Survival of human lymphoma cells requires B-cellreceptor engagement by self-antigensRyan M. Younga, Tianyi Wua, Roland Schmitza, Moez Dawooda, Wenming Xiaoa,1, James D. Phelana, Weihong Xua,Laurence Menardb, Eric Meffreb, Wing-Chung C. Chanc, Elaine S. Jaffed, Randy D. Gascoynee, Elías Campof,Andreas Rosenwaldg, German Otth, Jan Delabiei, Lisa M. Rimszaj, and Louis M. Staudta,2

aLymphoid Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892; bDepartment ofImmunobiology, Yale University School of Medicine, New Haven, CT 06511; cDepartment of Pathology, City of Hope National Medical Center, Duarte, CA91010; dLaboratory of Pathology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892; eCentre forLymphoid Cancer, British Columbia Cancer Agency, Vancouver, BC, Canada V5Z 4E6; fHematopathology Section, Department of Pathology, Hospital Clinic,University of Barcelona, 08036 Barcelona, Spain; gDepartment of Pathology, University of Würzburg, 97080 Würzburg, Germany; hDepartment of ClinicalPathology, Robert-Bosch-Krankenhaus and Dr. Margarete Fischer-Bosch Institute for Clinical Pharmacology, 70376 Stuttgart, Germany; iDepartment ofPathology, Oslo University Hospital, 0310 Oslo, Norway; and jDepartment of Pathology, University of Arizona, Tucson, AZ 85724

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2013.

Contributed by Louis M. Staudt, August 31, 2015 (sent for review July 28, 2015; reviewed by Christopher C. Goodnow and Klaus Rajewsky)

The activated B-cell–like (ABC) subtype of diffuse large B-cell lym-phoma (DLBCL) relies on chronic active B-cell receptor (BCR) signal-ing. BCR pathway inhibitors induce remissions in a subset of ABCDLBCL patients. BCR microclusters on the surface of ABC cells re-semble those generated following antigen engagement of normalB cells. We speculated that binding of lymphoma BCRs to self-anti-gens initiates and maintains chronic active BCR signaling in ABCDLBCL. To assess whether antigenic engagement of the BCR is re-quired for the ongoing survival of ABC cells, we developed isogenicABC cells that differed solely with respect to the IgH V region of theirBCRs. In competitive assays with wild-type cells, substitution of aheterologous V region impaired the survival of three ABC lines.The viability of one VH4-34

+ ABC line and the ability of its BCRto bind to its own cell surface depended on V region residues thatmediate the intrinsic autoreactivity of VH4-34 to self-glycoproteins.The BCR of another ABC line reacted with self-antigens in apoptoticdebris, and the survival of a third ABC line was sustained by reactiv-ity of its BCR to an idiotypic epitope in its own V region. Hence, adiverse set of self-antigens is responsible for maintaining the malig-nant survival of ABC DLBCL cells. IgH V regions used by the BCRs ofABC DLBCL biopsy samples varied in their ability to sustain survivalof these ABC lines, suggesting a screening procedure to identifypatients who might benefit from BCR pathway inhibition.

lymphoma | cancer biology | B-cell receptor

Theory and indirect evidence have supported the notion of an-tigenic stimulation in the pathogenesis of human B-cell lym-

phomas for the past half century (1). Human B-cell lymphomasselectively retain expression of the B-cell receptor (BCR) even inthe face of frequent chromosomal translocations that disrupt the Igheavy chain (IgH) locus, suggesting that the signaling pathwaysemanating from the BCR may sustain the survival of malignant Bcells (2). Foreign antigens have been implicated in certain lym-phomas, including the hepatitis C virus in splenic marginal zonelymphoma (3) and Helicobacter pylori in mucosa-associated lym-phoid tissue-type lymphomas (4). However, no discernible foreignantigen is present in the majority of lymphoma cases, suggesting apossible role for self-antigens in lymphoma etiology.Examination of the Ig variable (V) regions has lent further

support to the concept of antigen-dependent BCR signaling inlymphoid malignancies. In chronic lymphocytic leukemia, forexample, a small subset of germ-line–encoded IgH variable gene(VH) segments are rearranged recurrently (5). The same observa-tion has been made in mantle cell lymphoma (MCL), although therecurrent VH segments in these lymphomas are not fully concor-dant with those in chronic lymphocytic leukemia (CLL) (6). Nearlyone-third of CLL cases use “stereotyped” BCR sequences in which

malignant cells from different patients have almost identical IgH Vsequences, including the third complementarity determining region(CDR3) that is diversified during VH-DH-JH (VDJ) joining (5).This observation suggests that CLL BCRs may bind to an antigensbecause CDR regions typically dictate antibody reactivity. Indeed,CLL BCRs can react with many different self-antigens (7), in-cluding antigens released by apoptotic cells (8, 9). Additionally,BCRs derived from CLL patients can bind to a conserved epitopewithin the second framework region (FR2) of their own IgVH (10).Because a large component of the germ-line IgVH repertoire canform antibodies with self-reactivity (11), these findings mightmerely reflect the derivation of malignant B cells from self-reactiveB cells. An alternative, nonmutually exclusive hypothesis is that aself-reactive BCR is essential to maintain the malignant phenotypein an ongoing fashion. This hypothesis has not yet been testedbecause of the absence of an appropriate model system.Chronic active BCR signaling drives NF-κB signaling in cell line

models of the activated B-cell–like (ABC) subtype of diffuse largeB-cell lymphoma (DLBCL), which is essential for their viability(12). Unlike “tonic” BCR signaling (13), which is presumed tobe antigen-independent, chronic active BCR signaling in ABC

Significance

Signals emanating from the B-cell receptor (BCR) promote survivalof malignant B cells, such as the activated B-cell–like subtype ofdiffuse large B-cell lymphoma (DLBCL). Although there is evidencethat many lymphoid cancers arise from antigen selected B cells, itis not known whether these malignant cells require ongoing an-tigenic engagement of the BCR for their survival. In activatedB-cell–like DLBCL, we identified key determinants of self-antigenrecognition by the BCR that are required to augment the relativeviability of these malignant cells. Our findings imply that similarmechanisms may contribute to other B-cell malignancies andsuggest precision medicine strategies to identify patients whosecancers may respond to BCR pathway inhibitors, such as ibrutinib.

Author contributions: R.M.Y., R.S., and L.M.S. designed research; R.M.Y., T.W., and R.S.performed research; R.M.Y., T.W., M.D., W. Xiao, W. Xu, L.M., E.M., W.-C.C.C., E.S.J., R.D.G.,E.C., A.R., G.O., J.D., L.M.R., and L.M.S. contributed new reagents/analytic tools; R.M.Y., T.W.,R.S., M.D., W. Xiao, J.D.P., and L.M.S. analyzed data; and R.M.Y. and L.M.S. wrote the paper.

Reviewers: C.C.G., The Garvan Institute of Medical Research; and K.R., Max-DelbruckCenter for Molecular Medicine.

The authors declare no conflict of interest.1Present address: Division of Bioinformatics and Biostatistics, National Center for Toxicolog-ical Research, Food and Drug Administration, Jefferson, AR 72079.

