Switch Recombination in a Transfected Plasmid Occurs Preferentially in a B Cell Line That Undergoes Switch Recombination of Its Chromosomal Ig Heavy Chain Genes¹

When mature B lymphocytes, which express IgM and IgD on their surface, are activated by Ag and receive accessory signals, they switch to express downstream Ig heavy chain constant region (C_H) genes while maintaining the same expressed variable region. Because the C_H region determines the Ab effector function, class switching allows an adaptive humoral immune response to a variety of different infectious organisms. Ab class switching is caused by an intrachromosomal DNA recombination event called switch recombination, which occurs between switch (S) regions consisting of G-rich tandem repeats, located 5′ of each C_H gene, except Cδ (reviewed in Refs. 1, 2). The means by which the DNA is cut, aligned, and rejoined is unknown, as are the roles of several nuclear proteins that have been shown to bind to S regions (3, 4, 5, 6, 7, 8, 9, 10, 11, 12).

Base substitutions, duplications, and deletions found near switch recombination junctions led to the proposal that an illegitimate priming event, followed by error-prone DNA synthesis, creates the recombination (2, 13, 14). Recently, it has been shown that double-strand breaks are rapidly induced in the Sγ3 region when mouse splenic B cells are treated with mitogens that induce class switch recombination (CSR)³ to IgG3 (15). These data suggest that switch recombination is initiated by a double-strand break. Consistent with this is the finding that DNA-dependent protein kinase (DNA-PK), Ku70, and Ku80 are required for switch recombination (16, 17, 18). The Ku complex is known to bind to DNA double-strand breaks, nicks, gaps, and hairpins and is required for double-strand break repair (19) and for V(D)J recombination (20, 21, 22). However, because Ku proteins and DNA-PK are required for viability of mitogen-activated cells presumably due to induced double-strand breaks, it is possible that the effects of these mutations on CSR are indirect.

Numerous studies have established that transcription of unrearranged C_H genes occurs before switch recombination, producing what are termed germline, or switch, transcripts (23, 24, 25, 26). Transcription of germline RNA is regulated by the cytokines IL-4, IFN-γ, and TGF-β1, in concert with B cell activators, and serves to direct recombination to specific S regions (reviewed in Refs. 26, 27, 28). Transcription initiates at the I exon located 5′ to each S region. Deletions of the I exons or segments of the I exons and their upstream regulatory elements by gene targeting have demonstrated that germline transcripts are required for switch recombination (29, 30, 31, 32). Although cytokines induce germline transcripts, cytokines themselves do not induce switch recombination.

To provide an assay for the enzymes and proteins involved in cutting and recombining switch regions, several investigators have reported attempts to develop plasmid and retroviral substrates for assaying switch recombination. These substrates contain two switch region segments separated by a DNA segment whose deletion allows selection for recombination events. One approach has been to use transiently transfected plasmids that can replicate as episomes by virtue of the presence of a fragment from the Polyoma (Py) virus containing an origin of replication, early and late transcription initiation sites and encoding the T Ag. Using a drug selection assay, Leung and Maizels (33, 34) reported that such a plasmid undergoes recombination in LPS-activated spleen cells at a high frequency (21% of transfectants). Using a similar plasmid, Lieber and coworkers (35, 36, 37) found they could measure B cell-specific recombination events in the plasmid by subtracting the recombination that occurs immediately after transfection and only considering recombination events occurring in the window between 20 to 50 h after transfection. B cells of the sIg⁺ stage continue to accumulate recombination events during this window, whereas in most pre-B, plasmacytoma, and nonlymphoid lines no further recombination events are detected after 20 h. However, the high background levels of recombination occurring in non-B cells before the 20-h time point suggested that these plasmid assays lack some of the specificity expected for an assay of switch recombination enzymes. Furthermore, it was not determined whether these plasmids recombine preferentially in B cells that undergo CSR of their endogenous genes. Another approach was taken by Marcu and coworkers (38, 39, 40), who demonstrated that integrated retroviral vectors containing two switch region fragments undergo recombination in pre-B and B cell lines within the tandem repeats, and that the frequency of recombination is much lower in non-B cell lines. Recently, they found that treatment of splenic B cells with CD40 ligand induces recombination of the retroviral switch substrate (41). We combined these two approaches, using plasmids that lack the Py origin and assayed by the PCR for switch recombination after stable integration into the genome.

As a model system in which to study the mechanism of switch recombination, we use the mouse I.29μ B cell line, which can be induced to switch from expression of surface IgM (sIgM) to expression of sIgA by treatment with LPS plus TGF-β1 (42, 43, 44). This model resembles splenic B cells in that switching to IgA in mouse splenic B cells is also induced by a combination of LPS plus TGF-β1 (45, 46, 47).

Our plasmid, p273, undergoes recombination between its two switch region sequences (Sμ and Sα) in I.29μ cells, but at a reduced level in two T cell lines and infrequently in the B lymphoma M12.4.1 or in the plasmacytoma J558L, none of which undergo class switching of their endogenous genes. Recombination of the plasmid in I.29μ cells does not require treatment with inducers of switch recombination, suggesting that recombinase activity is constitutive in I.29μ cells.

