of August 30, 2018.This information is current as

E2A in a VDJ Recombination ComplexAllelic Exclusion of IgH through Inhibition of

GrundströmJannek Hauser, Christine Grundström and Thomas

http://www.jimmunol.org/content/192/5/2460doi: 10.4049/jimmunol.1302216January 2014;

2014; 192:2460-2470; Prepublished online 27J Immunol 

MaterialSupplementary

6.DCSupplementalhttp://www.jimmunol.org/content/suppl/2014/01/25/jimmunol.130221

Referenceshttp://www.jimmunol.org/content/192/5/2460.full#ref-list-1

, 18 of which you can access for free at: cites 52 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2014 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

Allelic Exclusion of IgH through Inhibition of E2A in a VDJRecombination Complex

Jannek Hauser, Christine Grundstrom, and Thomas Grundstrom

A key feature of the immune system is the paradigm that one lymphocyte has only one Ag specificity that can be selected for or

against. This requires that only one of the alleles of genes for AgR chains is made functional. However, the molecular mechanism of

this allelic exclusion has been an enigma. In this study, we show that B lymphocytes with E2A that cannot be inhibited by calmodulin

are dramatically defective in allelic exclusion of the IgH locus. Furthermore, we provide data supporting that E2A, PAX5, and the

RAGs are in a VDJ recombination complex bound to key sequences on the Igh gene. We show that pre-BCR activation releases the

VDJ recombination complex through calmodulin binding to E2A. We also show that pre-BCR signaling downregulates several

components of the recombination machinery, including RAG1, RAG2, and PAX5, through calmodulin inhibition of E2A. The

Journal of Immunology, 2014, 192: 2460–2470.

The diversity of AgRs is generated by each B and T cellprecursor assembling one variable (V), one joining (J),and, on some receptor genes, a diversity (D) gene segment

from vast arrays of these segments. This V(D)J recombination of Iggenes starts with IgH. Most IgHs do not become functional becauseof the random nature of the fusion of gene segments in V(D)J re-combination. Therefore, the expression of a pre-BCR with membrane-bound IgH associated with surrogate L chain proteins is a criticalcheckpoint that monitors for functional IgH rearrangement. Later inthe development, the functional rearrangement of Ig L chain genesegments is verified by the complete Ab of the BCR. V(D)J re-combinations are initiated by introduction of dsDNA breaks by therecombination activating gene enzymes RAG1 and RAG2 at re-combination signal sequences (RSSs). The RAG1/RAG2 complexbinds sequence specifically to RSS sequences, and RAG2 specifi-cally recognizes histone H3 trimethylated at lysine 4 (H3K4me3).Recognition of RSSs and H3K4me3 and accessibility of the chro-matin are all necessary for V(D)J recombination (1–4). DNA elementsregulating the accessibility include promoters and enhancers, andsome transcription factors were shown to play a role (1–4). Thestrongest link with a transcription factor is for E2A, which has thetwo splice forms E12 and E47 (3). These bind to “E-box” sequencesin Ig genes and are required for both RAG expression and Ig re-combination. E-boxes in the Igk L chain intronic enhancer arecrucial for activation of V-J rearrangement at Igk loci (5), and E12and E47 regulate the sequential rearrangement of the Ig L chain loci(6). Ectopic expression of E2A, together with RAG1/RAG2, induces

D to J recombination at Igh loci in cells that normally never undergoV(D)J recombination (7). The V segments of the Igh locus havebinding sites for several transcription factors, including E2A andPAX5 (8, 9), and PAX5 is essential in inducing and orchestratingIgh recombination in the V region (10–14). Embryonic kidney cellsshow V to DJ recombination at the Igh loci exclusively whenRAG1/RAG2 and E2A are ectopically coexpressed with PAX5(14). Therefore, RAG1, RAG2, PAX5, and E2A are all indis-pensable for VDJ recombination of the Igh locus.A key feature of the immune system is the paradigm that one

lymphocyte has only one Ag specificity that can be selected for oragainst to enable targeting of the pathogen while avoiding collateraldamage and wasted resources. The monospecificity requires thatonly one allele of genes for AgR chains ismade functional. This allelicexclusion requires that only one allele at a time undergoes completeV(D)J recombination. Multiple mechanisms most likely contribute toallelic exclusion, but the relative role of the proposed models remainscontroversial (15–17). The allelic-exclusion system ensures that suc-cessful V(D)J recombination on one allele shuts off recombination onthe other allele by feedback inhibition to prevent a second successfulrearrangement (15–17). This feedback inhibition from AgRs has anestablished major role in allelic exclusion, but little is known aboutthemolecularmechanism(s) bywhich it is achieved (15–17).However,expression of RAG1 and RAG2 is lower in pre-B cells expressing thepre-BCR than in the preceding and succeeding developmental stages,suggesting negative feedback of these key recombination proteins(18), but the molecular mechanism(s) of the allelic exclusion by feed-back inhibition has remained largely enigmatic.When the main Ca2+-sensor protein calmodulin (CaM) is Ca2+

loaded, it can bind to and inhibit transcriptional activities of E2A(19–21). In this study, we analyzed expression of the two Ighalleles in B cells expressing CaM-resistant (CaMR) E2A and showthat they are dramatically defective in allelic exclusion. Onemechanism of the dependence of allelic exclusion on CaM inhi-bition of E2A was shown to be that pre-BCR activation down-regulates several components of the recombination machinery,including RAG1, RAG2, and PAX5, through CaM inhibition ofE2A. Furthermore, data support that E2A, PAX5, and the RAGsare in a complex bound to key sequences on the Igh gene, and theyinstead became bound to CaM after pre-BCR activation. The re-combination complex is directly released from the key Igh sitesthrough CaM inhibition of E2A.

Department of Molecular Biology, Umea University, SE-901 87 Umea, Sweden

Received for publication August 19, 2013. Accepted for publication December 20,2013.

This work was supported by grants from the Swedish Research Council and theSwedish Cancer Society.

Address correspondence and reprint requests to Dr. Thomas Grundstrom, Departmentof Molecular Biology, Umea University, SE-901 87 Umea, Sweden. E-mail address:Thomas.Grundstrom@molbiol.umu.se

The online version of this article contains supplemental material.

Abbreviations used in this article: B6, C57BL/6; CaM, calmodulin; CaMR, CaMresistant; ChIP, chromatin immunoprecipitation; H3K4me3, histone H3 trimethylatedat lysine 4; PLA, proximity ligation assay; RSS, recombination signal sequence; shRNA,short hairpin RNA; TdT, terminal deoxynucleotidyl transferase; WT, wild-type.

Copyright� 2014 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/14/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1302216

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Materials and MethodsMice and allelic-exclusion analysis

C57BL/6 (B6) male mice heterozygous for E2A knockout (E2A+/2) (22)were crossed with BALB/c females (Taconic, Ry, Denmark). Total bonemarrow was isolated from resulting E2A+/2 F1 mice at age 6–8 wk andimmediately infected with retrovirus expressing wild-type (WT) or CaMR

mutant E12 in culture with complete RPMI 1640 medium, IL-7 (10 ng/ml;PeproTech), and polybrene (10 mg/ml). Twenty-four hours postinfection,bone marrow cells were supplemented with fresh total bone marrow(a quarter of the infected amount) from B6 3 BALB/c F1 E2A+/2 mice, andthe mixture was transplanted into lethally irradiated (9.5 Gy) B63 BALB/cF1 E2A+/2 recipients at age 6–8 wk (up to 2 3 107 cells in 150–200 mlHBSS/mouse) by tail-vein injection. Recipient mouse spleen cells werecollected 2–3 wk posttransplantation. The effect of E12 WT and mutantexpression on allelic exclusion was analyzed by FACS using the IgHallotype-specific Abs IgMA-PE (specific for IgH of the BALB/c strain;clone DS-1) and IgMB-biotin (specific for IgH of the B6 strain; clone AF6-78), anti–CD19-allophycocyanin (clone 1D3), and streptavidin-PerCP (allfrom BD Biosciences) or streptavidin–PerCP–Cy5.5 (eBioscience). Allcells were treated with CD16/CD32 Fc Block before staining, and deadcells were excluded by propidium iodide (both from BD Biosciences). Flowcytometry was performed with a FACSCalibur, and cell sorting was donewith a FACSAria III instrument (both from BD Biosciences). IgMA+ IgMB+

double-positive B cells, as displayed in Fig. 1A, were sorted at 4˚C directlyinto culture in 500 ml complete RPMI 1640 medium supplemented with20% FCS in a 24-well culture plate containing a layer of S17 feeder cells,irradiated at 30 Gy. Approximately 20,000–50,000 double-positive cellswere isolated from a total of 10–20 million splenocytes. After sorting, themedium was adjusted to 1 ml with 25 mg/ml LPS (Calbiochem) and 10%FCS, and the cells were cultivated for 3 d before reanalysis by FACS andimmunostaining. All mouse research was carried out under a Swedishproject license and was subject to local ethical review.

Plasmids, viruses, cell culture, and transfections

The human pre-B cell line Nalm-6 was maintained in RPMI 1640 mediumsupplemented with 5% FCS and antibiotics. The EBV-based pMEP4 shuttlevector expressing a short hairpin RNA (shRNA) that interferes with E12/E47-specific human mRNA and the pMEP4 derivatives encoding WT E12 and theCaMR mutant m8N47 of E12 were described previously, and the MSCV-IRES-GFP–based retroviruses were described previously (21, 23, 24). Nalm-6cells transfected to express pMEP4 derivatives were selected with Hygro-mycin B (Roche) for 5 d, as described previously (25). Their pre-BCR wasstimulated for 5 h, unless otherwise specified, with 5 mg goat F(ab9)2 anti-human IgM (Southern Biotechnologies). For the C-terminal Flag con-structs, the GFP-encoding DNA of the MSCV-IRES-GFP plasmid wasreplaced by oligonucleotides comprising a tandem repeat of the Flagepitope–coding sequence (21) and restriction enzyme sites for insertionof full-length cDNAs of RAG2 and PAX5 (GenScript). E12 WT andCaMR mutant, with or without the C-terminal Flag epitope, were clonedinto MSCV-IRES, as described (23). Production of retroviruses was in293T cells, as described (23).

Activation, infection, and analysis of pre-B lymphocytes frommouse bone marrow

Bone marrow cells were collected by flushing femur and tibia of C57BL/6mice. The cells were grown on S17 stromal cells with IL-7, and the pre-BCRof pre-B cells, where indicated, was activated by goat F(ab9)2 anti-mouseIgM (5 mg/ml; 1022-01; Southern Biotechnologies), as previously de-scribed (25). Pre-B lymphocytes were purified from the bone marrow cellswith biotinylated rat Ab to CD25 (7D4; BD Pharmingen) using magneticanti-biotin MicroBeads, a MidiMACS Separator, and LS columns (all fromMiltenyi Biotec), according to the manufacturer’s recommendations. In-fections with MSCV-IRES-GFP–based retroviruses were by centrifugationin the presence of polybrene (5 mg/ml), as previously described (25),followed by 18 h of culture in fresh complete medium with elevated IL-7(25 ng/ml).

