The catalytic subunit of DNA–protein kinase (DNA-PKcs) is not required for Ig class-switch recombination

Abstract

The joining of DNA ends during Ig class-switch recombination (CSR) is thought to involve the same nonhomologous end-joining pathway as used in V(D)J recombination. However, we reported earlier that CSR can readily occur in Ig transgenic SCID mice lacking DNA-dependent protein kinase (DNA-PK) activity, a critical enzymatic activity for V(D)J recombination. We were thus led to question whether the catalytic subunit of DNA-PK (DNA-PKcs) is essential for CSR. To address this issue, we asked whether class switching to different Ig isotypes could occur in a line of Ig transgenic mice lacking detectable DNA-PKcs protein. The answer was affirmative. We conclude that joining of DNA ends during CSR does not require DNA-PKcs and can occur by an alternative repair pathway to that used for V(D)J recombination.

The Ig H chain locus undergoes two kinds of rearrangement during B cell differentiation. Assembly of V, D, and J elements into VDJ coding segments for the variable region of H chains (VDJH recombination) occurs early in B cell differentiation. In late B cell differentiation, a given VDJH coding segment may be joined to different H chain constant region genes, thus enabling B cells to express the same H chain variable region with different classes of Ig.

Both class-switch recombination (CSR) and VDJH recombination entail pair-wise cuts of DNA at targeted sites, deletion of the intervening DNA and nonhomologous DNA end-joining (NHEJ) (for reviews, see refs. 1–4). Moreover, there is evidence that joining of DNA ends in CSR involves the same (classical) NHEJ pathway used in V(D)J recombination (5–9). For example, class switching to different Ig isotypes has been reported defective in Ig transgenic mice lacking Ku70 or Ku80 (5, 6, 9), two proteins known to play a critical role in the NHEJ pathway (10). Class-switching to different Ig isotypes (excepting IgG1) has also been reported to be defective in B cells of an Ig transgenic mouse line lacking the catalytic subunit of DNA-dependent protein kinase (DNA-PKcs) (8), another key protein in the NHEJ pathway. Interestingly, CSR has been shown to occur in DNA-PKcs-deficient B cells in Ig transgenic SCID mice (11, 12). SCID cells express a truncated and enzymatically dead DNA-PKcs protein (13–16). Thus, the SCID results could be taken to imply that DNA-PKcs plays only a structural role in CSR. An alternative interpretation, and one that we favor, is that DNA-PKcs is dispensable for CSR because, based on the work of others (17–21), any structural role for DNA-PKcs in NHEJ is likely to depend on its ability to phosphorylate itself and other factors involved in NHEJ.

Whether DNA-PKcs is in fact essential for class-switching to different Ig isotypes is still unresolved and controversial. In an effort to resolve this controversy, we tested the ability of B cells from a line of transgenic mice lacking DNA-PKcs to undergo class-switching to different Ig isotypes. The mice were obtained by selectively crossing into BALB/c DNA-PKcs^−/− mice the VDJH and VJκ coding sequences of the 3H9 (22, 23) and Vκ8 (24, 25) transgenes (tgs). Together, these two tgs code for an antibody with anti-self (DNA) specificity (22, 26). As shown previously, mice hemizygous for these two tgs and lacking DNA-PKcs activity (3H9Vκ8 SCID mice) contain inactive B cells that can be induced to undergo class-switching with nearly the same efficiency as 3H9Vκ8 wild-type controls (11). Similar results were obtained in this study with 3H9Vκ8 DNA-PKcs^−/− mice, with 3H9Vκ8 SCID and 3H9Vκ8 RAG1^−/− mice as positive controls. We found that 3H9Vκ8 DNA-PKcs^−/− mice contained ≈70% the normal number of B cells but generally lacked serum Ig, indicating that the B cells in these mice were inactive. However, B cell activation, differentiation, and class-switching readily occurred in these mice when they were engrafted with T cells from B cell-deficient donors (JH^−/− mice). This was evident from the appearance of sera IgG1, IgG2a, IgG2b, and IgA commensurate with the appearance of donor T cells in the peripheral blood of engrafted recipients. Our results indicate class-switching to different Ig isotypes can readily occur by a DNA-PKcs-independent pathway.

Results

3H9Vκ8 DNA-PKcs^−/− Mice: Selective Breeding and Confirmed Absence of DNA-PKcs.

