Solution structure of the DNA-binding domain of NtrC with three alanine substitutions1

Article No. jmbi.1999.3140 available online at http://www.idealibrary.com on

J. Mol. Biol. (1999) 292, 1095±1110

Solution Structure of the DNA-binding Domain of NtrC with Three Alanine Substitutions Jeffrey G. Pelton1, Sydney Kustu2 and David E. Wemmer1,3* 1

Physical Biosciences Division Lawrence Berkeley National Laboratory, 1 Cyclotron Road Berkeley, CA, 94710, USA 2

Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720-1460, USA 3

Department of Chemistry University of California Berkeley, CA, 94720-1460, USA

The structure of the 20 kDa C-terminal DNA-binding domain of NtrC from Salmonella typhimurium (residues Asp380-Glu469) with alanine replacing Arg456, Asn457, and Arg461, was determined by NMR spectroscopy. NtrC is a homodimeric enhancer-binding protein that activates the transcription of genes whose products are required for nitrogen metabolism. The 91-residue C-terminal domain contains the determinants necessary for dimerization and DNA-binding of the full length protein. The mutant protein does not bind to DNA but retains many characteristics of the wild-type protein, and the mutant domain expresses at high yield (20 mg/l) in minimal medium. Three-dimensional 1H/13C/15N triple-resonance, 1H-13C-13C-1H correlation and 15N-separated nuclear Overhauser effect (NOE) spectroscopy experiments were used to make backbone and side-chain 1H, 15N, and 13 C assignments. The structures were calculated using a total of 1580 intra and inter-monomer distance and hydrogen bond restraints (88 hydrogen bonds; 44 hydrogen bond restraints), and 88 f dihedral restraints for residues Asp400 through Glu469 in both monomers. A total of 54 ambiguous restraints (intra or inter-monomer) involving residues close to the 2-fold symmetry axis were also included. Each monomer consists of four helical segments. Helices A (Trp402Leu414) and B (Leu421-His440) join with those of another monomer to form an antiparallel four-helix bundle. Helices C (Gln446-Leu451) and D (Ala456-Met468) of each monomer adopt a classic helix-turn-helix DNAbinding fold at either end of the protein. The backbone rms deviation for Ê . Structural differences the 28 best of 40 starting structures is 0.6( 0.2) A between the C-terminal domain of NtrC and the homologous Factor for Inversion Stimulation are discussed. # 1999 Academic Press

*Corresponding author

Keywords: helix-turn-helix; FIS; four-helix bundle; NMR spectroscopy; protein structure

Introduction Nitrogen regulatory protein C (NtrC) is a bacterial enhancer-binding protein that activates the transcription of genes encoding enzymes required for nitrogen metabolism. One of the best studied Abbreviations used: FIS, Factor for Inversion Stimulation; HCCH, 1H-13C-13C-1H correlation; HMQC, heteronuclear multiple-quantum coherence; HSQC, heteronuclear single-quantum coherence; NtrC, nitrogen regulatory protein C; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; TOCSY, total correlation spectroscopy; TPPI, time-proportional phase incrementation; rmsd, root-mean-square deviation. E-mail address of the corresponding author: [email protected] 0022-2836/99/401095±16 $30.00/0

examples is its activation of the gene glnA, which encodes the enzyme glutamine synthetase. When combined nitrogen is limiting, phosphorylation of Asp54 of the N-terminal or receiver domain of NtrC is increased. Phosphorylation initiates oligomerization of NtrC at the glnA enhancer and phosphorylated NtrC oligomers contact the s54-RNA polymerase holoenzyme at the promotor by means of DNA-looping. Transcription is activated by the interaction of the central or motor domain of NtrC with the s54-RNA polymerase holoenzyme, hydrolysis of ATP or GTP, and conversion of the polymerase-promotor complex from the closed to the open form (for reviews, see Kustu et al., 1991; Porter et al., 1995; Reitzer & Magasanik, 1987). NtrC is composed of three domains (Drummond et al., 1986; North et al., 1993). The 124 residue # 1999 Academic Press

1096 N-terminal domain is homologous to receiver domains of other response regulator proteins in two-component signal transduction systems (Ninfa et al., 1987; Nixon et al., 1986). The three-dimensional structure of the unphosphorylated form of the N-terminal domain of NtrC was recently solved by NMR spectroscopy (Volkman et al., 1995). It is similar to that of CheY, a much studied receiver domain that controls ¯agellar rotation in bacteria (Ravid et al., 1986). We have also examined structural and functional changes associated with constitutively active mutant forms of the N-terminal domain of NtrC (Nohaile et al., 1997), which can generate output without being phosphorylated. The 240 residue central domain of NtrC is homologous to a domain found in all activators of the s54-RNA polymerase holoenzyme (Kustu et al., 1989; Thony & Hennecke, 1989). Although the three-dimensional structure of the domain is not known, a model based on the structure of the purine nucleotide-binding protein EF-Tu was recently proposed (Osuna et al., 1997). Early sequence comparisons with the proteins NifA (nitrogen ®xation protein A) and DctD (dicarboxylate transport protein D) suggested that the C-terminal domain of NtrC contained a helix-turnhelix (HTH) DNA binding motif (Drummond et al., 1986). Mutations in the putative HTH abolished DNA-binding, supporting the hypothesis (Contreras & Drummond, 1988; North et al., 1993; North & Kustu, 1997; Porter et al., 1995). Experiments on a 90 residue C-terminal fragment showed that the domain forms a stable dimer in solution (Porter et al., 1993). The af®nity of this fragment for DNA (Porter et al., 1993) and rates of monomer interchange (Klose et al., 1994) are approximately the same as for the full-length protein, indicating that the C-terminal fragment contains the determinants necessary for both DNAbinding and dimerization of full-length NtrC. Subsequent sequence comparisons revealed that the C-terminal domain of NtrC is homologous to the Factor for Inversion Stimulation (FIS) (Johnson et al., 1988). FIS consists of a single domain and is involved in many cellular processes, including sitespeci®c DNA inversion, excisive recombination of bacteriophage lambda, and regulation of gene expression (reviewed by Finkle & Johnson, 1992). It has a highly degenerate consensus binding site (Pan et al., 1996a,b), and bends DNA to different extents depending on sequence (Thompson & Landy, 1998). The structure of FIS and several mutant forms have been determined by X-ray crystallography (Kostrewa et al., 1992; Safo et al., 1997; Yuan et al., 1991, 1994). The ®rst 26 residues were disordered in all of the structures, except one. In that case the segment Val10 to Gln24 folded into a b-hairpin (Safo et al., 1997). Models for the binding of FIS to DNA have been proposed (Kostrewa et al., 1992; Yuan et al., 1991), but no structural studies of complexes have been reported.

Solution Structure of NtrCC-term(3Ala)

Recently, phylogenetic evidence has been used to show that FIS originated from the C-terminal domain of an ancestral a-proteobacterial NtrC (Morett & Bork, 1998). If the 26 residues that are disordered in the FIS X-ray structures are not included, FIS shows an average of 50 % sequence identity with the C-terminal domain of NtrC from a-proteobacteria and 33 % identity with that from g-proteobacteria. The high sequence identity suggests that the C-terminal fragment of NtrC and FIS would have similar structures. Our long-term goal is to study the structure of the C-terminal domain of NtrC from Salmonella typhimurium and its complexes with DNA by NMR spectroscopy. The high sequence similarity between the C-terminal domain of NtrC and FIS, combined with differences in their DNA-binding behavior, make a comparative study of these systems ideal for understanding how helix-turn-helix motifs interact with their DNA-binding sites. We successfully expressed a 91 residue C-terminal fragment of wild-type NtrC on Luria-Bertani (LB) medium, albeit with low yields (5 to 8 mg/liter). With a molecular weight for the dimer of 20 kDa, isotope enrichment was essential. However, we were unsuccessful at expressing the fragment in minimal medium containing ammonium sulfate as the sole nitrogen source. We were also unable to express the protein in minimal medium supplemented with glutamine, or medium containing an algal protein lysate (Martec-9, Martec Bioscience, Columbia, MD). The other possibility, of trying to express the protein in a rich labeled medium such as celtone (Martec), was rejected due to the poor observed yields in LB. Assuming that DNA-binding by the C-terminal domain was inhibitory to growth, we turned ®rst to a stable, soluble mutant form of this fragment that was essentially unable to bind DNA. We successfully expressed this mutant form of the C-terminal domain of NtrC, denoted NtrCC-term (3Ala), in which alanine replaced Arg456, Asn457, and Arg461 of the putative recognition helix of the wild-type protein (Figure 1) (Porter et al., 1993). Extensive biochemical experiments have shown that the three alanine substitutions in NtrC described above result in at least a 5000-fold reduction in DNA-binding af®nity (North & Kustu, 1997). However, in the full-length form, the mutant protein retains many characteristics of the wildtype protein. It forms a stable homodimer that can be phosphorylated by its cognate autokinase NtrB. The ATPase activity of the mutant is comparable to that of the wild-type protein in the absence of DNA, and most importantly, the mutant protein was able to stimulate transcription of the glnA gene when present at high concentrations (100 nM) (North & Kustu, 1997). Thus, the biochemical data suggest that the alanine substitutions affect only the DNA-binding capacity of the protein, and that NtrCC-term(3Ala) would serve as a suitable alternative for the study of the structure of the DNA-binding domain.

Solution Structure of NtrCC-term(3Ala)

Figure 1. Amino acid sequences of NtrCC-term(3Ala) derived from the g-proteobacterium S. typhimurium (Porter et al., 1993) with substitution of alanine for Arg456, Asn457, and Arg461 (top), and FIS derived from E. coli (Johnson et al., 1988) (bottom). The secondary structure of NtrCC-term(3Ala) is shown above the sequence. Residue numbers for NtrCC-term(3Ala) and FIS are shown above and below the sequence, respectively.

Herein we report the structure of the triple alanine substituted C-terminal domain of NtrC from the g-proteobacterial species S. typhimurium using mutidimensional heteronuclear NMR methods. The structure is interpreted in light of recent mutagenesis data and is compared with the previously reported structures of FIS. Several structural differences that relate to sequence differences between the C-terminal domain of NtrC from g-proteobacterial species and both the C-terminal domain of NtrC from a-proteobacterial species and the homologous protein FIS are discussed.