FACS and DNA-binding analysis

Intracellular immunostaining of harvested cells was performed with 2%paraformaldehyde and ethanol, as previously described (26). Flow cytometrywas performed with a FACSCalibur instrument, and analysis was done withCellQuest software (Becton Dickinson Biosciences). Nuclear extracts wereprepared, as previously described (21), from the Nalm-6 human pre-B cellline, and EMSAs were performed as described (20). The EMSAs wereperformed with probe sequences (E2A binding sites are underlined) from

the human RAG1 promoter, 59-AAGAGCCAGGTGGCAGCTGGAGCTG-39; the RAG2 promoter, 59-GCCAAGGCAGGTGGGTCACC-39; the RAGenhancer Erag, 59-AGAGGGACAGCTGCTTTCTA-39; and the TDT pro-moter, 59-AGGGCAGCACCTGTGAAGCC-39 (27–30). The control probefrom the human GM-CSF promoter was described previously (25).

Real-time RT-PCR

RNA was extracted using TRIzol reagent (Invitrogen), and quantitativereal-time RT-PCR analysis was performed, as previously described, usingGAPDH as internal control (23). The specificity of the PCR amplificationswas verified by melt-curve analysis, as described by the manufacturer(Bio-Rad). Some of the primer pairs used were described previously(25, 31). The real-time RT-PCR primer pairs for mouse and human genesthat have not been described previously were as follows: mouse Rag1:59-GACAAAGAAGAAGGTGGAGATG-39 and 59-AGGAAGGTATTGA-CACGGATG-39; mouse Rag2: 59-TTGCCCTACTTGTGATGTTG-39 and59-ACTTGTTGCTTCCTTCTGAC-39; mouse TdT: 59-GACATCTCCTG-GCTCATCG-39 and 59-TGGTTCTTCGCTGGCAAG-39; mouse Pax5: 59-AGTCTCCAGTGCCGAATG-39 and 59-TCCGTGGTGGTGAAGATG-39;mouse Ets1: 59-CACAACACAAATCCAAGAGTC-39 and 59-GGTGGT-AACAGGTCAAGTG-39; human RAG1: 59-CCAGCAAAAGAGTGCA-ATGA-39 and 59-TCAGCGACAGAAGATGTTGG-39; human RAG2: 59-CAGACAACATTCACAAACAG-39 and 59-ACATCACAAGTAGGGCA-G-39; human TDT: 59-GGGAAAAGAAGGGATTACTT-39 and 59-TC-TTTGACGAGGCAATTTGA-39; human FOS: 59-GAGCACGGAGGA-GGGATGAG-39 and 59-AGTCTCCAGAATGAACTCGCTTT-39; humanOCT2: 59-AGTGATCTGGAGGAGCTGGAG-39 and 59-CTCGCGGCA-TCTATCAACTGG-39; and human IL6: 59-GTCCAGTTGCCTTCTCCCT-39 and 59-GCCTCTTTGCTGCTTTCAC-39.

Proximity ligation assay

Harvested mouse primary pre-B cells were fixed with 4% paraformalde-hyde, permeabilized with 0.2% Triton X-100, and blocked with 5% FCS inPBS. Proximity ligation assay (PLA) (32–35) was performed with theDuolink system using in situ PLA kits from Olink Biosciences. In brief,cells were stained for intracellular proteins using Abs, as listed below.Secondary Abs conjugated with oligonucleotides (PLA probes) weresubsequently used, according to the manufacturer’s protocol, to generatefluorescence signals only when the two PLA probes were in close prox-imity (#40 nm). The fluorescence signal from each detected pair of PLAprobes was visualized as a distinct individual red dot in an epifluorescencemicroscope (Nikon 90i). The Abs were used at dilutions that yielded a lownumber of dots/cell to have a low risk for missing dots in the quantifica-tions. Nuclei were counterstained with DAPI dye. To enhance detection ofinfected cells, anti–GFP-FITC Ab (Novus Biologicals) was added duringthe PLA signal-detection step.

Chromatin immunoprecipitation

Infected primary mouse pre-B cells from E2A+/2 B6 mice were cross-linkedwith 1% formaldehyde for 10 min at room temperature. Cross-linking wasquenched with a 1/20 volume of 2.5 M glycine, and cells were washed withPBS. Cells were lysed on ice with 250 ml RIPA buffer (50 mM Tris-HCl[pH 8], 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxy-cholate, and 5 mM EDTA) supplemented with protease inhibitor mixture(Roche Applied Science). The lysate was sonicated at 60% amplitude for103 10 s on a Vibra-Cell instrument (Sonics) and clarified by centrifugation.The average size of sheared chromatin fragments was determined by agarosegel electrophoresis to be 180 bp. A total of 5 ml was saved as input, and thelysate was incubated overnight at 4˚C with mouse anti–Flag-tag Ab (Gen-Script). Immunocomplexes were captured for 1 h by 20 ml magneticDynabeads Protein A (Invitrogen) at 4˚C. Beads were washed three timeswith RIPA buffer and three times with Tris-EDTA (pH 8) on a magnet(Invitrogen). Cross-links were reversed with 1% SDS in Tris-EDTA (pH 8) at65˚C overnight, and samples were treated with Proteinase K (20 mg/ml) for1 h at 45˚C. DNAwas phenol-chloroform extracted, ethanol precipitated, andanalyzed by quantitative real-time PCR. Each sample was quantified relativeto its input chromatin DNA. The following primers were used for thechromatin immunoprecipitation (ChIP) analysis of the mouse Igh gene:Em: 59-TCAGAACCAGAACACCTGCAGCA-39 and 59-GGTGGGGC-TGGACAGAGTGTTTC-39; Hs3b: 59-TGGTTTGGGGCACCTGTGC-TGAG-39 and 59-GGGTAGGGCAGGGATGTTCACAT-39; VHA.1: 59-CCTTCGCCCCAATCCACC-39 and 59-CAAGTAACCCTCAAGAGAA-TGGAGACTC-39; VH10: 59-CTCTCTGCCAATGTAGGACCAG-39 and59-GCCTGAATTTCAGGGTCAGGG-39; VHJ558.47 (2): 59-CTACAAC-CAGAAGTTCAAGGGCAA-39 and 59-TCAGGCTGTGATTACAACACTG-TGT-39; DQ52: 59-AGTCCCTGTGGTCTCTGA-39 and 59-AGCCTCTCTA-

The Journal of Immunology 2461

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

CTTCCTCA-39; JH1: 59-GCTCAGGTCACTCAGGTCAGG-39 and 59-GA-GAATATCTTTTCCCGTTTCAGA-39; JH2: 59-TACTTTGACTACTGGGG-C-39 and 59-CCCTAGTCCTTCATGACC-39; JH3: 59-GCCTGGTTTGCTT-ACTGG-39 and 59-GACAAAGGGGTTGAATCT-39; JH4: 59-CACCAGGA-ATTGGCATAA-39 and 59-CCTGAGGAGACGGTGACT-39; V1: 59-GGTG-AAGCTGGTGGAATCTGGAGGA-39 and 59-CTCCATGTAGAAATCACT-GAAGG-39; V11: 59-GGTGAAGCTGGTGGAATCTGGAGGA-39 and 59-ATAAAACCCAACCACTCAAGTGCCTTTC-39; V13: 59-CAGTGTGAGG-TGAAGCTGATGGAG-39 and 59-ATCAAAGCCAACCACTCAGGTGACT-TCC-39; VH7183: 59-GCTGGTGGAGTCTGGGGGA-39 and 59-TGGTATT-GTCTCTCGAGATGATGAATCG-39; VHJ558: 59-GTGAAGATGTCCTGC-AAGGCTTCT-39 and 59-GGATTTGTCTGCAGTCAGTGTGGCCTTG-39;VH7183 (2): 59-CTCCAGAGACAATACCAAGAAGACCC-39 and 59-CCC-CCTGCTGGTCCTAGATG-39; VAR34: 59-CCTGGGATGTCACTGATATA-CACTCTG-39 and 59-GTAGTAGCCAGTAAATGAGTAACCAGAAGC-39;VHJ558.47: 59-GCAAGGCTTCTGGATACACATTCAC-39 and 59-CCTTG-CCTTTGAACTTCTCATTGTACTTAG-39; VHSM7: 59-CGTTATCCTCAT-TGCTACTACCACC-39 and 59-CCAAGTCCTAACTCTGTCTGAAGAAC-39;VH81X: 59-GAGATGAGATTCTGTCTGTTGTATGCAC-39 and 59-CTGC-AAACAAAGAGTGCTGGTCAG-39; DFL16.1: 59-CCCCCAATGCTCTCT-ACA-39 and 59-TATAAACCCCGTGATGCC-39; Cm: 59-CACCATTTCCTT-CACCTG-39 and 59-TGTTTTTGCCTCCGTAGT-39; and Cd: 59-CCATCA-CTTTTTGTCCAT-39 and 59-AGCAAGAGGTGTAAGGTT-39.

The Em and hs3b primers were from (36); the VHA.1 and VH7183 (2)primers were from (11); the VH10, VAR34, VHJ558.47, VHSM7, and VH81Xprimers were from (37); the DQ52, JH1, JH2, JH3, and JH4 primers were from(38); and the V1, V11, V13, VH7183, VHJ558, DFL16.1, Cm, and Cd primerswere from (14).

Western blot

Nuclear extracts were prepared as previously described (21), and Westernblot of purified mouse pre-B cells was performed using SuperSignalChemiluminescence Substrate (Pierce), according to the manufacturer’sinstructions.