2To whom correspondence should be addressed. Email: lstaudt@mail.nih.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1514944112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1514944112 PNAS | November 3, 2015 | vol. 112 | no. 44 | 13447–13454

MED

ICALSC

IENCE

SINAUGURA

LART

ICLE

DLBCL has the hallmarks of antigen-dependent BCR signaling innormal B cells (14), including prominent clustering of the BCR onthe cell surface (12). Moreover, ∼20% of DLBCL patients havegain-of-function mutations affecting the immunoreceptor tyro-sine-based activation motif (ITAM) signaling motifs of the BCRsubunits CD79A and CD79B, providing genetic evidence thatBCR signaling is integral to the pathogenesis of ABC DLBCL(12). Although these mutations increase the amplitude of BCRsignaling, they cannot initiate BCR signaling de novo (12), leadingus to consider a role for antigen in initiating and maintainingchronic active BCR signaling in ABC DLBCL.This possibility was supported by a clinical trial in relapsed/

refractory DLBCL of ibrutinib, an inhibitor of Bruton’s tyro-sine kinase, which links BCR activity to the NF-κB pathway(15). In ABC DLBCL, ibrutinib produced responses in 37% ofpatients, lengthening their survival. Although ABCDLBCL tumorswith CD79A or CD79B mutations responded more frequently toibrutinib, responses were also observed in 30% of cases withoutthese mutations, suggesting that the BCR activity of these tumorsmay depend on a nongenetic process, such as self-antigen en-gagement of the BCR (15). Moreover, ibrutinib has also provedeffective in other B-cell malignancies, such as CLL (16), in whichno genetic mechanisms of BCR activity have been reported. In thepresent study, we sought to provide experimental evidence that theIgVH regions of ABC DLBCL BCRs are required for their survivaland to elucidate the role of self-antigens in this process.

ResultsRestricted IgVH Repertoire in ABC DLBCL. We first investigated thenature of the rearranged IgVH segments in ABC DLBCL tumorsand compared them to those in the other prominent DLBCLsubtype, germinal center B-cell–like (GCB) DLBCL, and innormal human B cells. IgVH sequences in DLBCL tumors wereassembled from high throughput RNA sequencing data, andclassified by ImMunoGeneTics (IMGT)/High V-quest (www.imgt.org/HighV-QUEST) (Dataset S1). Use of IgVH gene seg-ments in normal human B cells was taken from a previous report(17). Remarkably, among 281 ABC DLBCL samples, over 30%used a single IgVH segment, VH4-34. In contrast, the expressionof this IgVH segment was significantly less prevalent among 117GCB DLBCL samples (9%; P < 0.0001), and normal B cells(4%; P = 0.0014) (Fig. 1A). Previous studies also found an en-richment for VH4-34 in limited sets of DLBCL (18) or non-GCBDLBCL cases (19, 20). Besides VH4-34, the only IgVH that wasmore prevalent in ABC DLBCL than in GCB DLBCL andnormal B cells was VH3-7, which was expressed in 7.8% of ABCDLBCL cases (Fig. 1A). Together, the two ABC-enriched IgVHsegments, VH4-34 and VH3-7, constitute ∼39% of all cases,demonstrating that these tumors have a remarkably restrictedIgVH repertoire and suggesting shared antigenic reactivity. Ofnote, VH4-34 and VH3-7 are among the IgVH segments that areoverrepresented in CLL (5), suggesting that ABC DLBCL andCLL may derive from B cells with similar specificity. In contrast,GCB DLBCL samples were only modestly enriched for threeIgVH segments (VH3-30, VH3-48, and VH4-59) compared withABC DLBCL and normal B cells, consistent with the fact thatthis subtype does not typically engage NF-κB through chronicactive BCR signaling (12), and seldom responds to ibrutinib (15).The IgVH regions in GCB and ABC DLBCL tumors have typ-

ically undergone somatic hypermutation (21), in accord with theirputative derivation from germinal center and postgerminal centerB cells, respectively (22). In the present set of ABC and GCB tu-mors, somatic hypermutation altered 8.7% and 9.8% of the IgVHbases, respectively (P = 0.084) (Fig. 1B). However, a substantialnumber of ABC cases (28 of 281; 10%) had IgVH sequences thatdiverged by <2% from the nearest germ-line IgVH sequence, whichwe defined as “nonmutated.”Among these nonmutated ABC casesthere was an overutilization of the VH1-69 segment (5 of 16;

31.2%) compared with the remainder of ABC DLBCLs (24 of 265;9.1%). Of note, VH1-69 is frequently used in the Ig-unmutatedform of CLL (5), perhaps indicating a shared origin and BCRreactivity of CLLs and ABC DLBCLs with VH1-69.The enzyme activation-induced cytidine deaminase that medi-

ates somatic hypermutation is also responsible for switch recom-bination from IgM to one of the other IgH classes in germinalcenter B cells (23). Despite their putative origin from postgerminalcenter B cells, 80% of ABC DLBCL tumors expressed IgM com-pared with 34% of the GCBDLBCL tumors (Fig. 1C). This findingis consistent with previous work demonstrating that ABC but notGCB DLBCL tumors have genetic lesions that disrupt the switch μregion of the IgH locus that is necessary for Ig class-switch re-combination (19, 24). Of note, the frequency of IgM BCRs amongVH4-34

+ ABC samples was nearly 90%, which was higher than inVH3-7

+ samples (77%) and in all other ABC samples (75%) (Fig.1D). In normal B cells, IgM and IgG BCRs deliver qualitativelydifferent signals, with IgG-BCR signaling promoting terminal plas-macytic differentiation (25–27). Thus, utilization of the IgM constantregion in ABC DLBCL tumors, especially those with VH4-34

+

BCRs, may be necessary to block terminal differentiation and favorpathological proliferation.

ABC DLBCL Viability Depends on IgH Variable Region Specificity.Previous results demonstrated that most ABC lines are depen-dent on BCR signaling for survival, but the origins of this signalinghave been unclear (12). To test the possibility that these lines usethe BCR to interact with self-antigens, we developed a knockdown-rescue strategy to create isogenic cell populations with BCRs thatdiffer only in their IgH V regions (Fig. 2A). To do this, ABC lineswere concurrently transduced with two vectors, one that coex-presses a doxycycline-inducible small hairpin RNA (shRNA) tar-geting human IgM (shIgM) along with GFP, and a “rescue” vectorthat coexpresses a mouse IgM constant region fused to a humanIgH V sequence along with an LYT2 (CD8a) surface marker.

DCB

A

Normal B cells (27)

GCB DLBCL (117)

ABC DLBCL (281)35

30

25

20

15

3-74-34IgVH: 3-23 1-69 4-393-30 3-33 4-59 3-48 3-74 5-51 3-21 1-81-2

10

5

0

% Ig

VH

mut

atio

n 70

80

90

100

60

50

40

30

20

0

10

ABC(263)

GCB(102)

IgM

IgG

IgA

IgE

VH3-7(22)

VH4-34(76)

VH-others(165)

% A

BC

cas

es

70

80

90

100

60

50

40

30

20

0

10

0

10

20

30

ABC(281)

GCB(117)

% to

tal c

ases

% to

tal c

ases

Fig. 1. IgH analysis of DLBCL tumors. (A) The IgVH repertoire in ABC andGCB DLBCL samples compared with normal naïve B cells. (B) Percentage ofIgVH bases mutated in ABC and GCB tumors. (C) IgH constant region isotypesin ABC and GCB DLBCL samples. (D) IgH constant region isotypes in ABCDLBCL cases expressing the indicated VH segments.

13448 | www.pnas.org/cgi/doi/10.1073/pnas.1514944112 Young et al.

