Solution Structure of the IIAChitobiose-IIBChitobiose Complex of the N,N′-Diacetylchitobiose Branch of the Escherichia coli Phosphotransferase System*

The solution structure of the IIA-IIB complex of the N,N′-diacetylchitobiose (Chb) transporter of the Escherichia coli phosphotransferase system has been solved by NMR. The active site His-89 of IIAChb was mutated to Glu to mimic the phosphorylated state and the active site Cys-10 of IIBChb was substituted by serine to prevent intermolecular disulfide bond formation. Binding is weak with a KD of ∼1.3 mm. The two complementary interaction surfaces are largely hydrophobic, with the protruding active site loop (residues 9–16) of IIBChb buried deep within the active site cleft formed at the interface of two adjacent subunits of the IIAChb trimer. The central hydrophobic portion of the interface is surrounded by a ring of polar and charged residues that provide a relatively small number of electrostatic intermolecular interactions that serve to correctly align the two proteins. The conformation of the active site loop in unphosphorylated IIBChb is inconsistent with the formation of a phosphoryl transition state intermediate because of steric hindrance, especially from the methyl group of Ala-12 of IIBChb. Phosphorylation of IIBChb is accompanied by a conformational change within the active site loop such that its path from residues 11–13 follows a mirror-like image relative to that in the unphosphorylated state. This involves a transition of the φ/ψ angles of Gly-13 from the right to left α-helical region, as well as smaller changes in the backbone torsion angles of Ala-12 and Met-14. The resulting active site conformation is fully compatible with the formation of the His-89-P-Cys-10 phosphoryl transition state without necessitating any change in relative translation or orientation of the two proteins within the complex.

The bacterial phosphoenolpyruvate:sugar phosphotransferase system (PTS) 3 couples a phosphorylation cascade involv-ing a sequential series of bimolecular protein-protein complexes to active sugar translocation across the membrane and to regulation of an array of cellular processes, including carbon catabolite repression (1)(2)(3)(4)(5)(6). The first two steps of the PTS, involving autophosphorylation of enzyme I (EI) by phosphoenolpyruvate and subsequent phosphoryl transfer to the histidine phosphocarrier protein (HPr), are common to all branches of the pathway. The downstream components of the PTS comprise four major classes of sugar-specific enzymes II corresponding to the glucose (Glc), mannitol (Mtl), mannose (Man), and lactose/chitobiose (Chb) branches of the PTS. The enzymes II are generally organized into two cytoplasmic domains (IIA and IIB), and one transmembrane domain (IIC), which may or may not be covalently linked to one another. The phosphoryl group is transferred from HPr to IIA, from IIA to IIB and finally from IIB onto the incoming sugar bound to IIC. Despite their similar organization, the IIA and IIB domains of the different sugar-specific branches of the PTS bear no sequence similarity to one another and, with the exception of IIB Mtl (7,8) and IIB Chb (9 -11), no structural similarity either. Whereas structures of many of the individual cytoplasmic components of the PTS have been solved either by crystallography (9,(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25) or NMR (7, 8, 10, 11, 26 -32), the complexes of the PTS have proved refractory to crystallization, presumably because of their weak and transient nature. Weak binding, however, is not an impediment to NMR spectroscopy, and over the last 10 years we have solved the solution structures of the N-terminal domain of enzyme I (EIN) complexed to HPr (33), and the IIA-HPr and IIA-IIB complexes of the glucose, mannitol, and mannose branches of the PTS (30, 31, 34 -37). These complexes provide a paradigm for understanding the structural basis of protein-protein interactions and how individual proteins can recognize multiple, structurally dissimilar, partners.
In the present report, we present the solution structure of the IIA-IIB complex of the Escherichia coli N,NЈ-diacetylchitobiose-specific enzyme II (II Chb ), a representative of the lactose/ chitobiose branch of the PTS (38 -41). The A, B, and C domains of II Chb are encoded by a single operon and expressed as three individual proteins. The solution NMR structure of E. coli IIA Chb (26) and the crystal structure of the related family mem-ber IIA Lac from Lactobacillus lactis (21) are symmetric trimers of ϳ35 kDa with three equivalent IIB binding sites. The active site residue, His-89, is located deep within a crevice formed by the interface of two helices from adjacent subunits. IIB Chb is a small 11-kDa protein that has been studied by both x-ray crystallography (9) and NMR (10,11) and is structurally similar to IIB Mtl (7,8), despite the absence of any significant sequence similarity (Ͻ10%). The active site residue, Cys-10, of IIB Chb is located within an 8-residue protruding loop (residues 9 -16) whose conformation is very similar to that of the low molecular weight protein-tyrosine phosphatases (42), including hydrogen-bonding interactions in the phosphorylated state between the phosphoryl group and backbone amide protons (11). (Note, throughout the text residues of IIB Chb are shown in italics.) Wild-type IIA Chb is highly prone to nonspecific aggregation promoted by a disordered 13-residue N-terminal tail and by metal ions that coordinate three buried aspartic acid residues (Asp-92), one from each subunit, at the center of the trimer interface (26). It has previously been shown that aggregation can be completely eliminated by removing the N-terminal tail and mutating Asp-92 to Leu to generate a mutant, which we refer to hereafter as IIA Chb* (26). In the present work, we made use of an active site H89E mutation introduced into IIA Chb* to mimic the phosphorylated state. For IIB Chb we introduced a C10S mutation to prevent intermolecular disulfide bridge-mediated dimer formation (9,10). The IIA Chb* -IIB Chb complex is transient and weak with an equilibrium dissociation constant (K D ) in the millimolar range. The affinity of IIA Chb* (H89E) for IIB Chb (C10S) is a factor of ϳ1.5 higher than that of IIA Chb* (K D ϳ1.3 versus ϳ2 mM) making the phosphomimetic mutant more suitable for NMR structural studies by increasing the population of the complex at the concentrations used in the NMR experiments. The structure of the IIA Chb* (H89E)-IIB Chb (C10S) complex reveals the structural basis of specific recognition and the interactions involved in phosphoryl transfer.

