Title: Structural Basis for the Differential Regulation of DNA by the Methionine Repressor MetJ
Abstract: The Met regulon in Escherichia coli encodes several proteins responsible for the biosynthesis of methionine. Regulation of the expression of most of these proteins is governed by the methionine repressor protein MetJ and its co-repressor, the methionine derivative S-adenosylmethionine. Genes controlled by MetJ contain from two to five sequential copies of a homologous 8-bp sequence called the metbox. A crystal structure for one of the complexes, the repressor tetramer bound to two metboxes, has been reported (Somers, W. S., and S. E. Phillips (1992) Nature 359, 387-393), but little structural work on the larger assemblies has been done presumably because of the difficulties in crystallization and the variability in the number and sequences of metboxes for the various genes. Small angle neutron scattering was used to study complexes of MetJ and S-adenosylmethionine with double-stranded DNA containing two, three, and five metboxes. Our results demonstrate that the crystal structure of the two-metbox complex is not the native solution conformation of the complex. Instead, the system adopts a less compact conformation in which there is decreased interaction between the adjacent MetJ dimers. Models built of the higher order complexes from the scattering data show that the three-metbox complex is organized much like the two-metbox complex. However, the five-metbox complex differs significantly from the smaller complexes, providing much closer packing of the adjacent MetJ dimers and allowing additional contacts not available in the crystal structure. The results suggest that there is a structural basis for the differences observed in the regulatory effectiveness of MetJ for the various genes of the Met regulon. The Met regulon in Escherichia coli encodes several proteins responsible for the biosynthesis of methionine. Regulation of the expression of most of these proteins is governed by the methionine repressor protein MetJ and its co-repressor, the methionine derivative S-adenosylmethionine. Genes controlled by MetJ contain from two to five sequential copies of a homologous 8-bp sequence called the metbox. A crystal structure for one of the complexes, the repressor tetramer bound to two metboxes, has been reported (Somers, W. S., and S. E. Phillips (1992) Nature 359, 387-393), but little structural work on the larger assemblies has been done presumably because of the difficulties in crystallization and the variability in the number and sequences of metboxes for the various genes. Small angle neutron scattering was used to study complexes of MetJ and S-adenosylmethionine with double-stranded DNA containing two, three, and five metboxes. Our results demonstrate that the crystal structure of the two-metbox complex is not the native solution conformation of the complex. Instead, the system adopts a less compact conformation in which there is decreased interaction between the adjacent MetJ dimers. Models built of the higher order complexes from the scattering data show that the three-metbox complex is organized much like the two-metbox complex. However, the five-metbox complex differs significantly from the smaller complexes, providing much closer packing of the adjacent MetJ dimers and allowing additional contacts not available in the crystal structure. The results suggest that there is a structural basis for the differences observed in the regulatory effectiveness of MetJ for the various genes of the Met regulon. Transcriptional levels of several proteins in Escherichia coli that are involved in the biosynthesis and transport of methionine are under the common control of the met repressor protein MetJ (1Old I.G. Phillips S.E. Stockley P.G. Saint Girons I. Prog. Biophys. Mol. Biol. 1991; 56: 145-185Crossref PubMed Scopus (58) Google Scholar, 2Gal J. Szvetnik A. Schnell R. Kalman M. J. Bacteriol. 2002; 184: 4930-4932Crossref PubMed Scopus (40) Google Scholar). MetJ is a 12-kDa protein that is reported to form a homodimer in its native state (3Smith A.A. Greene R.C. Kirby T.W. Hindenach B.R. Proc. Natl. Acad. Sci. U. S. A. 1985; 82: 6104-6108Crossref PubMed Scopus (31) Google Scholar). Repressor activity results from MetJ binding to specific 8-bp DNA sequences called metboxes, located in the promoter regions of genes in the Met regulon. Although MetJ selectively binds metbox sequences alone, its affinity for metbox DNA is enhanced severalfold by its co-repressor, S-adenosylmethionine (SAM), 3The abbreviations used are: SAM, S-adenosylmethionine; SANS, small angle neutron scattering. an end product of methionine biosynthesis (4Saint-Girons I. Belfaiza J. Guillou Y. Perrin D. Guiso N. Barzu O. Cohen G.N. J. Biol. Chem. 1986; 261: 10936-10940Abstract Full Text PDF PubMed Google Scholar, 5Shoeman R. Redfield B. Coleman T. Greene R.C. Smith A.A. Brot N. Weissbach H. Proc. Natl. Acad. Sci. U. S. A. 1985; 82: 3601-3605Crossref PubMed Scopus (41) Google Scholar, 6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar, 7Hyre D.E. Spicer L.D. Biochemistry. 