Differential requirements for cis and trans V(D)J cleavage: effects of substrate length

Abstract

The assembly of productive synaptic complexes is a critical, but poorly understood, regulatory step in V(D)J recombination. Several lines of evidence suggest that there may be important differences between recombination involving sites situated in cis (on the same DNA molecule) and in trans (on separate molecules). Because biochemical experiments using both purified RAG proteins and crude extracts have failed to detect trans cleavage of plasmid substrates it has been thought that there is a substantial bias against trans synapsis. In conflict with these results are more recent studies showing that purified RAG proteins can catalyze trans cleavage of short oligonucleotide substrates. Furthermore, recent experiments have detected efficient trans cleavage of plasmid substrates in vivo. We sought to investigate why these different systems yield such divergent results. We found that, unexpectedly, the ability of both purified RAG proteins and crude extracts to cleave DNA substrates in trans is a function of substrate length. Our data raise two critical issues: first, oligonucleotides, which are the most commonly used substrates to study V(D)J recombination in vitro, do not mimic the behavior of plasmid substrates; second, in the trans cleavage reaction current purified RAG systems do not accurately reflect the in vivo situation. We propose a unifying model to explain the effects of substrate length and coniguration (cis or trans) on the efficiency of synapsis.

INTRODUCTION

The ability of the vertebrate immune system to recognize a wide variety of antigens depends upon the generation of a diverse repertoire of antigen receptors. This diversity is produced by V(D)J recombination, a site-specific DNA rearrangement process in which particular DNA segments are recognized by the recombinase, assembled into a synaptic complex, cleaved and rejoined in a new configuration. The recombination machinery recognizes DNA sequences, called recombination signal sequences (RSS), which consist of conserved heptamer and nonamer motifs separated by 12 or 23 nt spacer sequences. RSS are located adjacent to each of the variable (V), diversity (D) and joining (J) coding elements and serve to target these gene segments for recombination during lymphocyte differentiation. The recombinase generates a double-strand break (DSB) between the RSS and the coding element, resulting in two types of DNA ends: covalently sealed (hairpin) coding ends and blunt, 5′-phosphorylated signal ends (1–3). These broken DNA ends are subsequently joined with the help of DNA repair factors to form coding joints and signal joints (reviewed in 4).

The cleavage step of V(D)J recombination is carried out by the RAG-1 and RAG-2 proteins (5). Efficient DSB formation requires a pair of RSSs, one of each spacer length (the 12/23 rule) (6–8), which are brought together prior to cleavage to form a synaptic complex (9). While it is clear that synaptic complex formation plays a critical regulatory role in V(D)J recombination, the steps leading to formation of this complex remain obscure. An early event involves binding of a RAG-1/RAG-2 complex to an RSS however, it is not clear how the RAG proteins locate their target sequences in the context of a large excess of genomic DNA. Furthermore, it is not known whether the next step in synaptic complex formation involves recruiting a second RAG/RSS DNA–protein complex or whether the second RSS is captured as naked DNA by the nascent complex. Given the central role of synaptic complex formation in the regulation of V(D)J recombination, it is critical to elucidate the pathways responsible for its proper assembly.

In vivo, and under appropriate conditions in vitro, most cleavage occurs in a coupled fashion at a 12/23 RSS pair (7,8,10–12), as predicted by the 12/23 rule. After cleavage the broken ends are thought to remain associated with the RAG proteins as a post-cleavage complex (9,13); joining may take place in the context of this complex (14). Although cleavage is performed by the RAG proteins, the non-specific DNA-bending proteins, HMG-1 or HMG-2, serve as cofactors, stimulating cleavage of the 23-RSS by purified RAG proteins and enhancing coupled cleavage in cis (12,15).

Productive V(D)J recombination events take place between partner RSS located on the same DNA molecule (recombination in cis). Less commonly, V(D)J rearrangements involve RSS residing on distinct DNA molecules (joining in trans). Given the potential dangers associated with trans V(D)J recombination events (including oncogenic chromosome translocations), it is not surprising that trans V(D)J rearrangements (i.e. formation of completed recombination products) are relatively rare (16–18). The mechanism responsible for suppressing trans rearrangements was previously believed to operate at the synapsis step, because trans cleavage of linear DNA substrates had not been observed in vitro in cell-free systems either with purified RAG proteins or with crude extracts (7,8). More recent work, however, has shown that purified RAG proteins can catalyze trans cleavage of oligonucleotide substrates (9). Furthermore, we found that trans cleavage of plasmid substrates in vivo proceeds efficiently both in fibroblasts and in lymphoid cells (18), an observation that was recently confirmed by others (19). Thus, there appears to be no substantial barrier to trans synapsis in vivo. These seemingly contradictory observations raise several interesting questions: (i) are cis and trans cleavage distinguished by differences in substrate, cofactor or reaction condition requirements?; (ii) why don’t in vitro studies using non-oligonucleotide substrates recapitulate the high levels of trans cleavage seen in vivo? (iii) Are there important differences between the way the V(D)J recombinase interacts with short oligonucleotides and longer, more physiologically relevant substrates?

