INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

V(D)J recombination assembles the exons that encode the antigen binding domain of immunoglobulin and T-cell receptor genes during the generation of B and T cells (12, 27). In the genome, each of the elements to be recombined, V (variable), D (diversity), and J (joining), is located adjacent to a specific DNA sequence called a recombination signal sequence (RSS), which directs the recombination. The RSS contains a conserved heptamer (5'-CACAGTG-3') and nonamer (5'-ACAAAAACC-3') separated by a nonconserved spacer of either 12 or 23 bp, termed 12RSS and 23RSS, respectively (18). The nucleotides of the V, D, or J segment that are immediately adjacent to the heptamer are referred to as the coding end nucleotides. During V(D)J recombination, DNA double-strand breaks are introduced at the junction between the RSS and the adjacent coding V, D, or J segment; the corresponding broken ends are called either the signal ends or the coding ends, accordingly. V(D)J recombination typically involves two DNA double-strand breaks: one at an element with a 12-RSS and the other at an element with a 23-RSS. This feature is known as the 12/23 rule (45). Completion of V(D)J recombination depends on the proper joining of the two coding ends to each other and of the two signal ends to one another through a general DNA double-strand break repair pathway known as nonhomologous end joining (26).

The genes encoding the recombination enzymes that cleave the DNA and initiate V(D)J recombination are called recombination activation genes (RAGs). The RAG protein complex consists of two gene products: RAG1 and RAG2 (4, 20, 34, 40, 44). RAG-mediated cleavage occurs in two steps (30). First, a nick is introduced at the heptamer-coding sequence junction 5' to the heptamer. This results in a 3' hydroxyl group at the end of the coding sequence and a 5' phosphate group attached to the heptamer end. Second, the 3' hydroxyl group of the coding sequence attacks the antiparallel strand in a direct transesterification reaction (12), which cleaves the DNA into a hairpin-structured coding end and a blunt signal end. This general type of DNA cleavage is also observed in other DNA transposition systems, such as Tn10 transposition (24). The organization of the RAG1 and RAG2 genes in the genome is consistent with an origin from a DNA transposon (35). Indeed, it was found recently that recombinant RAG proteins have transposase activity, in that they can insert DNA fragments with signal ends into other DNA molecules (1, 21, 35, 37), though such activity has not been documented in vivo. It is believed that V(D)J recombination evolved from an ancient DNA transposition event (35, 37).

In all of the transposition reactions characterized thus far, synapsis of two transposon ends is required before any strand nicking or strand transfer (double-strand breakage) occurs (2, 15, 31, 33, 38, 41, 48). It is important to note that two like ends (e.g., two right ends) can equivalently supplant a left and a right end for all steps in some transposition systems (3).

For in vitro V(D)J recombination, it has been clear that nicking does not require the presence of both a 12- and a 23-signal in solution (47). Based on this, some have inferred that nicking is entirely independent of synapsis (12). This conjecture overlooks the possibility that two identical ends (e.g., two 12-signals) could provide synapsis equivalent to a 12/23 synapsis. In fact, direct and indirect (competition) assays of synapsis have documented that 12/12 or 23/23 synapsis occurs either at only moderately (10-fold) reduced levels (19) or at levels similar to those for 12/23 synapsis (51).

In contrast with the assumption that nicking is independent of synapsis, data from others suggests that synapsis can actually stimulate nicking in V(D)J recombination (9). However, the inference of stimulation is based on a control which assumes that RAG binding at one signal is independent of RAG binding at a second signal located 89 bp away on the same DNA fragment. There is no data assuring that such close signals do not interfere with one another (8, 9, 42). In addition, the interpretation of this type of data is complicated by the fact that binding and nicking at one signal may result in rapid collisional rates with the second tethered signal (zero order kinetics), eliminating the ability to determine the kinetic dependence of nicking on synapsis (see Discussion).

