Heat Capacity Changes and Disorder-to-Order Transitions in Allosteric

Article pubs.acs.org/biochemistry

Heat Capacity Changes and Disorder-to-Order Transitions in Allosteric Activation William J. Cressman and Dorothy Beckett* Center for Biomolecular Structure and Organization, Department of Chemistry & Biochemistry, University of Maryland, College Park, Maryland 20742, United States ABSTRACT: Allosteric coupling in proteins is ubiquitous but incompletely understood, particularly in systems characterized by coupling over large distances. Binding of the allosteric effector, bio-5′-AMP, to the Escherichia coli biotin protein ligase, BirA, enhances the protein’s dimerization free energy by −4 kcal/mol. Previous studies revealed that disorder-to-order transitions at the effector binding and dimerization sites, which are separated by 33 Å, are integral to functional coupling. Perturbations to the transition at the ligand binding site alter both ligand binding and coupled dimerization. Alanine substitutions in four loops on the dimerization surface yield a range of energetic effects on dimerization. A glycine to alanine substitution at position 142 in one of these loops results in a complete loss of allosteric coupling, disruption of the disorder-to-order transitions at both functional sites, and a decreased affinity for the effector. In this work, allosteric communication between the effector binding and dimerization surfaces in BirA was further investigated by performing isothermal titration calorimetry measurements on nine proteins with alanine substitutions in three dimerization surface loops. In contrast to BirAG142A, at 20 °C all variants bind to bio-5′-AMP with free energies indistinguishable from that measured for wild-type BirA. However, the majority of the variants exhibit altered heat capacity changes for effector binding. Moreover, the ΔCp values correlate with the dimerization free energies of the effector-bound proteins. These thermodynamic results, combined with structural information, indicate that allosteric activation of the BirA monomer involves formation of a network of intramolecular interactions on the dimerization surface in response to bio-5′-AMP binding at the distant effector binding site.

A

loops on the ligand binding and dimerization surfaces function in substrate and corepressor binding, dimerization, and allostery.11−15 Disorder-to-order transitions at the corepressor binding site contribute to allosteric activation of the BirA monomer. Two loops in the bio-5′-AMP binding site, the biotin binding loop (BBL, residues 116−128) and the adenylate binding loop (ABL, residues 206−233), are disordered in apoBirA (Figure 110). Although BBL folding is coupled to biotin binding, the binding does not activate BirA dimerization.8,16 Effector binding is accompanied by folding of both the BBL and ABL with concomitant formation of a network of interacting hydrophobic side chains, and alanine substitutions in this network perturb loop folding as well as bio-5′-AMP synthesis and binding.13,15,17 These alanine substitutions also result in changes to the coupling free energy between binding and dimerization that range from 0.5 to 1.5 kcal/mol.13,15 Thus, the disorder-to-order transition at the effector binding site contributes to allosteric activation of dimerization. Folding transitions on the dimerization surface are also coupled to self-association and corepressor binding by BirA. Five loops on this surface, which is separated from the corepressor binding site by 33 Å, function in dimerization.11,14 Among these loops, the BBL and loops containing residues 140−146 and

lthough allosteric regulation is ubiquitous in biology, the mechanisms by which allosteric communication is achieved remain to be elucidated. Specifically, despite intense research, the molecular details of how allosteric signals are transmitted through the protein matrix to functionally couple, frequently distant, sites are the subject of much debate.1,2 A prerequisite for determining transmission mechanisms is elucidation of the structural and functional changes at the end points of the communication process that are integral to allosteric control. The Escherichia coli biotin repressor provides a model system for investigating allosteric communication mechanisms. The protein functions as both an essential metabolic enzyme and a transcription repressor.3−5 As an enzyme, BirA binds to biotin and ATP to synthesize the intermediate, bio-5′-AMP, from which biotin is transferred to the biotin carboxyl carrier protein of acetyl-CoA carboxylase.6 Alternatively, the enzyme−intermediate complex homodimerizes and binds to DNA to repress the initiation of transcription at the biotin biosynthetic operon.7 In the latter function, bio-5′-AMP acts as an allosteric effector/ corepressor by enhancing the BirA dimerization free energy by −4.0 kcal/mol (Figure 18,9). Multiple surface loops that are disordered in apoBirA play important functional roles (Figure 1). The BirA monomer is a three-domain protein in which the DNA binding domain adopts a winged helix−turn−helix fold, the central domain is characterized by a core mixed β sheet surrounded by α helices, and the C-terminal domain forms an all-β fold that is similar to an SH3 domain (Figure 110). In addition to the folded core, several © XXXX American Chemical Society

Received: August 25, 2015 Revised: November 27, 2015

A

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

Table 1. Dimerization Properties of Bio-5′-AMP-Bound BirA Variantsa protein wt E140A G142A P143A T195A G196A D197A I280A G281A K283A

Kdim (M) −6

(7 ± 3) × 10 (22 ± 5) × 10−6 (17 ± 52) × 10−3 (4 ± 1) × 10−5 (10 ± 4) × 10−5 (5 ± 4) × 10−6 (5 ± 3) × 10−4 (4 ± 3) × 10−3 (9 ± 4) × 10−5 (12 ± 7) × 10−6

ΔG°dim (kcal/mol)

fraction dimerb

−6.8 ± 0.3 −6.2 ± 0.2 −2.4 ± 0.9 −5.9 ± 0.2 −5.3 ± 0.2 −7.1 ± 0.4 −4.5 ± 0.3 −3.3 ± 0.7 −5.4 ± 0.3 −6.6 ± 0.4

0.272 0.125 0.0001 0.077 0.034 0.327 0.007 0.001 0.037 0.195

a Dimerization parameters were previously measured in standard buffer at 20 °C,11 with values reported for G142A and I280A BirA lower limits. bCalculated at the 1.8 μm final protein concentration present in the bio-5′-AMP titrations.

