PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Biochemistry. Author manuscript; available in PMC 2010 August 18.
Published in final edited form as:
PMCID: PMC2881472
NIHMSID: NIHMS199698

Ligand-induced changes in the structure and dynamics of E. coli peptide deformylase

Abstract

Peptide deformylase (PDF) is an enzyme that is responsible for removing the formyl group from nascently synthesized polypeptides in bacteria, attracting much attention as a potential target for novel antibacterial agents. Efforts to develop potent inhibitors of the enzyme have progressed based on classical medicinal chemistry, combinatorial chemistry, and structural approaches. Yet, the validity of PDF as an antibacterial target hangs, in part, on the ability of inhibitors to selectively target this enzyme in favor of structurally related metallohydrolases. We have used 15N NMR spectroscopy and isothermal titration calorimetry to investigate the high-affinity interaction of EcPDF with actinonin, a naturally occurring potent EcPDF inhibitor. Backbone amide chemical shifts, residual dipolar couplings, hydrogen-deuterium exchange, and 15N relaxation reveal structural and dynamic effects of ligand binding in the immediate vicinity of the ligand binding site as well as at remote sites. A comparison of the crystal structures of free and actinonin-bound EcPDF with the solution data suggests that most of the consequences of ligand binding on the protein are lost or obscured during crystallization. The results of these studies improve our understanding of the thermodynamic global minimum and have important implications for structure-based drug design.

Peptide deformylase (PDF) is an essential and highly conserved enzyme that functions in protein maturation by removing the N-formyl group from the methionine of nascently synthesized polypeptides in bacteria, protists and eukaryotic organelles (19). Because protein translation in bacteria is initiated with N-formyl-methionine, PDF has emerged as a target of efforts to develop novel antibacterial agents (1016). The extensively characterized enzyme from E. coli (EcPDF) exhibits some substrate selectivity in vitro (4, 6, 1721); this property has guided the design of several effective substrate analog inhibitors (1823). Current work seeks to identify compounds with broad-spectrum activity against bacterial PDF while avoiding inhibition of other cellular targets, including the recently identified human mitochondrial protein (7, 9, 24).

As a member of the metallohydrolase superfamily that includes the matrix metalloproteins (MMPs) thermolysin and stromelysin (24), PDF binds divalent metals in a tetrahedral arrangement involving coordination with two histidines in a conserved QHExDH motif and a cysteine from a conserved EGCLS motif; a water molecule serves as the fourth ligand (Figure 1). Although the active site metal in most members of the superfamily is zinc, evidence suggests the native metal in EcPDF is iron (5). Crystal structures of EcPDF free (2527) and in complex with several competitive inhibitors capable of metal coordination have been reported (22, 23), as well as an NMR structure of the conserved catalytic core (residues 1–147) lacking the C-terminal helix, α3 (2, 28, 29). In addition, crystal structures are now available of PDF from several bacteria, including Streptococcus pneumoniae, Staphylococcus aureus, Haemophilus influenzae, Pseudomonas aeruginosa, Bacillus stearothermophilus (13, 15, 30, 31), as well as that of the protist Plasmodium falciparum (32).

Figure 1
Peptide deformylase (a) Ribbon diagram of the EcPDF crystal structure, 1BS5, chain A. The bound metal is blue, metal ligands are Cys90, His132 and His136; the oxygen atom from the water molecule that comprises the fourth metal ligand is shown as a red ...

In search of inhibitors of PDF, much work has focused on complexes of EcPDF with derivatives of actinonin (Figure 1b), a hydroxamic acid-containing actinomycete-derived natural product that serves as a competitive inhibitor (13, 23). The available crystal structures of free and actinonin-bound EcPDF revealed measurable, if subtle, structural differences between the free and bound state, suggesting that induced-fit recognition may play a role in inhibitor binding. Such protein plasticity is increasingly being recognized as obfuscatory to structure-based drug design efforts (33) since it obscures the structure of the free energy minimum and makes it difficult to meaningfully predict thermodynamic relationships.

To better understand binding determinants for PDF it is necessary to complement structural insight with dynamic and thermodynamic knowledge. In particular, isothermal titration calorimetry (ITC) can be used to directly measure the enthalpy of binding (ΔHbind) and obtain the association equilibrium constant (KA), from which the binding free energy (ΔGbind) can be derived. The overall thermodynamics of ligand binding can be described by partitioning enthalpic and entropic contributions to the binding free energy:

ΔGbind=ΔHbindTΔSbind=ΔHbindT(ΔSsolv+ΔSconf+ΔSt/r)

where ΔSbind is the entropy change upon binding. The entropy term can be further partitioned into intermolecular and intramolecular components: ΔSsolv is the entropy change due to the changes in solvation of both binding partners, ΔSconf is the conformational entropy change due to changes in structure and dynamics, while ΔSt/r refers to changes in translational and rotational degrees of freedom (34). Formation of an enzyme-ligand complex is usually accompanied by an enthalpy gain through the formation of favorable intermolecular interactions (hydrogen bonds, van der Waals forces), but overall affinity is also affected by changes in conformational entropy at the binding interface, exclusion of ordered water molecules (hydrophobic effect) and gains in entropy at remote sites throughout the protein.

High-resolution NMR spectroscopy provides a means of observing and quantifying protein flexibility and the effect on ligand binding on protein structure in solution. For proteins weakly aligned in a magnetic field, residual dipolar couplings (rDCs)(35) provide a direct measurement of the orientation of individual bond vectors relative to the molecular frame, allowing ligand-induced structural changes to be precisely measured. Native state hydrogen-deuterium exchange measurements enable quantification of local protein flexibility and stability, yielding thermodynamic insights that could be used to dissect individual contributions to binding free energies (3644). Moreover, heteronuclear relaxation rates (45) provide a measure of the amplitudes of bond-vector motions occurring on the ns-ps timescales (the Lipari-Szabo order parameter, S2 (46)) that can be used to estimate the conformational entropy of individual residues in the protein (36, 37), thus providing a means to better characterize binding sites and to quantify the thermodynamic effect of ligand binding on the conformational entropy change, ΔSconf, of the affected sites.

In order to gain insights into the role of protein flexibility in molecular recognition, we have characterized the effect of actinonin binding on the solution structure and dynamics of EcPDF using NMR measurements of backbone rDCs, hydrogen-deuterium exchange rates, and 15N relaxation rates. Although we found a correlation between proximity to the binding site and induced structural and dynamic changes, rDC measurements highlighted differences between the crystal and solution structures, while relaxation and H/D exchange rate data suggest that localized changes in structure, dynamics and thermodynamics are transmitted to other parts of the protein. These data underscore the importance of understanding protein structure and dynamics in solution and demonstrate that local effects of ligand binding have consequences at remote sites in ways that are not entirely understood and are difficult to predict.

MATERIALS AND METHODS

Protein and ligand samples

The catalytic core of E. coli PDF (residues 1–147) (28) was expressed and purified as described (47). All experiments were performed with the conserved catalytic core of EcPDF (residues 1–147); for convenience, unless explicitly stated, EcPDF is used throughout the text to indicate this domain. Actinonin (3-[[1-[(2-(hydroxymethyl)-1-pyrrolidinyl)carbonyl]-2-methylpropyl]carbamoyl]octanohydroxamic acid; CAS No. 13434-13-4) (48) was generously provided by Dr. Z. Yuan (Versicor, Inc, San Diego, CA) or purchased from Sigma-Aldrich (St. Louis; Cat. # A6671), and used without further purification; purity and integrity were checked by electrospray mass spectrometry and one-dimensional WATERGATE (49, 50) NMR spectroscopy. NMR samples were 0.6–1.0 mM protein in 20 mM Tris, pH 7.2, 10% D2O, 90% H2O, and 0.02% NaN3; EcPDF/actinonin complex samples contained a slight excess of actinonin (~1.1 equivalents).