Abs

Abs and related reagents used included IgMA-PE (DS-1), IgMB-biotin (AF6-78), CD19-allophycocyanin and CD19-allophycocyanin-H7 (1D3), CD16/CD32Fc-block (2.4G2), CD25-biotin (7D4), and SAv-PerCP (all from BD Bio-sciences); SAv–eFluor 660, SAv-allophycocyanin, and SAv-PerCP-Cy5.5(eBioscience); anti–PE-TRITC (AP09206TC-N; Acris Abs); terminal deoxy-nucleotidyl transferase (TdT; A1-84083; Affinity BioReagents); anti-goat–allophycocyanin, anti-rabbit–PE, anti-rabbit–allophycocyanin, anti-mouse–PE,and anti-mouse–allophycocyanin (Jackson ImmunoResearch); anti–GFP-FITC(NB100-1771; Novus Biologicals); and anti–DYKDDDDK-tag (Flag; A00187;GenScript). The following Abs are labeled with superscript numbers whererequired to distinguish them when two or more primary Abs against the sameprotein were used in PLA: CaM (N-19), CaM2 (FL-149), E2A (V-18), E2A2

(Yae), E2A3 (H-208), E2A4 (N-649), RAG1 (P-20), RAG12 (H-300), RAG22

(D-20), RAG23 (M-300), PAX5 (C-20), PAX52 (E-9), ETS1 (N-276), and SMAD3(38-Q) (all from Santa Cruz Biotechnology); RAG2 (11825-1-AP; ProteintechGroup); and H3K4me3 (ab8580), H3K4me32 (ab1012), and H3K4me33

(ab71998) (Abcam). PAX52 Ab was biotinylated when used with PAX5 Ab.Secondary PLA probes were used as anti-goat/anti-rabbit, anti-goat/anti-biotin,or anti-rabbit/anti-mouse combinations, depending on each primary Ab pair.

Single-cell PCR

Single GFP+ IgMA+ IgMB+ apparently double-positive B cells of experi-ment 2 of Fig. 1B (as shown in Fig. 1A) were deposited directly into wellsof a PCR plate (Sarstedt) containing 20 ml PCR buffer (Bio-Rad) and 0.5mg/ml Proteinase K (PCR grade; Roche). To prepare DNA, plates wereincubated at 50˚C for 1 h, followed by heat inactivation at 95˚C for 15 min.Igh rearrangements were detected by two rounds of PCR amplificationwith semi-nested primers (39). In the first round, PCR amplification wasperformed with 20 ml single-cell lysate in 60 ml total volume containing40 ml 13 HotStarTaq Master Mix (QIAGEN), 5 mg/ml yeast tRNA, 3 mMfinal concentration of MgCl2, and 1 mM each first-round primer. In thesecond round, 2 ml first-round product was added to 10 ml HotStarTaqMaster Mix using the conditions above, but the reactions were in separatetubes, each containing a pair of second-round primers (5 mM each). Bothrounds of PCR used the following conditions: 95˚C for 15 min; 35 cyclesof 94˚C for 1 min, 60˚C for 1 min, and 72˚C for 2 min; and a final extensionat 72˚C for 10 min. A total of 5 ml of the second-round PCR products wereexamined on 1.2% agarose gels. Single-cell PCR primers were as described(39).

Protein expression and binding assay

The cDNAs for full-length murine PAX5 and amino acids 431–652 of murineE12 were obtained from Genscript and subcloned into pET21b+(2), a deriv-ative of pET21b+ (Novagen), in which the N-terminal T7-tag is replaced withthe pelB leader from pET20b+ (Novagen). The His6-tagged proteins wereexpressed in Escherichia coli BL21(DE3) pLysS. E12-expressing cells werecentrifuged and lysed by sonication in 50 mM Tris-HCl (pH 8), 1 M KCl,1 mM EDTA, 1 mM DTT, 1 mM PMSF. E12 was purified from the solublefraction by Ni-NTA agarose chromatography (QIAGEN) under native con-ditions, according to the manufacturer’s instructions, followed by purificationon Heparin-Sepharose. E12 was eluted with 20 mM HEPES (pH 8), 0.01%Triton X-100, 0.5 M NaCl, 10% glycerol, and 2 mM DTT. PAX5-expressingcells were centrifuged, and the pellet was solubilized and lysed by sonicationin 100 mMNaH2PO4, 10 mMTris (pH 8), and 8 M urea and centrifuged. ThePAX5-containing supernatant was purified by Ni-NTA agarose chroma-tography, according to the manufacturer’s instructions. Purified PAX5 wasrenatured by dialysis against 10 mM Tris (pH 8), 1 mM EDTA, and 5 mMDTT, with gradually decreasing concentrations of urea; 0.4 M arginine wasadded in the last dialysis without urea.

Purified PAX5 dialyzed against 1M potassium acetate (pH 8.3) was coupledto CNBr-activated Sepharose 4B (GE Healthcare), following the manu-facturer’s instructions, and using 1 M potassium acetate (pH 8.3) as couplingbuffer. Whole-cell extract from 1.5 million mouse B cells or purified WT orCaMR E12 or ETS1 (0.2 mg) was incubated with 10 ml PAX5-Sepharose orcontrol Sepharose beads in a binding buffer containing 50 mM Tris-HCl(pH 8), 100 mM NaCl, 1.5% Triton X-100, 1 mM DTT, and protease in-hibitor mixture tablet without EDTA (Roche) and either 0.1 mM CaCl2 or0.1 mM EGTA with rotation for 30 min at room temperature. The PAX5-Sepharose and control Sepharose were washed twice with binding buffer, andbound proteins were eluted by boiling in Laemmli loading buffer. Sampleswere separated by 12% SDS-PAGE and analyzed byWestern blot using Tetra-His Ab (QIAGEN) for E. coli–produced E12; ETS1/ETS2 Ab (C-275) forE. coli–produced ETS1; and b-actin Ab (ACTBD11B7), CD19 Ab (E-20), amixture of E2A Abs (V-18, H-208, and Yae) (all from Santa Cruz Biotech-nology), and a-tubulin Ab (B-5-1-2; Sigma) for proteins from whole-cell ex-tract. Pixel intensities of E12 bands were quantified by ImageJ software andplotted as percentage of E12 bound to PAX5.

Statistical analysis

The data are expressed as means 6 SD. The results were analyzed usingthe Student t test with two-tailed distribution, with the exception that datain Fig. 1E were analyzed using the Pearson x2 test.

ResultsDefective allelic exclusion in B cells expressing CaMR E2A

To analyze whether inhibition of E2A by CaM was relevant forallelic exclusion, we analyzed double expression of the IgH allelesin mice. C57BL/6 (referred to as B6) male mice heterozygous forE2A knockout (E2A+/2) were crossed with BALB/c females, andtotal bone marrow was isolated from resulting E2A+/2 F1 mice andinfected with retrovirus expressing either WT or CaMR mutant ofthe E12 splice form of E2A together with GFP. Infected bone marrowcells were transplanted into lethally irradiated B6 x BALB/c F1E2A+/2 recipients. Recipient mouse spleen B cells were analyzedby FACS for BALB/c (IgMA) and B6 (IgMB) IgH allotype expres-sion. The frequency of live-gated mouse splenic CD19+ B cellsappearing to express both Igh alleles was 1–1.5% for the GFP2

(i.e., noninfected) control cells from infection with virus expressingWT E2A (Fig. 1A, 1B, Supplemental Fig. 1A), as previously re-ported for FACS analysis of double expression of IgH (40). Im-portantly, in sharp contrast, both Igh alleles were detected in 7.1760.53% of the cells infected with virus expressing CaMR mutantE2A in the first experiment and in 6.336 0.64% of the cells in thesecond experiment (Fig. 1A, 1B, Supplemental Fig. 1A). This dra-matic increase in double-expressing cells also was in sharp con-trast to the cells infected with virus expressing WT E2A: 1.81 60.35% and 2.506 0.37%, respectively, appeared to double expressthe Igh alleles (Fig. 1A, 1B, Supplemental Fig. 1A). Compared withGFP2 cells from the infection with virus expressing WT E2A, ap-parently double-expressing cells were slightly more frequent both

2462 ALLELIC EXCLUSION BY INHIBITION OF E2A

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

among cells infected with virus expressing WT E2A and amongGFP2 cells from infections with virus expressing CaMR E2A(Fig. 1A, 1B, Supplemental Fig. 1A). These slight increases mightbe due to a slightly less efficient inhibition of E2A in some cells

that slightly overexpress WT E2A and due to a slight expression ofCaMR E2A in some cells selected as GFP2 after the infections with virusexpressing CaMR E2A. Nevertheless, the p values for the differencein the double-expression of the two Igh alleles between the GFP+

cells of the infections with virus expressing WTand CaMR E2Awere0.0003 and 0.00002, respectively, in the two experiments (Fig. 1B),and the p values for the differences in double expression between cellsexpressing the CaMR E2A and the two GFP2 controls were even lower.The increased frequency of apparent double expression of the

Igh alleles in the mutant implies that CaMR E2A either dramati-cally increases allelic inclusion of IgH or, much less likely, throughan unknown mechanism radically increases the background of cellsthat appear double positive in the FACS analysis but are not truedouble expressors. To discriminate between the alternatives, sortedapparently double-positive cells were cultivated for 3 d and rean-alyzed. Most apparently double-positive cells in FACS analysis arebackground, because they do not show double expression in thesecond FACS analysis (40), and this also was the case with theB cells infected with virus expressing WT E2A (Fig. 1C, 1D). Im-portantly, most IgH double-positive cells with CaMR E2A in sharpcontrast showed double expression also in the second FACS anal-ysis of IgH expression (Fig. 1C, 1D). Furthermore, we also analyzedthe sorted double-positive cells, cultivated for 3 d, that remainedapparently double positive in the second FACS analysis, by intra-cellular immunostaining after the second sorting. Most apparentlydouble-positive cells after this second sorting of WT E2A–infectedcells were background because they were not double stained,whereas most resorted double-positive cells with CaMR E2A insteadshowed both IgMA and IgMB immunostaining (Supplemental Fig.1B). Thus, the mutation of E2A increased true double expressionand not the background. To further discriminate between the twoalternatives, VDJ recombination of the Igh locus of sorted GFP+

(i.e., infected) apparently double-positive cells, selected as shownin Fig. 1A and Supplemental Fig. 1D, was analyzed by single-cellPCR. Approximately two thirds of cells with both alleles detectedthat were infected with virus expressing WT E2A, and the controlcells, had only one VDJ recombined allele, whereas the other allelewas DJ recombined (Fig. 1E, Supplemental Fig. 1C). In contrast,75% of the cells expressing CaMR E2A had both alleles VDJ re-combined (Fig. 1E, Supplemental Fig. 1C). The p values for thisdifference between the mutant and the WTs were#0.00016. Thus,CaM resistance of E2A dramatically increases allelic inclusion of IgH.Next, we analyzed whether CaMR E2A increases allelic inclu-

sion only when overexpressed or whether it also inhibits the allelicexclusion at normal expression levels. E2A+/2 mice have 50%reduced expression of E2A (22). FACS analysis showed increasedE2A expression level with increased GFP fluorescence level (Sup-plemental Fig. 1F). Interestingly, significantly increased allelic inclu-sion was seen in the CaMR mutant over a broad range of expressionlevels, and the peak level of allelic inclusion in the mutant wasaround the normal WT level of E2A (Supplemental Fig. 1E, 1F). Thus,CaMR E2A strongly increases allelic inclusion of IgH at normalE2A expression levels. We also noted that strong overexpressionof either WT or CaMR E2A distorted the control of VDJ recom-bination, leading to a high level of cells not expressing IgMA orIgMB (Supplemental Fig. 1F), which further supports that E2A hasa critical role in the control of VDJ recombination. These stronglyGFP-fluorescent and heavily E2A-overexpressing cells were ex-cluded from the analyses in Fig. 1 and Supplemental Fig. 1B, 1C(as shown in Supplemental Fig. 1D). These cells were present2 wk posttransplantation (Supplemental Fig. 1D–F) but were al-most completely absent 1 wk later (Fig. 1B, left panel, Supplemen-tal Fig. 1A), indicating that cells heavily overexpressing E2A arerapidly selected away.