In this work, results of isothermal titration calorimetry measurements of bio-5′-AMP binding by BirA variants are reported. As stated above, disruption of ABL folding, either through direct alanine substitutions in the ABL or via the G142A substitution, elicits penalties ranging from 3 to 4 kcal/mol to the bio-5′-AMP binding free energy.13,15,18 Therefore, alanine substitutions at dimerization surface residues that disrupt the ABL disorder-to-order transition are expected to have large effects on the bio-5′-AMP binding free energy. However, combined direct and displacement titration calorimetry measurements of biotin and bio-5′-AMP binding to dimerization surface variants reveal that at 20 °C only the substitution at G142 has a significant effect on the Gibbs free energy of corepressor binding. Molar heat capacity changes of bio-5′-AMP binding to the variants, which can provide information about structural changes that accompany binding, were obtained from measurements of the temperature dependence of corepressor binding enthalpy.19,20 The results reveal a correlation between the magnitude of the heat capacity change associated with adenylate binding to the BirA monomer and the Gibbs free energy of dimerization. This correlation in combination with structural data suggests that allosteric activation of BirA for dimerization involves formation of a network of intramolecular interactions that incorporates multiple protein segments that are disordered in apoBirA.

Figure 1. Bio-5′-AMP binding allosterically activates the BirA monomer for homodimerization. The biotin binding (blue) and adenylate binding (red) loops are disordered in apoBirA and ordered in holoBirA. Surface loops comprised of residues 140−146 (green), 170−175 (orange), 193−199 (pink), and 280−283 (brown) participate in homodimerization. The 140−146 and 193−199 loops are partially disordered in apoBirA. The bottom image shows the cross section of the dimer interface viewed from the top of the dimer image shown above it. Models were created in Pymol using input files 1BIA (apoBirA) and 2EWN (holoBirA).39

193−199, respectively, undergo disorder-to-order transitions upon adenylate binding and dimerization.17 An alanine substitution at residue G142 completely abolishes allosteric coupling.11 The structure of the holoBirAG142A variant indicates that the disorder-to-order transitions in the 140−146 and 193−199 loops are perturbed, consistent with a functional role for these transitions in activation of dimerization.18 Furthermore, 33 Å from the alanine substitution, the disorderto-order transition in the effector binding site ABL is also disrupted.18 The energetic penalty to bio-5′-AMP binding to the variant is 2.9 kcal/mol. The combined results indicate that folding events at the two distant functional surfaces are required to obtain the full −4 kcal/mol coupling between corepressor binding and dimerization. The origins of functional effects of alanine substitutions at the majority of the dimerization loop positions are not known. Energetic effects on dimerization could potentially reflect perturbation of allosteric activation of the monomer and/or contacts at the dimer interface. Furthermore, dimerization loop residues other than G142 may function in modulating the affinity of BirA for bio-5′-AMP. Five surface loops, including those comprised of residues 116−128 (BBL), 140−146, 170−176, 193−199, and 280−283, function in holoBirA dimerization.11,14 Alanine substitutions in the four latter loops yield a broad range of effects on BirA dimerization energetics (Table 1). Of the substitutions that result in penalties to the dimerization free energy, only two, D197A and T195A, are at residues that directly participate in intermolecular interactions in the dimer interface. Thermodynamic analysis of bio-5′-AMP binding by the dimerization loop variants will reveal their roles in both effector binding and allosteric activation.

■

MATERIALS AND METHODS Chemicals and Biochemicals. All chemicals and biochemicals were at least reagent grade. D-Biotin (Sigma-Aldrich) was prepared in standard buffer [10 mM Tris (pH 7.5) at 20 °C, 2.5 mM MgCl2, and 200 mM KCl] at concentrations of 500), at which quantitative binding of ligand to protein is achieved for multiple injections. The method, which has also been termed model-free ITC, can yield highly accurate values of the binding enthalpies.19,28 For TAPS titrations performed in this work, five 7 μL injections of 20 μM bio-5′-AMP into 2 μM protein were followed by a single 180 μL injection that brings the ligand to a concentration sufficient to fully saturate the protein. Finally, six postsaturation 10 μL injections were made. The integrated heat from each of the first five injections, when normalized to the number of moles injected, yields the molar heat of ligand binding and dilution: ΔH °TOT = ΔH °binding + ΔH °dil

⎡ nM tΔHV0 ⎢ Xt 1 ⎢1 + nM + nK M 2 t A t ⎣ ⎤ 2 ⎛ Xt 4X t ⎥ 1 ⎞ + ⎜1 + ⎟ − nM t nKAM t ⎠ nM t ⎥ ⎝ ⎦

Kbio‐5′‐AMP 1 + Kbiotin[biotin]

(4)

The heats of the final six injections yield the molar heat of dilution, ΔH°dil. The difference between these terms provides the bio-5′-AMP binding enthalpy. By subtracting the average experimental molar heat of dilution, we took five measurements of the molar enthalpy at each temperature. The TAPS method yields highly reproducible binding enthalpies provided that care is taken in the preparation of protein and ligand solutions and loading of these solutions into the reaction cell and injection syringe, respectively. In our hands, the method has yielded reproducible molar enthalpies of bio-5′-AMP binding to wtBirA

(1)

to obtain the association constant, KA, binding stoichiometry, n, and molar enthalpy change, ΔH°, of binding. Mt and Xt represent C

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

Figure 2. Biotin binding parameters are similar for all variants. (A) ITC data for I280 BirA obtained at 20 °C: (top) raw titration data and (bottom) data and best-fit curve from analysis using a single-site binding model. (B) Differences in the Gibbs free energies of biotin binding to the variants from that measured for wtBirA (ΔΔG°bio‑5′‑AMP = ΔG°bio‑5′‑AMP,variant − ΔG°bio‑5′‑AMP,wt). (C) Enthalpic (dark gray) and entropic, −TΔS°, contributions to the Gibbs free energies of biotin binding to the variants.

