BioMed Research International

BioMed Research International / 2015 / Article
Special Issue

Sequence and Structure Analysis of Biological Molecules Based on Computational Methods

View this Special Issue

Research Article | Open Access

Volume 2015 |Article ID 716945 | 12 pages | https://doi.org/10.1155/2015/716945

Redesigning Protein Cavities as a Strategy for Increasing Affinity in Protein-Protein Interaction: Interferon-γ Receptor 1 as a Model

Academic Editor: Yuedong Yang
Received02 Oct 2014
Revised22 Dec 2014
Accepted28 Dec 2014
Published28 Apr 2015

Abstract

Combining computational and experimental tools, we present a new strategy for designing high affinity variants of a binding protein. The affinity is increased by mutating residues not at the interface, but at positions lining internal cavities of one of the interacting molecules. Filling the cavities lowers flexibility of the binding protein, possibly reducing entropic penalty of binding. The approach was tested using the interferon-γ receptor 1 (IFNγR1) complex with IFNγ as a model. Mutations were selected from 52 amino acid positions lining the IFNγR1 internal cavities by using a protocol based on FoldX prediction of free energy changes. The final four mutations filling the IFNγR1 cavities and potentially improving the affinity to IFNγ were expressed, purified, and refolded, and their affinity towards IFNγ was measured by SPR. While individual cavity mutations yielded receptor constructs exhibiting only slight increase of affinity compared to WT, combinations of these mutations with previously characterized variant N96W led to a significant sevenfold increase. The affinity increase in the high affinity receptor variant N96W+V35L is linked to the restriction of its molecular fluctuations in the unbound state. The results demonstrate that mutating cavity residues is a viable strategy for designing protein variants with increased affinity.

1. Introduction

In studying specificity and affinity of protein-protein interactions, the main focus is traditionally on the structural properties of the interface, for example, complementarity of the residue composition, hydrogen-bonding networks, and the role of hydration [1]. However, there is also a significant contribution of the conformational dynamics to the binding affinity. Analysis of molecular dynamics simulations of 17 protein-protein complexes and their unbound components with quasi-harmonic analysis [2] concluded that the protein flexibility has an important influence on the thermodynamics of binding. Moreover, changes in the protein conformational dynamics may lead to substantial changes in affinity to binding partners without an apparent structural change of the complex. For example, reorganization of the hydrogen bonding networks and solvent bridges of the interacting molecules upon mutation, which was accompanied only by subtle structural changes, leads to radically different binding free energy [3, 4]. A recent work [5] shows that the apparent change in the amino acid dynamics determined by NMR spectroscopy is linearly related to the change in the overall binding entropy and also that changes in side-chain dynamics determined from NMR data can be used as a quantitative estimate of changes in conformational entropy [6, 7]. Also, an analysis of crystallographic B-factors has revealed a significant decrease of flexibility of residues exposed to solvent compared to flexibility of residues interacting with another biomolecule and further compared to their flexibility in the protein core [8]. This “freezing” of atoms upon complexation and in the protein core is only slightly larger for the side chain atoms than for the main chain atoms. Entropic cost specific for side-chain freezing has been computationally evaluated as a small, but important contribution to the thermodynamics of binding [9, 10]. These results indicate that changes in amino acid conformational entropy upon binding contribute significantly to the free energy of protein-protein association.

However important the interaction interface is for the affinity, the interaction is influenced by the whole composition of the cognate molecules, so that modulation of affinity can be achieved by changing other residues than residues at the interface. One such possible alternative approach would be filling cavities in one of the binding partners, thus influencing the stability and dynamics of the interacting proteins [1114]. Thermodynamic consequences of introducing cavity-filling mutations have been discussed for residues at the interaction interface [1517] showing that filling the interfacial cavity increases affinity due to both gain in binding enthalpy and a loss in binding entropy, the latter being attributed to a loss of conformational degrees of freedom. It has been shown that interactions between the internal “core” residues is responsible for the folding and thermal stability of a protein [18]. Here, we decided to test whether the protein-protein affinity could be increased by mutations not on the interface, but in cavities inside one of the cognate protein molecules.