To address these questions, we systematically examined cis and trans cleavage by purified RAG proteins using a variety of substrates and conditions. We show that cis and trans cleavage indeed have different requirements. Short oligonucleotide substrates are cleaved in trans, but larger DNA fragments, assayed under the very same conditions, are not. Our finding that trans cleavage is specifically sensitive to the length of the DNA substrate explains the inability of previous studies to detect trans cleavage of plasmid substrates in vitro and suggests an additional constraint that must be incorporated into models describing the assembly of synaptic complexes. Moreover, our results support a new model that explains the specific effects of substrate length on trans, but not cis, synapsis by purified RAG proteins.

Because previous work failed to detect cleavage of short DNA fragments in crude extracts containing RAG proteins (7), we also investigated cis and trans cleavage in a crude extract system. Like the purified system, crude extracts support trans cleavage of oligonucleotides, but not of longer (700–1000 bp) DNA molecules, even though these substrates were cleaved efficiently in cis. We found that trans cleavage was restored when the substrate length exceeded 1000 bp. This is the first in vitro system in which both cis and trans cleavage of plasmid substrates occurs efficiently, as in vivo. We discuss models to explain the biphasic dependence on DNA length observed in this system.

MATERIALS AND METHODS

Purified proteins

Polyhistidine-tagged core RAG-1 and RAG-2 proteins (R1 and R2) were prepared as previously described (3,5,12,20,21). HMG-1 was isolated from cultures of BL21(DE3)pLysS (Stratagene) containing pDVG83, which encodes a truncated form of HMG-1 (missing the acidic tail).

Crude extracts

Chinese hamster ovary fibroblasts (RMP41) were transfected with 21 µg of each truncated RAG-1 (pMS127) and truncated RAG-2 (pMS216) expression vector (22,23) per T150 flask using the FuGene 6 transfection reagent (Boehringer Mannheim). The transfection efficiency was 85–90%, as determined by assaying for β-galactosidase activity. After 48 h cells were harvested and the cell pellets were subjected to three cycles of freezing and thawing. The cells were then extracted in buffer (10⁷ cells/ml) containing 25 mM K-HEPES, pH 7.0, 260 mM KCl, 40 mM NaCl, 0.1% NP-40, 20% glycerol, 1 mM DTT and 0.5 mM PMSF for 2 h at 4°C (24). Insoluble material was removed by centrifugation at 25 000 g for 25 min at 4°C. Aliquots of the supernatant were frozen in liquid nitrogen and stored at –80°C. Extracts typically contained 1–3 mg/ml total protein and ∼20 ng/µl RAG-1 and RAG-2, as determined by western blotting and Coomassie staining.

DNA substrates

Recombination substrates p12/23 (pJH290), p12 (pGL1) and p23 (pGL2) have been described previously (6,18,25). Nicked p12/23, p12 and p23 were made by incubating 2 µg plasmid DNA with 366 µg/ml ethidium bromide and 2 µg/ml DNase I in a buffer containing 20 mM Tris–HCl, pH 7.5, 50 mM NaCl and 10 mM MgCl₂ for 30 min at 30°C. Relaxed p12/23, p12 and p23 were made by incubating 2 µg plasmid DNA with 10 U calf thymus topoisomerase I (Gibco BRL) in a buffer containing 10 mM Tris–HCl, pH 7.9, 50 mM KCl, 0.1 mM EDTA and 5 mM MgCl₂ for 1 h at 37°C. Eight kilobase linear substrates were made by digesting p12/23, p12 or p23 with NsiI; 700 bp linear fragment substrates (fp12/23, fp12 and fp23) were made by digesting p12/23, p12 or p23 with PvuII followed by gel purification of the fragment containing the RSS; 3, 2 and 1 kb substrates were made by PCR using 10 ng p12 or p23 as template. The labeled 140 bp 12-RSS substrate was made by radioactively labeling the 5′-end of one primer and amplifying the product using the labeled primer, an unlabeled primer and 10 ng p12 as template. Unlabeled 140 bp 12-RSS and 23-RSS were also made by PCR. PCR substrates were purified using the Qiaquick PCR purification kit (Qiagen). PAB4-12 was constructed by inserting a 12-RSS oligonucleotide (annealing of the complementary 37 nt oligonucleotides results in a 12-RSS flanked by two AflII compatible ends) into the AflII site of pAB4. Stretches of sequence identity between pAB4 and p23 are <10 bp, as determined using SeqWeb v.1.1. Equimolar amounts of pAB4 or pAB4-12 and p23 were used in reactions. The 12-RSS and 23-RSS oligonucleotide substrates corresponding to the sequences found in pJH290 were formed by annealing SK5 (5′-TTGGGCTGCAGGTCGACACAGTGCTACAGACTGGAACAAAAACCCTGCAG-3′) with its complement SK6 and SK39 (5′-ATCGATGAGAGGATCCCACAGTGGTAGTACTCCACTGTCTGGCTGTACAAAAACCCTCGG-3′) with its complement SK40, respectively. The 12-RSS (DAR39 and DAR40), the 23-RSS (DG61 and DG62) and the non-specific substrate (DAR81 and DAR82) have been previously described (5,26).

