The Ig heavy chain class switch in B lymphocytes involves a unique genetic recombination that fuses specific regions within the Ig locus and deletes intervening sequences. Here we describe a novel exonuclease activity in nuclear lysates of B cells in an in vitro assay. This activity was induced in B lymphocytes after treatment with either LPSs or CD40 ligand/anti-δ-dextran, both of which induce switch recombination, and considerably less activity was detected in untreated or anti-δ-dextran-treated B cells, Con A-stimulated spleen cells, liver cells, or a number of cell lines. The exonuclease activity was dependent on divalent cations, and both 3′ and 5′ labels were efficiently removed from DNA substrates. The presence of RNase A, but not RNase H, inhibited exonucleolytic digestion, suggesting that a ribonucleoprotein is responsible for the exonucleolysis. The DNA digestion appears to be nonspecific, since DNA substrates with either switch-μ or unrelated sequence were hydrolyzed with comparable efficiency. Germ-line switch region transcripts (Igγ1, Igγ3, and Igα) strongly inhibited the exonucleolysis of switch-μ DNA but not that of unrelated control DNA, while switch antisense RNA or tRNA were much less effective inhibitors.
Stimulation of B lymphocytes with Ag normally leads to proliferation and differentiation into memory and plasma cells. During this process, some of the stimulated B cells switch from the initially expressed IgM to one of six other Ig isotypes (reviewed in Refs. 1 and 2). This switch is accomplished by a genetic recombination that fuses switch regions, characteristically rich in repetitive sequence, each of which is located several kilobases upstream of an Ig constant region (3).
Several cell culture systems have been developed to mimic the process of isotype switch in vitro (reviewed in 1 . Ags that stimulate B cells are usually divided into one of two classes: T dependent and T independent, depending on the additional requirement for the presence of T lymphocytes to activate B cells. In an analogous manner, switching can be induced in murine splenocytes in vitro by the presence of T-independent LPS. T-dependent activation can be mimicked by treatment with the T lymphocyte surface protein CD40 ligand (CD40L)3 the activity of which is potentiated by dextran-conjugated anti-IgD (αδ-dex). In addition to native mouse spleen cells, a few B cell lines have been isolated that switch their Ig isotypes in vitro (4, 5).
The mechanisms that mediate the Ig heavy chain isotype switch recombination are largely unknown. It is commonly assumed that the process is initiated by DNA strand breaks in one of the switch regions (6). In addition, sequence analysis of switch recombination joints has identified the presence of point mutations, suggesting that the process involves an error-prone DNA synthesis (7, 8, 9).
Switching is preceded by transcription of the specific switch regions that are targeted by the recombination (10, 11, 12, 13, 14, 15, 16, 17, 18, 19). We recently proposed a model in which switch recombination involves a reverse transcriptase-mediated generation of chimeric switch region DNA (20). We detected an activity in B lymphocytes that uses the free ends of an artificial switch-μ (Sμ) DNA substrate to prime the reverse transcription of non-Sμ switch region RNAs. In this model, switch recombination would be accomplished by the homologous recombination of the newly generated cDNA with the respective switch region. Sequence analysis of in vitro-generated chimeric DNA, however, indicated that Sμ DNA is processed before its extension by reverse transcription. Both Sμ-specific endonucleases and an exonuclease could be responsible for the generation of the free DNA ends in Sμ (20).
Several genetic recombinations are known to be mediated in part by an exonuclease. Most notably, homologous recombination in Escherichia coli involves the products of genes recB, recC, and recD (reviewed in 21 . The RecD subunit of this protein complex is responsible for exonucleolytic digestion of target DNA.
Kenter and Tredup (22) detected a 3′→5′ exonuclease activity that was specific for B lymphocytes, but its activity was independent of B cell activation. Here we describe a novel B lymphocyte-specific exonuclease that is induced in activated B cells. We also show that different switch region germ-line transcripts strongly inhibit this enzymatic activity.
