VH replacement in primary immunoglobulinrepertoire diversificationAmy Suna,1, Tatiana I. Novobrantsevab,1,2, Maryaline Coffrea, Susannah L. Hewitta, Kari Jensenb, Jane A. Skoka,Klaus Rajewskyb,3, and Sergei B. Koralova,4

aDepartment of Pathology, New York University School of Medicine, New York, NY 10016; and bImmune Disease Institute, Boston, MA 02115

Edited by Frederick W. Alt, Howard Hughes Medical Institute, Boston Children’s Hospital and Harvard Medical School, Boston, MA, and approved December19, 2014 (received for review September 18, 2014)

The genes encoding the variable (V) region of the B-cell antigenreceptor (BCR) are assembled from V, D (diversity), and J (joining)elements through a RAG-mediated recombination process thatrelies on the recognition of recombination signal sequences (RSSs)flanking the individual elements. Secondary V(D)J rearrangementmodifies the original Ig rearrangement if a nonproductive originaljoint is formed, as a response to inappropriate signaling from aself-reactive BCR, or as part of a stochastic mechanism to furtherdiversify the Ig repertoire. VH replacement represents a RAG-mediated secondary rearrangement inwhich an upstreamVH elementrecombines with a rearranged VHDHJH joint to generate a new BCRspecificity. The rearrangement occurs between the cryptic RSS ofthe original VH element and the conventional RSS of the invadingVH gene, leaving behind a footprint of up to five base pairs (bps)of the original VH gene that is often further obscured by exonu-clease activity and N-nucleotide addition. We have previouslydemonstrated that VH replacement can efficiently rescue thedevelopment of B cells that have acquired two nonproductiveheavy chain (IgH) rearrangements. Here we describe a novelknock-in mouse model in which the prerearranged IgH locusresembles an endogenously rearranged productive VHDHJH allele.Using this mouse model, we characterized the role of VH replace-ment in the diversification of the primary Ig repertoire throughthe modification of productive VHDHJH rearrangements. Ourresults indicate that VH replacement occurs before Ig light chainrearrangement and thus is not involved in the editing of self-reactive antibodies.

VH replacement | receptor editing | lymphocyte development |secondary rearrangement | V(D)J

Ahallmark of the adaptive immune system is its ability togenerate a large antibody repertoire despite a limited ge-

nome through the rearrangement of variable (V), diversity (D),and joining (J) genes during B-cell development in a processcalled V(D)J recombination (1). Each antibody-producing cellexpresses a single pair of heavy and light chains generated bythis process during the pro–B-cell and pre–B-cell stages of de-velopment, respectively. V(D)J recombination is mediated bythe recombination activating proteins RAG1 and RAG2, whichbind and cleave recombination recognition sequences (RSSs)flanking V, D, and J genes. Because there are several membersof V, D (for the heavy chain), and J segments, the combinatorialnature of V(D)J rearrangement allows for the generation of avastly diverse antibody repertoire from a relatively modest amountof genetic information present in the germline DNA.Ig gene rearrangement progresses in an ordered stepwise manner

during B-cell development, with Ig heavy (IgH) chain assemblybefore Ig light (IgL) chain assembly. In-frame rearrangement ofa VHDHJH joint leads to expression of an IgH chain that pairswith an invariant surrogate light chain, and, in association withthe Igα and Igβ signal-transducing subunits, these proteins formthe pre–B-cell receptor (pre-BCR) (2, 3). Signaling througha functional pre-BCR allows the cell to progress from the pro–B-cell stage to the pre–B-cell stage, where IgL rearrangement can

occur. This progression is accompanied by a burst of prolif-eration, which ensures that the functional IgH rearrangement isnot lost in the event of unsuccessful IgL recombination andincreases diversity by allowing a given IgH chain to pair withmultiple IgL chains. Only cells that acquire a functional B-cellreceptor (BCR), with an in-frame rearranged IgH chain thatsuccessfully pairs with a productively rearranged IgL chain,progress to become mature B cells.The imprecise nature of the joining process in V(D)J re-

combination contributes to the diversity of the antibody rep-ertoire, but also leads to a significant number of nonfunctionalrearrangements, with approximately two-thirds of the joints be-ing out-of-frame. At the IgH locus, VH replacement can rescue“dead-end” pro-B cells that have acquired nonproductive IgHjoints on both alleles by rearranging an upstream VH gene witha nonfunctional VHDHJH (4–7).Ordered rearrangement of the IgH and IgL loci is mediated by

the RAG1/2 proteins that recognize RSSs flanking V, D, andJ segments. The consensus RSS is composed of a heptamer(CACTGTG) and a nonamer (GGTTTTTGT) separated by ei-ther a 12-bp or 23-bp spacer (8). V(D)J recombination occurspreferentially between gene segments flanked by RSSs of dis-similar lengths, thus directing the order of recombination; thisis known as the 12/23 rule. Because the VHDHJH recombinationprocess eliminates any remaining DH genes and their flanking12-bp RSSs, further editing of this locus requires a recombination

Significance

The recombinatorial process of V(D)J rearrangement generatesa vast antibody repertoire from a limited number of genes. Thejoints generated in the course of V(D)J recombination are im-precise thus yielding greater diversity but also resulting infrequent generation of nonproductive VDJ rearrangements.We have previously shown that B cells with two nonproductiveIgH rearrangements can be efficiently rescued by a form ofsecondary V(D)J recombination called VH replacement. We nowdemonstrate that VH replacement also contributes to the di-versity of the immune repertoire by modifying productive IgHrearrangements. Results presented herein suggest that VH re-placement occurs exclusively during early stages of B-cell de-velopment and therefore does not contribute to the editing ofself-reactive antibodies.