Isothermal Titration Calorimetry

ITC experiments were performed at 37°C on a MicroCal VP-ITC instrument, with the cell containing EcPDF at a concentration of 32–40 µM, and the syringe containing 320–400 µM of actinonin. Each experiment consisted of 59 injections of 5 µL actinonin with a 240 sec interval between each injection. The protein samples were dialyzed overnight and the dialysate was used to dissolve the actinonin sample. Both experimental solutions were thoroughly degassed before each titration.

The data were analyzed using a one-site binding model with the Origin (V.7 SR4) software package (MicroCal, Inc.). The baseline at the beginning and end of each injection peak were adjusted manually. The last few points of the titration, after the solution was saturated, were averaged and subtracted from each data point to correct for heat of dilution. Origin uses non-linear least squares minimization and the concentrations of the titrant, LT, and the analyte, MT, to fit the energy flow per injection, Qi, to the equilibrium binding equation KA=Θ(1Θ)(LTnΘMt), where Θi=QiMTV0nΔH is the fraction bound after the ith injection and V0 is the cell volume, generating best fit values of the stoichiometry (n), binding enthalpy (ΔH), and the binding constant (KA). The binding free energy is obtained from ΔGbind = −RT ln KA = RT lnKD, while the entropy of binding results from the Gibbs relation ΔGbind = ΔHbindTΔSbind.

The heat capacity change (ΔCp) of actinonin binding to EcPDF was obtained from ITC experiments performed at 13°C, 19°C, 25°C, 31°C, 37°C, and 43°C, and a linear fit of the enthalpy change at each temperature to ΔH = ΔCp (T –ΔTh) where Th is the temperature at which the binding enthalpy is zero. Titrations were repeated at varying salt concentrations (10, 150, 300 mM NaCl) to test for ion linkage, and in buffers with different ionization enthalpies (tris(hydroxymethyl)aminomethane (TRIS; hion = −11.3 kcal mol−1), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; hion = −4.8 kcal mol−1), and cacodylic acid (hion = −0.7 kcal mol−1)) to test for proton linkage. In all calculations, the differences between the standard state (1 M) and those of the NMR and ITC experiments were neglected.

NMR Spectroscopy

Backbone resonance assignments of free EcPDF were obtained from the BioMagRes bank (accession # 4089) (2, 29) and verified by 15N-separated TOCSY and NOESY spectra (51). 15N chemical shifts were adjusted by +3 ppm to correct for a systematic offset (47) resulting from differences in referencing between the published work and that reported here. NMR data for resonance assignments were recorded at 310 K on Bruker DMX-600 or DRX-600 spectrometers equipped with triple-resonance three-axis shielded gradient probes. Proton chemical shifts were referenced externally to DSS (2,2-dimethylsilapentane-5-sulfonic acid) and 15N/13C chemical shifts were referenced indirectly based on their nuclear gyromagnetic ratios 52. Sequence-specific backbone assignments for HN, N, Cα, Cβ and C' were made from HNCACB, and CBCA(CO)NH, and HNCO spectra (52). Backbone assignments were further confirmed from the aliphatic side chain proton and carbon resonances, which were assigned by incorporating data from HBHA(CO)NH, HCCH-TOCSY, HCCH-COSY, 15N-separated NOESY, 15N-separated TOCSY, HNHA, 13C-separated NOESY, and C(CO)NH-TOCSY spectra (51, 52). Time domain data were zero filled to twice the number of acquired points in all dimensions to improve spectral resolution. NMR data were processed with NMRPipe (53) and analyzed using NMRVIEW (54).

Weighted-average 1H/15N shift perturbations induced by actinonin were examined by comparing peak positions in two dimensional NMR spectra of free and actinonin bound EcPDF and analyzed according to (55):

ΔδHN=(ΔδH2+(ΔδN/5)2)2

were ΔδH and ΔδN is the change in chemical shift for 1H and 15N respectively.

NMR Relaxation Measurements

Relaxation spectra were recorded at a 1H frequency of 600 MHz and 800 MHz on spectrometers equipped with triple resonance single-axis gradient cryoprobes. 15N R1 (longitudinal) and R2 (transverse) relaxation rate constants were obtained from standard inversion recovery and CPMG-based experiments (45, 56) acquired with 16 scans per time point, 2048 × 256 complex data points. The recovery delay for both experiments was 1.5 s. Nine relaxation time points were sampled in random order for each experiment (10, 50, 150, 300, 500, 750, 1000, 1500 and 2000 ms) and (0, 17, 34, 51, 68, 85, 102, 136 and 187 ms) for R1 and R2, respectively. Data were zero-filled to 4096 × 512 in the direct and indirect dimensions, respectively, and apodized with a square cosine window function in the direct dimension and in the indirect dimension. Relaxation rates were obtained from one-parameter exponential fits to the peak intensities at each time point. The standard deviation of the baseline spectral noise was taken to be the uncertainty in peak heights and uncertainties in the fit parameters were obtained from Monte-Carlo simulations with a confidence interval of 0.68. Heteronuclear NOE spectra (57) were acquired (at 600 MHz) in an interleaved fashion with 32 scans per increment and a 5 s recycle delay. The NOE values were determined from the ratios of the peak intensities in spectra acquired with and without proton saturation. Proton saturation was achieved via a train of 120° proton pulses applied every 18 msec for the last 3 sec of the recycle delay.

The relaxation data were analyzed in terms of the extended “model-free” formalism (58) using the relax software with a modified model-free optimization protocol (59), in which, prior to optimization of the diffusion tensor, model-free modes are determined. Residues 1–147 of the crystal structures of free EcPDF (1BS5, chain A) or the EcPDF-actinonin complex (residues 1–147 of 1G2A, chain A), protonated with MOLMOL (60), were used to test fitness of the data to various diffusion models. Amides were fit to models in which diffusion was assumed to be isotropic (spherical), axially symmetric (oblate and/or prolate) or fully anisotropic (ellipsoid), and the pairwise and overall fit of the data to each model was evaluated. The appropriateness of each model was evaluated by use of the Bayesian information criteria (BIC) (61) analysis and 500 Monte-Carlo simulations, as implemented in the program relax (62). A value of −172 ppm was used for the amide 15N chemical shift anisotropy (CSA) and 1.02 Å for the N–H bond distance (63). For both free and actinonin-bound EcPDF, there was no significant improvement in fit to anisotropic models over isotropic treatment.

The internal dynamics parameters for each amide resonance were obtained by fitting the experimental relaxation data to five different motional models. In accord with the extended model free formalism (58), in addition to the common overall correlation time, τm, the parameters optimized for each model are: (1) S2, (2) S2 and τe,(3) S2 and Rex, (4) S2, τe and Rex, and (5) S2f, S2s and τe, where Rex is a phenomenological exchange term introduced to account for chemical exchange line-broadening (affecting R2) and τe is the effective correlation time for internal motions. The five models have the following implicit assumptions: (1) S2s =1 and τf ~ 0, (2) S2s=1 and slower motions are negligible (i.e., τe = τf), (3 and 4) Rex > 0, and (5) only τf is ~ 0. Model selection was performed based on BIC scores (61). After excluding residues with significant overlap for the free and actinonin complex, motional parameters were obtained for 109 of the free and 104 of the bound form of 141 non-proline residues.

Ligand-induced conformational entropy changes, ΔSconf, were estimated for each amide by assuming a “diffusion in a cone” motional model (37, 64) as described by

ΔSconf=kbNATln{3(1+8Sb)1/23(1+8Sf)1/2}

where Sf and Sb are the order parameters for the free and bound forms (i.e. (S2)1/2), kb is Boltzman’s constant, NA is Avogadro’s number, and T the absolute temperature, 310 K for these measurements (for residues with S2 > 0.95, S was set to 0.98). After considering residues whose amides could be measured for the free and the bound form, the thermodynamics parameters were obtained for 97 residues.

