Phosphoprotein Crh-Ser46-P Displays Altered Binding to CcpA to Effect Carbon Catabolite Regulation*

In Gram-positive bacteria, the catabolite control protein A (CcpA) functions as the master transcriptional regulator of carbon catabolite repression/regulation (CCR). To effect CCR, CcpA binds a phosphoprotein, either HPr-Ser46-P or Crh-Ser46-P. Although Crh and histidine-containing protein (HPr) are structurally homologous, CcpA binds Crh-Ser46-P more weakly than HPr-Ser46-P. Moreover, Crh can form domain-swapped dimers, which have been hypothesized to be functionally relevant in CCR. To understand the molecular mechanism of Crh-Ser46-P regulation of CCR, we determined the structure of a CcpA-(Crh-Ser46-P)-DNA complex. The structure reveals that Crh-Ser46-P does not bind CcpA as a dimer but rather interacts with CcpA as a monomer in a manner similar to that of HPr-Ser46-P. The reduced affinity of Crh-Ser46-P for CcpA as compared with that of HPr-Ser46 P is explained by weaker Crh-Ser46-P interactions in its contact region I to CcpA, which causes this region to shift away from CcpA. Nonetheless, the interface between CcpA and helix α 2 of the second contact region (contact region II) of Crh-Ser46-P is maintained. This latter finding demonstrates that this contact region is necessary and sufficient to throw the allosteric switch to activate cre binding by CcpA.

Carbon catabolite repression/regulation (CCR) 2 is a global regulatory mechanism utilized by bacteria to select, out of a mixture of compounds, the carbon source providing the optimal growth advantage (1)(2)(3). CCR is mediated largely at the level of transcription. The master transcriptional regulator of CCR in bacilli and other Gram-positive bacteria with low GC content is the catabolite control protein A (CcpA) (4 -10). CcpA binds to catabolite responsive elements (cre) to mediate its effect (11,12). Approximately 10% of the Bacillus subtilis genome is under regulation by CcpA, underscoring its vital metabolic role (13).
CcpA is a member of the LacI-GalR family of transcription regulators (14). LacI-GalR proteins contain a 60-residue N-terminal DNA binding domain that connects to a larger C-terminal domain. The C-terminal domain consists of N-and C-subdomains connected by a hinge region. Movements about the hinge lead to "open" and "closed" states of the domain. For most LacI-GalR members, small molecule ligands act as either inducers or corepressors by binding in the cavity between the subdomains to stabilize the open or closed form (15)(16)(17)(18)(19)(20). Although some studies suggest that glucose 6-phosphate and fructose 1,6bisphosphate act as corepressors of CcpA, these data are not unequivocal (21)(22)(23). By contrast, multiple studies have demonstrated that the Ser 46 -phosphorylated form of HPr binds CcpA as a corepressor, activating CcpA to bind cre sites (6, 24 -26). The recent structure of the CcpA-(HPr-Ser 46 -P)-cre revealed the mechanism by which HPr-Ser 46 -P functions as a corepressor for CcpA (27). Strikingly, this DNA binding activation mechanism is different from those of PurR and LacI in that CcpA utilizes a two-component phosphoprotein that is induced and stabilized by closure of its N-and C-subdomains. This mechanism involves both a rotation of CcpA subdomains as well as a relocation of the key residue Thr 61 , which is located at the interface of the DNAbinding and corepressor-binding domains. The repositioning of this residue leads to a juxtaposition of the DNA-binding domains to permit hinge helix formation in the presence of cognate DNA (15,16). In addition to HPr, a structural and functional homologue, Crh (for catabolite repression HPr) has been identified, but only in bacilli (28). Crh shows 45% sequence identity with HPr but lacks the His 15 residue found in HPr that is phosphorylated to function in a phosphoenolpyruvate:sugar phosphotransferase system. Crh, instead contains a conservative substitution to glutamine at this position (29,30). However, like HPr, Crh does contain the Ser 46 residue that is phosphorylated by the enzyme HPr kinase/phosphorylase, and Crh has been demonstrated to function in CCR (28,31). Interestingly, Crh-Ser 46 -P can only partially substitute for HPr-Ser46-P in CCR, whereas HPr-Ser 46 -P can completely substitute for Crh-Ser 46 -P in this pathway (28,32). These findings appear to be explained by the finding that HPr is produced much more efficiently than Crh under strong conditions of CCR, resulting in up to 100-fold more HPr than Crh (33). Indeed, when bacteria utilize succinate or citrate as their major carbon source, the difference in HPr and Crh levels is only ϳ10-fold and under these conditions citM, a gene encoding the Mg 2ϩ -citrate transporter, is specifically repressed by Crh but not by HPr (33).