EXPERIMENTAL PROCEDURES
Protein Expression and Mutagenesis-Genes encoding IIA Chb* (corresponding to a N⌬13/D92L mutant of wild-type IIA Chb ) and IIB Chb (kindly provided by Dr. Saul Roseman, Johns Hopkins University, Baltimore) were cloned into the pET-11 vector. Additional H89E and C10S mutations of the active site residues of IIA Chb* and IIB Chb , respectively were introduced using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA). The H89E mutation in IIA Chb* was designed to mimic the charge effects of phosphorylation of His-89, and the C10S mutation of IIB Chb was introduced to prevent any potential complications arising from possible intermolecular disulfide bridge formation.
The IIA Chb* , IIA Chb* (H89E), IIB Chb , and IIB Chb (C10S) plasmids were introduced into E. coli BL21(DE3) (Novagen) cells for protein expression and induced at an A 600 ϳ0.8 with 1 mM isopropyl-␤-D-thiogalactopyranoside at 37°C. Cells were grown in either Luria-Bertani medium or minimal medium (in either H 2 O or D 2 O) with 15  IIA Chb* and IIA Chb* (H89E) Purification-The cell pellet was resuspended in 50 ml of buffer A (20 mM Tris, pH 8.0, 1 mM EDTA, 0.2 mM sodium azide) with 1 mM phenylmethylsulfonyl fluoride. The cell suspension was lysed by two passages through a microfluidizer at 15,000 -23,000 psi and centrifuged at 75,600 ϫ g for 30 min. The supernatant was loaded onto a 5-ml HiTrap QFF column (Amersham Biosciences), and IIA Chb* was eluted with a gradient of buffer B (20 mM Tris, pH 8.0, 1 mM EDTA, 0.2 mM sodium azide, 1 M NaCl). The eluted protein was subsequently denatured with 4 M guanidine-HCl for 15 min. The protein solution was then dialyzed against 2 liters of buffer A overnight. After centrifugation of the dialyzed solution to remove precipitated proteins, the supernatant was purified by size exclusion chromatography on a Superdex-75 column (Amersham Biosciences) in buffer C (20 mM Tris, pH 8.0, 1 mM EDTA, 0.2 mM sodium azide, 0.5 M NaCl). The fractions containing IIA Chb* were exchanged into buffer A using an Amicon Ultra-15 (Millipore) filter, loaded onto a mono Q 10/100GL column (Amersham Biosciences), and eluted with a gradient of buffer B.
IIB Chb and IIB Chb (C10S) Purification-The cell pellet was resuspended in 50 ml of buffer D (20 mM sodium phosphate, pH 4.5, 0.2 mM sodium azide) with 1 mM phenylmethylsulfonyl fluoride. The cell suspension was lysed by two passages through a microfluidizer at 15,000 -23,000 psi and centrifuged at 75,600 ϫ g for 30 min. The supernatant was loaded onto a 5-ml HiTrap SP FF column (Amersham Biosciences), and IIB Chb (C10S) was eluted with a gradient of buffer E (20 mM sodium phosphate, pH 4.5, 0.2 mM sodium, 1 M NaCl). The eluted protein was further purified by size exclusion chromatography in buffer F (20 mM sodium phosphate, pH 6.5, 0.2 mM  15 N correlation spectroscopy, which showed large chemical shift perturbations of residues 10 -16 comprising the active site loop (11), relative to unphosphorylated IIB Chb . Full-length Enzyme I and HPr were expressed and purified as described previously (44).