Mice with disrupted DNA-PKcs alleles (DNA-PKcs^−/−) and hemizygous for the site-directed 3H9 and Vκ8 tgs (23, 25) were obtained by selective crossing of the tgs from 3H9/3H9, Vκ8/Vκ8 SCID mice (11) into BALB/c DNA-PKcs^−/− mice (provided by G. Taccioli, Boston University School of Medicine, Boston, MA). The latter mice were homozygous for a disrupted DNA-PKcs allele derived from a line of 129/C57BL6 chimeric mice that lacked detectable DNA-PKcs (27). The initial selection was for mice lacking the mutant scid allele. Segregation of the scid allele and its counterpart (i.e., lacking the scid mutation and bearing the upstream DNA-PKcs disruption) was monitored by using an allele-specific PCR assay with confronting two-pair primers (28). PCR products corresponding to an allele with and without the scid mutation (180 and 101 bp, respectively) enabled us to unambiguously identify mice homozygous for the disrupted DNA-PKcs allele (see Fig. 1A). Further selective breeding to obtain 3H9/3H9, Vκ8/Vκ8, DNA-PKcs^−/− mice for the purpose of producing 3H9/+, Vκ8/+, DNA-PKcs^−/− mice (hereafter referred to as 3H9Vκ8 DNA-PKcs^−/− mice) is described in Materials and Methods.

Fig. 1.

Genotyping of transgenic DNA-PKcs^−/− mice. (A) 3H9/3H9, Vκ8/Vκ8 SCID mice (tg-s/s mice) were back-crossed several generations to BALB/c DNA-PKcs^−/− mice (n/n mice). Transgenic backcross mice (tg-s/n) were intercrossed to obtain tg-n/n mice. The segregation of the scid (s) and DNA-PKcs⁻ (n) alleles was monitored by using the two-pair confronting PCR assay of Muruyama et al. (28). The mutant scid allele gives a PCR product of 180 bp as shown in the DNA from C.B-17 scid (s/s) and tg-s/+ mice. Because the DNA-PKcs⁻ allele lacks the scid mutation, it types as a WT (+) allele in this assay and gives a PCR product of 101 bp as shown in DNA from C.B-17 WT (+/+), tg-s/n, and tg-n/n mice. The assay also gives a common PCR product (257 bp). Size markers (M) are shown on the far left. (B) Absence of detectable DNA-PKcs in 3H9Vκ8 DNA-PKcs^−/− mice. The Western blot contains spleen lysates from C57BL6 (B6), C.B-17 heterozygous for scid (s/+), 3H9Vκ8 RAG1^−/− (tg-r/r), 3H9Vκ8 DNA-PKcs^−/− (tg-n/n), and BALB/c (C) mice. Filters were overlaid with IgG1 anti-DNA-PKcs antibody (clone 18.2) followed by HRP-conjugated anti-mouse-IgG1 antibody and developed with ECL substrate. DNA-PKcs corresponds to the band at 460 kDa. (C) Western blot illustrating apparent greater abundance of DNA-PKcs in spleen lysates of B6 and (B6 × BALB/c) F1 mice, denoted as B6xC, than in spleen lysates of BALB/c (C) and s/+ mice. The 220-kDa band, seen in all lanes, represents a protein that cross-reacts with the 18.2 monoclonal anti-DNA-PKcs antibody (13, 27) and serves as a control for protein loading.

To confirm the absence of detectable DNA-PKcs in 3H9Vκ8 DNA-PKcs^−/− mice, whole spleen cell lysates from these mice and control mice were tested for DNA-PKcs protein by Western analysis. As illustrated in Fig. 1B, DNA-PKcs (460 kDa) was not detectable in lysates of 3H9Vκ8 DNA-PKcs^−/− mice. Lysates from the spleen of C57BL6, C.B-17 (an Ig congenic partner strain of BALB/c) mice heterozygous for scid and from 3H9Vκ8 RAG1^−/− mice were included as positive controls. Of particular interest is that the intensity of the protein band corresponding to DNA-PKcs in BALB/c and its Ig congenic partner strain (C.B-17) was always found to be weaker than the DNA-PKcs protein band in C57BL6 and (C57BL6 x BALB/c) F1 mice (Fig. 1C). We interpret this finding to reflect a lower abundance of DNA-PKcs in BALB/c and C.B-17 than in mice with a C57BL6 genetic background. As discussed later, this difference could have implications for potential mouse strain differences in the dependency of CSR on DNA-PKcs.