1995; 34: 3212-3221Crossref PubMed Scopus (49) Google Scholar). The MetJ protein dimer has been crystallized in the apo and holo forms and in complex with metbox DNA as a tetramer (8Rafferty J.B. Somers W.S. Saint-Girons I. Phillips S.E. Nature. 1989; 341: 705-710Crossref PubMed Scopus (166) Google Scholar, 9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar). The structure of the protein dimer remains essentially invariant in the three crystals and shows that MetJ is a member of the ribbon-helix-helix family of DNA-binding proteins that bind DNA through a two-stranded β-ribbon (9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar). At least 12 genes, scattered throughout the E. coli chromosome, are repressed by MetJ. Multiple MetJ dimers bind to operator sequences that contain two to five contiguous metboxes. A minimum of two tandem metboxes is required for efficient MetJ binding in vitro and repression of transcription in vivo (6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar). Adjacent MetJ dimers can interact with each other when they are bound to tandem metbox DNA sites, making assembly of higher order repressor complexes a cooperative process (7Hyre D.E. Spicer L.D. Biochemistry. 1995; 34: 3212-3221Crossref PubMed Scopus (49) Google Scholar, 9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar, 10He Y.Y. McNally T. Manfield I. Navratil O. Old I.G. Phillips S.E. Saint-Girons I. Stockley P.G. Nature. 1992; 359: 431-433Crossref PubMed Scopus (29) Google Scholar). There is a great deal of sequence variability among metboxes within each operator, but the shared consensus sequence is palindromic 5′-AGACGTCT-3′ (6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar, 11Belfaiza J. Parsot C. Martel A. de la Tour C.B. Margarita D. Cohen G.N. Saint-Girons I. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 867-871Crossref PubMed Scopus (137) Google Scholar), and in vitro selection has shown that this sequence is optimal for MetJ binding (12He Y.Y. Stockley P.G. Gold L. J. Mol. Biol. 1996; 255: 55-66Crossref PubMed Scopus (38) Google Scholar). The number of contiguous metboxes as well as the match of those sequences to the consensus are likely responsible for the differential levels of repression of the genes in the Met regulon (1Old I.G. Phillips S.E. Stockley P.G. Saint Girons I. Prog. Biophys. Mol. Biol. 1991; 56: 145-185Crossref PubMed Scopus (58) Google Scholar, 13Marincs F. Manfield I.W. Stead J.A. McDowall K.J. Stockley P.G. Biochem. J. 2006; 396: 227-234Crossref PubMed Scopus (33) Google Scholar). Despite the variety in length and sequence of the natural operators, the only high resolution structure determined for a complex of native MetJ with SAM and metbox DNA is the tetramer bound to DNA having two copies of the consensus sequence. Higher order complexes containing more than two metboxes have not been successfully crystallized. Here, we report the results of small angle neutron scattering (SANS) with contrast variation measurements of three MetJ complexes with SAM and DNA containing two, three, and five metboxes in solution. These results suggest that the solution conformation of the two-metbox complex differs significantly from the crystal structure. Furthermore, the higher order complexes, particularly the five-metbox complex, are not simple extensions of the arrangement seen in the smallest complex. Sample Preparation—MetJ was prepared by overexpression from an isopropyl-1-thio-β-d-galactopyranoside-inducible promoter in BL21(DE3) cells (Novagen, San Diego, CA). Cells were harvested and lysed with BugBuster (Novagen). MetJ was purified sequentially by ammonium sulfate precipitation, ion exchange with cellulose phosphate and DEAE columns, and gel filtration on a Sephacryl-200 column. The protein was determined to be pure by SDS-PAGE. MetJ was concentrated with a Vivaspin-20 concentrator (ISC Bioexpress, Kaysville, UT) and then dialyzed into buffer containing varying amounts of D2O for contrast variation studies. The buffer was 25 mm Tris, pH 7.0, 100 mm NaCl, 1 mm dithiothreitol, and 0.02% sodium azide. The DNA oligos used in this study were obtained from IDT (Coralville, IA). All sequences were flanked by GG-on the 5′-end and -CC on the 3′-end to "lock" the ends of the oligos and reduce the chances that the strands would separate. The two-metbox (5′-GGAGACGTCTAGACGTCTCC-3′) and