Effect of Environmental Antigens on the Ig Diversification andthe Selection of Productive V-J Joints in the Bursa1

Hiroshi Arakawa,2*† Kei-ichi Kuma,* Masahiro Yasuda,§ Shigeo Ekino,§ Akira Shimizu,† andHideo Yamagishi3*‡

In chickens, a single set of unique functional segments of both Ig H and L chain genes is rearranged during early embryogenesisto generate a pool of B cell progenitors that will be diversified in the bursa by gene conversion, forming the preimmune repertoire.After hatching, bursal cells are exposed to environmental Ags in the bursal lumen. We prepared B cells from each single bursalfollicle and used PCR-directed Ig L chain gene analysis to study the differentiation of B cells and the effect of antigenic stimulationfrom the bursal lumen on the neonatal chicken B cell repertoire formation. Selective amplification of B cell clones with a pro-ductive V-J joint was observed during the late embryonic stage, possibly by the interaction with ligands expressed on the bursalstroma and further accelerated in the neonatal chicken. Administration of the artificial Ags into the bursal lumen before theisolation of bursa by bursal duct ligation in the embryo caused a significant increase in lymphocytes with a productive V-J jointin the neonatal chicken bursa compared with the isolated bursa. Intra- and interclonal diversity of a complementarity-determiningregion measured by an evolutionary distance increased during bursal development. Clonal diversification did not require stim-ulation by artificial Ags from the bursal lumen. Thus, the preimmune repertoire in the bursa is generated by gene conversionduring Ag-independent B cell proliferation, and antigenic stimulation from the bursal epithelium to bursal B cells plays roles inthe selection of clones with a productive V-J joint. The Journal of Immunology, 2002, 169: 818–828.

T he bursa of Fabricius is a primary organ of differentiationfor the B cell lineage in birds (1). The bursa is composedof �104 lymphoid follicles that can be easily isolated (2).

Each follicle is colonized by a small number of prebursal stemcells that are committed to a particular Ig gene rearrangement inthe intraembryonic mesenchyme (3, 4). All chicken Ig L chains arerearranged by the same V and J gene segments and only one-thirdof the L chains derived from the bursal B cells up to day 13 ofembryonic development are in-frame (4, 5). No further V-J rear-rangements are ongoing in the embryonic bursa.

A major role of the bursa is to provide the necessary microen-vironment for the somatic diversification of rearranged V-J genesthrough a program of segmental gene conversion with a pool ofnoncoding pseudogenes being used as donors (6–8) and for theselective amplification of lymphocytes with productive gene rear-rangements (5). In posthatching bursal cells, nearly all V-J jointsare in-frame.

The bursa is a gut-associated lymphoid tissue and a major trap-ping site for environmental Ags from the gut (9–12). Exogenousand gut-derived Ags are actively transported across the bursal ep-ithelium into the lymphoid follicles of the bursa. Therefore, the

antigenic microenvironment of B cells in the bursa after hatchingdiffers from that of bursal cells developing in the embryo. Ligationof the bursal duct before hatching blocks the transport of gut-derived Ags into the bursa and results in reduced proliferation ofbursal cells after hatching (13). However, it remains unclearwhether the B cell repertoire is positively selected in the bursa insitu by environmental Ags trapped from the gut.

To study the effects of antigenic stimulation from the environ-ment, we closed the connection between the bursa and the gut bybursal duct ligation (BDL)4 on day 18 of incubation and injectedan artificial Ag, 4-hydroxy-3-nitrophenylacetyl (NP) coupled toBSA into the bursal lumen immediately before the BDL (NP-BDL). We used a PCR to amplify all Ig L chain genes in eachsingle bursal follicle from the 18-day chick embryo and the nor-mal, BDL, and NP-BDL 7-day-old chicken, and determined thenucleotide sequences. Using a computer-adaptable method for de-finitive assignments of gene conversion (14), we distinguishedbase modifications brought by templated gene conversion frompoint mutations. Clonally related genes carrying shared and uniquenucleotide changes can be explained by the intraclonal generationof Ab mutants during the expansion of individual B cell clones.

Although the proliferation of abortive cloness, possibly inducedby constitutive basal signaling, was observed in the bursa from theembryo and the posthatching normal and BDL chicken bursa, therewas a marked difference in the percentage of productively rear-ranged clones in the bursa between BDL and NP-BDL chicken.However, no significant differences were observed among the nor-mal, BDL, and NP-BDL chickens for the clonal diversity as mea-sured by the average evolutionary distances reflecting the aminoacid change from the germline and by the inter- and intraclonalamino acid difference in the complementarity-determining region(CDR) of Ig L chain. These findings provide direct evidence that

*Department of Biophysics, Graduate School of Science, †Center for Molecular Bi-ology and Genetics, Kyoto University, and ‡Health Research Foundation, Kyoto,Japan; and §Department of Anatomy, Kumamoto University Medical School, Kum-amoto, Japan

Received for publication June 18, 2001. Accepted for publication May 15, 2002.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by grants-in-aid for science research from the Ministry ofEducation, Science, Sports and Culture of Japan.2 Current address: Department of Cellular Immunology, Heinrich-Pette-Institute,D-20251, Hamburg, Germany.3 Address correspondence and reprint requests to Dr. Hideo Yamagishi, Health Re-search Foundation, Pasteur Building 5F, 103-5 Tanaka Monzen-cho, Sakyo-ku, Kyoto606-8225, Japan. E-mail address: yama@mail.taishitsu.or.jp

4 Abbreviations used in this paper: BDL, bursal duct ligation; NP, 4-hydroxy-3-ni-trophenylacetyl; CDR, complementarity-determining region; s, surface; NJ, neighborjoining.