3H9Vκ8 DNA-PKcs^−/− Mice Contain Near-Normal Numbers of B Cells but Lack Serum Ig.

FACS analysis showed 3H9Vκ8 DNA-PKcs^−/− mice, as well as 3H9Vκ8 SCID and 3H9Vκ8 RAG1^−/− mice, to contain ≈60–70% the normal number of splenic B cells (Fig. 2 and Table 1). However, these mice generally lacked sera IgM (data not shown), IgG, and IgA (see Ig isotype concentrations at the 0-week interval in Figs. 3 and 4). Although all transgenic mice tested were found to be deficient for B220⁻CD3⁺ splenic T cells (Fig. 2), a screen of the peripheral blood of many transgenic mice revealed “leaky” CD3⁺ T cells, representing 0.5–2.7% of peripheral blood lymphocytes, in ≈15% of 3H9Vκ8 DNA-PKcs^−/− and 3H9Vκ8 SCID mice; as expected, no CD3⁺ T cells were detected in 3H9Vκ8 RAG1^−/− mice (data not shown). Thus, except for the presence of a few leaky T cells in some mice, the B and T cell phenotype of 3H9Vκ8 DNA-PKcs^−/− mice was equivalent to that of 3H9Vκ8 SCID and 3H9Vκ8 RAG1^−/− mice.

Fig. 2.

FACS profiles for B and T cell phenotype of 3H9Vκ8 DNA-PKcs^−/− mice and control 3H9Vκ8 SCID and 3H9Vκ8 RAG1^−/− mice. Lymphocyte-gated spleen cells from individual mice were analyzed for CD45(B220) versus IgM and CD45(B220) versus CD3. The numbers indicate the percentage of B (B220⁺IgM⁺) and T (B220⁻CD3⁺) cells within the boxed areas. Forward and light-angle scatter gates excluded nonlymphoid cells. Five to eight mice of each genotype were analyzed, and the profiles shown are representative.

Table 1.

Spleen cellularity and number of splenic B cells in 3H9Vκ8 transgenic lines and nontransgenic controls

Genotype	No. of mice	Number (× 10−6)
		Spleen cells	B cells
non-tg r/r	7	3.3 ± 1.7	<0.2
3H9Vκ8 r/r	8	73 ± 7	22 ± 8
3H9Vκ8 n/n	6	90 ± 37	27 ± 3
3H9Vκ8 s/s	7	83 ± 25	25 ± 8
non-tg s/+	8	107 ± 13	38 ± 9

The values shown represent the mean number (± SEM) of spleen cells and B (B220⁺IgM⁺) cells for the indicated number of mice analyzed. Spleen cell counts were made in a hemacytometer, and the number of viable cells was scored on the basis of their ability to exclude trypan blue. The number of B220⁺IgM⁺ cells was determined by flow cytometry (see Materials and Methods). r/r, mice with both RAG1 alleles disrupted; n/n, mice with both DNA-PKcs alleles disrupted; s/+, mice heterozygous for the scid mutation; non-tg, nontransgenic.

Fig. 3.

Reconstitution of 3H9Vκ8 DNA-PKcs^−/− mice with T cells from C.B-17 JH^−/− donors results in production of serum IgG isotypes comparable with that in T cell reconstituted 3H9Vκ8 RAG1^−/− controls. Thymocytes (1 × 10⁶) and bone marrow cells (5 × 10⁶) from JH^−/− mice were admixed and injected i.v. into 8- to 10-week-old transgenic recipients. FACS was used to determine the percentage of T (B220⁻CD3⁺) cells in the peripheral blood of each recipient before (0 weeks) and 4 and 6 weeks after injection of donor cells. Serum IgG isotype concentrations were ascertained before (0 weeks) and 3, 5, and 7 weeks after injection of donor cells. The horizontal bar indicates the mean value for the points shown in each graph.

Fig. 4.

Adoptive transfer of JH^−/− bone marrow and thymus cells into 3H9Vκ8 DNA-PKcs^−/− recipients results in production of serum IgG isotypes. 3H9Vκ8 SCID recipients served as a positive control. Thymocytes (2 × 10⁶) and bone marrow cells (3 × 10⁶) from C.B-17 JH^−/− mice were admixed and injected i.v. into 8- to 10-week-old transgenic recipients. The concentration of sera IgG1, IgG2a, and IgG2b in each recipient was ascertained before (0 weeks) and 2, 4, and 6 weeks after injection of donor cells. The horizontal bar indicates the mean value for the points shown in each graph.