HPr and Crh have highly homologous sequences and are structurally similar, indicating that the phosphorylated forms of these proteins likely bind CcpA similarly and elicit the same allosteric DNA binding switch. Examination of the CcpA-(HPr-Ser 46 -P)-cre complex structure shows that, with the exception of two residues (HPr 3 Crh: H15Q, T20A), the expected CcpA binding interfaces of the two proteins should be identical. Modeling studies using the CcpA-(HPr-Ser 46 -P)-cre structure sug-gested that Crh residue Gln 15 would be able to interact with CcpA residue Asp 296 in a manner similar to the interaction observed between His 15 and Asp 296 in the CcpA-(HPr-Ser 46 -P)-cre structure (27). Interestingly, biochemical studies have revealed that Crh-Ser 46 -P binds CcpA with up to 10-fold reduction in affinity as compared with HPr-Ser 46 -P (34), thereby indicating differences in their binding mechanisms.
Previous structural studies on Crh reveal that it has essentially the same fold as HPr. However, a recent crystal structure of Crh showed a domain-swapped dimer and NMR studies have provided evidence that, at high concentrations, Crh can form a mixture of monomers and dimers (35,36). From these studies a dimer-dimer type of interaction between Crh and CcpA was suggested. By contrast, the CcpA-(HPr-Ser 46 -P)-cre structure, which showed two monomers of HPr-Ser 46 -P bound per CcpA dimer, indicates that dimerization is not likely relevant in the binding of Ser 46 -phosphorylated, HPr-like corepressors to CcpA (27). Moreover, recent biochemical studies, which examined the interaction of CcpA with HPr-Ser 46 -P and Crh-Ser 46 -P, indicate that both proteins bind CcpA as a monomer (34). The possible relevance of a Crh-Ser 46 -P dimerization in its binding to CcpA remains unclear and whether dimerization or other structural alterations of Crh explains the reduced binding affinity of Crh-Ser 46 -P for CcpA is not known. Therefore, to determine the molecular basis for Crh-Ser 46 -P binding to CcpA and thus, gain insight into its reduced affinity for CcpA as compared with HPr-Ser 46 -P, we determined the crystal structure of a CcpA-(Crh-Ser 46 -P)-cre complex to 2.96-Å resolution.

MATERIALS AND METHODS
Protein Preparation, Crystallization, and Data Collection-Bacillus subtilis and Bacillus megaterium CcpA and B. subtilis Crh-Ser 46 -P proteins were overexpressed and purified as described (27,34). Both B. subtilis and B. megaterium His-tagged CcpA were used in crystallization screens as sequence alignment of CcpA proteins revealed that the HPr-Ser 46 -P interacting surfaces of these proteins are identical (27). The B. megaterium and B. subtilis Crh proteins (both 88 residues) share 64% sequence identity but their predicted CcpA interacting surfaces share 100% sequence identity. 3 Data quality crystals were obtained using only B. megaterium CcpA, B. subtilis Crh-Ser 46 -P, and the 16-bp cre duplex with the sequence of one strand 5Ј-CTGTTAGCGCTTTCAG-3Ј. Crystals were grown at 298 K using the hanging drop vapor diffusion method by mixing the stoichiometric CcpA(dimer)-(Crh-Ser 46 -P)(2 monomers)-cre duplex complex 1:1 with a reservoir solution of 22% PEG MME 3350, 0.2 M sodium iodide and sealing the drop over 1 ml of the reservoir. The crystals are monoclinic, space group C2, with a ϭ 83.69 Å, b ϭ 158.10 Å, c ϭ 125.47 Å, and ␤ ϭ 100.73°. For cryoprotection, glycerol was added to a final concentration of 35%. X-ray intensity data were collected at the Advanced Light Source beamline 8.2.1 at 100 K, processed with MOSFLM, and scaled with SCALA (Table 1).