Author contributions: A.S., T.I.N., J.A.S., K.R., and S.B.K. designed research; A.S., T.I.N.,M.C., S.L.H., K.J., J.A.S., and S.B.K. performed research; A.S., T.I.N., and S.B.K. analyzeddata; and A.S., K.R., and S.B.K. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1A.S. and T.I.N. contributed equally to this work.2Present address: Jounce Therapeutics, Cambridge, MA 02138.3Present address: Max Delbrück Center for Molecular Medicine, Berlin, Germany 13092.4To whom correspondence should be addressed. Email: sergei.koralov@nyumc.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1418001112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1418001112 PNAS Early Edition | 1 of 9

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

event that violates the 12/23 rule. VH replacement occurs be-tween a 23-bp RSS of an upstream invading VH gene and ahighly conserved 7-bp cryptic RSS (cRSS; TACTGTG) in thebody of the recipient VH gene (6, 7). Because the cRSS is locatednear the 3′ terminus of the recipient VH gene, this process leavesbehind only 5 bps proximal to the highly variable CDR3 regionof the original VHDHJH joint, a minimal footprint that can bemodified by exonuclease “chewback” and N-nucleotide addition(9, 10).VH replacement was initially observed in mouse B-cell lines in

which a small fraction of cells regularly acquired alternate IgHrearrangements, thus “editing” their original specificity (6, 7).Since then, VH replacement has been studied in transgenicmouse models, human cell lines, and human blood. Recentstudies that bioinformatically examined the frequency of 4- to5-bp VH footprints at the VH-DH vs. DH-JH junctions have sug-gested evidence of VH replacement in at least 5% of the humanIg repertoire (9). Analysis of VH footprints in mouse IgH rear-rangements revealed a similar frequency of VH replacement inwild-type (WT) mice, and also demonstrated a significantlyhigher frequency of VH replacement footprints in IgH sequencesfrom patients with autoimmune diseases and animals of au-toimmune-prone backgrounds (11, 12). However, the randomnature of exonuclease and terminal deoxynucleotidyl transferase(TdT) activity at the CDR3 region of IgH joints makes it difficultto assess the contribution of VH replacement to the repertoire bysequence analysis, owing to loss of the VH footprint. Further-more, our earlier work with a mouse model carrying a predefinednonproductive IgH rearrangement in the IgH locus has demon-strated that up to one-third of VH replacement events are mediatedby sequence microhomology of the highly conserved 3′ bases of VHgenes (4). Replacement reactions of this nature lack a detectablefootprint and thus escape detection by sequence analysis.Although the nonproductive VHDHJH mouse model allowed

us to study how VH replacement rescues nonproductive VHDHJHrearrangements, it did not address the contribution of VH re-placement to the diversification of the primary immune reper-toire or its role in editing self-reactive antibody specificities. Thus,we generated a knock-in mouse with a productive VHDHJH re-arrangement knocked into its physiological position within theIgH locus. With this mouse model, we assessed the role of VHreplacement in modifying productive IgH rearrangements innonautoreactive and autoreactive settings.

ResultsGeneration of a Mouse Strain with an IgH Locus Containing a FunctionalVHDHJH Rearrangement. Using a gene targeting and screening strat-egy similar to that described previously (4), we generated a noveltransgenic mouse with an IgH locus that mimics a physiologicallyrecombined VHDHJH allele, called D23prod, encoding the IgH chainof a polyreactive antibody, D23 (Fig. 1A). This antibody, expressedby the D23 hybridoma, has a broad range of specificities, includingssDNA and dsDNA (13). The Q52.2.4 VH gene of the D23 IgHchain encodes a well-conserved cRSS toward the 3′ end of the VHgene, and previous studies, including our own, have demonstratedVH replacement of VHDHJH joints containing the Q52 family VHgenes (4, 6, 7).We knocked the D23 IgH rearrangement into the IgH locus

in 129 IB10 ES cells, preserving all VH elements upstream ofVH81X while deleting all downstream DH and JH elements,thereby generating a VHDHJH knock-in that faithfully repro-duces an endogenous productive VHDHJH rearrangement (Fig. 1A).The VHDHJH was introduced in the JH gene region, whereas adouble loxP site was knocked into the region upstream of theVH81X gene, the most 3′ VH element. Subsequent Cre-mediatedrecombination between the upstream double loxP and the loxPflanking the introduced VDJ rearrangement resulted in the deletionof the intervening DH elements. This is the same strategy used to

generate our previously published mouse model of VH replacement,D23stop, so named because of a stop codon introduced in the D23VH gene (4). Whereas we originally eliminated the downstreamJH genes to readily distinguish between the new D23prod alleleand the nonproductive D23stop allele, the lack of JH elementsalso precludes rare VH-to-JH rearrangements that we previouslyobserved in B cells harboring two D23stop alleles (4, 14).

Mice Heterozygous for D23prod Have a Reduced Pro–B-Cell Compartment.Analysis of the bone marrow of mice heterozygous for theD23prod allele (D23prod/+) revealed a reduced compartment ofprogenitor B lymphocytes (B220+ IgM−) compared with bonemarrow from WT mice (Fig. 1B). Examination of this subsetdemonstrated a dramatic reduction in pro-B cells (c-kit+ CD25−),with a concomitant increase in the pre–B-cell compartment (c-kit−

CD25+). Because D23prod progenitor B cells enter the pro–B-cellcompartment with an in-frame VHDHJH rearrangement, the transitof these cells through this stage of development likely is accel-erated. There was no difference in the proportions of recircu-lating (B220hi IgM+) and immature (B220lo IgM+) B cells betweenmutants and controls. In the periphery, we observed a sub-stantial accumulation of cells in the marginal zone compart-ment of D23prod/+ mice (Fig. 1C; CD1dhi CD23lo), whereasthe number of follicular B cells (B220+ CD23hi) was normal.This analysis reveals that despite accelerated B-cell develop-ment, peripheral B-cell compartments appear mostly normalin D23prod/+ mice.

Evidence of Inactivation of the D23prod IgH Allele. The D23prod

VHDHJH targeting was performed in IB10 ES cells from the129/ola background; thus, mice homozygous for the D23prod

allele expressed IgM of the a allotype on the surface of B cells.We crossed the D23prod allele onto the C56Bl/6 background,where the expressed IgM is of the b allotype. To monitor use ofthe WT IgH allele or D23prod IgH allele in D23prod/+ animals, weused cell surface antibodies specific for the IgM a and b allotypes(Fig. 2A). D23prod/+ mice maintained allelic exclusion, with allmature B lymphocytes expressing either IgMa or IgMb and nocells detectably expressing both IgH alleles. The presence ofIgMb-expressing cells, albeit at low frequency (∼4%), was sur-prising, given that all pro-B cells in these mice initially carry aproductive IgH knock-in of the a allotype. These cells inactivatedthe D23 IgH allele by VH replacement and underwent pro-ductive VHDHJH rearrangement of the WT IgH allele.FACS analysis cannot fully reveal the frequency of secondary

rearrangement at the D23prod allele, however, because cells thatundergo in-frame VH replacement remain IgMa+ and con-tinue to express the IgM constant region from the targetedallele. In addition, cells that have undergone out-of-frame replace-ment may fail to productively rearrange the WT IgH locus andundergo apoptosis. Although the frequency of IgMb+ B cells in thesecondary lymphoid tissues, such as spleen, mesenteric lymph nodes,and Peyer’s patches, was similar, with a slightly higher frequencyobserved in spleen (Fig. S1A), we consistently noted an increasedfrequency of IgMb+ cells in the marginal zone compared withfollicular B cells (Fig. S1B).