three-metbox (5′-GGAGACGTCTAGACGTCTAGACGTCTCC-3′) constructs contained multiple copies of the consensus sequence and were self-complementary. To eliminate the possibility that a long palindromic sequence might fold back on itself and form a hairpin, the five-metbox construct used was the natural operator for the metF gene (5′-GGCTTCATCTTTACATCTGGACGTCTAAACGGATAGATGTGCCC-3′; bases that match the consensus sequence are underlined). Concentrated DNA stock solutions were heat annealed before use to favor the double-stranded form. The SAM used was the chloride salt (A7007; Sigma). A concentrated stock solution was made up in water and then adjusted to pH ∼5.0 with NaOH to ensure that the addition of SAM did not affect the pH of the samples. Samples for SANS were made by using MetJ that had been dialyzed into the required buffers and adding small volumes of DNA and SAM from concentrated stocks. All samples contained 5 mg/ml MetJ and stoichiometric amounts of DNA and SAM, i.e. one metbox and two SAM molecules per dimer. Gel Shift Assay—Samples were prepared containing 0.4 mg/ml MetJ with stoichiometric amounts of SAM and various DNA oligos in the same buffer as that used for SANS. In addition to the oligos used for the neutron scattering experiments, we examined oligos containing four and five copies of the consensus metbox as well as the natural metA operator that contains four metboxes (5′-GGAGCTATCTGGATGTCTAAACGTATAAGCGTATCC-3′; bases that match the consensus sequence are underlined). After the samples were allowed to mix for 15-20 min at room temperature, an equal volume of 2× loading dye (20% sucrose, 0.02% xylene cyanol in 10 mm Tris, 1 mm EDTA) was added before loading on the gel. Electrophoresis was performed at room temperature on a 9% non-denaturing acrylamide gel that was made and run in 1× TBE buffer (90 mm Tris borate, 2 mm EDTA). The gel was stained with SYBR Green nucleic acid stain (Invitrogen). SANS Measurements—SANS data were collected at the National Institute of Standards and Technology Center for Neutron Research (Gaithersburg, MD) using the NG-7 30-m SANS instrument (14Glinka C.J. Barker J.G. Hammouda B. Krueger S. Moyer J.J. Orts W.J. J. Appl. Crystallogr. 1998; 31: 430-445Crossref Scopus (610) Google Scholar). All experiments were performed at room temperature. A neutron wavelength of 5 Å was used with a wavelength spread Δλ/λ of 0.22 to provide the maximum flux on the sample. Sample and background intensities were collected at sample-to-detector distances of 1.5 and 4.6 m to obtain data over the q-range necessary for structural modeling. Data reduction to one-dimensional scattered intensity profiles, I(q) versus q, followed standard procedures to correct for detector sensitivity and sample background (14Glinka C.J. Barker J.G. Hammouda B. Krueger S. Moyer J.J. Orts W.J. J. Appl. Crystallogr. 1998; 31: 430-445Crossref Scopus (610) Google Scholar). An empty beam measurement was used to calibrate the measurements into absolute units (1/cm) (15Wignall G.D. Bates F.S. J. Appl. Crystallogr. 1987; 20: 28-40Crossref Scopus (671) Google Scholar). The reduced intensities from the two detector distances were merged using routines included with the data reduction software provided by the National Institute of Standards and Technology. Samples of the two-metbox complex were prepared in 10 and 71% D2O by mixing stock solutions of the subunits. Samples of the three-metbox and five-metbox complexes were prepared in the same way but in 10, 32, 53, and 71% D2O. The samples and corresponding solvent blanks were measured in 1-mm path length cells. The precise D2O content of each sample was determined by comparing the measured neutron absorbance, μ, to that of pure H2O and D2O. The neutron absorbance is obtained from the measured sample transmission using the relationship μ = ln(T)/d, where T is the transmission, and d is the sample path length. The neutron absorbance of a sample varies linearly with the hydrogen content, which is dominated by the buffer solution. The solvent deuteration fraction of the sample, P, can be determined using the relation P = 1-((μsample - μD2O)/(μH2O-μD2O)). The D2O content of each sample determined in this manner was used throughout the analysis and modeling of the data. Small Angle Scattering Data Analysis—The small angle scattering intensity profile of monodisperse, identical particles in solution is shown in Equation 1 I(q)=|〈∫v(ρ(r→)-ρs)e-iq→⋅r→d3r〉|2(Eq. 1) where ρ(→r) is the scattering length density of the particle as a function of position →r within the particle volume V; ρs is the average scattering length density of the solvent; and →q is the momentum transfer, having the magnitude q = 4πsin(θ)/λ, where 2θ is the scattering angle and λ is the wavelength. The integration in Equation 1 is averaged over all conformations and orientations of particles in the incident beam and includes the time- and ensemble-averaged information of all of the particles in the sampled volume. In addition to determining the radius of gyration, Rg, according to Guinier (16Guinier A. Ann. Phys. (Paris). 