Oligonucleotide cleavage assays

Oligonucleotide cleavage assays (10 µl) using purified RAG proteins were performed as previously described (9). An aliquot of 0.02 pmol of ³²P-labeled 12-RSS (SK5/SK6) oligonucleotide substrate and an unlabeled 23-RSS (SK39/SK40) or non-specific (DAR81/DAR82) oligonucleotide substrate (equimolar or 10-fold excess) were preincubated with purified RAG proteins (∼100 ng) in a buffer containing 25 mM MOPS, pH 7.0, 2 mM DTT, 100 µg/ml BSA, 19 mM KOAc, 20 ng/µl HMG and 5 mM CaCl₂ for 10 min at 37°C. MgCl₂ was added to a final concentration of 5 mM and the reactions were incubated at 37°C for 1 h. Reactions were stopped by addition of an equal volume of formamide loading dye and the reaction products separated on a 12% denaturing polyacrylamide gel. Cleavage assays using the radioactively labeled 140 bp PCR substrate were also performed using these conditions. Oligonucleotide cleavage assays using crude extracts were performed as described for purified RAG proteins with some modifications. The reactions were performed using 0.02 pmol of ³²P-labeled 23-RSS (DG61/DG62) oligonucleotide substrate with unlabeled 12-RSS (DAR39/DAR40) or non-specific oligonucleotide substrate in a buffer containing 50 mM HEPES, pH 8.0, 30 mM K-glutamate, 4 mM KCl, 1 mM DTT and 5 mM CaCl₂. MgCl₂ was added to a final concentration of 5 mM following the preincubation step.

Plasmid cleavage assays

Plasmid cleavage assays with purified RAG proteins were performed in 10 µl and contained 20 fmol of each recombination substrate, 25 mM MOPS, pH 7.0, 2 mM DTT, 100 µg/ml BSA, 19 mM KOAc, 11 mM K-glutamate, 5 mM CaCl₂, 20 ng/µl HMG-1 and RAG proteins (∼100 ng each). Reactions were preincubated at 37°C for 15 min, at which time MgCl₂ was added to a final concentration of 5 mM and the reactions were incubated for an additional 2 h at 37°C. Reactions were stopped by addition of 90 µl of stop buffer (100 mM Tris–HCl, pH 8.0, 0.2% SDS, 0.25 mg/ml proteinase K and 10 mM EDTA) and incubated for at least 1 h at 55°C. The reactions were phenol/chloroform extracted, ethanol precipitated, resuspended in TE and digested with 1 U PvuII for 30 min at 37°C. The digested reaction products were separated on a 4.5% polyacrylamide gel and electrophoretically transferred to membranes. Blots were hybridized with a random primed, ³²P-labeled PvuII fragment from pJH290 as described previously (11). Similar results were obtained when the cleavage reactions were performed in a buffer containing 50 mM HEPES, pH 8.0, 26 mM KCl, 4 mM NaCl, 1 mM DTT, 100 µg/ml BSA and 5 mM CaCl₂ and when the preincubation step was omitted.

The standard crude extract cleavage reaction (10 µl) contained 4 fmol of each recombination substrate, 1 µl of RAG extract, 50 mM K-HEPES, pH 8.0, 5 mM MgCl₂, 1 mM ATP, 1 mM DTT and KCl and/or K-glutamate to give a final concentration of 45 mM potassium ions, including the potassium ions in the extract. Untransfected extract was substituted for RAG extract in the no-RAG samples. Cleavage reactions were incubated for 3 h at 30°C. Reactions were stopped by addition of 90 µl of stop buffer and treated as described above. Cleavage efficiencies were determined using a Molecular Dynamics PhosphorImager system.