Results NMR signal assignments Strong NOEs between adjacent amide protons (dNN NOEs) characteristic of helical secondary structure (WuÈthrich, 1986) were observed in a 3D NOESY-FHSQC spectrum for virtually all residues between Ser401 and Glu469. The NOE data, along with Cai /Cai ÿ 1 linkages observed in a 3D HNCA spectrum formed the basis for the backbone 1H, 15 N, and 13Ca signal assignments. 13COi/13COi ÿ 1

1097 linkages obtained from HCACO and HNCO spectra, and 1Hai /1Hai ÿ 1 linkages obtained from HA(CO)NH and 15N-TOCSY-HSQC spectra were used to con®rm and extend the assignments. With these data, backbone assignments were obtained for residues Asp400 through Glu469, with the exception of HN and 15N signals for Lys445 and Ala457. The 1H-15N signal assignments are shown with an 1H-15N FHSQC spectrum of the protein in Figure 2. Within the N-terminal segment (Asp380 through Pro399), backbone signals were identi®ed for Gly383 through Thr390. Obtaining assignments for the other residues in this region was complicated by the large number of proline residues (Pro382, Pro391, Pro394, Pro398, and Pro399). This is due to the fact that proline residues lack an amide proton, and no information about the residue preceeding proline is available from the triple resonance experiments. NH resonances for residues Gly383 through Thr390 are sharp in HSQC spectra (Figure 2) and show negative 1H-15N heteronuclear NOEs (see below). Together, the data indicate that these eight residues are ¯exible. The aliphatic side-chain signals were assigned by connecting backbone 1Ha/13Ca pairs with sidechain 1H signals in HCCH-COSY and HCCHTOCSY spectra as described previously (Pelton et al., 1991). Side-chain signals derived from the 15 N-TOCSY-HSQC spectrum were also used to correlate backbone and side-chain resonances. The structured region of NtrCC-term(3Ala) contains a large number of methyl-containing residues, including ten alanine, eight threonine, and 14 leucine residues. All of the side-chain 1H-13C signals for these residues, with the exception of 1Hg-13Cg for Leu451 and Leu463, were assigned using the three 3D data sets. Stereo assignments for all of the leucine methyl groups were obtained from a comparison of a 1H-13C HSQC spectrum of a 10 % uniformly 13C-labeled protein and a 1H-13C constanttime HSQC spectrum of a 100 %-uniformly 13Clabeled protein as described (Neri et al., 1989; Szyperski et al., 1992). Obtaining assignments for the aromatic signals of the three tryptophan residues was equally important. All of these signals were assigned by comparing homonuclear COSY, TOCSY, and 1H-13C HMQC spectra. In contrast, side-chain signals for the three serine, ®ve glutamine, ®ve arginine, and three lysine residues within the structured segment (Asp400-Glu469) were dif®cult to assign due to extensive overlap. 1 b 13 b H - C signals were not assigned for Ser401, Gln408, Arg415, Ser416, Gln419, Ser423, Pro427, Arg439, Lys445, Lys462, and Met468. 1Hg-13Cg signals were not assigned for Arg412, Arg415, Gln419, Pro427, Glu430, Arg431, Arg439, Lys445, Glu447, Leu451, Lys462, Leu463, Lys464, and Met468. The assignments have been deposited in the BioMagResBank (BMRB) under accession number 4348.

1098

Solution Structure of NtrCC-term(3Ala)

Figure 2. 1H-15N FHSQC spectrum of NtrCC-term(3Ala). Peaks are labeled with residue numbers. Unlabeled resonances are denoted with the letter U. Pairs of sidechain NH2 resonances are connected by horizontal lines. Indole NH signals for Trp31 (10.31 ppm, 130.5 ppm) and Trp76 (10.16 ppm, 129.2 ppm) are not shown. The indole NH signal for Trp24 was not observed. Arginine side-chain NHe signals are denoted by the letter R. These signals are negative as a result of folding.

Structure calculations A total of 1492 intra and inter-monomer distance restraints were obtained from assignment of NOEs in 3D 1H-15N NOESY-FHSQC, 4D 13C/13C HMQCNOESY-HMQC, 4D 13C/15N HMQC-NOESYFHSQC, and 2D 13C/12C double-half ®ltered NOESY (Figure 3) spectra. These data were used in conjunction with 88 hydrogen bond restraints, and 88 phi dihedral restraints to calculate 40 structures using simulated annealing protocols similar to those developed by Nilges (1993) and O'Donoghue et al., (1996). Restraints were applied for residues Asp400 through Glu469 in each of the two monomers, with an average of 12 restraints per residue. The result can be considered a structure determination at medium resolution. The 28 structures which showed the lowest energies and distance violations were chosen for comparison. The coordinates have been deposited in the Protein Data Bank under accession number 1ntc. Structure calculations for the NtrCC-term(3Ala) homodimer were performed with three stages of simulated annealing re®nement. Short-range restraints along with six intermolecular NOEs identi®ed in a 2D 13C/12C double half-®ltered

NOESY experiment (Figure 3) (Zwablan et al., (1997) and references therein) were included in the ®rst stage in order to de®ne the monomer fold and the overall axis of symmetry. The remaining medium and long-range NOEs were treated as ambiguous (intra or inter-monomer) restraints until it was clear from intermediate calculations that the NOEs were of one type or the other. The ambiguous restraints were included in the second and third stages of re®nement using a sum-average potential (Nilges, 1993). Once the rmsd for backbone atoms Ê , the homologous protein FIS reached about 2.0 A was also used as a guide in resolving ambiguous restraints. Convergence improved dramatically as the number of long-range NOEs were resolved. In the ®nal calculations 54 restraints involving residues close to the symmetry axis were included in the third stage of re®nement as ambiguous NOEs. Intra and inter-monomer NOEs have been resolved using one or more of the techniques described above in recent structure determinations of monocyte chemo-attractant protein 1 (Handel & Domaille, 1996), melanoma growth stimulating activity protein (Fairbrother et al., 1994), transforming growth factor b (Hinck et al., 1996), cryptogein

1099

Solution Structure of NtrCC-term(3Ala) Table 1. Structural statistics NtrCC-term(3Ala) structures Structural statistics

Figure 3. Expansion of a 2D 13C/12C double half-®ltered NOESY spectrum of a 1:1 mixture of uniformly 15 N/13C-labeled and unlabeled protein. The spectrum was used to obtain six inter-monomer NOEs (A403 Hb/ Thr432 Hg; Leu406 Hd2/Thr432 Hg; Leu406 Hd1/Leu433 Hd2; Leu421 Hb/Leu452 Hd2; Ala425Hb/Leu433 Hd1; Leu429 Hd1/Leu429 Hd2). For each NOE, a continuous or broken box connects the two 1H-13C auto correlations with the two symmetry-related inter-monomer NOE cross-peaks. Assignments are denoted by one-letter amino acid code and residue number. The correlations for Leu421 Hb (13Cb of 43.7 ppm) are not shown.

(Gooley et al., 1998), transcriptional activator PUT3 (Walters et al., 1997), HIV-1 integrase (Cai et al., 1997), and the DNA-binding domain of the human papillomavirus E2 protein (Liang et al., 1996). The different classes of restraints used and the statistics for these restraints are shown in Table 1. Two of the 28 structures each had two distance Ê , although violations that were greater than 0.2 A Ê . In none of the violations was greater than 0.35 A addition, there were no dihedral violations greater than 5  . As shown in Figure 4(a), the number of NOEs is distributed over the whole molecule. A periodicity is also present. This is consistent with residues in helices where some side-chains point toward the hydrophobic core, while other sidechains point toward the surface. For the ®nal calculations, most of the long-range NOEs could be resolved as being intra (124) or inter-monomer (96). A set of ambiguous NOEs (54) involving residues close to the symmetry axis were also included. Potential intra and inter-monomer contributions for these ambiguous NOEs were taken into account with a sum-average potential function (Nilges, 1993). rmsds for the distance and dihedral restraints, as well as the X-PLOR energies (Table 1)

and

rmsds

for

hSAia

rmsd (AÊ) from exptl distance and hydrogen bond restraintsb All (1580) 0.010  0.002 Intra-residue (532) 0.009  0.003 Inter-residue Sequential (382) 0.007  0.001 Medium-range (304) 1 < ji ÿ jj < 5 0.011  0.002 Long-range intra-monomer (124) 0.012  0.002 inter-monomer (96) 0.005  0.001 Ambiguous (54) 0.009  0.01 Hbond (88) 0.016  0.005 rmsd (deg) from exptl f dihedral 0.09  0.04 restraints (88) rmsd from idealized geometry used in X-PLOR Ê) Bonds (A 0.00226  0.00004 Angles (deg) 0.511  0.004 c Impropers (deg) 0.38  0.03 X-PLOR energies (kcal-molÿ1)d ENOE 83 ECDIH 0.05  0.06 13.9  0.5 EBond Eangle 203  3 31.0  0.7 Eimproper EVDW 41  2 0.1  0.1 ENCS ÿ75  27 EL-J Atomic r.m.s. differences (AÊ)e Backbone All structured regions 0.6 0.2 Four helices-dimer 0.6  0.2 Four helices-monomer 0.5  0.1 Backbone-side-chain Heavy atoms 1.2  0.2 < 20 % solvent exposed 0.6  0.2 < 40 % solvent exposed 0.9  0.3 a

hSAi is the ensemble of 28 ®nal structures. Two of the 28 hSAi structures had two distance violations Ê . No structures had distance violations greater than 0.2 A Ê . No structures had dihedral violations of greater than 0.35 A greater than 5  . No structures had bond, angle, or improper Ê , 5  , or 5  , respectively. angle violations of greater than 0.05 A c Improper torsion restraints maintain planarity and chirality. d The square-well portion of the NOE (ENOE), the torsion angle (Ecdih), the quartic van der Waals repulsion (EVDW), and the noncrystallographic symmetry energies (ENCS)were calculated using Ê ÿ2, 200 kcal molÿ1 radÿ2, 4 kcal force constants of 50 kcal molÿ1 A Ê ÿ2, and 10 kcal molÿ1 A Ê ÿ2, respectively, with van der molÿ1 A Waals radii set to 0.78 times values used in the CHARMM empirical energy function (Brooks et al., 1983). The LennardJones energy (EL-J) was calculated using the CHARMM (Brooks et al., 1983) empirical energy function. It was not included in either the simulated annealing or re®nement protocols. e N, Ca, and C' atoms were used for backbone superpositions. Structured regions include residues Trp402 to Met468. The four helical regions include residues Trp402 to Leu414 (helix A), Leu421 to His440 (helix B), Gln446 to Leu451 (helix C), and Ala456 to Met468 (helix D). Backbone-side-chain superpositions include all heavy atoms. Solvent accessibility was calculated using the program MOLMOL (Koradi et al., 1996). Fractional solvent accessibility was calculated with respect to a Gly-X-Gly tripeptide (Chothia, 1975). Residues that have less than 20 % solvent accessibility on average for the 28 structures include Ser401 to Ala403, Leu406, Trp409, Ala410, Ala413, Leu414, Leu421 to Ser423, Ala425 to Pro427, Leu429, Glu430, Leu433, Leu434, Thr436 to Leu438, Thr441, Ala448, Ala449, Leu451, Leu452, Trp454, Gly455, Thr458, Leu459, Lys462, Leu466. Residues that have less than 40 % solvent accessibility on average in the 28 structures include Asp400 to Thr404, Leu406, Ala407, Trp409 to Asp411, Ala413, Leu414, Gy417, His418, Asn420 to Glu430, Thr432 to Leu438, Thr441, Gly443, Lys445, Gln446 to Ala449, Leu451 to Lys462, Glu465 to Gly467. b