Table 2. Biotin Binding Parameters of BirA Variantsa protein wt E140A G142A P143A T195A G196A D197A I280A G281A K283A

KD (M)b −8

(4.7 ± 0.8) × 10 (4.0 ± 0.4) × 10−8 (4.0 ± 0.4) × 10−8 (4.5 ± 0.3) × 10−8 (4.5 ± 0.5) × 10−8 (5.1 ± 0.5) × 10−8 (4.2 ± 0.4) × 10−8 (4.6 ± 0.5) × 10−8 (4.1 ± 0.4) × 10−8 (3.9 ± 0.2) × 10−8

ΔG° (kcal/mol)c

ΔH° (kcal/mol)

−TΔS° (kcal/mol)d

n

−9.8 ± 0.1 −9.92 ± 0.05 −9.92 ± 0.06 −9.85 ± 0.04 −9.85 ± 0.06 −9.78 ± 0.06 −9.89 ± 0.05 −9.84 ± 0.06 −9.91 ± 0.05 −9.93 ± 0.03

−20.1 ± 0.4 −20.1 ± 0.2 −21.9 ± 0.2 −19.7 ± 0.1 −19.9 ± 0.2 −20.3 ± 0.2 −20.1 ± 0.2 −20.6 ± 0.2 −20.4 ± 0.2 −20.8 ± 0.3

10.3 ± 0.5 10.2 ± 0.2 12.0 ± 0.3 9.9 ± 0.1 10.1 ± 0.2 10.5 ± 0.2 10.2 ± 0.5 10.8 ± 0.2 10.5 ± 0.2 10.8 ± 0.3

0.93 ± 0.01 0.931 ± 0.006 1.085 ± 0.009 0.955 ± 0.004 0.894 ± 0.007 0.984 ± 0.007 1.260 ± 0.008 1.027 ± 0.007 0.878 ± 0.006 0.904 ± 0.009

All measurements were performed in standard buffer at 20 °C using the direct ITC method. bThe reported uncertainties were propagated from the individual uncertainties associated with at least two independent measurements of each parameter. cGibbs free energies of binding were calculated using the expression ΔG° = −RT ln KA. dThe entropic contribution to binding was calculated using the equation ΔG° = ΔH° − TΔS°.

a

Selection of Variants for Binding Measurements. Alanine substitutions in four dimerization surface loops yield a broad range of effects on the Gibbs free energy of holoBirA dimerization (Table 111). Only a subset of these variants were selected for ligand binding measurements. Because of the proximity of the 170−176 loop to the enzyme active site, the conservation of its sequence in all biotin protein ligases, and its demonstrated role in catalysis, variants with alanine substitutions in this sequence were excluded from this study.10,11 In the three remaining loops, only variants with dimerization free energies similar to or less favorable than that measured for wild-type (wt) holoBirA were selected. First, like BirAG142A, these variants are more likely to be defective in allosteric communication to the ligand binding surface. Second, weak dimerization minimizes the contribution of coupled protein association to the heats measured by ITC. On the basis of these criteria, nine variants with alanine substitutions in three of the dimerization surface loops were studied (Table 1). Variants and wtBirA Bind to Biotin with Similar Energetics. As a control for the presence of nonallosteric

in measurements performed with multiple different protein and ligand preparations.26,13 Heat Capacities of Bio-5′-AMP Binding. Heat capacities at constant pressure (ΔCp°) of bio-5′-AMP binding by the BirA variants were obtained from linear regression of the temperature dependence of the molar binding enthalpies obtained using the TAPS method. The change in heat capacity upon binding, ΔCp°, is equal to the slope extracted from the regression.

■

RESULTS Allosteric activation of the BirA monomer for dimerization occurs via communication of disorder-to-order transitions at the corepressor binding and dimerization surfaces. Alanine substitution of G142, a dimerization surface residue, disrupts the transitions on both surfaces with concomitant decreases in the affinity of BirA for the effector, bio-5′-AMP, and loss of coupling between effector binding and dimerization.18 A variety of ITC measurements were used to investigate the roles of other dimerization surface residues in bio-5′-AMP binding and allosteric regulation. D

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry effects on ligand binding, including loss of binding site integrity, biotin binding to all variants was first measured at 20 °C. Because this ligand does not elicit an allosteric response, perturbations to biotin binding should not reflect effects on intersurface communication. Biotin binding to all variants was measured using direct titrations in which the ligand is titrated into a solution of apoBirA. Analysis of the biotin binding titration for I280A BirA, which was conducted at 20 °C, indicates excellent agreement with a single-site binding model (Figure 2). Moreover, the values of the equilibrium dissociation constant and molar enthalpy of binding obtained from the analysis are similar to those measured for wtBirA, as are the calculated Gibbs free energy, ΔG°biotin, and entropy, −TΔS°biotin, of binding (Table 2). Biotin titrations performed on all other variants yield parameters that are, with a few exceptions, similar to those obtained for wtBirA (Table 2). The equilibrium dissociation constants and Gibbs free energies of binding are identical for all proteins. With the exception of BirAG142A, the enthalpies and entropies of biotin binding are also identical. For the G142A variant, the molar binding enthalpy and entropic contribution to binding, −TΔSbiotin, are −1.8 ± 0.4 kcal/mol more favorable and 1.7 ± 0.6 kcal/mol less favorable, respectively, than they are for wtBirA. The Gibbs Free Energy of Bio-5′-AMP Binding Is Altered for Only the G142A Variant. Alanine-substituted variants on the dimerization surface that are, like G142A, defective in dimerization were analyzed for communication, indicated by loss of bio-5′-AMP binding affinity, to the ligand binding site. The high affinity, KD,bio‑5′‑AMP = (5 ± 2) × 10−11 M, of bio-5′-AMP for wtBirA precludes the use of direct binding ITC at 20 °C. Therefore, effector binding measurements were performed using the displacement or competitive titration method27 in which biotin-saturated BirA is titrated with bio-5′AMP to displace the biotin. Nonlinear least-squares analysis of the resulting isotherm using a single-site model yields apparent binding parameters. The bio-5′-AMP binding parameters are calculated using the biotin binding parameters obtained from direct binding titrations described above. The agreement of the equilibrium binding constant obtained from the displacement method with that calculated from measured kinetic parameters governing binding of bio-5′-AMP to wtBirA indicates that the method yields reliable equilibrium constants.26 However, for the variants studied in this work, the binding enthalpies obtained using the competitive method were consistently more negative than those obtained from either direct titrations or the TAPS measurements described below. Therefore, only the equilibrium constants and Gibbs free energies for the displacement titration measurements are reported. Data obtained for the I280A BirA displacement titration performed at 20 °C are shown in Figure 3. In contrast to biotin binding titrations, the peaks associated with the initial injections are endothermic because the positive heat of biotin displacement (approximately 20 kcal/mol) is larger than the negative heat of bio-5′-AMP binding (approximately −14 kcal/mol). Consequently, the apparent molar binding enthalpy obtained from the titration is ∼6 kcal/mol. The small heat signal associated with this modest net binding enthalpy might explain the inability to obtain reliable binding enthalpies for this particular system using the displacement titration method. Because of the resulting increased contribution from dimerization, use of higher protein concentrations to enhance the signal to noise ratio was not a viable option for several of the proteins. Nonetheless, the