This study follows our previous article [21] in which we designed mutations increasing the affinity of human interferon-γ receptor 1 (IFNγR1) towards its natural cognate molecule interferon-γ (IFNγ), an important protein of innate immunity [22, 23]. Here, we retain this model system and the main contours of the protocol but replace the search for interface mutations by searching for mutations in the receptor cavities in order to further increase its interaction affinity to IFNγ and our computer analysis revealed four such cavity mutants. Combining one of these cavity mutations with the best variant designed in our previous study led to a sevenfold increase in affinity compared to the wild-type receptor. We show that the affinity increase in this mutant is related to the restricted flexibility of amino acids in the unbound state of IFNγR1.

2. Materials and Methods

2.1. Outline of the Protocol

Our computational predictions are based on the analysis of crystal structures of complexes between IFNγ and the extracellular part of IFNγR1, namely, the structures of PDB codes 1fg9 [19] and 1fyh [20] that contain four crystallographically independent IFNγ/IFNγR1 complexes. Throughout the paper, IFNγR1 residues are numbered as in UniProt entry P15260. We used the empirical force field implemented in the software FoldX [24] to search for mutations within the positions lining the internal cavities of IFNγR1 molecule that would increase its stability and/or its affinity to IFNγ. All designed mutants of IFNγR1 were subsequently expressed and purified and their affinity to a “single-chain” form of IFNγ (IFNγSC, [25]) was measured. Individual steps of the computational protocol as well as experimental procedures are described below.

2.2. In Silico Design of Variants

The program 3V [26] was used to identify internal cavities in all four available structures of IFNγR1 molecules complexed with IFNγ. In total, 52 cavity-lining residues, which were identified as encapsulating the cavities in at least one of the four structures, were extracted using the VMD program [27]. Each of 52 amino acid residues identified as lining the internal receptor cavities was mutated in all four crystal IFNγ/IFNγR1 complexes to 20 amino acid residues using the “positionscan” and “analyzecomplex” FoldX keywords. This represented mutations (including self-mutations leading to ). Three types of changes of free energy () were calculated using the program FoldX:(1) of folding of IFNγR1 in complex” gauged the influence of mutations on the stability of the whole IFNγ/IFNγR1 complex;(2) of folding of free IFNγR1” estimated the effect of mutations on the stability of the isolated receptor;(3) of binding” of complex between IFNγR1 and IFNγ estimated the change of the interaction between the receptor molecule and the rest of the complex.

2.3. Modeling

IFNγR1 models are based on PDB structures 1fg9 [19] and 1fyh [20]. Missing residues in both structures were added using Modeller suite of programs [28]. The lowest energy loop models were used for further calculations.

2.4. Molecular Dynamics (MD) Simulations

MD simulations were run using GROMACS suite of programs to test the stability and dynamic properties, including analysis of values of root means square fluctuations (RMSF) [29] and the effect of variable geometry on prediction of changes of interaction free energy (ΔΔGs), of the IFNγ/IFNγR1 complexes (PDB codes 1fyh and 1fg9). More detailed protocol of MD and FoldX calculations follows.

2.5. Protocol of Molecular Dynamics (MD) Calculations

For the MD simulations the following setup was used: protonation state was determined by pdb2gmx program using parameters provided by the OpenMM [30] Zephyr [31] program. Implicit solvation (GBSA, ε = 78.3, with collision interval of 10.99 fs) was used in combination with parm96 force field [32]. OpenMM Zephyr implementation of GPU accelerated version of GROMACS [29] suite of programs was used to simulate the systems. The initial crystal structures were optimized and the simulation was propagated at 300 K with the time step of 2 fs. RMSF (root-mean square fluctuations) of atoms in the analyzed proteins were calculated from the 100 ns trajectory to estimate flexibility of residues; they were calculated by g_rmsf program in 5 ns windows.

2.6. Construction, Expression, and Purification of Recombinant IFNγR1 Variants

We followed the protocols from our previous study [21] for all proteins produced in this study. All selected IFNγR1 variants were prepared, expressed, and successfully purified to homogeneity by the following protocol.