FIGURE 1. CaMR E2A leads to allelic inclusion of IgH. B6 male mice

heterozygous for E2A knockout (E2A+/2) were crossed with BALB/c

females, and total bone marrow was isolated from resulting E2A+/2 F1 mice

and infected with retrovirus coding for either WT or CaMR mutant E12

splice form of E2A, together with GFP, which resulted in 5–10% of infected

GFP-expressing cells (data not shown). Twenty-four hours postinfection, the

bone marrow cells were supplemented with fresh total bone marrow from

B63 BALB/c F1 E2A+/2 mice, and each cell mixture was transplanted into

a lethally irradiated B6 3 BALB/c F1 E2A+/2 mouse. (A) Recipient spleen

B cells were analyzed 2 wk posttransplantation for BALB/c and B6 allotypes of

IgH to study the effect of expression of WT and CaMR mutant E12 on

allelic inclusion of IgH. Typical FACS plots of the GFP2 controls (upper

panels) and the GFP-expressing retrovirus-infected cells (GFP+, lower panels)

of live-gated recipient mouse splenic CD19+ B cells. (B) Quantification of

experiments as in (A) from five mice (Experiment 2) and as in Supple-

mental Fig. 1A from three mice (Experiment 1). Results are mean 6 SD.

(C) Apparently IgMA+ IgMB+ double-positive cells shown in (A) were sorted

to culture and activated with LPS (25 mg/ml) for 3 d. Surviving GFP+ cells

were reanalyzed for IgMA and IgMB expression, revealing a much higher

percentage (indicated in the upper right quadrant) of double-positive cells

in the CaMR E2A–infected samples. (D) Quantification of experiments, as

in (C). Results are mean 6 SD (n = 3). (E) Single-cell PCR analysis of Igh

rearrangements in apparent IgMA+ IgMB+ double-positive cells in (A) and

Experiment 2 of (B) summarized from Supplemental Fig. 1C. GFP+ IgMA+

IgMB+ single B cells were sorted directly into PCR plates and analyzed for

VHDJH rearrangements by two-round semi-nested PCR. WT splenic B

cells of B6 3 BALB/c F1 mice also were analyzed as controls.

The Journal of Immunology 2463

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Inhibition of expression of components of the recombinationmachinery upon pre-BCR activation dependent on CaMsensitivity of E2A

A possible reason for the defect in allelic exclusion when E2A isCaMR is that sensitivity of E2A to CaM is needed for the signal

from the pre-BCR that arrests further recombination. To test thishypothesis, we characterized the effects of pre-BCR signaling on

the expression of proteins relevant for the recombination. We usedmouse bone marrow–derived pre-B cells grown on limiting numbers

of stromal cells. We showed previously that the pre-BCR is notsignaling constitutively at these conditions and is inducible upon

activation by cross-linking with Ab against the m H chain (25, 41).The pre-BCR stimulation efficiently induced increased intracel-

lular Ca2+ concentration within ,1 min (Supplemental Fig. 2), andseveral of the effects of pre-BCR stimulation on gene expressionare through Ca2+ signaling (25, 41). Quantitative RT-PCR showed

that the pre-BCR stimulation reduced the mRNA levels of RAG1,RAG2, and TdT, which adds nucleotides at the RAG-generated

DNA ends, as well as those of the PAX5 and ETS1 transcriptionfactors, between 2- and 4-fold within 2 h (Fig. 2A). The levels of

the proteins also were reduced in most cases, but with a delay.These reductions were $2-fold within 5–8 h (Fig. 2A, Supplemental

Fig. 3A). However, the level of E2A mRNA was reduced to amuch lesser extent, and the E2A protein level was not reduced by

the pre-BCR stimulation (Fig. 2A, Supplemental Fig. 3A). To studywhether Ca2+ signaling is important for the identified downregu-

lation, we inhibited all Ca2+ signaling with the Ca2+ chelator BAPTA-AM and blocked IP3 receptor Ca2+ channels with TMB-8 and/or

blocked L-type Ca2+ channels with nifedipine. For RAG1, RAG2,TdT, PAX5, and ETS1, the reduction in mRNA level after pre-BCR stimulation was partially or completely blocked by TMB-8

or nifedipine, and it was completely blocked by the Ca2+ chelatoror by combining the channel blockers (Fig. 2B). Thus, the regu-

lation of these proteins by the pre-BCR is dependent on Ca2+

signaling. The Ca2+-dependent downregulation of RAG1, RAG2,

and TdT is not species specific, because we obtained similar resultsusing the human pre-B cell line Nalm-6 (Supplemental Fig. 3B, 3C).

Inhibitors of other signaling pathways from AgRs (PD98059,MEK1/2 inhibitor; bisindolylmaleimide, PKC inhibitor) showed that

Ca2+ plays a dominant, but not exclusive, role in pre-BCR regulationof these genes. PAX5 and ETS1 mRNAs could not be analyzed in

this way, because their levels remained unchanged upon pre-BCRstimulation as a result of unknown causes, perhaps coupled to the

fact that Nalm-6 is a tumor cell line.We determined whether activation of the pre-BCR affected the

amount of protein that could bind to E2A binding sites in the RAG1,

RAG2, and TDT promoters, as well as the RAG enhancer Erag, byEMSA of nuclear extracts in the absence of Ca2+. No reduction in

the amount of any of the E2A–DNA complexes was seen during 5 hof anti-m treatment (Fig. 2C). This was expected, because pre-BCR

stimulation did not have much effect on the expression of E2AmRNA or protein in primary pre-B cells (Fig. 2A) or in Nalm-6

cells (25) (data not shown). The shifted bands were identified asE2A–DNA complexes using neutralizing Abs (Supplemental Fig.

3D). The binding of E2A to the RAG1, RAG2, TDT, and Eragprobes, but not binding of other transcription factors to a control

probe, was much reduced under conditions that enable Ca2+ loadingof CaM (Fig. 2C). Id proteins are inhibitors of DNA binding of E2Aand other E-proteins. However, induction of Id protein did not

contribute to the rapid inhibition of recombination-coupled genesand proteins upon pre-BCR stimulation, because the number of

E2A–DNA complex with either of the RAG/TdT sites was not re-duced upon pre-BCR stimulation in the absence of Ca2+ (Fig. 2C).

Furthermore, none of the Id mRNA or protein levels shows anincrease after pre-BCR stimulation either in primary pre-B cells orNalm-6 cells (25, 41) that can account for the relatively fast reduc-tions in the mRNAs and proteins in this study.To examine whether Ca2+ signaling from the pre-BCR inhibited

the expression of recombination components through CaM in-hibition of E2A, we first used an EBV-based shuttle vector thatstably directs synthesis of an shRNA interfering with human E2AmRNA. This shRNA plasmid reduced the E2A mRNA and pro-tein levels in Nalm-6 cells by ∼85%, as previously reported (25).The shRNA expression plasmid reduced the expression of RAG1,RAG2, and TdT mRNAs by 70–80% (Fig. 2D), which confirmedthat the expression of these genes is E2A dependent. These re-ductions were specific, because the shRNA expression plasmid didnot affect the expression of the pre-BCR–regulated genes FOS,OCT2, and IL6 (Fig. 2D). Importantly, stimulation of the pre-BCRdid not inhibit RAG1, RAG2, or TdT expression further in thepresence of the shRNA against E2A (Fig. 2D), indicating that theinhibitions by pre-BCR engagement were entirely dependent oninhibition of E2A. The reduction in human E2A mRNA could becomplemented by cotransfection with the expression plasmid forthe E12 isoform of mouse E2A with which the shRNA does notinterfere. This complementation fully restored the expression levelsof RAG1, RAG2, and TdT mRNA and their sensitivity to pre-BCRstimulation (Fig. 2D). Importantly, anti-m had no effect on any ofthese genes when complementation was performed instead withthe CaMR mutant of E12 (Fig. 2D). This loss of effect of pre-BCRengagement was specific, because the mutation had no effect onthe inhibition of the control genes (Fig. 2D). Thus, CaM resistanceof E2A specifically causes the expression of the genes for thecomponents of the recombination process to be unaffected by pre-BCR stimulation. To further examine the importance of CaM in-hibition of E2A, we infected primary pre-B cells with retrovirusesencoding WT or CaMR E12, together with GFP. It was not pos-sible to study the effects of overexpression of CaM, because noinfected mouse pre-B cells with substantial overexpression of thisprotein were obtained (data not shown), presumably because CaMoverexpression affects viability and/or differentiation of pre-B cells.Infection with both WT and mutant E12 resulted in, on average,only a slight increase in total E2A protein compared with vector-infected cells (detected by FACS with E2A Ab) in this system, aspreviously reported (25). Stimulation of the pre-BCR of the pri-mary pre-B cells infected with retrovirus expressing WT E12or with vector control for 7 h resulted in significant (p # 0.01)∼2-fold inhibition of RAG1, RAG2, TdT, and PAX5 protein expres-sion (Fig. 2E). Importantly, in cells infected with virus expressingCaMR E12, the protein levels became completely resistant to pre-BCR stimulation for the RAGs and partially resistant for TdT andPAX5 (Fig. 2E). Thus, CaMR E12 renders the expression of thegenes for components of the recombination machinery resistant topre-BCR stimulation, implying that the inhibition of their expres-sion is through CaM inhibition of E2A.