1100

Solution Structure of NtrCC-term(3Ala)

are similar to those obtained for other protein dimers (Cai et al., 1997; Hinck et al., 1996; Liang et al., 1996; Walters et al., 1997). Structures

Figure 4. (a) Distribution of NOE restraints for NtrCC-term(3Ala). The height of each bar represents the total number of NOEs associated with that residue in the structure calculations. Shaded bars denote longrange NOEs. The locations of the four helices are shown at the top of the Figure. (b) The rmsd of heavy-atom backbone coordinates for the 28 best structures relative to the average structure. Backbone atoms for residues in the four helical segments (Table 1) were used for superposition. The rmsds for residues Pro398-Asp400 are off scale. (c) The rmsd for heavy-atom backbone and side-chain coordinates for the 28 best structures relative to the average structure. Heavy atoms for residues that show less than 40 % solvent accessibility (Table 1) were used for the superposition. The rmsds for residues Pro398-Pro399 are off scale. S(f) (d), S( ) (e) and S(w1) (f) angular order parameters (Hyberts et al., 1992) for the 28 best structures. An S value of 0.9 corresponds to a standard deviation in the angle of 24  . No  angles were calculated for Pro398, Pro399, or Pro427. No w1 angles were calculated for Pro398, Pro399, Gly417, Pro427, Gly443, Gly453, Gly455, or Gly467. (g) Solvent accessibility expressed as a percentage of the possible surface area for the residue in a Gly-X-Gly peptide (Chothia, 1975).

Structure calculations reveal that each monomer consists of four helical segments (A-D) (Figure 5). Helices A and B (residues Trp402 to Leu414 and Leu421 to His440) join with those of another monomer to form an antiparallel four-helix bundle. The long (20 residue) B helices align so that the sidechain of Leu422 packs against the side-chain of Leu433 in the other monomer. The side-chains of Leu429 from each monomer also make contact and de®ne a point of 2-fold symmetry. The A helices pack against both B helices with the N terminus of helix A close to Leu429 (B helix) of the other monomer. Compared to the classic four-helix bundle protein ROM (Banner et al., 1987; Eberle et al., 1991), the A helices are shorter in NtrCC-term(3Ala). They make contact with both B-helices, but do not overlap signi®cantly with one another (see below). Each B helix is kinked at Pro427, which is located about one-third of its length from its N terminus. The angle between helix axes for segments Leu421-Gln426 and Glu428-His440 is 14( 3) . Main-chain hydrogen bonds (Ni ÿ Oi ÿ 4) were identi®ed on the basis of slow NH exchange rates and short HN-O distances for residues surrounding Pro427. Hydrogen bonds were identi®ed for Gln426/Leu422, Leu429/Ala425, and Glu430/ Gln426, but not for Glu428/Glu424. A survey of distortions in helices caused by proline residues revealed an average kink angle of 26( 5) and a similar H-bonding pattern (Barlow & Thornton, 1988). Helices C and D (residues Gln446 to Leu451 and residues Ala456 to Met468) adopt a classic helixturn-helix DNA-binding fold (Brennan & Matthews, 1989) at either end of the molecule. The orientation of each HTH motif is de®ned by intramonomer packing interactions between each recognition helix (D helix) and the C-terminal half of its own B helix. Additional inter-monomer contacts between the C-terminal residues of helix C and residues that form the turn between helices A and B also help orient the HTH motifs. Superposition of backbone coordinates for the dimer (residues Trp402 to Met468) (Figure 6) yields Ê (Table 1). A plot an average rmsd of 0.6( 0.2) A of backbone rmsds versus residue number (Figure 4(b)) shows that the long B helices are the most precisely de®ned structural components. The N-terminal portion of helix A, and the turn between helices A and B (Arg415 to Asn420) show somewhat larger deviations. Overall, however, no segments show substantially larger rmsds, indicating that the backbone conformation for residues Trp402 to Met468 and the symmetry-related residues in the dimer are well de®ned. Superposition of backbone coordinates for the monomers yields a slightly lower rmsd than for

1101

Solution Structure of NtrCC-term(3Ala)

Figure 5. Stereo view of backbone N, Ca, and C' coordinates for the 28 best structures of the NtrCC-term(3Ala) dimer. Backbone coordinates for residues in helical segments (Table 1) were superimposed. Residues Asp400 through Met468 are shown. Helices A, C, and D are denoted by letters for the monomer colored in blue. The B helix is in the center of the moleclue. The Figure was created with the program MOLMOL (Koradi et al., 1996).

Ê . Phi and the dimer with an average of 0.5( 0.1) A psi torsion angle order parameters (Hyberts et al., 1992), which relate to the precision of the backbone for each monomer, are shown in Figure 4(d) and (e). The majority are close to one, again indicating that the monomer is well-de®ned. Lower backbone order parameters are observed for residues in the turns between helices A and B (Ser416, Gly417), between helices B and C (Lys445), and between

Figure 6. Expansion of one of the family of NtrCC-term (3Ala) structures showing the interaction between the side-chains of Ala410, Trp409, and Leu421 of one monomer (red), and Thr436 of the other monomer (blue). Backbone atoms are shown in ribbon format. The sidechains are shown in ball-and-stick format. The C-terminal portion of helix A, the turn between helices A and B, and the N-terminal portion of helix B are shown in red. The C-terminal portion of helix B of the other monomer is shown in blue. When glutamate replaces Ala410, the charged side-chain points into the hydrophobic patch created by the methyl groups of Thr436 and Leu421. The Ala410Glu mutant is monomeric in solution.

helices C and D (Trp454, Gly455). These regions also have larger coordinate rmsds (Figure 4(b) and (c)), and are less well de®ned by the data. The stereochemical quality of the backbone coordinates for the family of structures was analyzed using the program Procheck_nmr (Laskowski et al., 1993). For the 28 structures, backbone dihedral angles for over 98 % of the residues fall within the most favored (83.6 %) or additionally allowed regions (15 %) of the Ramachandran map. Moreover, the Lennard-Jones energy for each of the structures is negative (ÿ75( 27) kcal molÿ1) (Table 1), indicating that no extreme van der Waals contacts exist. Procheck nmr analysis revealed an average of 5.5 bad contacts per 100 residues. This is equivalent to the number expected for an X-ray Ê resolution structure determined to about 2.3 A (Morris et al., 1992). Heteronuclear 1H-15N NOEs were measured for 68 of the 75 assigned backbone amides (not shown). NOEs for residues Asp400 through Met469 were positive and relatively uniform with an average value of 0.76( 0.12). By comparison, the theoretical value in the spin diffusion limit is 0.82 (Grasberger et al., 1993). These data indicate that the helical and turn segments are relatively well ordered in solution. In contrast, heteronuclear NOEs for residues in the N-terminal region (residues Gly383 through Thr390) ranged from zero to ÿ0.5. Negative NOEs are indicative of 1H-15N bond vectors that undergo rapid, large amplitude motions, and show that these residues are ¯exible. Side-chains Approximately one-half of the residues in the structured C-terminal fragment (Asp400-Glu469) show less than 20 % solvent accessibility (Figure 4(g)). The side-chains for these residues are relatively well ordered (Figure 4(c), (f), and (g);

1102 Table 1) with an rmsd for heavy atoms of Ê . We were able to stereo assign the 0.6( 0.2) A methyl groups for the 17 leucine residues using fractional labeling methods. We also attempted to measure w1 and w2 angles for the leucine residues using long-range carbon-carbon (LRCC) (Bax et al., 1992) and long-range carbon-proton (LRCH) (Vuister & Bax, 1993) scalar couplings. Given the low signal-to-noise and poor spectral dispersion observed in 2D versions of these experiments, we decided not to include the measurements in the structure calculations, nor did we attempt to measure w1 angles for other residues. Thus, the conformational ordering is due solely to NOEs and packing interactions. Measurement of many additional 1H-1H, 13C-13C and 1H-13C coupling constants, which can be dif®cult to measure in a 20 kDa protein, would be required to obtain a high-resolution structure.

Discussion The C-terminal domain of NtrC is connected to the central or motor domain by the segment Asp380-Pro399. Residues Gly383 through Thr390 showed negative 1H-15N heteronuclear NOEs, indicating that this region is ¯exible. The remaining residues in this segment (Asp380 to Pro382 and Pro391 to Pro399) were not assigned, due in part to the presence of ®ve proline residues. We speculate that the unassigned residues are also ¯exible. Whether this segment is structured or ¯exible in the full-length protein is unknown. Though the role of the connecting segment in communication between the C-terminal and central domains is not known, there are two lines of evidence that such communication occurs: ®rst, binding to the enhancer stimulates oligomerization and the ATPase activity of NtrC (Rombel et al., 1998); second, some lesions affecting the Switch 1 region of the central domain result in profound inhibition of the interaction between NtrC and s54-holoenzyme only when NtrC is bound to DNA (D. Yan & S. Kustu, personal communication). Comparison of monomer interchange rates for full-length NtrC and various deletion forms indicates that the primary dimerization determinants are located in the C-terminal domain (Klose et al., 1994). The structure calculations on NtrCC-term(3Ala) reveal that the majority of intermonomer interactions occur between the A and B helices, which fold into an antiparallel four-helix bundle (Figure 5). Indeed, a truncated form of wild-type NtrC that lacks the B, C, and D helices is monomeric in solution (Klose et al., 1994). An Ala410Glu mutant is also monomeric (Klose et al., 1994). In this case an unfavorable interaction occurs when the charged glutamate side-chain is inserted into a hydrophobic pocket formed by Trp409 and Leu421 of one monomer, and Thr436 of the other monomer (Figure 6). Additional intermonomer interactions occur between residues at

