Recombinant core RAG1, expressed and purified from bacteria as a fusion protein with maltose binding protein (MBP), was previously shown to be active in in vitro DNA cleavage assays when combined with RAG2 purified from 293T cells 24. Recombinant core RAG1 (residues 384–1008) is produced in a soluble form at relatively high levels from this system, which has been useful for characterizing fundamental physicochemical properties of this protein. In contrast, production of the full length RAG1 protein in bacteria, to date, only yielded aggregated and inactive protein (unpublished data).

The final purification step of the MBP-core RAG1 fusion protein (MCR1) was size exclusion chromatography (SEC) using a 120 ml Superdex 200 gel filtration column (see Methods). The SEC purification step was used to separate the dimeric form of MCR1 from aggregated material that eluted in the void volume. In addition to the dimeric form of the protein, higher order oligomeric forms of the protein were generated. In this study, and also as done previously 21, we focused on the characterization of three fractions of MCR1 (referred to as Fractions A, B, and C) that were pooled after the preparative SEC gel filtration step (Figure 1).

The separate fractions were further characterized using multi-angle laser light scattering combined with SEC, referred to as MALLS-SEC. The SEC was performed using a 20 ml analytical Superdex 200 column that resulted in a significantly higher resolution for the separation of the MCR1 oligomeric species than the preparative SEC column, especially for Fraction C. The elution profiles of all three fractions obtained from MALLS-SEC at 10°C are shown in Figure 2. Consistent with our previous results 21, MCR1 self-associated as tetrameric and octameric forms, in addition to the dimeric form. Specifically, the elution profile of Fraction A showed two distinct peaks (resolved from the void volume) that corresponded in molecular mass to tetrameric [elution volume at maximum peak height (V_e max) of 10.4 mL] and dimeric (V_e max of 11.8 mL) MCR1 (Figure 2A &2B, top panel). The elution profile of Fraction B also showed two distinct peaks (Figure 2A); however, the first resolved peak eluted over a broad volume range (V_e from 9.3 to 11.0 mL) and was more polydisperse in molecular mass (Figure 2B, middle panel), indicating co-elution of multiple oligomers (from tetramer to octamer). Lastly, a resolved peak (V_e max of 9.8 mL) in the elution profile of Fraction C (Figure 2A) corresponded to octameric MCR1 as determined by MALLS-SEC analysis (Figure 2B, bottom panel). The elution profile of Fraction C also included a small peak at V_e max of 11.7 mL. The concentration of protein eluting at this volume was too low to yield an accurate molecular mass; however, the V_e max is consistent with dimeric MCR1. The first peak in each of the chromatograms in Figure 2B corresponded to protein that eluted in the void volume of the column, and likely included misfolded and/or non-specifically aggregated protein.

Previously, the relative proportions of the dimer, tetramer, and octamer forms of MCR1 as a function of ionic strength were examined 21. At high ionic strengths (>0.5 M) it was found that the dimer form was favored; however, at near physiological ionic strengths (0.2 M) the total protein was comprised of an increased proportion of both the tetramer and octamer forms. While the different oligomeric MCR1 forms appeared to re-establish equilibria after changes in ionic strength, the redistribution occurred slowly (over ~48 hours). Therefore, the octamer could dissociate to smaller oligomers, albeit slowly, indicating that it was not formed by nonspecific aggregation of misfolded protein 21. It is not likely MCR1 oligomers of a higher order than octamer exist, since protein that eluted in the void volume did not dissociate to lower oligomeric forms even at high ionic strengths 21.