Inducible in vivo proximities of the recombination-coupledcomponents E2A, RAG1, RAG2, and PAX5 with CaM and lossof proximities with H3K4me3 upon pre-BCR stimulation

The downregulation of recombination-coupled components was rel-ativelymodest and tookmany hours, which suggested the presence ofanother effect of pre-BCR stimulation to achieve efficient arrest ofIgh recombination. To further analyze the regulation of the recom-bination, we performed PLAs, which enable visualization and quan-tification of protein interactions. Primary mouse pre-B cells were stainedwith Abs against different combinations of two target proteins andsubsequently with secondary Abs conjugated with oligonucleotides

2464 ALLELIC EXCLUSION BY INHIBITION OF E2A

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

(PLA probes) to generate a strong red fluorescence dot when the twoPLA probe–marked proteins are in close proximity (#40 nm)(Fig. 3A). PLA is a binary assay in which the number of dots is pro-portional to the number of proximities in the cell (32–35). We obtainedproximity for all pairs of CaM and one of the recombination-coupledproteins analyzed: E2A, RAG1, RAG2, and PAX5. Anti-m treat-ment for 30 min increased the number of proximities detected by$2-fold, and the increases remained for 5 h (Fig. 3, SupplementalFig. 4A, 4B). Instead, the pre-BCR stimulation reduced proximitiesthat were found between H3K4me3, which is a binding site for RAG2,and E2A, RAG1, RAG2, and PAX5 $3-fold (Fig. 3, SupplementalFig. 4A, 4B). The background was minimal in control cells lackingthe probe-marked protein, as well as when one or both of the pri-mary Abs were omitted (Supplemental Fig. 4). The pre-BCR effecton each proximity was analyzed with similar results using two ormore different Ab pairs (distinguished by superscript numbers on theAbs in Figs. 3, 4, Supplemental Fig. 4). The results show that, uponpre-BCR stimulation, all of the analyzed recombination-coupled pro-teins quickly leave the active chromatin that is targeted by RAG2.The findings indicate that the recombination-coupled proteins are ina complex, so we also studied the proximities between these proteins.We found proximities between each pair of PAX5, RAG1, RAG2,and E2A. Importantly, these proximities remained after stimulationof the pre-BCR receptor (Fig. 3B, Supplemental Fig. 4A–C). Thesmall reductions in some of the proximities after 5 h are not morethan expected, considering the reductions in the levels of most of theproteins (Fig. 2A). Taken together, the data suggest the existence ofa recombination complex consisting of at least E2A, RAG1, RAG2,and PAX5 that remains intact when CaM binds to the complex afterpre-BCR stimulation. This suggested complex is on H3K4me3-modified chromatin before pre-BCR stimulation, but it leaves thismodified chromatin upon the stimulation that activates CaM to jointhe complex. Some of the pairs of H3K4me3 and/or a RAG proteinyielded more than two specific PLA dots in some of the cells (Fig.3A, Supplemental Fig. 4A–C). This finding that the number of specificPLA dots can exceed the number of Igh alleles implies that some ofthe detected proximities occur at other RAG-binding H3K4me3 mod-ification sites than at the Igh locus. Therefore, we analyzed inter-actions of E2A, PAX5, and RAG with the Igh locus (see below).

In vivo regulation of the proximities of E2A, PAX5, and theRAG proteins through CaM inhibition of E2A

To further study the regulation of the suggested recombinationcomplex, we performed PLA of primary pre-B cells infected with

FIGURE 2. Downregulation of key components of the recombination

process after pre-BCR stimulation dependent on CaM sensitivity of E2A. (A)

Levels of RAG1, RAG2, TdT, PAX5, E2A, and ETS1 mRNA and protein

after stimulation of the pre-BCR of primary bone marrow–derived mouse

pre-B cells by anti-m were determined by quantitative real-time RT-PCR and

Western blotting, respectively. Levels before addition of anti-m were set as

100%. Results are mean 6 SD (n $ 3). With the exception of E2A, the

reductions were significant for all mRNA after 1 h of anti-m (p, 0.012) and

for all proteins after 5 h of anti-m (p , 0.047). (B) Calcium signaling–de-

pendent downregulation of the indicated genes after pre-BCR activation.

Anti-m was added to one half of BAPTA-AM–treated (15 mM), TMB-8–

treated (50 mM), nifedipine-treated (50 mM), or TMB-8 plus nifedipine–

treated or untreated primary pre-B cells for 3 h, followed by quantitative

real-time RT-PCR. Values represent normalized mRNA levels with anti-m as

the percentage of the levels without anti-m. Results are mean 6 SD (n = 3).

(C) No change in DNA binding activity of E2A in the absence of Ca2+ after

pre-BCR activation with anti-m (left panel) and inhibition of formation of

the E2A–DNA complex in the presence of Ca2+ (right panel). Amounts of

E2A–DNA complexes (indicated by arrow) using nuclear extracts (NE) from

the Nalm-6 human pre-B cell line and probes containing E2A-binding sequences

from the human RAG1 promoter, RAG2 promoter, the RAG enhancer Erag,

the TdT promoter, and a control probe from the human GM-CSF promoter

(25) (nucleotides 262 to 225), which binds many transcription factors

(including AML1/RUNX, ETS, NFAT, and JUN/FOS proteins), were ana-

lyzed by EMSA in the presence of the chelator EDTA (0.5 mM) or Ca2+ (0.5

mM CaCl2). (D) Effects of a shuttle vector that expresses shRNA targeting

human E2A mRNA and of complementation by expression of WT or mutant

mouse E12 on the expression of RAG1, RAG2, TdT, Fos, Oct-2, and IL-6

mRNA. Nalm-6 cells were transfected with pMEP4 vector expressing

shRNA against human E2A mRNA or the empty shuttle vector or with the

shRNA expression vector plus the corresponding pMEP4 vector express-

ing WT or CaMR mutant of mouse E12, all together with CMV-EBNA

expression plasmid. Where indicated (shaded bars), cells were treated with

anti-m for 5 h before the harvest. Levels of mRNA were determined by

quantitative real-time RT-PCR. Results are mean 6 SD (n = 3). (E) Loss of

anti-m sensitivity of RAG1, RAG2, TdT, and PAX5 expression by ex-

pression of CaMR E12. Changes are shown in the indicated protein levels

upon pre-BCR stimulation with anti-m for 7 h in mouse primary pre-B

cells grown with IL-7 on S17 stromal cells and infected with retrovirus

encoding WT E12 or CaMR mutant of E12, together with GFP. Proteins

were quantified by intracellular immunostaining and flow cytometry. Cells

with a large increase in GFP fluorescence were considered retrovirus

infected and were used to calculate average protein levels. The expression

levels are calculated as the percentage of the level of the corresponding

cells not treated with anti-m. Data are mean 6 SD (n = 4). *p , 0.05,

**p , 0.01.

The Journal of Immunology 2465

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

retrovirus expressing either WT or CaMR E12, together with GFP.As for noninfected cells, the proximity between CaM and eachof E2A, RAG1, RAG2, and PAX5 was induced upon pre-BCRstimulation for 30 min in the cells infected with virus express-ing WT E12 (Fig. 4A). Importantly, all of these inductions werecompletely lost in cells infected with virus expressing the CaMR

E12 mutant (Fig. 4A). Proximity between H3K4me3 and each ofE2A, RAG1, RAG2, and PAX5 was inhibited upon pre-BCRstimulation for 30 min in the cells infected with virus express-ing WT E12 (Fig. 4A), as shown for noninfected cells. Thesmaller reductions in the infected cells compared with noninfectedones might reflect that the higher level of E2A, and, thereby, lessCaM relative to E2A, could reduce the efficiency of the inhibi-tion. In contrast to cells infected to express WT E2A, no reductionin the proximity between H3K4me3 and E2A, RAG1, RAG2, orPAX5 was seen after pre-BCR stimulation in the cells infected

with virus expressing CaMR E12 (Fig. 4A). These data supportthat CaM binds to E2A in a complex of E2A, RAG1, RAG2, andPAX5 after pre-BCR stimulation, thus inhibiting the DNA bindingof E2A and, thereby, making the recombination complex leave theH3K4me3-modified chromatin.

FIGURE 3. Inducible in vivo proximities of the recombination-cou-

pled components E2A, RAG1, RAG2, and PAX5 with CaM and rapid

reduction of the proximities with H3K4me3 upon pre-BCR stimulation.

(A) The pre-BCR of mouse primary pre-B cells grown with IL-7 on S17

stromal cells was stimulated with anti-m for the indicated times. The

detection of the proximal location of each protein pair analyzed (visu-

alized as red dots) was done by in situ PLA (original magnification360).

The controls are shown in Supplemental Fig. 4. (B) Summary of PLA

results on regulation of proximities of the recombination-coupled com-

ponents E2A, RAG1, RAG2, and PAX5 with CaM, with H3K4me3, and

with each other from (A) and Supplemental Fig. 4A–C. Different Abs

against the same protein are distinguished by superscript numbers.

FIGURE 4. CaMR E12 blocks pre-BCR stimulation–induced changes of

in vivo proximity of E2A, PAX5, and RAG proteins with CaM and

H3K4me3, and E12 interacts directly with PAX5. (A) Mouse pre-B cells were

infected with retrovirus expressing WTor CaMR mutant of E12, as in Fig. 2E.

Infected cells were identified by their GFP fluorescence in microscopy and

used to quantify the numbers of proximities (dots)/cell. The Abs were used at

dilutions that yielded a low number of dots/cell to have a low risk for missing

dots in the quantifications. The number of dots/cell in pre-B cell samples

before stimulation of the pre-BCR were for all pairs—E2A-CaM (4.4 dots/cell

for WT and 9.5 dots/cell for CaMR), RAG1-CaM (4.0 dots/cell for WT E2A

and 11.0 dots/cell for CaMR E2A), RAG2-CaM (8.7 dots/cell for WT E2A and

20.2 dots/cell for CaMR E2A), PAX5-CaM (11.1 dots/cell for WT E2A and

24.9 dots/cell for CaMR E2A)—set as 100%. The results are mean 6 SD

(n = 3). (B) E12 binds directly to PAX5. His6-tagged full-length PAX5

was produced in E. coli, and the purified PAX5 was immobilized on

cyanogen bromide–activated Sepharose 4B. Binding to the PAX5-

Sepharose is shown for proteins from whole-cell extracts of mouse B cells

(upper panels) and for purified proteins (middle and lower panels). The

whole-cell extract from 1.5 million cells or 0.2 mg of purified His6-tagged

WT or CaMR E12 (aa 431–652) or ETS1 from expression in E. coli was

allowed to interact with 10 ml of PAX5-Sepharose for 30 min. The

incubations of whole-cell extract were in the presence of 0.1 mM EGTA,

and those of purified proteins were in the presence of either 0.1 mM

EGTA or 0.1 mM CaCl2. After washing, the bound proteins were detected

by Western blot against the protein or, in the case of E. coli–produced E12

proteins, against the His6-tags. Preloads were 10% of the amount of protein

extract, E12, or ETS1 added to the binding reactions. Negative controls

were with empty Sepharose (Seph.) and Protein-A Sepharose (Prot.A-Seph.)

and without addition of protein or with addition of ETS1 instead of E12 to

PAX5-Sepharose (PAX5-Seph.). Quantification of the E12 bands from three

to nine experiments were plotted as the percentage of E12 bound to PAX5

(bottom panel). Results are mean 6 SD. The multiple bands in the panel

with E2A from cell extract are due to alternative splicing and multiple

phosphorylation sites in E2A, and the lower band in the panels with E. coli–

produced E12 is due to partial protease cleavage N-terminal to the bHLH

domain during E. coli production. *p , 0.05, **p , 0.01, ***p , 0.001.