The 1.8-kb HindIII mouse Sμ segment (BALB/c) from pM2-20 (48) was inserted into the HindIII site of the Bluescript SK⁻ vector (Stratagene, La Jolla CA). The 2.1-kb HaeIII Sα fragment cloned from the I.29μ lymphoma (3, 13) was inserted into the BamHI site of the vector after ligation of BamHI linkers. Both of these switch fragments are made up of tandem consensus repeats, although the Sμ fragment also includes about 330 bp 5′ to the SacI site where the tandem repeats begin. Both the Sμ and Sα fragments have a binding site for the B cell-specific paired domain transcription factor, BSAP (Pax-5), near their 5′ ends (3, 8, 49). The herpes virus thymidine kinase (TK) gene (1.8 kb) was excised from pZN(Sμ-Sγ2b)tk.1 (38) with BamHI, cloned into Bluescript KS⁻, excised with EcoRI and BamHI, and inserted between the Sμ and Sα segments at the EcoRI and BamHI sites. The G418 resistance gene (neo^r) was excised from pPGKneo^cbpA (received from Dr. Allan Bradley, Baylor College of Medicine, Houston TX) with XhoI, sticky ends filled in with T4 polymerase, KpnI linkers ligated on, and the fragment inserted into the KpnI site. The 1.0-kb cassette (EP) which has the 0.7-kb XhoI/EcoRI fragment containing most of the Ig μ intron enhancer and a 0.3-kb fragment containing the V_H 186.2 promoter was excised from pEP.B (50) (received from Dr. Fred Alt, Harvard Medical School, Boston MA) with SphI and XhoI, sticky ends filled in with Klenow and T4 polymerase, and cloned into the XhoI site (filled-in). A 0.5-kb BglII fragment containing the germline α promoter (Iα promoter) segment from −489 to +46 relative to the first RNA initiation site (51) was inserted in the forward direction into the BamHI site between the TK and Sα segments. The plasmids were linearized at the NotI or SacII site before transfection.

Cells (maximally 50 × 10⁶ per transfection) were electroporated as previously described (52) with 1 μg plasmid DNA per 10⁶ cells. Cells were cultured at 37°C in bulk culture for 2 days in the absence or presence of inducers, before drug selection. The components of the complete medium and culture conditions for I.29μ were described previously (43). All other cell types were cultured in the same medium in a 5% CO₂ incubator, using 10% FBS, without insulin. The concentrations and sources of LPS (50 μg/ml), TGF-β1 (2 ng/ml), and nicotinamide (10 mM) are as previously reported (44). EL-4 cells were either untreated or treated with 10 mM PMA (Sigma, St. Louis, MO) and ionomycin (1 ng/ml; Sigma)

To select for stably transfected cell lines, cells were pelleted and resuspended in medium (without inducers) containing G418 (Life Technologies, Gaithersburg, MD) at 400 μg/ml for all cells, except at 1.2 mg/ml for BW5147, and plated in 96-well plates at limiting dilution. For I.29μ, 2 × 10⁴ cells per well were plated. For the M12.4.1, J558L, and EL-4 lines, 4 × 10³ cells/well were plated, and for BW5147, 2 × 10² cells/well were plated. Cells were re-fed with medium supplemented with G418 subsequently about every 3 days. Transfected lines were identified and expanded. DNA was isolated as soon as the cultures contained 5–10 million cells. This took 2–3 wk. The transfected lines obtained should be clonal because they were generally obtained from experiments in which fewer than 10% of the wells yielded G418-resistant cells. Since for the purposes of this assay it is unimportant whether the transfected lines are clonal, they have not been further subcloned.

Class switching was assayed on day 2 by immunofluorescence microscopy as described (44).

The PCR to detect Sμ-Sα recombination was conducted for 30 cycles using the Taq Expand High Fidelity System from Boehringer Mannheim (Indianapolis, IN) in 2.0 mM MgCl₂ under conditions recommended by the manufacturer. The annealing temperature was 64°C. A total of 400 ng of genomic DNA, 0.2 μM primers, and 250 μM dNTPs were used. Primers used were: μ-1A, 5′-CTCTACTGCCTACACTGGACTGTTC-3′, identical with nt 5269–5293 of the germline BALB/c Sμ sequence (MUSIGCD07) (15); μ-3, 5′-TGGCTTAACCGAGATGAGCC-3′, identical with nt 5206–5226 of MUSIGCD07 (41); and R10, 5′-CTCTATCTAGGTCTGCCCCGTCTAGATAAG-3′, complementary to nt 62–91 upstream of the 3′ end of the 2.1-kb HaeIII Sα segment (GenBank accession no. AF069385) in the switch plasmids. The sequences of both μ primers differ from the sequence of the endogenous I.29 Sμ sequence (MUSIGHMY) at several nucleotides. The μ-3 and R10 primer combination was used for all PCR amplifications results presented, except for the clones used for nucleotide sequencing that were amplified earlier using the μ-1A and R10 primers (see Fig. 4). For amplifying the PCR product from clone 153 for sequencing, a second set of nested primers was used: μ-2, 5′-CCTGGGGTGAGCTCAGCTATGCTACGC-3′, identical with nt 5307–5333 of the BALB/c Sμ sequence (MUSIGCD07) (15); and R10B, 5′-GGAATTCATAAGCTCAGCCTTGTTCAGCCCATTCCATC-3′, complementary to nt 87–117 upstream of the 3′ end of the Sα segment, except for the addition of an EcoRI site at the 5′ end. The other PCR products that were sequenced were amplified by a single round of PCR using the μ-1 and R10 primers.

For amplification of a 550 bp segment of the neo^r gene from the plasmid a primer at the T7 promoter of Bluescript: 5′-GTAATACGACTCACTATAGGGC-3′ and a primer for the neo^r coding region 5′-ATGGCCGCTTTTCTGGATTC-3′ were used. Amplification was for 30 cycles using HotStar Taq polymerase (Qiagen, Chatsworth, CA). The annealing temperature was 55°C and the concentration of MgCl₂ was 2.5 mM.

Primers for amplification of the TK-Iα segment to produce an 1.8-kb fragment for use as a probe on Southern blots of PCR products contain sequences from the TK gene (TKF2): 5′-TGACTTACTGGCAGGTGCTG-3′ (corresponding to nt 769–789 of GenBank accession no. V00467) and sequences complementary to the Iα segment at −33/−14 (GLαR1): 5′-GGCTGCATGACTGTGTGTCT-3′ (GenBank accession no. L04145).