NMR Data Collection and Analysis-All NMR samples were prepared in a buffer of 20 mM sodium phosphate, pH 6.5, 100 mM NaCl, 0.2 mM sodium azide, and either 90% H 2 O/10% D 2 O or 99.99% D 2 O. IIA Chb* is a symmetric trimer with three equivalent binding sites for IIB Chb (26). To achieve optimal linewidths for NMR spectroscopy, a 1:1 mixture of IIA Chb* (H76E) trimer to IIB Chb (C10S) monomer was employed. NMR spectra were recorded at 20 and 35°C on Bruker DMX500, DMX600, DRX600, DRX800, and DRX900 spectrometers equipped with either x-, y-, and z-shielded gradient triple resonance probes or z-shielded gradient triple resonance cryoprobes. Spectra were processed with the NMRPipe package (45) and analyzed using the program PIPP (46). Sequential and side chain assignments of IIA Chb* (H89E) and IIB Chb (C10S) were derived from the following three-dimensional double and triple resonance through-bond correlation experiments (47)(48)(49) 13 C-separated, and 13 C/ 13 C-separated nuclear Overhauser enhancement (NOE) experiments were used to facilitate side chain assignments (47,48). Backbone 1 D NH residual dipolar couplings (RDC) were obtained from the difference in 1 J HN scalar couplings measured in dilute liquid crystalline medium (phage pf1 (50,51)) and isotropic (water) medium, measured using two-dimensional in-   (71), where D obs are the observed RDCs, and D calc are the calculated RDCs obtained by singular value decomposition against the coordinates of the indicated protein (72). The values for the magnitude of the principal component of the alignment tensor (D a NH ) and the rhombicity () are as follows. For the RDCs measured on free IIB Chb (C10S), the values are 25.6 Hz and 0.4, respectively, in 100 mM NaCl and 10 mg/ml pf1, and Ϫ8.9 Hz and 0.3, respectively, in 400 mM NaCl and 17 mg/ml phage pf1; the normalized scalar product between the alignment tensors (80) at the two salt concentrations is Ϫ0.1, indicating that the two alignment tensors are effectively independent of one another. For the RDCs measured on free phospho-IIB Chb , the values are 21.2 Hz and 0.53, respectively in 100 mM NaCl and 10 mg/ml phage pf1, and Ϫ9.6 Hz and 0.64, respectively in 200 mM NaCl and 15 mg/ml phage pf1, with a normalized scalar product of 0.96 between the two alignment tensors. b The PDB accession codes are 1IIB and 1E2B for the x-ray (9) and NMR (10) structures of IIB Chb (C10S), respectively; and 1H9C for the NMR structure of phospho-IIB Chb (11).
The active site residue is located at position 10, and the active site loop comprises residues 9 -16. c The number of experimentally measured 1 D NH RDCs is shown in parentheses. d RDCs were measured for residues 9, 10, 11, 12, 13, and 16. The cross-peaks for the backbone amide groups of Met-14 and Ser-15 were too broad to permit measurement of the 1 J NH splitting in the alignment media. e RDCs were measured for residues 9, 10, 11, 12, 13, 14, and 16. The cross-peak for the backbone amide group of Ser-15 was too broad to permit measurement of the 1 J NH splitting in the alignment medium.  (52). Intermolecular NOEs were observed on the IIA Chb* (H89E)-IIB Chb (C10S) complex in D 2 O buffer using three-dimensional 12 C-filtered(F 1 )/ 13 C-separated(F 2 ) or 13 C-separated(F 2 )/ 12 Cfiltered(F 3 ) NOE experiments, and in H 2 O buffer using twodimensional 15 N-separated/ 13 C-edited and 13 C-separated/ 15 N-edited NOE experiments (53,54). Nine different combinations of isotope-labeled complexes were used for analysis of intermolecular NOEs (Table 1).
Structure Calculations-NOE-derived interproton distance restraints were classified into loose approximate distance ranges of 1.8 -2.7, 1.8 -3.5, 1.8 -5.0, and 1.8 -6.0 Å corresponding to strong, medium, weak, and very weak NOE cross-peak intensities, respectively (55); an empirical correction of 0.5 Å was added to the upper distance bounds of distance restraints involving methyl groups to account for the higher apparent intensity of methyl resonances (56). NOEs involving non-stereospecifically assigned methyl, methylene, and aromatic protons were represented by a (⌺r Ϫ6 ) Ϫ1/6 sum (57). Backbone torsion angle restraints for the active site loop of IIB Chb (C10S) and phospho-IIB Chb , and for the  interfacial side chains and to the active site loop (residues 9 -16) of IIB Chb (C10S). b The 1 D NH RDCs were measured for free IIB Chb (C10S) in two alignment media (10 mg/ml pf1 and 100 mM NaCl; and 17 mg/ml phage pf1 and 400 mM NaCl) (see Table  2). The active site loop (residues 9 -16) is given complete torsional degrees of freedom, while the remainder of the backbone is treated as a rigid body comprising the x-ray coordinates of IIB Chb (C10S) (PDB code 1IIB (9)); thus, the orientation of the alignment tensor is determined by the rigid portion of IIB Chb (C10S), while the conformation of the active site loop is determined by the 1 D NH RDCs within the loop. The RDC R-factors for the rigid portion of IIB Chb (C10S) are given in Table 2, whereas those for the active site loop of the restrained regularized mean structure of the complex are provided in Table 4. c The 13 C␣ and 13 C␤ chemical shift restraints involve only the active site loop (residues 9 -16) of IIB Chb (C10S). d None of the structures exhibit interproton distance violations Ͼ0.3 Å or torsion angle violations Ͼ5°. e The intermolecular repulsion energy is given by the value of the quartic van der Waals repulsion term calculated with a force constant of 4 kcal⅐mol Ϫ1 ⅐Å Ϫ4 and a van der Waals radius scale factor of 0.78. The intermolecular Lennard-Jones van der Waals interaction energy is calculated