RESULTS

Trans cleavage by purified RAG proteins depends on substrate length

We first examined cleavage of oligonucleotide substrates by polyhistidine-tagged RAG proteins purified from baculovirus-infected insect cells. Cleavage in trans (defined here as cleavage that is stimulated by the presence of a 12/23-RSS pair) was observed using 20 fmol each of a 12- and a 23-RSS-containing oligonucleotide (Fig. 1, lane 2). Cleavage was substantially enhanced when a 10-fold molar excess of the unlabeled 23-RSS oligonucleotide was added (lane 6), in agreement with previous observations (27).

Figure 1.

Trans cleavage of oligonucleotide substrates by purified RAG proteins. Double-stranded oligonucleotide substrates containing the coding flank and 12-RSS sequences found in the pJH290 recombination substrate were radioactively labeled on the 5′-end of the top strand and incubated with purified RAG proteins, HMG-1 and an equimolar amount of the indicated unlabeled oligonucleotide substrates (lanes 2–5) or with 10-fold excess of unlabeled oligonucleotide substrates (lanes 6–9) (see Materials and Methods). –R lanes (lanes 5 and 9) contain a labeled 12-RSS and an unlabeled 23-RSS incubated without purified RAG proteins. Products were separated on a 12% denaturing polyacrylamide gel. Diagrams on the left indicate the positions of the uncleaved substrate, the hairpin product and the nicked product. The asterisk indicates a radioactive label.

The behavior of plasmid substrates was quite different. Cis cleavage was measured using 20 fmol of a standard substrate containing a 12/23-RSS pair (p12/23); trans cleavage was assayed using an equimolar mixture (20 fmol each) of two plasmids that each contain only a single RSS, p12 or p23 (Fig. 2A). Cleavage of p12/23 in cis was efficient (20–30%, in several experiments) and single cleavage of p12/23 and p23 was observed (Fig. 2B, lanes 1 and 4). In multiple experiments cleavage of p23 was somewhat more efficient than cleavage of p12, in agreement with previous observations (19) (in Fig. 2B cleavage of p12 was not observed at the exposure shown; inefficient cleavage of this substrate was detected in other experiments). Coupled cleavage in trans, however, was not observed, as an equimolar mixture of p12 and p23 gave no more efficient cleavage than the two substrates alone (lanes 3–5). Similar results were obtained using lower substrate concentrations (4 fmol/reaction), in several different buffer conditions and using MBP- and GST-tagged RAG proteins purified from insect and mammalian cells, respectively (data not shown). Furthermore, addition of a 10-fold excess of a 12-RSS-containing substrate does not stimulate cleavage at the 23-RSS (data not shown). Thus, in contrast to the situation with oligonucleotides, there appears to be a substantial barrier to trans cleavage of plasmid substrates by purified RAG proteins.

Figure 2.

Cleavage of plasmid and non-oligonucleotide substrates by purified RAG proteins. (A) Diagram of plasmid substrates. Triangles represent the 12-RSS (open) and 23-RSS (filled). Open and filled squares symbolize the coding flanks. P indicates the PvuII restriction sites flanking the RSS. p12/23, p12 and p23 are identical except for the presence or absence of RSS. Linear fragment substrates (fp12/23, fp12 and fp23) were made by digesting the plasmid substrates with PvuII followed by gel purification of the RSS-containing fragment. The approximate size of the substrates is indicated to the right of the diagrams. (B) Plasmid substrates (lanes 1 and 3–5) or fragment substrates (lanes 2 and 6–8) were incubated with purified RAG proteins and HMG-1. The reactions were phenol extracted, digested with PvuII (plasmid substrates), separated using a 4.5% polyacrylamide gel and transferred to nylon membranes. The blots were hybridized to a random primed ³²P-labeled probe derived from the relevant PvuII fragment of p12/23. Diagrams of the resulting cleavage products are shown on the left. (C) A 5′-³²P-labeled primer was used with an unlabeled primer to amplify an end-labeled 140 bp 12-RSS-containing PCR substrate from p12. Unlabeled primer sets were used to generate unlabeled 12-RSS and 23-RSS-containing PCR substrates from p12 and p23, respectively. These substrates have the same coding flank length as the oligonucleotide substrates. Cleavage reactions were performed as with oligonucleotide substrates (Fig. 1). –R indicates reactions incubated without RAG proteins. The asterisk indicates a radioactive label. The products were separated on a 12% denaturing polyacrylamide gel. Diagrams on the left indicate the positions of the uncleaved 12-RSS substrate, the hairpin product and the nicked product.