Two days after transfection of I.29μ cells by electroporation, total cell RNA was prepared using the Ultraspec RNA Isolation System (Biotecx Laboratories, Houston TX) or by the hot-phenol method. cDNA was synthesized from 3 μg RNA using Moloney murine leukemia virus (M-MLV)-Reverse-Transciptase (200 U; Promega, Madison, WI) with the 3′ primer R10 and with the 3′ μ-dn2 primer, 5′-CTTGGTTCTTGGCCAGCCAGCTCTAC-3′ (positions 1037/1062 in MUSIGCD09), under reaction conditions recommended by Promega. PCR was performed for 30 cycles using the Taq Expand High Fidelity System, as described for analysis of Sμ-Sα recombination, except that for amplification of the TK-Sα segment a 5′ primer specific to the HSV-TK gene, TKF2, was used along with the 3′ primer R10. For amplification of the neo^r-Sμ segment, PCR was performed using a 5′ primer complementary to the neo^r gene, Neo4, 5′-CGTCCAGATCATCCTGATC-3′, and the μ-dn2 primer. An equal proportion of the cDNA products were amplified for each sample and 3 μCi [α-³²P]dCTP was included in each reaction. As a positive control for the RT-PCR using the R10 primer, an aliquot of the cDNA obtained from p273-transfected cells was amplified using the 5′ primer IαF, 5′-ACAGGCAATCACACACAGAG-3′, corresponding to the Iα region +13/+33 relative to the first RNA initiation site, along with the 3′ primer R10. As a positive control for the RT-PCR using the μ-dn2 primer, an aliquot of the cDNA obtained from p273- and p200-transfected cells was amplified using a 5′ primer for the V_H sequence in the EP segment, 5′VHN, 5′-TGTTCTCTTTACAGTTACTGAGCACACAGGACC-3′. To control for DNA contamination, PCR was performed on an equivalent amount of RNA which had not been transcribed by reverse transcription. To control for the PCR and to provide m.w. markers, PCRs were also performed on each plasmid DNA (0.5 ng) with the TKF2/R10 primers, with the 5′ Neo/μ-dn2 primers and also on p273 using the IαF/R10 primers and on p273 and p200 using 5′VHN/μ-dn2 primers. PCR products were separated on 1.0% agarose gels, dried, and autoradiographed.

Southern blotting of PCR products was performed as previously described for genomic Southern blotting (42, 44). After each hybridization and exposure to autoradiographic film, the probes were removed from the blots by heating to nearly 100°C in 0.1× SSC and 0.1% SDS. Successful removal of the probes was always verified by autoradiography. Probes were restriction enzyme fragments or PCR products that were labeled by random priming using hexamer primers (Boehringer Mannheim) and Klenow DNA polymerase. The fragments used for plasmid construction were used as hybridization probes in Southern blotting experiments. The Sμ, Sα, and TK-Iα probes were labeled to a similar specific activity (50 × 10⁶ cpm per 25 ng), and equal amounts of cpm were used for each hybridization (with one exception which is noted). The hybridized blots were all autoradiographed for equivalent times.

PCR products obtained after one or two rounds of amplification of DNA from stably transfected clones were subcloned directly into pGEM-T (all except clone 153) or digested with SacI, purified from gels, and subcloned into Bluescript (clone 153). Nucleotide sequencing was performed using an Applied Biosystems (Foster City, CA) model 373 Stretch DNA Sequencer using Applied Biosystems Prism version 2.1.1 software. Sequences were determined at least two times from every template, in both directions when possible. Analysis and alignments of DNA sequences were performed using software from the GCG Wisconsin Package (Genetics Computer Group, Madison WI) and/or the lfasta, lalign and SIMs programs placed on the internet by the Human Genome Center at the Baylor College of Medicine and/or by the Clustal 1.4 program (53).

Quantitation of the PCR fragments detected by Southern blot analyses was performed by measuring by densitometry the signal in the entire lane extending from the maximal size observed for Sμ-Sα fragments down to the smallest detectable Sμ-Sα segments. The densitometer used was a Molecular Dynamics (Sunnyvale, CA) Personal Densitometer SI and analysis was by Molecular Dynamics Image Quant 1.2. Quantitation of the RT-PCR products was performed by a Bio-Rad (Richmond, CA) PhosphorImager.

The maps of the three switch plasmid substrates used in these studies are shown in Fig. 1,A. Each plasmid has a neo^r gene and the Sμ and Sα segments described in Materials and Methods, separated by a herpes virus TK gene. Two of the plasmids, p273 and p200, also have a 1.0-kb fragment (EP) containing the Ig μ intron enhancer and an Ig V_H promoter (50), inserted upstream of the Sμ segment. One plasmid (p273) also has a 535-bp fragment (Iα) containing the promoter and first RNA initiation site for the germline α transcripts inserted upstream of the Sα segment (51). Fig. 1 B diagrams an expected switch recombination event in p273.

The switch plasmids p273 and p200 were transfected into the 22D or 22A10 clones of I.29μ (54, 55), and stably transfected clones were selected by limiting dilution in the presence of G418 (see Materials and Methods). To examine the copy number of the transfected plasmids, we performed Southern blot analyses of genomic DNA from the transfected clones, hybridizing with a probe for the neo^r gene (Fig. 2, A and B), and/or used PCR to amplify a segment extending from the vector backbone into the neo^r gene segment (Fig. 2,C). The location of the primers for the PCR are indicated in Fig. 1 B.

