ABSTRACT
The molecular mechanism of immunoglobulin switch recombination is poorly understood. Switch recombination occurs between pairs of switch regions located upstream of the constant heavy chain genes. Previously we showed that switch recombination breakpoints cluster to a defined subregion in the S[gamma]3, S[gamma]1 and S[gamma]2b tandem repeats. We have developed a strategy for direct amplification of S[mu]/S[gamma]3 composite fragments as well as S[mu] and S[gamma]3 regions by PCR. This assay has been used to analyze the organization of S[mu], S[gamma]3 and a series of S[mu]/S[gamma]3 recombination breakpoints from hybridomas and normal mitogen-activated splenic B cells. DNA sequence analysis of the switch fragments showed direct joining of S[mu] and S[gamma]3 without deletions or duplications. Mutations were found in two switch junctions on both sides of the crossover point, suggesting that template switching is the most likely model for the mechanism of switch recombination. Statistical analysis of the positions of the recombination breakpoints in the S[gamma]3 tandem repeat indicates the presence of two sub-clusters, suggesting non-random usage of DNA substrate in the recombination reaction.
Immunoglobulin (Ig) variable regions are encoded by germline gene segments V, D and J, which are assembled during early B cell development by the process of V(D)J joining (1 ). The VDJ portion of the Ig protein is responsible for antigen binding while the constant (C) region is involved in biological effector function. The VDJ region is initially expressed with the C[mu] region to produce IgM. The murine Ig heavy (IgH) locus is composed of eight constant region (CH) genes which are arrayed as follows: 5'-VDJ-C[mu]-C[delta]-C[gamma]3-C[gamma]1-C[gamma]2b-C[gamma]2a-C[epsilon]-C[alpha]-3' (2 ). When B cells are stimulated by antigen and the appropriate lymphokines they produce new isotypes while maintaining the same antigen binding specificity. This phenomenon, the IgH chain class switch, permits expression of a variable region with new constant regions (3 -5 ). The recombination event focuses on the switch (S) DNA, regions of repetitive sequence upstream of each CH gene (with the exception of C[delta]) and produces a new hybrid DNA combination, S[mu]-Sx. The hybrid S[mu]-Sx DNA configuration is formed on the chromosome while the intervening genomic material is looped out and excised as a circle, confirming the importance of switch regions in the recombination process (7 -11 ).
Sequence analyses of the switch recombination joints of switched IgH genes from normal B cells and switch circles have revealed that for both the S[mu] (donor) and acceptor S region breakpoints fall within the tandem repeats (7 -10 ,12 ). No obvious consensus recombination signal sequences have been identified, though the pentamers GAGCT, GGGGT and GGTGG, components of all the switch regions, are often found at or close to recombination joints (6 ,12 ). Sequence comparison of S DNAs has shown that while all are highly repetitive, there has been significant sequence divergence between them (6 ). Several views on the mechanism of switch recombination have been articulated, which include switch-specific recombination, homologous recombination, illegitimate priming, template switching and illegitimate recombination (12 -18 ). It is currently unclear which of these models best describes the mechanism of switch recombination.