Following shRNA induction for 2 d, the proportion of viable,transduced mouse IgM+/GFP+ cells was scored over time by flowcytometry, which evaluates the relative degree of cell death andproliferation of the transduced versus untransduced cells. Previousstudies revealed that BCR signaling in ABC DLBCL engages theNF-κB pathway, which primarily functions to inhibit apoptosis butalso promotes cell cycle progression (28, 29). For this analysis, wegated on cell populations with equivalent cell surface expression ofthe transduced murine IgM chains. Because pairing of Ig heavy andlight chains is required for cell surface expression of the BCR, thisgating ensures that the ectopic mouse IgM chain effectively pairswith the endogenous human Ig light chain. Furthermore, by gatingon equivalent surface BCR expression, any functional differencesbetween rescue constructs can be attributed to differences inV-region specificity and not merely to differences in BCR surfaceabundance. In establishing this system, we noted that the degree ofsurvival of transduced cells was proportional to the level of ectopicmouse IgM expression (Fig. S1), demonstrating the value of gatingon defined surface BCR levels in our survival assays.The IgH rescue system was used to evaluate the IgVH depen-

dence of three IgM+ ABC DLBCL lines: HBL1 (VH4-34+), OCI-

Ly10 (VH3-7+), and TMD8 (VH3-48

+). In cells receiving a controlempty rescue vector, shRNA-mediated knockdown of endogenousIgM expression was toxic, as expected (Fig. 2B). The viability ofeach cell line was fully or partially rescued by ectopic provision of amouse IgM chain bearing the V region that is naturally producedby that cell line. Mouse heavy chains bearing different V regionsrescued viability less well, or not at all. To quantify these differ-ences, we normalized the data such that the viability of cells at day8 following transduction with a vector expressing the endogenousIgH V region was set to 100%, and the viability of cells at day 8following transduction with an empty vector was set to 0% (Fig.2C). From this representation, it is clear that the OCI-Ly10 cell linehas a pronounced preference for the OCI-Ly10 IgH V region. Inthe HBL1 and TMD8 lines, the self IgH V region was also pre-ferred, whereas some heterologous IgH V regions partially rescuedviability and others were inactive. Hence, the mere presence of aBCR on the cell surface is insufficient to sustain survival of theseABC cell lines, implying that the BCR signaling in these ABC celllines is unlikely to represent “tonic” signaling, which requires thepresence of a BCR on the cell surface but not strong antigenicengagement (13).

Self-Antigen Recognition by VH4-34+ BCRs in ABC DLBCL. VH4–34 is

the most frequently used IgH V gene segment in ABC DLBCL(Fig. 1A) and is also prevalent in other BCR-dependent malignan-cies, including CLL (5, 30) andMCL (6). The ABC cell line HBL1 isBCR-dependent (12) and expresses a VH4-34

+ IgH (Fig. 3A). Wegenerated mutants targeting the CDRs of the HBL1 IgH and testedtheir ability to sustain viability of HBL1 cells using the rescue assayoutlined above. Although reversion of the HBL1 CDR1 to the germ-line–encoded VH4-34 sequence had no effect on its ability to rescueviability, reversion of the HBL1 CDR2 to its germ-line sequencecaused a moderate (20%) but reproducible impairment, as did re-version of both CDR1 and CDR2 (Fig. 3B). The CDR2 mutation inHBL1 disrupts a motif for N-linked glycosylation (N52H53S54), whichis often mutated in normal memory B cells, presumably to improveantigen-binding affinity (31). This motif is frequently altered amongVH4-34

+ ABCDLBCL tumors, with 59% of IgM cases and 100% ofIgG cases acquiring mutations (Fig. 3D). Finally, to probe thefunction of the HBL1 CDR3 region, we mutated its arginine andaspartic acid residues to alanine. Mutation of all three aspartic acidresidues, the arginine alone, or all charged residues had a partial(30–40%) effect on rescue ability in this assay.The fact that mutations within the CDRs of the HBL1 BCR

did not yield a more dramatic impairment in cell survival sug-gested the possibility that an invariant aspect of VH4-34 mightcontribute to BCR signaling. Consistent with this possibility,several distinct VH4-34

+ sequences derived from ABC DLBCLpatient biopsy samples were able to rescue the viability of HBL1cells (Fig. 3C). The germ-line–encoded VH4-34 sequence confersintrinsic autoreactivity, which is directed against self-antigens onthe surface of red blood cells in cold agglutinin disease (32), or Bcells in systemic lupus erythematosis (SLE) (33). This auto-reactivity requires two motifs within the FR1, Q6W7 and A24V25Y26,which in other VH4 family members are instead E6S7 and T24V25S26,respectively (33) (Fig. 3A). Among VH4-34 sequences fromIgM+ ABC DLBCL cases, this AVY motif is mutated in only7% of cases compared with 40% in IgG+ cases (Fig. 3D) (P = 0.056).Similarly, the FR1 QWmotif is only mutated in 2% of IgM+VH4-34

+

ABC DLBCL cases (Fig. 3D). This finding implies that theautoreactive potential of VH4-34 is retained in the majority ofIgM+ ABC DLBCL cases. Expression of HBL1 heavy-chainconstructs in which either of these autoreactive FR1 motifs wasmutated to the consensus VH4 family sequence failed to rescue

OCI-Ly10 (VH3-7) TMD8 (VH3-48)

OCI-Ly10(VH3-7)

TMD8(VH3-48)

Live

IgV

H+

(G

FP

+)

cells

(%

day

0)

B C

A

0 4 40862 2 6 8

ABC DLBCLcell line

Knockdown endogenous IgH

Rescue withendogenous or heterologous

human IgVH fused tomouse IgM constant region

mouse IgM

Gate on equivalentsurface BCR expression

Survivalassay

OCI-Ly10HBL1

endogenous IgVH

heterologous IgVH

TMD8

OCI-Ly19vectoronly

OC

I-Ly1

0H

BL1

TM

D8

OC

I-Ly1

9

OC

I-Ly1

0

HB

L1T

MD

8

OC

I-Ly1

9

OC

I-Ly1

0H

BL1

TM

D8

OC

I-Ly1

9

Nor

mal

ized

% r

escu

e (d

ay 8

)

HBL1 (VH4-34)

HBL1(VH4-34)

Days post shRNA induction

120

100

80

60

40

20

00 42 6 8

120

100

80

60

20

-20

0

40

IgVHrescuevector

Rescue vector

shIgM

+

+

+

+

+

ABC

ABC

ABC

GCB

–

DLBCLsubtype

Rescuevector:

Fig. 2. Ectopic expression of IgH chains to determine the role of IgH V regions in maintaining the viability of ABC DLBCL cell lines. (A) A knockdown-rescuestrategy to create isogenic ABC cell lines that express different IgH V regions. See Results for details. (B) Ability of the indicated IgH V regions to sustainviability of ABC cell lines following knockdown of endogenous IgM expression. For each indicated ABC cell line, ectopic re-expression of the endogenous IgHV region was most efficient at sustaining survival (mean ± SEM, n ≥ 8). (C) Data from B plotted as the percent rescue of survival at 8 d after shIgM induction:100% is scaled to the degree of rescue by re-expression of the endogenous IgH V region and 0% is scaled to the degree of rescue by an empty vector.

Young et al. PNAS | November 3, 2015 | vol. 112 | no. 44 | 13449

MED

ICALSC

IENCE

SINAUGURA

LART

ICLE

HBL1 survival efficiently (Fig. 3E). The superiority of the wild-type HBL1 IgVH sequence over these FR1 mutant sequences wasevident at each level of ectopic mouse IgH expression (Fig. S2).Moreover, mutation of the QW motif in the context of a patient-derived VH4-34 heavy chain completely abrogated its ability tosustain HBL1 viability (Fig. 3F).The FR1 region of the VH4-34 IgH chain is thought to bind to

cell surface glycoproteins that bear the n-acetyl-lactosamine Iiblood group determinant (34). To determine if the HBL1 BCR canbind antigens on the surface of malignant B cells, we prepared asoluble human IgG1 antibody bearing the HBL1 V region, eitherwith or without the IgH FR1 mutation of QW to ES. The HBL1-IgG bound to the surface of three different ABCDLBCL cell lines,including HBL1 itself (Fig. 3G). In contrast, the HBL1-IgG with amutation disrupting the FR1 QW motif had little if any ability tobind to the surface of these lines. These data indicate that theHBL1 VH4-34 BCR uses its unique FR1 elements to recognizeantigens expressed on its own cell surface.