The data suggest that the copy number of integrated plasmids does not differ more than 4-fold among most clones, although occasional clones have many copies (Fig. 2 and data not shown). As a loading control, we show beneath Fig. 2,B the signals obtained from hybridization of the 5-kb cellular TK SacI fragment on the same blot. Similar results were obtained upon analysis of the cellular TK signal for the DNA samples shown in Fig. 2,A (data not shown). There are no consistent differences in the endogenous TK gene or neo^r gene copy numbers among the various cell lines and treatments. In most p273- and p200-transfected clones, the neo^r probe detects a 2.9-kb SacI fragment which contains the 3′ portion of the neo^r gene, the 1-kb EP segment, and 0.3 kb from the 5′ portion of Sμ (see Fig. 1,A). As shown in Fig. 2,A (clone 432) and Fig. 2,B (clone 6), rearrangements in this fragment are occasionally observed. Such rearrangements may be due to recombinations in the EP segment, in the 5′ portion of Sμ or within the neo^r gene, as we have not mapped them. PCR amplification of a 550 bp segment of the neo^r gene from stably transfected plasmids demonstrated that all G418-resistant clones analyzed in this study contain the neo^r segment, including M12.4.1 clone 1 in which the neo fragment was not detected on the genomic Southern blot (Fig. 2,C). To normalize between different PCR experiments, an identical sample of p273 DNA was amplified in every experiment (Fig. 2 C, left-most lanes). To attempt to ensure that these neo gene amplifications can serve as internal controls for the analysis of S-S recombination, the same samples of diluted DNA prepared for the neo amplifications are used in all the experiments described below.

A characteristic of switch recombination that differs from other types of recombination is its inducibility by certain B cell mitogens. Switching from IgM to IgA expression occurs infrequently in I.29μ cells in the absence of LPS plus TGF-β1, as cells expressing both IgM and IgA are rarely observed in untreated cultures (42, 43, 44). TGF-β1 induces transcription from the promoter for germline α transcripts, but the role of LPS is unknown. In addition, nicotinamide has been shown to stimulate class switch recombination in I.29μ cells by inhibiting the nuclear enzyme, poly(ADP-ribose) polymerase (44).

I.29μ cells were electroporated with p273 or p200, divided into aliquots that were treated with LPS, LPS plus TGF-β1, LPS plus TGF-β1 plus nicotinamide, or were left untreated and stable transfectants were selected and cloned by limiting dilution. We verified that switching of the endogenous Ig genes had indeed been induced by using immunofluorescence microscopy to measure sIgM and sIgA expression on day 2 of induction (44), the day the cells were distributed into wells for selection of clones (data not shown). In the optimally induced cultures, an average of 6.5% of the cells expressed IgA (data not shown).

Switch recombination occurs by direct recombination between switch regions and thus we wanted to evaluate whether the switch plasmids could undergo direct Sμ-Sα recombination. We developed a PCR assay for recombination between the Sμ and Sα segments, using oligonucleotide primers complementary to the 5′ end of Sμ and to the 3′ end of Sα (see Fig. 1,B). This strategy was previously used to amplify endogenous Sμ-Sγ3 recombinants (56). The μ primers are specific for the BALB/c Sμ sequence in the plasmid, but the Sα primer is complementary to both the plasmid and endogenous Sα sequences. The primers amplify Sμ-Sα recombinants from the transfected plasmid, but not from endogenous genes of the stably transfected clones (see below). After one round of PCR amplification, the products were analyzed by Southern blotting, hybridizing sequentially with three probes, Sμ, Sα, and a probe containing both the TK and Iα segments from p273. The probes correspond to the segments indicated in the plasmid maps (Fig. 1). PCR fragments that are due to direct Sμ-Sα recombination and therefore lack the TK and Iα segments should be ≤3.3 kb in size.

Southern blots of PCR products obtained from the genomic DNA of I.29μ clones stably transfected with p273 that were cultured in medium alone (untreated) and from clones cultured with the indicated inducers of switching are shown in Fig. 3. The blots on the left were hybridized with the Sμ probe, the blots in the middle panels are the identical blots hybridized with the Sα probe, and the blots on the right are the identical blots hybridized with the TK-Iα probe. The dots on the edge of the bands in the Sμ blots indicate bands that also hybridize with the Sα probe. The great majority of the amplified segments are detected by both the Sμ and Sα probes, but not with the TK-Iα probe. However, from three of the transfected clones that were not treated with inducers, 5.8-kb fragments that hybridize with the Sμ, Sα, and TK-Iα probes and which correspond to unrecombined plasmid were amplified (Fig. 3). (The 5.8-kb band in clone 438 can be seen on the original autoradiograph.) None of the amplified fragments from clones induced to undergo switching hybridized with the TK-Iα probe, except for two fragments from one clone induced with LPS without TGF-β1. These data suggest that the transfected switch plasmid has undergone numerous direct Sμ-Sα recombination events in I.29μ cells. An average of 7.9 (±0.6) Sμ-Sα PCR products were amplified from each p273-transfected I.29μ clone in two experiments, whether or not the cells were treated with inducers of class switch recombination (Table I).

Because I.29μ cells undergo switch recombination on their endogenous chromosomes, it was important to determine whether the comigrating Sμ-Sα bands detected in this experiment could be due to endogenous switch recombination. To examine this possibility, we performed the same PCR and Southern blot analysis on DNA from a line of I.29μ cells in which 50% of the cells express IgA (see Fig. 9 below, left-most lanes in all three panels). A faint band (∼4 kb) was detected with both the Sα and TK-Iα probes (perhaps due to amplification of an endogenous Iα-Sα fragment), and a different sized band (∼2.5 kb) was detected with the Sμ probe. Thus, the primers used do not appear to amplify the endogenous Sμ-Sα junctions in IgA⁺ cells derived from the I.29μ cell line. Further evidence for this comes from an examination of one of the p273-transfected clones shown in Fig. 3 (no. 458), which was derived from cells induced with LPS plus TGF-β1. This clone did not yield Sμ-Sα PCR products, although it yielded a neo^r PCR product (Fig. 2 C) and contains endogenous genomic Sμ and Sα fragments, as shown by genomic Southern blot analysis (data not shown). Nucleotide sequencing data described below further support the conclusion that the Sμ-Sα fragments amplified from p273-transfected I.29μ cells are due to recombination in the plasmid and are not due to recombination of endogenous Sμ and Sα sequences.