1939; 12: 161-237Google Scholar), data were analyzed for the distance distribution function, P(r), between scattering centers within an object. The scattered intensity I(q) is related to P(r) as shown in Equation 2. P(r)=12π2∫0∞dq⋅(qr)⋅I(q)sin(qr)(Eq. 2) The indirect Fourier transform algorithm originally described by Moore (17Moore P.B. J. Appl. Crystallogr. 1980; 13: 168-175Crossref Google Scholar) was used to determine P(r) from measured intensity profiles. The boundary conditions P(r)/r = 0at r = 0 and the maximum linear dimension, dmax, are applied to ensure proper behavior. In addition to providing dmax, P(r) fitting also provides a second measure of Rg, which is the second moment of P(r). Structural Models from SANS Data—Modeling of the complex was based on the high resolution crystal structure of a two-metbox complex (Protein Data Bank code 1CMA) (9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar) with hydrogen atoms added using MOLPROBITY (18Davis I.W. Murray L.W. Richardson J.S. Richardson D.C. Nucleic Acids Res. 2004; 32: W615-W619Crossref PubMed Scopus (820) Google Scholar). The structure of a single MetJ dimer bound to the metbox sequence from the crystal structure was used as the basic unit for building models, which was maintained as a rigid body when constructing models of the higher order assemblies. This approach maintained the appropriate distortion of the DNA induced by the binding of the MetJ dimer. The two-, three-, and five-metbox complexes were constructed by sequential assembly of the basic unit. To facilitate the construction, a coordinate system was defined using the A3/T18 base pair of the first metbox in the crystal structure and the standard reference frame for nucleic acids (19Olson W.K. Bansal M. Burley S.K. Dickerson R.E. Gerstein M. Harvey S.C. Heinemann U. Lu X.J. Neidle S. Shakked Z. Sklenar H. Suzuki M. Tung C.S. Westhof E. Wolberger C. Berman H.M. J. Mol. Biol. 2001; 313: 229-237Crossref PubMed Scopus (476) Google Scholar). To assemble the two-, three-, and five-metbox complexes, the translation and rotation applied to the next copy of the basic unit is such that the A3/T18 base pair of the basic unit is properly placed in the space of the standard reference frame of the A11/T10 base pair of the previous basic unit. To test for structural deviations of the two-, three-, and five-metbox complexes from the organization implied by the crystal structure (9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar), the roll angle, ω, of the basic unit about the vector from C′1,T18 to C′1,A3 of the metbox was varied. An examination of the high resolution structure suggests that this particular rotation axis affords the greatest degree of flexibility for the modeling and avoids interference between adjacent MetJ dimers. In the models of the three- and five-metbox complexes, the roll angles between all of the basic units were assumed to be identical, thereby ensuring that the interaction between subsequent MetJ dimers remains identical and minimizing the number of free parameters. The modeling tested ω values every 5° from -90° to +90°, where 0° is defined by the angle present in the crystal structure. Negative values of ω increase the separation of adjacent MetJ dimers bound to the DNA, whereas positive values bring the centers of mass of adjacent dimers closer together. The program ORNL_SAS 4W. T. Heller and E. Tjioe, manuscript in preparation. was used to calculate the intensity profiles from the generated model structures and evaluate the quality of the fit against the experimental data. ORNL_SAS simultaneously calculates multiple small angle scattering intensity profiles from high resolution structures using a Monte Carlo approach and is capable of comparing the results against several sets of input data from x-ray and neutron scattering experiments. If the series of input data sets are Ij(q) and the model intensity profiles are Im,j(q), the quality of fit is evaluated using the reduced χ2 parameter as shown in Equation 3. χ2=1(∑jNj,pts)−Nf∑j∑Nj,pts(Ij(q)−Im,j(q))2σj(q)2(Eq. 3) Nj,pts is the number of data points modeled against in the measured intensity Ij(q). σj(q) is the experimental uncertainty in the measured intensity Ij(q). Nf is the number of parameters used for the scaling of the model profiles to the data. There are two scaling parameters per data set, being the overall scale factor and a base line, which