Solution Structure of NtrCC-term(3Ala)

the end of helix C and residues in the loop between helices A and B (Figure 5). When these interactions are removed by deleting either the D or both the C and D helices, the protein is largely monomeric (Klose et al., 1994). Similar conclusions were derived from comparison of the mutagenesis data for NtrC with the structure of the homologous protein FIS, before the structure of NtrCC-term(3Ala) was known (Klose et al., 1994). The structure calculations con®rm arguments based on homology (Drummond et al., 1986) and mutagenesis (Contreras & Drummond, 1988) that residues Gln446 to Met468 (helices C and D; Figure 5) adopt an HTH DNA-binding fold. The HTH motif satis®es classic stereochemical constraints proposed by Brennan & Matthews (1989). Position 9 is glycine (Gly453). Positions 4 (Ala448), 8 (Leu452), 10 (Trp454), and 15 (Leu459) are hydrophobic. Position 5 (Ala449) contains an unbranched side-chain, and no residues of the HTH are proline. Superposition of Ca coordinates for the HTH of NtrCC-term(3Ala) with similar residues of the canonical HTH in the lambda Cro repressor yields an Ê . By comparison, the rmsd for rmsd of 1.1( 0.1) A similar residues between the catabolite activator Ê (Brennan & protein (CAP) and Cro is 0.9 A Matthews, 1989). If three helices are used for classi®cation (Wintjens & Rooman, 1996), NtrCC-term (3Ala) belongs to the toxin family of HTH DNAbinding proteins. Other members of the family include diphtheria toxin repressor and FIS. In several protein-DNA complexes side-chains for residues at positions 1, 2, and 6 of the recognition helix make contact with DNA bases (Drummond et al., 1986). For NtrC, DNA-binding was abolished when these three residues (Arg456, Asn457, and Arg461) were replaced with alanine (North & Kustu, 1997). In the structures, the alanine methyl groups point into solution as would be expected for the Arg and Asn side-chains of the wild-type protein. This gives us con®dence that the structure of the triple-alanine mutant is a good model for the structure of the wild-type protein. Point mutations at positions Thr460 and Arg461 have also been shown to reduce transcriptional activity, presumably due to reduced af®nity for DNA (Contreras & Drummond, 1988). Examination of the structures suggests that Lys462, Lys464, Glu465, M468, and Glu469 are also candidates for speci®c or non-speci®c DNA contacts. The C-terminal domain of NtrC is homologous to FIS. FIS is a single domain protein of 98 amino acids that functions in a number of cellular processes, many of which involve the binding to and bending of DNA (Finkle & Johnson, 1992). Phylogenetic evidence suggests that FIS originated from the C-terminal domain of an ancestral a-proteobacterial NtrC gene (Morett & Bork, 1998). If the ®rst 26 N-terminal residues are left out of the comparison, FIS sequences are 50 % identical to a-proteobacterial NtrC proteins and 33 % identical to g-proteobacterial NtrC proteins, of which S. typhimurium NtrC studied here, is an example.

Solution Structure of NtrCC-term(3Ala)

The phylogenetic results suggest that the structure of FIS (derived from either a or g-proteobacteria) would be very similar to the C-terminal domain of NtrC derived from a-proteobacteria and also similar, but less so, to the C-terminal domain of NtrC derived from g-proteobacteria. The structure of FIS from Escherichia coli, as well as the structures of several mutant forms, have been solved by X-ray crystallography (Kostrewa et al., 1992; Safo et al., 1997; Yuan et al., 1991, 1994). Comparison of the structure of NtrCC-term(3Ala) from the g-proteobacterium S. typhimurium determined here with the X-ray results on FIS (Figure 7) shows that the proteins are indeed similar. Both form antiparallel four-helix bundles with HTH DNAbinding motifs extending from the long B helices. As for NtrCC-term(3Ala), in all but one X-ray structure of FIS, the ®rst 26 residues were disordered. In the only exception, residues Val10 to Gln24 folded into a b-hairpin (Safo et al., 1997). Closer examination of NtrCC-term(3Ala) and FIS (Figure 7) reveals three signi®cant structural differences that can be related to sequence variations. First we noted that the A helices for NtrCC-term (3Ala) were shorter than those of FIS. Helix A in NtrCC-term(3Ala) starts either between Asp400 and Ser401 or between Ser401 and Trp402. Weak dNN NOEs are observed between the latter pair, whereas degeneracy of amide proton shifts prevents identi®cation of similar NOEs between the former. In any case, Pro398 and Pro399 must disrupt the helix. On the other hand, the A helices in FIS are disrupted by Pro26, which is equivalent to Thr397 in NtrC. The shorter A helices in NtrCC-term (3Ala) result in signi®cantly less overlap between them. For FIS Leu27 and Val31 of opposing helices

Figure 7. Ribbon diagrams of the structured regions of NtrCC-term(3Ala) (a) and FIS (b) (Yuan et al., 1991). The structures were oriented relative to one another by superimposing the long B helices. The N and C termini are denoted N and CO, respectively. Helices for the blue monomer for each protein are denoted as A, B, C, or D. The Figure was created with the program MOLSCRIPT (Kraulis, 1991).

1103 interdigitate, whereas for NtrCC-term(3Ala) the equivalent interaction would be between Pro398 and Trp402. We speculate that the reduced overlap may lead to greater ¯exibility within the long B helices, and as a consequence, make it easier to adjust the orientation or spacing of the HTH motifs relative to one another. An alternative interpretation is that the hydrophobic side-chains of Pro398 and Pro399 pack against hydrophobic side-chains of the other helix (Ala403), providing some additional stabilization to the dimer. These ideas might be tested by removing Pro398 and Pro399, extending the A helices with suitable hydrophobic residues, and probing the DNA-binding behavior of the mutant forms. A second signi®cant difference between NtrCC-term(3Ala) and FIS relates to a prolineinduced kink in the long B helices. Sequence alignment of several species of both a and g-proteobacterial NtrCs as well as FIS, shows that with one exception, the B helices in all of the proteins contain one prolyl residue (Morett & Bork, 1998). All ®ve gproteobacterial species of NtrC contain conserved prolyl residues at position 427. In contrast, all seven of the a-proteobacterial species and all six FIS sequences contain a conserved prolyl residue at position 432 of NtrC or the equivalent position 61 in FIS. The sequence conservation suggests that the proline residues play an important structural or functional role. The B helices for NtrCC-term(3Ala) and FIS are compared in Figure 8. Both show proline-induced kinks. The kink angles (14(3) in NtrCC-term(3Ala); 16 to 30  in FIS (Kostrewa et al., 1992; Yuan et al., 1991)) are similar in magnitude to kinks observed in other proline-containing helices (26(5) ; Barlow & Thornton, 1988). As shown in Figure 8, the prolyl residue in NtrCC-term(3Ala) occurs ®ve residues before that in FIS. A structural consequence of this sequence difference is that the B helices kink at different places. For NtrCC-term(3Ala) the kink occurs about one third of the distance from the N terminus of its B helix, whereas for FIS the kink

Figure 8. Comparison of B helices in NtrCC-term(3Ala) (blue) and FIS (red) (Yuan et al., 1991) in stereo. The helices were oriented by superimposing the ®rst six residues in the B helix of NtrCC-term(3Ala) with the equivalent residues in FIS. The Figure was created with the program MOLMOL (Koradi et al., 1996).

1104 is closer to the center of its B helix. In addition, the spacing of the proline residues does not correspond to an integral number of helical turns, so that the phase of the kink is different in the two proteins (Figure 8). The differences in kink position and phase lead, at least in part, to differences in the orientations of the HTH (C and D helices) motifs. This may in turn lead to differences in the DNA-binding properties of the two proteins. It is known that NtrCC-term and FIS have different DNA-binding speci®cities (Porter et al., 1993; Pan et al., 1996a). In order to better understand the differences, side-by-side comparisons of the DNAbinding behavior for both proteins are needed. Mutational studies have shown the importance of Pro61 in FIS function. FIS mutants Pro61Ser and Pro61Leu bind to DNA but have no Hin-mediated DNA inversion activity (Osuna et al., 1991). Interestingly, a Pro61Ala mutant retains much of the function of the wild-type protein. X-ray analysis of the latter mutant protein showed that the B helix was kinked even without the prolyl residue (Yuan et al., 1994). It was concluded that hydrophobic and hydrogen bond interactions associated with the B helices also determine the conformation of the helix. Given the high structural similarity, the same conclusions may apply for g-proteobacterial species of NtrC as well. The last signi®cant difference between the structures of NtrCC-term(3Ala) and FIS concerns the relative spatial arrangement of the HTH domains. Figure 9 shows a superposition of the HTH for one monomer in each of the NtrCC-term(3Ala) structures with similar residues in FIS. The individual HTH domains for NtrCC-term(3Ala) are very well de®ned Ê , and are very with an rmsd of only 0.2(0.1) A similar to the single HTH domain in FIS (rmsd Ê ). In contrast, the spatial relationship of 1.0(0.1) A one HTH domain relative to the other in the NtrC dimer is less well de®ned by the data, and differs substantially from the conformation observed in FIS (Figure 9). In NtrCC-term(3Ala) the C and D helices are displaced and rotated such that the distance between recognition helices is shorter Ê ) than in FIS (25 A Ê ). The altered relation(23 A ship between HTH domains can be attributed in part to differences in the position and phase of proline-induced kinks in the B helices. Compared to FIS, the kink in NtrC is closer to the N terminus of its B helix. This leads to greater displacement of the C-terminal end of each B helix, and as a consequence, brings the HTH domains closer together. The short distance between recognition helices in NtrCC-term(3Ala) has implications for DNA binding and DNA bending. Following the arguments of Yuan et al. (1991) and Kostrewa et al. (1992), the shortest distance between major grooves in unbent Ê . Many HTH proteins can be B-form DNA is 29 A docked into the major groove of unbent DNA, whether the protein actually bends DNA or not. If bending occurs, it is often centered at the ends of the dyad-symmetric binding sites, and in the ¯anking sequences. The CAP-DNA complex is an

Solution Structure of NtrCC-term(3Ala)

Figure 9. Comparison of C and D helices in NtrCC-term(3Ala) (blue) and FIS (red) (Yuan et al., 1991) in stereo. One of the pair of HTH domains for each of the NtrCC-term(3Ala) dimers (Lys445 through Gly467) was superimposed with the equivalent residues in FIS (bottom). The symmetry-related HTH domains in NtrCC-term(3Ala) and FIS are also shown (top). Black lines are used to indicate that only residues that form the HTH domains in each of the structures are displayed. The Figure was created with the program MOLMOL (Koradi et al., 1996).