VH Replacement Leads to Inactivation of the D23prod Allele. To assessthe rearrangements occurring at the D23prod IgH allele, we iso-lated splenic B cells from D23prod heterozygous and homozygousmice. Using a mixture of degenerate forward primers specific fordifferent families of VH genes (excluding the Q52 family, whichincludes the Q52.2.4 VH gene used in the D23prod rearrange-ment) with a reverse primer specific for the DH-JH junction ofD23prod, we were able to amplify secondary rearrangements fromthe D23prod locus while avoiding amplification of the originalD23prod allele or rearrangements from the WT IgH locus (Fig. 2B).Analysis of these sequences revealed that upstream VH genes had

2 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1418001112 Sun et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

invaded and replaced the recipient D23prod VH gene (Fig. 2Cand Dataset S1). We observed in-frame as well as out-of-frame re-placement events. Sequences also revealed extensive exonuclease

chewback of the replaced VH gene, occasionally eliminating thefootprint of the original VH element, and N-nucleotide addition atthe newly formed VH-to-VHDHJH joint.

9.1

1125

12

1319

82

0.4

65

5

IgM

B220

c-kit

CD25

54 48

32

63

22

74

76

20

84

9.8

CD23

CD1d

B220

CD19

B220

AA4.1

D23 VDJNeor

DH elements JH elementsDQ52VH elements 81x

Neor

double loxP

frt frt

DH elementsVH elementsNeor D23 VDJ

VH elementsD23 VDJ

LA SALA SA

WT IgH allele

Targeting vector 1 and 2

Targeted IgH locus

D23prod allele

Cre mediated deletion

Eμ

Eμ

A

B CWT D23prod/+ WT D23prod/+

Fig. 1. Generation of the mouse model with a physiologically relevant productive IgH rearrangement. (A) Targeting scheme to generate an IgH locus that closelyresembles a recombined VHDHJH allele. Targeting vector 1 was used to replace the JH elements with the D23prod VHDHJH joint and a single loxP site. Targeting vector2 introduced a double loxP into the VH 81X site. Cre transduction followed by Southern blot analysis identified ES cells harboring both insertions on the samechromosome. Cre-mediated deletion resulted in the D23prod allele mimicking an endogenously rearranged IgH sequence. (B and C) Representative FACS analysis ofbonemarrow (B) and spleen (C). B-cell subsets of 7- to 8-wk-oldWT and D23prod/+mice demonstrate contraction of the pro–B-cell subset (c-kit+ CD25−) in D23prod/+.c-kit vs. CD25 plots are gated on B220+ IgM− bone marrow cells. AA4.1 vs. B220 plots are gated on B220+ CD19+ splenocytes. CD1d vs. CD23 plots are gated onB220+ AA4.1lo mature cells. Data are representative of more than five independent experiments, with more than five mice in each group.

Sun et al. PNAS Early Edition | 3 of 9

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

We sorted IgMa+, IgMb+, and total B cells from D23prod/+

mice and sequenced the replacement joints of the D23prod allele.B cells expressing IgMb, and thus the WT Ig allele, carried ex-clusively out-of-frame VH replacements of the D23prod knock-inallele (Fig. 3); thus, the out-of-frame replacement events of theD23prod allele inactivated that allele, and the WT IgMb+ allelewas subsequently rearranged and expressed. Replacement eventsamplified from IgMa+ B cells contained primarily in-framereplacements; the few out-of-frame replacement events werelikely a result of contamination by IgMb+ cells.We previously described a mouse model in which nearly the

entire Ig repertoire was generated as a result of VH replacementof the nonproductive D23stop allele. In that mouse model, weobserved that ∼30% of the replacement events were mediated bysequence microhomology at the VH-to-VHDHJH junction andthus lacked N/P nucleotide addition (4). Sequences generatedby replacement of the D23prod allele harbored N/P nucleotidesin >95% of the cases, however. In the absence of any selectionpressure, roughly two-thirds of such replacement events areexpected to be out-of-frame, owing to the stochastic nature ofexonuclease chewback and N-nucleotide addition. Our observa-tion that the vast majority (∼95%) of replacement events fromtotal B cells of D23prod/+ mice were out-of-frame suggests posi-tive selection of cells expressing a rearrangement from the WTIgH allele, rather than a VH-replaced D23prod allele (Fig. 3).

Preferential Use of Proximal V Genes in VH Replacement of theProductive VHDHJH Allele. Sequence analysis of invading VH genesrevealed a strong preference for genes in the VH7183 family, theupstream VH gene family most proximal to the D23prod rear-rangement (Fig. 4). Indeed, the most frequently invading VH genewas the VH7183 family member VHD6.96, the VH gene im-mediately upstream of the knock-in site (Fig. S2 and DatasetS1). In contrast, in D23stop knock-in mice (4), the distribution ofJ558 and VH7183 donor VH genes in D23stop/D23stop replacement

events was similar to that of primary VDJ recombination at WTloci. One possible explanation for the preferential use of proxi-mal VH genes in the replacement events at the D23prod locus wasthe limited window of opportunity for pro-B cells with a readilyexpressed VHDHJH to open the VH locus to initiate secondaryrecombination. Indeed, in D23prod/D23prod pro-B cells in whichIgα signaling was abolished (ΔTm/ΔTm) (15), distal VH gene usewas restored to levels observed in the primary WT IgH reper-toire (Fig. 4B).Chromatin accessibility is a major mechanism regulating VH-

to-DHJH recombination (16). We suspected that VH gene useduring VH replacement also might be regulated, at least in part,by chromatin accessibility. To evaluate this possibility, we quantifiedVH gene sterile transcripts extracted from pro-B cells of D23prod/D23prod and D23stop/D23stop mice. These noncoding RNA tran-scripts of germline Ig genes reflect the accessibility of VH genesto RNA polymerase and RAG enzymes (17, 18). We comparedJ558 and VH7183 VH gene sterile transcripts by quantitativePCR using degenerate primers recognizing J558-specific andVH7183-specific genes but excluded the targeted VH gene(VH81x) and the D23prod knock-in VH gene (Q52.2.4). Weobserved a smaller fraction of J558 and more VH7183 steriletranscripts in D23prod/D23prod pro-B cells compared withD23stop/D23stop pro-B cells (Fig. S3). The correlation between dif-ferential sterile transcription and preferential proximal VH gene usein D23prod/D23prod B cells suggests a need for open chromatin andactive sterile transcription to mediate VH replacement reactions.