Our findings of the three distinct oligomeric forms of RAG1 coupled with the contradictory reports of RAG1 stoichiometry in catalytically-active complexes 111213 led us to further investigate the self-association properties of this protein. In this study we examined how the self-association properties of MCR1 varied with temperature (previous studies were performed at a single temperature with MCR1 samples incubated at 4°C prior to SEC 21). Fractions A, B, and C were each incubated at the indicated temperatures for 30 min prior to analysis by SEC. With increasing temperature, the proportion of tetrameric protein appeared to decrease in Fraction B (Figure 3), as well as in Fraction A (Figure 3, inset). In Fraction A, the only retained peak at the higher temperatures contained dimeric MCR1. Likewise, in Fraction B, the proportion of tetrameric MCR1 in the broad peak at 10°C appeared to decrease at the higher temperatures, yielding resolved peaks with V_e values consistent with the dimer and octamer forms only (Figure 3). It is likely that the reduction in the tetrameric form at temperatures higher than 10°C was due either to dissociation to the dimer form, or more likely, association to higher order oligomeric forms. With Fraction C, the octamer peak was present in the elution profiles over the entire temperature range of 10–37°C (not shown).

Of the resolved MCR1 species from SEC (excluding the void volume), Fractions A and C contained predominantly dimer or octamer species, respectively (Figure 2A). In contrast, Fraction B contained a mixture of oligomers, with the tetramer form only detectable at the lower temperatures (Figure 3). It is feasible that octameric RAG1 could contribute to RAG1 function in V(D)J recombination, due to its presence at physiological temperatures. To assess this possibility, we compared Fractions A and C in RAG1-related functions, including binding to 12-RSS, complex formation with RAG2, and DNA cleavage in a single 12-RSS cleavage assay.

To test for the affinity of binding to a single 12-RSS, electrophoretic mobility shift assays (EMSA) were performed. Increasing concentrations of MCR1 Fractions A and C were incubated with radiolabeled 12-RSS prior to electrophoresis under non-denaturing conditions. Fraction A (containing dimeric MCR1) formed multiple complexes with the RSS at protein concentrations of less than 0.2 μM (Figure 4A), consistent with our previous results in which we showed that MCR1 can form two major complexes with a single 12-RSS prior to saturation of the DNA 24. In previous studies it was determined that the fastest migrating MCR1:RSS complex in the nondenaturing gel contained a single MCR1 dimer 24, whereas the slower migrating complex contained four MCR1 subunits bound to the 12-RSS 20.

The band labeled in Figure 4A as the 4-subunit MCR1:12-RSS complex may form from the interaction of the pre-formed tetramer or from two separate MCR1 dimers with the RSS. While there is some evidence that this complex may contain pre-formed tetramer 20, it is more likely that it is due to the binding of two separate dimers to the RSS, since (as shown in Figure 3) the pre-formed tetramer is less evident at 25°C (the typical incubation temperature used in the mobility shift experiments). However, in the event that two separate MCR1 dimers bind the RSS, it is not clear if both dimers form contacts with the RSS. Additionally, if the two dimers form contacts with each other on the RSS, then it is not possible to determine by EMSA if the interface is the same as that in the pre-formed tetramer. Due to these limitations, the slower mobility complex will be referred to as a 4-subunit MCR1:RSS complex rather than a tetrameric MCR1:RSS complex (Figure 4A).

A greater than 10-fold increase in the concentration of MCR1 Fraction C was required to shift an equivalent fraction of the RSS as compared to Fraction A (Figure 4A). It is likely, however, that the DNA binding observed in Fraction C was due to small amount of dimer present in this sample (see Figure 2A). Moreover, the relative migration of the protein-DNA complexes in the gel was the same between Fractions A and C. Since the fastest migrating band corresponds to dimeric MCR1 bound to the 12-RSS, this would be inconsistent with pre-formed MCR1 octamer forming a complex with the 12-RSS.

To compare the interaction of Fractions A and C with RAG2, a pull-down assay using core RAG2 fused to glutathione S-transferase (GST) was performed. GST-core RAG2 was incubated with the MCR1 fractions in the presence of glutathione-sepharose resin. After extensive washing of the resin with binding buffer, the proteins were released from the resin, resolved by SDS-PAGE, and analyzed by Western blot (Figure 4B). MCR1 Fraction A, but not Fraction C, bound to GST-core RAG2 under the conditions used here. It is possible that within the octameric form, the RAG1 subunits may be associated in a way that shields both the RSS and RAG2 binding sites.