Multiple proteins bind specific sites in the DNA and are required for alignment of DNA recombination regions in highly ordered structures (19 ,20 ). We have identified a DNA binding protein complex, SNUP, which specifically recognizes a site in the S[mu] tandem repeat (21 ), and two DNA binding proteins, SNIP and SNAP, which specifically interact with motifs within the S[gamma] tandem repeats (22 ). SNIP is indistinguishable from NF-[kappa]B p50 homodimer, a member of the rel family of transcription factors, while SNAP contains epitopes in common with E47, a helix-loop-helix transcription factor (22 ,23 ). Recombination breakpoints cluster in the region spanning the SNIP and SNAP recognition motifs (22 ,24 ,25 ). Since this correlation is based on a statistical analysis, we sought to expand the sample size of available S[mu]/S[gamma]3 recombination junctions to more fully analyze the recombination breakpoint distribution in S[gamma]3. We report here a PCR method which permits direct amplification of S[mu]/S[gamma]3 composite molecules from genomic DNA without the necessity of using nested primers. Using this method we have derived nine new S[mu]/S[gamma]3 recombination junction sequences from hybridomas and normal lipopolysaccharide (LPS)-activated splenic B cells. Statistical analysis of the positions of S[gamma]3 recombination breakpoints in the tandem repeat confirms the correlation of clustered breakpoints with the region spanning the SNIP and SNAP motifs and identifies two statistically significant subclusters of switch junctions. Based on these findings a new model describing the initiating cleavage steps in the recombination reaction is proposed. The presence of mutations surrounding the switch junctions is one of the most striking features of switch recombination. The illegitimate priming and the template switching models have been proposed to accommodate the presence of mutations in switching (13 ,14 ). The illegitimate priming model predicts that mutations will be found on only one side of the switch junction while the template switching model predicts the presence of mutations on both sides of the crossover site. We find that mutations occur on both sides of the crossover site in several switch junctions newly described here. These findings support the the template switching model of switch recombination.
Single cell suspensions prepared from the spleens of 8-12-week-old female BALB/c nu/nu mice were stimulated in culture with LPS as previously described (26 ). The BALB/c nu/nu mice contain a T cell deficiency, thus T cell depletion to obtain B cells is unnecessary. Cells were maintained in RPMI-1640 (Gibco BRL Life Technologies Inc., Grand Island, NY) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Hyclone Laboratories, Logan, UT), 100 U/ml penicillin, 100 [mu]g/ml streptomycin and 4 mM glutamine in 5% CO2 at 37oC. The cell lines 8A5.4A5.II.88 (27 ), TIB 114 and HB 70 (28 ) were grown in RPMI-1640 supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin, 100 [mu]g/ml streptomycin and 4 mM glutamine. 8A5.4A5.II.88 was suppplemented with 50 [mu]M [beta] mercaptoethanol and 1 mM sodium pyruvate and TIB 114 with 50 [mu]M [beta] mercaptoethanol. The cells were maintained in culture at densities of 0.5-2.0 * 106 cells/ml. The TIB 114 and HB 70 hybridomas were obtained from the American Type Culture Collection.
Primers were designed using Amplify 1.0 software (University of Wisconsin) with the default settings. Once primers which fit these criteria were identified they were further tested for cross-hybridization with their own switch region and all other switch regions in the IgH locus. Only primers which did not or only weakly hybridized to other sites in S DNA were chosen for use. Primer [mu]-1 (5'-CTCTACTGCCTACATGGACTGTTC-3'), located 5' of S[mu] and which anneals to positions 5269-5293 (MUSIGCD07) on the germline [mu] sequence, and primer [mu]-2 (5'-CTGGCTCACTTAGCCTAATTGATTCTTGG-3'), located 3' of S[mu] and which anneals to positions 1058-1086 (MUSIGCD09) on the germline [mu] sequence, were used to amplify S[mu] regions. Primer [gamma]3-1 (5'-CAGGCTAAGATGGATGCTACAGGGA-3'), located 5' of S[gamma]3 and which anneals to positions 403-428 (MUSIGHANA) on the germline [gamma]3 sequence, and primer [gamma]3-2 (5'-TACCCTGACCCAGGAGCTGCATAAC-3'), located 3' of S[gamma]3 and which anneals to positions 2603-2627 on the germline [gamma]3 sequence, were used to amplify S[gamma]3 regions. The combination of primers [mu]-1 and [gamma]3-2 was used to amplify S[mu]/S[gamma]3 hybrid fragments. The PCR reactions for amplification of S[mu], S[gamma]3 and S[mu]/S[gamma]3 hybrid fragments from genomic Balb/c DNA contained 200 [mu]M each dNTP, 50 mM KCl, 1.5 mM MgCl2, 10 mM Tris-HCl, pH 8.3, 0.01% gelatin, 0.2 mg/ml BSA, 10 pmol each PCR primer, 0.3-0.5 [mu]g genomic template DNA, 2.5 U AmpliTaq DNA polymerase (Boehringer Mannheim, Indianapolis, IN) in a 50 [mu]l volume under 50 [mu]l mineral oil. The reaction was first heated to 95oC for 3 min and cycled in a Perkin-Elmer/Cetus DNA thermal cycler for 30 cycles each consisting of 1 min at 95oC, 45 s at 64oC and 2 min at 72oC. After the final cycle the reaction was incubated at 72oC for 5 min. Alternatively, Vent polymerase (New England Biolabs, Beverly, MA) was substituted for Taq where indicated and the buffer used was according to the manufacturer's instructions.