The Journal of Immunology

Copyright © 2002 by The American Association of Immunologists, Inc. 0022-1767/02/$02.00

environmental Ags play a significant role in the selective ampli-fication of B cells with a productive V-J rearrangement but nocritical role in the process to create the diversity required in theadult immune repertoire. Thus, the preimmune repertoire in thebursa is generated by the Ag-independent B cell proliferation, butantigenic stimulations to the bursa accelerate expansion of produc-tive bursal cells driven by interactions with Ag.

Materials and MethodsIsolation of single bursal follicles

White Leghorn HB-15 chickens (15) were sacrificed on day 18 of incuba-tion and day 7 of posthatching, and each single bursal follicle was isolatedas described previously (2), with a few modifications. In brief, bursa ofFabricius was cut open and gently minced in HBSS in a 10-cm plastic dish.The fragments were teased with the backside of a curved tweezer. Indi-vidual follicles released into the medium were visible under a dissectingmicroscope. A single follicle was transferred into a microcentrifuge con-taining 30 �l of 1� LA Taq buffer (Takara, Kyoto, Japan) with 1 mg/mlproteinase K and 0.5% Tween 20. To avoid DNA degradation, the isolatedfollicles were immediately incubated for 45 min at 56°C for proteinaseK-mediated proteolysis, followed by a 10-min incubation at 95°C to inac-tivate proteinase K. This crude lysate of a single follicle was stored at�20°C before use.

Bursal duct ligation

BDL was performed on day 18 of incubation as described previously (16).To prepare the NP-BDL chicken, 0.3 mg of sterile NP-BSA resuspended in8 �l saline was administered into the bursal lumen via a vinyl tube im-mediately before BDL. Each single bursal follicle of BDL and NP-BDLchickens was isolated on day 7 of posthatching and stored as crude lysateprepared as described.

PCR amplification

A frozen lysate of a single bursal follicle was divided into three portions(10 �l each) and independently subjected to two rounds of PCR using pairsof nested primers. The primary PCR of 30 cycles was conducted with thenested L1/L4 primers in a 50-�l volume. The secondary PCR of 20 cycleswas conducted with the L20/L21 primers internal to those used for primaryPCR in separate reaction tubes using 2 �l of the first-round reaction mix-ture in a 50-�l volume. PCR were monitored by the PCR product definedas a visible band with an expected 350-bp length on an ethidium bromide-stained 0.8% agarose gel. Because the PCR product reached its maximumafter 30 cycles of amplification, PCR-introduced mutations are minimized.The primers were designed from the registered nucleotide sequenceM24403 (European Molecular Biology Laboratory/GenBank/DNA DataBase in Japan). Primary PCR primers, L1 5� of V�1 in the leader intronand L4 3� of J� in the J-C intron, were described previously (17). Thesecondary PCR primers, L20 (TCCAAGCTTTCCTCTCCCTCTCCAGG)and L21 (GGCTCTAGATCACGATGGGGGAAGAA) were flanked byHindIII and XbaI cloning sites, which facilitate the cloning of PCR prod-ucts. Neither of the restriction sites HindIII and XbaI were found in theregion including the �V�, V�1, and J� genes. PCR amplification wasperformed with LA Taq polymerase (Takara) as follows: an initial 4-minincubation at 94°C, 30 or 20 cycles consisting of 95°C for 30 s, 60°C for30 s, and 72°C for 90 s, with a final 5-min elongation step at 72°C.

DNA cloning and sequencing

Amplified DNA of expected length was purified by 0.8% agarose gel elec-trophoresis. Cloning was performed in pUC119 vector after digestion withHindIII and XbaI. Sequencing was done using the BigDye terminator cyclesequencing ready reaction kit (PE Applied Biosystems, Foster City,CA) and the autosequencer model 373 (PE Applied Biosystems). Allsequences were confirmed by sequencing both strands using the uni-versal and reverse oligonucleotide primers specific for M13. GenBankaccession numbers for the sequences reported in this paper areAB061547-061652 and AB061654-061667.

PCR-introduced mutations

PCR amplification artifacts were measured by sequencing 14 cloned prod-ucts of an analogous two rounds of amplification of a known L chainplasmid, p2115 (352 bp), equivalent to 5000 copies. We found only twosubstitutions in 4928 nucleotides (14 � 352 bp) during in vitro amplifi-cation. Therefore, these infrequent PCR artifacts are unlikely to account forthe majority of the observed mutations.