Residual dipolar coupling measurements

Residual dipolar couplings (rDCs) (35) of EcPDF and its complex with actinonin were measured from the difference in multiplet splittings of signals in 2D IPAP spectra (65) recorded in isotropic solution and in 20 mg/ml of Pf1 phage (ASLA BIOTECH Ltd.). Spectra were recorded at 800 MHz with 2048 × 256 complex points and spectral widths of 12,019 × 4865 Hz in the direct (1H)and indirect (15N) dimensions. The data were apodized with a cosine bell and zero filled to 4096 points in each dimension prior to Fourier transformation. Peak positions were obtained by least-squares fitting as implemented in NMRPipe (53); amides with significant multiplet overlap were excluded from the analysis. The magnitudes of the alignments tensors were estimated from a histogram of the rDC data (66), using the relation

rDC=S2DaNH[(3cos2θ1)+32Rsin2θcos2φ]

where DaNH is the axial component of the residual dipolar coupling tensor, R is the rhombicity and S2 is the square of the experimentally determined order parameter; the values were confirmed with the program PALES (67), using as input the first chain of structures 1BS5 (residues 1–147) and 1G2A (residues 1–147) for free and actinonin-bound EcPDF. Only amides for which both upfield and downfield multiplet components could be resolved were considered for analysis. Error estimates were obtained by propagating the uncertainties in the peak positions of each resonance obtained from the ratio of the linewidth at half-height to signal-to-noise (35). In order to compare the rDC values of free and actinonin-bound EcPDF, rDC values were normalized by the axial component of the corresponding tensor.

Hydrogen-Deuterium Exchange

Protein samples for measurement of amide exchange rates were prepared by dilution of highly concentrated samples into a buffer containing 90% D2O to final concentration of 0.8 mM. 15N-edited HSQC spectra were acquired every hour for a 10-hour exchange time course; each acquisition was 11 minutes. The first time point was approximately 15 minutes after exchanging into D2O. Data were recorded on a Bruker DRX 800 MHz spectrometer at 37°C. Exchange rates were obtained by fitting peak intensities to single exponential function. Residues that exchanged within the first time point were assigned as on the order of the fastest rate measured (~0.1 s−1), and residues for which the intensity did not change more that 10% after 10 hrs hours were assigned the magnitude of the slowest rate measured (0.0001 s−1).

Site-specific changes in thermodynamic stability due to actinonin binding were obtained from changes in hydrogen/deuterium exchange rates based on

ΔΔG=RTlnkckf=RTlnPcPf

where kf and kc are the measured exchange rates and Pf and Pc are the derived protection factors for free and actinonin-bound EcPDF (68).

Structure thermodynamic calculations

Solvent-accessible surface areas were determined from the structures 1BS5 and 1G2A for free and actinonin-bound EcPDF using the program STC (v. 5.3) (69). The solvation entropy predicted from the structural data was calculated from the change in ASA using

ΔSsol(T)=0.45ΔAnpln(T/384.15)+0.26ΔApolln(T/335.15)

where ΔAnp is the change in non-polar accessible surface area upon binding and ΔApol is the change in polar accessible surface area (69).

Per-residue differences in the average crystal structures of the free (1BS5) and actinonin-bound PDF (1G2A) were computed with MOLMOL (60). First, average coordinates were computed for each structure with three molecules in the asymmetric unit. Then, the RMSDs between these averaged coordinates were computed. Averaged coordinates were used without further regularization.

UV-visible spectroscopy of Co2+ EcPDF

Absorption spectra were acquired on a Hewlett Packard 8452A UV-Visible diode array spectrophotometer at room temperature. Protein samples were 50 µM (in 20 mM Tris, pH 7.2 at 25°C, 10 mM NaCl). The EcPDFactinonin complex was formed by the addition of actinonin (45 µl of 17 mM stock solution water) to 1 mL of protein sample resulting in a slight excess of inhibitor.

RESULTS

Isothermal Titration Calorimetry

Isothermal titration calorimetry measurements of the binding of actinonin to zinc-loaded EcPDF in NMR buffer (20 mM Tris, pH 7.2, 10 mM NaCl) revealed a single, enthalpically favorable (exothermic) binding event (Figure 2). This, and unpublished findings (70), is consistent with expectations for the mode of actinonin binding; no evidence was found for thermodynamically distinct species of the enzyme (71). The enthalpy of actinonin binding (ΔHbind) was measured to be −1.644 ± 0.012 kcal mol−1 at 37°C, and the fitted dissociation constant (Kd) is 110 ± 15 nM; these results indicate a c value (KA·MTot) of ca. 290, slightly above the ideal range for accurate fitting of KA (72, 73). From the fitted parameters we obtained ΔG of −9.8 ± 1.3 kcal mol−1 and ΔS of 27 ± 4 cal mol−1 K−1. The heat capacity change ΔCp associated with actinonin binding to EcPDF was obtained from the temperature dependence of the enthalpy, yielding a value of −180 ± 10 cal mol−1 K−1.

Figure 2
Isothermal titration calorimetry of Zn-PDF with actinonin at 37°C. Heats of reaction from each injection were fit to a single binding site model (Origin V.7) to obtain the best-fit values of binding stoichiometry (n = 0.954 ± 0.004), affinity ...

To account for contributions by salt bridge formation in the measured thermodynamics parameters (7476) the salt dependence of binding was examined by performing ITC experiments with a range of different salt concentrations; if the rearrangement of atoms in EcPDF involved the formation or disruption of a salt bridge, the binding affinity (KA) of the reaction would change as the salt concentration increases due to ion linkage. From experiments at 10 mM NaCl, 150 mM NaCl, and 300 mM NaCl (Supporting Information), no significant correlation between salt concentration and binding affinity was apparent, implying that ion linkage does not contribute to the binding of actinonin to EcPDF.

Proton linkage (77) was examined by performing calorimetry experiments in buffers with different heats of ionization (78). In the event of proton linkage, the number of protons taken (or released) from the buffer can be calculated from (77):

ΔHobs=ΔHind+nHhion

where ΔHobs is the measured enthalpy, ΔHind is the ionization independent contribution, hion is the heat of ionization from the buffer, and nH is the number of protons linked to the binding event. The change in enthalpy (Supporting Information) yielded an nH value of 0.05 ± 0.02, indicating that, at this pH, proton linkage does not significantly contribute to the thermodynamics of EcPDF binding to actinonin. These controls indicated the measured thermodynamic parameters (i.e., ΔHbind, ΔCp) do not need to be corrected for ion or proton linkage.

Effect of actinonin on the structure of EcPDF

To examine the effects of actinonin binding to EcPDF, we performed a series of analyses of the atomic-level difference between structural and dynamic properties of free and actinonin-bound EcPDF. The key findings are summarized in Figure 3 and Figure 4 (and Supporting Information), and elaborated on in detail below.

Figure 3
Effect of actinonin on the structure and dynamics of EcPDF. (a) Pairwise backbone atom RMSDs between the mean structures from the three molecules in each of the asymmetric units of the free (1BS5) and actinonin-bound (1G2A) EcPDF crystal structures. (b) ...
Figure 4
Effect of actinonin binding on EcPDF structure and dynamics. (a) Actinonin-induced amide shift perturbations. Residues are colored by a linear ramp from grey for unperturbed amides (Δδ = 0) to red for amides perturbed greater than twice ...