PCR amplification products were purified using the QIAquick Spin PCR Purification Kit (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. The purified fragments were then cloned into plasmid pCR3.1 (Invitrogen Corp., San Diego, CA). Alternatively, PCR product ends were polished using T4 DNA polymerase (Promega, Madison, WI) and cloned into the SmaI site of pBlueScript KS and then transformed into STBL2 (Gibco BRL, Gaithersberg, MD) or DH5[alpha] competent cells. The clones that contained inserts were verified by minipreparation, digestion with EcoRI and the inserts visualized by gel electrophoresis. Clones containing inserts were randomly chosen for further analysis by DNA sequencing. DNA sequencing reactions were performed using Sequenase sequencing kits (US Biochemical Corp., Cleveland, OH) according to the manufacturer's instructions and/or by automated DNA sequence analysis (Genetic Engineering Facility, University of Illinois, Urbana, IL). Cloned S[mu]/S[gamma]3 fragments from TIB 114 and HB70 were sequenced from independent isolates derived from two individual PCR reactions or PCR products were directly sequenced (US Biochemical Corp., Cleveland, OH) according to the manufacturer's instructions. The genomic S[gamma]3 region from Balb/c spleen DNA was PCR amplified using the S[gamma]3 locus-specific primers [gamma]3-1 and [gamma]3-2, described above. PCR was performed using Vent polymerase (New England Biolabs, Beverly, MA) for 30 cycles. Each cycle consisted of 1 min denaturation at 95oC, 45 s annealing at 65oC and 2 min elongation at 72oC. The amplified PCR fragment was gel purified and electroeluted from the agarose using Gene Capsule (Midwest Scientific, St Louis, MO) and blunt end cloned into the pBluescript KS cloning vector (Stratgene, La Jolla, CA); this is referred to as pS[gamma]3. The 5'- and 3'-ends were submitted to DNA sequence analysis and the resulting sequence was found to match the published sequence for S[gamma]3 (MUSIGHANA) (29 ).
High molecular weight genomic DNA was prepared from splenic B cells and from the cell lines using the QIAamp Tissue Kit (Qiagen, Germany). pS[gamma]3 described above and pM2-20 (a gift from K.Marcu; 30 ) were used as the sources of probes for hybridization with genomic S[gamma]3 and S[mu] respectively. The plasmids were digested with appropriate restriction enzymes and the resulting DNA fragments were separated by gel electrophoresis. The S[gamma]3 and S[mu] fragments were isolated from the gel and prepared as probes for Southern blot hybridization using a random primers kit (New England BioLabs, Beverly, MA) as recommended by the manufacturer.
The 8A5.4A5.II.88 cell line is a [mu] heavy (H) chain-producing pre-B cell line while TIB 114 and HB 70 are IgG3-producing hybridomas; all these cell lines are derived from BALB/c mice (27 ,28 ). Southern analysis was carried out to determine the organization of the S[mu] and S[gamma]3 regions in these cell lines. There may be four or more chromosomes containing the IgH locus in the hybridomas, making analysis of the switch DNA complex. Nonetheless, in DNA from TIB 114 digested with SacI or with XbaI (data not shown) the S[mu]-associated restriction fragments differ from the comparable fragment from BALB/c liver, indicating that these loci have undergone rearrangement (Fig. 1 A, lanes 1 and 3). In DNA from HB 70 digested with XbaI all of the S[mu]-associated restriction fragments are smaller than the 6.0 kb germline S[mu] fragment (31 ), indicating that in HB 70 the S[mu] loci have undergone rearrangement. Southern analysis of genomic DNA from 8A5.4A5.II.88 cells digested with XbaI or with SacI (data not shown) demonstrates that the XbaI fragment spanning the S[mu] region is 3.4 kb, as compared with the 6.0 kb fragment from BALB/c liver DNA (Fig. 1 B; 31 ). This finding implies that the S[mu] region in 8A5.4A5.II.88 contains an internal deletion.