Structure Determination-The CcpA-(CrH-Ser 46 -P)-cre structure was determined by molecular replacement using the CcpA-(HPr-Ser 46 -P)-cre structure, with the solvent removed (27), as a search model and the program EPMR (37). Searching with a single subunit of CcpA, its bound HPr-Ser 46 -P corepressor, and a cre half-site produced three clear solutions: two of which formed complex CcpA-(CrH-Ser 46 -P)-cre dimer, and the third, a monomer, that when the crystallographic symmetry was applied produced a dimer. This starting model was first subjected to rigid body refinement, in which each CcpA subunit, Crh-Ser 46 -P molecule and DNA half-site, were treated as rigid units (38,39). This was followed by multiple cycles of simulated annealing and positional/thermal parameter refinement in CNS and rebuilding in O (38,39).

RESULTS AND DISCUSSION
Structure of the CcpA-(Crh-Ser 46 -P)-cre Complex-The CcpA-Crh-Ser 46 -P-cre DNA complex was crystallized using equimolar amounts of CcpA monomer, Crh-Ser 46 -P (monomer), and the 16-bp cre site (with one strand of the sequence 5Ј-CTGTTAGCGCTTTCAG-3Ј). This sequence was also used in the determination of the CcpA-(HPr-Ser 46 -P)-cre structure to allow comparison of the CcpA-DNA contacts when bound by the different CcpA-phosphoprotein corepressor complexes (27). The structure was solved by molecular replacement using the CcpA-(HPr-Ser 46 -P)-cre structure as a search model ("Materials and Methods," Fig. 1, A and B, and Table 1). The crystallographic asymmetric unit contains one dimeric CcpA-(Crh-Ser 46 -P)-cre complex and one monomeric complex in which crystallographic symmetry generates the dimer. Thus, the structure provides three independent views of CcpA-(Crh-Ser 46   other two CcpA subunits, residues 2-84 of the three Crh-Ser 46 -P subunits, all nucleotides of the three 16-bp cre strands, 11 iodide atoms, and 46 solvent molecules. The R work and R free are 23.2 and 29.8%, respectively, to 2.96-Å resolution and selected model statistics (40) are given in Table 1.
The structure of CcpA, like LacI-GalR members PurR and LacI, consists of a DNA-binding domain (residues 1-59) and a dimerization/ corepressor (Crh-Ser 46 -P)-binding domain (residues 60 -332) (15)(16)(17)(18)(19). The DNA-binding domain can be divided into two regions, ␣1 through ␣3, which form the HTH containing three-helix bundle, and ␣4, which forms the minor groove binding hinge helix (Fig. 1A). As is characteristic of the LacI-GalR proteins, there is a dramatic kink in the conserved central CpG step of the cre DNA, which is induced by the two conserved symmetry related leucines (Leu 56 ) located on the hinge helices. The conformation of CcpA in the CcpA-(Crh-Ser 46 -P)-cre structure is essentially identical to that found in the CcpA-(HPr-Ser 46 -P)-cre structure; superimpositions of the C␣ atoms of each CcpA subunit from the CcpA-(Crh-Ser 46 -P)-cre structure onto a CcpA subunit from the CcpA-(HPr-Ser 46 -P)-cre structure results in root mean square deviations (r.m.s. deviations) of ϳ0.70 Å. Similar overlays of the CcpA dimers from each structure result in r.m.s. deviations of ϳ1.0 Å for all C␣ atoms. The larger deviation in superimposition of the dimer reflects the flexible attachment between helix ␣3 and the hinge helix, ␣4 (see below).