In Vivo Induction of CSR in B Cells of 3H9Vκ8 DNA-PKcs^−/− Mice.

In our earlier work (11), we showed that 3H9Vκ8 SCID^−/− mice could give rise to IgG- and IgA-producing cells when provided with a source of exogenous T cells from JH^−/− mice. The use of donor mice genetically unable to generate B cells ensured that engrafted recipients were reconstituted with lymphocytes of the T lineage only. Using the same biologically relevant system, we asked whether T cell engraftment of 3H9Vκ8 DNA-PKcs^−/− mice would also result in production of IgG and IgA. Fig. 3 shows the extent of T cell repopulation and concentration of IgG1, IgG2a, IgG2b, and IgA in 3H9Vκ8 DNA-PKcs^−/− and 3H9Vκ8 RAG1^−/− recipients as a function of time after the adoptive transfer of thymocytes and bone marrow cells from JH^−/− donor mice. The RAG proteins are not required for CSR (29); therefore, 3H9Vκ8 RAG1^−/− mice served as the positive control. As indicated in Fig. 3, T cells were detectable in the peripheral blood at 4 weeks after cell transfer. At 6 weeks, JH^−/− donor T cells represented ≈25% and 40% of the peripheral blood lymphocytes in 3H9Vκ8 RAG1^−/− and 3H9Vκ8 DNA-PKcs^−/− recipients, respectively. In parallel with T cell reconstitution, the concentrations of sera IgG1, IgG2a, IgG2b, and IgA increased dramatically and reached comparable levels in both groups of recipients, except in the case of IgG2a, which in most 3H9Vκ8 DNA-PKcs^−/− recipients was 3- to 4-fold less than the IgG2a levels in the 3H9Vκ8 RAG1^−/− controls.

In another cell transfer similar to the one described above, we used 3H9Vκ8 SCID mice as the positive control because B cells of these mice were shown in a previous extensive analysis (11) to class-switch to different Ig isotypes with close to the same efficiency as B cells of transgenic WT mice (3H9Vκ8 SCID/+ mice). As shown in Fig. 4, serum IgG1, IgG2a, and IgG2b concentrations were found to increase with the same time course in 3H9Vκ8 DNA-PKcs^−/− recipients as in 3H9Vκ8 SCID control recipients, reaching two to three orders of magnitude over background (<10 μg/ml) 6 weeks after cell transfer. We conclude from the results of Figs. 3 and 4 that T cell-dependent class-switching to different isotypes by B cells in 3H9Vκ8 DNA-PKcs^−/− mice can occur with little or no impairment.

In Vitro Induction of CSR in B Cells of 3H9Vκ8 DNA-PKcs^−/− Mice.

Molecular evidence that class-switching to different Ig isotypes can occur independently of DNA-PKcs is shown in Fig. 5. Splenic B cells from 3H9Vκ8 DNA-PKcs^−/− and 3H9Vκ8 SCID control mice were stimulated for 2 days with LPS alone or with LPS/TGF-β to induce recombination between the μ region and the γ3, γ2b, or α switch regions. Using RT-PCR as previously described (11, 30), we found the relative abundance of postswitch transcripts resulting from recombination between the above switch regions to be similar in stimulated cells from 3H9Vκ8 DNA-PKcs^−/− and 3H9Vκ8 SCID mice. In previous work we showed the abundance of postswitch transcripts in mitogen/cytokine-stimulated 3H9Vκ8 SCID splenic cells to be within 2-fold of that in control 3H9Vκ8 SCID/+ splenic cell cultures (11). We conclude that B cells of 3H9Vκ8 DNA-PKcs^−/− mice can be induced in vitro to undergo CSR to different Ig isotypes with close to the same efficiency as B cells of 3H9Vκ8 SCID/+ mice.

Fig. 5.

Induction of CSR in 3H9Vκ8 DNA-PKcs^−/− and 3H9Vκ8 SCID splenic B cells. Iμ-Cγ2b, Iμ-Cγ3, and Iμ-Cα postswitch transcripts were PCR-amplified from 3-fold serial dilutions of cDNA from splenic cells that were stimulated for 2 days with LPS or LPS/TGF-β. β-2 microglobulin (β2M) served as an internal control for input cDNA. Transcripts were detected by Southern blot analysis of gel-separated PCR products.