The crystal structures of EcPDF in the absence (1BS5, 2.5 Å resolution) (27) and presence of actinonin (1G2A, 1.75 Å) (23) were obtained from the same crystallographic space group, with three molecules per asymmetric unit. Comparison of the free and actinonin bound structures was complicated by variability between molecules in the asymmetric unit. These differences mask the actinonin-induced changes in the structure of EcPDF. In order to distinguish structural variations due to crystal packing from those due to actinonin binding, the average structures for the free and bound proteins were separately calculated and compared. This analysis revealed a mean deviation of 0.23 ± 0.11 Å for residues 1–147, 0.35 ± 0.10 Å for the β4-β5 loop, and 0.27 ± 0.12 for residues within 5 Å of the binding site (Glu42 - Ala47, Gln50, Ile86 - Ser92, Arg97, Leu125, Ile128 -Cys129, His132 - Glu133 and His136)(Figure 3a). Although not examined in detail here, it is worth noting that the ligand-induced structural perturbations are larger at the sidechain level, with a mean displacement of 0.48 ± 0.31 for all heavy atoms of residues in the proximity of the binding site (Supporting Information). However, since the NMR data reported below directly probe the backbone and not sidechains, the analysis in this paper is limited to backbone atoms.

Resonance assignments

Resonance assignments of the EcPDF-actinonin complex were determined from standard double- and triple-resonance NMR spectra, particularly three-dimensional HNCACB, CBCA(CO)NH, CC(CO)NH-TOCSY, and HBHA(CO)NH and HNCO spectra (51, 52). With the exception of the first residue, all amide proton and nitrogen chemical shifts of non-proline residues could be assigned. The high quality of the data allowed for the assignment of 99% of HN, Cα, Cβ, Hα, Hβ and C' positions, including 140 backbone amide resonances (of 147 residues, six are prolines).

Spectral perturbation induced by actinonin binding to EcPDF

Actinonin binding changes the coordination state of the bound metal. The substitution of Co2+ into the active site of metalloenzymes is a common method for characterizing binding events due to the metal’s unique spectral properties (79). The electronic absorption spectrum in the visible region (d-d transitions) is sensitive to the coordination state of Co2+ and therefore a useful probe of the environment of the bound metal in EcPDF (80). Binding of actinonin to EcPDF results in a change in coordination state for the metal from tetrahedral to penta-coordinate, as the hydroxamate moiety of the ligand serves as a bidentate chelator of the bound metal, displacing the water molecule that serves as the found ligand in the free protein (13, 23). Consistent with this expectation, upon actinonin binding we observe a decrease in absorption at 565 nm (Figure 5a).

Figure 5
Spectral perturbations induced by actinonin binding. (a) UV-visible spectra of Co2+-EcPDF recorded before (solid line) and after addition of actinonin (dashed line). The change in coordination geometry is evident from the decreased absorbance at 560 nm ...

Actinonin induces profound changes in the 1H-15N spectrum of the protein (Figure 5b). This in spite of the small surface area buried by the ligand upon binding to EcPDF (~775.76 Å2), and the apparently subtle structural deformations. The weighted average of the actinonin-induced shift perturbations are shown in Figure 3b and are mapped onto residues 1–147 of the EcPDF-actinonin complex crystal structure in Figure 4a. The correspondence between proximity to the ligand and shift perturbation is quite evident in the data. Clearly, the largest shift perturbations are concentrated in the immediate vicinity of the ligand binding site, particularly, Glu41 to Gly47, Cys90 to Ser92, and His132 to Glu133. However, significant shift perturbations are also evident at sites substantially removed (> 8 Å) from the site of actinonin binding (e.g., His7, Gln55, Val100, Phe142 and Met143; Figure 3b, Figure 4). This finding was unexpected because of the subtle structural differences between the crystal structures of free and actinonin-bound EcPDF, and absence of highly shielding or deshielding functionalities on actinonin.

The effect of actinonin binding on the residual dipolar couplings (rDCs) obtained under partial alignment obtained by addition of Pf1 filamentous phage was also assessed (Figure 5c), as was the degree of agreement between the observed rDC and those predicted based on the crystal structures (1BS5, 1G2A; Figure 5d); Q-factors (67) were 0.257 and 0.281 for chain A of free and actinonin-bound PDF, respectively. Due to overlap of the peak multiplets, ligand-dependent changes in rDC values could be obtained for 112 of 140 backbone amide resonances. Although care was taken to use identical conditions for sample preparation of the free and actinonin-bound EcPDF, the resulting alignment tensors differed slightly; the magnitudes of the alignment tensors (DaNH) were 2.33 and 2.01 for free and actinonin-bound EcPDF, with rhombicities (R) of 0.63 and 0.51, respectively.

After normalizing the measured rDC values by DaNH, the residual dipolar couplings of the free and actinonin-bound samples were compared (Figure 3c). Although the values were similar for the protein, significant differences (larger than twice the error) were noted for 35 residues, while significant changes were observed in residues within 5 Å of the ligand-binding site, especially near residues 43–50 and 126–136, but also at sites more remote from the binding site, such as 69–71 and 105–110. The actinonin-induced absolute changes in rDC values are mapped onto the structure of the EcPDF-actinonin complex in Figure 4b. Experimentally measured rDC values did not agree well with those calculated from the crystal structures (Figure 5d), as evident from the least-squares minimization of the target function,

c2=ij((DijmeasDijcalc)2σij2)

of 7,543 for free and 3,146 for the complex, where Dij is the rDC between pairs of spins and σij is the uncertainty in the values (81).

Backbone amide 15N relaxation

Backbone amide 15N R1, R2 and NOE relaxation rate constants for free and actinonin-bound EcPDF are shown in Supporting Information. High NOE values (η > 0.8) and relatively uniform R2/R1 ratios for both the free and actinonin-bound protein are indicative of a highly ordered structure throughout the core of the protein. Measurable actinonin-induced changes in the R2/R1 ratios are seen in the EGCLS motif (residues 88–92) and the QHExDH motif (residues 131–136), both of which are involved in metal binding. However, ligand-induced changes in relaxation rates were not limited to residues in or near the ligand-binding site.

Analysis of the 15N relaxation for free EcPDF data resulted in a global rotational correlation time of 6.02 ns, with no significant improvement in the fit for anisotropic over isotropic treatments of molecular tumbling. The dynamics of 109 backbone amides could be investigated (out of 140 possible, excluding the amino terminal and proline residues). Of these, the relaxation rates for 69 amides were best fit by model 1 (S2), 17 by model 2 (S2 and τe), 14 by model 3 (S2 and Rex), 5 by model 4 (S2, τe and Rex) and 4 by model 5 (Sf2, Ss2 and τe).

Overall, the backbone of free EcPDF exhibits restricted mobility on the ns-ps timescale, with an average S2 of 0.86 ± 0.06 (Figure 6). Locally, the loop containing the conserved GxGLAAxQ motif (residues 43–50), and helix α2, bearing the QHExDH motif (residues 131–136), exhibit average or above-average S2 values. However, there are regions of the protein that are more flexible on this timescale, such as residues 97–102 in the loop between β4- β5 near the actinonin binding site. Other residues displaying lower than average error-adjusted order parameters include Asp10, Asn24, Glu68, Arg97, Arg102, Val106, Lys107, Asp123, Gly124, and the penultimate Leu146 (Figure 6). Of the residues within 5 Å of the actinonin binding site or involved in metal binding, Ile44, Gly45, Leu46, Glu88 and Glu133 required (Rex) chemical exchange terms to fit the relaxation data, while Gly42, Glu87 and Glu88 were best fit by including an internal correlation time (τe).

Figure 6
Backbone amide 15N ‘model-free’ parameters for free (black) and actinonin-bound EcPDF (red). S2, square of the generalized order parameter; τe, effective correlation time for internal motions; Rex, phenomenological exchange term ...