Gene conversion search and other statistical comparisons

According to the Conversion Search computer program, as previously de-scribed (14), linked base modifications with counterparts in the pseudogenepool were assigned as templated gene conversions. An Ig L chain pseu-dogene mini-database was constructed from M12437, M15097–15099,M15137–15156, and AH002536 (European Molecular Biology Labora-tory/GenBank/DNA Data Base in Japan). Other all single-base changeswere assigned as point mutations. The average number of base modi-fications, point mutations, and gene conversion events among relatedsequences in genealogical trees were calculated as shown previously(17). The evolutionary trees were inferred by the neighbor joining (NJ)method (18) and calculations of the evolutionary distance were as de-scribed (14).

ResultsSelection of productive Ig L chain in the bursa

We used a PCR to amplify all V�1-J� fusions of B cells isolatedfrom individual single follicles of the bursa from the day-18 em-bryos and from the 7-day-old normal, BDL, and NP-BDL chick-ens. Amplified DNAs were cloned into plasmid vectors, and the Vregion inserts of individual PCR clones were sequenced. We an-alyzed a total of 215 L chain rearrangements obtained from 16single bursal follicles (Table I). Despite the extensive sequencediversity by base modifications, each sequence can be related byunique V-J joint sequences. Because most follicles are popu-lated by a very few prebursal stem cells that are committed toa particular Ig gene rearrangement at the very beginning of thedevelopment of the embryonic bursa (2– 4), sequences relatedby a unique V-J joint have likely originated from the sameprecursor cell. Junctional diversification is achieved by some Pbase additions and moderate exonucleolytic nibbling of the cod-ing ends (5– 8).

When we examined the sequence of the V-J joints, about one-third of the L chains derived from the day-18 embryo and from the7-day-old normal chickens were out-of-frame. This proportion ofnonproductive V-J joints was significantly lower than the two-thirds observed at the earliest bursal stage (4, 5). Antigenic stim-ulation from the bursal lumen (NP-BDL) caused a more significantdecrease in nonproductive joints to one-fifth (Table I). This indi-cates that a selection of B cells with a productively rearranged Lchain locus within the bursa occurred during embryonic develop-ment and was further amplified by stimulation of artificial envi-ronmental Ags after hatching.

Selective amplification of B cells with productive gene rear-rangements suggests that the productive V-J coding sequencesmay be amplified during extensive bursal clonal diversifications.We examined the proportion of abortive clones with out-of-framejoints or with the loss of a productive rearrangement by modifyingthe reading frame in all clones analyzed. Neonatal normalchickens stimulated possibly by the natural environmental Agsafter hatching showed a significant decrease to one-fourth in theproportion of abortive clones (Table I). The neonatal NP-BDLchickens stimulated by the NP Ags showed a more pronounceddecrease to less than one-tenth in the proportion of abortiveclones (Table I).

Clonal diversification of B cells in the bursa during development

To reveal the clonal expansions present during late embryonic lifeand on day 7 after hatching, clonally related L chain sequencesobtained from each single bursal follicle were aligned with germ-line sequence and their representative related clones from e2 andp2 are shown in Fig. 1. Germline precursor segments are identifiedin e2 clones (Fig. 1A). For the quantitative assignments of geneconversion, a computer program, Conversion Search, was used for

819The Journal of Immunology

this study (14). Clonally related L chain sequences carried sharedand unique nucleotide changes. The mutations in this collection ofgenes included both templated gene conversions and point muta-tions. Single-base changes were either templated or untemplated inthe known pseudogene pool. The mutational patterns reflect anintraclonal generation of Ab mutants during the expansion of in-dividual B cell clones. All the Ig L sequences derived from thesingle bursal follicles, e2 and p2, are represented in the form ofgenealogical trees (Fig. 2). The unrelated single clones of uniqueV-J joints may well be considered representative of a minor pop-ulation of less diversification in the bursa.

As shown in Fig. 1A, gene conversion events with �V16(e2204) and �V23 (e2103 and e2109) resulted in the out-of-framesequences by shifting the reading frame of productive V-J joints.Repeated gene conversion events with �V23 (e2202) corrected theout-of-frame joints of e2103 or e2109. Repeated use of the geneconversion donors �V7 and �V10 was also observed in the CDR1to CDR2 (Fig. 1B).

Effect of Ags in the bursal lumen on the clonal diversification ofB cells after hatching

We aligned the clonally related L chain sequences from each sin-gle bursal follicle of BDL and NP-BDL chickens with the germlinesequence. Representative related clones from the bursal follicles ofb3 of BDL chickens and n1 of NP-BDL chickens are shown in Fig.3. Putative precursor segments shared by BDL b3-I clones (Fig.3A) show an out-of-frame V-J joining event, suggesting that non-productive joining is not a lethal event by itself. The pseudogenes�V2, �V6, �V8, and �V23 are preferentially and repeatedly usedas the gene conversion donor in the CDR1 and CDR3. The more

complicated gene conversions were observed in the n1-I clones ofNP-BDL chickens (Fig. 3B). The first common mutational event inthis group was the point mutation A to T in the CDR3 flankingregion. All the Ig L sequences derived from the single bursal fol-licles b3 and n1 are represented in the form of genealogical trees(Fig. 4).