Materials and Methods
Spleen cell suspensions were prepared from 5- to 12-wk-old BALB/cAn mice. Erythrocytes were lysed in 0.15 M NH4Cl, 1 mM KHCO3, and 0.1 mM Na2EDTA (pH 7.2), and B lymphocytes were enriched using anti-B220 mAb (Miltenyi Biotec, Bergisch Gladbach, Germany) and magnet-activated cell sorting. Cells were cultured at 3 × 105 cells/ml in RPMI 1640 supplemented with 2 mM glutamine, 50 μM 2-ME, penicillin, streptomycin, and 10% FCS. Two methods of activation were used. 1) Total spleen cell suspensions were activated by adding 50 μg/ml LPS (Sigma Chemical Co., St. Louis, MO). 2) Sorting-enriched splenic B cells were presented with CD40L, as the membrane fraction of baculovirus-infected Sf21 cells (provided by Dr. Marilyn Kehry, Boehringer Ingelheim, Ridgebury, CT) (23), together with 3 ng/ml anti-δ-dex (provided by Dr. Clifford Snapper, U.S. Public Health Service, Bethesda, MD). LPS-activated cells were lysed for analysis after 48 and 96 h, and CD40L-treated cells were lysed after 72 h. Routine FACScan analysis of surface Ig expression (IgG1, IgG3) was performed after 72 and 120 h, respectively. Nuclear extracts were also prepared from freshly isolated B lymphocytes, αδ-dex-treated B cells, Con A-stimulated spleen cells, liver cells, the pre-B lymphoma line 70Z/3 (24), the mature B cell lines CH12.LX (25) and I.29 (26), the T lymphoma EL4 (27), monocytic leukemia cells FDJ2 (28), and the fibroblast line NIH 3T3 (29).
Nuclear extracts were prepared as described previously (20). Briefly, we washed the cells three times in PBS, and 2 × 107 cells were resuspended in 400 μl of 10 mM HEPES (pH 7.9), 10 mM KCl, 0.1 mM EDTA, and 0.1 mM EGTA and incubated on ice for 15 min. After adding Triton X-100 (final concentration, 0.6%), cells were vortexed (10 s) and centrifuged twice (500 × g for 3 min). The pellet was resuspended in 50 μl of 20 mM HEPES (pH 7.9), 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, and 1 mM DTT supplemented with the Protease Inhibitors Set (Boehringer Mannheim, Indianapolis, IN), gently rocked for 15 min at 4°C, and centrifuged (10,000 × g, 5 min). The resulting supernatants were adjusted to a protein concentration of 2 μg/μl, aliquoted, and stored at −70°C.
DNA substrates and in vitro transcription
Two Sμ substrates were used to determine the presence of exonucleases. Generation of the Sμ minisubstrate (Mμ) has been described (20). It consists of an 169-bp fragment with 42 bp of Sμ sequence that is flanked by unrelated DNA (113 bp upstream and 14 bp downstream). We also used a 692-bp Sμ substrate (MD3) that we generated by PCR amplification, using murine spleen DNA as template followed by cloning into pCR2.1 (Invitrogen, Carlsbad, CA). This fragment (MUSIGCD07 5368-5541/MUSIGCD09 708-1225) consists mainly of repetitive Sμ pentamers, and it includes a large deletion of repetitive switch sequence. In addition to the Igμ substrates we used a control DNA (LTR) that did not show sequence characteristics of a switch region (GenBank MUSFLIAP 7-190).
The inserts Mμ and LTR were released from the pCR2.1 plasmid by HindIII/XbaI digestion, and MD3 was released by EcoRI digestion. The inserts were isolated by electrophoresis on low melting point agarose gels. The 3′ ends were radiolabeled by filling in 5′ overhangs (Mμ and LTR) with Klenow polymerase using [α-32P]dCTP and cold dNTPs. To generate 5′ end-labeled DNA, both overhangs of Mμ were filled in with cold dNTPs and then phosphorylated with [γ-32P]ATP and T4 kinase on both ends. Unincorporated nucleotides were removed by passage through Sephadex G-50 columns (5 Prime-3 Prime, Boulder, CO). The digestion of unlabeled fragment MD3 was detected by Southern blotting using [α-32P]dCTP-labeled MD3 (labeled by nick translation) as a hybridization probe.
To generate switch region RNAs, portions of several Ig switch regions (Sγ1, GenBank entry MUSIGHANB 3664-4983; Sγ3, MUSIGHANA 54-2561; Sα, MUSIALPHA 1861-4693) were PCR amplified and cloned into pCR2.1. Clones were sequenced to identify those with inserts that were in the appropriate orientation for use of the plasmid T7 promoter site. RNAs corresponding to the sense orientation of downstream coding regions were generated with T7 RNA polymerase (Promega, Madison, WI) using 9 μg of linearized plasmid DNA as template. After polymerization (4 h, 37°C), the plasmid DNA was destroyed by DNase digestion (Promega). Unincorporated nucleotides were removed by passage through Sephadex G-50 columns (5 Prime-3 Prime, Boulder, CO), and proteins were removed by phenol/chloroform extraction and isopropanol precipitation. The concentrations of in vitro-generated RNAs were estimated on ethidium bromide-stained agarose gels. As a negative control, we also generated antisense Sα RNA by this same procedure.