The precedent in the other transposition systems makes it even more important that the dependence of nicking on synapsis be directly tested. The only suggestion that there might be a difference between V(D)J recombination and the other biochemically characterized transposition reactions concerning synapsis prior to any catalysis is that the hairpin step is much more efficient with both the 12- and 23-signals present in the solution than with reactions having only one type of signal present (19, 47, 51). In contrast, the nicking step appears to be similarly efficient whether one or both types of signals are present (9, 47, 51, 53). It is important to note that this in no way indicates that synapsis is unnecessary for nicking. It simply indicates that if synapsis is required for both nicking and hairpinning, then the 12/23 rule is operating only at the hairpinning step. This could simply be indicative of a more stringent three-dimensional structure of the protein-DNA complex at the later step.

The uncertainty about synapsis at the nicking step affects how one views all of the subsequent steps. If nicking is independent of synapsis, then a synaptic step that has functional consequences must occur between the key catalytic steps of nicking and hairpinning (53). Hence, the ordering of the steps requires a test of the synaptic requirement at the point of nicking.

In the present study, we unambiguously demonstrate that nicking is independent of synapsis in a system where synapsis was prevented by immobilization of the DNA substrate on streptavidin beads. Moreover, the initial rate of nicking is consistent with a unireactant enzyme-catalyzed kinetic model. The kinetic analysis provides additional insights regarding the time required for nicking. These findings distinguish the mechanism of the RAG proteins from other characterized transposases in a key aspect, and they have implications for the development of an immune receptor repertoire.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA substrates. Nonbiotinylated oligonucleotides were synthesized by Operon Technologies, Inc. (Alameda, Calif.). A DNA substrate containing either a 12RSS or a 23RSS was made by annealing two complementary oligonucleotides. The 12-substrate for the initial rate assay was made by annealing KY28 (5' GAT CAG CTG ATA GCT ACC ACA GTG CTA CAG ACT GGA ACA AAA ACC CTG CT 3') and KY29 (5' TAG CAG GGT TTT TGT TCC AGT CTG TAG CAC TGT GGT AGC TAT CAG CTG AT 3'). The 23-substrate for the initial rate assay was made by annealing KY36 (5' GAT CAG CTG ACA GTA GCA CAG TGG TAG TAC TCC ACT CTC TGG CTG TAC AAA AAC CCT GCT 3') and KY37 (5' TAG CAG GGT TTT TGT ACA GCC AGA GAG TGG AGT ACT ACC ACT GTG CTA CTG TCA GCT GAT 3').

View larger version (44K): [in this window] [in a new window]   FIG. 1.   Cleavage with immobilized DNA substrate indicates that nicking is independent of synapsis. (A) Immobilization of DNA substrates on streptavidin agarose beads. The 12-substrate is depicted by an open triangle. B designates the biotin group. A star indicates the position of the radioisotope label. (B) Biotinylation does not interfere with 12/23 regulated in vitro nicking (N) and hairpin (HP) formation. S, substrate. All panel B reactions were performed in the absence of streptavidin agarose beads. The combinations of 12- and 23-substrates are indicated on the top of the gels. A dash indicates no DNA substrate. Other symbols are the same as described above. (C) Nicking and hairpin formation on an immobilized DNA substrate. Lane 1, both 12- and 23-substrates were biotinylated and immobilized on the streptavidin agarose beads. Lane 2, biotinylated 12-substrate was immobilized on the beads. Freely diffusible nonbiotinylated 23-substrate was added separately after the immobilization of the 12-substrate.

Protein expression and purification. Fusion proteins of maltose-binding protein (MBP) and core regions of RAG1 (amino acids 384 to 1008) and RAG2 (amino acids 1 to 388) were expressed and purified from baculovirus-infected insect cells as previously described (30, 47, 51). Glutathione S-transferase (GST)-fused truncated mouse RAG1 (amino acids 330 to 1040) and RAG2 (amino acids 1 to 383) were coexpressed in the human epithelial cell line 293T and purified as previously described (5, 39, 49). C-terminally truncated mouse high-mobility-group 1 protien (HMG1) was expressed in bacteria as a six-histidine-tagged protein and purified over a Ni-nitrilotriacetic acid column (Qiagen Inc., Valencia, Calif.) (51). Protein concentrations were determined using the Bradford method (Bio-Rad, Hercules, Calif.). The molar concentration of the RAGs was calculated assuming two RAG1 and two RAG2 polypeptides (tetramer). (In this regard, it is noteworthy that recent experiments indicate a trans cleavage mechanism by a complex with two RAG1s and one or two RAG2s [P. Swanson, personal communication], suggesting a trimer or tetramer stoichiometry.)