Figure 3. Binding of bio-5′-AMP to BirA variants. (A) Displacement titration obtained for binding of bio-5′-AMP to BirA I280A: (top) raw displacement data and (bottom) titration curve with best-fit curve to a single-site model. (B) Differences between the Gibbs free energies of bio-5′-AMP binding to each variant from that measured for wtBirA (ΔΔG°bio‑5′‑AMP = ΔG°bio‑5′‑AMP,variant − ΔG°bio‑5′‑AMP,wt).

isotherms obtained from data analysis are described well by a single-site model (Figure 3). Moreover, the equilibrium constant and free energy of bio-5′-AMP binding derived from the analysis indicate that BirAI280A is similar to the wild-type protein (Table 3 and Figure 3). With the exception of BirAG142A, displacement titrations were used to measure binding of bio-5′-AMP to all variants. These measurements yield binding free energies identical, within error, to that obtained for wtBirA. The titration of BirAG142A with the corepressor, which was performed using the direct method, indicates considerably weaker binding with an equilibrium dissociation constant of (8 ± 1) × 10−9 M, approximately 200-fold greater than that measured for wtBirA. Enthalpic and Entropic Contributions to Bio-5′-AMP Binding. The bio-5′-AMP binding enthalpies for the variants were measured using the total association at partial saturation or the TAPS titration method, which decouples measurement of binding enthalpy from determination of the equilibrium association constant and stoichiometry.19,28 For any binding system, this decoupling allows more accurate determinations of molar enthalpies.28 The specific advantage of the method for the BirA−bio-5′-AMP system is that, as a direct titration, it is characterized by a heat signal magnitude significantly higher than that of the displacement titration (see the previous section). TAPS titration measurement of BirAG196A at 20 °C is shown in Figure 4. Analogous to the direct biotin titration, each of the initial five peaks yields the combined heat of ligand binding and E

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

dilution. The large central peak results from injection of an excess of bio-5′-AMP to ensure complete protein saturation with ligand. Each of the final six peaks yields the heat of ligand dilution. Subtraction of the average of the integrated heat for these final peaks from that for each of the first five peaks, followed by normalization to the number of moles of ligand injected, yields the molar binding enthalpy. A single titration yields five measurements of the bio-5′-AMP binding enthalpy. The bio-5′-AMP binding enthalpies for all variants, which range from −12.0 kcal/mol for wtBirA to −13.3 kcal/mol for BirAG142A (Table 4 and Figure 4B), are all similar in magnitude.

Table 3. Thermodynamics of Binding of Bio-5′-AMP to BirA Variants protein wt E140A G142A P143A T195A G196A D197A I280A G281A K283A

KD (M)a −11

(5 ± 2) × 10 (7 ± 3) × 10−11 (8 ± 1) × 10−9 (8 ± 2) × 10−11 (1.0 ± 0.4) × 10−10 (2 ± 3) × 10−11 (9 ± 3) × 10−11 (9 ± 2) × 10−11 (8 ± 3) × 10−11 (9 ± 2) × 10−11

ΔG° (kcal/mol)b,c

n

−13.8 ± 0.2 −13.6 ± 0.2 −10.85 ± 0.09 −13.5 ± 0.2 −13.4 ± 0.2 −14.2 ± 0.8 −13.5 ± 0.2 −13.5 ± 0.1 −13.6 ± 0.2 −13.5 ± 0.3

0.93 ± 0.01 0.90 ± 0.01 1.010 ± 0.01 0.96 ± 0.01 0.88 ± 0.01 1.09 ± 0.03 1.21 ± 0.02 1.06 ± 0.01 0.89 ± 0.01 0.89 ± 0.01

Table 4. Enthalpies of Binding of Bio-5′-AMP to BirA Variantsa

a

With the exception of those of the G142A variant, reported values were obtained using the displacement ITC method in standard buffer at 20 °C. The G142A values, which were obtained from direct titrations, were previously reported.17 bGibbs free energies were calculated using the expression ΔG° = −RT ln KA. cThe reported uncertainties were propagated from the individual uncertainties associated with at least two independent measurements of each parameter.

protein

ΔH°bio‑5′‑AMP (kcal/mol)

ΔH°bio‑5′‑AMP,corr (kcal/mol)b

wt E140A G142A P143A T195A G196A D197A I280A G281A K283A

−12.0 ± 0.3 −12.43 ± 0.06 −13.3 ± 0.1 −12.20 ± 0.01 −12.58 ± 0.07 −11.9 ± 0.1 −12.4 ± 0.2 −12.6 ± 0.1 −12.7 ± 0.2 −13.3 ± 0.2

−15 ± 1 −13.7 ± 0.5 −13.2 ± 0.1 −12.9 ± 0.2 −12.9 ± 0.2 −16 ± 1 −12.2 ± 0.2 −12.6 ± 0.1 −13.1 ± 0.3 −15.3 ± 0.7

a Enthalpies, ΔH°bio‑5′‑AMP, were measured using the TAPS method at 20 °C with reported errors being the standard deviation obtained from the average of five values. bThe corrected enthalpies were calculated as described in Discussion.