A VH3-7 BCR in ABC DLBCL Recognizes an Antigen Present in ApoptoticCell Debris. In addition to VH4-34, the VH3-7 gene segment is usedmore often in ABC than in GCB DLBCL cases or in normal Bcells (Fig. 1A). The ABC cell line OCI-Ly10 is dependent on aVH3-7 IgM-BCR for its survival (Fig. 2B). A serendipitous cluethat the OCI-Ly10 BCR binds a self-antigen came from observingits cell surface expression under different culture conditions. OCI-Ly10 cells are typically cultured in 20% (vol/vol) human plasma butcan be adapted to grow in 20% (vol/vol) FBS. We detected sub-stantially lower levels of BCR expression on the surface of OCI-Ly10 cells cultured in human plasma compared with those culturedin FBS, suggesting the possibility that human plasma may containan autoantigen that induces BCR internalization (Fig. 4B). Giventhat plasma contains debris from apoptotic cells, especially in pa-tients with SLE (35), and that autoantigens from apoptotic cellscan bind some BCRs in CLL (9), we tested whether the OCI-Ly10BCR reacts to antigens present in apoptotic debris, created asdescribed previously (9). Provision of apoptotic debris to OCI-Ly10cells reduced surface BCR expression after overnight incubation,

which was also true for the ABC cell line U2932, but not for otherABC lines (HBL1, TMD8) or for two GCB cell lines (Fig. 4C).We next investigated whether apoptotic debris induces BCR

signaling by examining tyrosine phosphorylated proteins (p-Tyr)in OCI-Ly10 and U2932 ABC cells. In both lines, exposure toapoptotic debris stimulated brisk tyrosine phosphorylation ofmultiple proteins, most of which were also phosphorylated uponanti-IgM–induced BCR cross-linking (Fig. 4D). To verify thatthe apoptotic debris increased p-Tyr levels by engagement of theBCR, we used an shRNA to knock down the expression of the BCRcomponent CD79A in OCI-Ly10 cells, thereby preventing BCRsurface expression and signaling (12). In contrast to cells induced toexpress a control shRNA, CD79A knockdown prevented the in-crease in p-Tyr levels by apoptotic debris (Fig. 4E). Hence, one ormore autoantigens in apoptotic debris appear to interact with theBCR, causing it to signal and internalize.To determine if the survival of OCI-Ly10 cells depends on the

ability of its BCR to bind an antigen, we made a series of mutationsin the CDRs of the OCI-Ly10 IgH V region (Fig. 4A). Expressionof an OCI-Ly10 IgH in which CDR1 was reverted to the VH3-7germ-line sequence modestly increased its ability to rescue OCI-Ly10 cells following IgM knockdown (Fig. 4F). In contrast,mutation of the OCI-Ly10 CDR2 to its germ-line configurationsignificantly reduced its ability to sustain OCI-Ly10 viability, sug-gesting that this sequence element may contribute to antigenbinding (Fig. 4F). Expression of an OCI-Ly10 IgH with both theCDR1 and CDR2 sequences reverted to germ line had an in-termediate phenotype (Fig. 4F). We next performed alanine mu-tagenesis of several residues in the OCI-Ly10 CDR3 region.Mutation of either of the two charged residues within the CDR3substantially reduced the ability of this heavy chain to maintainOCI-Ly10 survival. Alteration of K98 at the border of CDR3 andFR2, which was likely created by junctional diversification duringVDJ recombination, had the greatest effect, reducing rescue effi-ciency by over 60% (Fig. 4F). These data suggest that nongerm-line–encoded amino acids in the OCI-Ly10 IgH chain may havebeen selected to sustain chronic active BCR signaling.Because the OCI-Ly10 IgH construct bearing the CDR3 lysine

to alanine substitution was unable to sustain OCI-Ly10 viability, we

A

B C D E F

G

Fig. 3. Reactivity of VH4-34+ BCRs with self-antigens in ABC DLBCL. (A) The HBL1 IgH V region sequence compared with the germ-line VH4-34 sequence.

Mutations introduced in this study are highlighted in red. (B) Substitutions of CDR residues in the HBL1 IgH V region affect HBL1 viability (mean ± SEM, n = 3).(C) Ability of VH4-34

+ V regions derived from ABC DLBCL biopsies to sustain HBL1 survival (mean ± SEM, n ≥ 3). (D) Frequency of mutations in ABC DLBCLbiopsies targeting the Q6W7 and A24V25Y26 FR1 motifs and the N52H53S54 CDR2 motif in the germ-line VH4-34 gene segment, for ABC cases using either an IgMor IgG constant region. (E) Mutation of the FR1 Q6W7 and A24V25Y26 motifs in the HBL1 IgH V region fail to sustain viability of HBL1 cells (mean ± SEM, n = 3).(F) Mutation of the FR1 Q6W7 motif in a VH4-34

+ ABC DLBCL biopsy impairs rescue of HBL1 cells (mean ± SEM, n = 2). (G) FACS analysis of a soluble humanIgG1 antibody bearing the HBL1 V region binding to the surface of the indicated ABC DLBCL cell lines. The IgH V region of the recombinant antibody waseither identical to the HBL1 IgH V region (WT) or was mutated in the FR1 Q6W7 motif (representative data from three experiments).

13450 | www.pnas.org/cgi/doi/10.1073/pnas.1514944112 Young et al.

speculated that it would also be unable to interact with auto-antigens present in apoptotic debris. To investigate this, we engi-neered OCI-Ly10 cells to express mouse heavy chains bearingeither the wild-type or K98A mutated IgH V regions. FollowingshRNA-mediated knockdown of endogenous IgM expression, cellswere incubated overnight with fresh media, apoptotic debris, oranti-mouse IgM antibodies as a positive control. Both the wild-typeand mutant OCI-Ly10 BCRs were internalized following anti-mouse IgM cross-linking. In contrast, the wild-type OCI-Ly10 BCRwas down-modulated upon provision of apoptotic debris while theK98A isoform was unresponsive (Fig. 4G).Some but not all VH3-7

+ IgH chains derived from ABC DLBCLpatient biopsies could sustain OCI-Ly10 viability (Fig. 4H). In thecase of one patient-derived IgH V region (ABC 2488) that wasparticularly efficient in rescuing OCI-Ly10 survival, mutation of anaspartate acid residue in its CDR3 region to alanine abrogatedrescue, again suggesting that BCR V region specificity is essentialfor ABC DLBCL cell survival. However, the CDR3 of ABC 2488differs substantially from the CDR3 of OCI-Ly10 cells in that itcontains no positively charged amino acids (Fig. S3). The cellsurface expression of the ABC 2488 BCR was not modulated byapoptotic debris (Fig. S3), suggesting that this V region may sustainthe viability of OCI-Ly10 cells by recognizing a different auto-antigen than is recognized by the OCI-Ly10 BCR.