The Incidence of VH Replacement Is Not Increased During ReceptorEditing. We next addressed the contribution of VH replacementto editing of self-reactive BCR specificities. Receptor editing iscentral to rescuing B cells in which the IgH chain fails to pairwith the IgL chain or the IgH and IgL chain pairing yieldsautoreactive specificities (19, 20). Previous studies have usedtransgenic mice with a VHDHJH joint knocked into the IgH locus

0.2

99

50

44

95

0.1

93

4

IgMaIgMb

C57Bl/6(IgMb/IgMb)

WT F1(IgMa/IgMb)

D23prod/D23prod(IgMa/IgMa)

D23prod/+(IgMa/IgMb)

D23prod rearrangementQ52.2.4 DFL16.2

Mix of FR1 VHprimers

A

B

...TGTGCCAGAGAGAAACTACGGCTAAGGTACTTTGACTACTGG...

JH2

Reverse primer for specific for D23 allele

...TGTGCAAGGCTCCAGAGAGAACTCCGGCTAAGG...

...TGTGCAAGAGAGAACTCCGGCTAAGGTACTTTG...IFOF

...TGTGCAAGACAATCGAGAGAACTCCGGCTAAGG...OF

VHD6.96J558.427183.14

Invading VH Sequence after VH ReplacementC

Fig. 2. Inactivation of the knock-in VHDHJH. (A) Splenic B cells examined by flow cytometry for expression of IgMa and IgMb allotype reflect the use of theknock-in and WT IgH allele, respectively. Cells were first gated on B220+ IgM+ expression. Data are representative of more than 10 independent experiments.(B) PCR strategy for amplifying VH replacement events from D23prod/+ mice. A combination of forward degenerate primers for VH genes, excluding the Q52 VH

gene family and D23 VH gene, and a reverse primer specific for the D23 DH-JH sequence was used to selectively amplify VH replacement events at the knock-inlocus. (C) Sequencing revealed both in-frame and out-of-frame VH replacement by the invading V gene (italicized nucleotides) with N-nucleotide addition(underlined nucleotides) and variable exonuclease chewback of the D23 VH footprint (bolded nucleotides).

4 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1418001112 Sun et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

to demonstrate IgH receptor editing by VH replacement (21).Furthermore, a recent study by the Zhang group found thatantibodies isolated from patients suffering from autoimmunedisease and animals on autoimmune-prone backgrounds havehigher frequencies of VH replacement footprints, suggesting thatediting of the IgH repertoire either contributes to self-reactivespecificities or plays a role in editing of self-reactive Ig rear-rangements (11, 12). Intriguingly, the same group demonstratedthat immature B cells can be induced to undergo VH replace-ment by BCR cross-linking, implying that receptor editing by VHreplacement may occur after the B cells have egressed from thebone marrow (22). With our D23prod model, we can assess for anincrease in B cells that have undergone VH replacement in re-sponse to a self-antigen and determine the stage at which thesecondary rearrangements occur.We took advantage of the ssDNA/dsDNA autospecificity of

the paired D23 IgH and IgL chains to investigate whether VHreplacement can modify and edit autoantibodies (13). Nearly allB cells in D23prod/+ D23κ/+ mice expressed the κ IgL chain, andthe autospecificity to DNA resulted in a lower level of IgMa ex-pression compared with D23prod IgH chain expression alone(Fig. S4A). No increase in VH replacement was observed, how-ever, as indicated by the maintained low fraction of cells expressingthe WT IgH locus (IgMb). Consistent with a lack of editing at eitherIgH or IgL loci, we observed a severe reduction of pro-B and pre–B-cell compartments, given that the knock-in IgH and IgL chainsare readily expressed and mediate accelerated B-cell development(Fig. S4B).To introduce an even more robust autoreactive signal in D23prod

B cells, we bred the D23prod allele onto a κ-macroself back-ground. The κ-macroself allele encodes a ubiquitously expressedchimeric cell surface protein with a Fab fragment that spe-

cifically recognizes the mouse Igκ constant region (23). As a result,developing B cells expressing Igκ antibody will bind to κ-mac-roself and undergo IgL chain editing. In D23prod/+ mice, Igλ+ Bcells constitute <15% of total B cells (Fig. 5). In contrast, inκ-macroself mice, the majority of B cells are Igλ+, owing to re-ceptor editing triggered by cross-linking of all Igκ BCRs. Theremaining 10–20% of λ− κ− B cells are expressing primarily VλX(24). Similarly, in D23prod/+ κ-macroself mice, most B cellsexpressed Igλ, indicating that IgL chain editing had occurred;however, the ratio of IgMa+ to IgMb+ B cells remained un-changed compared with that in D23prod/+ mice, indicating thatthe frequency of VH replacement did not increase in response toautoreactivity. This finding suggests that VH replacement doesnot contribute significantly to the editing of self-reactive BCRspecificities.

Accumulation of IgMb+ Cells with Age. Despite the apparent ab-sence of a role for VH replacement in the editing of autoimmuneantibody specificities in our mouse model, there is a noticeableage-dependent increase in B cells that have undergone inacti-vating VH replacement of the knock-in locus and now express theWT IgH allele (Fig. 6). Newborn pups have nearly 0% of B cellsexpressing the WT IgH chain, but this steadily increases to 5–10% as the mice age. This suggests that in the D23prod mice,IgMb+ cells are positively selected for and maintained, likely toincrease the antibody repertoire. To address this issue more di-rectly, we examined germinal center B cells in Peyer’s patcheswhere B cells are confronted with microflora- and food-derivedantigens. Peyer’s patches of unimmunized D23prod/+ and WT micecontained comparable numbers of germinal center B cells (CD19+

B220+ Fas+ PNA+CD38lo), suggesting that VH replacement is ableto generate an antibody repertoire sufficiently diverse to recog-nize gut-associated antigens. Indeed, examination of the allotypeof the IgM+ germinal center B cells revealed that, in contrast toB cells not in the germinal center, the majority were IgMb+ and

94

4

IgMaIg

Mb

D23prod/+ (IgMa/IgMb)

0

20

40

60

80

100D23prod/+

IgMa sorted

D23prod/+IgMb sorted

D23prod/+total Ig+B cells

WTtotal Ig+ B cells

% o

f in-

fram

e re

plac

emen

t seq

uenc

es

n = 31 n = 27 n = 52 n = 43

Fig. 3. Out-of-frame VH replacement of the knock-in allele promotes en-dogenous IgH rearrangement. D23prod/+ splenic B cells were FACS-sortedaccording to their IgM allotype expression. The knock-in IgH locus was am-plified and sequenced using VH replacement-specific primers. Alignment ofIg genes was performed using the IgBlast algorithm, and in-frame/out-of-frame status was determined based on analysis of the CDR3 region. WT IgHrearrangements were amplified from splenic WT B cells using a degenerateVH primers and a reverse JH2 primer. Total number of sequences analyzedfrom sorted cells is listed for each compartment. Data are accumulated frommore than three independent experiments, with at least two mice fromeach genotype.