using the CHARMM19/20 parameters and is not included in the target function used to calculate the structures. f Defined as the average r.m.s. difference between the final 90 simulated annealing structures and the mean coordinates. The value quoted for the backbone (which excludes residues 75-84 of IIA Chb that are disordered and have a coordinate precision of 1.9 Å) provides a measure of the precision with which the orientation and translation of the two proteins in the complex have been determined, and does not take into account the backbone accuracy of the NMR coordinates of IIA Chb* (H89E) (26) and the x-ray coordinates of IIB Chb (C10S) (9) used for conjoined rigid body/torsion angle dynamics docking. The accuracy of the restrained regularized mean coordinates of IIA Chb* (H89E), excluding the loop residues 75-84 that were given backbone torsional degrees of freedom can be estimated from the coordinate precision (0.4 Å (26)) and the value of 8.7% for the crossvalidated RDC R-factor for the NH bond vectors (this work), which suggests a coordinate accuracy comparable to a 1.5-2 Å resolution crystal structure (73,74). The accuracy of the x-ray coordinates of IIB Chb (C10S) (excluding the active site loop that was given backbone torsional degrees of freedom) is likely to be around 0.3 Å judging from the crystallographic resolution (1.8 Å) and R-factor (18.7 and 24.1% for R free ) (9), as well as the RDC R-factors (this work,  Structures were calculated using conjoined rigid body/torsion angle-simulated annealing (59) with the program Xplor-NIH (60,61). The minimized target function comprises NOEderived interproton distance restraints, torsion angle restraints, RDC restraints (62), 13 C␣/ 13 C␤ chemical shift restraints (63), a quartic van der Waals repulsion term for the non-bonded con-tacts (64), a multidimensional torsion angle data base potential of mean force (65), and a gyration volume potential to ensure optimal packing (66). Structure figures were generated using the programs VMD-XPLOR (67) and GRASP (68). Reweighted atomic probability density maps were calculated as described previously (69)

RESULTS AND DISCUSSION
Equilibrium Binding of IIA Chb* -(H89E) and IIB Chb (C10S)-The binding of IIA Chb* and the active site IIA Chb* (H89E) phosphomimetic mutant (at natural isotopic abundance) to U-15 N-labeled IIB Chb (C10S) was monitored by 1 H-15 N correlation spectroscopy. Exchange between the complex and free proteins is fast on the chemical shift time scale. The pattern of 1 H N / 15 N chemical shift perturbations observed for the binding of IIA Chb* and IIA Chb* (H89E) to IIB Chb (C10S) is very similar but the magnitude of the perturbations is smaller for IIA Chb* than IIA Chb* (H89E). This is caused by the fact that the binding of IIA Chb* to IIB Chb (C10S) is weaker (K D ϳ2.1 Ϯ 0.5 mM) than that of IIA Chb* (H89E) (K D ϳ1.3 Ϯ 0.3 mM; see Fig. 1), and hence the fraction of complex formed under the same experimental conditions (protein concentrations Յ 1 mM) is considerably reduced. For this reason, all structural studies were conducted with the IIA Chb* (H89E) phosphomimetic mutant.
Structural studies were carried out on samples comprising 1 mM IIA Chb* (H89E) trimer and 1 mM IIB Chb( C10S). Under these conditions, 51.6% of IIA Chb* (H89E) and 64.4% of IIB Chb (C10S) are in the bound state. Because IIA Chb* is a symmetric trimer with three equivalent binding sites for IIB Chb , one can calculate that the percentage of IIA Chb* (H89E) with one, two, and three IIB Chb (C10S) molecules bound is 39.7, 10.9, and 1.0%, respec- , and C subunits of the IIA Chb* (H89E) symmetric trimer in blue, gold, and green, respectively, and IIB Chb (C10S) in red. The active site residues, H89E and C10S are shown in purple and cyan, respectively, and the pink meshes represent the reweighted atomic density probability map (69) for these two residues (drawn at a value of 20% maximum). Only a single IIB Chb molecule binding at the interface of the A and C subunits of IIA Chb* (H89E) is shown; because IIA Chb* (H89E) is a symmetric trimer there are three identical binding sites formed at the interfaces between the A and C subunit, the C and B subunits and the B and A subunits. B, ribbon diagram of the complex showing two IIB Chb (C10S) molecules bound to the IIA Chb* (H89E) trimer. C, ribbon diagram of the complex with a view orthogonal to that shown in B depicting three molecules of IIB Chb (C10S) bound to the IIA Chb* (H89E) trimer. The color coding in B and C is the same as that in A. D, stereoview showing a reweighted atomic probability density map (drawn at a value of 20% maximum and calculated from the final 90 simulated annealing structures) for some of the interfacial side chains displayed as purple and green meshes for IIB Chb (C10S) and the C chain of IIA Chb* (H89E), respectively. The backbones are shown as tubes color coded as in A; the side chains of the restrained regularized mean structure are color coded according to atom type (carbon, gray; oxygen, red; and nitrogen, blue). Residues of IIB Chb (C10S) are labeled in italics. FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 tively; the corresponding percentages, expressed in terms of IIB Chb (C10S) are 39.7, 21.7, and 3.0%, respectively. Given molecular masses of 33.6 kDa and 11.4 kDa for free IIA Chb* (H89E) and IIB Chb (C10S), respectively, and the fact that all species are in fast exchange with one another, the linewidths of IIA Chb* (H89E) and IIB Chb (C10S) in the NMR sample are determined by population weighted-average molecular masses of 41.0 and 36.2 kDa, respectively.