The CcpA-(Crh-Ser 46 -P)-cre structure reveals that each Crh-Ser 46 -P binds the surface of CcpA with a stoichiometry of one Crh-Ser 46 -P molecule per CcpA subunit. As anticipated, the CcpA-(Crh-Ser 46 -P) binding interface is similar to the CcpA-(HPr-Ser 46 -P) interface, but not identical (Fig. 1A) (6,27). Crh-Ser 46 -P binds CcpA as a monomer in all three crystallographically independent complexes, clearly demonstrating that a Crh dimer is not utilized in formation of the CcpA-(Crh-Ser 46 -P)-cre ternary complex. Moreover, modeling of the domainswapped Crh dimer into the complex reveals that steric clash between the N-subdomain of CcpA and the second subunit of Crh (data not shown) would preclude an identical binding mode.
Comparison of the CcpA-bound HPr-Ser 46 -P and Crh-Ser 46 -P Molecules-The three crystallographically independent Crh-Ser 46 -P proteins bound to CcpA have the same structures as evidenced from the average r.m.s. deviation of 0.51 Å for the pairwise superimpositions of all 84 Crh C␣ atoms. Crh-Ser 46 -P contains three ␣-helices (␣1, residues 18 -27; ␣2, residues 47-52; ␣3, residues 70 -81) and four ␤-strands (␤1, residues 2-6; ␤2, residues 32-37; ␤3, residues 41-44; ␤4, residues 61-65) (Fig. 1). Comparison of the Crh-Ser 46 -P structure in our ternary complex to that of HPr-Ser 46 -P in the CcpA-(HPr-Ser 46 -P)-cre structure results in a r.m.s. deviation of 1.2 Å for 80 corresponding C␣ atoms, showing that Crh-Ser 46 -P and HPr-Ser 46 -P adopt essentially the same structure when bound to CcpA. Only two significant differences are found between the HPr-Ser 46 -P and Crh-Ser 46 -P structures bound to CcpA. The first is the conformation of helix ␣3, which is tilted slightly differently in the two proteins and the second is the absence of HPr-Ser 46 -P ␤ strand ␤5 (residues 86 -88) from Crh-Ser 46 -P. These secondary elements are located distal to the regions that interact with CcpA and likely have little impact on Crh-Ser 46 -P and HPr-Ser 46 -P binding to CcpA.
Two structures of Crh, but not Crh-Ser 46 -P, have been reported; one determined by NMR and the other by crystallography. The NMR structure shows a monomer with the same overall fold as Crh in our structure (36). However, the NMR analysis suggested that Crh dimerizes at high concentrations and a subsequent crystal structure revealed a domainswapped dimer in which the N-terminal ␤ strand, ␤1, exchanges subunits (35). Superimposition of the C␣ atoms of our Crh-Ser 46 -P structure onto the corresponding C␣ atoms of a Crh subunit in the domain-swapped structure, excluding residues 1-12, which are involved in domain exchange, results in r.m.s. deviation of 1.1 Å. Intriguingly, the crystal structure of Crh-Ser 46 -P is also identically domain swapped. 4 What role, if any, Crh dimerization may play in vivo is unclear. However, our structure, obtained under high protein concentrations (200 M) clearly reveals that the monomer of Crh-Ser 46 -P functions as a corepressor for CcpA, a finding consistent with recent studies examining CcpA binding to Crh-Ser 46 -P and HPr-Ser-46 -P by surface plasmon resonance (34).
Flexible DNA Binding by the CcpA HTH Elements-The CcpA-(Crh-Ser 46 -P)-cre structure provides three crystallographically independent views of the interaction between CcpA and the 16-bp cre. CcpA, like other LacI-GalR proteins, kinks its DNA binding site to allow formation of operator-specific, HTH-major groove contacts (41,42). As expected, these contacts are essentially identical to those observed in the CcpA-(HPr-Ser 46 -P)-cre structure, in which the same oligodeoxynucleotide was used in crystallization (27) (Fig. 2A). The global DNA bend angle, induced by partial interaction of the dyad-related hinge helix residues Leu 55 , the "leucine levers," 31°, is the smallest bend angle observed thus far for a LacI-GalR protein; the bend angles of the DNA bound by PurR and LacI are ϳ50 and ϳ40°, respectively, whereas the global bend angle observed in the CcpA-(HPr-Ser 46 -P)-cre structure, which, again, is the same DNA site as in the CcpA-(Crh-Ser 46 -P)-cre structure, is 35° (15)(16)(17)(18)(19)27).