important example (Schultz et al., 1991). In contrast, the recognition helices for both NtrCC-term (3Ala) and FIS are too close to permit docking without bending. FIS has been shown to bend DNA to different extents depending on sequence (Thompson & Landy, 1998). Similarly, a constitutive mutant form of NtrC with phenylalanine substituted for Ser160 has been shown to bend plasmid DNA that contains the glnA promotor-regulatory region including the enhancer, and binds to negatively supercoiled DNA (Revet et al., 1995). The spacing of recognition helices in NtrCC-term(3Ala) is consistent with DNA bending by single dimers, and may account for the observed binding to negatively supercoiled DNA. However, in both studies the NtrC binding unit was either a tetramer or higher order oligomer, and it is unclear whether bending results from oligomerization, or is induced by a single dimer. Studies of NtrC with plasmids that contain a single strong binding site should help resolve this ambiguity. The structural comparison of NtrCC-term(3Ala) and FIS shows that although the proteins have very high sequence identity, sequence differences result in structural differences. Particularly striking are the different lengths of the A helices, the different kink positions in the B helices, and the different spacing of the recognition helices. These factors may relate to differences in the DNA-binding behavior for the two proteins. Given the shorter A helices in NtrCC-term(3Ala), another question relates to how ¯exible the B helices are and how easy it is to reorient the C and D helices upon DNA-binding. Given their overall structural similarity, further investigations of the differences in the DNA-binding behavior of NtrCC-term and FIS should yield new insights into how HTH domains recognize DNA.

1105

Solution Structure of NtrCC-term(3Ala)

Materials and Methods Protein expression and purification Plasmid pJES1092 (Porter et al., 1993) codes for a mutant form of the DNA-binding domain (91 residue C-terminal amino acids 379-469) of NtrC from S. typhimurium. In this construct, Arg456, Asn457, and Arg461 are substituted with alanine. The triply-substituted domain, denoted NtrCC-term(3Ala), was over-expressed using E. coli strain BL21 DE3 pLys-S harboring plasmid pJES1092 under control of phage T7 RNA polymerase. 15 N, 15N/13C, and 10 % uniformly 13C-labeled samples were obtained from bacteria grown on M9 minimal medium (Nohaile et al., 1997) supplemented with 15N-labeled ammonium chloride, both 15N-labeled ammonium chloride and uniformly 13C-labeled glucose (Martec, Columbia, Maryland), or 10 % uniformly 13C-labeled glucose, respectively. An innoculant was produced by growing the bacteria overnight at 37  C on Luria-Bertani broth (LB) supplemented with ampicillin (50 mg/ml) and chloramphenicol (34 mg/ml). The cells were twice pelleted by centrifugation and resuspended in 20 ml of M9 medium to remove residual LB. The cells were then added to one liter of M9 to an absorbance of 0.03 at 600 nm. The bacteria were allowed to grow at 37  C with vigorous shaking to an absorbance of 0.4 at 600 nm. Protein production was initiated with the addition of isopropylb-D-thiogalactopyranoside (IPTG, 1 mM). After six hours of growth, the cells were harvested by centrifugation and resuspended in 20 ml of breakage buffer (50 mM Tris-acetate (pH 8.2), 200 mM potassium chloride, 1 mM EDTA, 1 mM DTT, and the protease inhibitors phenylmethylsulfonyl¯oride (PMSF, 1 mM) and leupeptine (10 mg/ml)) (Keener, 1988). The suspension was stored at ÿ80  C until use. Puri®cation of NtrCC-term(3Ala) was initiated by sonication of the thawed cell suspension at 100 W for three one-minute intervals. The cell debris was removed by centrifugation at 12,000 g for 30 minutes. The lysate was fractionated with 40 % and 75 % ammonium sulfate. The material that precipitated between 40 and 75 % saturation was re-suspended and dialyzed against buffer A (10 mM Tris-acetate (pH 8.2), 25 mM NaCl, 1 mM EDTA, 5 % (v/v) glycerol, 1 mM PMSF, 10 mg/ml leupeptine). The dialysate was diluted to 200 ml with buffer A, and the pH was reduced to 3.4 using small aliquots of 1 M HCl. The solution was applied to a 5 ml S-Sepharose cation exchange column (SP-Sepharose fast-¯ow, Pharmacia, Uppsala, Sweden) at a ¯ow rate of 2 ml/minute. The column was washed with buffer A containing 300 mM NaCl, and the protein was eluted with buffer A containing 600 mM NaCl. The pH of the solution was raised to 7.5 with small aliquots of 1 M NaOH and was dialyzed overnight against buffer B (10 mM sodium phosphate (pH 7.5), 25 mM NaCl, 0.1 mM EDTA). After lyophilization, the protein was resuspended in H2O, the pH was adjusted to 6.1, and the buffer was exchanged for sample buffer (20 mM sodium phosphate (pH 6.1), 50 mM potassium chloride, 0.1 mM EDTA, 0.02 % (w/v) sodium azide) using a 2 ml centricon ultracentrifuge concentration unit (Amicon, Millipore, Bedford, MA.). Protein concentrations were determined using a calculated extinction coef®cient of 16,500 Mÿ1 cmÿ1 at 280 nm (Pace et al., 1995). Approximately 20 mg of NtrCC-term(3Ala) were obtained per liter of M9 medium. Mass spectral analysis con®rmed that the correct

sequence was produced. The molecular weight of the over-expressed protein was within two mass units of the theoretical value (9979 Da per monomer). Sample purity was greater than 95 % as judged by polyacrylamide gel electrophoresis and mass spectral analysis. NMR samples in H2O were prepared by adding 10 % 2 H2O to the protein solutions. NMR samples in 2H2O were prepared by dissolving lyophilized protein in 99.96 % 2H2O, re-lyophilizing the protein, and re-dissolving it again in 99.96 % 2H2O. Typical protein concentrations ranged from 0.8 to 1.5 mM. Samples were stored in NMR tubes at 4  C. NMR spectroscopy Unless otherwise stated, NMR spectra were recorded on a Bruker AMX-600 NMR spectrometer equipped with a triple-axis gradient ampli®er and probe at 35  C. Most experiments were recorded using the States-TPPI method (Marion et al., 1989) of quadrature detection for indirectly detected dimensions. For heteronuclear experiments, 15N and 13C decoupling were achieved by WALTZ16 (Shaka et al., 1983) modulation of a 1.6 kHz rf ®eld (15N) or GARP (Shaka et al., 1985) modulation of a 3.8 kHz rf ®eld (13C). 1H chemical shifts are referenced indirectly to DSS using the H2O resonance (4.73 ppm at 37  C). 15N and 13C signals were referenced indirectly to DSS and liquid ammonia (Wishart et al., 1995). Data were processed with the NMRPipe suite of programs (Delaglio et al., 1995). Typically, data were apodized with phase-shifted sine-bell window functions followed by zero-®lling to twice the original data size. For constant-time evolution periods linear prediction was used to double the number of points before apodization. Spectra were analyzed using the programs CAPP and PIPP (Garrett et al., 1991). Two-dimensional homonuclear experiments 2D homonuclear NOESY (Macura et al., 1981) spectra of unlabeled NtrCC-term(3Ala) in 2H2O and H2O were recorded on a Bruker DRX spectrometer operating at 500 MHz with mixing times of 80 ms. 2D homonuclear DQFCOSY (Rance et al., 1983) and TOCSY (Bax & Davis, 1985) spectra of unlabeled NtrCC-term(3Ala) were acquired in 2 H2O on a General Electric GN/OMEGA spectrometer operating at 500 MHz. For the TOCSY experiment 1H isotropic mixing was achieved by WALTZ16 modulation (Shaka et al., 1983) of a 7 kHz rf ®eld. Two-dimensional heteronuclear experiments 2D 15N-1H HSQC spectra of NtrCC-term(3Ala) in H2O were acquired using the FHSQC sequence (Mori et al., 1995). 2D 13C-1H constant-time (ct) HSQC spectra (Vuister & Bax, 1992) of uniformly 13C/15N labeled NtrCC-term(3Ala) were recorded in 2H2O. A matching 13 C-1H HSQC spectrum (Bax et al., 1990b) of 10 % uniformly 13C-labeled NtrCC-term(3Ala) was recorded in a similar manner. Comparison of the two 13C-1H HSQC spectra was used to stereoassign the methyl groups of the 17 leucine residues (Neri et al., 1989; Szyperski et al., 1992). Amide exchange rates were measured by lyophilizing the protein from H2O, redissolving it in 2H2O (pH 6.1), and acquiring a series of 2D 1H-15N FHSQC spectra. At 37  C most amide protons had exchanged before the ®rst spectrum could be recorded. The relatively fast amide