Among several point mutations with no counterpart in theknown L chain pseudogene pool, we found a clear indication ofuntemplated mutations in the J� and J�-C� intron of the rear-ranged L chain genes (Figs. 1A and 3B) as shown in the bursa of3-wk-old chickens (7, 19).

Mutation mechanism of B cells specified in the bursa

In Table II, the number of mutation events was calculated for eachpair-group of IgL sequences as shown in the genealogical trees(Figs. 2 and 4). The number of conversion events increased withtime: two on day 18 of incubation and five on 7 days after hatch-ing. The gene conversion events were not different between BDLand NP-BDL chickens and thus were unaffected by the environ-mental Ags. A significant number of point mutations were ob-served during the late embryonic stage. However, the increase inpoint mutations was very small during development and was notaffected by the environmental Ags. Total base modifications dur-ing development were achieved mainly by a mechanism of geneconversion and were independent of the clonal selections by theenvironmental Ags.

The bar graph shown in Fig. 5 illustrates the frequency of pseu-dogene usage in the unique gene conversion events specified in thebursal follicles listed in Table I. Except for a striking increase inthe usage of �V23 segments at the late embryonic stage and in the

Table I. Number of sequences characterized by the unique V-J junction and coding sequence

Bursal FollicleaNo. of Sequences

AnalyzedAbortive

Clones (%)

Independent V-J Joints

Total Nonproductive (%)

e1 8 3 3 2e2 30 13 3 1e3 8 6 3 1e4 8 1 3 0Total (day-18 embryo) 54 23 (42.6)b 12 4 (33.3)

p1 7 1 5 1p2 30 6 10 3p3 8 2 5 2p4 8 4 8 4Total (day-7 neonatal) 53 13 (24.5)b 28 10 (35.7)

b1 8 2 4 1b2 8 2 6 2b3 30 13 11 6b4 8 3 4 2Total (BDL) 54 20 (37.0)c 25 11 (44.0)b

n1 30 1 6 1n2 8 0 5 0n3 8 2 5 2n4 8 1 4 1Total (NP-BDL) 54 4 (7.4)c 20 4 (20.0)b

a Bursal follicles were isolated from day-18 embryos (e1–e4) or 7 days after hatching from normal (p1–p4), BDL (b1–b4)and NP-BDL (n1–n4) chickens. Each set of follicles (e1 and e2; e3 and e4; p1 and p2; p3 and p4; b1 and b2; b3 and b4; n1 andn2; n3 and n4) was isolated from the bursa of a single chicken.

b Each percentage is significantly different ( p � 0.05).c Each percentage is significantly different ( p � 0.002). Data evaluation is based on the two-tailed Student t test.

820 DIVERSIFICATION AND SELECTION OF BURSAL B CELLS

BDL chicken (Fig. 5, A and C), the preferential usage of pseudo-genes located in the inverted orientation and more proximal to therearranged V-J locus was generally confirmed as described previ-ously for the random IgL clones (7, 20). The local follicle prefer-ence of the pseudogene donor as �V23 may be an accidental eventin a total of �104 follicles per bursa. Only the pseudogene �V22,carrying the shortest coding segment between CDR1 and CDR2,was not used for the gene conversion donors.

Evolutionary distance of Ig L chain protein from the germline

We translated the PCR sequences of the productive clone to thecorresponding amino acid sequences. Then we searched aminoacid sequence similarities between a pair of neighbor sequencesand constructed a unique evolutionary tree under the principle ofminimum evolution according to the NJ method (18) (Fig. 6). TheNJ method provided not only the topology but also the horizontalbranch lengths representing evolutionary distance of the final tree.

FIGURE 1. Represenative nucleotide sequences of L chain V segments cloned from a single bursal follicle, e2 of a day-18 embryo (A) and p2 of a7-day-old normal chicken (B). Clonally related sequences are compared with a precursor sequence. Only those codons that differ from the germline sequenceare shown; dashes represent bases identical with the germline sequence and dots represent gaps. Asterisks show termination codons or frame-shiftmutations. Deleted bases are shown in brackets. Codons are numbered according to the system of Reynaud et al. (7). We referred to the nucleotide sequenceof M15155 (European Molecular Biology Laboratory/GenBank/DNA Data Base in Japan), revised version of sequences by Reynaud et al. (7). There wasa single base conflict in �V23. Conversion sequences are underlined with homologous pseudogenes, the germline orientation of which is indicated byhorizontal arrows. Insertions introduced to pseudogenes to maximize homology are shown with vertical arrows. Point mutations, templated or untemplatedin the known pseudogene pool, are shown by filled or open superscript dots, respectively.

821The Journal of Immunology

FIGURE 2. Genealogical relationship of IgL sequences derived from a single bursal follicle, e2 of a day-18 embryo (A) and p2 of a 7-day-old normalchicken (B). E, The mutated sequence clone; U, hypothetical intermediates. Crippling mutations as termination or frame-shift mutations are shown bynotched circles. The two numbers alongside the branches refer to the additional numbers of gene conversion events/point mutations, respectively. Thenumbers in parentheses represent the additional base modifications. Clones e2101–e2113 and e2201–e2217 in A and clones p2101–p2115 and p2201–p2215 in B were derived from independent PCR, respectively.