The smaller DNA bend angle observed in the CcpA-(Crh-Ser 46 -P)cre structure as compared with the other LacI proteins appears to be the result of the different docking modes of their HTH motifs onto the major grooves rather than any differences in the partial intercalation of the hinge helices. Superimpositions of the dimer/corepressor binding domains reveal that the positions of the hinge helices (as well as the DNA around the central CpG step) are highly conserved structurally (Fig. 2B). By contrast, the three-helix bundle, composed of the HTH motif and ␣3, rotate as a unit about a flexible loop that links helices ␣3 and ␣4. Such flexibility permits the CcpA HTH motifs to adjust individ-  ually to accommodate to the precise major groove sequence of the DNA as well as the conformation of the DNA, the latter of which may be influenced by the adjacent DNA or adjacently bound proteins. Similar plasticity has been observed in structures of the Lac repressor (41,42).
However, there appears to be a thermodynamic cost associated with such binding flexibility as CcpA binds the amyE cre nearly 12-fold more tightly than the xyl cre, both of which are responsive to CCR (43). This plasticity allows CcpA to bind half-sites with altered sequences and provides an explanation for how CcpA can bind such a large number of degenerate cre elements that are found in different contexts throughout bacilli genomes (43,44).

CcpA-Crh-Ser 46 -P Interactions-Each
Crh-Ser 46 -P monomer docks onto a CcpA subunit in a manner similar to that utilized by HPr-Ser 46 -P. Indeed, the primary interaction interface is forged between Crh-Ser 46 -P residues located on helices ␣1 and ␣2 and the turn between ␤ strand ␤1 and helix ␣1 and the surface exposed region of CcpA that includes helices I and IX from the N-subdomain of CcpA. Thus, CcpA interaction interfaces of HPr-like proteins can be broken into two regions: contact regions I (CRI), encompassing residues 15-28 (helices ␣1 and the preceding loop) and contact region II (CRII), comprised primarily of ␣2 and phosphorylated residue Ser 46 (Fig. 3).
Contact Region II: Conserved CcpA-HPr/Crh Interactions-Residues from CRII of Crh-Ser 46 -P include the phosphoresidue Ser 46 -P, which is located one residue N-terminal to ␣2, and the remainder of ␣2. Residues from this region are completely conserved between HPr and Crh and in addition to Ser 46 -P include Ile 47 , Met 48 , and Met 51 (Fig. 3, A and B). As in the CcpA-(HPr-Ser 46 -P)-cre structure, Crh-Ser 46 -P residue Ser 46 -P is contacted specifically by two CcpA basic residues, Arg 303 and Lys 307 , which are located near the center of CcpA helix IX (Fig. 4A). This interaction is buttressed by Tyr 89 . CcpA cannot bind HPr-like proteins at physiologically relevant concentrations unless they are phosphorylated on residue Ser 46 (6,26,34). Such phosphorylation of Ser 46 increases the binding affinity of CcpA for cre DNA, indicating the critical nature of this interaction in both binding specificity and affinity (26,34). In addition, several hydrophobic residues on ␣2 interact with hydrophobic residues on CcpA helix IX to form a water-free interface. These Crh-Ser 46 -P residues, which are conserved in HPr, include Ile 47 , Met 48 , and Met 51 and interact with CcpA residues Ala 299 , Val 300 , and Leu 304 (Fig. 3B).