reflects difficulties in determining the exact base line of SANS data when the solvents contain a large amount of hydrogen and is generally not larger than the uncertainty in the measured forward scatter, I(0), of a properly background-subtracted data set. SANS data were collected on solution samples of two-, three-, and five-metbox complexes. To verify the formation of the MetJ-DNA-SAM complexes, we performed a gel shift assay, shown in Fig. 1. These results show that MetJ forms only one stable complex with each of the DNA oligos, because, in the presence of MetJ, there is only a single band that is shifted in mobility relative to the DNA alone. The size of the complexes cannot be determined from this type of experiment, but by examining the full series of complexes the trend in the shift of the bands for the complexes clearly indicates that longer DNA fragments bind more MetJ than the shorter ones. In addition, the complexes formed with the four-consensus metbox sequence and metA run identically, indicating that the same number of MetJ dimers are bound to both. This is also true for the five-consensus metbox sequence and metF, justifying our assumption that the five-metbox complex used for SANS binds five MetJ dimers. One other observation to note is that the long palindromic sequences show signs of hairpin formation, indicated by an additional fast-running band in the lanes without MetJ. It was to avoid this hairpin potential that we decided to use a natural sequence for the five-metbox construct. The reduced SANS intensity profiles for the two-, three-, and five-metbox complexes are shown in Fig. 2, A, B, and C, respectively. The differential spacing between the data points (on the left and right side) of each profile results from the two detector distances used to provide the desired q-range. The P(r) for the three complexes extracted from the data collected in 10 and 71% D2O, which have the strongest signals, are shown in Fig. 3. The trends in Rg and dmax with increasing numbers of MetJ dimers can be seen in Fig. 4. The trend in Rg is not simple to interpret, but it is clear from the trend in dmax that the length of the complex, going from 80-90 to 120 Å, is not a linear function of the number of base pairs of DNA. The results suggest that the binding of five MetJ dimers has produced a compaction of the complex beyond any structural change induced in the two- and three-metbox complexes. The overall lengths for all three complexes are shorter than expected for idealized B-form DNA.FIGURE 3P(r) curves derived from the 10% (A) and 71% (B)D2O data sets using the Moore algorithm (17Moore P.B. J. Appl. Crystallogr. 1980; 13: 168-175Crossref Google Scholar). The symbols correspond to the two-metbox (▪), three-metbox (○), and five-metbox (▴) complexes. The curves have been given an arbitrary scaling to make it easier to compare the peak positions and shapes of the curves.View Large Image Figure ViewerDownload Hi-res image Download (PPT)FIGURE 4Plots of Rg (A) and dmax (B) derived from the 10% (▪) and 71% (○)D2O data sets showing that, as the length of the DNA increases, the increase in Rg and dmax is not linear.View Large Image Figure ViewerDownload Hi-res image Download (PPT) A closer inspection of the P(r) curves gives more insight into the structures of the complexes. The easiest to interpret are those from the samples in 71% D2O (Fig. 3B), which are dominated by the scattering of the MetJ in the complex because the scattering length density of DNA is very nearly identical to the solution scattering length density. In contrast, at 10% D2O both the MetJ and DNA scatter strongly, with the DNA scattering ∼3.5 times stronger than MetJ on a per unit volume basis. The larger relative total volume of the MetJ means that it contributes more to the scattered intensity than the DNA, but the DNA contribution is not negligible, as it is in 71% D2O. The P(r) curve of the two-metbox complex shown in Fig. 3B is consistent with a compact structure, such as would be expected based on the crystal structure. The five-metbox complex P(r) is typical of a rod-like structure, as might be expected for an assembly of the proteins on a longer piece of DNA. The P(r) of the three-metbox complex is the most interesting. In addition to the main peak at 27 Å, there is a shoulder near 50 Å. Such a shoulder suggests some distinction between groups of structural elements that leads to a secondary peak in P(r), but the effect is subtle. The modeling of the complexes provides an effective means of visualizing the conformational changes that occur in the higher order complexes. A plot of χ2 as a function of the rotation angle ω is shown in Fig. 5. The two-metbox- and three-metbox complexes essentially have a