822 DIVERSIFICATION AND SELECTION OF BURSAL B CELLS

The average evolutionary distance of IgL from the germline was3.3% during late embryonic life and expanded to 15% 1 week afterhatching. There were no significant differences in IgL in the evo-lutionary distance from germline between BDL and NP-BDLchickens.

Intra- and interclonal amino acid differences in CDR

Because the primary structure responsible for Ag binding is lo-cated in CDRs of each Ig chain, we examined the average evolu-tionary distances between clones of the same group and betweenclones of different groups in the region corresponding to the CDRgermline of the Ig L chain (26 aa of CDR1, CDR2, and CDR3) asshown in Table III. Clonal diversity in CDRs was expanded duringbursal development. No significant differences were observed be-tween intra- and interclonal diversities or in the clonal diversitiesbetween BDL and NP-BDL chickens. The evolutionary distancefrom the germline was expanded 2-fold more in the CDRs than inthe full length of IgL.

DiscussionAg-independent bursal diversifications

In chickens, the Ig gene rearrangement is not the key event for Igdiversity, but the postrearrangement diversification by gene con-version in bursa generates the B cell repertoire. Therefore, the

bursa of Fabricius is the primary site of B cell preimmune reper-toire formation in the chicken. The evolutionary distance of theIgL sequence from the germline increases with time in the bursa:3.3% at the intraembryonic phase and 15% at the posthatchingperiod (Fig. 6). These bursal diversifications are expanded to theaverage evolutionary distance of 22% in the periphery, as shownby Ag-activated B cells migrating into germinal centers (14). Bothbursal and germinal center diversifications contribute equally tothe evolutionary distance at the periphery.

In the pseudogene sequences used for gene conversion, frame-work regions are well conserved but the CDRs are more diversified(1). Accordingly, the evolutionary distance from the germline isenlarged to �30% in neonatal bursal cells when compared with theCDR sequences (Table III). However, repertoire of bursal lym-phocyte specificities shown by both intraclonal and interclonalevolutionary distances in CDRs of the Ig L chain was not signif-icantly affected by the NP antigenic administration. These resultsbased on the sequence diversity are compatible with the functionalevidence that the BDL treatment did not impede the rate of Abdiversification during embryonic development (21).

The postbursal Ig diversifications in the germinal center areequally induced by gene conversion and by point mutations in theearly phase of Ag stimulation, but gene conversion events arestrongly suppressed during the late stage (14). Although we iden-

FIGURE 3. Representative nucleotide sequences of L chain V segments cloned from a single bursal follicle, b3 of 7-day-old BDL chicken (A) and n1of 7-day-old NP-BDL chicken (B). Other indications are as described in Fig. 1.

823The Journal of Immunology

FIGURE 4. Genealogical relationship of IgL sequences derived from a single bursal follicle, b3 of BDL chicken (A) and n1 of NP-BDL chicken (B).Clones b3101–b3115 and b3201–b3215 in A and clones n1101–n1114 and n1201–n1216 in B are derived from independent PCR, respectively. Otherindications are as described in Fig. 2.

824 DIVERSIFICATION AND SELECTION OF BURSAL B CELLS

tified base modifications induced by gene conversion and pointmutations separately, somatic point mutations remained sup-pressed to a low level during bursal development (Table II). Point

mutations occurred at sites distant from gene conversion events(Figs. 1 and 3). The generation of the bursal preimmune repertoireis mostly brought out by gene conversion events.

FIGURE 5. Frequency of �VL usage as gene conversion donors. The number of occurrences of each �VL segment as possible conversion donors wastabulated in 31 gene conversion events specified in 28 IgL sequences from embryonic bursal follicles, e1–e4 (A); 118 conversion events in 30 IgL sequencesfrom neonatal normal bursal follicles, p1–p4 (B); 119 conversion events in 30 IgL sequences from BDL bursal follicles, b1–b4 (C); and 108 conversionevents in 30 IgL sequences from NP-BDL bursal follicles, n1–n4 (D). For events with more than two potential donors, each was counted as a fractiondivided by the number of possible donors. �VL segments are identified by number assignment, and orientation is distinguished by filled bars (antisenseorientation) and open bars (sense orientation).

Table II. Average mutational events for each pair-group of IgL sequences in genealogical trees

Bursal FollicleClonal Group

(no. of sequences)Gene Conversion

EventsPoint

MutationsBase

Modifications

e2 (day-18 embryo) I (26) 1.7 0.7 7.1

p2 (day-7 neonatal) I (7) 4.1 0.9 22.6II (7) 6.1 1.9 24.0III (6) 4.7 0.7 20.2IV (4) 7.5 2.3 32.5Average 5.4 1.3 24.0

b3 (BDL) I (5) 4.4 3.4 18.4II (5) 5.5 1.3 18.3III (5) 6.8 1.2 25.0IV (4) 6.8 1.8 27.8Average 5.8 1.9 22.1

n1 (NP-BDL) I (19) 5.1 2.8 22.0II (5) 4.0 1.4 21.6III (3) 4.0 1.3 19.7Average 4.7 2.4 21.7