Discussion

Is DNA-PKcs Essential for CSR?

In contrast to our results, others have reported Ig class-switching to be strongly dependent on DNA-PKcs (7, 8). Rolink et al. (7) found class-switching to IgE to be severely reduced (at least 100- to 250-fold) in SCID B lineage cell lines compared with RAG2^−/− control cell lines. However, because class-switching normally occurs in mature B cells with a productive VDJH rearrangement, the results for IgE switching in immature B lineage cell lines lacking a VDJH rearrangement (7) may not apply (equally) to mature B cells. Indeed, T cell- and cytokine-induced class-switching to IgE has been shown to occur in B cells of H/L-chain transgenic mice (11, 12) and was found to be only moderately reduced (≈3-fold) relative to that in H/L-chain, RAG1^−/− transgenic control B cells (12).

In H/L-chain transgenic mice totally lacking DNA-PKcs (129/C57BL6 DNA-PKcs^−/− mice), Manis et al. (8) were able to detect class-switching to IgG1 but not to other isotypes. Sera and supernatants of appropriately stimulated splenic B cell cultures from these mice were found to contain low levels of IgM and IgG1 compared with the transgenic WT controls but lacked detectable IgG2a, IgG2b, IgG3, and IgA (and IgE as well in the B cell cultures). Because the secreted IgM and IgG1 levels were both proportionately lower in transgenic DNA-PKcs^−/− mice and B cell cultures than in the transgenic WT controls, the authors concluded that class-switching to IgG1 was relatively unimpaired in the absence of DNA-PKcs. Furthermore, the results were interpreted to reflect a distinct DNA-PKcs-independent mechanism for class-switching to IgG1. It is difficult to understand, however, why CSR involving the γ1 switch region should differ mechanistically from that involving the γ2a, γ2b, or γ3 switch regions. All of these switch regions share 49- to 52-mer repeats (31–34) and would therefore appear near-equivalent substrates for the CSR recombinase machinery. A possible alternative interpretation of the Manis et al. (8) results is that switching to isotypes other than IgG1 may have been below the threshold of detection. Even for IgG1, the most prominent serum IgG isotype, the concentrations of IgG1 (≈5 μg/ml) in the transgenic DNA-PKcs^−/− mice were very low relative to those in the transgenic WT controls (≈100–200 μg/ml) (8). Given this alternative interpretation, one could argue that class-switching to IgG2a, IgG2b, and IgG3, for example, is as unimpaired as switching to IgG1 in B cells of H/L-chain transgenic 129/C57BL6 DNA-PKcs^−/− mice.

In contrast to the results of Manis et al. (8), we found that B cells in H/L-chain transgenic DNA-PKcs^−/− mice with a BALB/c genetic background (3H9Vκ8 DNA-PKcs^−/− mice) were present at near-normal numbers and could be readily induced to class-switch not only to IgG1 but to other Ig isotypes as well. Possible reasons for the apparent discrepancy in results may relate in part to the tgs used in each model and to genetic differences in mouse strain background. For example, the responsiveness of transgenic B cells to stimuli that induce differentiation and class-switching could depend greatly on the specificity of the tg-coded antigen receptor. In this respect, 3H9Vκ8 DNA-PKcs^−/− mice might serve as a more robust model for class switching than H/L-chain transgenic 129/C57BL6 DNA-PKcs^−/− mice. For example, we showed that B cells in 3H9Vκ8 DNA-PKcs^−/− mice could be readily activated to differentiate and undergo class-switching in the presence of T cells from JH^−/− mice. We believe this process is very efficient because the donor T cell population from JH^−/− mice would be expected to include naïve T cells that recognize host 3H9Vκ8 anti-DNA B cells and self-antigen presented by these cells as foreign (35). Indeed, the concentrations of sera IgG1, IgG2a, IgG2b, and IgA in T cell-reconstituted 3H9Vκ8 DNA-PKcs^−/− mice (shown in Figs. 3 and 4) were in most cases higher than the mean concentration of these isotypes in nontransgenic C.B-17 WT mice (11).