2466 ALLELIC EXCLUSION BY INHIBITION OF E2A

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The interactions in vivo between the RAGs, PAX5, E2A, andCaM could all be direct; alternatively, one or more interactionscould be indirect through another protein in the suggested complex.Zhang et al. (14) reported a direct interaction between the RAG1/RAG2 complex and PAX5, and we (19–21) reported and charac-terized the direct interaction between E2A and CaM. However, nointeraction between PAX5 and E2A has been reported. To analyzewhether PAX5 and E2A can interact directly with each other, werecombinantly expressed full-length PAX5, the E12 splice form ofE2A (C-terminal 222 aa), and ETS1 in E. coli, purified the pro-teins, and determined the ability of E12 and the control protein tobind to PAX5. We also analyzed the binding of E2A and otherproteins from a mouse B cell extract to the PAX5 Sepharose. E2Afrom the whole-cell extract showed very efficient binding to thePAX5 Sepharose, which was in sharp contrast to the poor bindingof b-actin, a-tubulin, and CD19 control proteins (Fig. 4B). TheE. coli–produced E12 bound to PAX5 Sepharose beads but not tocontrol Sepharose beads with Protein A or without coupled protein,and E12 bound PAX5 Sepharose much more than did ETS1 controlprotein (Fig. 4B). The binding of E12 to PAX5 was equally effi-cient in the presence of Ca2+ and when Ca2+ was absent, because ofthe presence of the Ca2+ chelator EGTA, and the CaMR mutation ofE12 did not affect the efficiency of the binding to PAX5 (Fig. 4B).In summary, the RAG1/RAG2 complex can interact directly withPAX5, PAX5 can interact directly with E2A, and E2A can interactdirectly with CaM.

Binding of E12, RAG, and PAX5 at the Igh locus is sensitive toCaM inhibition of E2A

We used retrovirus-infected primary pre-B cells and performedChIP to analyze the effects of CaM resistance of E12 on the lo-calization of E12, RAG, and PAX5 in the Igh locus before and afterpre-BCR stimulation. We added a C-terminal sequence encodinga tandem Flag-epitope to the full-length E12 (E12-Flag) in theretrovirus. Lysates of infected pre-B cells were subjected to ChIPagainst the Flag-epitope. The recovered DNA was quantified byPCR using primer pairs amplifying regions of the Igh locus con-taining binding sites for E2A (E-box sequences), RAG1/RAG2(RSS), RAG2 (H3K4me3-rich sequences), and PAX5. E2A hasa strong preference for E-boxes with the sequence CAGCTG, butthere are also preferences for the surrounding nucleotides, andthere are only a few E2A sites/million bp in the genome (42). Asexpected, we observed ChIP signals for the E-box–containing Ighenhancers Em and hs3b (36) in cells expressing WT E12-Flag overthe background of cells infected with empty vector (Fig. 5A).Importantly, anti-Flag signals for E12-Flag also were above thevector background for most amplicons containing RAG1, RAG2,or PAX5 binding sites but not for control sequences havingacetylated (i.e., active) chromatin without a binding site for E2A,RAG, or PAX5 (Fig. 5A). Thus, E2A binds not only to E2A sites butalso to many RAG and PAX5 sites. Next, we added a C-terminaltandem Flag-epitope to RAG2 and PAX5 in retroviruses. E12remained without the Flag-tag, thereby allowing expression of bothE12 and either flagged RAG2 or flagged PAX5 from the same virus.We did not construct the corresponding retrovirus for RAG1 be-cause of the large size of its cDNA. As expected, cells infected withvirus expressing RAG2-Flag plus E12 gave anti-Flag signals overthe background of empty virus in the ChIP for most of the RAG1and RAG2 binding sites (Fig. 5B). Importantly, signals also werestrong for the E-box–containing hs3b, as well as three of five PAX5binding sites (Fig. 5B). Finally, cells infected with PAX5-Flag plusE12 gave anti-Flag signals much over the vector control for all fivePAX5 binding sites (Fig. 5C). Importantly, specific signals also wereobtained for the E-box–containing Em and hs3b, as well as all of the

RAG1 and RAG2 binding sites (Fig. 5C). Taken together, these datastrongly support the notion that E12, RAG2, and PAX5 interactin vivo in a complex on the Igh locus through the binding sites ofthese proteins, and we infer that RAG1 is also part of this proteincomplex. The fraction of anti-Flag immunoprecipitated DNA variedfor the analyzed probe sites for the different Flag-labeled proteins(Fig. 5). This probably occurs, at least in part, because the detectionof the interaction of the suggested complex with the chromatinizedIgh DNA varies because it can bind with any of its DNA-bindingdomains and the H3K4me3-binding domain of RAG2. The levels ofcross-linking in ChIP depend on the specific amino acid side chainthat is located at different points of close contact of the protein witha reactive group at the base at the DNA. Thus, the efficiency of ChIPcan change when either the nucleotide sequence at the site and/or theprotein(s) binding there is changed or has a changed conformation.Furthermore, in this case, the accessibility for the Ab, and, thereby,the efficiency of the ChIP, might be different when the Tag is ondifferent proteins in the suggested complex. In addition, all protein

FIGURE 5. ChIP of E2A, PAX5, and RAG proteins at the Igh locus.

Mouse pre-B cells were infected with empty retrovirus (vector) or viruses

expressing E12-Flag (A), E12 plus RAG2-Flag (B), or E12 plus PAX5-Flag

(C) on the same MSCV-IRES virus and subjected to ChIP using the Flag-

epitope of the tagged protein. The recovered DNA was analyzed by quan-

titative real-time PCR using primer pairs amplifying regions of the Igh locus

containing binding sites for E2A (E-box sequences), RAG1/RAG2 (RSS),

RAG2 (H3K4me3-rich sequences), and PAX5 (PAX5-consensus sequences)

or acetylated active chromatin. The E2A and PAX5 sites were confirmed by

ChIP (8, 14, 36). The amplified VHA.1, VH10, and VhJ558.47(2) segments

contain the RSS where V and DJ segments are known to join after cleavage

by bound RAG1/RAG2 and, thus, are known RAG binding sites, and RAG

binding to the DQ52 and JH1–JH4 sites was verified by ChIP (38). The

identical amount of DNA, in mass, was used in the analysis of all samples

from the same mouse, all samples were treated identically, and all quantifi-

cations were from the early log phase of the PCR. Three mice were analyzed,

and all PCR samples from these three mice were measured in triplicate. The

figure shows the amount of PCR product obtained for each site after immu-

noprecipitation with anti-Flag (expressed as& of the amount of PCR product

before immunoprecipitation). All data are mean 6 SD.

The Journal of Immunology 2467

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

contacts in a complex will not be cross-linked when precipitatinga large complex of proteins on DNA; therefore, differential stabilityof noncross-linked protein interactions in the complex, in vivo orduring the ChIP procedure, also could affect the strength of the ChIPsignals. Nevertheless, it is notable that Em did not precipitatesignificantly more than the vector control in the RAG-Flag ChIP(Fig. 5B). This could be a result of the technical reasons describedabove that the combination of Flag on RAG and the E-box site ofEm gives weak signals for the complex. However, it is also possiblethat there is less RAG protein at this site compared with sites whereRAG-Flag gives significant ChIP signals. Our data suggesting thatE2A, PAX5, and RAG are together in a complex at many sites on theIgh locus does not necessarily imply that all of these proteins arepresent in stoichiometric amounts at all of the sites. In that case, itwould make sense if less RAG is located at Em, or if the binding ofRAG is less stable during the ChIP procedure at Em, compared withDNA segments that have RAG binding sites (RSS and/or H3K4me3)and are RAG cleavage sites.To analyze the effects of pre-BCR stimulation and the role of

CaM inhibition of E2A, we infected the pre-B cells with retrovirusexpressing either WT or CaMR E12 and performed anti-Flag ChIPbefore and after pre-BCR stimulation. We analyzed those IghDNA sequences that resulted in signals clearly over background inFig. 5. In cells infected with the retrovirus expressing WT E12-Flag, 30 min of pre-BCR stimulation reduced the ChIP signal withanti-Flag between 2- and 4-fold for all E2A binding sites, RSSsites (RAG1/RAG2 binding sites), H3K4me3 sites (RAG2 binding

sites), and PAX5 binding sites (Fig. 6). These reductions werestatistically significant (p , 0.035 in all cases) and remainedapproximately equally large after 2 h. Thus, pre-BCR stimulationrapidly reduces the binding of E2A to the Igh locus both on E2Abinding sites and on RAG and PAX5 sites where the localizationof E2A must be indirect. Importantly, the reductions in anti-FlagChIP signals were abolished both on the E2A binding sites and onsites for RAG1, RAG2, and PAX5 by the expression of CaMR

E12-Flag. All of the differences between WT and mutant E12were statistically significant after 30 min of pre-BCR stimulation(Fig. 6). These results suggest that the recombination complexleaves the Igh locus upon pre-BCR stimulation through CaM inhi-bition of E2A. All of the reductions in ChIP signal after pre-BCRstimulation when expressing E12 plus RAG2-Flag were 2–10-foldand were highly statistically significant (after 30 min all p , 0.025;after 2 h all p # 0.014), with the exception of the hs3b site whereno significant reduction was obtained (Fig. 6). A possible reason forthis exception is that other proteins binding together with E2A atthis DNase-hypersensitive site might retain RAG2 there also wheninhibition by CaM reduces the DNA binding of E2A. Importantly,all reductions in ChIP signals were completely or almost completelyabolished by expression of CaMR E2A (Fig. 6). The differencebetween expression of WT and mutant E12 was significant (p #

0.046) on all sites, with the exception of hs3b, after 30 min of pre-BCR stimulation (Fig. 6). ChIP analyses postinfection with virusexpressing E12 plus PAX5-Flag resulted, similarly to E2A-Flagand RAG1-Flag, in $2-fold reductions for the WT E12 that were

FIGURE 6. The binding of E2A, PAX5, and RAG proteins to the Igh locus is sensitive to CaM inhibition of E2A. Mouse pre-B cells were infected with

retrovirus constructs, as in Fig. 5, using either WT or CaMR E12. The pre-BCR of infected cells was stimulated with anti-m for the indicated times and then

subjected to ChIP using the Flag-epitope of the tagged protein and analyzed by quantitative real-time PCR, as in Fig. 5. The figure shows the amount of

PCR product obtained for each site after immunoprecipitation with anti-Flag compared with the amount of PCR product before immunoprecipitation. These

values for anti-Flag–immunoprecipitated DNA are expressed relative to the immunoprecipitated DNA from the corresponding unstimulated cells expressing

either WT or CaMR E12 that were set as 100% (bars not shown). All data are mean 6 SD (n = 3). *p , 0.05, **p , 0.01, ***p , 0.001.