Nicking and hairpinning of DNA substrates bound on streptavidin beads. Exactly 0.2 pmoles of biotinylated DNA substrates in 100 µl of buffer (25 mM morpholinepropanesulfonic acid [pH 7.0], 5 mM MgCl₂, 30 mM KCl, 30 mM potassium glutamate) was mixed with 1 µl of streptavidin-agarose suspended in 100 µl of the same buffer and incubated at 37°C for 30 min. The beads were collected by low-speed centrifugation and washed twice with 100 µl of cleavage buffer. The cleavage reaction was initiated by the addition of 100 nM HMG1 and 200 ng of the RAG proteins and incubated for 30 min at 37°C. The mixture was then extracted with phenol and chloroform followed by ethanol precipitation. The DNA pellet was redissolved in 10 µl of formamide, heated for 5 min at 100°C, and then immediately quenched with ice water.

Initial rate of nicking. A 10-µl reaction mixture containing 25 mM morpholinepropanesulfonic acid (pH 7.0), 5 mM MgCl₂, 30 mM KCl, 30 mM potassium glutamate, 100 nM HMG1, 10 nM RAG proteins, 2 nM ³²P-labeled 12-substrate, and various amounts of unlabeled identical 12-substrate was incubated for periods of up to 5 min at 37°C. The reaction was stopped by the addition of 0.1% sodium dodecyl sulfate and 20 mM EDTA. The mixture was then extracted with phenol and chloroform followed by ethanol precipitation. The DNA pellet was redissolved in 10 µl of formamide, heated for 5 min at 100°C, and then immediately quenched with ice water.

Denaturing polyacrylamide gel electrophoresis. Reaction products were separated on 15% polyacrylamide gels containing 7 M urea in 1× Tris-borate-EDTA buffer. Gels were visualized by autoradiography by using a Molecular Dynamics (Sunnyvale, Calif.) PhosphorImager 445SI and quantified with ImageQuaNT software (version 1.0).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Nicking does not require synapsis. As a first test of a synaptic requirement for nicking, we immobilized the biotinylated 12-substrate on streptavidin agarose beads. The amount of biotinylated 12-substrate loaded onto the streptavidin beads was less than 1% of the binding capacity, ensuring that each DNA molecule tethered on the beads was separated maximally from other DNA molecules (Fig. 1A). This allowed us to perform the nicking and hairpinning assay under conditions such that synapsis was prevented.

Initial rate kinetics for the nicking step. If nicking is independent of synapsis, kinetically it should be a unireactant enzymatic reaction (in which the enzyme binds one substrate at a time and acts on it catalytically), and it should fit the rate equation v = k_N[E₀][S]/(K_m+[S]). In this equation, k_N is the catalytic constant for nicking, K_m is the Michaelis-Menten constant for the binding of the 12- or 23-substrate, [E₀] is the concentration of the total active RAGs, and [S] is the concentration of the free 12- or 23-substrate. To determine if the unireactant kinetic model fits the nicking reaction, we measured the initial rate of the nicking reaction at different substrate concentrations. The substrates used in this assay contain consensus RSSs (18) and optimal coding end sequences for maximal nicking and hairpinning (53). When plotting the initial velocity against the substrate concentration, we found that the data fit the kinetic model of a unireactant enzyme-catalyzed reaction quite well (Fig. 2). The constants for the kinetic equation were determined by curve fitting (Table 1). The initial rate measurements were performed with both GST-fused RAG proteins (Fig. 2A) and MBP-fused RAG proteins (Fig. 2B and C). Importantly, we found that the type of fusion protein does not alter the unireactant nature of the enzyme-catalyzed reaction.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Initial rates of nicking as a function of substrate concentration. The plot of the initial rate (v0) against the initial 12-substrate or 23-substrate concentration (S0). (A) Nicking of the 12-substrate by the GST-RAGs. (B) Nicking of the 12-substrate by the MBP-RAGs. (C) Nicking of the 23-substrate by the MBP-RAGs. Curve fitting was done with DeltaGraph 4.5, and the kinetic constants of best fit are listed in Table 1.