Codon-optimized synthetic gene (GenScript) encoding extracellular domain of human IFNgR1 (residues 18–245) was cloned into the pET-28b(+) vector (Novagen) using NcoI and XhoI restriction enzymes in frame with N-terminal start codon and C-terminal HisTag. The QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies) was used for mutating the IFNγR1 gene according to manufacturer’s manual using primers listed below. Primers were designed by web-based PrimerX program (http://www.bioinformatics.org/primerx/).

The recombinant IFNγR1 variants were expressed in Escherichia coli BL21 (λDE3) in LB medium containing 60 μg/mL of kanamycin at 37°C for 4 hours after induction by 1 mM IPTG. Harvested cells by centrifugation (8,000 g, 10 min, 4°C) were disrupted by ultrasound in 50 mM Tris buffer pH 8 and centrifuged at 40,000 g, 30 min, 4°C, and inclusion bodies were dissolved in 50 mM Tris buffer pH 8 containing 8 M urea and 300 mM NaCl to extract protein that was further affinity-purified on Ni-NTA agarose (Qiagen) in the same buffer. Protein was eluted from resin by 250 mM Imidazole pH 8 in previous buffer and refolded by dialysis against 100 mM Tris-HCl pH 8, 150 mM NaCl, 2.5 mM EDTA, 0.5 mM Cystamine, and 2.5 mM Cysteamine overnight at 4°C. Final purification of monomeric receptor variants was performed at 4°C on a HiLoad 16/600 Superdex 200 pg (GE Healthcare) equilibrated by PBS buffer pH 7.4 (Figure 1). Monodispersity of the purified receptor protein was verified by dynamic light scattering (DLS) using Malvern Zetasizer Nano ZS90 instrument (data not shown).

2.7. Primers

Mutagenesis primers are designed for the introduction of single residue substitution into IFNγR1 WT. Mutated nucleotides are underlined. We have the following:V35LForward: 5′-GTCCCGACCCCGACCAACTTGACGATTGAAAGTTACAAC-3′Reverse: 5′-GTTGTAACTTTCAATCGTCAAGTTGGTCGGGGTCGGGAC-3′A114EForward: 5′-GAAAGAATCAGCGTATGAAAAATCGGAAGAATTCGCC-3′Reverse: 5′-GGCGAATTCTTCCGATTTTTCATACGCTGATTCTTTC-3′D124N Forward: 5′-CGCCGTGTGCCGTAATGGCAAAATCG-3′Reverse: 5′-CGATTTTGCCATTACGGCACACGGCG-3′H222YForward: 5′-CTGAAGGCGTTCTGTATGTCTGGGGTGTC-3′Reverse: 5′-GACACCCCAGACATACAGAACGCCTTCAG-3′

2.8. Construction, Expression, and Purification of IFNγSC

Recombinant interferon gamma in so-called single chain form (IFNγSC) described by [25] was cloned into pET-26b(+) vector (Novagen) using NdeI and XhoI restriction enzymes in frame with N-terminal start codon not to have no peptide leader nor tag.

The recombinant IFNγSC was expressed in E. coli BL21 (λDE3) in LB medium containing 60 μg/mL of kanamycin at 30°C for 4 hours after induction by 1 mM IPTG. Harvested cells by centrifugation (8,000 g, 10 min, 4°C) were disrupted by ultrasound in 20 mM Na-Phosphate buffer pH 7.3 and centrifuged at 40,000 g, 30 min, 4°C, and soluble fraction was further purified on SP Sepharose HP (GE Healthcare) using linear gradient of NaCl and further purified to homogeneity by gel filtration in same procedure as IFNγR1 receptor (see above).

2.9. Biophysical Characterization of the Studied Proteins

Melting temperatures of the receptor variants were measured using fluorescence-based thermal shift assay and for selected mutants by CD melting experiments. Interactions between IFNγR1 variants and IFNγSC were measured by the technique of surface plasmon resonance (SPR) as discussed in our previous study [21]. Experimental procedures are detailed below.