As expected from the lack of RSS or RAG2 binding, MCR1 fraction C was inactive in a single RSS cleavage assay (when combined with RAG2). In this assay, both MCR1 fractions were incubated separately with GST-core RAG2 followed by the addition of double-stranded WT 12-RSS. As shown in Figure 4C (lane 6), dimeric MCR1 (Fraction A) was catalytically active and formed the hairpin and nicked products, whereas the octameric MCR1 (Fraction C) was inactive even with increased concentrations of protein from Fraction C (lanes 3&4).

It was somewhat surprising that Fraction C showed no evidence of DNA cleavage activity or interaction with RAG2, given that this fraction contained a small proportion of dimeric MCR1. It is possible that dissociation of the octameric protein to a dimeric form resulted in protein in an inactive conformation. Nevertheless, it is apparent that Fraction A isolated by preparative SEC contains the active form of MCR1.

Although the dimeric form of MCR1 was the most effective at carrying out RAG1 functions, the relative proportion of the dimeric form in Fraction A was substantially decreased with increasing temperature, with the majority of protein in an aggregated state at 25 and 37°C (Figure 5A). Specifically, ~60% of the total protein in Fraction A was dimeric at 10°C, whereas <10% was dimeric at temperatures of 25°C and greater, with the remainder aggregated in an inactive form. The aggregation occurred within an incubation period of 30 min at two temperatures tested, with no further aggregation occurring between 30 to 150 min incubation times (Figure 5B).

To determine if the aggregation of dimeric MCR1 was due to protein unfolding, we measured the thermal denaturation of dimeric MCR1 (Fraction A) by circular dichroism (CD) spectroscopy. The protein showed two transitions with a minor transition occurring between 20 to ~37°C and a major transition at ~55°C (Figure 6A). Both transitions were irreversible under the conditions used here. Notably, the major protein unfolding transition occurred at higher temperatures (~50–60°C) than the range of interest (10–37°C). The denaturation curve of MBP alone was also measured, with the transition temperature for MBP unfolding at ~57°C, with a relatively constant CD signal from 10–50°C (Figure 6A). Therefore, it is consistent that the minor transition between 20–37°C is due to the RAG1 portion of MCR1, and thus this transition may reflect conformational changes that occur upon oligomerization of the protein.

A less likely possibility for the minor CD transition between 20–37°C could be due to a temperature dependent disruption in an interface formed between the core RAG1 and MBP portions of the fusion protein. A change in CD signal of the magnitude observed here would result from significant changes in the secondary structure of one or the other proteins upon complex dissociation. We believe such an interaction to be unlikely in this case since previous attempts to detect association between MBP and core RAG1 (after proteolytic cleavage from MBP) by various methods, such as chromatography and protein-protein crosslinking, were unsuccessful (not shown).

The CD spectra for MCR1 incubated at 20°C, 30°C, and 37°C for 60 min are shown in Figure 6B. At 10–20°C, MCR1 is predicted to contain 52% α-helices and 14% β-sheet, whereas at 37°C, the proportion of α-helical structure is decreased by 10%. Since the transition between 20–37°C was irreversible, there may be a time-dependent shift in the CD signal within this temperature range. Additionally, it was important to confirm that protein incubated at temperatures between 20–37°C did not enter into the major unfolding transition over time. To analyze time-dependent changes in the conformation within this temperature range, CD spectra were collected over a 2-hour incubation period at 20°C, 30°C, and 37°C. At temperatures representing the beginning and end of the first minor transition (20 and 37°C), there were no changes in either set of CD spectra over a 2 hr time period (Figure 6C, left and right panels). Therefore, MCR1 remained in its low temperature folded state at 20°C, while at 37°C the protein had obtained a stable intermediate conformation. After incubation at 30°C, the CD spectra changed slowly with time, with the final spectrum at 2 hr incubation matching the spectrum obtained at 37°C. This was expected, given the irreversibility of the minor transition. In conclusion, the majority of aggregation of dimeric MCR1 occurred between 20–37°C, which coincided with an irreversible minor transition in the thermal denaturation profile. The minor transition may be due to unfolding of a small region in core RAG1 that increases the propensity for nonspecific aggregation.