Anti-Idiotype Reactivity in ABC DLBCL.Although many ABC DLBCLcases express the VH4-34 or VH3-7 V gene segments, the majorityof ABC cases express a nondominant V gene segment yet maynonetheless rely on antigen-dependent BCR signaling for theirsurvival. We therefore examined antigen specificity in the ABCline TMD8, which uses a nondominant VH3-48 IgH V genesegment and relies on BCR signaling for survival (Fig. 5 A andB). Reversion of CDR2 in the TMD8 V region to the VH3-48germ-line sequence resulted in a slight loss of its ability to rescueTMD8 survival after IgM knockdown (Fig. 5B). In contrast,mutation of charged residues in the CDR3 region to alanine

substantially affected its ability to rescue survival, particularlywhen the sole aspartic acid was mutated to alanine (D99A) (Fig.5B). Thus, the CDR portions of the TMD8 BCR are critical toits function, suggesting that BCR signaling in TMD8 requiresantigen binding.Recent work demonstrated that BCRs of CLL cells bind to a

V37R38Q39 epitope in their own FR2 domain (10), effectively anidiotope/anti-idiotope interaction (36). TMD8 IgH V regions withmutations targeting this FR2 idiotope (V37G or R38A) were se-verely crippled in their ability to sustain TMD8 survival (Fig. 5C,Left), implicating a similar mechanism in ABC DLBCL. In con-trast, introduction of the V37G mutation into the OCI-Ly10 Vregion did not affect its ability rescue survival of OCI-Ly10 cells(Fig. 5C, Right). This finding suggests that the V37G mutation doesnot cause gross misfolding of the IgH chain and is consistent withthe fact that the OCI-Ly10 V region recognizes a different antigenin apoptotic cell debris. To show directly that the TMD8 BCR canbind to itself through the V37R38 idiotope, we transduced HBL1cells with vectors expressing mouse IgM fused to either the wild-type or V37G isoform of the TMD8 V region. After knockdown ofendogenous HBL1 IgM, we labeled cells with a soluble TMD8-IgG1 antibody. The TMD8-IgG antibody bound to cells expressingthe wild-type TMD8 V region but not to cells expressing the V37Gisoform or to cells transduced with an empty vector. These dataestablish that the TMD8 BCR binds directly to its own FR2 re-gion, thus demonstrating that the TMD8 BCR functions as itsown self-antigen.To determine if ABC DLBCL patient cases use the V37R38 FR2

idiotope as a self-antigen, we assayed seven IgH V regions derivedfrom VH3-7

+ABC patient biopsies for their ability to sustain TMD8survival. Only two of these IgH V regions rescued TMD8 survival bymore than 50% compared with rescue by the TMD8 V region (Fig.5E). Introduction of the V37G FR2 mutation into these two IgHchains abrogated their ability to rescue viability, demonstrating thatthis anti-idiotypic mechanism is active in some ABC cases. Becauseit has been reported that many CLL BCRs signal through this anti-

ED

H

CB

A

F G

Fig. 4. Reactivity of ABC DLBCL BCRs to antigenspresent in apoptotic debris. (A) Sequence of theOCI-Ly10 IgH V region compared with the germ-lineVH3-7 sequence. Mutations introduced in this studyare highlighted in red. (B) FACS analysis of surfaceIgM in OCI-Ly10 cells cultured in FBS or humanplasma. (C) FACS analysis of surface IgM expressionin the indicated DLBCL cell lines cultured with orwithout apoptotic cell debris (see SI Materials andMethods for details). (D) Western blot analysis oftyrosine phosphorylated proteins (pY 4G10) in OCI-Ly10 (Left) and U2932 (Right) following stimulationwith apoptotic debris or anti-IgM (0.5 μg/mL) for theindicated times. (E) Western blot analysis of pY inOCI-Ly10 cells exposed to apoptotic debris as in-dicated. Cells expressed either an shRNA targetingCD79A or a control shRNA, as indicated. (F) Substi-tutions of CDR residues affect rescue of OCI-Ly10viability (mean ± SEM, n ≥ 3). (G) FACS analysis ofOCI-Ly10 cells ectopically expressing mouse IgMfused to either the OCI-Ly10 IgH V region (WT) or thisV region with a K98A substitution, in the indicatedculture conditions (Left); relative mean fluorescenceintensity is plotted (Right) (mean ± SEM, n ≥ 3).(H) Rescue by VH3-7

+V regions derived fromABCDLBCLbiopsies in OCI-Ly10 cells (mean ± SEM, n = 3). Right twobars indicate the degree of rescue by ABC DLBCL case2488 (WT) compared with this V region with a D99

A substitution in its CDR3 region (mean ± SEM, n = 3).

Young et al. PNAS | November 3, 2015 | vol. 112 | no. 44 | 13451

MED

ICALSC

IENCE

SINAUGURA

LART

ICLE

idiotypic mechanism in a mouse pre–B-cell system (10), we testedwhether CLL-derived IgH V regions would sustain the viability ofABC cell lines. Among seven VH3-7

+ V regions from CLL cases(30), six were able to maintain TMD8 viability, to varying de-grees (Fig. 5G), whereas none were able to maintain the viabilityof OCI-Ly10 cells (Fig. S4). Introduction of the V37G mutationinto the two most potent CLL V regions (CLL38 and CLL135)reduced their ability to rescue the viability of TMD8 cells, sug-gesting that these V regions signal via the anti-idiotypic mecha-nism (Fig. 5G). However, the V37G isoform of CLL 38 retainedsubstantial activity, perhaps indicating that anti-idiotypic re-activity is not the only way in which this V region promotes BCRsignaling.

DiscussionOur studies demonstrate that the ongoing, malignant survival of Blymphoma cells can depend critically on the ability of their BCRsto bind self-antigens. Previously, various types of indirect evidencepointed to a potential role for autoreactivity in B-cell malignancies,including the restricted use of particular V regions (5, 6, 30) andthe autoreactivity of antibodies derived from malignant B cells(10). However, such data cannot distinguish between autoreactivitythat is continuously required for malignant cell survival andautoreactivity that is merely a vestigial remnant of the normal Bcell from which the lymphoma arose. By creating isogenic ABCDLBCL cells that differ only in their utilization of particular IgHV regions, we have demonstrated directly that interaction of theBCR with a self-antigen is essential for the survival of theselymphomas cells. We can now understand the previously reportedBCR microclusters in the plasma membrane of ABC DLBCLcells (12) as the consequence of self-antigen binding. The presentwork also illuminates clinical trials that have shown activity of theBCR pathway inhibitor ibrutinib in ABC DLBCL and otherB-cell malignancies.The self-antigens recognized by ABC DLBCL BCRs are diverse

and theoretically could either be coexpressed by the same cell asthe BCR (“cis” model) or by a neighboring cell (“trans model”)(Fig. 6). In the three ABC lines tested, each IgH V region dictatesbinding to a different self-antigen (Fig. 6), which is perhaps notsurprising given the diversity of autoantibodies that are potentiallyencoded by germ-line VH segments (11), or acquired somatically

during VDJ joining or activation-induced cytidine deaminase-dependent somatic hypermutation (37). Two of the lymphoma BCRsrecognized apparently distinct antigens present on their own cellsurface. The VH4-34

+ BCR of HBL1 cells recognizes a cell surfaceantigen using FR1 residues that have previously been implicatedin binding to n-acetyl-lactosamine chains, suggesting that the self-antigen may be one or more glycoproteins (Fig. 6A). VH4-34