Solution Structure of the IIA Chb -IIB Chb Complex
Structure Determination-Because the chemical shift perturbations observed upon complex formation are small, one can conclude that there are no significant backbone structural changes (within the limits of the NMR method) induced within either IIA Chb* (H89E) or IIB Chb (C10S). We, therefore, proceeded to solve the structure of the complex using conjoined rigid body/torsion angle dynamics simulated annealing (59,70), largely on the basis of intermolecular NOE data. In this approach, the backbone (except for certain selected regions, see below) and non-interfacial side chain coordinates are treated as rigid bodies with rotational and translational degrees of freedom, whereas the interfacial sidechains are given full torsional degrees of freedom.
The chemical shift differences between IIA Chb* and IIA Chb* (H89E) are limited to the immediate vicinity of the mutation, and 1 D NH RDC measurements on free IIA Chb* (H89E) indicate excellent agreement between observed RDCs for helical residues and those calculated from the NMR structure (restrained regularized mean coordinates) of IIA Chb* (26). The RDC R-factor is 8.7% (defined as {Ͻ(D obs Ϫ D calc ) 2 Ͼ/ 2ϽD obs 2 Ͼ)} 1/2 (71), where D obs are the observed RDCs, and D calc are the calculated RDCs obtained by singular value decomposition against the coordinates of the protein (72)). The alignment tensor of IIA Chb* (H89E) is axially symmetric, as expected for a symmetric trimer, with a magnitude of Ϫ11.3 Hz for the axial component (D a NH ). We, therefore, used the NMR structure of IIA Chb* in the conjoined rigid body/torsion angle dynamics simulated annealing calculations. However, because the loop connecting helices 2 and 3 is partially disordered (i.e. highly mobile) in solution and contributes to the interface with IIB Chb , the backbone of residues 75-84 was also give torsional degrees of freedom.
IIB Chb (C10S) strongly interacts with the alignment medium (phage pf1) in the absence of salt; RDC measurements were therefore carried out in 100 and 400 mM NaCl using 10 and 17 mg/ml phage pf1, yielding values of 25.5 and Ϫ8.5 Hz for D a NH , respectively, and a rhombicity of ϳ0.4. The RDC R-factors are summarized in Table 2 and allow one to conclude the following: (a) both the 1.8-Å resolution x-ray structure of IIB Chb (C10S) (9) and the NMR structure of phospho-IIB Chb (11) provide a much better representation of the actual solution structure of IIB Chb (C10S) than does the NMR structure of IIB Chb (C10S) (10), reflecting the lower coordinate accuracy of the latter; (b) there are very large discrepancies between observed and calculated RDCs within the active site loop (residues 9 -16) for both the x-ray structure of IIB Chb (C10S) and the NMR structure of phospho-IIB Chb , reflected in very high RDC R-factors for this region; (c) removing the RDCs for the active site residues results in excellent agreement between observed and calculated RDCs with lower RDC R-factors for the x-ray structure of IIB Chb (C10S). Therefore, the coordinates of the x-ray structure of IIB Chb (C10S) (9) were used in the calculations, with the backbone of the active site loop (residues 9 -16), given torsion degrees of freedom. The observed RDC R-factors of 8 -9% observed for the x-ray structure of IIB Chb (C10S), excluding the active site loop, are as expected for a crystal structure solved at 1.5 to 2 Å resolution (73,74). It is worth noting that phosphorylation of IIB Chb is accompanied by a large conformational change within the active site loop as manifested both by RDCs (Table 2) and by significant differences in backbone (N, C␣, H␣, and C␤) chemical shifts (11). Excluding the active site loop, the crystal structure of IIB Chb (C10S) still displays lower RDC R-factors than the NMR structure of phospho-IIB Chb (Table 2). However, the RDCs within the active site loop are now in excellent agreement with the NMR structure of phospho-IIB Chb but exhibit very large discrepancies with respect to the x-ray structure of IIB Chb (C10S) ( Table 2).