2.10. CD Measurements

CD spectra were recorded using “Chirascan-plus” (Applied Photophysics) spectrometer in steps of 1 nm over the wavelength range of 190–260 nm. Samples at a concentration of 0.2 mg/mL were placed into 0.05 cm path-length quartz cell to the thermostated holder and individual spectra were recorded at the temperature of 25°C. The CD signal was expressed as the difference between the molar absorption of the right- and left-handed circularly polarized light and the resulting spectra were buffer subtracted. To analyze the ratio of the secondary structures we used the CDNN program provided with Chirascan CD spectrometer [33]. For CD melting measurements, samples at a concentration of 1.5 mg/mL were placed into 10 mm path-length quartz cell to the thermostated holder and CD signal at 280 nm was recorded at 1°C increment at rate of 1.0°C/min over the temperature range of 25 to 65°C with an averaging time of 10 seconds. CD melting curves were normalized to relative values between 1.0 and 0.0.

2.11. Thermostability of the IFNγR1 Variants by Thermal-Based Shift Assay

Melting temperature () curves of the WT and selected variants were obtained from fluorescence-based thermal shift assay (TSA) using fluoroprobe. Experiment was performed in “CFX96 Touch Real-Time PCR Detection System” (Bio-Rad) using FRET Scan Mode. The concentration of fluorescent SYPRO Orange dye (Sigma Aldrich) was 8-fold dilution from 5000-fold stock and protein concentration was 2 μL in final volume of 25 μL. As a reference we used only buffer (PBS buffer pH 7.4) without protein. Thermal denaturation of proteins was performed in capped “Low Tube Strips, CLR” (Bio-Rad) and possible air bubbles in samples were removed by centrifugation immediately before the assay. The samples were heated from 20°C to 75°C with stepwise increment of 0.5°C per minute and a 30 s hold step for every point, followed by the fluorescence reading. Data subtraction by reference sample was normalized and used for first derivative calculation to estimate the melting temperature.

2.12. SPR Measurements

His-tagged receptor molecules were diluted to concentration of 10 μg/mL in PBST running buffer (PBS pH 7.4, 0.005% Tween 20) and immobilized on a HTG sensor chip activated with Ni2+ cations at a flow rate 30 μL/min for 60 s to gain similar surface protein density. Purified IFNγSC was diluted in running buffer to concentrations ranging from 0.1 to 9 nM and passed over the sensor chip for 90 seconds at a flow rate 100 μL/min (association phase). Dissociation was measured in the running buffer for 10 min at the same flow rate. Correction for nonspecific binding of IFNγSC to the chip surface was done by subtraction of the response measured on uncoated interspots and reference channel coated with His-tagged Fe-regulated protein D (FrpD) from Neisseria meningitides [34]. Data were processed in the ProteOn Manager software (version 3.1.0.6) and the doubly referenced data were fitted to the 1 : 1 “Langmuir with drift” binding model.

3. Results and Discussion

3.1. Internal Cavities Identified in IFNγR1

The cavity analysis revealed generally different number and size of cavities for each IFNγR1 crystal structure; their characteristics are listed in Table 1; their location in a representative receptor molecule (PDB entry 1fg9, chain C [19]) is highlighted in Figures 2(a) and 2(b). All amino acid residues lining cavities in all four IFNγR1 proteins complexed with IFNγ were combined, resulting in 52 residues used in subsequent in silico analysis.


Surface [Å2]*Number of residues lining the cavityResidues selected for mutationCavity observed in IFNγR1 chain of
1fg91fyh

11347V35, A114 C D
21335B E
347014D124C D
42629H222C DB E
51206C DE
61657C DE
71777DB E
81385CB

Surface calculated with a probe radius of 0.25 Å for cavities combined from all relevant receptor chains.
Some residues are shared by neighboring cavities.
3.2. In Silico Design of Variants