+ Vregions from patient samples also sustained HBL1 survival in amanner that depended on motifs in the FR1 that confer thisautoreactivity. These FR1 motifs are not encoded in other germ-line VH segments, potentially accounting for the fact that nearlyone-third of all ABC DLBCL tumors use VH4-34. The cell surfaceantigen recognized by the BCR of TMD8 cells is the BCR itself:survival of TMD8 cells depended upon an idiotope in the FR2 ofthe IgH V region, and cells bearing BCRs in which this idiotopewas disrupted were not bound by a soluble antibody designed tomatch the TMD8 BCR (Fig. 6B). This anti-idiotypic mechanismwas shared by other IgH V sequences from other ABC DLBCLcases and some, but not all CLL cases, as previously described(10). Chronic active BCR signaling in OCI-Ly10 is apparentlydriven by yet another self-antigen that is present in apoptoticdebris (Fig. 6C). This observation is reminiscent of the auto-reactivity to antigens expressed on the surface of apoptotic cellsor in their debris that has been described in CLL (8, 9). Histolog-ically, most malignant lymphomas contain evident apoptotic cells,most likely as a byproduct of their disordered regulatory biology,and this dead cell debris could furnish self-antigens that react withsome lymphoma BCRs.The interaction of lymphoma BCRs with self-antigens relies in

part on the intrinsic autoreactivity encoded in the germ-line VHrepertoire. The frequent utilization of VH4-34 in ABC DLBCLcases is consistent with previous work demonstrating an enrich-ment for VH4-34 among DLBCL cases classified as “non-GCB” byimmunostaining (20). Primary central nervous system lymphoma(PCNSL) is a type of DLBCL that has an ABC phenotype (38) andacquires several of the pathognomonic mutations of ABC DLBCL(39–42). VH4-34 is used in 55% of PCNSL cases, an enrichmentthat suggests that this VH segment confers a selective advantage inPCNSL, as in nodal ABC DLBCL. To a lesser extent, VH4-34 useis enriched in CLL (8.9% of cases) and MCL (14.6% of cases) (5,6), suggesting that the intrinsic autoreactivity of VH4-34 could play

A

B

E F G H

DCFig. 5. Anti-idiotypic reactivity of ABC DLBCL BCRs.(A) Sequence of the TMD8 IgH V region comparedwith the germ-line VH3-48 sequence. Mutations in-troduced in this study are highlighted in red. (B) Sub-stitutions of CDR residues in the TMD8 IgH V regionaffect TMD8 viability (mean ± SEM, n = 3). (C, Left)Substitutions of FR2 residues in the TMD8 IgH V regionfail to sustain TMD8 viability. (Right) Substitution of aFR2 residue in the OCI-Ly10 IgH V region maintainsOCI-Ly10 viability (mean ± SEM, n = 5). (D) Surfacebinding of a soluble human IgG1 antibody bearing theTMD8 V region to HBL1 cells transduced with eitherthe IgH V region from TMD8 cells (WT), the TMD8 FR2V37G, or empty vector, as indicated (FACS plots repre-sentative of three experiments). (E) Rescue by VH3-7

+ Vregions derived from ABC DLBCL biopsies and OCI-Ly10 in TMD8 (mean ± SEM, n ≥ 3). (F) Rescue by ABCDLBCL biopsies with the FR2 V37G substitutions inTMD8 (mean ± SEM, n ≥ 3). (G) Rescue by IgH V re-gions derived from CLL patient samples in TMD8(mean ± SEM, n ≥ 3). (H) Effect of a FR2 V37G sub-stitution in two CLL patient samples on rescue in TMD8(mean ± SEM, n ≥ 3).

13452 | www.pnas.org/cgi/doi/10.1073/pnas.1514944112 Young et al.

a role in these malignancies as well. Although ∼5–10% of naïve Bcells have VH4-34

+ BCRs, these B cells do not often progressthrough the germinal center reaction to generate memory B cellsand plasma cells (43). The intrinsic autoreactivity of VH4-34

+ an-tibodies to n-acetyl-lactosamine moieties on glycoproteins is pre-sumed to anergize these B cells, potentially accounting for theirrelative inability to enter productive immune responses. Of theVH4-34

+ B cells that do make it into the memory B-cell pool inhealthy individuals, 43% acquire mutations that disrupt the FR1AVY motif that is essential for autoreactivity, presumably as onemechanism to escape anergy (31). In contrast, 93% of the VH4-34sequences from IgM+ ABC DLBCLs were wild-type for this FR1motif. Autoreactivity of VH4-34 also requires the FR1 QW motif,which is retained in 98% of VH4-34 sequences from IgM+ ABCDLBCL cases. Together, these findings suggest that most VH4-34

+

BCRs in ABC DLBCL retain autoreactivity. Consistent with thisnotion, experimental mutation of either the AVY or QW motif inthe FR1 diminished the ability of the VH4-34

+ IgM chains tosustain HBL1 viability and blocked autoreactive binding to thesurface of lymphoma cells. Patients with SLE have an increasedrisk of DLBCL (44, 45), which is notable because the autoreactivepotential of VH4-34 antibodies contributes substantially to diseaseflares in SLE (46). SLE epidemiological studies to date have notdistinguished between the ABC and GCB DLBCL subtypes, butour studies would predict that patients with SLE would have aparticular predisposition to ABC DLBCL, given the importance ofVH4-34 to both diseases.Lymphoma BCRs also rely on acquired/altered autoreactivity

that is a consequence of somatically generated residues in IgH Vregions. The activity of the HBL1 VH4-34

+ BCR is influenced bya somatic mutation in CDR2 that disrupts an NHS motif forN-linked glycosylation. Attachment of carbohydrate chains to theV region can impair antigen binding, presumably because of sterichindrance (31). Notably, the CDR2 NHS motif is mutated in 60%of IgM+ ABC DLBCL cases, suggesting that this is a frequentmechanism to increase antigen binding by lymphoma BCRs.Consistent with this view, this motif is also mutated in 72% of VH4-34

+

PCNSL tumors (47). The phenomenon we observe here appearsto be different from the reported increase in BCR signaling at-tributed to the somatic acquisition of N-linked glycosylation sitesin the IgVH sequences from follicular lymphoma (48). The HBL1BCR also requires charged residues in CDR3 for optimal activity,either to promote reactivity to the same antigens as are recog-nized using the FR1 motifs or to other antigens. These chargedresidues are generated by the DH and JH segments and by N-region addition during VDJ joining. In the TMD8 BCR, chargedresidues in CDR3 are also important for activity, especially D100,which is also generated by junctional diversification during VDJ

joining. Finally, the OCI-Ly10 BCR acquired somatic mutationsaltering residues in CDR2 (A57I58) and in CDR3 (K98), which arerequired for its ability to promote OCI-Ly10 survival and, in thecase of K98, for its ability to interact with self-antigens in apoptoticdebris. Thus, in each case, full BCR function requires contributionsfrom somatically generated CDR residues that should be present inonly a small minority of normal B cells.From this perspective, somatically acquired changes in the IgH

V regions of B cells could be viewed as one of the earliest geneticevents in multistep lymphomagenesis. In some cases, the first on-cogenic “hit” may occur in the bone marrow as a result of junc-tional diversification during VDJ joining and in other cases, thenecessary genetic alteration of the V region may occur in the pe-riphery as a result of somatic hypermutation. Thus, although manymature B cells are autoreactive (11), leading to clonal anergy (49),we speculate that only a minority of B cells acquire the particularstrength or quality of autoreactivity that is necessary to promotemalignant B-cell survival.Our results provide a fresh perspective on clinical trials of BCR

pathway inhibitors in lymphoid malignancies. A recent clinical trialof the BCR pathway inhibitor ibrutinib in relapsed/refractoryDLBCL demonstrated significant activity in the ABC but notGCB subtype (15). Within ABC DLBCL, 37% of patientsresponded, and those whose tumors had gain-of-function muta-tions targeting the BCR subunit CD79B responded somewhatmore frequently. However, most ibrutinib responses in this trialwere observed in patients whose tumors had wild-type CD79B,suggesting a nongenetic mechanism for BCR pathway addiction(15). Our present findings suggest that BCR signaling in ABCDLBCL is typically initiated by autoreactivity of the BCR to a self-antigen, and the ensuing chronic active BCR signaling makes thetumors sensitive to ibrutinib. Indeed, the mutant CD79B isoformsthat occur in ABC DLBCL cannot initiate BCR signaling de novo,but rather increase the amplitude of already ongoing BCR sig-naling (12). Ibrutinib also has clinical activity in CLL and MCL,neither of which are characterized by gain-of-function mutations inCD79B (50–52), suggesting that the type of autoreactivity that wedescribe here may also initiate BCR signaling and maintain via-bility in these other lymphoid malignancies. Indeed, some but notall CLL IgH V regions were able to sustain the survival of theTMD8 cell line, and this was a result of their self, anti-idiotypicreactivity. Finally, our findings suggest a potential precision medi-cine approach to optimize the deployment of BCR pathway in-hibitors for the therapy of lymphoid malignancies. Specifically, themodel systems we have developed could provide a means to gaugewhether the IgH V region expressed in a particular tumor isautoreactive in a fashion that can sustain lymphoma survival. Indeed,patient-derived V regions varied considerably in their ability to