plateau centered near ω = -60°, but the minimum is relatively shallow and there is considerable noise. The minimum for the 5-metbox complex is the most clearly defined of the three complexes measured and best fits the data at an ω of +30°. The model intensity profiles for the best-fit models are plotted in Fig. 2 with the data. The fit is excellent in all cases. The model profiles that appear to end before the end of the plotted range reflect the small shift in the base line allowed for the scaling of the intensity profiles to account for possible difficulties in the data subtraction. It is important to point out that the original crystal structure, after having hydrogens appropriately added using MOLPROBITY (18Davis I.W. Murray L.W. Richardson J.S. Richardson D.C. Nucleic Acids Res. 2004; 32: W615-W619Crossref PubMed Scopus (820) Google Scholar) and removing water molecules found in the crystal structure, yields a χ2 of 2.29 when tested against the SANS data of the two-metbox complex. This value is significantly higher than the best value found in the modeling (χ2 = 1.05), indicating that the model structure fits the data significantly better than the crystal structure. The best-fit models for the three complexes are shown in Fig. 6 and provide a way to understand the nature of the sign of the rotation angle ω. In the models, the structure of a single metbox and the associated MetJ dimer with SAM bound is considered fixed. As ω changes, the extent of the interaction between the consecutive dimers changes. A negative value of ω increases the separation between the centers of mass of the MetJ dimers, whereas a positive value of ω causes the separation between the MetJ dimers to decrease, enhancing the potential for interaction between the dimers. The most obvious feature of the structures of the two-metbox and three-metbox complexes is that the amount of contact along Helix A of the adjacent MetJ dimers is very minimal, being reduced to only a few interactions near the ends of the helix. If the structure were to be continued indefinitely, the resulting superhelical arrangement of the protein around the DNA would be very wide with only minimal protein-protein contacts. In contrast, the structure of the five-metbox complex is such that there is a great deal of interaction between the adjacent MetJ dimers and the points of contact are no longer limited to Helix A. The nature of the superhelical packing of the structure can be seen clearly in Fig. 6D and is significantly more compact than if the structures of the smaller (two- and three-metbox) complexes were to be extended. Fig. 7 shows a close-up of the interface between adjacent dimers highlighting the main difference between the models of the smaller complexes and that of the five-metbox complex.FIGURE 7Close-up of the interface of adjacent MetJ dimers from the crystal structure (A), the model of the two-metbox complex (B), and one of the interfaces from the model of the five-metbox complex (C). Relative to the crystal structure, the model of the two-metbox complex shows that the dimers are farther apart and the number of contacts across Helix A have been reduced. For the model of the five-metbox complex, the dimers have moved closer to each other and contacts are not limited to Helix A.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The results of the SANS studies with contrast variation presented here provide new insight into the cooperative regulation of genes in the Met regulon by the methionine repressor MetJ and its cofactor SAM. One important finding is that the calculated scattering curve based on coordinates from the crystal structure is not consistent with the SANS data collected from the two-metbox complex in solution. The models derived from the SANS data indicate that the centers of mass of the adjacent MetJ dimers are further apart than they are in the crystal structure. The DNA conformation also differs significantly in that there is a significant bend between adjacent metboxes as can be seen in Fig. 6. As a result of the opening of the structure, the interaction between the adjacent MetJ dimers through Helix A is drastically reduced as shown in Figs. 6 and 7. Contacts along the A helix are thought to be involved in the cooperativity observed for binding multiple MetJ dimers to adjacent metboxes (7Hyre D.E. Spicer L.D. Biochemistry. 1995; 34: 3212-3221Crossref PubMed Scopus (49) Google Scholar, 9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar, 10He Y.Y. McNally T. Manfield I. Navratil O. Old I.G. Phillips S.E. Saint-Girons I. Stockley P.G. Nature. 