Model-free parameters for the actinonin-EcPDF complex were obtained for the 104 backbone amides; relaxation data of 73 amides were best fit by model 1, 20 by model 2, 7 by model 3, 2 by model 4 and 2 by model 5. The overall ns-ps dynamics of EcPDF in complex with actinonin is similar to that of the free protein, with an average S2 of 0.86 ± 0.05 (Figure 6). In the complex, lower than average error-adjusted order parameters were still seen for Gln4, Asp10, Val16, Asn24, Arg69, Glu79, Thr84, Arg97, Arg102, Asp123, Gly124, and the penultimate Leu146, though most of the other residues near actinonin exhibit average or above average S2 values. In the complex, the entirety of the central QHExDH-bearing helix α2 displays above-average S2 except Gly124, at the beginning of the helix, which is more dynamic than the average (Figure 6).

Although the protein experienced clear and measurable ligand-induced differences in the amplitudes of internal motions (Figure 3), the difference in S2 values are relativity small, implying no large changes in overall backbone dynamics. In all, 22 of 97 amides exhibited significant (error-adjusted) changes in order upon actinonin binding. Inspection of the differences in backbone order parameters (Figure 6, Supporting Information), or as translated to conformational entropy and mapped onto the structure of the complex (Figure 4c), reveals decreased dynamics for Thr49, Gln50, Lue91, Ile93, Met134, and Leu137 all of which are in the ligand binding site. However, equally noteworthy are the increased dynamics both within the binding site and in other parts of the protein (e.g., Arg12, Gly45, Leu47, Arg97, Val106, Leu125 and Glu133). It is clear from these observations that the effects of ligand binding are not limited to the residues immediately surrounding the binding cavity. In addition, several residues experience change in dynamics manifested by a decrease in the complexity of the model needed to describe their molecular motion. In particular, several residues do not contain an Rex and τe term upon complex formation, which implies a change in motions on the µs-ms timescale (Figure 6).

Hydrogen/Deuterium Exchange

Backbone amide hydrogen-deuterium exchange rates under native conditions report on the local thermodynamic stability of protein folded states (68), and thus can be used to probe ligand-induced local stabilization. Deuterium exchange experiments show significant regions of both free EcPDF and actinonin-bound EcPDF for which backbone amides are completely protected from solvent and do not exchange after 10 hours, as well as regions that are exposed to solvent and exchange within 20 minutes. Changes in the solvent exchange rate due to complex formation were obtained for 107 residues (Figure 3). Figure 4d shows the measured change in exchange rates mapped to the structure of EcPDF.

As expected, actinonin binding resulted in significant protection from exchange for several residues within the ligand binding site, including Gly43, Gly45, Ala47, Thr49, Gly89, Cys90, Leu91, and Ser92 (Figure 3, Figure 4). However, these local stabilizing effects appear to be compensated in part by destabilizing effects at remote sites in the protein (e.g., Ile27, Gln28, Arg29, Val62, Glu76, Leu78, Glu79, Ser81, Ile93, Val100, Val106, Ile108, Ala110, Leu111, Asp112, Gly115, Phe118, and Leu120).

Structural thermodynamics

Thermodynamic parameters associated with water release from the molecular surfaces (i.e. the hydrophobic effect) were estimated from the change in exposed surface area of the crystal structures (69). The total change in accessible surface area for both the protein and ligand is 775.76 Å2 (polar 284.69 Å2 and non-polar 491.07 Å2), calculated using a solvent radius of 1.4 Å. The solvation entropy was estimated to be: ΔSsol = 43 cal mol−1 K−1 (TΔS = 13.3 kcal mol−1 K−1 at 310 K). These same structure-based thermodynamic calculations predict an overall entropic contribution to binding, including entropy losses from changes in internal and overall rotational and translational degrees of freedom, of 7.8 kcal mol−1 at 310 K (Supporting Information), which compares favorably to the calorimetrically-measured 8.4 ± 1.3 kcal mol−1.

DISCUSSION

Effect of actinonin on the structure of EcPDF

At the level of the peptide backbone, EcPDF is well structured and rigid in the absence of ligands. This rigidity contrasts with related members of the broader metallohydrolase superfamily, many of which exhibit a high degree of protein flexibility at the level of the peptide backbone (33, 82). This difference may be related to the relatively narrow substrate specificity of PDF, for which the substrates are peptides bearing N-formylated methionine residues (21, 83), while other metallohydrolases have a broader range of substrates that may necessitate greater flexibility in the substrate binding site.

Despite the atomic level detail provided by the crystal structure of the EcPDF-actinonin complex of the interactions between the protein and inhibitor (Figure 1), it is nevertheless difficult to appreciate from the crystal structure the extent to which entropic effects contribute to binding. The large entropic effects of desolvation in the protein ligand interface notwithstanding, the effect of the more subtle but concerted ligand-induced deformation on thermodynamics of ligand binding are difficult to appreciate from static structures. Indeed, the differences between molecules in the asymmetric unit are on the same order of magnitude as the differences between the free and ligand bound protein (Supporting Information). Although backbone atom displacements are emphasized in this paper (since the NMR measurements employed here probe the backbone), the variability in the sidechain displacements in the crystal structures follow a similar pattern, albeit with larger amplitudes.

NMR chemical shift is very sensitive to local electronic effects, which depend on the equilibrium local structure and are sensitive to rapid conformational sampling. The strong correlation between the chemical shift data and distance to the ligand-binding site (Figure 3 and Figure 4a) indicates that the largest ligand-induced changes occur in the immediate vicinity of the binding site. However, the NMR data also clearly show that ligand binding affects the structure of the protein at sites distant from the ligand-binding pocket; unfortunately, the chemical shift data alone do not reveal the nature of the change, since there is not a simple relationship between structure and chemical shift. Nevertheless, the chemical shift data suggest that structural changes are propagated from the ligand-binding site to remote sites, presumably via a network of interactions within the protein (main-chain and/or sidechain).

Residual dipolar couplings (rDCs) directly report the average orientation in solution of individual bond vectors (relative to the field). Comparing the crystallographic (Figure 3a) and rDC data (Figure 3c) it is clear that the crystallographic data fail to reveal some details of the structural changes induced by the ligand. The low correlation between the actinonin-induced shift perturbations and rDC data in solution, and the pairwise differences between the free and actinonin-bound structures in the crystal, suggest that some actinonin-dependent structural differences that are detected by the spectroscopic measurements in solution are lost to lattice constraints during crystallization (Figure 5d).

Effect of actinonin on EcPDF thermodynamics

Under native conditions (i.e., the EX2 regime (68)), hydrogen/deuterium exchange rates reflect local thermodynamic stability. As expected, there is an increase in stability for residues adjacent to the site of actinonin binding (e.g., α1-β1 and β4-β5). However, there are several regions of EcPDF for which binding of the ligand is destabilizing (i.e., experience increased H/D exchange rates), particularly in β-strands 3, 6 and 7, on the opposite side of the protein of the ligand binding site (Figure 3, Figure 4). Thus, some of the local stabilizing effects seem to be countered by destabilization elsewhere in the protein.

An objective of this study was to use the changes in backbone 15N order parameters (S2) for individual backbone amides to obtain an estimate of the overall change in conformational entropy imposed by ligand binding (64). ‘Model-free’ analysis of the backbone 15N relaxation data has provided a map, albeit incomplete, of the residual entropy of EcPDF before and after binding of actinonin (Figure 3d and Figure 4c). Two general features are noteworthy for this analysis (summarized in Figure 3d): First, the overall change in ΔSconf is small. Of the 97 residues for which order parameters could be measured in both free and bound EcPDF, only 22 amides exhibit entropy changes (TΔSconf) greater than the error in the measurements; averaged over all amides with changes greater than the error, the magnitude of TΔSconf is 0.21 kcal mol−1. Second, local loss of conformational entropy is almost entirely compensated by entropy gains at other sites. Thus, as measured from the ps-ns dynamics of the backbone amides, given the experimental uncertainties, and ignoring possible contribution from residues for which order parameters could not be measured, the net effect of actinonin binding on the overall conformational entropy of EcPDF appears to be negligible.