Y

Y YP

PSFK

Y

P

PSYK

Y

Y YP

PSFK

IgH

IgH

Ig

L

IgL

Y

P

PSYKSYK

SFK

CD79A

CD79B

SYK

SFKY

Y YP

PSFK

Y

P

PSYK

Y

Y YP

PSFK

Y

P

PSYKSYK

SFK

CD79A

CD79B

SYK

SFK

Y

Y

P

PSFK

IgH

IgH

Ig

L

IgL

Y

Y Y

Y

P

PSYKSYK

SFK

CD79A

CD79B

Y

Y YP

PSFK

Y

P

PSYK

Y

Y YP

PSFK

IgH

IgH

Ig

L

Y

P

PSYKSYK

SFK

CD79A

CD79B

SYK

SFKY

Y YP

PSFK

Y

P

PSYK

Y

Y YP

PSFK

IgH

IgH

Ig

L

IgL

Y

P

PSYKSYK

SFK

CD79A

CD79B

SYK

SFK

P

P

Y P

P

P

P PP

PSYK SYK SYK

SFK SFK SFK SFK

Y

Y

Y

Y

Y

Y

YSYK

A CB

B cell receptor

cis trans transcis

OCI-Ly10 (VH3-7)TMD8 (VH3-48)HBL1 (VH4-34)

Y

Y Y

Y

IgH

IgH

Ig

L

IgL

P

PSYK

SFK

P

PSFK

SYK

self-antigen

self-antigen

self-antigen

Fig. 6. Models of self-antigen–dependent survival signaling in ABC DLBCL. (A) HBL1 (VH4-34) recognizes self-antigens expressed on the surface of malignantB cells, either in a cis or trans fashion. Binding to self-antigens is potentially mediated through IgH FR1 and CDR3 interactions with n-acetyl-lactosamineglycans attached to cell surface proteins. (B) TMD8 (VH3-48) binds to an idiotope in the IgH FR2 region of its BCR, either in a cis or trans fashion. (C) The OCI-Ly10 (VH3-7) BCR binds to self-antigens released from apoptotic cells.

Young et al. PNAS | November 3, 2015 | vol. 112 | no. 44 | 13453

MED

ICALSC

IENCE

SINAUGURA

LART

ICLE

sustain the survival of our ABC DLBCL models. In the future, itwill be interesting to assess whether the autoreactivity measuredby these models systems is able to predict which patients willbenefit from ibrutinib and other BCR pathway inhibitors.

Materials and MethodsIgH V gene segment sequences were assembled from RNA-seq data using Trinity(53) and processed by IMGT High V-Quest (54). Patient samples were obtainedafter informed consent and were treated anonymously during RNA-seq anal-ysis. ABC cell lines were grown in Iscove’s modified Dulbecco’s medium (IMDM)

with 20% heparinized human plasma, penicillin/streptomycin, L-glutamine, and2-mercaptoethanol. Knockdown-rescue experiments were performed as pre-viously described (12). Additional details are available in the SI Materials andMethods. A complete list of G-blocks used in this study is in Dataset S2.

ACKNOWLEDGMENTS. This study used the high-performance computationalcapabilities of the Biowulf Linux cluster at the National Institutes of Health.This research was supported by the Intramural Research Program of theNational Institutes of Health, National Cancer Institute, and Center forCancer Research; and the Dr. Mildred Scheel Stiftung für Krebsforschung(Deutsche Krebshilfe) (R.S.).

1. Dameshek W, Schwartz RS (1959) Leukemia and auto-immunization—Some possiblerelationships. Blood 14:1151–1158.

2. Küppers R (2005) Mechanisms of B-cell lymphoma pathogenesis. Nat Rev Cancer 5(4):251–262.

3. Quinn ER, et al. (2001) The B-cell receptor of a hepatitis C virus (HCV)-associated non-Hodgkin lymphoma binds the viral E2 envelope protein, implicating HCV in lym-phomagenesis. Blood 98(13):3745–3749.

4. Wotherspoon AC, et al. (1993) Regression of primary low-grade B-cell gastric lym-phoma of mucosa-associated lymphoid tissue type after eradication of Helicobacterpylori. Lancet 342(8871):575–577.

5. Agathangelidis A, et al. (2012) Stereotyped B-cell receptors in one-third of chroniclymphocytic leukemia: A molecular classification with implications for targetedtherapies. Blood 119(19):4467–4475.

6. Hadzidimitriou A, et al. (2011) Is there a role for antigen selection in mantle cell lym-phoma? Immunogenetic support from a series of 807 cases. Blood 118(11):3088–3095.

7. Hervé M, et al. (2005) Unmutated and mutated chronic lymphocytic leukemias derivefrom self-reactive B cell precursors despite expressing different antibody reactivity.J Clin Invest 115(6):1636–1643.

8. Catera R, et al. (2008) Chronic lymphocytic leukemia cells recognize conserved epi-topes associated with apoptosis and oxidation. Mol Med 14(11-12):665–674.

9. Chu CC, et al. (2010) Many chronic lymphocytic leukemia antibodies recognize apo-ptotic cells with exposed nonmuscle myosin heavy chain IIA: Implications for patientoutcome and cell of origin. Blood 115(19):3907–3915.

10. Dühren-von Minden M, et al. (2012) Chronic lymphocytic leukaemia is driven byantigen-independent cell-autonomous signalling. Nature 489(7415):309–312.

11. Wardemann H, et al. (2003) Predominant autoantibody production by early human Bcell precursors. Science 301(5638):1374–1377.

12. Davis RE, et al. (2010) Chronic active B-cell-receptor signalling in diffuse large B-celllymphoma. Nature 463(7277):88–92.

13. Lam KP, Kühn R, Rajewsky K (1997) In vivo ablation of surface immunoglobulin onmature B cells by inducible gene targeting results in rapid cell death. Cell 90(6):1073–1083.

14. Tolar P, Hanna J, Krueger PD, Pierce SK (2009) The constant region of the membraneimmunoglobulin mediates B cell-receptor clustering and signaling in response tomembrane antigens. Immunity 30(1):44–55.

15. Wilson WH, et al. (2015) Targeting B cell receptor signaling with ibrutinib in diffuselarge B cell lymphoma. Nat Med 21(8):922–926.

16. Byrd JC, et al. (2013) Targeting BTK with ibrutinib in relapsed chronic lymphocyticleukemia. N Engl J Med 369(1):32–42.

17. Boyd SD, et al. (2010) Individual variation in the germline Ig gene repertoire inferredfrom variable region gene rearrangements. J Immunol 184(12):6986–6992.