To test for the presence of exonuclease activity in nuclear lysates, 32P-labeled DNA substrate (105 cpm, ∼20 ng) was mixed with 4 μg of nuclear lysates in 10 mM Tris (pH 8.3), 50 mM KCl, 4 mM MgCl2 (unless indicated otherwise), and 1 mM DTT in a total volume of 50 μl. If indicated, we preincubated the reaction mixture for 10 min (37°C) with 40 μg of proteinase K, 40 μg of RNase A, 8 U of RNase H, 2 μg of unrelated DNA (EcoRI-digested pCR2.1), or 4 μg of RNA (in vitro generated switch region RNAs or yeast tRNA) before adding the DNA substrate. The mixture was incubated at 37°C for 40 min unless indicated otherwise. We terminated the reaction by phenol/chloroform extraction. The reaction mix was precipitated with ethanol in the presence of 40 μg of glycogen, analyzed by electrophoresis on 8% polyacrylamide/8 M urea gels, and exposed to x-ray film. The amount of undigested substrate was determined with a phosphorimager.
In an alternative approach, approximately 100 ng of unlabeled Sμ DNA (MD3) was exposed to nuclear lysates as described above. After precipitation and rehydration, the reaction was size fractionated by electrophoresis on 1% agarose gels followed by blotting onto nylon membranes (Hybond-N, Amersham, Arlington Heights, IL). The membranes were hybridized in Hybrisol (Oncor, Gaithersburg, MD) with a probe that we generated by nick translating MD3 with [α-32P]dCTP, washed, and exposed to x-ray film.
To detect exonuclease activity, nuclear lysates from activated B lymphocytes were incubated with radiolabeled Sμ DNA, and the reaction products were analyzed by electrophoresis on 8% polyacrylamide/8 M urea gels. Figure 1,A shows that treatment with nuclear extracts led to the appearance of a second band a few base pairs shorter than the 3′ end-labeled substrate. Both DNAs were subsequently digested to near completion within 60 min of exposure to nuclear extracts from LPS-activated total spleen cells. No products shorter than the substrate were observed if a longer 692-bp Sμ substrate was used (Fig. 1 B), indicating the processive character of the exonuclease. Only dsDNA appeared to be digested, because denaturing the DNA substrate by heating (95°C, 5 min) before exposure to nuclear lysates prevented its digestion (not shown).
Many nucleases require the presence of divalent cations for their full activity. As shown in Figure 1 C, the exonuclease activity was optimal in the presence of 4 mM MgCl2 (or 4 mM MnCl2, not shown). No activity was detected below 0.6 mM magnesium/manganese.
To determine the tissue and cell type specificities of this exonuclease activity, equal amounts of various nuclear preparations were incubated with the DNA substrate (Fig. 2). Treatment of spleen cells with LPS for 4 days or activation of B lymphocytes with CD40L induced an exonuclease activity that digested the DNA substrate to near completion. In contrast, nuclear extracts from unstimulated or αδ-dex-treated splenic B lymphocytes, Con A-stimulated spleen cells, liver, or a number of cell lines digested only between 6 and 58% of the DNA substrate. This corresponds to equal or less enzymatic activity than we detected in 1/4 dilutions of CD40L/αδ-dex-activated B cells (Fig. 2).
Several substances inhibited the exonucleolytic digestion of the Sμ substrate (Fig. 3). Preincubation of nuclear lysates with proteinase K abrogated exonuclease activity, and RNase A partially inhibited DNA digestion (Fig. 3, A and B). If RNA (either switch RNA or tRNA) was added before RNase A to the reaction, exonuclease activity could be partially restored (Fig. 3,B). Adding RNA after the nuclear lysates were preincubated with RNase A did not affect the inhibition of the exonuclease reaction by RNase A (Fig. 3 C).
Substantial inhibition of exonuclease activity was also observed if the reaction mixture contained switch region sense RNAs (Sγ1, Sγ3, or Sα) that were generated by in vitro transcription (Fig. 3, A and B). Between 60 and 75% of the DNA substrate was left unhydrolyzed if either of the switch region transcripts were present (compared with about 15% in the absence of inhibitors). In contrast, Sα antisense RNA produced a much smaller inhibitory effect (on the average, 22% undigested substrate; Fig. 3, A and B). RNA that was entirely homologous to the template strand of the DNA substrate was less inhibitory than switch sense RNAs (∼25%; lane 4 in Fig. 3,B), indicating that the inhibition cannot entirely be explained by a homologous interaction of RNA with the DNA substrate. This is further supported by the finding that the presence of RNase H, which digests RNA that is annealed to DNA, did not affect the exonucleolysis inhibition by switch RNAs (Fig. 3,B). The presence of tRNA had almost no effect on the digestion of the Sμ substrate (Fig. 3, A and B).