A

B

Fig. 4. Preferential use of proximal VH genes during VH replacement inD23prod mice. (A) Diagram of the VH gene families located upstream of theD23prod knock-in site. The J558 VH gene family is located most distally andthe VH7183 gene family (with interspersed Q52 VH genes) most proximallyto the knock-in rearrangement. (B) Ratio of J558 and VH7183 V genes usedin VH replacement events from WT, D23stop/D23stop, D23prod/D23prod splenicB cells, and D23prod/D23prod Igα signaling-deficient (ΔTm/ΔTm) pro-B cells. VH

replacement sequences were specifically amplified and analyzed as de-scribed above. IgH sequences from WT mice were amplified from splenicB cells. The number of sequences analyzed from sorted cells is listed; data arefrom two or more independent experiments.

Sun et al. PNAS Early Edition | 5 of 9

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

thus expressed the WT IgH allele, underscoring the importanceof VH replacement in diversification of the Ig repertoire in theknock-in mice and highlighting a possible selection mechanismfor B cells as the mice age (Fig. S5).

DiscussionWe previously described a mouse model with a nonproductiveVHDHJH rearrangement (D23stop) knocked into the IgH locus.Analysis of these mice revealed that VH replacement can rescueB cells that have undergone out-of-frame rearrangement on bothIgH alleles, and that although VH replacement is not as efficientas conventional 12/23-bp V(D)J recombination, it nonethelessis able to generate a large complement of B cells with a highlydiverse Ig repertoire (4). Furthermore, analysis of replacementjoints revealed that the majority of the cells were generated earlyin B-cell development, because many sequences carried evidenceof TdT activity. In the present study, we examined the potentialfor VH replacement to further diversify the Ig repertoire bymodifying a productively rearranged IgH locus and its possiblerole in editing the BCR in response to autospecificity. For thispurpose, we generated a mouse strain in which a productiveVHDHJH rearrangement is knocked into the IgH locus in a physi-ological manner. Our targeting strategy eliminated all remainingunrearranged D and J segments at the IgH locus, preventing anysecondary IgH rearrangement except VH replacement. Although weobserved a severe contraction in the D23prod pro–B-cell compart-ment, this stage represents the time when cells with a primary IgHrearrangement may linger in the pro–B-cell stage and undergo VHreplacement.Although B cells of D23prod/+ mice encode a functional IgH

chain and thus should express the corresponding VHDHJH seg-ment exclusively, we observed a small but significant populationof B cells expressing novel VHDHJH joints in peripheral lym-phoid organs. In cases where VH replacement resulted in anonproductive VHDHJH segment, the cells were able to rear-range and express the WT IgH allele, thereby diversifying the

antibody repertoire in this mouse model. Furthermore, someB cells that maintained the expression of the knock-in allele al-tered the specificity of the D23prod rearrangement through pro-ductive replacement of the knock-in VH gene. The lack of B cellsexpressing IgH from both alleles indicated that D23prod/+ pe-ripheral B cells maintain allelic exclusion.Sequencing of secondary VHDHJH rearrangements revealed

evidence for 5′ exonuclease activity at the site of VH invasion andTdT activity that further diversified the VH-to-VHDHJH jointthrough N-nucleotide addition. Other groups have analyzed thesequences of mouse and human VHDHJH joints for footprints ofVH replacement to assess the contribution of VH replacement tothe antibody repertoire (9, 11, 12). These studies distinguishedIgH genes with VH replacement by identifying 5- or 4-bp foot-prints at the VH-to-DH border and compared the frequency ofVH footprints at VH-DH joints with the frequency at which thesesequences are found at the DH-JH junction. However, becauseexonuclease and TdT activity in pro-B cells alter the VH-DHjunction, such approaches likely underestimate the frequency ofthese events, as demonstrated by our finding of VH replacementsequences with footprints of ≤3 bp.In addition, our previous analysis of VHDHJH sequences from

B cells of D23stop/D23stop mice demonstrated that many VH re-placement events did not carry a footprint of the recipient VH gene,because VH replacement may occur via DNA microhomology-mediated joining (4). The same study also highlighted that VHfootprints are counterselected at the pre-B to immature transi-tion owing to the presence of charged amino acids, specificallyarginine, commonly encoded by VH replacement footprints, fur-ther suggesting that assessment of the VH replacement prevalencebased on VH footprint detection may greatly underestimate itstrue frequency (4). Because the sequence of the recipient VHDHJHjoint is defined in our D23prod knock-in mice, we can confidentlyidentify all secondary rearrangement events and assess theircontribution to this mouse model’s antibody repertoire. Ournew mouse model allows us to directly assess the contributionof VH replacement to the further diversification of the primaryB-cell repertoire, as well as its potential to edit autoimmunespecificities.A significant fraction of VH replacement in D23stop/D23stop

mice is mediated by microhomology between the invading andrecipient VH genes, wherein the footprint of the original VHgene was missing and no N/P-nucleotides were observed (4). Incontrast, only two out of more than 100 VH replacement eventsthat we analyzed in D23prod mice appear to have been medi-ated by microhomology. The absence of VH replacement by

95

3

13 81 91

0.2

98

94

3

λ1,2,3

B220

IgM

b

IgMa

D23prod/+ k-self(IgMa/IgMb)

k-self(IgMb/IgMb)

D23prod/+(IgMa/IgMb)

Fig. 5. Lack of increase in secondary IgH rearrangement in B cells frommice expressing ubiquitous anti-κ receptor. Editing of the D23prod allele wasassessed in mice expressing a κ-macroself chimeric protein that cross-linksIgκ. Splenic B cells from 9- to 10-wk-old D23prod and κ-macroself mice wereassessed for surface expression of the IgL chain and IgM by flow cytometry.The increase in λ+ B cells highlights the dramatic editing at the IgL locus,whereas the lack of increase in WT IgH expression in D23prod/+ mice carryingthe κ-macroself allele indicates no change in IgH receptor editing. Data arerepresentative of three independent experiments, with at least four mice ineach group.