The intermolecular NOE data were derived from a large series of isotope-filtered/isotope-separated intermolecular NOE experiments (53). Because of the relatively large size of the complex and extensive chemical shift overlap, nine different labeling combinations (Table 1), including amino acid specific labeling, were used to eliminate any ambiguities in intermolecular NOE assignments. Examples of the quality of the intermolecular NOE data are shown in Fig. 2.
In protein-protein complexes of the PTS that we have solved previously (30,31,(33)(34)(35)(36)(37), it is usually possible to derive sidechain torsion angle restraints for interfacial side chains based on heteronuclear 3 J scalar couplings and short mixing time NOE data (53). In this instance, the complex is not fully saturated because of weak binding (K D ϳ 1.3 mM), and there is a significant proportion of each component in the free state. We therefore employed a heuristic approach in which the interfacial side chains were given torsional degrees of freedom but restrained within the 1 and, where appropriate, 2 rotamers occupied in the free structures, unless these were inconsistent with the intermolecular NOE data.
Unfortunately we were not able to use RDC data to provide information on the relative orientation of the two components  Table 2, Footnote a for definition. b The number of experimentally measured 1 D NH RDCs is shown in parentheses. c RDCs were measured for residues 9, 10, 11, 12, 13, and 16 (see Table 2, Footnote d). d RDCs were measured for residues 9, 10, 11, 12, 13, 14, and 16 (see Table 2, Footnote e).

Solution Structure of the IIA Chb -IIB Chb Complex
within the complex. In a simple case of a weak binding binary complex in fast exchange on the chemical shift scale, the observed RDCs are weighted averages of the RDCs in the free and bound state, so that it is possible in principle to back-calculate the RDCs for the pure complex providing one knows exactly the fraction of the bound species (35,75). However, for the IIA Chb* (H89E)-IIB Chb (C10S) complex, the bound species comprises a mixture of three states with one, two and three IIB Chb (C10S) molecules bound to the IIA Chb* (H89E) trimer, each with its own alignment tensor. Deconvolution of the align-ment tensors for the individual bound states is not feasible, thereby precluding the use of RDCs in this system. Moreover, the alignment media that we explored (pf1, strained gels and PEG/hexanol (72)) all displayed differential interaction with one of the partners, making any extrapolation of average RDCs for the bound states unreliable. A summary of the structural statistics is given in Table 3, and a bestfit superposition of the final 90 simulated annealing structures is shown in Fig. 3A. The NOE-derived interproton distance restraints comprised 40 intermolecular NOEs (per bound IIB Chb molecule), as well as intramolecular NOEs related to those portions of the IIB Chb (C10S) backbone that were given torsional degrees of freedom. The agreement of the RDCs within the active site loop of IIB Chb (C10S) is comparable to that of the rest of the protein ( Table 4). The relative orientation of IIB Chb (C10S) to IIA Chb* (H89E) is well defined with a precision of 0.3 Ϯ 0.1 Å for the backbone of the complete complex (IIA Chb* and IIB Chb best-fitted overall), 0.9 Ϯ 0.3 Å for the backbone of IIB Chb with best-fitting to IIA Chb* , and 1.2 Ϯ 0.4 Å for IIA Chb* with best-fitting to IIB Chb . The coordinate precision for the interfacial side chains is 0.9 Ϯ 0.1 Å.
Overall Structure of the IIA Chb* (H89E)-IIB Chb (C10S) Complex-The overall structure of the complex is shown in Fig. 3 with one, two, and three molecules of IIB Chb (C10S) bound.
IIA Chb* is a symmetric trimer (26). Each subunit comprises a three-helix bundle (␣1, residues 14 -44; ␣2, residues 46 -74, ␣3, residues 84 -114) in an up-down topology with ␣2 antiparallel to ␣1 and ␣3. The trimer interface consists of a parallel coiled-coil formed by helix ␣3 of each subunit. The active site residue, His-89, is located in helix ␣3 and lies in a deep groove bounded by helix ␣1 of its own subunit and helices ␣2 and ␣3 of an adjacent subunit. In the representations shown in this report with subunits A, B, and C of IIA Chb* colored in blue, gold, and green, respectively, there are three identical binding sites for IIB Chb , located at the interface of subunits A and C, subunits C and B, and subunits B and A, with the first named subunit of each binding site bearing the active site His-89 (Fig. 3, B and C). Solution Structure of the IIA Chb -IIB Chb Complex FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 IIB Chb is a mixed ␣/␤ protein comprising 5 helices (residues 12-30, 43-49, 63-71, 80 -86, and 88 -105), and 4 ␤-strands (residues 4 -9, 34 -39, 53-56 and 76 -78) arranged in a -1x, -2x, -1x topology (9 -11). The active site residue, Cys-10, is located in an exposed active site loop (residues 9 -16) that forms a protrusion on the surface of the protein. The overall topology of IIB Chb is similar to that of IIA Mtl (7) although the percentage sequence identity is only 8%. In the NMR structure of phospho-IIB Chb (11) but not the x-ray structure of IIB Chb (C10S) (9), the active site loop has a conformation that is similar to that of IIB Mtl (both unphosphorylated (7) and phosphorylated (8)) and the low molecular weight protein-tyrosine phosphatases (42), except that the residues located at iϩ1 and iϩ2 of the active site cysteine are replaced by only a single residue in IIB Chb .