1106 exchange is probably due to rapid monomer exchange (half-life two to three minutes) (Klose et al., 1994) with concomitant unfolding. To reduce the amide exchange rates, these experiments were conducted at 25  C. Spectra were recorded at 7, 17, 27, 47, 67, 87, 107, 127, and 147 minutes after addition of 2H2O. Upper bounds on amide proton exchange rates were estimated from peak intensities in the spectrum recorded after seven minutes of exchange, assuming that a signal could be detected if 5 % of the protons remained. The upper bounds were compared to calculated random coil exchange rates (Zhang Y.-Z. & Roder, H. at the sphere hydrogen exchange estimation web site (http://dino.fold.fccc.edu:8080/sphere.html) (Bai et al., 1993; Connelly et al., 1993). Amide protons that showed protection factors (ratio of actual exchange rate to that of a random coil peptide) of at least 25 were assumed to be hydrogen bonded. 3 JHNHA coupling constants were measured semi-quantitatively (either less than 6 Hz or greater than 8 Hz) from a 15N-HMQC-J spectrum (Kay & Bax, 1990) recorded with 512 (1H) and 475 (15N) complex points (t1 max ˆ 400 ms). Exponential line broadening (15 Hz) and exponential line narrowing (ÿ2 Hz) were applied in the 1H and 15N dimensions, respectively. After zero®lling (2048 points (1H) and 1024 points (15N)) and Fourier transformation, the ®nal digital resolution was 4.1 Hz (1H) and 1.1 Hz (15N). These data were used to estimate phi torsion angles for 44 residues. This led to 88 phi torsion restraints for the dimer. 1 H-15N heteronuclear NOEs were measured with experiments that employ sensitivity-enhanced coherence selection and water-¯ipback pulses to return H2O to the ‡Z axis before acquisition (Farrow et al., 1994). Protons were saturated by repeated application of a 120  ¯ip angle pulse followed by a 10 ms delay. The total proton saturation period was set to three seconds. The recycle delay between scans was set to 1.8 seconds. Three-dimensional heteronuclear experiments 3D NOESY-FHSQC (Talluri & Wagner, 1996) and TOCSY-HSQC (Driscoll et al., 1990) experiments were recorded on 15N-labeled samples of NtrCC-term(3Ala) dissolved in H2O. The HSQC portion of the NOESY experiment was based on the FHSQC sequence (Mori et al., 1995) as suggested (Talluri & Wagner, 1996). The sequence achieves water-¯ip back excitation of the amide signals using purge-type gradients (Bax & Pochapsky, 1992) and a 3-9-19 watergate sequence (Sklenar et al., 1993). For the TOCSY-HSQC experiment, 1 H isotropic mixing was achieved by DIPSI-2rc (Cavanagh & Rance, 1992) modulation of a 8.3 kHz rf ®eld. Mixing times for the NOESY and TOCSY experiments were 100 ms and 45 ms, respectively. 3D constant-time HNCA (Grzesiek & Bax, 1992b), HNCO (Grzesiek & Bax, 1992b), and HA(CO)NH (Grzesiek & Bax, 1992a) experiments were recorded as described on uniformly 15N/13C labeled NtrCC-term(3Ala). Triple-axis purge-type gradients (Bax & Pochapsky, 1992) were used to remove artifacts and a 3-9-19 watergate sequence (Sklenar et al., 1993) replaced the last 180  1 H pulse of each sequence to suppress the solvent signal. 3D constant-time gd-HCACO (Zhang & Gmeiner, 1996), HCCH-COSY (Ikura et al., 1991), and HCCHTOCSY (Bax et al., 1990a) experiments were recorded as described on a uniformly 15N/13C sample of NtrCC-term(3Ala) dissolved in 2H2O. In the HCCH-

Solution Structure of NtrCC-term(3Ala) TOCSY experiment isotropic 13C mixing was achieved via DIPSI-2 modulation (Shaka et al., 1988) of a 5.4 kHz rf ®eld that was applied for 22.4 ms. In all three experiments purge-type gradients (Bax & Pochapsky, 1992) were used to remove artifacts and suppress the residual H2HO signal. 4D 13C/13C HMQC-NOESY-HMQC (Vuister et al., 1993) and 13C/15N HMQC-NOESY-FHSQC (Kay et al., 1990) experiments were recorded on uniformly 15N/13Clabeled samples of NtrCC-term(3Ala) dissolved in 2H2O and H2O, respectively. The 13C/15N HMQC-NOESYFHSQC experiment was derived from the 13C/13C HMQC-NOESY-HSQC experiment by replacement of the HSQC step with the FHSQC sequence (Mori et al., 1995). The NOE mixing time in both experiments was 80 ms. The 1H, 13C and 15N carrier frequencies were set to H2O (4.67 ppm), 63.7 ppm (13C), and 118.75 ppm (15N), and either 16 (13C/13C NOESY) or 32 (13C/15N NOESY) scans per complex t1, t2, t3 point. For the 13C/13C HMQCNOESY-HMQC experiment the 1H and 13C spectral widths were 11 ppm and 20.7 ppm, respectively, with 64 (t1, 1H), 15 (t2, 13C), 15 (t3, 13C), and 512 (t4, 1H) complex points. For the 13C/15N HMQC-NOESY-FHSQC experiment the spectral widths were 14.1, 20.7, and 18.5 ppm in the 1H, 13C, and 15N dimensions with 61 (t1, 1H), 10 (t2, 13C), 10 (t3, 15N) and 256 (t4, 1H) complex points. The program PIPP (Garrett et al., 1991) was used to peak pick the 4D NOESY data sets. The peak tables were used as input to in-house written software that matches each NOESY peak with possible chemical shift assignments and the distance for the assignment in a representative structure. A 2D 13C-1H double half-®lter experiment was recorded on a DRX-500 spectrometer, using a 1:1 mixture of unlabeled and uniformly 13C/15N-labeled protein in 2 H2O. Monomer exchange is rapid (half-life of two to three minutes at 37  C; Klose et al., 1994), and no special procedures were necessary to promote formation of heterodimers. The pulse sequence was derived from the 4D 13 C/13C HMQC-NOESY-HMQC sequence (Vuister et al., 1993) set to record the originating 13C and destination 1H (bound to 12C) chemical shifts. Structure calculations NOEs identi®ed in 3D 15N-edited NOESY-FHSQC, 4D N/13C-edited HMQC-NOESY-FHSQC, and 4D 13 C/13C-edited HMQC-NOESY-HMQC spectra were Ê ), medium (1.8-3.5 A Ê ), or classi®ed as strong (1.8-2.7 A Ê ) (Clore et al., 1986). A correction was weak (1.8-5.0 A added to the upper limit for constraints involving methyl protons and non-stereospeci®cally assigned protons (Clore et al., 1986; WuÈthrich et al., 1983). Phi torsion angles were constrained to ÿ60( 30) (Pardi et al., 1984) for 3JHNHA values less than 6 Hz as measured in an HMQC-J spectrum (Kay & Bax, 1990). Hydrogen bonds were identi®ed on the basis of low amide proton exchange rates (protection factors greater than 25) and Ê ) calcushort distances in sets of structures (rmsd 1 A lated without these restraints. Restraints were applied Ê ) and between HN between N and O atoms (2.8-3.3 A Ê ) (Clore et al., 1986). and O atoms (1.8-2.3 A Structures were calculated using ab initio simulated annealing methods as implemented in X-PLOR version 3.851 (BruÈnger, 1992). The ideas of Nilges (1993) and O'Donoghue et al. (1996) were also implemented for calculation of the NtrCC-term(3Ala) homodimer. At all stages of re®nement, symmetry was imposed using both 15

1107

Solution Structure of NtrCC-term(3Ala) a non-crystallographic symmetry pseudo-energy term (NCS restraints) and symmetry restraints on inter-monomer Ca-Ca distances (Nilges, 1993). Each simulation was computed in 0.005 ps steps for a total of 100 ps at a temperature of 2000 K followed by cooling in 25 K steps to 100 K. Structures were output after 250 steps of Powell energy minimization. In both the simulated annealing and energy minimization procedures only repulsive energy terms, NOE, and dihedral restraints were used. Starting structures with randomized phi and psi torsion angles were generated using the protocol random_monomer (Nilges, 1993). In the starting conformation the two monomers are identical and overlapped. Three stages of simulation were then used to convert the starting structures into a family of converged structures. In stage I the secondary structure of each monomer was de®ned and its orientation with respect to the other monomer was determined. Conformational restraints input at this stage included intra-residue NOEs, sequential NOEs between amide protons (dNN), sequential NOEs between alpha and amide protons (daN), and medium-range NOEs between alpha and amide protons (daNi,i‡3), which were identi®ed in a 3D NOESY-FHSQC spectrum. These NOEs were assumed to be intra-monomer. Phi torsion angle restraints derived from the 15N HMQC-J experiment were also included. Together, these NOE and torsion angle restraints de®ne the four helical segments of each NtrCC-term(3Ala) monomer. In addition, six inter-monomer restraints derived from NOEs observed in a 2D 13C/12C double-half ®ltered NOESY spectrum were used. These restraints order the monomers with respect to one another by forcing the B helices to align in an antiparallel fashion. The distance restraints were applied with a soft square-well potential, with harmonic walls, that switches to a linear function (slope Ê . The 0.1 kcal molÿ1 Aÿ1) at deviations larger than 0.5 A harmonic force constant increases from 2 to 150 kcal Ê ÿ2 during the calculations. Center averaging was molÿ1 A used for NOEs to methyl and unresolved methylene protons. In stage II the global fold of each monomer and the symmetry of the dimer were de®ned in detail. In addition to the distance and torsion angle restraints described above, a set of ambiguous restraints (intra or inter-monomer) derived from NOEs in 4D 13C/13C NOESY and 13C/15N NOESY spectra was included. A sum-average distance function was applied to these restraints in order to account for the possibility of intraor inter-monomer contributions to the NOE (Nilges, 1993). These NOEs were included in the calculations with a soft-square-well potential that switches from harÊ ÿ2) to linear monic (force constant 2 to 50 kcal molÿ1 A ÿ1 Ê ÿ1 Ê. (slope 2 kcal mol A ) for deviations greater than 1 A Ambiguous restraints to methyl, methylene, and methine groups were resolved to the nearest carbon atom. Intramonomer distance restraints from stage I were included using a square-well potential with a ®xed force constant Ê ÿ2 and center averaging. of 50 kcal molÿ1 A In stage III the structures output from stage II were re-re®ned using the protocol rere®ne.inp (Nilges, 1993) with force constants held ®xed to their ®nal values. The resulting structures were used to assign additional NOEs in 4D 15N/13C-edited and 13C/13C-edited NOESY spectra using in-house software to match NOE cross-peaks with potential assignments and distances in representative structures. Once the backbone rmsd for the family of structures Ê , many of the ambiguous NOEs could be reached 2 A

resolved as being intra- or inter-monomer. At this point the structure of FIS (Yuan et al., 1994) was also used as a guide in resolving ambiguous restraints. This prompted modi®cation of the restraint lists and potential functions. Restraints previously identi®ed as ambiguous, but that now could be resolved as intra-monomer, were included in stage I using the same potential that was applied to intra-monomer NOEs. These restraints helped de®ne the structure of each monomer, which in turn led to better overall convergence. Restraints that could be resolved as inter-monomer were included in stage II using center averaging and a soft-square well potential. Finally, restraints that remained ambiguous were added in stage III. As reported previously (O'Donoghue et al., 1996), better convergence was achieved as the monomer fold was more precisely de®ned in stage 1. Convergence was only about 30 % in early stages of re®nement. With the addition of resolved long-range intra-monomer NOEs, convergence increased to 80 %.

Acknowledgements We thank Mr David King for performing mass spectral analysis, and Ms Andrea Shauger for help with protein expression. This work was supported by the Of®ce of Energy Research, Of®ce of Health and Environmental Research, Health Effects Research Division of the U.S. Department of Energy under contract no. DE-AC0376SF00098 to D.E.W., and through instrumentation grants from the U.S. Department of Energy (DE FG0586ER75281), and the National Science Foundation (DMB 86-09035 and BBS 87-20134) to D.E.W. This work was also supported by NIH grant GM38361 to S.K.

References Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W. (1993). Primary structure effects on peptide group hydrogen exchange. Proteins: Struct. Funct. Genet. 17, 75-86. Banner, D. W., Kokkinidis, M. & Tsernoglou, D. (1987). Ê resolStructure of the ColEI ROP protein at 1.7 A ution. J. Mol. Biol. 196, 657-675. Barlow, D. J. & Thornton, J. M. (1988). Helix geometery in proteins. J. Mol. Biol. 201, 601-619. Bax, A. & Davis, D. G. (1985). MLEV-17 based twodimensional homonuclear magnetization transfer spectroscopy. J. Magn. Reson. 65, 355-360. Bax, A. & Pochapsky, S. (1992). Optimized recording of heteronuclear multidimensional NMR spectra using pulsed ®eld gradients. J. Magn. Reson. 99, 638-643. Bax, A., Clore, G. M. & Gronenborn, A. M. (1990a). 1 H-1H correlation via isotropic mixing of 13C magnetization, a new 3D approach for assigning 1H and 13 C spectra of 13C-enriched proteins. J. Magn. Reson. 88, 425-431. Bax, A., Ikura, M., Kay, L. E., Torchia, D. A. & Tschudin, R. (1990b). Comparison of different modes of two-dimensional reverse-correlation NMR for the study of proteins. J. Magn. Reson. 86, 304318. Bax, A., Max, D. & Zax, D. (1992). Measurement of long-range 13C-13C J couplings in a 20 kDa proteinpeptide complex. J. Am. Chem. Soc. 114, 6923-6925.

1108 Brennan, R. G. & Matthews, B. W. (1989). The helixturn-helix DNA-binding motif. J. Biol. Chem. 264, 1903-1906. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. & Karplus, M. (1983). CHARMM: a program for macromolecular energy, minimisation, and dynamics calculations. J. Comput. Chem. 4, 187-217. BruÈnger, A. T. (1992). X-PLOR Manual, Version 3.0. A System for Crystallography and NMR, Yale University, New Haven. Cai, M., Zheng, R., Caffrey, M., Craigie, R., Clore, G. M. & Gronenborn, A. M. (1997). Solution structure of the N-terminal zinc binding domain of HIV-1 integrase. Nature Struct. Biol. 4, 567-577. Cavanagh, J. & Rance, M. (1992). Suppression of cross relaxation effects in TOCSY spectra via a modi®ed DIPSI-2 mixing sequence. J. Magn. Reson. 96, 670678. Chothia, C. (1975). Structural invarients in protein folding. Nature, 254, 304-308. Clore, G. M., Nilges, M., Sukumaran, D. K., BruÈnger, A. T., Karpulus, M. & Gronenborn, A. M. (1986). The three-dimensional structure of a-1-purothionin in solution: combined use of nuclear magnetic resonance, distance geometry, and restrained molecular dynamics. EMBO J. 5, 2729-2735. Connelly, G. P., Bai, Y., Jeng, M. F. & Englander, S. W. (1993). Isotope effects in peptide group hydrogen exchange. Proteins: Struct. Funct. Genet. 17, 87-92. Contreras, A. & Drummond, M. (1988). The effect on the function of the transcriptional activator NtrC from K. pneumoniae of mutations in the DNA-recognition helix. Nucl. Acids Res. 16, 4025-4039. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J. & Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR, 6, 277-293. Driscoll, P. C., Clore, G. M., Marion, D., Wing®eld, P. T. & Gronenborn, A. M. (1990). Complete resonance assignments for the polypeptide backbone of interleukin-1b using 3D heteronuclear NMR spectroscopy. Biochemistry, 29, 3542-3556. Drummond, M., Whitty, P. & Wootton, J. (1986). Sequence and domain relationships of NtrC and NifA from K. pneumoniae: homologies to other regulatory proteins. EMBO J. 5, 441-447. Eberle, W., Pastore, A., Sander, C. & RoÈsch, P. (1991). The structure of ColEI ROP in solution. J. Biomol. NMR, 1, 71-82. Fairbrother, W. J., Reilly, D., Colby, T. J., Hesselgesser, J. & Horuk, R. (1994). The solution structure of melanoma growth stimulating activity. J. Mol. Biol. 242, 252-270. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G., Shoelson, S. E., Pawson, T., Forman-Kay, J. D. & Kay, L. E. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src Homology 2 domain studied by 15N NMR relaxation. Biochemistry, 33, 5984-6003. Finkle, S. E. & Johnson, R. C. (1992). The FIS protein: It's not just for DNA inversion anymore. Mol. Micobiol. 6, 3257-3265. Garrett, D. S., Powers, R., Gronenborn, A. M. & Clore, G. M. (1991). A common sense approach to peak picking in two-, three-, and, four-dimensional spectra using automatic computer analysis of contour diagrams. J. Magn. Reson. 95, 214-220.

Solution Structure of NtrCC-term(3Ala) Gooley, P. R., Keniry, M. A., Roumen, A. D., Marsh, D. E., Keizer, D. W., Gayler, K. R. & Grant, B. R. (1998). The NMR solution structure and characterization of pH dependent chemical shifts of the b-elicitin, cryptogein. J. Biomol. NMR, 12, 523-534. Grasberger, B. L., Gronenborn, A. M. & Clore, G. M. (1993). Analysis of backbone dynamics of interleukin-8 by 15N relaxation measurements. J. Mol. Biol. 230, 364-372. Grzesiek, S. & Bax, A. (1992a). An ef®cient experiment for sequential backbone assignment of mediumsized isotopically enriched proteins. J. Magn. Reson. 99, 201-207. Grzesiek, S. & Bax, A. (1992b). Improved 3D triple-resonance NMR techniques applied to a 31 kDa protein. J. Magn. Reson. 96, 432-440. Handel, T. M. & Domaille, P. J. (1996). Heteronuclear (1H, 13C, 15N) NMR assignments and solution structure of the monocyte chemoattractant protein 1 (MCP-1) dimer. Biochemistry, 35, 6569-6584. Hinck, A. P., Archer, S., Qian, S. W., Roberts, A. B., Sporn, M. B., Weatherbee, J. A., Tsang, M. L.-S., Lucas, R., Zhang, B.-L., Wenker, J. & Torchia, D. A. (1996). Transforming growth factor b1: three-dimensional structure in solution and comparison with the X-ray structure of transforming growth factor b2. Biochemistry, 35, 8517-8534. Hyberts, S. G., Goldberg, M. S., Havel, T. F. & Wagner, G. (1992). The solution structure of eglin c based on measurements of many NOEs and coupling constants and its comparison with X-ray structures. Protein Sci. 1, 736-751. Ikura, M., Kay, L. E. & Bax, A. (1991). Improved threedimensional 1H-13C-1H spectroscopy of a 13Clabeled protein using constant-time evolution. J. Biomol. NMR, 1, 299-304. Johnson, R. C., Ball, C. A., Pfeffer, D. & Simon, M. I. (1988). Isolation of the gene encoding the Hin recombinational enhancer binding protein. Proc. Natl Acad. Sci. USA, 85, 3484-3488. Kay, L. E. & Bax, A. (1990). New methods for the measurement of NH-CaH coupling constants in 15 N-labeled proteins. J. Magn. Reson. 86, 110-126. Kay, L. E., Clore, G. M., Bax, A. & Gronenborn, A. M. (1990). Four-dimensional heteronuclear triple-resonance NMR spectroscopy of interleukin-1b in solution. Science, 249, 411-414. Keener, J. W. (1988). Nitrogen regulation in entric bacteria: protein kinase and phosphoprotein phosphatase activities of the NtrB and NtrC proteins, Ph.D. Thesis. University of California. Klose, K. E., North, A. K., Stedman, K. M. & Kustu, S. (1994). The major dimerization determinants of the nitrogen regulatory protein NtrC from enteric bacteria lie in its carboxyl-terminal domain. J. Mol. Biol. 241, 233-245. Koradi, R., Billeter, M. & WuÈthrich, K. (1996). MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graphics, 14, 51-55. Kostrewa, D., Granzin, J., Stock, D., Choe, H.-W., Labahn, J. & Saenger, W. (1992). Crystal structure Ê of the factor for inversion stimulation FIS at 2.0 A resolution. J. Mol. Biol. 226, 209-226. Kraulis, P. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946-950. Kustu, S., Santero, E., Keener, J., Popham, D. & Weiss, D. (1989). Expression of sigma-54 (ntrA)-dependent

Solution Structure of NtrCC-term(3Ala) genes is probably united by a common mechanism. Microbiol. Rev. 53, 367-376. Kustu, S., North, A. K. & Weiss, D. S. (1991). Prokaryotic transcriptional enhancers and enhancer-binding proteins. Trends Biochem. Sci. 16, 397-402. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). Procheck: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283-291. Liang, H., Petros, A. M., Meadows, R. P., Yoon, H. S., Egan, D. A., Wlater, K., Holzman, T. F., Robins, T. & Fesik, S. W. (1996). Solution structure of the DNA-binding domain of a human papillomavirus E2 protein: evidence for ¯exible DNA-binding regions. Biochemistry, 35, 2095-2103. Macura, S., Huang, Y., Suter, D. & Ernst, R. R. (1981). Two-dimensional chemical exchange and crossrelaxation spectroscopy of coupled nuclear spins. J. Magn. Reson. 43, 259-281. Marion, D., Ikura, M., Tschudin, R. & Bax, A. (1989). Rapid recording of 2D NMR spectra without phase cycling: application to the study of hydrogen exchange in proteins. J. Magn. Reson. 85, 393-399. Morett, E. & Bork, P. (1998). Evolution of new protein fuction: Recombinational enhancer FIS originated by horizontal gene transfer from the transcriptional regulator NtrC. FEBS Letters, 433, 108-112. Mori, S., Abeygunawardana, C., O'Neil, Johnson M. & van Zijl, P. C. M. (1995). Improved sensitivity of HSQC spectra of exchanging protons at short interscan delays using a new fast HSQC (FHSQC) detection scheme that avoids water saturation. J. Magn. Reson. 108B, 94-98. Morris, A. L., MacArthur, M. W., Hutchinson, E. G. & Thornton, J. M. (1992). Stereochemical quality of protein structure coordinates. Proteins: Struct. Funct. Genet. 12, 345-364. Neri, D., Szyperski, T., Otting, G., Senn, H. & WuÈthrich, K. (1989). Stereospeci®c NMR assignments of the methyl groups of valine and leucine in the DNAbinding domain of the 434 repressor by biosynthetic directed fractional 13C labeling. Biochemistry, 28, 7510-7516. Nilges, M. (1993). A calculation strategy for the structure determination of symmetric dimers by 1H NMR. Proteins: Struct. Funct. Genet. 17, 297-309. Ninfa, A. J., Reitzer, L. J. & Magasanik, B. (1987). Initiation of transcription at the bacterial glnAp2 promoter by puri®ed E. coli components is facilitated by enhancers. Cell, 50, 1039-1045. Nixon, B. T., Ronson, C. W. & Ausubel, F. M. (1986). Two-component regulatory systems responsive to environmental stimuli share strongly conserved domains with the nitrogen assimulation regulatory genes ntrB and ntrC. Proc. Natl Acad. Sci. USA, 83, 7850-7854. Nohaile, M., Kern, D., Wemmer, D., Stedman, K. & Kustu, S. (1997). Structural and functional analyses of activating amino acid substitutions in the receiver domain of NtrC: evidence for an activating surface. J. Mol. Biol. 273, 299-316. North, A. K. & Kustu, S. (1997). Mutant forms of the enhancer-binding protein NtrC can activate transcription from solution. J. Mol. Biol. 267, 17-36. North, A. K., Klose, K. E., Stedman, K. M. & Kustu, S. (1993). Prokaryotic enhancer-binding proteins re¯ect eukaryotic-like modularity: the puzzle of nitrogen regulatory protein C. J. Bacteriol. 175, 4267-4273.