The exonuclease activity was not specific for Sμ sequences, since an unrelated DNA substrate was efficiently digested (Fig. 3,B, lanes 10–13). This digestion could be inhibited by the presence of proteinase K or RNase A. However, in contrast to the exonucleolysis of Sμ DNA, the presence of switch region RNAs did not affect the digestion of the unrelated DNA (Fig. 3 B).
We next investigated whether 5′ labels would be removed from DNA substrates (Fig. 3 D). Nuclear extracts from activated B cells efficiently used 5′ labeled substrates for digestion. The disappearance of the labeled DNA could be theoretically caused by a phosphatase activity. However, as with the 3′ labeled DNA substrates, the presence of switch sense RNAs inhibited the activity.
The mechanisms that mediate switch recombination in B lymphocytes remain largely unknown (reviewed in Refs. 1 and 2). Switching occurs in regions that are transcribed before the recombination event (10, 11, 12, 13, 14, 15, 16, 17, 18, 19). The relevance of these transcripts is unclear, since they usually lack significant open reading frames. We recently postulated that these RNAs are used as templates for the synthesis of chimeric switch DNAs by an Sμ-primed reverse transcription (20). Here we show a second novel activity of primary germ-line switch region transcripts: Sμ DNA-specific inhibition of an LPS- and CD40L/αδ-dex-inducible DNA exonuclease activity. This inhibitory effect was observed with in vitro-generated Sγ1, Sγ3, and Sα sense RNAs, which differ considerably in primary structure. The exonuclease was much less affected by Sα antisense RNA or by an RNA that corresponded to the nontemplate strand of the Sμ DNA substrate, indicating that the inhibition was not the sole consequence of homologous interactions between homologous DNA and RNA. Digestion with RNase A and abrogation of RNase A activity by the inclusion of RNA indicates that the exonuclease activity depends upon an RNA moiety. The exonuclease activity is also dependent on the presence of magnesium or manganese ions.
Kenter and Tredup (22) described an exonuclease in B lymphocytes. Several findings indicate that this enzyme activity and the one described here are different. 1) Their expression patterns differ, in that the previously described exonuclease was in high quantity in resting B cells, whereas the enzyme described here was induced in activated B lymphocytes. 2) The two enzymes have substantially different magnesium optima (0.1 vs 4 mM). 3) Kenter and Tredup detected only products caused by incomplete DNA digestion. We found some digestion products a few base pairs shorter than the substrate, particularly if switch RNAs were present (Fig. 3). This could have been caused by the enzyme described previously (22). However, DNA was subsequently hydrolyzed to near completion even if a longer Sμ substrate (692 bp) was used, indicating that the newly described exonuclease is processive. 4) In contrast to the previously described 3′->5′ exonuclease, the activity described here probably removes both 3′ and 5′ labels, since both activities can be inhibited by switch region RNAs.
A well-described experimental system in which recombination occurs only with exonuclease activity is the RecBCD enzyme in Escherichia coli (reviewed in 21 . RecBCD represents a powerful exonuclease, the activity of which is inhibited upon encountering the recombination hot spot sequence χ. This sequence converts the enzyme complex into a recombinase, possibly by ejecting RecD (21). A lack of χ sequence prevents recombination. We suggest that a similar enzymatic activity is involved in switch recombination (Fig. 4). The process is initiated by a strand break in a switch region. Exonuclease activity digests switch DNA unless germ-line switch RNA is present. In the absence of the RNA, the switch region DNA will be digested and will not be accessible for recombination. The presence of the RNA stops the digestion, which could be the equivalent of RecBCD encountering a χ sequence. In contrast to homologous recombination in Escherichia coli, the recombining Ig switch loci are not homologous. However, we recently demonstrated that nuclear extracts of switch competent B cells possess a reverse transcriptase-like activity that is capable of generating a chimeric switch DNA intermediate (20). This chimeric intermediate could then be used for homologous recombination with the endogenous switch locus, generating the switch recombination products. Identification of the inducible exonuclease activity in this report further supports our working model for switch recombination (20).
We thank Dr. J. Fred Mushinski for many helpful discussions throughout these studies. J.R.M. gratefully acknowledges Dr. M. Potter for his support during the course of this work.
This work was supported in part by U.S. Public Health Service Grant GM26939 (to K.B.M.).
Abbreviations used in this paper: CD40L, CD40 ligand; αδ-dex, dextran-conjugated anti-IgD; Sμ, switch-μ; LTR, long terminal repeat.