2468 ALLELIC EXCLUSION BY INHIBITION OF E2A

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

statistically significant (all p # 0.024 after 30 min; all p # 0.038after 2 h). Importantly, the reductions in ChIP signals were com-pletely abolished by coexpression of the CaMR E2A with PAX5-Flag (Fig. 6). The differences between WT and mutant E12 werestatistically significant (p values # 0.042 after 30 min of pre-BCRstimulation), with the exception of JH2 (p = 0.054) and VHJ558(p = 0.074). The loss of inhibition after pre-BCR stimulation by theexpression of CaMR E2A remained after 2 h (Fig. 6), including onthe JH2 and VHJ558 sites, which supports the results after 30 min.The reductions with CaM-sensitive E2A within 30 min, which ismuch before any significant changes in levels for proteins stronglyaffected by AgR stimulation (41, 43), argue against the alternativethat the reductions are indirect through changed expression of an-other unknown protein. Therefore, ChIP data strongly support thenotion that E12, RAG2, and PAX5 interact in vivo in a complex onthe Igh locus. The coimmunoprecipitation of the binding site of oneprotein with Ab against the Flag on the other protein impliesa physical interaction between the Flagged protein and the proteinthat binds to the site. Furthermore, mutation of E12 to CaM re-sistance affects the level of this coimmunoprecipitation, evenwhen RAG2 binding sites are analyzed with Flagged PAX5 andwhen PAX5 binding sites are analyzed with the Flag on RAG2,which strongly suggests that all three proteins are together ina complex on the Igh locus. Therefore, the results further supportthat the effect of CaM is through E2A that is together with PAX5and RAG2 at PAX5, RAG, and E2A sites of the Igh locus. Insummary, ChIP analyses show that WT E12 allows the recombi-nation complex to leave E2A, RAG, and PAX5 binding sites in theIgh locus upon pre-BCR stimulation, whereas CaMR E12 sequestersthe complex on the Igh DNA.

DiscussionFeedback inhibition from AgRs has a well-established major rolein allelic exclusion, but the molecular mechanism has remainedlargely enigmatic (15–17). In this study, we show that CaM resis-tance of E2A dramatically increases the normally very low fre-quency of double expression of the Igh alleles at the cell surface to5–7%. This is close to the theoretical maximum in the absence offeedback inhibition that was estimated at ∼6.4%, because mostVDJ recombinations do not lead to a functional IgH (15–17). Atthe genomic level, CaM resistance of E2A resulted in a remarkable∼7-fold increase in the ratio between B cells VDJ recombined onboth alleles and B cells that had one allele VDJ recombined and theother allele only DJ recombined. Thus, the feedback inhibition is al-most completely lost in B cells with E2A to which CaM cannot bind.RAG1, RAG2, PAX5, and E2A are all indispensable for VDJ

recombination at Igh (2–4, 10–14). In this study, signaling fromthe pre-BCR was shown to downregulate the expression of severalcomponents of the recombination machinery, including RAG1,RAG2, and PAX5, through inhibition of E2A by CaM. Data alsowere provided supporting the notion that E2A, PAX5, and RAGare in a complex bound together to E2A, PAX5, RAG1, and RAG2sites on the Igh gene before pre-BCR activation. Upon activationof the pre-BCR, this complex instead bound to CaM and left theseIgh sequences when E2A was CaM sensitive. What is the relativerole in the allelic exclusion for the two negative effects on VDJrecombination? One argument that the latter is more important isthat downregulation of expression of PAX5, RAG1, and RAG2takes several hours, whereas the inhibition of the binding of thecomplex to the Igh gene was complete after 30 min, and Ca2+/CaM inhibits DNA binding of E2A in vitro in ,5 min (44). Thus,CaM inhibition of the binding of the recombination complexcould inhibit recombination of the second Igh allele much fasterthan the reductions in protein levels could. In addition, the re-

ductions in PAX5, RAG1, and RAG2 were ∼2-fold, and the re-ductions in their pair-wise proximities with each other and withE2A also were relatively modest. Nevertheless, the identifieddownregulation of recombination-coupled components are likelyto contribute to the allelic exclusion. The findings support thatinhibition of binding of the complex to the key sites on Igh hasa dramatic inhibitory effect on further recombination. The pre-BCR–induced reductions in binding of proteins in the complex toindividual key sites in the Igh locus were 2–10-fold in ChIP an-alyses (Fig. 6). However, reductions in binding of the complex toimportant sites upon pre-BCR stimulation could be even larger forseveral reasons: 1) overexpression of E2A in a part of the infectedcells could make inhibition of E2A by CaM appear less efficient,as indicated by the smaller effect of pre-BCR stimulation in in-fected cells than in noninfected cells in PLA (Figs. 3, 4, Sup-plemental Fig. 4A–C); 2) reductions in the joining of the manybinding sites relevant for recombination through the complexcould be much larger than reductions in protein binding to indi-vidual sites; and 3) the total number of PAX5, E2A, RAG1, andRAG2 sites in the locus is very high, and reductions in bindingthat are as large, or even larger, than those detected in this studymight occur at some of the other sites.Li et al. (45) reported a cyclin-dependent kinase site in RAG2

and that accumulation of RAG2 is regulated during the cell cycle,which raises the question of whether increased degradation ofRAG2 could be relevant for the reduction in RAG2 after pre-BCRstimulation. However, the downregulation of proteins that we foundwas not limited to RAG2 and was preceded by reductions in themRNAs that occurred much more rapidly and were of approximatelythe same magnitude. Thus, the reductions are through reductionsin the mRNA and are not through induction of degradation of theproteins. The reductions in the mRNAs after pre-BCR stimulationalso occur too rapidly to be through a shift in the cell cycle fromG1 to S and G2/M, where degradation of RAG2 was reported tooccur. Li et al. (45) showed that coupling of V(D)J recombinationto the cell cycle is not essential for enforcement of allelic exclu-sion. However, our data showing that CaM sensitivity of E2A isessential for allelic exclusion does not exclude the participation ofother mechanisms. After the Ca2+ signal from the pre-BCR hasterminated the VDJ recombination of Igh, something also enablesinitiation of the recombination of Igk or Igl. It should be notedthat a Ca2+ signal is time limited, which is suitable for an inactivationthat should be temporary. Because E2A also is needed in recombi-nation of L chains (3, 5, 6), it is likely that the retargeting to a L chaingene happens after the Ca2+ signal from the pre-BCR has ended.When Igh is accessible to the recombination machinery, chromo-

somal contraction and DNA looping enable recombination, whereaschromosomal expansion inhibits the recombination (46, 47). DuringVDJ recombination of Igh, the chromatin of the locus is organizedin compartments containing clusters of loops separated by linkers,and the entire repertoire of V regions merges and juxtaposes to Delements (48, 49). Productive V-DJ recombination at the Igh locusleads to reversal of locus compaction, and this decontraction waspostulated to enforce allelic exclusion (47). The large-scale chromatinstructures and the changes in them upon productive V–DJ recombi-nation indicate the existence of protein complexes binding to multipleplaces in Igh and, thereby, enabling looping between the critical DNAelements during the VDJ recombination but not after feedbackinhibition from the pre-BCR. The results presented in this study areconsistent with PAX5, E2A, RAG1, and RAG2 being key participantsin such protein complexes. The complexes also might contain CTCF,cohesin, YYI, and/or Ikaros, because studies (46, 50–52) supportthat these proteins contribute to the shaping of the conformation ofthe Igh locus.

The Journal of Immunology 2469

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The RAG proteins are essential for V(D)J recombination of allIg and TCR loci. Also, E2A and E-boxes control many aspects ofV(D)J recombination of Ig L chain and the TCR loci (3). Therefore,related protein complexes, as at the Igh locus, containing at least E2A,RAG1 and RAG2 might play corresponding roles on other AgR loci.Elucidation of this will be an interesting subject for future research.

AcknowledgmentsWe thank Dr. Per-Arne Oldenborg (Umea University) for tail-vein injections.

PAX52/2 pro-B cells were a kind gift of Drs. A. Ebert and M. Busslinger

(Institute of Molecular Pathology, Vienna Biocenter, Vienna, Austria).

DisclosuresThe authors have no financial conflicts of interest.

References1. Bergman, Y., and H. Cedar. 2010. Epigenetic control of recombination in the

immune system. Semin. Immunol. 22: 323–329.2. Matthews, A. G., and M. A. Oettinger. 2009. RAG: a recombinase diversified.

Nat. Immunol. 10: 817–821.3. Osipovich, O., and E. M. Oltz. 2010. Regulation of antigen receptor gene as-

sembly by genetic-epigenetic crosstalk. Semin. Immunol. 22: 313–322.4. Subrahmanyam, R., and R. Sen. 2010. RAGs’ eye view of the immunoglobulin

heavy chain gene locus. Semin. Immunol. 22: 337–345.5. Inlay, M. A., H. Tian, T. Lin, and Y. Xu. 2004. Important roles for E protein

binding sites within the immunoglobulin kappa chain intronic enhancer in ac-tivating Vkappa Jkappa rearrangement. J. Exp. Med. 200: 1205–1211.

6. Beck, K., M. M. Peak, T. Ota, D. Nemazee, and C. Murre. 2009. Distinct rolesfor E12 and E47 in B cell specification and the sequential rearrangement ofimmunoglobulin light chain loci. J. Exp. Med. 206: 2271–2284.

7. Romanow, W. J., A. W. Langerak, P. Goebel, I. L. Wolvers-Tettero, J. J. vanDongen, A. J. Feeney, and C. Murre. 2000. E2A and EBF act in synergy with theV(D)J recombinase to generate a diverse immunoglobulin repertoire in non-lymphoid cells. Mol. Cell 5: 343–353.

8. Ebert, A., S. McManus, H. Tagoh, J. Medvedovic, G. Salvagiotto, M. Novatchkova,I. Tamir, A. Sommer, M. Jaritz, andM. Busslinger. 2011. The distal V(H) gene clusterof the Igh locus contains distinct regulatory elements with Pax5 transcriptionfactor-dependent activity in pro-B cells. Immunity 34: 175–187.

9. Johnston, C. M., A. L. Wood, D. J. Bolland, and A. E. Corcoran. 2006. Completesequence assembly and characterization of the C57BL/6 mouse Ig heavy chain Vregion. J. Immunol. 176: 4221–4234.

10. Fuxa, M., J. Skok, A. Souabni, G. Salvagiotto, E. Roldan, and M. Busslinger.2004. Pax5 induces V-to-DJ rearrangements and locus contraction of the im-munoglobulin heavy-chain gene. Genes Dev. 18: 411–422.