In previous studies it was shown that RAG2 could bind to both the 2-subunit MCR1:12-RSS and the 4-subunit MCR1:12-RSS complexes following incubation at 25°C 20. Since these complexes may be due to either one or two active RAG1 dimers bound to the 12-RSS, we asked if both of these complexes may be relevant to RAG1 function at 37°C. Here, we monitored the formation of the MCR1:RAG2:12-RSS complexes as a function of temperature. Upon incubation at 10 and 25°C, two complexes were formed (Figure 7A, lane 3), as has been previously observed under these conditions 20. The addition of GST-core RAG2 substantially increased the level of protein-DNA complexes formed (lane 1 versus lanes 2–7), which may be due either to RAG2-induced conformational changes in RAG1 or to additional contacts formed between RAG2 and the RSS.

Significantly, upon incubation at 37°C, the higher mobility MCR1:GST-core RAG2 complex with 12-RSS (complex 1) was maintained, while the level of complex 2 was significantly diminished (Figure 7A, lane 4). There appears to be a slightly lower binding affinity overall at 37°C given the increased amount of unbound 12-RSS (lane 4). However, the addition of HMGB1 also showed decreased formation of the slower mobility complex 2 at 37°C, but with no significant increase in unbound 12-RSS (Figure 7A, lanes 5–7). Overall, in both lanes 4 and 7, there was robust formation of complex 1, but with little evidence of complex 2. This is in contrast to lower temperatures where both complexes are observed, particularly after saturation of the oligonucleotide duplex (as in Figure 7A). Complexes 1 and 2 were both catalytically active, with nicking and hairpin formation detected by an in-gel DNA cleavage assay (Figure 7B). In this assay, the gel slices containing Complexes 1 and 2 (formed after incubation at 25°C as in Figure 7A, lane 3) were tested for DNA cleavage activity. Moreover, it should be noted that since DNA cleavage activity must be performed at 37°C, it is possible that a RAG1 dimer could dissociate from Complex 2, effectively forming Complex 1.

MCR1 purified from mammalian cells showed similar results, as MCR1 and GST-core RAG2 protein co-expressed in 293T cells primarily formed complex 1 (with only low levels of complex 2 evident upon overexposure of the gel) with incubation at 25°C (Figure 8A). Further, complex 1 remained the predominant complex formed even after saturation of the 12-RSS substrate with increasing concentrations of the co-expressed RAG proteins (Figure 8B). Overall, it appears that complex 1 is the most stable RAG1:RAG2:RSS complex observed under various conditions, which is consistent with the findings from the physicochemical characterizations that the dimeric form of MCR1 is the relevant oligomer that functions in V(D)J recombination. We have previously assigned RAG1:RAG2:RSS complexes 1 and 2 (in Figure 7A) as GST-core RAG2 bound to the 2-subunit and 4-subunit MCR1:12-RSS complexes, respectively 20. There is a possibility that complex 2 could be due to the binding of additional GST-core RAG2 subunits to the 2-subunit MCR1:12-RSS complex, which would correspond to the previously reported complexes "SC1" and "SC2" that contain 1 and 2 subunits of RAG2 bound to the RAG1 dimer:RSS complex, respectively 12. However, we do not believe that this is the case, but rather, under the conditions used here, only one physiologically relevant RAG1:RAG2:12-RSS complex can form (complex 1 in Figure 7A). This may be due to the use of a GST tag on RAG2 in our system, which could preclude formation of an 'SC1' complex containing a single subunit of RAG2 given that the GST moiety is known to dimerize 25.