To examine the nature of the recombination events that produced the comigrating Sμ and Sα fragments, we determined the nucleotide sequences of six Sμ-Sα segments amplified from I.29μ cells stably transfected with p273. The PCR products were cloned from five independent clones of p273-transfected I.29μ cells, 153, EW4, EW7, EW18, and JI. Southern blot analyses of the PCR products from three of these clones are shown in Fig. 4. These DNA samples were chosen for analysis due to the presence of PCR products that could be detected on an ethidium bromide stained gel (data not shown). PCR products were excised from the gel, cloned, and sequenced. The sequences show direct Sμ to Sα junctions with 0, 1, 2, 6, or 8 nt of identity between the Sμ and Sα sequences at the recombination sites (Fig. 5), quite comparable to switch recombination junctions in genomic DNA (2).

The nucleotide sequences of three products from three independent nested PCRs from clone 153 were identical and demonstrate direct Sμ-Sα recombination. Fig. 5,A presents the sequence surrounding the Sμ-Sα junction of clone 153 aligned with the plasmid Sμ and Sα sequences. To search for evidence of nucleotide substitutions, deletions, or insertions in comparison with the parental p273 plasmid, as would be expected if switch recombination activities created the Sμ-Sα junction, we compared the entire Sμ and Sα sequences of clone 153 with the p273 sequence (Fig. 6 A and data not shown). The Sμ region of clone 153 has undergone one or more deletion events during or subsequent to switch recombination. The deletions were not generated during PCR amplification because they are present in PCR products obtained from three independent reactions. Due to deletions in the 153 sequence, one cannot be completely confident of the entire alignment with the p273 Sμ sequence. However, in the regions at the 5′ and 3′ ends of the sequence, the alignments obtained by three programs using different algorithms are identical. In these regions the 153 Sμ sequence has 9-nt differences from the p273 sequence (in bold type). The Sα sequence of 153 is identical with the Sα sequence in p273 (data not shown).

Due to the finding of differences from the p273 Sμ sequence, we considered the possibility that the Sμ sequence of clone 153 may have been derived by recombination with endogenous Sμ sequences, rather than due to recombination within the p273 plasmid. We compared the Sμ sequence of clone 153 with the endogenous I.29 Sμ sequence and with the p273 sequence and found that the 153 sequence matches the p273 sequence better (Fig. 6,B) The Sμ segments of all other Sμ-Sα PCR products in Fig. 5 were also compared with the I.29 Sμ sequence and found not to match it as well as they match the p273 Sμ sequence (data not shown). Therefore, all of the PCR products analyzed by nucleotide sequencing originated from Sμ-Sα recombination events in the switch plasmid.

Unlike the 153 PCR products, the nucleotide sequences of two independent PCR products (EW 4.1) from another p273-transfected I.29μ clone (EW 4) show no differences from the plasmid Sμ and Sα sequences. The sequence of EW 4.1 surrounding the Sμ-Sα junction is shown in Fig. 5 B. The junction occurs within a 6-nt identity (underlined) between the Sμ and Sα sequences.

The nucleotide sequences of four other PCR products were also sequenced. These products were each obtained from only one PCR. Fig. 5, C--F, presents segments of the sequences surrounding the sites of Sμ-Sα recombination. Although they have several nucleotide differences in comparison to the plasmid Sμ and Sα sequences, we do not know whether these were generated during the PCR or occurred during switch recombination. Altogether, the characteristics of the sequences of the switch recombination junctions from transfected p273 are typical of junctions from endogenous genes, including the finding of deletions, both mutated and unmutated sequences and the finding that recombination occurs at sites of no or short sequence identities between the Sμ and Sα sequences.

To determine whether transcription through the Sα segment is required for switch recombination in the plasmid, as it is for endogenous class switch recombination, we tested the ability of p200, a plasmid which lacks the germline (Iα) promoter upstream of the Sα segment, to undergo Sμ-Sα recombination. Twenty clones of I.29μ cells stably transfected with p200, which were untreated or treated with LPS plus TGF-β1, were analyzed by PCR followed by Southern blot analysis. An average of 3.6 (±0.36) Sμ-Sα fragments that were devoid of TK-Iα signal were detected per clone, whether untreated or treated with LPS plus TGF-β1 (Fig. 7, Table I). Although the difference between the numbers of S/S fragments obtained from p273 and p200 is only about 2-fold, the difference is statistically significant (p < 0.001).

As true for p273, the numbers of Sμ-Sα fragments obtained per p200-transfected clone are similar in untreated and induced cells. However, unlike p273, the intensities of the hybridization signals of the amplified switch fragments from induced cells is greater than from uninduced cells. This result suggests that the proportion of cells in the clone which have each fragment is greater in the induced cells than in the uninduced cells. The difference in intensities of the Sμ-Sα fragments in untreated and induced cells is unlikely to be due to different amounts of DNA used in the amplifications or due to different amounts of plasmids in these cells, as attested to by the similar signals obtained from the neo^r gene amplification of these same DNA samples (Fig. 2 C). Thus, although S/S recombination occurs in p200 in the absence of LPS plus TGF-β1 treatment, it is possible that it occurs earlier or its frequency might be enhanced in induced cells.

We also examined Sμ-Sα recombination in a plasmid lacking transcriptional activator elements upstream of either S region (p218, see Fig. 1) and found an average of 5.6 (±0.54) Sμ-Sα PCR products per clone (Table I). Therefore, it appears that Sμ-Sα recombination within the plasmid does not require the germline Iα promoter segment upstream of the Sα segment nor the EP segment located upstream of the Sμ segment. However, it appears to be somewhat enhanced by these elements. The finding that transcriptional activator elements are not required appears to differ entirely from chromosomal switch recombination, as it has been shown that transcription of germline RNA is essential for CSR (29, 30, 32).