Another factor that may have contributed to efficient class-switching in DNA-PKcs-deficient 3H9Vκ8 mice is their BALB/c genetic background. In the course of confirming the absence of detectable DNA-PKcs in 3H9Vκ8 DNA-PKcs^−/− mice by Western analysis, we repeatedly observed the protein band corresponding to DNA-PKcs to be less intense in spleen lysates of C.B-17 and BALB/c control mice than in spleen lysates of C57BL6 mice, another positive control (see Fig. 1C). This finding leaves open the possibility that a low level of DNA-PKcs in the C.B-17 strain relative to that in mice with a C57BL6 genetic background may have resulted in a compensatory increase of another protein (or proteins) involved in NHEJ of DNA ends during CSR. Accordingly, one could argue that such compensation makes DNA-PKcs dispensable for CSR in C.B-17 mice. Although we found little or no impairment of class-switching in the B cells of 3H9Vκ8 SCID and 3H9Vκ8 DNA-PKcs^−/− mice (ref. 11 and this study), it is of interest to note that Cook et al. (12) reported that class-switching to different IgG isotypes in B cells of H/L-chain transgenic C57BL6 SCID mice bearing tgs that code for antibody to hen-egg-lysozyme was reduced ≈2- to 3-fold compared with H/L transgenic C56BL RAG1^−/− controls. Here too, both the tgs and strain background of the Cook et al. (12) transgenic model differ from that of the 3H9Vκ8 model used in this study.

Although the preceding considerations may help explain why our results differ from those of others, the inescapable conclusion from this study is that anti-self B cells lacking DNA-PKcs can be induced to undergo efficient class-switching in vivo in a biologically relevant system; i.e., in the presence of T cells and self-antigen.

DNA-PKcs-Independent NHEJ.

Whether DNA-PKcs is required for processing and joining of DNA ends in a given recombination reaction may depend in large part on the nature of the DNA substrate. For example, plasmids with noncomplementary termini have been found to join end to end with normal efficiency in DNA-PKcs deficient fibroblast and lymphoid cell lines (36). Furthermore, the joining of chromosomal DNA ends, such as blunt signal ends in V(D)J recombination, is relatively unimpaired in DNA-PKcs-deficient cells (37–40). In contrast, the joining of V, D, and J coding ends, which involves cleavage of hairpinned ends (41) by the DNA-PKcs-dependent enzyme, Artemis (42–46), has been estimated to be suppressed by >1,000-fold in DNA-PKcs-deficient cells (36). Because no obvious impairment of class-switching was found in the B cells of 3H9Vκ8 DNA-PKcs^−/− mice, we suggest that CSR in these mice could primarily involve the joining of open DNA ends, consistent with the reported detection of blunt DNA double-strand breaks in the γ3 switch region after the initiation of CSR (47). More recently, Rush et al. (48) reported DNA double-strand breaks in the μ switch region. These breaks consisted of staggered or blunt ends. However, the broken DNA species with staggered ends were much more abundant than the species with blunt ends, leaving open the possibility that some (or many) staggered ends may be made blunt before end-joining. Indeed, Schrader et al. (49) also detected DNA double-strand breaks with blunt or staggered ends in the μ (and γ3) switch regions and proposed a model for how staggered ends might be converted to blunt ends.

In summary, our results show that class-switching to different Ig isotypes can occur by a DNA-PKcs-independent mechanism. Although there is evidence for usage of the classical DNA-PKcs-dependent NHEJ pathway in CSR, our results indicate that joining of DNA ends during CSR can efficiently occur by an alternative repair pathway.

Materials and Methods

Mice.

3H9/3H9, Vκ8/Vκ8 C.B-17 SCID mice (C.B-17 is an Ig congenic partner strain of BALB/c) were selectively crossed with BALB/c mice homozygous for a disrupted DNA-PKcs allele derived from the original line of 129/C57BL6 chimeric mice described by Taccioli et al. (27). BALB/c DNA-PKcs^−/− mice were provided by G. Taccioli and were derived after six backcrosses of the disrupted DNA-PKcs allele onto the BALB/c genetic background; mice were then intercrossed to recover BALB/c DNA-PKcs^−/− mice. Ig transgenic offspring homozygous for the disrupted DNA-PKcs allele were identified as described in Results (see Fig. 1A) and selectively crossed to establish a transgenic foundation line of 3H9/3H9, Vκ8/Vκ8, DNA-PKcs^−/− mice. Segregation of the tg and WT IgH and IgL(κ) alleles was monitored by using previously described PCR assays (23, 25). The transgenic foundation line was used for crosses with BALB/c DNA-PKcs^−/− mice to obtain DNA-PKcs^−/− offspring hemizygous for both tgs (3H9/+, Vκ8/+, DNA-PKcs^−/− mice), here simply designated as 3H9Vκ8 DNA-PKcs^−/− mice.