Burst kinetics and functional stoichiometry. The maximal initial rate, V_max, of a particular enzymatic reaction depends on the enzyme concentration. The catalytic constant, or k_cat (named k_N for nicking), which reflects the catalytic efficiency, can be deduced based on the equation V_max = k_cat [E]. However, it is difficult to determine the actual RAG protein concentration in a cleavage reaction. First, the stoichiometry of the active RAG complex has not been clear (4, 10, 44); this means that the real molar concentration cannot be directly deduced from the measured protein concentration. Second, the RAG preparation is not homogeneous because it consists of two gene products that are usually coexpressed. It can contain RAG1-RAG2 tetramers, RAG1-RAG2 dimers, RAG1-RAG1 dimers, and RAG2 oligomers (4, 5). Although it is reported that a tetramer of two RAG1 and two RAG2 molecules is the active RAG complex, it is not currently possible to purify the tetramer in the quantities necessary for kinetic studies (T. Bailin and M. Sadofsky, personal communication).

View larger version (30K): [in this window] [in a new window]   FIG. 3.   Coomassie staining of the RAG fusion proteins used in kinetic studies. The GST-RAGs were purified using glutathione agarose (see Materials and Methods). The MBP-RAGs were purified using nickel-nitrilotriacetic acid resin followed by amylose resin (see Materials and Methods). Molecular weight (MW) markers, given in thousands, are shown on the left side of the gel. The position of each RAG protein band is indicated on the right side of the gel.

View larger version (60K): [in this window] [in a new window]   FIG. 4.   Determining the active RAG concentration by burst kinetics. The top inset shows the nicking of a 12-substrate with 20, 40, and 80 nM total concentrations of RAGs for a 30-min time course. These results were graphed as the nicked product versus time (lower inset); the y intercepts from this plot are the burst at each enzyme concentration. The major figure shows the burst (y intercept from the lower inset plot) as a function of the RAG concentration; the slope of this plot gives the fraction of active MBP-RAGs in the protein preparation. The slope is 0.215, indicating that 21.5% of the RAGs are catalytically active. CaCl2 was the source of the 1 mM Ca2+, and MgCl2 was the source of the 5 mM Mg2+.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have shown that immobilization of DNA substrates on streptavidin beads, which blocks synapsis, still permits their nicking but not their hairpinning. These are the first data to demonstrate that synapsis of two signals is not required for the nicking step, making it unique in this respect among transposition-type reactions. Our results demonstrate that the nicking step of V(D)J recombination occurs as a one-substrate enzymatic reaction. The nicking of the 12-substrate alone is a unireactant enzyme-catalyzed reaction, and the same is true for the nicking of the 23-substrate alone. This indicates that even if a RAG complex binds to a 12- and a 23-RSS simultaneously, it would still nick them independently. Based on the results here, we can infer a kinetic scheme for the RAG-mediated nicking and hairpinning during the initiation of V(D)J recombination (Fig. 5).

View larger version (8K): [in this window] [in a new window]   FIG. 5.   Kinetic scheme for the binding, nicking, synapsis, and hairpin formation steps mediated by the RAG complex (RAG1, RAG2, and HMG1). 12 represents a 12-substrate (12RSS and adjacent DNA). Equilibrium constants are depicted with K, and rate constants are depicted with k. R represents the active RAG complexes. R:12 is the RAG complex bound to the substrate. 12N is the nicked form of the 12-substrate. 12H is the hairpinned form of the 12-substrate. R:12N:23N is the synaptic complex of nicked 12- and 23-substrates.