Contact Region I: Discrimination in the Binding of HPr-like Proteins to CcpA-CRI, consisting of the loop-␣1 motif, contains loop residues Gln 15 and Ala 16 and ␣1 residues Arg 17 , Ala 20 , Val 23 , and Gln 24 (Fig. 3). Crh-Ser 46 -P residues Val 23 and Gln 24 make similar contacts with CcpA in the CcpA-(HPr-Ser 46 -P)-cre and CcpA-(Crh-Ser 46 -P)-cre structures. However, unlike CRII, in which all CcpA interacting residues are conserved between HPr and Crh, CRI contains two positions, 15 and 20, that differ between HPr and Crh. In all Crh proteins sequenced thus far these residues are glutamine and alanine, whereas in all HPr proteins they are histidine and threonine (Fig. 3A). In the CcpA-(HPr-Ser 46 -P)cre structure, His 15 of HPr hydrogen bonds to the side chain of CcpA residue Asp 296 (N ␦1 -O ␦1 , 2.5 Å), whereas the O ␦2 of Asp 296 engages in a hydrogen bond with the amide nitrogen of HPr residue Ala 16 . Just as the interactions between HPr CRII residue Ser 46 -P and CcpA basic residues Arg 303 and Lys 307 serve as key tethering points for HPr-Ser 46 -P CRII, the interactions between HPr-Ser 46 -P residues His 15 and Ala 16 and CcpA residue Asp 296 , serve as the anchor for HPr-Ser 46 -P contact region I to CcpA. In addition, these interactions function to partition the high energy CCR and low energy phosphoenolpyruvate:sugar phosphotransferase pathways because CcpA cannot bind HPr-His15-P and HPr-Ser 46 -P cannot bind the phosphoenolpyruvate:sugar phosphotransferase enzyme, E1 (45).
We predicted that the conservative H15Q substitution should retain the key hydrogen bond with CcpA residue Asp 296 (27). Surprisingly, only two of the three crystallographically independent CcpA-Crh-Ser 46 -P complexes show an interaction between Crh-Ser 46 -P residue Gln 15 and CcpA residue Asp 296 . Moreover, these contacts are weak (between 3.5 and 4.5 Å). Indeed, the structure shows that the primary contact to Crh-Ser 46 -P residue Gln 15 is from the side chain of CcpA residue Arg 324 (Figs. 3, B and C, and 4A). By contrast, in the CcpA-(HPr-Ser 46 -P)-cre complex, Arg 324 is greater than 5.0 Å from His 15 (Fig. 3C). In the CcpA-(Crh-Ser 46 -P)-cre structure the Arg 324 side chain rotates down into the CcpA-Crh interface to hydrogen bond with Gln 15 (aver-  ; Q8ENL6). Shown below the alignment are the secondary structure elements of Crh with ␤ strands represented as arrows and ␣ helices as rectangles. The secondary structural elements and loop region not involved in CcpA binding are colored black and gray, respectively. Strand ␤5, which is present in HPr but absent in Crh, is shown as a gray arrow. Secondary structural elements located in contact regions I and II are colored green and red, respectively. Residues that interact with CcpA are identical between Crh-Ser 46 -P and HPr-Ser 46 -P are indicated by black asterisks under the alignment, whereas the two residues that are different between Crh and HPr and make different contacts to CcpA are indicated by orange asterisks. B, view of the CcpA-(Crh-Ser 46 -P) interface. One CcpA subunit is gray and Asp 99 Ј from the other CcpA subunit is blue. CcpA helices I and IX are labeled and Crh-Ser 46 -P contact regions I and II are colored green and red, as in Fig. 3A. Residues from the elements of each protein that interact are shown as colored sticks and labeled. C, superimposition of the loop-␣1 contact regions of HPr-Ser 46 -P (yellow) and the three Crh-Ser 46 -P molecules in the asymmetric unit of the CcpA-(Crh-Ser 46 -P)-cre structure (blue, green, and red). The shortest distances (Å) between residue 15 of Crh and HPr and residues Asp 296 and Arg 324 of CcpA are indicated as solid lines of respective color. age contact distance of 3.1 Å) (Fig. 3C). Interestingly, sequence alignments show that Arg 324 is conserved only in CcpA proteins from bacilli. This suggests that the Arg 324 -Gln 15 interaction represents a bacillispecific CcpA-(Crh-Ser 46 -P) contact, which is supported further by the fact that Crh has, thus far, been identified only in bacilli. By contrast to the HPr-Ser 46 -P His 15 interaction with CcpA residue Asp 296 , the interaction between Crh-Ser 46 -P residue Gln 15 and CcpA residue Arg 324 is far weaker. This is evidence by the long contact distances between Gln 15 and Arg 324 as well as the poorer electron density and higher B-factors for these residues in all independent CcpA-(Crh-Ser 46 -P)-cre structures. Interestingly, CcpA does not engage in a similar contact to His 15 of HPr-Ser 46 -P, although Arg 324 is free to do so. The underlying reason is unknown, but perhaps might originate in the residual positive electrostatic nature of His 15 , which would repel the guanidinium group of Arg 324 .