Switch recombination is a deletional process in which the 5'-end of the S[mu] region and the 3'-end of the targeted S region join together to form a new hybrid S region (Fig. 2 ). When composite S[mu]/S[gamma]3 regions are formed by recombination, the distance between the PCR primers, located 5' of S[mu] and 3' of S[gamma]3, is greatly shortened, permitting amplification of the hybrid switch DNA. The use of PCR to directly amplify germline and rearranged switch regions provides an opportunity to confirm the organization of the S[mu], S[gamma]3 and S[mu]/S[gamma]3 fragments in specific cell lines and to characterize the switch/switch recombination junction sites by nucleotide sequence analysis.
Genomic DNA from 8A5.4A5.II.88, TIB 114, HB 70 and spleen cells was tested for the presence of S[mu]/S[gamma]3 hybrid switch regions by amplification using primers [mu]-1 and [gamma]3-2 (Figs 2 and 3 ). Amplification of DNA from HB 70 and TIB 114 cells reproducibly gave rise to one or two S[mu]/S[gamma]3 fragments respectively (Fig. 3 , lanes 6 and 9). Two S[mu]/S[gamma]3 co-migrating fragments were detected by Southern analysis of genomic DNA from HB 70 cells (Fig. 1 ), whereas a single S[mu]/S[gamma]3 fragment resulted from PCR amplification. This difference may be due to loss of one or more priming sites for this fragment. No S[mu]/S[gamma]3 fragments were observed following amplification of DNA derived from either 8A5.4A5.II.88 or spleen cells (Fig. 3 , lanes 3 and 12). These results suggest that hybrid S[mu]/S[gamma]3 regions can be specifically amplified from genomic DNA using locus-specific primers.
The three composite S[mu]/S[gamma]3 switch fragments which were reproducibly amplified from DNA of TIB 114 and HB 70 cells were purified and cloned. Independent clones from two separate PCR reactions were analyzed and/or PCR amplification products were directly sequenced. The clones TIB 114.1 (1.2 kb), TIB 114.2 (1.8 kb) and HB 70.1 (2.0 kb) share the 5'-S[mu] sequence starting from position 5268 of the germline S[mu] sequence (MUSIGCD07). TIB 114.1 was sequenced from both the 5'- and 3'-ends and confirmed to contain S[mu] and S[gamma]3 sequence. Restriction mapping of TIB 114.1 helped identify the region in which the crossover occurred and the S[mu]/S[gamma]3 crossover point was located at position 2133 in genomic S[gamma]3 DNA (MUSIGHANA) by automated DNA sequence analysis (Fig. 4 ). Three sequences, germline S[mu], germline S[gamma]3 and the recombined S[mu]/S[gamma]3 region, are shown in the area surrounding the breakpoint (Fig. 4 B). TIB 114.1 contained a total of 735 bp of S[mu] DNA and 480 bp of S[gamma]3 DNA. It is not possible to identify the actual genomic position of the S[mu] breakpoint in TIB 114.1 since the sequence available for genomic S[mu] in this region contains many internal deletions. However, a good match for the sequence immediately upstream of the crossover point, located at position 588, was found (MUSIGCD09). In clone TIB 114.2, 118 bp of DNA sequence were determined starting at the 5'-end; the S[mu] crossover point is located at position 5351 (MUSIGCD07) in genomic S[mu] while the S[gamma]3 breakpoint is located at position 1025 in genomic S[gamma]3 DNA (Fig. 4 ). There is a direct joining of S[mu] with S[gamma]3 DNA in these cloned fragments around the S[mu]/S[gamma]3 junctions (Fig. 4 ). At the S[mu]/S[gamma]3 junction of the TIB 114.1 and TIB 114.2 clones, no nucleotide addition is seen adjacent to the recombination site. The S[mu]-S[gamma]3 homology of the germline sequence at the breakpoint was one base for TIB 114.1 and two bases for TIB 114.2.