825The Journal of Immunology

In both embryonic and posthatching stages, intraclonal CDRdiversity shown by average evolutionary distances between clonesof the same group is not markedly different from the interclonal

CDR differences between clones of different groups (Table III). Thisis characteristic of the Ag-independent bursal diversifications gener-ating the preimmune repertoire. In the Ag-dependent germinal center

FIGURE 6. The minimum evolution trees of 31 L chain amino acid sequences from embryonic bursal follicles, e1–e4 (A); 40 L chain sequences fromneonatal normal bursal follicles, p1–p4 (B); 34 L chain sequences from BDL bursal follicles, b1–b4 (C); and 50 L chain sequences from NP-BDL bursalfollicles, n1–n4 (D). The trees were inferred by the NJ method (18) after the alignment of 355 bp positions. The horizontal branch length represents thecorrected percentage of amino acid substitution among sequences including the germline sequence. The numbers above the tree refer to the averageevolutionary distance from the germline sequence in each stage.

826 DIVERSIFICATION AND SELECTION OF BURSAL B CELLS

reaction, in which Ag-activated B cells are diversified during the earlyphase and are subsequently subjected to selection for oligoclonality,intraclonal diversity was restricted to a narrow range in contrast withthe expanded interclonal diversity (14).

Nonproductive clones are not lethal

Joining events were identified by unique V-J junctions. Thesenumbers may be different from the actual number of independentjoints, because many of the joints are clonally amplified and arecounted as one independent event. Many nonproductive V-J jointsand abortive clones were observed in sequences derived from thesingle bursal follicles at different developmental stages (Table I).Nonproductive rearrangements cloned at day 18 of incubationwere 42% in total sequences and 33% in independent joints. Thispercentage is higher than the previous values, 6% of all V-J joints(5) and 4% of all V-D-J joints (22) cloned from pools of bursalcells. Minor nonproductive clones in each single follicle may notbe counted in a large pool of bursal follicle cells.

Clonal expansion of nonproductively rearranged IgL alleles and thebase modifications by gene conversion continued during bursal de-velopment without artificial Ag stimulation (Figs. 2B and 4A). Wealso characterized gene conversion events that result in either the lossof a productive rearrangement or correct out-of-frame joints by shift-ing the reading frame during late embryonic life (Fig. 1A). The non-productive V-J joining event is not a lethal event by itself in bursa.These findings would account for the Ag-independent clonal diversi-fication by gene conversion. Constitutive basal signaling may be suf-ficient to support B cell development in the bursa. In contrast, noabortive clones carrying crippling mutations in the Ig L chain havebeen diversified in the Ag-induced germinal centers (14).

Ag-dependent positive selection of B cells

The proportion (one-third) of nonproductive V-J joints at the lateembryonic stage is lower than the two-thirds expected if V-J join-ing is random at the earliest bursal stage (5). Thus, selective am-plification of cells with productive gene rearrangements could be

FIGURE 7. Schematic representation of bursal Bcell development and the IgM�IgG� immune complexafter hatching. After hatching, environmental Ags aretransported from the bursal lumen across the follicle-associated epithelium (FAE) into the medulla and formthe immune complex with yolk-derived maternal IgGentering possibly either from the lumen or from thecirculation. The Ag-IgG immune complex is trapped byfollicular dendritic cells (FDCs). Bursal B cells arestimulated by binding to the immune complex throughsIgM receptors and migrate to the cortex, where prolif-erating B cells and CD4� T cells are abundant.

Table III. Percentage of average amino acid difference in CDRs of IgL chaina

Bursal FolliclesNo. of CDRsComparedb Intraclonal Distance Interclonal Distance Distance from Germline

e1, e2, e4 25 8.8 � 7.6 11.6 � 8.3 5.9 � 5.8p1, p2, p3 30 35.6 � 9.0 47.0 � 10.2 38.3 � 7.8b1, b2, b3, b4 29 36.6 � 7.0 42.2 � 10.6 35.5 � 8.2n1, n2, n3, n4 40 38.3 � 11.4 48.3 � 12.3 31.2 � 10.6

a Percentage of amino acid difference is shown by a weight-average evolutionary distance, giving weights of a numericalfactor of CDRs.

b Clonal groups are shown with the number of CDR sequences (26 aa of CDR1, 2, and 3) in parentheses as follows: e1-II (4), e2-I(16), e4-I (5); p1-I (2), p1-II (2), p2-I (7), p2-II (7), p2-III (6), p3-I (2), p3-II (2), p3-III (2); b1-I (4), b2-I (2), b2-II (2), b3-II (5), b3-III(5), b3-IV (4), b3-V (2), b4-I (2), b4-III (3); n1-I (19), n1-II (5), n1-III (3), n2-I (3), n2-II (2), n3-I (3), n4-I (5).

827The Journal of Immunology

induced by an interaction between the surface receptor and ligandsin the bursal microenvironments.