The T20A substitution in Crh is the only other difference between the CcpA interacting residues of HPr-Ser 46 -P and Crh-Ser 46 -P (Fig. 3A). In the CcpA-(HPr-Ser 46 -P)-cre structure Thr 20 hydrogen bonds to the phenolic side chain of CcpA residue Tyr 295 . Obviously, this contact is not possible in the CcpA-(Crh-Ser 46 -P)-cre complex. The last key CcpA interacting residue in CRI is Arg 17 . In the CcpA-(HPr-Ser 46 -P)-cre structure, the guanidinium side chain of Arg 17 makes two contacts to residues Asp 69 Ј and Asp 99 Ј, which are located on the other subunit of the CcpA dimer. These "cross-contacts" have been postulated to be important in the allosteric DNA-binding mechanism of CcpA. In the CcpA-(Crh-Ser 46 -P)-cre structure, the cross-contacts are weakened as Arg 17 interacts with only side chain of Asp 99 Ј (Fig. 3B). Thus, the interactions between CcpA and CRII of both HPr-Ser 46 -P and Crh-Ser 46 -P are highly conserved, but the contacts between CcpA and the CRI of these CCR coregulators are different.
Basis for Differential Binding of Crh-Ser 46 -P and HPr-Ser 46 -P to CcpA-The CcpA-(Crh-Ser 46 -P)-cre structure demonstrates that although Crh-Ser 46 -P and HPr-Ser 46 -P bind CcpA in the same manner globally, there are several differences, specifically in the CRI-CcpA interface, that may explain the reduced affinity of Crh-Ser 46 -P for CcpA compared with HPr-Ser 46 -P (Fig. 4A). Underscoring these differences, superimpositions of the C␣ atoms of one CcpA subunit of the CcpA-(Crh-Ser 46 -P)-cre structure onto those of a CcpA subunit of the CcpA-(HPr-Ser 46 -P)-cre structure reveals that whereas the CRII of HPr and Crh dock nearly identically onto CcpA, the CRIs do not. Specifically, the CRI of Crh-Ser 46 -P is shifted away from the CcpA molecule as compared with the corresponding region of HPr-Ser 46 -P (Fig. 4B). This relocation results from the loosening of interactions between loop-␣1 (CRI) residues of Crh-Ser 46 -P to CcpA, primarily the weaker Gln 15 -Arg 324 contact, and the loss of the Thr 20 -Tyr 295 hydrogen bond. The shift of CRI away from CcpA also explains the loss of the cross-contact from Arg 17 to CcpA residue Asp 69 Ј, which in this conformation is too far (Ͼ4 Å) to form a meaningful interaction. The different docking mode of Crh-Ser 46 -P is highlighted quantitatively by the buried surface area between CcpA and Crh-Ser46-P, which, whereas still sizeable, is 100 Å 2 less than that observed between HPr-Ser 46 -P and CcpA (1300 Å 2 compared with 1400 Å 2 ). An additional contribution to the different binding affinities of Crh-Ser 46 -P and HPr-Ser 46 -P to CcpA may be the difference in the unliganded/liganded states of these phosphoproteins, if Crh actually exists in a monomer to dimer equilibrium in vivo. Indeed, we find that the domain-swapped Crh dimer would not be competent for CcpA binding. However, the physiological relevance, if any, of the domainswapped Crh dimer remains to be determined.