Because the three IIB Chb binding sites on IIA Chb* are identical, we will simply consider the interaction of IIB Chb (C10S) with the A and C subunits of IIA Chb* (H89E), where the contributing active site histidine (His-89) originates from the A subunit (Fig.  3A, 4 and 5). A total of 1836 Å 2 is buried upon complexation, 927 Å 2 originating from IIA Chb* (H89E) and 909 from IIB Chb (C10S). The interface accessible surface area contributed by the A and C subunits of IIA Chb* (H89E) is approximately equal. The interface comprises 15 residues each from the A and C subunits of IIA Chb* (H89E) and 29 residues from IIB Chb (C10S); 4 residues of the latter interact simultaneously with the A and C subunits of IIA Chb* (H89E) (Fig. 4C). The inter-face is ellipsoidal in shape with an eccentricity of ϳ0.3 (where values of 0 and 1 indicate a perfect circle and a straight line, respectively), the gap volume index (ratio of gap volume to interface accessible surface area) is 3.3 Å, and the r.m.s. deviation of the interface atoms from a least-squares plane through these atoms is 3.7 Å (concave; Fig. 5, left panel) for IIA Chb* (H89E) and 3.3 Å (convex; Fig. 5, right panel) for IIB Chb (C10S), These values are typical of transient complexes (76,77).
Stereoviews depicting details of the interface are shown in Fig. 4, A and B, and a summary of the intermolecular contacts is provided in Fig. 4C. The active site loop, a 3 10 helix (residues 58 -62) and helix ␣4 of IIB Chb (C10S) are in contact with helices ␣1 and ␣3 of the A subunit of IIA Chb* (H89E) (Fig.  4A). Helices ␣1 and ␣2, the C-terminal end of strand ␤2 and the subsequent 3 10 helix (residues 40 -42) of IIB Chb (C10S) contact helices ␣2 and ␣3 of the C subunit of IIA Chb* (H89E), as well as some residues in the disordered loop connecting these two helices in IIA Chb* (H89E) (Fig. 4B). The intermolecular interactions are largely hydrophobic with 50 -60% of the interfacial residues being nonpolar. Met-14 of the active site loop of IIB Chb (C10S) is involved in a very large number of intermolecular hydrophobic interactions with methyl group clusters of Val and Leu residues, specifically Val-21 A and Val-86 A of the A subunit and Val-83 C , Leu-87 C , and Val-88 C of the C subunit of IIA Chb* (H89E). Two other methionine residues, Met-22 A and Met-81 C of IIA Chb* (H89E) are also involved in an array of intermolecular hydrophobic interactions, with C10S, Pro-58, Gln-59, and Tyr-84, and with Leu-18, Lys-22, and Val-87, respectively, of IIB Chb (C10S). There are only two intermolecular salt bridge/ hydrogen bonding interactions: between Arg-24 and Glu-73 C , and between Ser-17 and the backbone carbonyl of Ile-72 C (Fig.  4B). These are supplemented by 7 longer range electrostatic interactions, 5 involving the A chain and 2 the B chain of IIA Chb* (H89E) (Fig. 4C). Of these, two involve interactions between charged side chains (Glu-15 A and Lys-86, and Lys-70 C and Glu-37). The remainder involve interactions either between polar groups or between polar and charged groups. An example of the former would include the interaction between the sulfur atom of Met-22 A and the hydroxyl group of Tyr-84 (Fig. 4A). An example of the latter is the interaction between the carboxylate of Glu-19 A and the hydroxyl group of Ser-81 and the amide group of Gln-59 (Fig. 4A).
As in previously solved complexes of the PTS (30, 31, 33-37), the center of each interface is largely hydrophobic, surrounded The left panel display the interaction surface (formed by the A and C subunits) on IIA Chb* (H89E) for IIB Chb (C10S); the right panel shows the interaction surface on IIB Chb (C10S) for IIA Chb* (H89E). The surfaces are color coded as follows: hydrophobic residues, green; uncharged residues bearing a polar functional group, cyan; negatively charged residues, red; positively charged residues, blue. Relevant portions of the backbone and active site residue of the interacting partner are displayed as tubes and bonds, respectively. Residues of IIB Chb (C10S) are labeled in italics. Residues from the C subunit of IIB Chb* (H89E) are indicated by an apostrophe after the residue number; in addition, the surfaces of the A and C subunits of IIA Chb* that do not constitute the interaction surface are colored in dark gray and light gray, respectively.