To test the oligomeric state of GST-core RAG2, MALLS-SEC was performed on the purified protein (Figure 8C). In summary, the major oligomeric form of GST-core RAG2 detected was dimer (eluting in Peak 2), with possibly minor amounts of tetramer (Peak 1), but no obvious monomeric species present. Specifically, protein eluting in the predominant peak (V_e max of 12.7 mL; labeled Peak 2) was at a molar mass consistent with the dimer form. Material eluting in a shoulder of this peak (V_e from 13.5 to 14.5 mL) was also at a molar mass consistent with dimeric GST-core RAG2. (Material eluting at this position may be due to an alternate conformation of the protein or to interactions with the column matrix.) Lastly, a peak with V_e of 15.9 mL (Peak 3) yielded a molar mass of 59 kD. Peak 3 is likely due to the presence of a contaminating protein, which is evident on Coomassie blue-stained SDS polyacrylamide gels at ~55 kDa. The contaminating protein does not appear to be derived from either GST (at 26.3 kDa) or the core RAG2 portion (at 42.6 kDa) of the fusion protein. Moreover, protein eluting in Peak 3 could be not be detected by Western blotting against a monoclonal α-GST antibody (not shown). Finally, MCR1 combined with material eluting in Peak 2, but not Peak 3, resulted in the formation of complex 1 (as labeled in Figure 7A) by EMSA and formed hairpin products in DNA cleavage assays (not shown). As the protein concentration eluting in Peak 2 varied between 40–200 nM, which is within the range of protein concentration used in these studies (for example in Figure 7A), we surmise that GST-core RAG2 is in the dimeric state. These results are consistent with previous studies, which showed that dimeric GST unfolds in a two-state mechanism to denatured monomers, indicating that a monomeric folded form of GST is highly unstable 26.

The MBP-core RAG1 (MCR1) fusion protein was expressed in Escherichia coli and purified as described previously 18. MBP was expressed and purified as described 20. GST-core RAG2, either alone or co-expressed with MCR1, was transiently expressed and purified from 293T cells as previously reported 1627.

The concentrations of MCR1 and MBP purified from E. coli were determined from the absorbance at 280 nm using extinction coefficients of 129.5 and 66.5 mM^-1cm^-1, respectively. The concentrations of proteins purified from 293T cells were determined by quantitative Western blotting as previously described 27.

Multi-Angle Laser Light scattering with Size-Exclusion Chromatography (MALLS-SEC) was accomplished using a DAWN DSP laserphotometer coupled with an Optilab DSP interferometric refractometer (Wyatt Technology; Santa Barbara, CA) and combined in-line with a Superdex 200 gel filtration column (GE Healthcare; Piscataway, NJ).

Before each injection MCR1 was incubated at 10, 25, or 37°C as indicated, followed by chromatographic separation at room temperature. The column buffer for MCR1 MALLS-SEC experiments was comprised of 20 mM Tris, pH 8.0; 50 μM ZnCl₂; 5 mM β-mercaptoethanol; and 200 mM NaCl. For each experiment, between 0.06 and 0.13 mg of MCR1 was injected. Prior to SEC, GST-core RAG2 samples (at 0.01–0.03 mg) were kept at 4°C. The column buffer for GST-core RAG2 MALLS-SEC analysis was 25 mM Tris, pH 8.0; 150 mM KCl; and 2 mM dithiothreitol. Analysis of MALLS-SEC data was done as previously described 28.

The circular dichroism (CD) spectroscopy experiments were performed using a JASCO J715 spectropolarimeter with a PTC-348WI peltier temperature controller (Jasco, Corp.; Tokyo, Japan). The parameters for the spectra measurements are: 1.0 nm bandwidth, 1 nm resolution, 16 sec. response time, and 5 accumulations. The protein concentration in each sample was in the range of 1.7–2.0 μM and the buffer consisted of 20 mM Tris pH 8.0, 200 mM NaCl. Analysis of secondary structural content was done as previously described 2021.