These enthalpies, combined with the Gibbs free energies obtained from the displacement titrations, were used to calculate the entropic contributions, −TΔS°, to the binding free energies. With the exception of the G142A variant, bio-5′-AMP binding is characterized by a modest, favorable entropy. At 20 °C, the G142A variant binds to bio-5′-AMP with an unfavorable entropy of approximately 3.0 kcal/mol (Figure 4B). Variants Show a Range of Heat Capacity Changes for Effector Binding. The structural basis of BirA allosteric activation was further probed by determining the heat capacity changes associated with binding of bio-5′-AMP to the variants. Results of these measurements can reveal information about changes in solvent accessibility that might be coupled to binding.20,28,29 Binding enthalpies were measured over the temperature range from 10 and 25 °C,27 and linear regression of the resulting enthalpy versus temperature data yielded the heat capacity changes at constant pressure. The dimerization surface variants exhibit a range of heat capacity changes for bio-5′-AMP binding. For the G196A variant, linear regression indicates a heat capacity change of −0.240 ± 0.006 kcal mol−1 K−1 [r2 = 0.95 (Figure 5A)]. The r2 values associated with linear regression of the molar enthalpy versus temperature data for the remaining variants indicate constant heat capacity changes over the temperature range employed. The ΔCp,bio‑5′‑AMP value for the G196A variant is more positive than that measured for wtBirA (Figure 5B). Four of the variants are characterized by ΔCp,bio‑5′‑AMP values that are more negative than that measured for the wild-type protein. Heat capacity change values for all variants in the 280−283 loop are, within error, identical. Additionally, no significant differences in ΔCp,bio‑5′‑AMP are observed for variants with E140A, P143A, T195A, and D197A substitutions. The tightest dimerizing variant, BirA-

Figure 4. Enthalpic and entropic driving forces for binding of bio-5′AMP to the BirA variants at 20 °C. (A) Titration of the G196A variant under total association at partial saturation (TAPS) conditions: (top) raw titration data and (bottom) integrated heats for each injection normalized to the moles of bio-5′-AMP injected. (B) Enthalpic (dark gray) and entropic (−TΔS°, light gray) contributions to binding of bio5′-AMP to the variants at 20 °C. The entropic contribution was calculated with the equation ΔG° = ΔH° − TΔS° using the Gibbs free energies obtained from the displacement titrations and enthalpies measured using the TAPS method.

F

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

The range of molar heat capacity changes associated with binding of bio-5′-AMP to the variants suggests that corepressor binding is coupled to different structural transitions in each. To obtain clues about the nature of these structural changes and their relationship to the allosteric activation process, the correlation between the perturbations to molar heat capacity change and the Gibbs free energy of dimerization for the variants was examined (Figure 6A). No correlation was observed.

Figure 5. Heat capacity changes associated with binding of bio-5′-AMP to the BirA variants. (A) Linear regression of the temperature dependence of the bio-5′-AMP binding enthalpies for the G196A variant. (B) Heat capacity changes for binding of bio-5′-AMP to the variants (wtBirA, light gray bar).

G196A, exhibits the most positive value for the heat capacity change. However, bio-5′-AMP binds to BirAG142A, the variant characterized by the weakest dimerization, with a ΔCp,bio‑5′‑AMP value identical to that measured for wtBirA.

■

DISCUSSION Contributions of dimerization surface residues to bio-5′-AMP binding and coupled allosteric activation of the BirA monomer were investigated. Among the dimerization surface variants, only BirAG142A is perturbed in its free energy of bio-5′-AMP binding. Thus, the glycine at position 142 plays a critical role in communication between the two coupled sites in the protein. Although alanine substitutions at other residues on the dimerization surface have no impact on the Gibbs free energy of adenylate binding at 20 °C, several of the substitutions significantly alter the heat capacity change associated with the binding process. The glycine at position 142 is the only dimerization surface residue investigated in this work that influences the structure and function of the corepressor binding site. Previous studies indicate that any perturbation of ABL folding, either through direct alanine substitution of loop residues or through the G142A substitution, is accompanied by a 3.0−4.0 kcal/mol penalty to bio-5′-AMP binding.13,15,18 Consequently, the absence of effects of alanine substitutions on the bio-5′-AMP binding free energy at 20 °C for the remaining dimerization surface variants provides strong evidence of the preservation of the ABL disorder-to-order transition in these proteins. Structural data indicate that the G142A substitution is also accompanied by disruption of a coilto-helix transition in residues 143−146 that occurs in wtBirA (Figure 1). Moreover, the correlation between the absence of this transition in the G142A variant and disruption of effector-linked ABL folding suggests that the two processes are coupled.18 Although structures are not available for the remaining dimerization surface variants, the thermodynamic data are consistent with preservation of the helical transition in these proteins.

Figure 6. Relationship of the Gibbs free energy of holoBirA dimerization to the heat capacity change associated with binding of bio-5′-AMP to the BirA monomer. (A) Correlation plot obtained using the uncorrected heat capacity changes. (B) Heat capacity changes, uncorrected (dark gray) and corrected (light gray) for dimerization, of binding of bio-5′AMP to the variants. (C) Relationship of the dimerization free energy to the corrected heat capacity change for binding of bio-5′-AMP to the variants. The two outlier points are for BirAI280A and BirAG142A.

However, this analysis neglects the possible contribution of the dimerization enthalpy to the apparent enthalpies of binding of adenylate to the variants. ITC measurements were performed under conditions in which, for all variants, the majority of the holoprotein was monomeric (Table 1). Nevertheless, given the 41 ± 3 kcal/mol enthalpy of wild-type holoBirA dimerization,30 even a modest extent of dimerization can make a significant contribution to the heat signal measured via ITC. Corrections for contributions of dimerization to the enthalpy and heat capacity changes for binding of bio-5′-AMP to wtBirA are significant. On the basis of Hess’s law, the total measured binding enthalpy is the sum of the contributions from binding and dimerization.31 Thus, at any single temperature, the total binding enthalpy can be corrected for the dimerization contribution using the following equation: ΔH °bio‐5′‐AMP,app = ΔH °bio‐5′‐AMP + 0.5χdim ΔH °dim G

(5)