Why are plasmids good substrates for cis, but not trans, cleavage under conditions that foster trans cleavage of oligonucleotide substrates? Oligonucleotides differ from plasmid substrates in several significant ways: they are linear, they are short and the RSS is located a short distance from a DNA end. To test the effects of each of these parameters independently on the efficiency of trans cleavage we analyzed additional substrates. We first examined the behavior of short linear fragments created by digesting the various plasmid substrates with PvuII and gel purifying the resulting 694 bp fragments containing the RSS (fp12/23, fp12 and fp23). Although these molecules were good substrates for cis cleavage (Fig. 2B, lane 2), they did not allow trans cleavage (compare lane 8 with lanes 6 and 7). Further shortening the substrates to 300 bp failed to promote trans cleavage, even when the heptamer of the RSS was located the same distance from the end as in the oligonucleotide substrates (data not shown), suggesting that neither substrate topology nor the distance of the RSS from a DNA end determine the efficiency of trans cleavage by purified RAG proteins. Substrates ∼140 bp in length did, however, undergo trans cleavage with an efficiency similar to that observed with oligonucleotide substrates (Fig. 2C, lane 3). Thus, under these conditions, which support efficient cleavage in cis, efficient trans cleavage requires substrates shorter than 300 bp.

These data indicate that substrates greater than a few hundred base pairs in length provide a barrier to trans synapsis. One possibility is that longer substrates provide more phosphate backbone charges, contributing more electrostatic repulsion energy that must be overcome to allow synapsis. In the cis situation the high local concentration of the two sites may overcome this electrostatic barrier. In the case of short oligonucleotides the barrier may be substantially lowered by charge shielding resulting from binding of the RAG proteins to the DNA (see Discussion).

Efficient trans cleavage of plasmid substrates by crude extracts

The experiments described above (summarized in Table 1) demonstrate that plasmid substrates are not efficiently cleaved in trans by purified RAG proteins. These results conflict with our previous discovery that trans cleavage of these same substrates is quite efficient in vivo (18). Therefore, we investigated the behavior of crude extracts, which might contain factors missing from the purified protein system. We prepared crude extracts from Chinese hamster ovary cells transiently transfected with RAG expression vectors (the same cells used previously to study trans cleavage in vivo). These extracts support robust cleavage of supercoiled plasmid substrates (4 fmol) in cis (Fig. 3, lane 1). Cleavage in crude extracts is quite efficient: typically 30–40% of the substrate is cleaved at both RSS, an efficiency comparable to that observed in vivo using the same substrate (11). Most cleavage in this extract system was coupled (i.e. both RSS are cleaved), but some uncoupled, single RSS cleavage of p12/23 occurred (lane 1), as it does in vivo (6,11,28). Very little cleavage of plasmids containing a single RSS was observed (Fig. 3, lanes 2 and 3), even at long exposures.

Table 1. Summary of cleavage of various substrates.

	Purified RAG proteins	Crude extract
Cis substrates
8 kb p12/23	+++	++++
8 kb Lp12/23	nt	++++
0.7 kb fp12/23	+++	++++
Trans substrates
8 kb Lp12 + Lp23	nt	++++
8 kb p12 + p23	–	++++
3 kb p12 + p23	nt	+++
2 kb p12 + p23	nt	++
1 kb p12 + p23	nt	+
0.7 kb p12 + p23	–	+/–a
140 bp 12 + 23	++	nt
50 bp 12 + 23	+	+++b

Pluses indicate the relative efficiency of cleavage of each substrate. Values shown are representative of several experiments.

nt, not tested.

^aA slight increase in cleavage is observed at higher substrate concentrations.

^bCleavage assays with the 50 bp oligonucleotide substrates required higher substrate concentrations than the plasmid cleavage assays to observe trans cleavage.

Figure 3.

Trans cleavage of plasmid substrates in crude extracts. Plasmid substrates (shown in Fig. 2A) were incubated with crude extracts containing RAG proteins (lanes 1–4) or without RAG proteins (lane 5) for 3 h at 30°C. –R indicates incubations in extract without RAG proteins. Reactions were treated as described in Figure 2B. Diagrams of the reaction products are shown on the right.