After hatching, environmental Ags are transported along thebursal duct connecting the intestinal lumen to the bursal lumenacross the follicle-associated epithelium into the medullary areasof the bursal follicle (23). Thus, bursal cells with the prediversifiedreceptor are exposed to foreign Ags at a time and place in whichthe B cell repertoire is being generated. Although the proportion ofproductive joints was not changed during this time period, theselective expansion of B cell clones with in-frame V-J joints wasobserved (Table I). To clarify whether the selective amplificationof bursal lymphocyte with productive rearrangements could resultas a consequence of exposure to environmental Ags, we comparedthe IgL sequences of B cells from a single bursal follicle of BDLand NP-BDL chickens. Antigenic administration into the bursallumen immediately before BDL caused a significant decrease inlymphocytes with a nonproductive V-J joint at the IgL locus (Ta-ble I) and a remarkable decrease in abortive IgL clones in the bursa(Table I) in comparison with nonimmunized BDL chickens. How-ever, administration of bacterial Ag into the bursal lumen did notcause an increase in the proliferation rate of bursal cells in com-parison with the bursa of BDL chickens (13). There is no evidencefor Ag-specified B cell proliferation in bursa. Antigenic stimula-tion into the bursal lumen before BDL caused a significant increasein bursal cells expressing IgM�IgG� double-positive immunecomplexes on their surface (12). These immune complexes wereverified as the surface (s)IgM-positive B cells bound to the com-plex of environmental Ag and maternal IgG trapped on the surfaceof follicular dendritic cells as illustrated in Fig. 7 (Ref. 24 and M.Yasuda, H. Arakawa, H. Yokoyama, Y. Yokomizo, and S. Ekino,unpublished results). Starting around the time of hatching, B cellsin the epithelial buds segregate to form an outer cortex of cellssurrounding the inner medulla, and most bursal cell proliferationoccurs in the cortex (13, 24–26). The medulla contains largelynondividing cells. CD4� T cells were distributed not in the me-dulla but in the cortex, where some cytoplasmic IgG� cells werefound (24). The frequency of apoptotic cells in the normal bursaincreases after hatching and is especially enhanced in B cells ex-pressing a truncated Ig� chain (27). Thus, we suggest that B cellsstimulated by enormous environmental Ags are escaped from ap-optotic cell death and potentiated to migrate to form the cortex ofrapidly dividing cells. Bursal B cells emigrate directly from thecortex to the periphery via lymph vessels (26, 28). Introduction ofAg into the bursal lumen has been shown to induce the specificproduction of Ab-forming cells in the periphery following subse-quent systemic challenge (11, 29).

Overall, these findings strongly suggest competition within thefollicle of bursal cells expressing endogenous sIg receptors drivenby environmental Ag. Thus, environmental Ag-induced amplifica-tion of bursal B cells with in-frame V-J joints may be accountedfor by the Ag-induced B cell survival depending on the signalingthrough endogenous sIg receptors rather than the Ag-specified pro-liferation. Two distinct populations of peripheral B cells have beendefined as major populations of short-lived B cells comprising adiverse repertoire which have not encountered the foreign Ag inthe bursa, and minor populations of B cells exposed to environ-mental Ag within the bursa, acquiring a longer-lived state (30). Bcell populations clonally expanded in the bursal microenvironmentand positively selected by environmental Ags may acquire thelonger-lived state and respond to the environmental Ags in theperiphery after emigrating from the bursa.

AcknowledgmentsWe thank Dr. Jean-Marie Buerstedde for critical reading of the manuscriptand stimulating discussions.

References1. Weill, J. C., and C. A. Reynaud. 1987. The chicken B cell compartment. Science

238:1094.2. Pink, J. R., O. Vainio, and A.-M. Rijnbeek. 1985. Clones of B lymphocytes in

individual follicles of the bursa of Fabricius. Eur. J. Immunol. 15:83.3. Weill, J. C., C. A. Reynaud, O. Lassila, and J. R. L. Pink. 1986. Rearrangement

of chicken immunoglobulin genes is not an ongoing process in the embryonicbursa of Fabricius. Proc. Natl. Acad. Sci. USA 83:3336.

4. Mansikka, A., M. Sandberg, O. Lassila, and P. Toivanen. 1990. Rearrangementof immunoglobulin light chain genes in the chicken occurs prior to colonizationof the embryonic bursa of Fabricius. Proc. Natl. Acad. Sci. USA 87:9416.

5. McCormack, W. T., L. W. Tjoelker, C. F. Barth, L. M. Carlson, B. Petryniak,E. H. Humphries, and C. B. Thompson. 1989. Selection for B cells with produc-tive IgL gene rearrangements occurs in the bursa of Fabricius during chickenembryonic development. Genes Dev. 3:838.

6. Thompson, C. B., and P. E. Neiman. 1987. Somatic diversification of the chickenimmunoglobulin light chain gene is limited to the rearranged variable gene seg-ment. Cell 48:369.

7. Reynaud, C. A., V. Anquez, H. Grimal, and J. C. Weill. 1987. A hyperconversionmechanism generates the chicken light chain preimmune repertoire. Cell 48:379.