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

values were then applied to calculate a corrected ΔH°bio‑5′‑AMP value at each temperature as described above (Figure 7B). As expected, the positive dimerization enthalpy for wild-type holoBirA results in a larger correction to the effector binding enthalpy with an increase in temperature. Linear regression of the corrected enthalpies versus temperature yields a corrected molar heat capacity change for corepressor binding that differs from the uncorrected value by −0.45 ± 0.07 kcal mol−1 K−1 (Figure 7B). The bio-5′-AMP binding enthalpies and heat capacity changes for the variants were also corrected for dimerization. Although the dimerization enthalpies have not been measured for the variants, measurements of the solvent isotope effect on dimerization indicate that the process is also governed by a large positive enthalpy for the variants. First, measurements of wtBirA dimerization performed in H2O and D2O over a range of temperatures indicate that solvent release is the major thermodynamic driving force for dimer formation.32 The large unfavorable dimerization enthalpy reflects the penalty associated with the removal of solvent from the protein surface, and the correspondingly large favorable entropy, −TΔS°, arises from the release of this solvent to the bulk. Second, consistent with solvent release also serving as the major driving force for BirA variant dimerization, the solvent isotope effect measured at 20 °C is identical for wtBirA and variants regardless of the location of the alanine substitution.32 Thus, in correcting for the contribution of protein association to the measured enthalpies of adenylate binding, we assumed the dimerization enthalpy was the same and constant over the relevant temperature range for wt and variant proteins. The broad range of dimerization free energies associated with the variants results in a correspondingly broad range of magnitudes for the corrections to the binding enthalpy at 20 °C. For weakly dimerizing variants with alanine substitutions at G142, P143, T195, D197, I280, and G281, the apparent and corrected values are similar in magnitude (Figure 7A and Table 4). By contrast, corrections to the enthalpies for the tightly dimerizing E140A, K283A, and G196A variants range from −1.3 ± 0.5 to −4 ± 1 kcal/mol (Figure 7A and Table 4). Extension of the molar enthalpy corrections to the entire temperature range permitted calculation of corrected heat capacity change values for binding of bio-5′-AMP to the variants. The results of these calculations yielded corrections ranging from 0 for G142A to −0.44 ± 0.06 kcal mol−1 K−1 for G196A, the most tightly dimerizing variant (Figure 6B and Table 5). Moreover, plots of the Gibbs free energy of dimerization versus the corrected heat

where ΔH°bio‑5′‑AMP,app is the apparent molar heat of binding obtained from the ITC TAPS measurements, ΔH°bio‑5′‑AMP is the actual molar heat of binding of bio-5′-AMP to the monomer, χdim is the fraction of the total protein, when saturated with bio-5′AMP, that is dimer, and ΔH°dim is the molar enthalpy of holoBirA dimerization. The apparent binding heat measured for each ligand injection in the TAPS titration was corrected by accounting for the incremental change in the fraction dimer that accompanies the injection. For wtBirA at 20 °C, the corrected bio-5′-AMP binding enthalpy differs from the apparent value by −3 ± 1 kcal/mol (Figure 7A). The known temperature

Figure 7. Enthalpies and heat capacity changes for bio-5′-AMP binding corrected for dimerization. (A) Apparent (black bars) and corrected (gray bars) molar enthalpies of binding of bio-5′-AMP to the BirA variants. The corrected enthalpies were calculated using eq 5. (B) Linear regression of the bio-5′-AMP binding enthalpy [uncorrected (▼) and corrected (■)] vs temperature for wtBirA.

independence of the dimerization enthalpy over the range utilized for the heat capacity change measurements30 allowed extension of the corrections to the binding enthalpies over this entire range. First, the van’t Hoff equation and the known dimerization enthalpy were used to calculate the equilibrium dimerization constant and the resulting fraction of protein in the dimer form, χdim, at each of the working temperatures. These

Table 5. Heat Capacity Changes for Binding of Bio-5′-AMP to BirA Variants protein

ΔCpa (kcal mol−1 K−1)

ΔΔCpb (kcal mol−1 K−1)

ΔCp,corra (kcal mol−1 K−1)

ΔΔCp,corrb (kcal mol−1 K−1)

wt E140A G142A P143A T195A G196A D197A I280A G281A K283A

−0.30 ± 0.01 −0.37 ± 0.03 −0.300 ± 0.008 −0.363 ± 0.008 −0.37 ± 0.01 −0.240 ± 0.006 −0.364 ± 0.009 −0.31 ± 0.01 −0.31 ± 0.01 −0.29 ± 0.02

− −0.07 ± 0.01 0.00 ± 0.03 −0.06 ± 0.01 −0.07 ± 0.01 0.06 ± 0.01 −0.06 ± 0.01 −0.01 ± 0.01 −0.01 ± 0.01 0.01 ± 0.02

−0.75 ± 0.06 −0.58 ± 0.05 −0.31 ± 0.01 −0.49 ± 0.02 −0.43 ± 0.02 −0.68 ± 0.06 −0.37 ± 0.01 −0.31 ± 0.01 −0.38 ± 0.01 −0.59 ± 0.04

− 0.17 ± 0.08 0.44 ± 0.06 0.26 ± 0.06 0.32 ± 0.06 0.07 ± 0.08 0.38 ± 0.06 0.44 ± 0.06 0.37 ± 0.06 0.16 ± 0.07

a

The heat capacities were obtained from linear regression of the enthalpies obtained from TAPS measurements or those corrected for the contribution from dimerization vs temperature using Prism 4.03 (GraphPad). Errors represent the 65% confidence intervals obtained from the regression. bΔΔCp = ΔCp,variant − ΔCp,wt with errors propagated from the errors associated with the individual values using the standard method. H