LPS is a specific B cell mitogen causing proliferation and differentiation of murine splenic B cells from surface IgM-positive to IgG-secreting cells (33 -35 ). B cells were activated with LPS for 92 h, DNA isolated and hybrid S[mu]/S[gamma]3 switch fragments amplifed. Switch recombination can occur anywhere within the S[mu] and S[gamma]3 regions and produces a population of S[mu]/S[gamma]3 switch fragments which are mixed with respect to size and structure. Amplification of DNA from LPS-activated B cells gives rise to a complex mixture of PCR products, whereas no fragments were amplified from DNA isolated from unstimulated splenic B cells (Fig. 3 ). The PCR products derived from DNA of LPS-activated B cells which fell in the size range 0.3-2.5 kb were purified and cloned. Twenty transformants containing inserts were randomly chosen for further analysis. These clones were shown by partial DNA sequence analysis to contain S[gamma]3 and S[mu] DNA sequence. Restriction maps were deduced for six clones, (B-1-B-6) containing S[mu]/S[gamma]3 fragments of >= 500 bp (Fig. 5 A). These six clones were taken for automated DNA sequence analysis. In all of these clones the switch junctions showed direct joining of S[mu] and S[gamma]3 DNA without deletion, duplication or insertion at the recombination breakpoint (Fig. 5 B). The S[mu]-S[gamma]3 homology of germline sequence at the breakpoint was one base for B-2, B-5 and B-6 and two bases for B-4. In clone B-6 a S[gamma]3 -> S[gamma]3 rearrangement was also found, which resulted from an internal deletion in S[gamma]3 downstream of the switch junction (Fig. 5 A). Mutations can be introduced as an artifact of PCR amplification. Since the B-1-B-6 clones were obtained from a heterogeneous population of S[mu]/S[gamma]3 recombinant molecules it is not possible to analyze them in independent PCR amplifications. Consequently, it is difficult to conclude that the apparent mutations reported in Figure 5 arose during the process of switching.
To study the mechanism of class switching we have sought to examine the distribution of switch recombination junctions in S[gamma]3 tandem repeats following S[mu] -> S[gamma]3 rearrangement. Southern blot analysis of DNA from the IgG3-producing hybridomas TIB 114 and HB 70 demonstrates that the S[mu] and S[gamma]3 regions co-migrate, suggesting that these regions are now structured as composite S[mu]/S[gamma]3 fragments. In contrast, the S[mu] and S[gamma]3 regions do not co-migrate in DNA from [mu] heavy chain-producing 8A5.4A5.II.88 cells. Using the direct PCR method, three S[mu]/S[gamma]3 hybrid fragments were amplified and analyzed from TIB 114 and HB 70. The size of these S[mu]/S[gamma]3 fragments was consistent with the organization of the rearranged genomic loci as observed by Southern blot analysis. DNA sequence analysis confirmed the origin of the S[mu]/S[gamma]3 fragments and located the switch breakpoints. Using locus-specific primers we found that the S[mu] and S[gamma]3 loci could be amplified in DNA from [mu] heavy chain-producing 8A5.4A5.II.88 cells and in splenic B cells but not from TIB 114 and HB 70, which had undergone isotype switching. Taken together, the absence of S[mu] and S[gamma]3 regions and the presence of new S[mu]/S[gamma]3 composite fragments provides direct evidence for switch recombination events. This PCR method was then applied to DNA derived from normal LPS-activated B cells. Twenty S[mu]/S[gamma]3 fragments were analyzed by partial sequence analysis and confirmed to contain S[mu] and S[gamma]3 DNA. Six of the S[mu]/S[gamma]3 fragments were further characterized and the recombination junctions defined. This PCR-based method provides a means to study low frequency switch events with small numbers of normal B cells. The selection of primers based on their high specificity for sites proximal or distal of the tandem repeats of the S[mu] and S[gamma]3 regions allows direct amplification of hybrid S[mu]/S[gamma]3 fragments without the necessity of using nested primers.