Thermal denaturation experiments were performed in 10 mM Tris pH 8.0, 5 mM MgCl₂, and 100 mM NaCl using the same protein concentrations as the wavelength scan described above. The CD signal was detected at 222 nm, and the parameters were as follows: 10–80°C temperature range, 1.0 nm bandwidth, 0.2°C resolution, and a response time of 16 sec.

An oligonucleotide duplex containing a 12-RSS, commercially synthesized and PAGE-purified (Integrated DNA Technologies), was used as the DNA substrate in the EMSA experiments. The sequence of the 12-RSS containing oligonucleotide duplex has been previously reported 29. The 12-RSS substrate was labeled with ³²P at the 5'-end of the top strand using γ³²P-ATP and T4 polynucleotide kinase, and the double-stranded substrate was prepared by annealing the top strand with its respective complementary oligonucleotide.

EMSA was performed as described previously 20. The binding buffer contained 20 mM Tris, pH 8.0; 5 mM MgCl₂; 6% glycerol; 100 mM NaCl; and 2 mM dithiothreitol. In samples containing both RAG1 and RAG2, the binding buffer contained 2 mM CaCl₂ in place of MgCl₂ to prevent DNA cleavage. Prior to loading the samples into the gel, samples were incubated for 30 minutes at room temperature unless otherwise specified.

Prior to the addition of the RAG proteins, glutathione-sepharose 4B resin (GE Healthcare; Piscataway, NJ) was blocked with 1 mg/ml bovine serum albumin in interaction buffer (20 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 10% glycerol, 10μM ZnCl₂, and 5 mM β-mercaptoethanol) for 45 min at 4°C, followed by washing the resin three times with interaction buffer. Subsequently, MCR1 Fractions A or C (at 0.1 μM) or MBP (at 0.5 μM) was combined with GST-core RAG2 or GST (at 0.15 μM), and the proteins incubated with resin in the interaction buffer for 1 h at 4°C. Following three washes with the interaction buffer, the bound proteins were eluted from the resin with SDS loading buffer, resolved by SDS-PAGE (10% polyacrylamide gel), and electrotransferred to polyvinylidene difluoride membrane. Western analysis was done for both MBP and GST proteins as previously described 18.

The single RSS cleavage assay was performed in a total volume of 10 μl with a buffer composed of 10 mM Tris (pH 8.0), 2 mM dithiothreitol, 6% glycerol, 100 mM NaCl and 5 mM MnCl₂. MCR1 samples were incubated with GST-core RAG2 for 30 min at 4°C, followed by addition of 1 nM ³²P-labeled double-stranded wild type (WT) 12-RSS and incubation for 2 hrs at 37°C. The remaining procedure was as described previously 30.

The in-gel cleavage assay was done as previously described 31. In summary, samples were electrophoretically separated by EMSA as described above. After electrophoresis, the nondenaturing polyacrylamide gel was submerged in reaction buffer (10 mM Tris, pH8, 100 mM NaCl, and 5 mM MnCl₂) at 37°C for 1.5 hour. Next the gel was electro-transferred to DEAE membrane at 4°C overnight (50 V, 100 mA). To visualize the supershifted bands, the membrane was exposed to autoradiographic film overnight at 4°C, and washed in sterile wash buffer (50 mM NaCl, 10 mM Tris pH 7.5, 1 mM EDTA). Using the autoradiographic film as a guide, the supershifted bands were cut from the membrane and incubated with DNA elution buffer (1 M NaCl, 10 mM Tris pH 7.5, 1 mM EDTA) at 65°C for 30 min. The eluted DNA was filtered from the DEAE membrane using 0.22 μm Cellulose Acetate Spin X Centrifuge Tubes. After elution, the DNA was precipitated with ethanol and resuspended in 10 μL TE/Formamide loading dye (50% TE + 50% formamide loading dye). Samples were heated 2 min at 72°C, followed by separation on a 10% denaturing polyacrylamide gel. Bands corresponding to nick and hairpin cleavage products were visualized by autoradiographic analysis.