We next assessed the ability of the extracts to perform cleavage in trans. Co-incubation of equimolar amounts (4 fmol each) of supercoiled plasmid substrates that contain only a single consensus RSS (p12 or p23) in the extract resulted in efficient trans cleavage (lane 4), with levels of signal and coding ends substantially above those seen with either substrate alone (lanes 2 and 3) and similar to those generated by cleavage in cis (compare lanes 1 and 4). The efficient trans cleavage of plasmid substrates observed in vivo and in our crude extract system contrasts sharply with its absence in experiments using purified RAG proteins (8; and see above). These data strongly suggest that cis and trans cleavage have distinct requirements and indicate that factors present in the crude cell extract may promote coupled plasmid cleavage in trans.

Why then did previous studies using a similar crude extract system fail to detect cleavage in trans (7)? Because those studies examined only fairly short (531 bp) linear substrates, we investigated the effect of substrate topology and length on the efficiency of trans cleavage. Linear p12/23 and circular supercoiled p12/23 substrates (8 kb in length) were cleaved in cis with similar efficiency (Fig. 4A, lanes 1 and 2). Trans cleavage was similarly unaffected by linearization of the substrates (compare lanes 7 and 8) and we observed no difference in the efficiency of trans cleavage of relaxed or nicked substrates (data not shown). Thus, linear substrates are competent for trans cleavage.

Figure 4.

Effects of substrate topology and length on trans cleavage in crude extracts. Cleavage reactions were performed as described in Figure 3, varying the substrate topology and length. (A) Long (8 kb) linear substrates (Lp12/23, Lp12 and Lp23) were made by linearizing p12/23, p12 and p23 at a unique restriction site. fp12/23, fp12 and fp23 are described in Figure 2A. All of the substrates were cleaved in cis with similar efficiencies (lanes 1–3). Although some degradation of the coding ends from the 700 bp cis substrate was observed (lane 3), they were readily detected in longer exposures. Trans cleavage of the 700 bp substrates was substantially reduced compared to the 8 kb substrates (lanes 7–9). Mock incubations (in cleavage buffer without extract) of the 700 bp substrates are shown in lanes 6 and 12. (B) Intermediate length substrates (3, 2 and 1 kb) were generated by PCR and used in cleavage reactions. The efficiency of trans cleavage gradually decreased with decreasing substrate length (lanes 1–5). All lanes are from the same blot. (C) fp12 and fp23 samples were incubated with a non-homologous plasmid, λdv, (lane 3) which accounts for the mass difference between the long and short recombination substrates. (D) A 12-RSS was cloned into the pAB4 backbone to generate a non-homologous 12-RSS substrate, pAB4-12. Cleavage of p23 was only observed upon co-incubation with pAB4-12 (lane 3). The probe was made from p23 and does not hybridize to pAB4.

We next investigated the effect of substrate length using the 694 bp linear fragments described above. While fp12/23 and circular p12/23 were cleaved in cis with similar efficiencies in multiple experiments (see for example Fig. 4A, lanes 1 and 3), trans cleavage of the fp12 and fp23 substrates was consistently and substantially reduced (lane 9). (Single RSS cleavage of fp12/23 was also reduced.) We considered the possibility that the crude extract might contain exonuclease activity that could cause some degradation of the short substrates, as described previously (7). Our data, however, provide three lines of evidence eliminating degradation as a cause of the reduced trans cleavage of short substrates. First, while the very short coding end fragments generated from cleavage of short cis substrates are heterogeneous in size (suggesting limited exonucleolytic degradation), they are readily detected (Fig. 4A, lane 3 and longer exposure). Second, there is no significant reduction in levels of the uncleaved cis (lane 3) or trans (lane 9) substrates (which are almost the same size as the products of trans cleavage) or the products of dual RSS cleavage in cis (lane 3) (see also Fig. 4B–D). Third, no degradation of short substrates incubated in extract was observed (compare lanes 9, 11 and 12). Thus, signal ends derived from trans cleavage, which are close to the size of the uncleaved substrate (and much larger than dual RSS cleavage products), should be readily detected if formed. Based on the data from the short linear (694 bp) fragments and other shortened substrates (see below), we conclude that, under our conditions, cis and trans cleavage are differentially affected by substrate length.

We next determined how the efficiency of trans cleavage varies with the length of the substrate. Linear substrates of varying length (3000, 2000 and 1000 bp) were generated from the substrate plasmids by PCR. Incubation of these substrates with the extract revealed a gradual decrease in the efficiency of trans cleavage as the substrates were shortened (Fig. 4B, lanes 1–5, and data summarized in Table 1). This length dependence was not observed with cis substrates (data not shown).