8. Reynaud, C. A., A. Dahan, V. Anquez, and J. C. Weill. 1989. Somatic hyper-conversion diversifies the single VH gene of the chicken with a high incidence inthe D region. Cell 59:171.

9. Schaffner, T., T. J. Muller, M. W. Hess, H. Cottier, B. Sodat, and C. Ropke. 1974.The bursa of Fabricius: a central organ providing for contact between the lym-phoid system and the intestinal content. Cell. Immunol. 13:304.

10. Sorvari, T., R. Sorvari, P. Routsalainen, A. Toivanen, and P. Toivanen. 1975.Uptake of environmental antigens by the bursa of Fabricius. Nature 253:217.

11. Ekino, S., K. Suginohara, T. Urano, H. Fujii, K. Matsuno, and M. Kotani. 1985.The bursa of Fabricius: a trapping site for environmental antigens. Immunology55:405.

12. Ekino, S., B. Riwar, F. G. M. Kroese, E. H. Schwander, G. Koch, andP. Nieuwenhuis. 1995. Role of environmental antigen in the development ofIgG� cells in the bursa of Fabricius. J. Immunol. 155:4551.

13. Ekino, S. 1993. Role of environmental antigens in B cell proliferation in the bursaof Fabricius at neonatal stage. Eur. J. Immunol. 23:772.

14. Arakawa, H., K. Kuma, M. Yasuda, S. Furusawa, S. Ekino, and H. Yamagishi.1998. Oligoclonal development of B cells bearing discrete Ig chains in chickensingle germinal centers. J. Immunol. 160:4232.

15. Kondo, T., H. Arakawa, H. Kitao, Y. Hirota, and H. Yamagishi. 1993. Signaljoint of immunoglobulin V�1-J� and novel joints of chimeric V pseudogenes onextrachromosomal circular DNA from chicken bursa. Eur. J. Immunol. 23:245.

16. Ekino, S., Y. Nawa, K. Tanaka, K. Matsuno, H. Fujii, and M. Kotani. 1980.Suppression of immune response by isolation of the bursa of Fabricius fromenvironmental stimuli. Aust. J. Exp. Biol. Med. Sci. 58:289.

17. Arakawa, H., S. Furusawa, S. Ekino, and H. Yamagishi. 1996. Immunoglobulingene hyperconversion ongoing in chicken splenic germinal centers. EMBO J.15:2540.

18. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method forreconstructing phylogenetic trees. Mol. Biol. Evol. 4:406.

19. Parvari, R., E. Ziv, F. Lantner, D. Heller, and I. Schechter. 1990. Somatic diver-sification of chicken immunoglobulin light chains by point mutations. Proc. Natl.Acad. Sci. USA 87:3072.

20. McCormack, W. T., and C. B. Thompson. 1990. Chicken IgL variable regiongene conversions display pseudogene donor preference and 5� to 3� polarity.Genes Dev. 4:548.

21. Lassila, O., I. Lefkovits, and A. Alanen. 1989. Immunoglobulin diversification inbursal duct-ligated chickens. Eur. J. Immunol. 19:1343.

22. Reynaud, C. A., V. Anquez, and J. C. Weill. 1991. The chicken D locus and itscontribution to the immunoglobulin heavy chain repertoire. Eur. J. Immunol.21:2661.

23. Toivanen, P., A. Naukkarinen, and O. Vainio. 1987. What is the function of bursa ofFabricius? In Avian Immunology: Basis and Practice, vol. 1. A. Toivanen and P.Toivanen, eds. CRC, Boca Raton, FL, p. 79.

24. Yasuda, M., S. Tanaka, H. Arakawa, Y. Taura, Y. Yokomizo, and S. Ekino. 2002.A comparative study of gut-associated lymphoid tissue in calf and chicken. Anat.Rec. 266:207.

25. Reynolds, J. D. 1987. Mitotic rate maturation in the Peyer’s patches of fetal sheepand in the bursa of Fabricius of the chick embryo. Eur. J. Immunol. 17:503.

26. Paramithiotis, E., and M. J. H. Ratcliffe. 1994. B cell emigration directly from thecortex of lymphoid follicles in the bursa of Fabricius. Eur. J. Immunol. 24:458.

27. Sayegh, C. E., and M. J. H. Ratcliffe. 2000. Perinatal deletion of B cells express-ing surface Ig molecules that lack V(D)J-encoded determinants in the bursa ofFabricius is not due to intrafollicular competition. J. Immunol. 164:5041.

28. Ekino, S., K. Matsuno, and M. Kotani. 1979. Distribution and role of lymphvessels of the bursa Fabricii. Lymphology 12:247.

29. Ekino, S., K. Matsuno, S. Harada, H. Fujii, Y. Nawa, and M. Kotani. 1979.Amplification of plaque-forming cells in the spleen after intracloacal antigenstimulation in neonatal chicken. Immunology 37:811.

30. Paramithiotis, E., and M. J. H. Ratcliffe. 1993. Bursa-dependent subpopulationsof peripheral B lymphocytes in chicken blood. Eur. J. Immunol. 23:96.

828 DIVERSIFICATION AND SELECTION OF BURSAL B CELLS