1109 O'Donoghue, S. I., King, G. F. & Nilges, M. (1996). Calculation of symmetric multimer structures from NMR data using a priori knowledge of the monomer structure, co-monomer restraints and interface mapping: the case of leucine zippers. J. Biomol. NMR, 8, 193-206. Osuna, J., Soberon, X. & Morett, E. (1997). A proposed architecture for the central domain of the bacterial enhancer-binding proteins based on secondary structure prediction and fold recognition. Protein Sci. 6, 543-555. Osuna, R., Finkel, S. E. & Johnson, R. C. (1991). Identi®cation of two functional regions in Fis: the N-terminus is required to promote Hin-mediated DNA inversion but not lambda excision. EMBO J. 10, 1593-1603. Pace, C. N., Vajdos, F., Fee, L., Grimsley, G. & Gray, T. (1995). How to measure and predict the molar extinction coef®cient of a protein. Protein Sci. 4, 2411-2423. Pan, C. Q., Finkel, S. E., Cramton, S. E., Feng, J.-A., Sigman, D. S. & Johnson, R. C. (1996a). Variable structures of Fis-DNA complexes determined by ¯anking DNA-protein contacts. J. Mol. Biol. 264, 675-695. Pan, C. Q., Johnson, R. C. & Sigman, D. S. (1996b). Identi®cation of new binding sites by DNA scission with FIS-1,10-phenanthroline-copper (I) chimeras. Biochemistry, 35, 4326-4333. Pardi, A., Billeter, M. & WuÈthrich, K. (1984). Calibration of the angular dependence of the amide proton-C alpha proton coupling constants, 3JHNHa, in a globular protein. Use of 3JHNHa for identi®cation of helical secondary structure. J. Mol. Biol. 180, 741-751. Pelton, J. G., Torchia, D. A., Meadow, N. D., Wong, C. Y. & Roseman, S. (1991). 1H, 15N, and 13C NMR signal assignments of IIIGlc, a signal-transducing protein of Escherichia coli, using three-dimensional triple-resonance techniques. Biochemistry, 30, 10043-10057. Porter, S. C., North, A. K., Wedel, A. B. & Kustu, S. (1993). Oligomerization of NtrC at the glnA enhancer is required for transcriptional activation. Genes Dev. 7, 2258-2273. Porter, S. C., North, A. K. & Kustu, S. (1995). Mechanism of transcriptional activation by NtrC. In Twocomponent Signal Transduction (Hoch, J. A. & Silhavy, T. J., eds), pp. 147-158, American Society for Microbiology, Washington, DC. Rance, M., Sorensen, O. W., Bodenhausen, G., Wagner, G. & Ernst, R. R. (1983). Improved spectral resolution in COSY 1H NMR spectra of proteins via double quantum ®ltering. Biochem. Biophys. Res. Commun. 117, 479-485. Ravid, S., Matsumura, P. & Eisenbach, M. (1986). Restoration of ¯agellar clockwise rotation in bacterial envelopes by insertion of the chemotaxis protein CheY. Proc. Natl Acad. Sci. USA, 83, 7157-7166. Reitzer, L. J. & Magasanik, B. (1987). Ammonia assimulation and the biosynthesis of glutamine, glutamate, aspartate, asparagine, L-alanine, and D-alanine. In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F. C., ed.), pp. 302-320, American Society for Micorbiology, Washington, DC. Revet, B., Brahms, S. & Brahms, G. (1995). Binding of the transcription activator NRI to a supercoiled DNA segment imitates association with the natural enchancer: an electron microscopic investigation. Proc. Natl Acad. Sci. USA, 92, 7535-7539.

1110 Rombel, I., North, A., Hwang, I., Wyman, C. & Kustu, S. (1998). The bacterial enhancer-binding protein NtrC as a molecular machine. Cold Spring Harbor Symp. Quant. Biol. 63, 157-166. Safo, M. K., Yang, W. Z., Corselli, L., Cramton, S. E., Yuan, H. S. & Johnson, R. E. (1997). The transactivation region of the FIS protein that controls site speci®c DNA inversion contains extended mobile bhairpin arms. EMBO J. 16, 6860-6873. Schultz, S. C., Shields, G. C. & Steitz, T. A. (1991). Crystal structure of a CAP-DNA complex: the DNA is bent by 90  . Science, 253, 1001-1007. Shaka, A. J., Keeler, J. & Freeman, R. (1983). Evaluation of a new broadband decoupling sequence: Waltz16. J. Magn. Reson. 53, 313-340. Shaka, A. J., Barber, P. B. & Freeman, R. (1985). Computer-optimized decoupling scheme for wideband applications and low level operation. J. Magn. Reson. 64, 547-552. Shaka, A. J., Lee, C. J. & Pines, A. (1988). Iterative schemes for bilinear operators: application to spin decoupling. J. Magn. Reson. 77, 274-293. Sklenar, V., Piotto, M., Leppik, R. & Saudek, V. (1993). Gradient-tailored water suppression of 1H-15N HSQC experiments optimized to retain full sensitivity. J. Magn. Reson. 102A, 241-245. Szyperski, T., Neri, D., Leiting, B., Otting, G. & WuÈthrich, K. (1992). Support of 1H NMR assignments in proteins by biosynthetic directed fractional 13 C-labeling. J. Biomol. NMR, 2, 323-334. Talluri, S. & Wagner, G. (1996). An optimized 3D NOESY-HSQC. J. Magn. Reson. 112B, 200-205. Thompson, J. F. & Landy, A. (1998). Empirical estimation of protein-induced DNA bending angles: applications to lambda site-speci®c recombination complexes. Nucl. Acids Res. 16, 9687-9705. Thony, B. & Hennecke, H. (1989). The ÿ24/ÿ 12 promotor comes of age. FEMS Micorbiol. Rev. 63, 341-358. Volkman, B. F., Nohaile, M. J., Amy, N. K., Kustu, S. & Wemmer, D. E. (1995). Three-dimensional solution structure of the N-terminal receiver domain of NtrC. Biochemistry, 34, 1413-1424. Vuister, G. & Bax, A. (1992). Resolution enhancement and spectral editing of uniformly 13C-enriched proteins by homonuclear broadband 13C decoupling. J. Magn. Reson. 98, 428-435. Vuister, G. W. & Bax, A. (1993). Measurement of twoand three-bond proton to methyl carbon J couplings in proteins uniformly enriched with 13C. J. Magn. Reson. 102B, 228-231. Vuister, G. W., Clore, G. M., Gronenborn, A. M., Powers, R., Garrett, D. S., Tschudin, R. & Bax, A. (1993). Increased resolution and improved spectral quality in 4D 13C/13C-separated HMQC-NOESYHMQC spectra using pulsed ®eld gradients. J. Magn. Reson. 101B, 210-213. Walters, K. J., Dayie, K. T., Reece, R. J., Ptashne, M. & Wagner, G. (1997). Structure and mobility of the PUT3 dimer. Nature Struct. Biol. 4, 744-750.

Solution Structure of NtrCC-term(3Ala) Wintjens, R. & Rooman, M. (1996). Structural classi®cation of HTH DNA-binding domains and proteinDNA interaction modes. J. Mol. Biol. 262, 294-313. Wishart, D. S., Bigam, C. Y., Yao, J., Abildgaard, F., Dyson, H. J., Old®eld, E., Markley, J. L. & Sykes, B. D. (1995). 1H, 13C, and 15N chemical shift referencing in biomolecular NMR. J. Biomol. NMR, 6, 135140. WuÈthrich, K. (1986). NMR of Proteins and Nucleic Acids, John Wiley & Sons, New York. WuÈthrich, K., Billeter, M. & Braun, W. (1983). Pseudostructures for the 20 common amino acids for use in studies of protein conformations by measurements of intramolecular proton-proton distance contraints with nuclear magnetic resonance. J. Mol. Biol. 169, 949-961. Yuan, H. S., Finkel, S. E., Feng, J.-A., KaczorGrzeskowiak, M., Johnson, R. E. & Dickerson, R. E. (1991). The molecular structure of a wild-type and a mutant FIS protein: relationship between mutational changes and recombinational enhancer function on DNA-binding. Proc. Natl Acad. Sci. USA, 88, 95589562. Yuan, H. S., Wang, S. S., Yang, W.-Z., Finkel, S. E. & Johnson, R. C. (1994). The structure of FIS mutant Pro61Ala illustrates that the kink within the long alpha helix is not due to the presence of the proline residue. J. Biol. Chem. 269, 28947-28954. Zhang, W. & Gmeiner, W. H. (1996). Improved 3D gdHCACO and gd-(H)CACO-TOCSY experiments for isotopically enriched proteins dissolved in D2O. J. Biomol. NMR, 7, 247-250. Zwablan, C., Legault, P., Vincent, S. J. F., Greenblatt, J., Konrat, R. & Kay, L. E. (1997). Methods for measurement of inter-molecular NOEs by multinuclear NMR spectroscopy. J. Am. Chem. Soc. 119, 67116721.

Edited by P. E. Wright (Received 22 June 1999; received in revised form 20 August 1999; accepted 20 August 1999)

http://www.academicpress.com/jmb Supplementary Material, comprising Tables of chemical shift assignments, is available from JMB Online.