To determine whether the ‘additional’ DNA present in the long substrates must be physically connected to the RSS, we added an equimolar amount of a plasmid that lacks RSS to reactions containing the short substrates. Addition of a 6 kb circular supercoiled plasmid, λdv, which does not share sequence homology with our substrates, to the cleavage reactions did not appreciably stimulate trans cleavage (Fig. 4C, lane 3). As an additional test, we digested the recombination substrates with PvuII, as before, but did not gel purify the fragment containing the RSS. The efficiency of trans cleavage with this substrate was comparable to that of the gel-purified fragments (data not shown). Together, these data demonstrate that the additional non-specific DNA present in the ‘long’ substrates must be physically connected to the RSS to promote efficient trans cleavage.

Because p12 and p23 have identical plasmid backbones, one explanation for the ability of additional DNA sequence to promote trans cleavage could be that homologous pairing (perhaps mediated by a factor present in the crude extract) stimulates trans cleavage. To test this possibility, we inserted a 12-RSS into the plasmid pAB4 backbone, which does not share significant homology with the p12/23 family (see Materials and Methods). Incubation of this substrate, pAB4-12, with p23 substantially stimulates cleavage of p23 (Fig. 4D, lane 3). Cleavage requires the 12-RSS, as it is not observed using equivalent molar amounts of pAB4 (lane 2). Thus, trans cleavage in the extract does not require homologous pairing of the two partner DNA molecules.

As shown above, purified RAG proteins are capable of carrying out trans cleavage of oligonucleotide substrates. Since substrates several hundred base pairs in length did not undergo efficient trans cleavage in extracts in our experiments, we tested whether our extracts are capable of cleaving oligonucleotide substrates in trans. Conversion of a radiolabeled 23-RSS oligonucleotide to hairpin form was observed in the presence of equimolar (20 fmol/reaction) unlabeled 12-RSS (Fig. 5A, lane 2). Cleavage was enhanced by a 10-fold molar excess of the partner RSS (lane 6), as shown above and by others (27) for purified RAG proteins. As expected, trans cleavage depended upon the presence of an authentic partner RSS, because cleavage was not observed using a non-specific partner oligonucleotide (lanes 4 and 8). These experiments clearly show that the crude extracts can catalyze trans cleavage of oligonucleotide substrates.

Figure 5.

Trans cleavage of oligonucleotide substrates and short non-oligonucleotide substrates in crude extracts. (A) Trans cleavage of oligonucleotide substrates by extracts. A double-stranded oligonucleotide substrate containing the 23-RSS was radioactively labeled on the 5′-end of the top strand and incubated with a RAG-containing whole cell extract and an equimolar amount of the indicated unlabeled oligonucleotide substrates (lanes 2–4) or with 10-fold excess of unlabeled oligonucleotide substrates (lanes 6–8) (see Materials and Methods). –R indicates samples incubated with an extract lacking RAG proteins (lanes 4 and 9). Products were separated on a 12% denaturing polyacrylamide gel. The uncleaved substrate, hairpin products and nicked products are indicated by the diagrams on the left. (B) Increased trans cleavage of short non-oligonucleotide substrates was observed at higher substrate concentrations. The 700 bp linear substrates were assayed for trans cleavage at 4 (lanes 2–4) and 20 fmol/reaction (lanes 5–8).

Why are extracts capable of trans cleavage of oligonucleotides but not of short linear substrates? One possibility is that the two types of substrate were assayed at different DNA concentrations. Oligonucleotide cleavage reactions are typically carried out using 20 fmol substrate/10 µl reaction (9,29). In our standard plasmid substrate cleavage reactions, however, the substrate concentration is 5-fold lower (4 fmol/10 µl reaction). As shown in Figure 5B, while trans cleavage is barely (if at all) detectable with 4 fmol substrate (lane 4), increased trans cleavage (2-fold, as measured by PhosphorImager analysis) is seen with 20 fmol substrate (lane 7). We observed no significant stimulation of cis cleavage of long or short substrates at the higher substrate concentration (data not shown). It is important to note that although the increased substrate concentration assists trans cleavage, the effect is quite modest: cleavage of the short substrate at the high concentration is only 30% as efficient as with lower concentrations of the long substrate. Thus, while increasing the substrate concentration can partially compensate for the trans cleavage DNA length requirement, there remains a substantial length dependence at both the low and the high DNA concentrations. Potential mechanistic explanations for this effect are discussed below.