0 day

2-8 w

ks

2-3 m

onths

3.5-6

months

7-12 m

onths

> 12 m

onths0

5

10

15

% o

f IgM

b + B

cel

ls

Fig. 6. Accumulation of B cells expressing the WT IgH in D23prod/+ mice withage. Splenic B cells from D23prod/+ mice of various ages were assessed for theproportion of B220+ cells expressing IgMb (WT Ig allele) by FACS analysis (n =53). P < 0.0001, one-way ANOVA.

6 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1418001112 Sun et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

microhomology in D23prod mice suggests that the replacementevents occur exclusively in the presence of TdT. The high fre-quency of microhomology-mediated replacement events inD23stop/D23stop mice may either result from rearrangementsoccurring in the fetal liver in the absence of TdT or representrare events in the bone marrow that are selected for preferen-tially owing to their lack of the VH footprint and thus the lowerfrequency of the positively charged arginine residues in CDR3(4). The nearly complete absence of IgMb+ cells observed innewborn D23prod/+ mice suggests that any replacement eventsthat do occur in the fetal liver of these mice are in-frame, con-sistent with the fact that TdT is not expressed during fetal he-matopoiesis. However, because we failed to amplify any re-placement joints from fetal livers of D23prod/+ mice, we concludethat replacement events at this stage of development are rare inthis model.FACS analysis of B cells from adult D23prod/+ mice revealed

that 3–5% of B cells were using the WT IgH allele, havinginactivated the knock-in D23 allele by out-of-frame VH re-placement. In addition, B cells that underwent an allele-inacti-vating VH replacement may fail to acquire an in-frame rear-rangement on the WT IgH allele and thus perish in the bonemarrow, so that the frequency of replacement events may beeven higher. We could not assess the true extent of VH re-placement until we also examined the frequency of in-frame VHreplacement, however. In D23prod/+ total splenic B cells, weobserved that out-of-frame replacement events were far morefrequent than in-frame secondary rearrangements, with in-frame VH replacement accounting for only 5% of secondaryrearrangements.Owing to the stochastic nature of TdT and exonuclease ac-

tivity, only two-thirds of VH replacement events are expected tobe out-of-frame, but since the vast majority were out-of-frame,there must be selection against in-frame and/or selection forout-of-frame replacement. In-frame VH replacements may beselected against due to the prevalence of charged amino acidsencoded by the footprint bases (as discussed above). On theother hand, the out-of-frame replacement events allow de novorearrangement of the WT IgH locus, and thus significantlydiversify the antibody repertoire, from which specificitiesexpressed in peripheral B cells can be selected. Indeed, weobserved steady accumulation of B cells that had undergoneout-of-frame VH replacement and now expressed IgMb (WT)heavy chain as the animals aged; it was this population that waspreferentially recruited into germinal centers of gut-associatedimmune organs. The preferential participation of IgMb+ B cellsin the gut-associated germinal centers likely reflects the broaderantibody repertoire of this population. The accumulation of such“escapee” cells with age is frequently observed in IgH knock-inmouse models.Most VH replacements at the D23prod allele used VH genes

from the VH7183 V gene family, predominantly VHD6.96, theVH gene most proximal to the knock-in site. This finding is instriking contrast to our previous study of D23stop/D23stop mice,where the entire Ig repertoire was generated by secondary rear-rangements of the nonproductive IgH alleles. In that case, VHreplacement events used the full spectrum of VH genes, and therepertoire of VH elements used closely resembled that seen inprimary WT IgH rearrangements (4). Pro-B cells blocked fromprogressing in development, such as those from D23stop/D23stop

mice or from Igα signaling-deficient animals, have a broader win-dow of opportunity to undergo secondary rearrangement. In con-trast, cells that undergo VH replacement in D23prod animalslikely initiate secondary IgH rearrangement before receiving a sig-nal through the pre-BCR.The highly restricted VH use during replacement observed in

the D23prod mice may reflect a lack of contraction at the IgHlocus during the limited time that these cells remain in the

pro–B-cell compartment and may be related to chromatin inac-cessibility of the distal VH genes in the early stages of B-cell dif-ferentiation (25, 26). This hypothesis is supported by the observeddisparity in sterile transcription of J558 and VH7183 gene seg-ments in pro-B cells from D23stop/D23stop mice vs. D23prod/D23prod

mice. The strong bias toward proximal VH gene use disappearsin D23prod mice with a defect in Igα signaling, likely becausethe VH locus becomes more accessible as a consequence of theblocked progression from the pro–B- to the pre–B-cell stage.The observed bias toward DH-proximal VH genes in D23prod

mice is consistent with the fact that other mouse models har-boring productive IgH transgenes show inefficient exclusion ofthe most 3′ VH to DH gene, suggesting that the 3′ end of the IgHlocus is accessible during early pro–B-cell development (27–29).Furthermore, because theVH-DH region harbors several CTCF-binding elements critical for allelic exclusion and for regulationof IgH locus accessibility (30), it is conceivable that the loss ofCTCF sites or other locus regulatory elements resulting from ourgene targeting strategy increases the accessibility of the 3′ VHgenes prematurely, thus allowing VH replacement to occur ear-lier then it would in WT pro-B cells.Along with diversifying the antibody repertoire, VH replace-

ment has been suggested as a mechanism of receptor editing inimmature B cells (11, 22, 31). We examined the frequency of VHreplacement of the D23prod allele on two autoreactive back-grounds, first by pairing this IgH with its cognate IgL chainD23κ, resulting in an antibody with ssDNA/dsDNA autospeci-ficity, and then by expressing the D23prod knock-in in mice thatubiquitously express a protein that cross-links all BCRs that carrya Igκ, thus triggering massive IgL editing. In both cases, we ob-served no change in the frequency of VH replacement.Although a recent study found evidence of VH replacement

breaks occurring in B cells that recently emigrated from the bonemarrow (22), the presence of N-nucleotides in ∼99% of VH re-placement joints analyzed from D23prod mice suggests that re-placement occurs at the pro–B-cell stage, at which the expressionof Tdt is limited (32). Our conclusion is consistent with an earlieranalysis of cRSS cleavage throughout the B-cell developmentalstages (33). Our finding that the presence of a prerearrangedD23κ did not alter the frequency of VH replacement on theD23prod IgH chain further suggests that this process occurs inearly B-cell progenitors, before IgL chain expression. The in-creased frequency of VH replacement footprints in autoimmunerepertoires may reflect positive selection of the additional posi-tively charged amino acids encoded by the VH footprint in theCDR3 region (4, 34), rather than evidence suggesting VH re-placement as a mechanism for editing of the primary repertoirein response to autoantigens.The majority of earlier IgH knock-in mice were of limited use

in the study of secondary IgH rearrangements, because thesemice had VHDHJH joints knocked-in ∼180 kb downstream of themost 3′ VH gene, replacing the endogenous JH elements andleaving the intervening DH elements intact. In contrast, theD23prod allele closely resembled an endogenous IgH rearrange-ment, allowing us to study the secondary rearrangement of aproductive VHDHJH joint under quasi-physiological conditions.The use of mice generated by nuclear transfer using B cells asdonors, as well as mice produced using iPS cells following thereprogramming of mature B lymphocytes, may provide furtherinsight into VH replacement, because B-cell progenitors in thesemice will have prerearranged Ig loci (35–37). Indeed, studies bythe Casola group (see the accompanying paper by Kumar et al.,ref. 38) reveal that animals generated by transnuclear transferfrom an IgA+ plasma cells have peripheral B cells with a veryhigh frequency of VH replacement of the original IgH rear-rangement. Those studies demonstrated that the contribution ofVH replacement to the diversification of the primary immunerepertoire is not limited to the D23prod knock-in mice. The