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry capacity change values for binding of adenylate to the monomer show near perfect correlation for the majority of the variants (Figure 6C). The heat capacity changes for binding of adenylate to the variants correlate with the predicted disruption of intramolecular interactions found in the holoBirA structure. Comparison of the apo- and holoBirA monomer structures reveals several intramolecular interactions on the dimerization surface of the liganded protein that are absent in the unliganded protein. These interactions cannot form in apoBirA because one or more of the participating amino acid residues are in a protein segment that is disordered in the apo species. For example, the 193−197 segment, which contributes to two of the interactions, is disordered in apoBirA.10,17 Likewise, R119, which interacts with the backbone carbonyl of residue G281 in holoBirA, is part of the BBL for which no electron density is observed for apoBirA.10 Finally, the alanine substitutions and G142 and I280 on the dimerization surface, which yield the largest perturbations to both the heat capacity change and the dimerization free energy, are at residues that participate in these intramolecular interactions. It should be emphasized that although the structural data allow prediction of the loss of specific intramolecular interactions in the variants, this loss may be coupled to disruption of additional interactions. The G142A variant exemplifies this phenomenon.18 The major contributions to negative heat capacity changes accompanying ligand binding by proteins include loss of exposure of a hydrophobic surface to solvent and disorder-toorder transitions.29,33−35 The large difference in the ΔCp values for binding of bio-5′-AMP to wtBirA and the G142A variant can be readily explained in terms of these two contributions. The corrected heat capacity change for binding of wtBirA to bio-5′AMP is −0.75 ± 0.06 kcal mol−1 K−1. Consistent with this large negative value, the apo/unliganded and holo/effector-bound wtBirA structures indicate coupled folding of multiple protein segments comprising a total of 29 amino acid residues, several of which have nonpolar side chains.10,17 Binding also results in removal of the bio-5′-AMP ligand, which contains significant nonpolar surface, from solvent. By contrast, the heat capacity change for binding of the effector to the G142A variant of −0.31 ± 0.01 kcal mol−1 K−1, although still negative, is considerably more modest than that measured for wtBirA. Consistent with this observation, the holoBirAG142A structure indicates that 20 of the 29 residues that fold upon binding of bio-5′-AMP to wtBirA remain disordered in the variant.18 The well-characterized wtBirA and BirAG142A can serve as a reference in proposing the structural origins of heat capacity changes associated with binding of bio-5′-AMP to the remaining variants. First, heat capacity changes for binding of bio-5′-AMP to variants E140A, G196A, and K283A are similar in magnitude to that measured for wtBirA (Table 5). These proteins also dimerize with Gibbs free energies similar to that of wtBirA (Table 1). On the basis of heat capacity changes, binding of the effector to these variants is likely accompanied by folding transitions similar to those observed for wtBirA. By contrast, the identical values of the heat capacity changes associated with binding of the effector to the I280A and G142A variants suggest that the coupled folding is significantly compromised for the former variant. Despite its location far from the dimerization interface, the I280A substitution also perturbs the dimerization free energy by 3.5 kcal/mol.11 The wild-type holoBirA structure indicates that the I280A substitution disrupts a hydrophobic interaction with the W309 side chain (Figure 8). The relationship between

Figure 8. Intramolecular interactions on the holoBirA dimerization surface between I280 and W309, K194 and D197, and R119 and G281 backbone carbonyl. The effector ligand is colored yellow. This figure was created using Pymol with input file 2EWN.39

this local disruption and the more extensive loss of folding upon binding predicted from the thermodynamics has yet to be determined. For variants characterized by more modest perturbations to the heat capacity change that range from 0.26 to 0.37 kcal mol−1 K−1, the locally disrupted interactions involve polar or charged side chains. For example, the alanine substitutions at D197 and G281 disrupt an electrostatic interaction with the K194 side chain and a hydrogen bond with the R119 side chain, respectively (Figure 8). The extent to which the accompanying effects on the heat capacity changes of bio-5′-AMP binding reflect changes in burial of hydrophobic surface area and/or disorder upon effector binding remains to be determined. The results reported in this work provide insight into structural changes on the dimerization surface and their functional effects on allosteric activation of the BirA monomer for homodimerization. The G142 residue plays a critical role in the helical transition of residues 142−146 that accompanies both corepressor binding and the disorder-to-order transition on the corepressor binding surface. This helical transition, which is coupled to the disorder-to-order transition in the 193−199 loop, allows formation of the network of intramolecular interactions on the dimerization surface that enhances the Gibbs free energy of BirA. The negative heat capacity change for effector binding reflects the contributions of both folding events that result in burial of the hydrophobic surface and disorder-to-order transitions in the protein. In the context of this model, communication between corepressor-linked folding on the ligand binding surface and formation of the helical extension on the dimerization surface constitutes the core of the allosteric activation process. The challenge to gaining a molecular understanding of allostery in this system is determining how these two folding events are communicated through the folded protein core. The results presented in this work reinforce the utility of measurements of heat capacity changes for gaining insight into allosteric mechanisms. The results also highlight the advantages of protein disorder for allosteric control. The structural origins of heat capacity changes associated with biomolecular processes, in general, and allostery, in particular, have been the subject of extensive discussion in the literature.33,36−38 The measurements presented in this work reveal that residues in a functional allosteric site serve distinct roles in allosteric activation. For example, heat capacity measurements reveal that the large effects that alanine substitutions at G142 and I280 have on allosteric activation of dimerization reflect their key roles in folding and I