Our studies previously demonstrated that S[gamma] recombination breakpoints were non-randomly distributed in the S[gamma] tandem repeat (22 ,24 ,25 ). Given that the significance of breakpoint clustering is dependent on statistical analyses, we wished to further examine this phenomenom using a larger sample pool of S[gamma]3 breakpoints derived from B cells. Compilation of nine new S[mu]/S[gamma]3 junctions with previously characterized S[gamma]3 recombination breakpoints revealed that 84% of S[gamma]3 breakpoints are localized in the region spanning the SNIP/SNAP binding sites and focused on two subclusters in the SNIP and SNAP bindings sites respectively. Statistical analysis of breakpoint position in the S[gamma]3 tandem repeat demonstrates the non-random character of this distribution and confirms our earlier findings (22 ,24 ,25 ). It should be noted that ~16% of S[gamma]3 recombination breakpoints fall outside the SNIP/SNAP region. Thus preferential usage of the SNIP/SNAP binding region for recombination does not reflect an absolute concordance of binding sites with recombination junctions.
Several models for the mechanism of switching have been articulated (12 -18 ). One model suggests that switch recombination is due to homologous recombination and is catalyzed by the short pentameric repeats common amongst all S regions (6 ,16 ,17 ). The homologous recombination hypothesis is poorly supported by the data since there is very limited homology found at the points of recombination. The lack of an identifiable recombination signal sequence in switch regions has led to the illegitimate recombination model, which involves joining of non-homologous DNA ends at random break sites (18 ). Observation of a clustered distribution of recombination breakpoints in S[gamma] DNA suggests an ordered switch mechanism. We propose that switching may be initiated by site-specific double-strand breaks and that cleavage may be followed by extensive processing which produces both blunt and staggered ends. This model would explain the absence of recombination signal sequences at switch breakpoint sites and focusing of breakpoints on discrete regions in the S[gamma] tandem repeat. The recognition site for endonucleolytic cleavage may occasionally occur at other sites in the tandem repeat, thus providing an explanation for the exceptional breakpoints which comprise ~15% of the total breakpoint population. One activity which could account for this processing is B cell-associated nuclease (BCAN) (36 ). BCAN is a B cell-specific 3' -> 5' exonuclease which is non-processive and is expressed in the nuclei of normal mitogen-activated B cells and mature B cell lines but not in quiescent splenic B cells, pre-B cells, plasma cells, T cells or in other cell types (36 ). Further work is needed to test this model of switch recombination.
Two models which could account for mutations found immediately flanking switch junctions are illegitimate priming and template switching (13 ,14 ). Both models predict that single- or double-strand breaks occur in switch regions, that error prone DNA synthesis is intrinsic to the mechanism of switching and that mutations are generated around the switch junctions (13 ,14 ). The illegitimate priming model predicts that mutations will be found only on one side of the switch junction, while the template switching model proposes that mutations will be found on both sides of the crossover site. We found mutations on both sides of the switch junctions for both TIB.1 and TIB.2 and confirmed the authenticity of these mismatches relative to the germline sequence for TIB.2. This is the first example of a switch junction for which mutations have been confirmed on both sides of the crossover site and supports the template switching model for isotype switching. However, this is a limited data set and further examples of switch junctions containing mutations on both sides of the crossover site are needed to further confirm this conclusion.
This work was supported by Tobacco Research Council Award 3175 and NIH grant GM39231 to A.L.K. We thank R.Wuerffel for designing the primers used for PCR and B.Thompson for intially testing the primers. We thank K.Marcu for the M2-20 plasmid. We thank R.Wuerffel for critical reading of this manuscript.
*To whom correspondence should be addressed. Tel: +1 312 996 5293; Fax: +1 312 996 6415; Email: amy.l.kenter@uic.edu
REFERENCES