Sun et al. PNAS Early Edition | 7 of 9

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

frequency of replacement events may vary between the two modelsowing to the differences in VHDHJH substrate sequence, thechromosomal location of the primary rearrangement relative tothe available donor VH genes or in relation to transcriptionregulatory sites, or to differential signaling properties of IgM andIgA cytoplasmic tails.In summary, our analysis of secondary rearrangements in a

mouse model with a naturally positioned IgH knock-in demon-strates that VH replacement can modify productively rearrangedVHDHJH joints as a mechanism of antibody repertoire diver-sification. Extrapolating from the D23prod allele to the entireantibody repertoire, the overall contribution of this secondarydiversification process is apparently minor, and the physiologicalimpact of VH replacement may be more significant for rescuingB-cell progenitors with nonproductive IgH rearrangements (4).In either case, however, VH replacement occurs in B progenitorcells independent of antigen specificity and is unlikely to con-tribute to receptor editing in the context of the counterselectionof autoreactive BCRs in pre-B or immature B lymphocytes.

Materials and MethodsGene Targeting. Our gene targeting strategy was similar to the previouslypublished D23stop targeting strategy (4). The rearranged D23 IgH genesegment was isolated by PCR amplification, and its promoter was cloned bya PCR walking strategy. The D23 VDJ together with the promoter was clonedinto the targeting vector 1 (Fig. 1A) containing a floxed neo cassette flankedby homology to the JH region. To generate targeting vector 2 (Fig. 1A),a double LoxP followed by an frt-flanked neomycin-resistance cassette wasinserted into a vector containing homology to the VH 81X gene. (The ori-entation of the VH element was not known at the time of targeting.) Se-quential targeting of first the double LoxP and then the D23 VDJ resulted inthe targeted IgH locus. The chimera with transmission of the targeted alleleinto the germline was crossed to the deleter strain to eliminate the in-tervening DH elements by Cre activity in germ cells (39).

Preparative and Analytical FACS. Fluorescence staining was performed as de-scribed previously (4). Cells isolated from κ-macroself mice were incubated in

unlabeled mouse IgG/κ antibody for blocking before antibody staining. Cellsorting was performed using a triple-laser flow cytometer (BD FACSVantage)and a four-laser flow cytometer (BD FACSAria). Analytical FACS was carriedout on a four-laser flow cytometer (BD LSRFortessa and LSRII).

PCR and RT-PCR. DNA from sorted populations was prepared by proteinase Kdigestion in Tris buffer for 3 h. IgH rearrangements were amplified usingamix of previously described forward primers and a reverse primer specific tothe D23H D–J sequence (5′-CCCCAGTAGTCAAAGTACC-3′) or a JH2 primer (5′-AGACTGTGAGAGTGGTGCC-3′) (40). High-fidelity polymerase was used tominimize point mutations during the PCR analyses (Roche). The expected-sized band (∼380 bp) was excised and gel-purified before being insertedinto the pGEM-T easy vector by TA cloning (Promega). The subclonedDNA was prepared by column purification (Qiagen) and sequenced usingT7 primer (Macrogen).

RNA was isolated from sorted populations by column purification (QiagenRNeasy Micro Kit). cDNA was prepared immediately using random hexamerprimers (Superscript VILO; Life Technologies). Forward and reverse primersspecific to but degenerate within the J558 (forward: 5′-CTTCTGGCTACA-CCTTCAC-3′; reverse: 5′-CTGAGCTGCATGTAGGCTG-3′) and 7183 (forward:5′- GCCTCTGGATTCACTTTCAG-3′; reverse: 5′-CATTGTCTCTGGAGATGGTG-3′)V gene families were used to quantitatively amplify germline transcriptionof V genes.

Mice. Mice were housed and cared for under specific pathogen-free con-ditions in accordance with protocols approved by the Animal Care and UseCommittees of Harvard Medical School and NYU School of Medicine.

Sequence Analysis of IgH Rearrangements. VDJ sequences were analyzedusing the National Center for Biotechnology Information’s IgBlast program(Dataset S1).

ACKNOWLEDGMENTS. We thank Stefano Casola for the open and enjoyablediscussions on the complementary projects. This work was supported by theArnold and Mabel Beckman Foundation, the Ralph S. French CharitableFoundation, and the National Institutes of Health [Grant R21AI110830(to S.B.K.)]. A.S. was supported by National Institutes of Health Grant5T32GM007308-36 and National Cancer Institute Grant 5T32CA009161-39.

1. Tonegawa S (1983) Somatic generation of antibody diversity. Nature 302(5909):575–581.2. Melchers F, et al. (1994) Roles of IgH and L chains and of surrogate H and L chains in the

development of cells of the B lymphocyte lineage. Annu Rev Immunol 12(1):209–225.3. Herzog S, Reth M, Jumaa H (2009) Regulation of B cell proliferation and differenti-

ation by pre-B-cell receptor signalling. Nat Rev Immunol 9(3):195–205.4. Koralov SB, Novobrantseva TI, Königsmann J, Ehlich A, Rajewsky K (2006) Antibody rep-

ertoires generated by VH replacement and direct VH to JH joining. Immunity 25(1):43–53.5. Lutz J, Müller W, Jäck HM (2006) VH replacement rescues progenitor B cells with two

nonproductive VDJ alleles. J Immunol 177(10):7007–7014.6. Kleinfield R, et al. (1986) Recombination between an expressed immunoglobulin

heavy-chain gene and a germline variable gene segment in a Ly 1+ B-cell lymphoma.Nature 322(6082):843–846.