Conclusions

Successful structure-based inhibitor design requires knowledge of the local energy minimum defined by the structures and interactions of the target molecule and ligand. High-resolution structures derived from X-ray crystallography and NMR spectroscopy can provide a description of the local energy minimum and serve as templates for design. However, limited resolution, induced-fit binding and structural plasticity complicate these objectives, partially because the structures provide limited insights into the steepness of the free energy well that surrounds this minimum, and thus little understanding of the cost of deforming the conformation of the protein (or ligand) away from this target structure. Measurement of dynamics parameters (relaxation and rDCs) and H/D exchange rates for individual bond vectors can provide important insights both into the local energetic cost of deforming the equilibrium structure and a more accurate picture of the global minimum.

Experimental data show that the catalytic domain of EcPDF adopts a well defined, globular structure that is held together by an interwoven network of intramolecular interactions. The site-specific measurements of the effect of actinonin binding on EcPDF (i.e., chemical shifts, backbone relaxation data, H/D exchange and rDC) highlight significant perturbations in two regions: the ligand binding site (residues 40–50, 85–95), and on the opposite side of the protein, particularly in strand β6 (Figure 3, Figure 4). It is apparent from these observations that local structural perturbations in the binding site are transmitted via these interaction networks to other parts of the protein. Given the apparent rigidity of the PDF active site, and the consequent ruling out of induced fit in recognition, it may be difficult to identify selective inhibitors of the enzyme. Nevertheless, the NMR measurements are acutely sensitive to these (even subtle) structural perturbations, providing unique insights into these interaction networks, and the shape of the global minimum.

Thus, it would seem that the distribution of conformational entropy upon actinonin binding to EcPDF is consistent with the overall thermodynamic picture: that local loss of flexibility is compensated for by increased dynamics elsewhere in the protein. As a general disclaimer, however, we note that conclusions regarding the overall change in conformational entropy are limited by several factors. Among these are incomplete sampling (changes in order parameters could not be measured for all bond vectors in the protein), insufficient precision (changes that are smaller than the error in the measurements), and the implicit assumption that backbone motions are not correlated (thus representing independent degrees of freedom)(37). Further, this analysis is not sensitive to motions that do not reorient amide bond vectors relative to the molecular frame, entropy changes associated with conformational states sampled on slower timescales (> µs), nor does it include changes in entropy associated with other bond vectors (i.e. those whom dynamics are not correlated to the backbone amides) throughout the enzyme.

In summary, assessment in EcPDF of actinonin-induced chemical shift perturbations and changes in backbone dynamics reveal significant structural and dynamic effects at sites both adjacent to and remote from that of ligand binding. Comparison of rDC data for free and actinonin-bound EcPDF and to the crystal structures reveals changes in the equilibrium structure of EcPDF that are not detected in the crystalline state, presumably due to crystal packing forces. Finally, quantitation in EcPDF of the conformational entropy change that accompanies actinonin binding indicates that net sum of the changes is small but measurable. This result is in agreement with previous findings that entropy changes can be readily redistributed from one site to another in a protein, with important thermodynamic implications. They underscore the premise that binding events are more complex than generally assumed, and that an accurate descriptor of molecular behavior will need to incorporate both local and long-range contributions to the global energy minimum and the depth and width of the local potential energy well.

Supplementary Material

ACKNOWLEDGEMENTS

We thank R. Wilson, M. Chan, D. Pei, R. Rajagopalan and Z. Yuan for reagents and helpful discussions, C. Cottrell and C. Yuan for help with NMR instrumentation.

This work was supported in part by grants to M.F. from the American Heart Association/Ohio Valley Affiliate (AHA 0060418B) and the National Science Foundation (MCB 0092962).

ABBREVIATIONS

EcPDF
E. coli peptide deformylase
NMR
nuclear magnetic resonance spectroscopy
rDC
residual dipolar couplings
ITC
isothermal titration calorimetry
CSA
chemical shift anisotropy
S2
square of the generalized order parameter from modelfree analysis
H/D exchange
exchange of hydrogen (protons) with deuterium
RMSD
root mean square difference.

Footnotes

SUPPLEMENTAL INFORMATION

Supporting Information. Tables of buffer ionization enthalpies and by-residue structure-thermodynamic parameters, parameters and results of structure-based thermodynamic calculations, graphs of by-residue relaxation data, temperature-dependent enthalpy, salt-dependence of ligand binding, representative best fit lines to relaxation data, H/D exchange, and order parameters. Protein sequence alignments, cartoon diagrams of free and actinonin-bound EcPDF, actinonin-induced displacements, order parameters mapped to the protein backbone, and diagram of actinonin-PDF contacts. Supplemental materials may be accessed free of charge online at http://pubs.acs.org.

