathogenetic implications and clinical correlations. Blood 32. Stilgenbauer S, et al. (2007) Clonal evolution in chronic lymphocytic leukemia: 109(1):259–270. Acquisition of high-risk genomic aberrations associated with unmutated VH, resistance 8. Hervé M, et al. (2005) Unmutated and mutated chronic lymphocytic leukemias derive to therapy, and short survival. Haematologica 92(9):1242–1245. from self-reactive B cell precursors despite expressing different antibody reactivity. J 33. Shanafelt TD, et al. (2008) Karyotype evolution on fluorescent in situ hybridization Clin Invest 115(6):1636–1643. analysis is associated with short survival in patients with chronic lymphocytic leukemia 9. Damle RN, et al. (1999) Ig V gene mutation status and CD38 expression as novel and is related to CD49d expression. J Clin Oncol 26(14):e5–e6. prognostic indicators in chronic lymphocytic leukemia. Blood 94(6):1840–1847. 34. Efremov DG, Gobessi S, Longo PG (2007) Signaling pathways activated by antigen- 10. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK (1999) Unmutated Ig V(H) receptor engagement in chronic lymphocytic leukemia B-cells. Autoimmun Rev 7(2): genes are associated with a more aggressive form of chronic lymphocytic leukemia. 102–108. Blood 94(6):1848–1854. 35. Chen L, et al. (2002) Expression of ZAP-70 is associated with increased B-cell receptor 11. Dühren-von Minden M, et al. (2012) Chronic lymphocytic leukaemia is driven by signaling in chronic lymphocytic leukemia. Blood 100(13):4609–4614. antigen-independent cell-autonomous signalling. Nature 489(7415):309–312. 36. Mockridge CI, et al. (2007) Reversible anergy of sIgM-mediated signaling in the two 12. Bichi R, et al. (2002) Human chronic lymphocytic leukemia modeled in mouse by subsets of CLL defined by VH-gene mutational status. Blood 109(10):4424–4431. targeted TCL1 expression. Proc Natl Acad Sci USA 99(10):6955–6960. 37. Kostareli E, et al. (2010) Intraclonal diversification of immunoglobulin light chains in 13. Narducci MG, et al. (2000) Regulation of TCL1 expression in B- and T-cell lymphomas a subset of chronic lymphocytic leukemia alludes to antigen-driven clonal evolution. and reactive lymphoid tissues. Cancer Res 60(8):2095–2100. Leukemia 24(7):1317–1324. 14. Yan XJ, et al. (2006) B cell receptors in TCL1 transgenic mice resemble those of 38. Ghia P, Scielzo C, Frenquelli M, Muzio M, Caligaris-Cappio F (2007) From normal to aggressive, treatment-resistant human chronic lymphocytic leukemia. Proc Natl Acad clonal B cells: Chronic lymphocytic leukemia (CLL) at the crossroad between neoplasia Sci USA 103(31):11713–11718. and autoimmunity. Autoimmun Rev 7(2):127–131. 15. Lanemo Myhrinder A, et al. (2008) A new perspective: Molecular motifs on oxidized 39. Gurrieri C, et al. (2002) Chronic lymphocytic leukemia B cells can undergo somatic LDL, apoptotic cells, and bacteria are targets for chronic lymphocytic leukemia hypermutation and intraclonal immunoglobulin V(H)DJ(H) gene diversification. J Exp antibodies. Blood 111(7):3838–3848. Med 196(5):629–639. 16. Catera R, et al. (2008) Chronic lymphocytic leukemia cells recognize conserved 40. Tumang JR, Hastings WD, Bai C, Rothstein TL (2004) Peritoneal and splenic B-1 cells epitopes associated with apoptosis and oxidation. Mol Med 14(11-12):665–674. are separable by phenotypic, functional, and transcriptomic characteristics. Eur J 17. Mercolino TJ, Arnold LW, Hawkins LA, Haughton G (1988) Normal mouse peritoneum Immunol 34(8):2158–2167. contains a large population of Ly-1+ (CD5) B cells that recognize phosphatidyl choline. 41. Calissano C, et al. (2009) In vivo intraclonal and interclonal kinetic heterogeneity in Relationship to cells that secrete hemolytic antibody specific for autologous B-cell chronic lymphocytic leukemia. Blood 114(23):4832–4842. erythrocytes. J Exp Med 168(2):687–698. 42. van Gent R, et al. (2008) In vivo dynamics of stable chronic lymphocytic leukemia 18. Cabiedes J, et al. (2002) Characterization of anti-phosphatidylcholine polyreactive inversely correlate with somatic hypermutation levels and suggest no major leukemic natural autoantibodies from normal human subjects. J Autoimmun 18(2):181–190. turnover in bone marrow. Cancer Res 68(24):10137–10144.

19. Bhat NM, Bieber MM, Spellerberg MB, Stevenson FK, Teng NN (2000) Recognition of 43. Gauld SB, Benschop RJ, Merrell KT, Cambier JC (2005) Maintenance of B cell anergy MEDICAL SCIENCES auto- and exoantigens by V4-34 gene encoded antibodies. Scand J Immunol 51(2): requires constant antigen receptor occupancy and signaling. Nat Immunol 6(11): 134–140. 1160–1167. 20. Guchhait P, López JA, Thiagarajan P (2004) Characterization of autoantibodies 44. Hayden RE, Pratt G, Roberts C, Drayson MT, Bunce CM (2012) Treatment of chronic against sulfatide from a V-gene phage-display library derived from patients with lymphocytic leukemia requires targeting of the protective lymph node environment systemic lupus erythematosus. J Immunol Methods 295(1-2):129–137. with novel therapeutic approaches. Leuk Lymphoma 53(4):537–549. 21. Arnold LW, McCray SK, Tatu C, Clarke SH (2000) Identification of a precursor to 45. Patten PE, et al. (2012) IGHV-unmutated and IGHV-mutated chronic lymphocytic phosphatidyl choline-specific B-1 cells suggesting that B-1 cells differentiate from leukemia cells produce activation-induced deaminase protein with a full range of splenic conventional B cells in vivo: Cyclosporin A blocks differentiation to B-1. J biologic functions. Blood 120(24):4802–4811. Immunol 164(6):2924–2930. 46. Hoellenriegel J, et al. (2012) Selective, novel spleen tyrosine kinase (Syk) inhibitors 22. Suljagic M, et al. (2010) The Syk inhibitor fostamatinib disodium (R788) inhibits tumor suppress chronic lymphocytic leukemia B-cell activation and migration. Leukemia growth in the Eμ- TCL1 transgenic mouse model of CLL by blocking antigen- 26(7):1576–1583. dependent B-cell receptor signaling. Blood 116(23):4894–4905. 47. Hoellenriegel J, et al. (2011) The phosphoinositide 3′-kinase delta inhibitor, CAL-101, 23. Casali P, Schettino EW (1996) Structure and function of natural antibodies. Curr Top inhibits B-cell receptor signaling and chemokine networks in chronic lymphocytic Microbiol Immunol 210:167–179. leukemia. Blood 118(13):3603–3612. 24. Chumley MJ, Dal Porto JM, Cambier JC (2002) The unique antigen receptor signaling 48. Ponader S, et al. (2012) The Bruton tyrosine kinase inhibitor PCI-32765 thwarts chronic phenotype of B-1 cells is influenced by locale but induced by antigen. J Immunol lymphocytic leukemia cell survival andtissuehominginvitroandinvivo.Blood 169(4):1735–1743. 119(5):1182–1189.

Chen et al. PNAS | Published online April 2, 2013 | E1507 Downloaded by guest on September 23, 2021