It was possible that recombination in the plasmid is not eliminated by deletion of the germline Iα promoter due to run-through transcription from the TK gene. We examined this possibility by examining RNA transcripts from plasmids in I.29μ cells 2 days after transfection. Total cell RNA was purified, transcribed with reverse transcriptase using the same Sα primer (R10) that was used for the PCR experiments. The reverse transcriptase products were amplified by PCR using the Sα primer along with either a 5′ primer that binds near the beginning of the TK gene or a 5′ primer that binds just downstream of the RNA initiation site within the Iα segment (see Fig. 1,A). The Iα/R10 primers successfully amplified a 2.3-kb product from p273-transfected cell cDNA, demonstrating the presence of the expected Iα-Sα transcript (Fig. 8 A, lane 11). The TK/R10 primers amplified very low amounts of 3.8- and 3.3-kb products from cDNA of cells transfected with p273, p200 or p218 (lanes 2, 5, and 8) corresponding to the sizes of comparable fragments amplified from the plasmids (lanes 1, 4, and 7). These data suggest that run-through transcripts from the TK gene are present in very low amounts relative to the Iα/Sα transcripts (0.5–1.0% of the Iα-Sα RNA, as determined by phosphorimaging).

Although the neo^r segment of the plasmid is oriented in the anti-sense direction relative to the Sμ segment, it was possible that transcription through the Sμ segment was also occurring in the plasmid lacking the EP segment (p218). To address this possibility, a similar RT-PCR experiment was performed, using cDNA transcribed from RNA of the transfected cells with a 3′ Sμ primer, μ-dn2, indicated in Fig. 1,A. The cDNA preparation was subjected to amplification using either a 5′ V_H primer (5′VHN) or a 5′ neo primer (Neo4) and the μ-dn2 primer (Fig. 8 B). The 5′VHN/μ-dn2 primers readily amplified cDNAs of the correct size from cells transfected with either p273 or p200 (lanes 11 and 13), indicating the presence of reverse transcriptase products in the reactions. However, no RT-PCR products were detected from any transfected cell RNA with the combination of the Neo4/μ-dn2 primers (lanes 2, 5, and 8). The very low level of transcription through the Sα segments of p218 and p200 may be sufficient to support recombination in the plasmid or, alternatively, transcription may not be required for switch recombination on the plasmid.

Because many B cell lines (and also non-B cell lines) do not undergo class switch recombination, we wanted to determine whether switch recombination in p273 maintains this cell type specificity. The mouse cell lines that we examined were 1) M12.4.1, a B lymphoma line that does not express sIg and was derived from a sIgG2a⁺ line (M12) that has not been reported to undergo class switching in culture; 2) the plasmacytoma line J558L, which secretes λ light chains; and 3) two T cell lines, EL-4 and BW5147. M12.4.1 and EL-4 cells both express a transiently transfected reporter gene driven by the germline α (Iα) segment in the presence or absence of the Ig μ intron enhancer, whereas J558L and BW5147 cells do not (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation) (M. J. Shi, laboratory observations).

To determine whether p273 undergoes recombination in M12.4.1 B lymphoma cells, stably transfected clones were derived and analyzed by PCR (Fig. 9 and Table I). An average of 1.3 (±0.7) fragments per clone that hybridize with both Sμ and Sα probes, but not with the TK-Iα probe, were detected. This is 6-fold fewer than in I.29μ cells. Another method for comparing the frequency of Sμ-Sα recombination is to compare the intensity of the hybridization signals obtained from the two cell types. We performed this analysis by comparative densitometric analysis of the Sμ and Sα signals from the entire lane of PCR products from the p273-transfected M12.4.1 clone 363 with those from all the I.29μ lanes on the same gel (bottom panels in Fig. 3). The intensity of the hybridization signals in the M12.4.1 lane were about 2.5% of those obtained from p273-transfected I.29μ cells. This difference in signal intensity does not appear to be due to a smaller amount of plasmid in the M12.4.1 cells because the amplified neo^r bands are of comparable intensities in the p273–transfected I.29μ and M12.4.1 clones (Fig. 2 C).

Altogether, these results indicate that Sμ-Sα recombination in the plasmid occurs less frequently and in a much smaller proportion of the cells in each M12.4.1 clone than in the I.29μ clones. The low frequency of Sμ-Sα recombination may be due to low amounts of putative switch enzymes or other proteins required for switch recombination. Surprisingly, the 5.8-kb band from unrecombined plasmid was not detected, suggesting that p273 undergoes frequent rearrangement events in M12.4.1 cells that do not result in Sμ-Sα recombination. This finding is consistent with a genomic Southern blot analysis in which it was found that no Sα band was detectable in 7 of 12 and no plasmid TK segment was detected in 6 of 12 p273-transfected M12.4.1 clones (data not shown).

We also used the PCR assay to examine p273-transfected J558L plasmacytoma clones. Plasmacytoma cells correspond to plasma cells that are differentiated to secrete large amounts of Ab and that do not undergo class switching (57). The 5.8-kb Sμ-TK-Iα-Sα segment was detected in 7 of the 12 clones examined, and only four very faint Sμ-Sα bands were detected (Fig. 10 and Table I). All of the clones contain a neo^r gene (Fig. 2 C). Thus, the frequency of Sμ-Sα recombination of p273 in J558L cells was 0.3 (±0.2) events per clone, which is 26-fold lower than in I.29μ cells. Thus, like M12.4.1 cells, J558L cells appear to contain only low levels of switch recombination activity.