7. Reth M, Gehrmann P, Petrac E, Wiese P (1986) A novel VH to VHDJH joining mechanismin heavy-chain-negative (null) pre-B cells results in heavy-chain production. Nature322(6082):840–842.

8. Jung D, Giallourakis C, Mostoslavsky R, Alt FW (2006) Mechanism and control of V(D)Jrecombination at the immunoglobulin heavy chain locus. Annu Rev Immunol 24:541–570.

9. Zhang Z, et al. (2003) Contribution of Vh gene replacement to the primary B-cellrepertoire. Immunity 19(1):21–31.

10. Zhang Z, Burrows PD, Cooper MD (2004) The molecular basis and biological signifi-cance of VH replacement. Immunol Rev 197(1):231–242.

11. Huang L, et al. (2013) Contribution of V(H) replacement products in mouse antibodyrepertoire. PLoS ONE 8(2):e57877.

12. Lange MDH, et al. (2014) Accumulation of VH replacement products in IgH genes derivedfrom autoimmune diseases and anti-viral responses in human. Front Immunol 5:345.

13. Baccala R, Quang TV, Gilbert M, Ternynck T, Avrameas S (1989) Two murine naturalpolyreactive autoantibodies are encoded by nonmutated germ-line genes. Proc NatlAcad Sci USA 86(12):4624–4628.

14. Koralov SB, Novobrantseva TI, Hochedlinger K, Jaenisch R, Rajewsky K (2005) Direct invivo VH to JH rearrangement violating the 12/23 rule. J Exp Med 201(3):341–348.

15. Torres RM, Flaswinkel H, Reth M, Rajewsky K (1996) Aberrant B-cell developmentand immune response in mice with a compromised BCR complex. Science 272(5269):1804–1808.

16. Roth DB, Roth SY (2000) Unequal access: Regulating V(D)J recombination throughchromatin remodeling. Cell 103(5):699–702.

17. Alessandrini A, Desiderio SV (1991) Coordination of immunoglobulin DJH transcriptionand D-to-JH rearrangement by promoter-enhancer approximation. Mol Cell Biol 11(4):2096–2107.

18. Yancopoulos GD, Blackwell TK, Suh H, Hood L, Alt FW (1986) Introduced T- cell re-

ceptor variable region gene segments recombine in pre-B cells: Evidence that B and

T cells use a common recombinase. Cell 44(2):251–259.19. Nemazee D (1999) Receptor editing in B cells. Advances in Immunology, ed Frank JD

(Academic Press, San Diego), Vol 74, pp 89–126.20. Halverson R, Torres RM, Pelanda R (2004) Receptor editing is the main mechanism of

B-cell tolerance toward membrane antigens. Nat Immunol 5(6):645–650.21. Chen C, Nagy Z, Prak EL, Weigert M (1995) Immunoglobulin heavy chain gene re-

placement: A mechanism of receptor editing. Immunity 3(6):747–755.22. Liu J, et al. (2013) Regulation of VH replacement by B-cell receptor-mediated sig-

naling in human immature B cells. J Immunol 190(11):5559–5566.23. Ait-Azzouzene D, et al. (2005) An immunoglobulin C κ-reactive single chain antibody

fusion protein induces tolerance through receptor editing in a normal polyclonal

immune system. J Exp Med 201(5):817–828.24. Dildrop R, Gause A, Müller W, Rajewsky K (1987) A new V gene expressed in lambda-2

light chains of the mouse. Eur J Immunol 17(5):731–734.25. Fuxa M, et al. (2004) Pax5 induces V-to-DJ rearrangements and locus contraction of

the immunoglobulin heavy-chain gene. Genes Dev 18(4):411–422.26. Stanhope-Baker P, Hudson KM, Shaffer AL, Constantinescu A, Schlissel MS (1996) Cell

type-specific chromatin structure determines the targeting of V(D)J recombinase ac-

tivity in vitro. Cell 85(6):887–897.27. Kurosawa Y, et al. (1981) Identification of D segments of immunoglobulin heavy-

chain genes and their rearrangement in T lymphocytes. Nature 290(5807):565–570.28. Iacomini J, Yannoutsos N, Bandyopadhay S, Imanishi-Kari T (1991) Endogenous im-

munoglobulin expression in mu transgenic mice. Int Immunol 3(2):185–196.29. Costa TE, Suh H, Nussenzweig MC (1992) Chromosomal position of rearranging gene

segments influences allelic exclusion in transgenic mice. Proc Natl Acad Sci USA 89(6):

2205–2208.30. Guo C, et al. (2011) CTCF-binding elements mediate control of V(D)J recombination.

Nature 477(7365):424–430.31. Itoh K, et al. (2000) Immunoglobulin heavy chain variable region gene replacement as

a mechanism for receptor revision in rheumatoid arthritis synovial tissue B lympho-

cytes. J Exp Med 192(8):1151–1164.32. Li YS, Hayakawa K, Hardy RR (1993) The regulated expression of B lineage- associated

genes during B-cell differentiation in bone marrow and fetal liver. J Exp Med 178(3):

951–960.

8 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1418001112 Sun et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022

33. Davila M, et al. (2007) Multiple, conserved cryptic recombination signals in VH genesegments: Detection of cleavage products only in pro-B cells. J Exp Med 204(13):3195–3208.

34. Sekiguchi DR, Eisenberg RA, Weigert M (2003) Secondary heavy chain rearrangement:A mechanism for generating anti-double-stranded DNA B cells. J Exp Med 197(1):27–39.

35. Dougan SK, et al. (2013) Antigen-specific B-cell receptor sensitizes B cells to infectionby influenza virus. Nature 503(7476):406–409.

36. Dougan SK, et al. (2012) IgG1+ ovalbumin-specific B-cell transnuclear mice show classswitch recombination in rare allelically included B cells. Proc Natl Acad Sci USA109(34):13739–13744.

37. Hanna J, et al. (2008) Direct reprogramming of terminally differentiated matureB lymphocytes to pluripotency. Cell 133(2):250–264.

38. Kumara R, et al. (2015) Antibody repertoire diversification through VH gene re-placement in mice cloned from an IgA plasma cell. Proc Natl Acad Sci USA, 10.1073/pnas.1417988112.

39. Schwenk F, Baron U, Rajewsky K (1995) A cre-transgenic mouse strain for theubiquitous deletion of loxP-flanked gene segments including deletion in germcells. Nucleic Acids Res 23(24):5080–5081.

40. Ehlich A, Martin V, Müller W, Rajewsky K (1994) Analysis of the B-cell progenitorcompartment at the level of single cells. Curr Biol 4(7):573–583.

Sun et al. PNAS Early Edition | 9 of 9

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

11, 2

022