Comparison to other studies on nicking in V(D)J recombination. It has been shown that in a solution containing only 12-substrates, those 12-substrates could undergo nicking (8, 9, 14, 46, 47). However, in all of those previous studies, it was unclear whether two 12-substrates might synapse to permit the nicking step or the 12-substrates were being nicked individually (and prior to any synapsis). The same applies to reactions involving only 23-substrates. Our immobilization and kinetics studies clearly indicate that the nicking step occurs as a one-substrate reaction, even when both 12- and 23-substrates are present.

Physiologic significance of a unireactant nicking step. Chromatin structure regulates the local accessibility of RAG recombinases (6, 16, 32, 43). The immunoglobulin intronic enhancers have been found to be essential for opening up the chromatin (6, 11, 16, 22, 23, 32, 43). The fact that DJ recombination typically precedes the V to DJ recombination at the immunoglobulin H locus may reflect a gradual chromosomal structural change initiated from the enhancer near the J_H region and extending into the distal V cluster. Our determination that nicking can occur without synapsis means that it is quite possible that the J_H segments may exist in the nicked form before any V_H and D segments become accessible to the RAGs.

Utility of a kinetic description of the nicking step. It is useful to compare the fraction of active RAGs from the functional stoichiometry and burst kinetics with electrophoretic mobility shift results using the same RAG preparation. When k_N is sufficiently small, the dissociation constant K_D is equal to K_m. It is not possible to determine K_D by a gel mobility shift without knowing the active RAG concentration (i.e., the concentration of RAGs that are able to bind the DNA substrate). If we substitute K_m for K_D, then from our previous electrophoretic mobility shift assay (see Fig. 5 of reference 53), it can be estimated that maximally about 7.5% of the GST-RAG proteins are able to form a complex with DNA substrates ([E] = K_D[ES]/[S]  K_m[ES]/[S]). This sets an upper limit for the active GST-RAG fraction, because K_D must be smaller than K_m. This is consistent with our burst kinetics (3.2%) and functional stoichiometry (2.0%) results on the GST-RAGs, which indicate that only a small fraction of the RAG protein preparation is active. Corresponding estimations can be done for the MBP-RAGs. Mobility shift analysis places an upper limit of 50% on the fraction of RAGs active for substrate binding (data not shown). The burst kinetics analysis determination of 21.5% is in line with this estimate. Binding and catalytic activity by only a fraction of the RAG protein preparation probably explain why efficient nicking and hairpinning typically require more RAG protein than DNA substrate (5, 19, 25, 47, 51).

1

1

1

1

1

1

Comparison of V(D)J recombination to other transposition and site-specific recombination reactions. All other site-specific and transposition recombination reactions require synapsis prior to initiation of the first catalytic step [nicking in the case of V(D)J recombination]. One might have expected that synapsis would also be essential for V(D)J recombination based on this evolutionary trend. Synapsis ensures that nicking at the two recombination sites is coordinated. However, V(D)J recombination is the first such reaction to break this rule. We can only speculate as to the reasons for this deviation from other similar reactions. First, the effects on the immune repertoire discussed earlier may be important forces that may have favored this evolutionary deviation from the other transposition reactions. Second, V(D)J recombination requires that the recombinase find the signals in a genome which is more complex than any other site-specific or transposition reaction. In many transposition reactions, such as in the case of retroviruses, the transposase is packaged within the nucleocapsid; when double-stranded DNA is generated from the infecting RNA, the transposase must bind to the ends of a genome that is only kilobases in length. In the case of prokaryotic transposases, the bacterial genome is relatively small. Based on nicking without synapsis, a potentially wider number of V, D, and J segments may contribute to the repertoire, as has been mentioned. The multisite nature of this transposition reaction and its use in an immune defense process make it advantageous to open the first step of this process to as many V, D, and J segments as possible. For this reason, a unireactant first step may have been a distinct evolutionary advantage.