To further explore the cell type specificity of Sμ-Sα recombination in p273, we assessed recombination between the plasmid switch regions in the T cell lines, EL-4 and BW5147, which were stably transfected and then treated with the T cell activators PMA and ionomycin (EL-4), or left untreated (EL-4 and BW5147). The PCR assay detected an average of 3.1 (±0.3) Sμ-Sα fragments that were devoid of detectable TK-Iα signal per EL-4 clone and an average of 2.4 (±0.3) such fragments in the BW5147 clones (Fig. 11). These values are 2.5- and 3.3-fold lower, respectively, than that obtained with I.29μ cells and are significantly different (p < 0.001; Table I). Unlike results with p273-transfected I.29μ cells, more than half of the PCR products from the T cell lines were ≤0.5 kb in length. These fragments were obtained in two sizes and appeared identical in every clone and were obtained in two different EL-4 and two different BW5147 transfection experiments (data not shown). These fragments were not obtained when DNA from p273-transfected I.29μ clone 472 was amplified in the same experiment with BW5147 (Fig. 11, left-most lanes). The ≤0.5-kb fragments are not due to primer dimers because they were not obtained in reactions performed without template DNA (lanes labeled N.T.). Although comparable ≤0.5-kb fragments are observed in some p273-transfected I.29μ cells (Fig. 3), they are more variable and contribute a much smaller portion of the total PCR products. Thus, EL-4 and perhaps also BW5147 cells may have some switch recombinase activity, although the activity appears to differ qualitatively from that in I.29μ cells.

The properties of the stably transfected plasmids described in this report suggest that they are undergoing switch recombination. The most compelling of these properties is that the frequency of Sμ-Sα recombination in the optimal switch plasmid, p273, is greater in the B cell line I.29μ, which undergoes switch recombination of its endogenous heavy chain genes, than in four cell lines that do not undergo switching of their endogenous genes. In addition to yielding a lower number of amplified plasmid Sμ-Sα fragments, three of the four nonswitching cell lines produced Sμ-Sα fragments that gave lower hybridization signals than did I.29μ cells, indicating they were present in a smaller proportion of the cells of the clone. The T cell line, EL-4, gave the highest level of Sμ-Sα recombination of the four nonswitching lines, although the difference from I.29μ is still statistically highly significant. Furthermore there are qualitative differences in the Sμ-Sα bands found in I.29μ cells and the two T cell lines, because the majority of the amplified fragments in the two T cell lines were small, ≤0.5 kb, and were identical among all clones. The finding of some switch activity in EL-4 cells is consistent with results obtained by Daniels and Lieber (35) using their switch plasmid. We conclude that one reason T cells do not undergo switch recombination in vivo may be because they express low levels of switch recombinase activity, and it is also likely that the switch regions are inaccessible to recombinase activity in the endogenous loci.

Recombination events which occur soon after transfection of the plasmids will be present in a high proportion of the cells in a clone and give intense hybridization signals. Cells in which the plasmid recombines frequently would have a higher probability of recombining soon after transfection and therefore would have more intense Sμ-S bands. Thus, the difference in intensity of the Sμ-Sα bands in I.29μ in comparison with M12.4.1, J558L, and BW5147 indicates that Sμ-Sα recombination begins more rapidly and/or occurs more frequently in I.29μ cells than in M12.4.1, J558L, and in BW5147 cells.

Because Sμ-Sα recombination can result in loss of targets for further switch recombination, cells in which switch recombination occurs frequently can saturate the assay. This is consistent with the finding that when switch recombination of p273 is assayed in a transient assay (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation) there is a greater difference in the numbers of plasmid switch fragments obtained from I.29μ and M12.4.1 cells than when assayed in the stable assay described here. It is possible that in cell lines in which the bands are faint, e.g. M12.4.1 and BW5147, that recombinations occur infrequently but continue to accumulate if the plasmid integrates at a permissive site in the chromosome. In I.29μ, recombination occurs frequently, thus resulting in intense bands and probably saturation of the assay.

Evidence from transient transfection experiments with these plasmids indicates that Sμ-Sα recombination begins before integration of the plasmids in I.29μ cells (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation). Whether recombination continues after integration has not been established. Data presented here suggest that it may continue after integration, because the G418-resistant clones obtained by limiting dilution cultures have different Sμ-Sα fragments. However, it is also possible that the different Sμ-Sα fragments could be due to integration of different recombined plasmids after cloning on day 2. To attempt to determine whether integrated plasmids can recombine, we subcloned a few clones that have integrated intact plasmids in the presence of gancyclovir and LPS plus TGF-β1. Very few gancyclovir-resistant cells were obtained, suggesting that these plasmids were not accessible to recombinase.

Similarly to the previously reported plasmid switch substrates (33, 34, 35, 36, 37), the plasmids reported here undergo recombination events in addition to switch recombination. These other recombinations occur in all cell lines we examined, except perhaps in J558L cells. This was learned from extensive genomic Southern blot analyses of stably transfected clones. For example, the TK gene of p273 is undetectable in many clones that yield very few or no Sμ-Sα PCR products. It was deleted from 50% of the M12.4.1 clones, 67% of the EL-4 clones, 70% of I.29μ clones, but present in unrearranged form in all of the J558L clones. The Sα segment of the plasmid is also frequently deleted, except in J558L.

Deletion of the Sα segment would result in loss of the 3′ primer binding sites. Deletion of the Sα segment can explain why we do not usually detect plasmid bands containing the TK gene in the PCR assay, even in cells which retain the plasmid TK gene (identified by genomic Southern blotting). However, the loss of primer binding sites cannot explain why the plasmid infrequently undergoes Sμ-Sα recombination in M12.4.1 or the T cell lines, because in experiments in which the plasmids were isolated 2 days after transfection, at which time most of the plasmids in nuclei are present in the intact form (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation), a greater difference is found between the yield of Sμ-Sα PCR products from I.29μ and the nonswitching cell lines than found here using the stable assay (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation). Further information about the other recombination events in the plasmids is reported by Shanmugam et al. (A. Shanmugam, M.-J. Shi, J. Stavnezer, and A. L. Kenter, manuscript in preparation). Because of these other recombination events, the use of drug selection against the TK gene does not provide a specific assay for Sμ-Sα recombination on the plasmid. However, the use of the PCR assay allows one to specifically assay switch recombination on the plasmid.