Crh-Ser 46 -P Is a Functional Corepressor for CcpA-Binding of HPr-Ser 46 -P or Crh-Ser 46 -P to CcpA allows the regulator to bind cre sites with high affinity and specificity. The CcpA-(HPr-Ser 46 -P)-cre and apo-CcpA structures reveal that HPr-Ser 46 -P mediates a novel two-component allosteric DNA-binding activation mechanism that involves rotation of the CcpA subdomains as well as a direct, coregulator-induced relocation of a pivot-point residue, Thr 61 , located between the N-domain and the DNA-binding domain of CcpA (27). Key to the relocation of Thr 61 is the interaction between Ser 46 -P and CcpA residue Arg 303 , which causes a rotation of Arg 303 and a concomitant shift in the position of Tyr 89 . This movement is translated to the entire Tyr 89 loop, which causes the hydrogen bond between Tyr 91 and Thr 306 to break. Tyr 91 then rotates toward the CcpA dimer interface, dislodging Thr 61 (27). The final result is the juxtaposition of the DNA-binding hinge regions, allowing formation of the hinge helices in the presence of cognate DNA. What is notable about this activation mechanism is the principal role played by residues of helix ␣2. Overlays of the CcpA-(Crh-Ser 46 -P)-cre and CcpA-(HPr-Ser 46 -P)-cre structures show that although ␣1 of the Crh complex is significantly shifted as compared with ␣1 in the HPr complex, ␣2 remains similarly docked against CcpA (Fig. 4B). Moreover, ␣2 residues of both Crh and HPr make essentially identical contacts to CcpA. Therefore, despite the weakening of the CRI interface, the interactions between the CRII of Crh-Ser 46 -P and CcpA remain adequately positioned to flip the DNA-binding switch to "on." This finding provides an atomic explanation for data showing that although Crh-Ser 46 -P binds CcpA with lower affinity, this phosphoprotein is fully capable of activating CcpA to bind DNA and hence to play a significant role in CCR.
Whereas HPr-Ser 46 -P and Crh-Ser 46 -P are clearly physiologically relevant corepressors for CcpA, recent data suggests that Glu-6-P and Fru-1,6-P may act as adjunct corepressors (34). These studies reveal that Glu-6-P and Fru-1,6-P can enhance the DNA binding of the CcpA-(HPr-Ser 46 -P) complex. Interestingly, however, they do not enhance DNA binding by the CcpA-(Crh-Ser 46 -P) complex. These results could be explained if Crh-Ser 46 -P binding somehow occludes the small molecule effector binding pocket of CcpA. Examination of the CcpA-(Crh-Ser 46 -P)-cre structure shows no obvious occlusion of the pocket by Crh-Ser 46 -P, however, the C terminus of Crh-Ser 46 -P is near the putative small molecule binding pocket of the other subunit of the CcpA dimer (Fig. 5). Intriguingly, the last few residues (81-84) of Crh-Ser 46 -P, which are acidic, are mostly disordered in the structure, suggesting the possibility that these residues could block the pocket or interfere electrostatically with binding of the negatively charged Fru-6-P and Glu-6-P. The C-terminal regions of Crh and HPr show the most structural divergence between the two proteins. Instead of the extended C-terminal coil found in Crh, HPr contains a short turn followed by ␤5, which folds back into the core of HPr, effectively preventing blockage of the CcpA pocket by these residues. In addition, unlike Crh, the C terminus of HPr is not negatively charged. Although these contrasting properties of the C-terminal regions of HPr and Crh could explain the differential CcpA DNAbinding enhancement by Glu-6-P and Fru-6-P, additional data are clearly needed to understand the roles of these adjunct corepressors and the conditions under which they may function.
In conclusion, the structure of the CcpA-(Crh-Ser 46 -P)-cre complex reveals that Crh-Ser 46 -P does not bind CcpA as a dimer but rather interacts with CcpA as a monomer in a manner similar to that of HPr-Ser 46 -P. The reduced affinity of Crh-Ser 46 -P for CcpA as compared with that of HPr-Ser 46 -P is explained readily by weaker Crh-Ser 46 -P interactions of its CRI to CcpA, which causes this region to shift away from CcpA. Nonetheless, the interface between CcpA and helix ␣2 (CRII) of Crh-Ser 46 -P is maintained and thereby demonstrates this contact region is necessary and sufficient to throw the allosteric switch and activate cre binding by CcpA. Finally, this study, which reveals how small differences in protein-protein interfaces can have a significant impact on the structure of a complex, underscores the need to take great caution in modeling protein-protein interactions even when a highly homologous structure is available.