18. Hsu FJ, Levy R (1995) Preferential use of the VH4 Ig gene family by diffuse large-celllymphoma. Blood 86(8):3072–3082.

19. Ruminy P, et al. (2011) The isotype of the BCR as a surrogate for the GCB and ABCmolecular subtypes in diffuse large B-cell lymphoma. Leukemia 25(4):681–688.

20. Sebastián E, et al. (2012) Molecular characterization of immunoglobulin gene re-arrangements in diffuse large B-cell lymphoma: Antigen-driven origin and IGHV4-34as a particular subgroup of the non-GCB subtype. Am J Pathol 181(5):1879–1888.

21. Lossos IS, et al. (2000) Ongoing immunoglobulin somatic mutation in germinal centerB cell-like but not in activated B cell-like diffuse large cell lymphomas. Proc Natl AcadSci USA 97(18):10209–10213.

22. Wright G, et al. (2003) A gene expression-based method to diagnose clinically distinctsubgroups of diffuse large B cell lymphoma. Proc Natl Acad Sci USA 100(17):9991–9996.

23. Honjo T, Kinoshita K, Muramatsu M (2002) Molecular mechanism of class switch re-combination: Linkage with somatic hypermutation. Annu Rev Immunol 20:165–196.

24. Lenz G, et al. (2007) Aberrant immunoglobulin class switch recombination and switchtranslocations in activated B cell-like diffuse large B cell lymphoma. J Exp Med 204(3):633–643.

25. Martin SW, Goodnow CC (2002) Burst-enhancing role of the IgG membrane tail as amolecular determinant of memory. Nat Immunol 3(2):182–188.

26. Horikawa K, et al. (2007) Enhancement and suppression of signaling by the conservedtail of IgG memory-type B cell antigen receptors. J Exp Med 204(4):759–769.

27. Dogan I, et al. (2009) Multiple layers of B cell memory with different effector func-tions. Nat Immunol 10(12):1292–1299.

28. Davis RE, Brown KD, Siebenlist U, Staudt LM (2001) Constitutive nuclear factor kappaBactivity is required for survival of activated B cell-like diffuse large B cell lymphomacells. J Exp Med 194(12):1861–1874.

29. Lam LT, et al. (2005) Small molecule inhibitors of IkappaB kinase are selectively toxicfor subgroups of diffuse large B-cell lymphoma defined by gene expression profiling.Clin Cancer Res 11(1):28–40.

30. Fais F, et al. (1998) Chronic lymphocytic leukemia B cells express restricted sets of

mutated and unmutated antigen receptors. J Clin Invest 102(8):1515–1525.31. Sabouri Z, et al. (2014) Redemption of autoantibodies on anergic B cells by variable-

region glycosylation and mutation away from self-reactivity. Proc Natl Acad Sci USA

111(25):E2567–E2575.32. Potter KN, et al. (1993) Molecular characterization of a cross-reactive idiotope on

human immunoglobulins utilizing the VH4-21 gene segment. J Exp Med 178(4):

1419–1428.33. Richardson C, et al. (2013) Molecular basis of 9G4 B cell autoreactivity in human

systemic lupus erythematosus. J Immunol 191(10):4926–4939.34. Cappione AJ, Pugh-Bernard AE, Anolik JH, Sanz I (2004) Lupus IgG VH4.34 antibodies

bind to a 220-kDa glycoform of CD45/B220 on the surface of human B lymphocytes.

J Immunol 172(7):4298–4307.35. Perniok A, Wedekind F, Herrmann M, Specker C, Schneider M (1998) High levels of

circulating early apoptic peripheral blood mononuclear cells in systemic lupus er-

ythematosus. Lupus 7(2):113–118.36. Jerne NK (1974) Towards a network theory of the immune system. Ann Immunol

(Paris) 125C(1-2):373–389.37. Shlomchik MJ, Marshak-Rothstein A, Wolfowicz CB, Rothstein TL, Weigert MG (1987)

The role of clonal selection and somatic mutation in autoimmunity. Nature 328(6133):

805–811.38. Camilleri-Broët S, et al. (2006) A uniform activated B-cell-like immunophenotype

might explain the poor prognosis of primary central nervous system lymphomas:

Analysis of 83 cases. Blood 107(1):190–196.39. Braggio E, et al. (2015) Genome-wide analysis uncovers novel recurrent alterations in

primary central nervous system lymphomas. Clin Cancer Res 21(17):3986–3994.40. Nakamura T, et al. (2015) Recurrent mutations of CD79B and MYD88 are the hallmark

of primary central nervous system lymphomas. Neuropathol Appl Neurobiol, 10.1111/

nan.12259.41. Vater I, et al. (2015) The mutational pattern of primary lymphoma of the central

nervous system determined by whole-exome sequencing. Leukemia 29(3):677–685.42. Yamada S, Ishida Y, Matsuno A, Yamazaki K (2015) Primary diffuse large B-cell lym-

phomas of central nervous system exhibit remarkably high prevalence of oncogenic

MYD88 and CD79B mutations. Leuk Lymphoma 56(7):2141–2145.43. Pugh-Bernard AE, et al. (2001) Regulation of inherently autoreactive VH4-34 B cells in

the maintenance of human B cell tolerance. J Clin Invest 108(7):1061–1070.44. Smedby KE, et al. (2006) Autoimmune and chronic inflammatory disorders and risk of

non-Hodgkin lymphoma by subtype. J Natl Cancer Inst 98(1):51–60.45. Bernatsky S, et al. (2014) Lymphoma risk in systemic lupus: Effects of disease activity

versus treatment. Ann Rheum Dis 73(1):138–142.46. Tipton CM, et al. (2015) Diversity, cellular origin and autoreactivity of antibody-

secreting cell population expansions in acute systemic lupus erythematosus. Nat

Immunol 16(7):755–765.47. Montesinos-Rongen M, Purschke F, Küppers R, Deckert M (2014) Immunoglobulin

repertoire of primary lymphomas of the central nervous system. J Neuropathol Exp

Neurol 73(12):1116–1125.48. Radcliffe CM, et al. (2007) Human follicular lymphoma cells contain oligomannose

glycans in the antigen-binding site of the B-cell receptor. J Biol Chem 282(10):

7405–7415.49. Quách TD, et al. (2011) Anergic responses characterize a large fraction of human

autoreactive naive B cells expressing low levels of surface IgM. J Immunol 186(8):

4640–4648.50. Quesada V, et al. (2012) Exome sequencing identifies recurrent mutations of the

splicing factor SF3B1 gene in chronic lymphocytic leukemia. Nat Genet 44(1):47–52.51. Beà S, et al. (2013) Landscape of somatic mutations and clonal evolution in mantle cell

lymphoma. Proc Natl Acad Sci USA 110(45):18250–18255.52. Wang L, et al. (2011) SF3B1 and other novel cancer genes in chronic lymphocytic

leukemia. N Engl J Med 365(26):2497–2506.53. Li S, et al. (2013) IMGT/HighV QUEST paradigm for T cell receptor IMGT clonotype

diversity and next generation repertoire immunoprofiling. Nat Commun 4:2333.54. Wang C, et al. (2014) Effects of aging, cytomegalovirus infection, and EBV infection

on human B cell repertoires. J Immunol 192(2):603–611.55. Haas BJ, et al. (2013) De novo transcript sequence reconstruction from RNA-seq using

the Trinity platform for reference generation and analysis. Nat Protoc 8(8):1494–1512.56. Tiller T, et al. (2008) Efficient generation of monoclonal antibodies from single human

B cells by single cell RT-PCR and expression vector cloning. J Immunol Methods 329(1-2):

112–124.

13454 | www.pnas.org/cgi/doi/10.1073/pnas.1514944112 Young et al.