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

dimerization: a case for divergent followed by convergent evolution. J. Mol. Biol. 357, 509−523. (18) Eginton, C., Cressman, W. J., Bachas, S., Wade, H., and Beckett, D. (2015) Allosteric Coupling via Distant Disorder-to-Order Transitions. J. Mol. Biol. 427, 1695−1704. (19) Bains, G., and Freire, E. (1991) Calorimetric determination of cooperative interactions in high affinity binding processes. Anal. Biochem. 192, 203−206. (20) Makhatadze, G. I., and Privalov, P. L. (1995) Energetics of protein structure. Adv. Protein Chem. 47, 307−425. (21) Strecher, P. G. E. (1960) The Merck Index of Chemicals and Drugs, 7th Ed., 150−151. (22) Abbott, J., and Beckett, D. (1993) Cooperative binding of the Escherichia coli repressor of biotin biosynthesis to the biotin operator sequence. Biochemistry 32, 9649−9656. (23) Gill, S. C., and von Hippel, P. H. (1989) Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182, 319−326. (24) VP-ITC MicroCalorimeter User’s Manual, MicroCal, LLC. (25) Biswas, T., and Tsodikov, O. V. (2010) An easy-to-use tool for planning and modeling a calorimetric titration. Anal. Biochem. 406, 91− 93. (26) Brown, P. H., and Beckett, D. (2005) Use of Binding Enthalpy To Drive an Allosteric Transition. Biochemistry 44, 3112−3121. (27) Sigurskjold, B. W. (2000) Exact analysis of competition ligand binding by displacement isothermal titration calorimetry. Anal. Biochem. 277, 260−266. (28) Haq, I., Jenkins, T. C., Chowdhry, B. Z., Ren, J., and Chaires, J. B. (2000) Parsing free energies of drug-DNA interactions. Methods Enzymol. 323, 373−405. (29) Prabhu, N. V., and Sharp, K. A. (2005) Heat capacity in proteins. Annu. Rev. Phys. Chem. 56, 521−548. (30) Zhao, H., Streaker, E., Pan, W., and Beckett, D. (2007) Proteinprotein interactions dominate the assembly thermodynamics of a transcription repression complex. Biochemistry 46, 13667−13676. (31) Hess, G. H. (1840) Recherches thermochimiques. Bulletin scientifique, Académie impériale des sciences 8, 257−272. (32) Eginton, C., and Beckett, D. (2013) A large solvent isotope effect on protein association thermodynamics. Biochemistry 52, 6595−6600. (33) Sturtevant, J. M. (1977) Heat capacity and entropy changes in processes involving proteins. Proc. Natl. Acad. Sci. U. S. A. 74, 2236− 2240. (34) Makhatadze, G. I., and Privalov, P. L. (1990) Heat capacity of proteins. I. Partial molar heat capacity of individual amino acid residues in aqueous solution: hydration effect. J. Mol. Biol. 213, 375−384. (35) Cooper, A., Johnson, C. M., Lakey, J. H., and Nollmann, M. (2001) Heat does not come in different colours: entropy-enthalpy compensation, free energy windows, quantum confinement, pressure perturbation calorimetry, solvation and the multiple causes of heat capacity effects in biomolecular interactions. Biophys. Chem. 93, 215− 230. (36) Vega, S., Abian, O., and Velazquez-Campoy, A. (2015) On the link between conformational changes, ligand binding and heat capacity. Biochim. Biophys. Acta, Gen. Subj., DOI: 10.1016/j.bbagen.2015.10.010. (37) Spolar, R. S., and Record, M. T., Jr. (1994) Coupling of local folding to site-specific binding of proteins to DNA. Science 263, 777− 784. (38) Cooper, A. (2010) Protein Heat Capacity: An Anomaly that Maybe Never Was. J. Phys. Chem. Lett. 1, 3298−3304. (39) DeLano, W. L. (2002) The PyMOL Molecular Graphics System, version 1.2r3pre, Schrödinger, LLC, Portland, OR.

ordering of the dimerization surface in the activation process. Second, the results support the idea that disordered segments that are not contiguous in a linear protein sequence can collectively contribute to allosteric activation through their ability to form a network of noncovalent interactions in response to effector binding. In the effector-free protein, this same disorder serves an important regulatory function as an entropic barrier to dimerization in the absence of the appropriate signal.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Motlagh, H. N., Wrabl, J. O., Li, J., and Hilser, V. J. (2014) The ensemble nature of allostery. Nature 508, 331−339. (2) Malmstrom, R. D., Kornev, A. P., Taylor, S. S., and Amaro, R. E. (2015) Allostery through the computational microscope: cAMP activation of a canonical signalling domain. Nat. Commun. 6, 7588. (3) Barker, D. F., and Campbell, A. M. (1981) The birA gene of Escherichia coli encodes a biotin holoenzyme synthetase. J. Mol. Biol. 146, 451−467. (4) Barker, D. F., and Campbell, A. M. (1981) Genetic and biochemical characterization of the birA gene and its product: evidence for a direct role of biotin holoenzyme synthetase in repression of the biotin operon in Escherichia coli. J. Mol. Biol. 146, 469−492. (5) Cronan, J. E., Jr. (1989) The E.coli bio operon:transcriptional repression by an essential protein modification enzyme. Cell 58, 427− 429. (6) Lane, M. D., Rominger, K. L., Young, D. L., and Lynen, F. (1964) The enzymatic synthesis of holotranscarboxylase from apotranscarboxylase and (+)-biotin. J. Biol. Chem. 239, 2865−2871. (7) Streaker, E. D., and Beckett, D. (2003) Coupling of protein assembly and DNA binding:biotin repressor dimerization preceeds biotin operator binding. J. Mol. Biol. 325, 937−948. (8) Eisenstein, E., and Beckett, D. (1999) Dimerization of the Escherichia coli biotin repressor:corepressor function in protein assembly. Biochemistry 38, 13077−13084. (9) Streaker, E. D., Gupta, A., and Beckett, D. (2002) The biotin repressor: thermodynamic coupling of corepressor binding, protein assembly, and sequence-specific DNA binding. Biochemistry 41, 14263− 14271. (10) Wilson, K. P., Shewchuk, L. M., Brennan, R. G., Otsuka, A. J., and Matthews, B. W. (1992) Escherichia coli biotin holoenzyme ynthetase/ bio repressor crystal structure delineates the biotin- and DNA- binding domains. Proc. Natl. Acad. Sci. U. S. A. 89, 9257−9261. (11) Adikaram, P. R., and Beckett, D. (2012) Functional versatility of a single protein surface in two protein:protein interactions. J. Mol. Biol. 419, 223−233. (12) Kwon, K., and Beckett, D. (2000) Function of a conserved sequence motif in biotin holoenzyme synthetases. Protein Sci. 9, 1530− 1539. (13) Naganathan, S., and Beckett, D. (2007) Nucleation of an allosteric response via ligand-induced loop folding. J. Mol. Biol. 373, 96−111. (14) Kwon, K., Streaker, E. D., Ruparelia, S., and Beckett, D. (2000) Multiple disordered loops function in corepressor-induced dimerization of the biotin repressor. J. Mol. Biol. 304, 821−833. (15) Eginton, C., Naganathan, S., and Beckett, D. (2015) Sequencefunction relationships in folding upon binding. Protein Sci. 24, 200−211. (16) Weaver, L. H., Kwon, K., Beckett, D., and Matthews, B. W. (2001) Corepressor-induced organization and assembly of the biotin repressor: a model for allosteric activation of a transcriptional regulator. Proc. Natl. Acad. Sci. U. S. A. 98, 6045−6050. (17) Wood, Z., Weaver, L. H., Brown, P. H., Beckett, D., and Matthews, B. W. (2006) Co-repressor induced order and biotin repressor J

DOI: 10.1021/acs.biochem.5b00949 Biochemistry XXXX, XXX, XXX−XXX