Transgenic mice with disrupted RAG1 alleles were obtained by selectively crossing the tgs from 3H9/3H9, Vκ8/Vκ8 C.B-17 SCID mice into BALB/c RAG1^−/− mice. BALB/c RAG1^−/− mice were provided by R. Hardy (Fox Chase Cancer Center) and were obtained after six backcrosses of the disrupted RAG1 allele described by Spanopoulou et al. (50) onto the BALB/c genetic background; mice were then intercrossed to recover BALB/c RAG1^−/− mice. Segregation of the RAG1^−/− allele was monitored by using the PCR assay described by Li et al. (51). Mice from this line (3H9/3H9, Vκ8/Vκ8, RAG1^−/− mice) were crossed with RAG1^−/− mice to produce mice hemizygous for both tgs, here simply designated as 3H9Vκ8 RAG1^−/− mice. SCID mice hemizygous for the 3H9 and Vκ8 tgs were produced as previously described (11) and are referred to as 3H9Vκ8 SCID mice. R. Hardy provided C.B-17 mice with deleted JH loci (JH^−/− mice) (52). All mice were bred and maintained as specific pathogen-free mice in the Laboratory Animal Facility of the Fox Chase Cancer Center. Mice were used between 8 and 16 weeks of age according to protocols approved by the Animal Care and Use Committee at this Institution.

Western Blot Analysis.

Whole-cell lysates of 30 × 10⁶ spleen cells were obtained by using 200 μl of M-Per lysis buffer (Pierce, Rockford, IL) containing 20 μl of a proteinase inhibitor mixture (Sigma, St. Louis, MO). After lysing the cells for 30 min at 37°C, the debris was spun down for 15 min at 14,000 rpm. One-fifth of the lysate was mixed with 4× SDS loading buffer and 2-mercaptoethanol, electrophoresed under denaturing conditions on a 3–8% Tris-acetate polyacrylamide gel (Invitrogen, Carlsbad, CA), and blotted onto a poly(vinylidene difluoride) filter (Millipore, Billerica, MA) for 3 h at 90 V. The filter was overlaid with IgG1 anti-DNA-PKcs antibody (clone 18-2; Lab Vision, Fremont, CA). After incubation with anti-mouse-IgG1 antibody conjugated to HRP, filters were developed with ECL substrate (Pierce).

Cell Preparation and Flow Cytometry.

Suspensions of bone marrow cells, splenocytes, thymocytes, and peripheral blood cells were prepared as previously described (11, 35, 53). For flow cytometry, cells were stained for CD45 (B220), IgM, and CD3 using allophycocyanin-conjugated anti-B220 (clone no. RA3–6B2), FITC-conjugated anti-IgM (clone no. 331.1), and PE-conjugated anti-CD3 (clone no. 500A2) (BD Pharmingen, San Diego, CA). Analyses were performed with a FACS VantageSE flow cytometer (Becton Dickinson, Franklin Lakes, NJ) using FlowJo software.

RT-PCR for Postswitch Transcripts.

Stimulation of spleen cells with LPS or LPS/TGF-β and culture of the cells in six-well plates (0.25 × 10⁶ cells per ml) was done as described previously (11). Postswitch transcripts of the μ intronic (I) exon spliced to the 5′ Cγ2b (Iμ-Cγ2b), Cγ3 (Iμ-Cγ3), or Cα (Iμ-Cα) exons were amplified from cDNA of 2-day cultures using oligonucleotide primer pairs described by Muramatsu et al. (30) and visualized by Southern analysis of gel-separated PCR products (11). β2M transcripts were PCR-amplified as described previously (11) and served as an internal control for input cDNA.

Serological Assay.

The concentration of sera IgG1, IgG2a, IgG2b, and IgA was ascertained by ELISA as described previously (11, 35) by using monoclonal antibodies and standards from BD Pharmingen.

Abbreviations

CSR

class-switch recombination

DNA-PKcs

catalytic subunit of DNA–protein kinase

NHEJ

nonhomologous DNA end-joining

tg

transgene.