References

1. Mazel D, Pochet S, Marliere P. Genetic characterization of polypeptide deformylase, a distinctive enzyme of eubacterial translation. EMBO J. 1994;13:914–923. [PubMed]
2. Meinnel T, Blanquet S, Dardel F. A new subclass of the zinc metalloproteases superfamily revealed by the solution structure of peptide deformylase. J Mol Biol. 1996;262:375–386. [PubMed]
3. Meinnel T, Lazennec C, Villoing S, Blanquet S. Structure-function relationships within the peptide deformylase family. Evidence for a conserved architecture of the active site involving three conserved motifs and a metal ion. J Mol Biol. 1997;267:749–761. [PubMed]
4. Rajagopalan PT, Datta A, Pei D. Purification, characterization, and inhibition of peptide deformylase from Escherichia coli. Biochemistry. 1997;36:13910–13918. [PubMed]
5. Rajagopalan PTR, Yu XC, Pei D. Peptide deformylase: a new type of mononuclear iron protein. J Am Chem Soc. 1997;119:12418–12419.
6. Ragusa S, Mouchet P, Lazennec C, Dive V, Meinnel T. Substrate recognition and selectivity of peptide deformylase. Similarities and differences with metzincins and thermolysin. J Mol Biol. 1999;289:1445–1457. [PubMed]
7. Bracchi-Ricard V, Nguyen KT, Zhou Y, Rajagopalan PT, Chakrabarti D, Pei D. Characterization of an eukaryotic peptide deformylase from Plasmodium falciparum. Arch Biochem Biophys. 2001;396:162–170. [PubMed]
8. Giglione C, Serero A, Pierre M, Boisson B, Meinnel T. Identification of eukaryotic peptide deformylases reveals universality of N-terminal protein processing mechanisms. EMBO J. 2000;19:5916–5929. [PubMed]
9. Serero A, Giglione C, Meinnel T. Distinctive features of the two classes of eukaryotic peptide deformylases. J Mol Biol. 2001;314:695–708. [PubMed]
10. Meinnel T. Peptide deformylase of eukaryotic protists: a target for new antiparasitic agents? Parasitol Today. 2000;16:165–168. [PubMed]
11. Yuan Z, Trias J, White RJ. Deformylase as a novel antibacterial target. Drug Discov Today. 2001;6:954–961. [PubMed]
12. Apfel CM, Locher H, Evers S, Takacs B, Hubschwerlen C, Pirson W, Page MG, Keck W. Peptide deformylase as an antibacterial drug target: target validation and resistance development. Antimicrob Agents Chemother. 2001;45:1058–1064. [PMC free article] [PubMed]
13. Guilloteau J-P, Mathieu M, Giglione C, Blanc V, Dupuy A, Chevrier M, Gil P, Famechon A, Meinnel T, Mikol V. The crystal structures of four peptide deformylases bound to the antibiotic actinonin reveal two distinct types: a platform for the structure-based design of antibacterial agents. J Mol Biol. 2002;320:951–962. [PubMed]
14. Waller AS, Clements JM. Novel approaches to antimicrobial therapy: peptide deformylase. Curr Opin Drug Discov Devel. 2002;5:785–792. [PubMed]
15. Smith KJ, Petit CM, Aubart K, Smyth M, McManus E, Jones J, Fosberry A, Lewis C, Lonetto M, Christensen SB. Structural variation and inhibitor binding in polypeptide deformylase from four different bacterial species. Protein Sci. 2003;12:349–360. [PubMed]
16. Guay DR. Drug forecast - the peptide deformylase inhibitors as antibacterial agents. Ther Clin Risk Manag. 2007;3:513–525. [PMC free article] [PubMed]
17. Meinnel T, Blanquet S. Enzymatic properties of Escherichia coli peptide deformylase. J Bacteriol. 1995;177:1883–1887. [PMC free article] [PubMed]
18. Wei Y, Yi T, Huntington KM, Chaudhury C, Pei D. Identification of a potent peptide deformylase inhibitor from a rationally designed combinatorial library. J Comb Chem. 2000;2:650–657. [PubMed]
19. Meinnel T, Patiny L, Ragusa S, Blanquet S. Design and synthesis of substrate analogue inhibitors of peptide deformylase. Biochemistry. 1999;38:4287–4295. [PubMed]
20. Huntington KM, Yi T, Wei Y, Pei D. Synthesis and antibacterial activity of peptide deformylase inhibitors. Biochemistry. 2000;39:4543–4551. [PubMed]
21. Hu YJ, Wei Y, Zhou Y, Rajagopalan PT, Pei D. Determination of substrate specificity for peptide deformylase through the screening of a combinatorial peptide library. Biochemistry. 1999;38:643–650. [PubMed]
22. Hao B, Gong W, Rajagopalan PT, Zhou Y, Pei D, Chan MK. Structural basis for the design of antibiotics targeting peptide deformylase. Biochemistry. 1999;38:4712–4719. [PubMed]
23. Clements JM, Beckett RP, Brown A, Catlin G, Lobell M, Palan S, Thomas W, Whittaker M, Wood S, Salama S, Baker PJ, Rodgers HF, Barynin V, Rice DW, Hunter MG. Antibiotic activity and characterization of BB-3497, a novel peptide deformylase inhibitor. Antimicrob Agents Chemother. 2001;45:563–570. [PMC free article] [PubMed]
24. Boularot A, Giglione C, Petit S, Duroc Y, Alves de Sousa R, Larue V, Cresteil T, Dardel F, Artaud I, Meinnel T. Discovery and refinement of a new structural class of potent peptide deformylase inhibitors. J Med Chem. 2007;50:10–20. [PubMed]
25. Chan MK, Gong W, Rajagopalan PT, Hao B, Tsai CM, Pei D. Crystal structure of the Escherichia coli peptide deformylase. Biochemistry. 1997;36:13904–13909. [PubMed]
26. Becker A, Schlichting I, Kabsch W, Groche D, Schultz S, Wagner AF. Iron center, substrate recognition and mechanism of peptide deformylase. Nat Struct Biol. 1998;5:1053–1058. [PubMed]
27. Becker A, Schlichting I, Kabsch W, Schultz S, Wagner AF. Structure of peptide deformylase and identification of the substrate binding site. J Biol Chem. 1998;273:11413–11416. [PubMed]
28. Meinnel T, Lazennec C, Dardel F, Schmitter JM, Blanquet S. The C-terminal domain of peptide deformylase is disordered and dispensable for activity. FEBS Lett. 1996;385:91–95. [PubMed]
29. Dardel F, Ragusa S, Lazennec C, Blanquet S, Meinnel T. Solution structure of nickel-peptide deformylase. J Mol Biol. 1998;280:501–513. [PubMed]
30. Baldwin ET, Harris MS, Yem AW, Wolfe CL, Vosters AF, Curry KA, Murray RW, Bock JH, Marshall VP, Cialdella JI, Merchant MH, Choi G, Deibel MR., Jr Crystal structure of type II peptide deformylase from Staphylococcus aureus. J Biol Chem. 2002;277:31163–31171. [PubMed]
31. Harris MS, Bock JH, Choi G, Cialdella JS, Curry KA, Deibel MR, Jr, Jacobsen EJ, Marshall VP, Murray RW, Jr, Vosters AF, Wolfe CL, Yem AW, Baldwin ET. Co-crystallization of Staphylococcus aureus peptide deformylase (PDF) with potent inhibitors. Acta Crystallogr D Biol Crystallogr. 2002;58:2153–2156. [PubMed]
32. Kumar A, Nguyen KT, Srivathsan S, Ornstein B, Turley S, Hirsh I, Pei D, Hol WGJ. Crystals of peptide deformylase from Plasmodium falciparum reveal critical characteristics of the active site for drug design. Structure. 2002;10:357–367. [PubMed]
33. Moy FJ, Chanda PK, Chen J, Cosmi S, Edris W, Levin JI, Rush TS, Wilhelm J, Powers R. Impact of mobility on structure-based drug design for the MMPs. J Am Chem Soc. 2002;124:12658–12659. [PubMed]
34. Spolar RS, Record MT., Jr Coupling of local folding to site-specific binding of proteins to DNA. Science. 1994;263:777–784. [PubMed]
35. Bax A, Kontaxis G, Tjandra N. Dipolar couplings in macromolecular structure determination. Methods Enzymol. 2001;339:127–174. [PubMed]
36. Li Z, Raychaudhuri S, Wand AJ. Insights into the local residual entropy of proteins provided by NMR relaxation. Protein Sci. 1996;5:2647–2650. [PubMed]
37. Yang D, Kay LE. Contributions to conformational entropy arising from bond vector fluctuations measured from NMR-derived order parameters: application to protein folding. J Mol Biol. 1996;263:369–382. [PubMed]
38. Stivers JT, Abeygunawardana C, Mildvan AS. 15N NMR relaxation studies of free and inhibitor-bound 4-oxalocrotonate tautomerase: backbone dynamics and entropy changes of an enzyme upon inhibitor binding. Biochemistry. 1996;35:16036–16047. [PubMed]
39. Kay LE, Muhandiram DR, Wolf G, Shoelson SE, Forman-Kay JD. Correlation between binding and dynamics at SH2 domain interfaces. Nat Struct Biol. 1998;5:156–163. [PubMed]
40. Stone MJ. NMR relaxation studies of the role of conformational entropy in protein stability and ligand binding. Acc Chem Res. 2001;34:379–388. [PubMed]