The fact that transcriptional promoters upstream of the Sμ or Sα segments do not appear to stimulate recombination between switch regions in the transfected plasmids is inconsistent with what is known about switch recombination of endogenous genes. The requirement for promoters upstream of S regions appears to be absolute in vivo. We examined whether this was because transcription from the TK promoter may continue through the Sα segment and found a very low level of run-through transcripts (0.5–1.0% of the level of transcription driven by the Iα promoter). Further support for a lack of requirement for a high rate of transcription was obtained by the finding that p218, which has no transcriptional activator segment upstream of either S region segments, undergoes Sμ-Sα recombination in I.29μ cells about as frequently as does p200.

The lack of requirement for specific transcriptional elements has several possible explanations. It is possible that low levels of run-through transcripts initiating in segments upstream of the S regions or initiating within the S regions themselves are sufficient for recombination. Alternatively, it is possible that whatever the role of germline transcripts, this role is not required for recombination of these plasmids. It is possible that the role for germline transcripts in vivo may be to contribute to synapsis of the switch regions and due to their close proximity in the plasmid, this is not required. Perhaps the chromatin structure of the transfected plasmid does not interfere with switch recombination, unlike the chromatin of endogenous switch regions and thus germline transcription is not required.

The presence of mutations and deletions on one side of the switch recombination junction in the sequence from clone 153 is similar to endogenous switch recombination events (2). These observations are predicted by a model for switch recombination involving error-prone DNA synthesis in which one switch region (donor) primes DNA synthesis on another (acceptor) switch region (2, 14). Simultaneously, the other strand of the acceptor switch region primes DNA synthesis on the donor switch region. According to this model, after recombination and segregation of the products into two different daughter cells, one of the two daughter cells has a newly synthesized acceptor switch region (Sα) which is mutated, and the other daughter has a newly synthesized (mutated) donor switch region (Sμ). The 153 sequence corresponds to the latter type of product.

The illegitimate priming model might seem to predict the finding of sequence identities between the Sμ and Sα sequences at the sites of switch recombination. Although such identities are found at switch recombination junctions which occur in endogenous genes and are also observed in the PCR products from the switch plasmid, there is no preference for junctions with long identities (2). Thus, switch recombination differs from homologous recombination. The finding of short bits of identity at recombination junctions is typical of illegitimate recombination, or end-joining, in mammals (58). It is also typical of V(D)J recombination (59). An observation that is consistent with the finding that error-prone DNA synthesis occurs across switch recombination junctions (2, 13, 14) and also has similarity to recombination by end-joining is the finding that end-joining in germinal vesicle extracts from Xenopus oocytes requires deoxynucleotide triphosphates and thus appears to involve DNA synthesis (60).

Although induction of switch recombination of the endogenous heavy chain genes in I.29μ cells requires addition of LPS, recombination of the switch regions in transfected p273 occurs without treatment of I.29μ cells with LPS. The function of LPS in inducing switch recombination is not understood, as it does not increase the rate of transcription of germline α transcripts (43), nor does it increase proliferation of I.29μ cells (data not shown). One possibility was that it induces recombinase activity. However, the finding that switch recombination of the plasmid does not require LPS treatment suggests that the switch recombinase activity is present constitutively in I.29μ cells, but the endogenous loci do not recombine due to a lack of accessibility. This result is reminiscent of the finding that the V(D)J recombinase enzymes, Rag-1 and Rag-2, are expressed in developing B lineage cells at stages in which endogenous loci are inaccessible to their activity. Specific chromatin changes are required to induce accessibility of the Ig genes to recombinase (61). The role of LPS may be to induce chromatin remodeling, resulting in accessibility. It is possible that the endogenous switch regions are not accessible to the hypothetical constitutive recombinase activities due to proteins bound to switch regions that protect them from recombinase activities, and that the role of LPS is to remove the repressor proteins.

Kinoshita et al. (62) recently demonstrated switch recombination in a plasmid substrate which was stably integrated into the B cell line CH12F3-2, a cell line capable of undergoing switch recombination of its endogenous genes. They found that intact integrated plasmids did not undergo constitutive recombination, but did recombine when cells were treated with reagents, e.g., cytokines and CD40 ligand, which induce switch recombination of the endogenous genes. We postulate that the function of the inducers is to stimulate accessibility of the integrated plasmid. In contrast, our plasmid begins to recombine before integration and thus a major difference between their experiments and ours is that our plasmid does not have the constraints of endogenous chromatin which may regulate the recombination. In conclusion, although switch recombination of the endogenous genes in I.29μ cells requires treatment with LPS, recombination of the plasmid does not. This result suggests that the switch recombinase is constitutive in I.29μ cells and that LPS may induce chromatin changes at the endogenous loci, allowing accessibility to the recombinase.

We thank Dr. Meng-Jiao Shi for examining the activity of reporter genes driven by transcriptional elements from the switch plasmids; Drs. Ako Jacintho, Gang Qiu, and Özlem Yazar and Mr. Curtis Barrett for performing some of the preliminary Southern blotting experiments; and Ms. Jennifer Nietupski for technical assistance. We thank Drs. Rachel Gerstein and Carol Schrader for helpful comments on the manuscript and members of our laboratories for helpful discussion. We thank Drs. Fred Alt for pEP.B, Allan Bradley for the pPGK-neo plasmid, and Ken B. Marcu for pZN(Sμ-Sγ2b)tk.1.