1992; 359: 431-433Crossref PubMed Scopus (29) Google Scholar) and have been postulated to explain why at least two metboxes are required for stable protein binding to DNA. In our SANS-based model for the tetramer assembly on two consecutive metbox sequences in solution, however, there is reduced contact at the dimer-dimer interface. The addition of a third metbox, which allows a third MetJ dimer to bind the complex, does not significantly change the relationship between the MetJ dimers from that seen in the two-metbox complex. The other surprising result is that the five-metbox complex appears to have a very different organization of MetJ dimers. In addition to contacts through the previously identified Helix A (residues 30-45), the model suggests that there are additional interactions through residues 17-20 in the loop that precedes the β strand. The most straightforward a priori assumption would be that increasing the number of metboxes should have a simple additive effect on the strength of MetJ binding, so that each additional dimer that can bind adds some fixed amount to the overall affinity. Our model of the five-metbox complex suggests, however, that additional factors associated with structural features of the assembly must be considered. Intriguingly, an earlier footprinting experiment with DNA having mutations in the metF operator supports this interpretation as well (20Davidson B.E. Saint Girons I. Mol. Microbiol. 1989; 3: 1639-1648Crossref PubMed Scopus (26) Google Scholar). DNase footprinting showed that of the five metboxes in the metF operator, four were preferentially bound initially whereas binding to a fifth (repeat 5) required higher concentrations of MetJ. Moreover, deletion of or mutations in repeat 5 could eliminate repression in vivo. This result was interpreted to mean that the additional cooperativity gained from binding the last MetJ dimer to repeat 5 allowed the complex to meet some "threshold" for tight binding (20Davidson B.E. Saint Girons I. Mol. Microbiol. 1989; 3: 1639-1648Crossref PubMed Scopus (26) Google Scholar). In light of our SANS results, we further postulate that MetJ binding to repeat 5, and possibly 4, modifies the structure of the assembly, creating a more compact complex that is likely bound more tightly. With the data presented here, we can only say that in solution the tight binding mode requires more than three dimers. It should also be noted that the DNA sequence used in the footprinting experiment as well as our SANS experiment was not five copies of the consensus metbox but rather the native metF operator sequence. It is possible, but unlikely, that the natural sequence variations alter the interaction between adjacent MetJ dimers in the complex. There is a long history of using the consensus sequence as well as natural operators to study MetJ binding (6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar, 7Hyre D.E. Spicer L.D. Biochemistry. 1995; 34: 3212-3221Crossref PubMed Scopus (49) Google Scholar, 9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar, 21Wild C.M. McNally T. Phillips S.E. Stockley P.G. Mol. Microbiol. 1996; 21: 1125-1135Crossref PubMed Scopus (13) Google Scholar, 22Lawrenson I.D. Stockley P.G. FEBS Lett. 2004; 564: 136-142Crossref PubMed Scopus (11) Google Scholar, 23He Y.Y. Garvie C.W. Elworthy S. Manfield I.W. McNally T. Lawrenson I.D. Phillips S.E. Stockley P.G. J. Mol. Biol. 2002; 320: 39-53Crossref PubMed Scopus (13) Google Scholar). Further evidence that MetJ may bind DNA differently depending on the number of metboxes comes from studies of the metC operator, the only one in E. coli that contains the minimum of two metboxes (11Belfaiza J. Parsot C. Martel A. de la Tour C.B. Margarita D. Cohen G.N. Saint-Girons I. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 867-871Crossref PubMed Scopus (137) Google Scholar). DNase footprinting experiments have shown that MetJ covers a 32-bp region centered around the two consensus metbox sequences, i.e. MetJ covers a region the size of four 8-bp metboxes (6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar, 21Wild C.M. McNally T. Phillips S.E. Stockley P.G. Mol. Microbiol. 1996; 21: 1125-1135Crossref PubMed Scopus (13) Google Scholar). Thus, it is possible that in vivo repression is only achieved by the binding of four to five MetJ dimers, and that this tight binding mode is different from the weak binding of two to three dimers. While most of the in vitro work has been done with synthetic minimal operators containing only two metboxes, sometimes embedded within longer DNA fragments, the metboxes used are always composed of repeats of the consensus sequence (6Phillips S.E. Manfield I. Parsons I. Davidson B.E. Rafferty J.B. Somers W.S. Margarita D. Cohen G.N. Saint-Girons I. Stockley P.G. Nature. 1989; 341: 711-715Crossref PubMed Scopus (96) Google Scholar, 7Hyre D.E. Spicer L.D. Biochemistry. 