41. Olejniczak ET, Zhou MM, Fesik SW. Changes in the NMR-derived motional parameters of the insulin receptor substrate 1 phosphotyrosine binding domain upon binding to an interleukin 4 receptor phosphopeptide. Biochemistry. 1997;36:4118–4124. [PubMed]
42. Lee AL, Sharp KA, Kranz JK, Song XJ, Wand AJ. Temperature dependence of the internal dynamics of a calmodulin-peptide complex. Biochemistry. 2002;41:13814–13825. [PubMed]
43. Kovrigin EL, Cole R, Loria JP. Temperature dependence of the backbone dynamics of ribonuclease A in the ground state and bound to the inhibitor 5'-phosphothymidine (3'-5')pyrophosphate adenosine 3'-phosphate. Biochemistry. 2003;42:5279–5291. [PubMed]
44. Arumugam S, Gao G, Patton BL, Semenchenko V, Brew K, Van Doren SR. Increased backbone mobility in beta-barrel enhances entropy gain driving binding of N-TIMP-1 to MMP-3. J Mol Biol. 2003;327:719–734. [PubMed]
45. Kay LE, Torchia DA, Bax A. Backbone dynamics of proteins as studied by 15N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry. 1989;28:8972–8979. [PubMed]
46. Lipari G, Szabo A. A Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. theory and range of validity. J. Am. Chem. Soc. 1982;104:4546–4559.
47. Byerly DW, McElroy CA, Foster MP. Mapping the surface of Escherichia coli peptide deformylase by NMR with organic solvents. Protein Sci. 2002;11:1850–1853. [PubMed]
48. Gordon JJ, Devlin JP, East AJ, Ollis WD, Wright DE, Ninet L. Studies concerning the antibiotic actinonin. Part I. The constitution of actinonin. A natural hydroxamic acid with antibiotic activity. J Chem Soc Perkin. 1975;1:819–825. [PubMed]
49. Piotto M, Saudek V, Sklenar V. Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions. J Biomol NMR. 1992;2:661–665. [PubMed]
50. Sklenar V, Piotto M, Leppik R, Saudek V. Gradient-tailored water suppression for 1H-15N HSQC experiments optimized to retain full sensitivity. J Magn Reson A. 1993;102:241–245.
51. Sattler M, Schleucher J, Griesinger C. Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Progress In Nuclear Magnetic Resonance Spectroscopy. 1999;34:93–158.
52. Cavanagh FPSR. Protein NMR Spectroscopy: Principles and Practice. 2006
53. Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, Bax A. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR. 1995;6:277–293. [PubMed]
54. Johnson BA. Using NMRView to visualize and analyze the NMR spectra of macromolecules. Methods Mol Biol. 2004;278:313–352. [PubMed]
55. Grzesiek S, Bax A, Clore GM, Gronenborn AM, Hu JS, Kaufman J, Palmer I, Stahl SJ, Wingfield PT. The solution structure of HIV-1 Nef reveals an unexpected fold and permits delineation of the binding surface for the SH3 domain of Hck tyrosine protein kinase. Nat Struct Biol. 1996;3:340–345. [PubMed]
56. Barbato G, Ikura M, Kay LE, Pastor RW, Bax A. Backbone dynamics of calmodulin studied by 15N relaxation using inverse detected two-dimensional NMR spectroscopy: the central helix is flexible. Biochemistry. 1992;31:5269–5278. [PubMed]
57. LeMaster DM, Kay LE, Brunger AT, Prestegard JH. Protein dynamics and distance determination by NOE measurements. FEBS Lett. 1988;236:71–76. [PubMed]
58. Clore GM, Szabo A, Bax A, Kay LE, Driscoll PC, Gronenborn AM. Deviations from the simple two-parameter model-free approach to the interpretation of nitrogen-15 muclear magnetic relaxation of proteins. J. Am. Chem. Soc. 1990;112:4989–4991.
59. d'Auvergne EJ, Gooley PR. Optimisation of NMR dynamic models II. A new methodology for the dual optimisation of the model-free parameters and the Brownian rotational diffusion tensor. J Biomol NMR. 2008;40:121–133. [PMC free article] [PubMed]
60. Koradi R, Billeter M, Wuthrich K. MOLMOL: a program for display and analysis of macromolecular structures. J Mol Graph. 1996;14:51–55. 29–32. [PubMed]
61. d'Auvergne EJ, Gooley PR. The use of model selection in the model-free analysis of protein dynamics. J Biomol NMR. 2003;25:25–39. [PubMed]
62. d'Auvergne EJ, Gooley PR. Optimisation of NMR dynamic models I. Minimisation algorithms and their performance within the model-free and Brownian rotational diffusion spaces. J Biomol NMR. 2008;40:107–119. [PMC free article] [PubMed]
63. Palmer AG., 3rd Nmr probes of molecular dynamics: overview and comparison with other techniques. Annu Rev Biophys Biomol Struct. 2001;30:129–155. [PubMed]
64. Yang D, Mok YK, Forman-Kay JD, Farrow NA, Kay LE. Contributions to protein entropy and heat capacity from bond vector motions measured by NMR spin relaxation. J Mol Biol. 1997;272:790–804. [PubMed]
65. Ottiger M, Delaglio F, Bax A. Measurement of J and dipolar couplings from simplified two-dimensional NMR spectra. J Magn Reson. 1998;131:373–378. [PubMed]
66. Clore GM, Gronenborn AM, Bax A. A robust method for determining the magnitude of the fully asymmetric alignment tensor of oriented macromolecules in the absence of structural information. J Magn Reson. 1998;133:216–221. [PubMed]
67. Zweckstetter M. Prediction of charge-induced molecular alignment: residual dipolar couplings at pH 3 and alignment in surfactant liquid crystalline phases. Eur Biophys J. 2006;35:170–180. [PubMed]
68. Dempsey CE. Hydrogen exchange in peptides and proteins using NMR spectroscopy. Progress in Nuclear Magnetic Resonance Spectroscopy. 2001;39:135–170.
69. Lavigne P, Bagu JR, Boyko R, Willard L, Holmes CF, Sykes BD. Structure-based thermodynamic analysis of the dissociation of protein phosphatase-1 catalytic subunit and microcystin-LR docked complexes. Protein Sci. 2000;9:252–264. [PubMed]
70. Simmons A. Department of Biochemistry. Columbus, OH: The Ohio State University; 2007. B.S. Thesis: Inhibition and thermodynamics of Escherichia coli peptide deformylase.
71. Berg AK, Srivastava DK. Delineation of alternative conformational states in Escherichia coli peptide deformylase via thermodynamic studies for the binding of actinonin. Biochemistry. 2009;48:1584–1594. [PMC free article] [PubMed]
72. Wiseman T, Williston S, Brandts JF, Lin LN. Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal Biochem. 1989;179:131–137. [PubMed]
73. Jelesarov I, Bosshard HR. Isothermal titration calorimetry and differential scanning calorimetry as complementary tools to investigate the energetics of biomolecular recognition. J Mol Recognit. 1999;12:3–18. [PubMed]
74. Missimer JH, Steinmetz MO, Baron R, Winkler FK, Kammerer RA, Daura X, van Gunsteren WF. Configurational entropy elucidates the role of salt-bridge networks in protein thermostability. Protein Sci. 2007;16:1349–1359. [PubMed]
75. Tsumoto K, Ogasahara K, Ueda Y, Watanabe K, Yutani K, Kumagai I. Role of salt bridge formation in antigen-antibody interaction. Entropic contribution to the complex between hen egg white lysozyme and its monoclonal antibody HyHEL10. J Biol Chem. 1996;271:32612–32616. [PubMed]
76. Prabhu NV, Sharp KA. Heat capacity in proteins. Annu Rev Phys Chem. 2005;56:521–548. [PubMed]
77. Baker BM, Murphy KP. Evaluation of linked protonation effects in protein binding reactions using isothermal titration calorimetry. Biophys J. 1996;71:2049–2055. [PubMed]
78. Goldberg RNKNLRM. Thermodynamic Quantities for the Ionization Reactions of Buffers. J. Phys. Chem. Ref. Data. 2002;31:231–370.
79. Maret W, Vallee BL. Cobalt as probe and label of proteins. Methods Enzymol. 1993;226:52–71. [PubMed]
80. Rajagopalan PT, Grimme S, Pei D. Characterization of cobalt(II)-substituted peptide deformylase: function of the metal ion and the catalytic residue Glu-133. Biochemistry. 2000;39:779–790. [PubMed]
81. Dosset P, Hus JC, Marion D, Blackledge M. A novel interactive tool for rigid-body modeling of multi-domain macromolecules using residual dipolar couplings. J Biomol NMR. 2001;20:223–231. [PubMed]
82. Cuniasse P, Devel L, Makaritis A, Beau F, Georgiadis D, Matziari M, Yiotakis A, Dive V. Future challenges facing the development of specific active-site-directed synthetic inhibitors of MMPs. Biochimie. 2005;87:393–402. [PubMed]
83. Ragusa S, Mouchet P, Lazennec C, Dive V, Meinnel T. Substrate recognition and selectivity of peptide deformylase. Similarities and differences with metzincins and thermolysin. J Mol Biol. 1999;289:1445–1457. [PubMed]