DISCUSSION

Length dependence of trans cleavage by purified RAG proteins

Our results reveal several effects of substrate length on the efficiency of trans cleavage (summarized in Table 1). The first observation is that purified RAG proteins are capable of efficiently cleaving oligonucleotides, but not longer DNA substrates, in trans. Thus, the length of the substrate substantially and specifically affects the efficiency of trans cleavage.

Several substrate length effects have been observed in the bacteriophage Mu transposase system. Intermolecular strand transfer by MuA is much more efficient with short oligonucleotides than with plasmid substrates (30). Furthermore, short substrates promote tetramerization of the MuA transposase more effectively than longer substrates (31). Finally, assembly of the Mu synaptic complex is stimulated by addition of an enhancer sequence (the IAS), which is normally present in cis. When the IAS is added in trans, a short IAS-containing DNA fragment is much more effective than a long fragment at stimulating assembly (32).

Based on these observations, Mizuuchi and co-workers proposed that trans synapsis of larger DNA fragments may be inhibited by electrostatic repulsion that can be overcome by charge shielding when the transposase binds to short DNA fragments (30). Similar considerations may pertain to assembly of synaptic complexes by purified RAG proteins: electrostatic repulsion by backbone phosphates may provide a significant barrier to synapsis. When the RSS are configured in cis this barrier is overcome by the fact that the two RSSs are tethered on the same molecule, leading to a very highly effective local DNA concentration. The barrier can also be overcome by catenating the two substrates (19) and, according to our results, at least partially by increasing the DNA concentration. This model is supported by our observation that, in comparison with cis cleavage, trans cleavage requires a higher concentration of divalent cations (data not shown), which are particularly effective at shielding the charge of the phosphate backbone (33,34).

Regardless of the mechanistic basis of the DNA length effect, our data demonstrate that there are significant differences in the behavior of short oligonucleotide substrates and longer DNA molecules. Thus, while oligonucleotides provide a convenient and potentially powerful way to study the biochemistry of V(D)J recombination, care must be taken when extending observations made with oligonucleotides to longer, more physiologically relevant substrates.

Cleavage in trans by crude extracts

Like the purified protein system, crude extracts are capable of cleaving oligonucleotides in trans, but fail to catalyze cleavage of the 700 bp linear substrates. As in the purified RAG system, productive synapsis of the 700 bp substrates in trans may be inhibited by electrostatic repulsion. Interestingly, efficient trans cleavage is observed with longer DNA substrates (>1 kb); plasmid substrates are efficiently cleaved both in cis and in trans, as observed in vivo. These data suggest that the crude extract system provides alternative mechanisms that can overcome the barrier to trans cleavage in a DNA length-dependent fashion.

There are several possible explanations for this observation. One obvious possibility is that increased substrate length promotes protein–DNA interactions that increase the effective concentration of the substrates, leading to more efficient trans synapsis. Such interactions might occur in several ways: (i) the RAG proteins could non-specifically interact with multiple DNA substrates; (ii) the RAG proteins could promote intermolecular synapsis via binding to cryptic, non-canonical RSS (it is worth noting that such sequences occur, on average, once every 600 nt; 35); (iii) other proteins in the extract could bind to both substrates, effectively bridging the two molecules. These models are not mutually exclusive and the length effect may depend on several mechanisms operating together.

Our experiments argue strongly against a role for homologous pairing, as completely non-homologous substrates are efficiently cleaved in trans. These data also suggest that the effect is not mediated by a particular DNA sequence. Furthermore, our analysis of substrates of varying length shows a gradual diminution in the efficiency of trans cleavage, rather than the sharp decrease one would expect upon removal of a discrete sequence, such as a recombination enhancer.

Another possibility is that the trans cleavage-enhancing effects of long DNA reflect the assembly pathway for synaptic complex formation. There are several precedents for substrate length affecting the association (and dissociation) of specific DNA–protein complexes. Assembly of the bacteriophage Mu transpososome is sensitive to both DNA topology and substrate length (31) and the kinetics of formation and dissociation of DNA protein complexes can be strongly influenced by the length of the target molecule (36).

In conclusion, our results show that the requirements of V(D)J cleavage in cis differ in some significant respects from cleavage in trans. Understanding these differences in detail may provide important information about the assembly of synaptic complexes, a process that remains poorly understood. Our studies of trans cleavage also provide a clear demonstration that the behavior of the system most commonly used for biochemical analysis of the V(D)J recombination reaction—cleavage of oligonucleotide substrates by purified RAG proteins—differs in significant ways from the in vivo situation. Further work using this system may allow the identification of important regulatory factors that assist proper assembly of synaptic complexes.