1995; 34: 3212-3221Crossref PubMed Scopus (49) Google Scholar, 9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar, 21Wild C.M. McNally T. Phillips S.E. Stockley P.G. Mol. Microbiol. 1996; 21: 1125-1135Crossref PubMed Scopus (13) Google Scholar, 22Lawrenson I.D. Stockley P.G. FEBS Lett. 2004; 564: 136-142Crossref PubMed Scopus (11) Google Scholar, 23He Y.Y. Garvie C.W. Elworthy S. Manfield I.W. McNally T. Lawrenson I.D. Phillips S.E. Stockley P.G. J. Mol. Biol. 2002; 320: 39-53Crossref PubMed Scopus (13) Google Scholar). Consequently, MetJ has an unnaturally high affinity for them. In E. coli, only one metbox, part of the metD operator, consists of the actual consensus sequence (2Gal J. Szvetnik A. Schnell R. Kalman M. J. Bacteriol. 2002; 184: 4930-4932Crossref PubMed Scopus (40) Google Scholar), while in the closely related Salmonella typhimurium there is also only one occurrence, part of the metER operator (24Wu W.F. Urbanowski M.L. Stauffer G.V. FEMS Microbiol. Lett. 1993; 108: 145-150Crossref PubMed Scopus (9) Google Scholar). The strength of binding is also affected significantly by the co-repressor, SAM, but all of the SANS experiments reported here were carried out in the presence of SAM and thus provide no information on the structural consequence of SAM binding. The different binding modes could be a result of many factors, including enhanced compaction of the DNA, which could lead to more extensive DNA-protein and protein-protein interactions, as well as conformational changes within MetJ. The models built to test against the SANS data are intended to be simple and assume that the structure of the MetJ dimer in solution is unchanged from the high resolution crystallographic structure in the tetramer complex (9Somers W.S. Phillips S.E. Nature. 1992; 359: 387-393Crossref PubMed Scopus (284) Google Scholar), which is reported to be essentially the same as that of the crystallized dimer alone (8Rafferty J.B. Somers W.S. Saint-Girons I. Phillips S.E. Nature. 1989; 341: 705-710Crossref PubMed Scopus (166) Google Scholar). SANS, being an inherently lower resolution technique than x-ray diffraction, is often well suited for determining whether the solution structure of a protein is identical to a crystal structure, such as when there are well defined domains connected by flexible linkers. In the case of the MetJ-DNA-SAM complexes, the system is significantly more complicated. The use of additional variables to model changes in the structures of the MetJ subunits would decrease confidence in the final models as the size and complexity of the search space increases geometrically with the number of variables used. The application of ab initio modeling (25Heller W.T. Krueger J.K. Trewhella J. Biochemistry. 2003; 42: 10579-10588Crossref PubMed Scopus (42) Google Scholar), in which no assumptions are made regarding the structure, to the data presented here is also unlikely to resolve the issue, given the inherent uncertainty of such structural reconstructions. The modeling approach chosen is a compromise to help visualize the solution conformations of the complexes with the minimum number of variables. Because this approach does not take into account any conformational changes that could occur in the MetJ dimers, the models do not preclude cooperativity in the assembly of the complex. In conclusion, we have used SANS with contrast variation to study the solution structures of two-, three-, and five-metbox complexes. Model building has demonstrated that the solution conformation adopted is a function of the number of tandem metbox sequences. Importantly, the SANS data collected for the two-metbox complex demonstrate that the crystal structure is not consistent with the solution conformation of the complex. In the shorter complexes measured, the centers of mass of the adjacent MetJ dimers are further apart than found in the crystal structure, producing a superhelical assembly that is relatively open. In contrast, the SANS data of the largest complex studied, the five-metbox complex, are consistent with a structure in which the adjacent MetJ dimers are in much closer proximity, producing a superhelical structure that is quite compact. In this case, SANS with structural modeling gives unique insight into the assembly of this large (150 kDa), important native repressor complex in solution. The results provide evidence for a structural mechanism controlling differences observed in the transcription regulation provided by MetJ and suggest the possibility that, in solution, MetJ is able to bind to DNA in at least two different ways. We acknowledge the support of the National Institute of Standards and Technology, U. S. Dept. of Commerce, in providing the neutron research facilities used in this work.