11. Hesslein, D. G., D. L. Pflugh, D. Chowdhury, A. L. Bothwell, R. Sen, andD. G. Schatz. 2003. Pax5 is required for recombination of transcribed, acety-lated, 59 IgH V gene segments. Genes Dev. 17: 37–42.

12. Johnson, K., D. L. Pflugh, D. Yu, D. G. Hesslein, K. I. Lin, A. L. Bothwell, A. Thomas-Tikhonenko, D. G. Schatz, and K. Calame. 2004. B cell-specific loss of histone 3 lysine9 methylation in the V(H) locus depends on Pax5. Nat. Immunol. 5: 853–861.

13. Johnson, K., K. L. Reddy, and H. Singh. 2009. Molecular pathways andmechanisms regulating the recombination of immunoglobulin genes during B-lymphocyte development. Adv. Exp. Med. Biol. 650: 133–147.

14. Zhang, Z., C. R. Espinoza, Z. Yu, R. Stephan, T. He, G. S. Williams,P. D. Burrows, J. Hagman, A. J. Feeney, and M. D. Cooper. 2006. Transcriptionfactor Pax5 (BSAP) transactivates the RAG-mediated V(H)-to-DJ(H) rear-rangement of immunoglobulin genes. Nat. Immunol. 7: 616–624.

15. Brady, B. L., N. C. Steinel, and C. H. Bassing. 2010. Antigen receptor allelicexclusion: an update and reappraisal. J. Immunol. 185: 3801–3808.

16. Cedar, H., and Y. Bergman. 2008. Choreography of Ig allelic exclusion. Curr.Opin. Immunol. 20: 308–317.

17. Vettermann, C., and M. S. Schlissel. 2010. Allelic exclusion of immunoglobulingenes: models and mechanisms. Immunol. Rev. 237: 22–42.

18. Grawunder, U., T. M. Leu, D. G. Schatz, A. Werner, A. G. Rolink, F. Melchers,and T. H. Winkler. 1995. Down-regulation of RAG1 and RAG2 gene expressionin preB cells after functional immunoglobulin heavy chain rearrangement. Im-munity 3: 601–608.

19. Corneliussen, B., M. Holm, Y. Waltersson, J. Onions, B. Hallberg, A. Thornell,and T. Grundstrom. 1994. Calcium/calmodulin inhibition of basic-helix-loop-helix transcription factor domains. Nature 368: 760–764.

20. Onions, J., S. Hermann, and T. Grundstrom. 1997. Basic helix-loop-helix proteinsequences determining differential inhibition by calmodulin and S-100 proteins.J. Biol. Chem. 272: 23930–23937.

21. Saarikettu, J., N. Sveshnikova, and T. Grundstrom. 2004. Calcium/calmodulininhibition of transcriptional activity of E-proteins by prevention of their bindingto DNA. J. Biol. Chem. 279: 41004–41011.

22. Zhuang, Y., P. Soriano, and H. Weintraub. 1994. The helix-loop-helix gene E2Ais required for B cell formation. Cell 79: 875–884.

23. Hauser, J., N. Sveshnikova, A. Wallenius, S. Baradaran, J. Saarikettu, andT. Grundstrom. 2008. B-cell receptor activation inhibits AID expression throughcalmodulin inhibition of E-proteins. Proc. Natl. Acad. Sci. USA 105: 1267–1272.

24. Hughes, K., A. Antonsson, and T. Grundstrøm. 1998. Calmodulin dependence ofNFkappaB activation. FEBS Lett. 441: 132–136.

25. Hauser, J., A. Wallenius, N. Sveshnikova, J. Saarikettu, and T. Grundstrom.2010. Calmodulin inhibition of E2A stops expression of surrogate light chains ofthe pre-B-cell receptor and CD19. Mol. Immunol. 47: 1031–1038.

26. Krutzik, P. O., and G. P. Nolan. 2003. Intracellular phospho-protein stainingtechniques for flow cytometry: monitoring single cell signaling events. Cytom-etry A 55: 61–70.

27. Bhaumik, D., B. Yang, T. Trangas, J. S. Bartlett, M. S. Coleman, andD. H. Sorscher. 1994. Identification of a tripartite basal promoter which regulateshuman terminal deoxynucleotidyl transferase gene expression. J. Biol. Chem.269: 15861–15867.

28. Fong, I. C., A. A. Zarrin, G. E. Wu, and N. L. Berinstein. 2000. Functionalanalysis of the human RAG 2 promoter. Mol. Immunol. 37: 391–402.

29. Hsu, L. Y., J. Lauring, H. E. Liang, S. Greenbaum, D. Cado, Y. Zhuang, andM. S. Schlissel. 2003. A conserved transcriptional enhancer regulates RAG geneexpression in developing B cells. Immunity 19: 105–117.

30. Zarrin, A. A., I. Fong, L. Malkin, P. A. Marsden, and N. L. Berinstein. 1997.Cloning and characterization of the human recombination activating gene 1(RAG1) and RAG2 promoter regions. J. Immunol. 159: 4382–4394.

31. Hauser, J., J. Verma-Gaur, A. Wallenius, and T. Grundstrom. 2009. Initiation ofantigen receptor-dependent differentiation into plasma cells by calmodulin in-hibition of E2A. J. Immunol. 183: 1179–1187.

32. Jarvius, M., J. Paulsson, I. Weibrecht, K. J. Leuchowius, A. C. Andersson,C. Wahlby, M. Gullberg, J. Botling, T. Sjoblom, B. Markova, et al. 2007. In situdetection of phosphorylated platelet-derived growth factor receptor beta usinga generalized proximity ligation method. Mol. Cell. Proteomics 6: 1500–1509.

33. Soderberg, O., M. Gullberg, M. Jarvius, K. Ridderstrale, K. J. Leuchowius,J. Jarvius, K. Wester, P. Hydbring, F. Bahram, L. G. Larsson, and U. Landegren.2006. Direct observation of individual endogenous protein complexes in situ byproximity ligation. Nat. Methods 3: 995–1000.

34. Soderberg, O., K. J. Leuchowius, M. Gullberg, M. Jarvius, I. Weibrecht, L. G. Larsson,and U. Landegren. 2008. Characterizing proteins and their interactions in cells andtissues using the in situ proximity ligation assay. Methods 45: 227–232.

35. Soderberg, O., K. J. Leuchowius, M. Kamali-Moghaddam, M. Jarvius,S. Gustafsdottir, E. Schallmeiner, M. Gullberg, J. Jarvius, and U. Landegren.2007. Proximity ligation: a specific and versatile tool for the proteomic era.Genet. Eng. (N. Y.) 28: 85–93.

36. Greenbaum, S., and Y. Zhuang. 2002. Identification of E2A target genes inB lymphocyte development by using a gene tagging-based chromatin immuno-precipitation system. Proc. Natl. Acad. Sci. USA 99: 15030–15035.

37. Chowdhury, D., and R. Sen. 2001. Stepwise activation of the immunoglobulinmu heavy chain gene locus. EMBO J. 20: 6394–6403.

38. Matthews, A. G., A. J. Kuo, S. Ramon-Maiques, S. Han, K. S. Champagne,D. Ivanov, M. Gallardo, D. Carney, P. Cheung, D. N. Ciccone, et al. 2007. RAG2PHD finger couples histone H3 lysine 4 trimethylation with V(D)J recombina-tion. Nature 450: 1106–1110.

39. ten Boekel, E., F. Melchers, and A. Rolink. 1995. The status of Ig loci rear-rangements in single cells from different stages of B cell development. Int.Immunol. 7: 1013–1019.

40. Barreto, V., and A. Cumano. 2000. Frequency and characterization of phenotypicIg heavy chain allelically included IgM-expressing B cells in mice. J. Immunol.164: 893–899.

41. Hauser, J., J. Verma-Gaur, and T. Grundstrom. 2013. Broad feedback inhibitionof pre-B-cell receptor signaling components. Mol. Immunol. 54: 247–253.

42. Lin, Y. C., S. Jhunjhunwala, C. Benner, S. Heinz, E. Welinder, R. Mansson,M. Sigvardsson, J. Hagman, C. A. Espinoza, J. Dutkowski, et al. 2010. A globalnetwork of transcription factors, involving E2A, EBF1 and Foxo1, that orches-trates B cell fate. Nat. Immunol. 11: 635–643.

43. Verma-Gaur, J., J. Hauser, and T. Grundstrom. 2012. Negative feedback regu-lation of antigen receptors through calmodulin inhibition of E2A. J. Immunol.188: 6175–6183.

44. Onions, J., S. Hermann, and T. Grundstrom. 2000. A novel type of calmodulininteraction in the inhibition of basic helix-loop-helix transcription factors. Bio-chemistry 39: 4366–4374.

45. Li, Z., D. I. Dordai, J. Lee, and S. Desiderio. 1996. A conserved degradationsignal regulates RAG-2 accumulation during cell division and links V(D)J re-combination to the cell cycle. Immunity 5: 575–589.

46. Bolland, D. J., A. L. Wood, and A. E. Corcoran. 2009. Large-scale chromatinremodeling at the immunoglobulin heavy chain locus: a paradigm for multigeneregulation. Adv. Exp. Med. Biol. 650: 59–72.

47. Roldan, E., M. Fuxa, W. Chong, D. Martinez, M. Novatchkova, M. Busslinger, andJ. A. Skok. 2005. Locus ‘decontraction’ and centromeric recruitment contribute toallelic exclusion of the immunoglobulin heavy-chain gene. Nat. Immunol. 6: 31–41.

48. Jhunjhunwala, S., M. C. van Zelm, M. M. Peak, S. Cutchin, R. Riblet, J. J. vanDongen, F. G. Grosveld, T. A. Knoch, and C. Murre. 2008. The 3D structure ofthe immunoglobulin heavy-chain locus: implications for long-range genomicinteractions. Cell 133: 265–279.

49. Jhunjhunwala, S., M. C. van Zelm, M. M. Peak, and C. Murre. 2009. Chromatinarchitecture and the generation of antigen receptor diversity. Cell 138: 435–448.

50. Chaumeil, J., and J. A. Skok. 2012. The role of CTCF in regulating V(D)J re-combination. Curr. Opin. Immunol. 24: 153–159.

51. Ribeiro de Almeida, C., R. Stadhouders, S. Thongjuea, E. Soler, and R. W. Hendriks.2012. DNA-binding factor CTCF and long-range gene interactions in V(D)J recom-bination and oncogene activation. Blood 119: 6209–6218.

52. Seitan, V. C., M. S. Krangel, and M. Merkenschlager. 2012. Cohesin, CTCF andlymphocyte antigen receptor locus rearrangement. Trends Immunol. 33: 153–159.

2470 ALLELIC EXCLUSION BY INHIBITION OF E2A

by guest on August 30, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from