His-89 A in the transition state, as described for the phosphoryl transition state of the IIA Mtl -IIB Mtl complex (36). The results are shown in Fig. 6, A and B. The active site loop of IIB Chb in the transition state adopts a very similar conformation to that seen in the previously determined NMR structure of phospho-IIB Chb (11) with a backbone r.m.s. difference of 0.8 Å, and the RDCs within the active site loop are well satisfied ( Table 4). The key conformational changes within the active site loop of IIB Chb are the transition of the / angles of Gly-13 from the right (ϳ Ϫ60°/Ϫ40°) to left ␣-helical region (ϳ100°/30°) of the Ramachandran plot, a shift in the / angles of Met-14 away from the ␣-helical region (from ϳ Ϫ50°/Ϫ50°to ϳ Ϫ110°/Ϫ70°), and a shift of the / angles of Ala-12 from the ␤ region (ϳ Ϫ150°/70°) to the helical region (ϳ Ϫ110°/Ϫ25°). The changes in the backbone of residues 87 A -91 A of IIA Chb* are minimal with atomic r.m.s. displacements of less than 0.6 Å (Fig. 6A). It can be readily seen that as a consequence of the conformational change in the active site loop of IIB Chb , Ala-12 no longer occludes the phosphoryl group and the path followed by the backbone readily permits the formation of a phosphoryl transition state His-89-P-Cys-10 intermediate.
Although the structure of phospho-IIA Chb* has not been solved, the phosphorylated state is easily modeled from the structure of unphosphorylated IIA Chb* (26): minimal changes in the 1 and 2 angles of His-89 A are all that are required to permit the formation of hydrogen bonds from the side chain amide of Gln-91 C and the N⑀2-H proton of His-93 A to the phosphoryl group. In the transition state, these distances are lengthened but bridging hydrogen bonds involving water molecules could clearly be formed (Fig.  6C). The phosphoryl group in the transition state is stabilized by an array of hydrogen bonds from the IIB Chb active site loop, including the backbone amides of Ala-12, Met-14, and Ser-15, and the hydroxyl groups of Ser-11 and Ser-15; in addition, hydrogen bonds from the backbone amides of Ser-11 and Gly-13 to the sulfur atom to Cys-10 further stabilize the conformation (Fig. 6, B and C). Many of these interactions are maintained upon shortening of the S-P bond to form phospho-IIB Chb (11). The interactions stabilizing the phosphoryl group in both the IIA Chb* -P-IIB Chb transition state and phospho-IIB Chb (11) are very similar to those observed in phospho-IIB Mtl (8) and the phosphorylated state of the low molecular weight protein-tyrosine phosphatases (42). Presumably, the reason that the active site loop of IIB Chb undergoes a conformational change upon phosphorylation, whereas that of IIB Mtl does not, resides in the fact that the single residue deletion within the active site loop of IIB Chb , relative to IIB Mtl , results in stereochemical strain that can only be overcome by numerous interactions between the protein backbone and the phosphoryl group.
Further examination of the IIA Chb* -P-IIB Chb transition state reveals that the phosphoryl group is buried within a largely hydrophobic environment provided by Ala-12 and Met-14 of IIB Chb and Val-21 A , Met-22 A , Ile-25 A , Ile-72 C , and Leu-87 C of IIA Chb* . This configuration is common to all protein-protein complexes of the PTS solved to date (30,31,(33)(34)(35)(36)(37), including the EIN-HPr complex that is common to all branches of the pathway, as well as the complexes involving sugar-specific components.
Concluding Remarks-We have determined the solution structure of the IIA Chb* (H89E)-IIB Chb (C10S) complex, and shown that this structure is fully compatible with the formation of a phosphoryl transition state when the active site loop of IIB Chb adopts the conformation found in phosphorylated IIB Chb . As previously noted, the overall topology of IIB Chb and IIB Mtl are remarkably similar, despite 8% sequence identity, and the C␣ atoms of 71 out of 106 residues can be superimposed with an atomic r.m.s. difference of 2 Å. Fig. 7 shows a comparison of the phosphoryl transition states of the two complexes with the coordinates of IIB Chb and IIB Mtl superimposed. The structures of the IIA Chb* trimer and IIA Mtl monomer bear no resemblance to one another, and with the exception of a single turn of helix that fortuitously overlaps, the structural elements making up the binding site are entirely different. Nevertheless, the position of the His-P-Cys phosphoryl transition state is remarkably similar. In addition, the distribution of hydrophobic, polar, and charged residues within the binding sites of IIA Chb* and IIA Mtl , and likewise IIB Chb and IIB Mtl , is broadly similar (compare Fig. 5 of this report with Fig. 6 of Ref. 36), although the interaction of IIA Chb* with IIB Chb involves a somewhat larger preponderance of hydrophobic residues than that between IIA Mtl and IIB Mtl , and the cleft in which the active site histidine of IIA Chb* is located is both deeper and narrower than that for IIA Mtl . Thus, one might argue that these two PTS complexes from distinct sugar branches of the pathway illustrate an example of convergent evolution of the surfaces of active sites (in terms of shape and distribution of residue type) generated by completely different underlying backbone structural elements.