All 52 amino acids lining the cavities of the receptor molecule were subject to the mutation analysis by FoldX. The resulting values indicated potential for mutation leading to increasing the receptor affinity to IFNγ. The mutations were ordered by their values and the first 50 best mutations from each crystal structure (200 mutations in total) were further analyzed. Of these 200 mutations, twelve positions were predicted in all four or at least three crystal structures. The twelve promising positions are highlighted in orange and yellow in Figure 2(c). Following the previous study [21], where we observed significant differences between predicted directly from the crystal structures and from structures after molecular dynamics (MD) relaxation, we performed short (10 ns) MD simulations of the four crystal structures of complexes between wild type IFNγR1 and IFNγ, and repeated the FoldX mutation analysis on 500 snapshots extracted from these MD trajectories. After averaging of the predicted values for the twelve selected positions, we made the final selection of the four candidate mutations. The averaged values resulting from these calculations for structure 1fg9, receptor chain C, are summarized in Figure 3. The final selection of the four variants is listed in Table 2 together with the changes of their binding free energies averaged over 500 MD snapshots from each of the four IFNγ/IFNγR1 complexes in crystal structures 1fg9 and 1fyh.


VariantΔΔ of folding of IFNγR1 in complex*ΔΔ of folding of free IFNγR1ΔΔ of binding of IFNγR1/IFNγ complexSequence conservation

V35L80%
A114E60%
D124N40%
H222Y40%

ΔΔ of folding of IFNγR1 bound to IFNγ measures the influence of mutations on the stability of the whole complex.
ΔΔ of folding of IFNγR1 alone represents changes of the stability of the isolated receptor.
ΔΔ of binding of the whole complex between IFNγR1 and IFNγ estimates the change of the affinity between the receptor molecule and the rest of the complex.
Sequence conservation of amino acid residues at positions 35, 114, 124, and 222. It was based on the global alignment of 32 sequences of the extracellular part of IFNγR1 (Figure 2(c)).

Finally, the four consensus candidate mutations, which resulted as the best replacements of the WT sequence, were expressed, and characterized by SPR, CD, and thermal-based shift assay. The relative affinities of these four cavity-filling single mutants are shown in Figure 4(a) together with relative affinities of the double mutants combining the four cavity-filling mutations with mutation N96W.

As Table 2 and in detail Figure 3 show, the calculations revealed only modest potential gains in interaction affinity, probably because of small cavity volumes as well as the fact that they are often lined by evolutionary highly conserved residues. As opposed to the interface mutations, where the predicted s of IFNγR1 stability and binding to IFNγ served as a sufficient criterion for the selection of affinity increasing mutations, there was no clear-cut rule for selecting internal cavity mutations that would result in improved interaction energy. We thus decided to test experimental consequences of combination of three types of values calculated from the MD snapshots. To identify potentially favorable mutations, we combined values of folding ( types (1) and (2) in the in silico protocol described in Materials and Methods) and of binding (type (3)). The first two mutations, V35L and H222Y, were predicted to increase of folding to a similar extent for both the complexed and free IFNγR1 ( (1) and (2)), while calculated values of their of binding were virtually zero. The other two selected mutations, A114E and D124N, were predicted to slightly improve of binding while both types of their of folding were destabilizing. In the latter case, of folding of free IFNγR1 (type 2) was more unfavorable than of folding of complexed IFNγR1 (type 1). This means that the complex is predicted to be relatively more stable compared to the free IFNγR1.

3.3. Experimental Determination of the Affinities between IFNγR1 Variants and IFNγSC

Computer-designed IFNγR1 variants were expressed and purified and their affinities to IFNγSC were determined by SPR measurements; relative affinities are plotted in Figure 4(a); SPR sensograms are depicted in Figure 4(b). The calculated values showed that the four selected “cavity” single amino acid mutation variants bind to the IFNγSC with similar affinity as WT; a modest increase was observed for the V35L variant. In line with our previous work, we decided to test to what extent the effect of two distant point mutations is additive. To this end, we combined the four cavity mutants designed here with the variant with the highest affinity designed previously, N96W. The results were quite encouraging: while affinity of one double mutant () is neutral and one () affinity actually decreased, two double mutants, N96W with A114E and V35L, had affinity increased compared to WT. The affinity increase of one of the double mutants, , is significant, seven times higher than affinity of WT.

The thermal stability (Figure 5) and secondary structure (Figure 6) of four IFNγR1 variants, V35L, N96W, , and WT, were studied by CD and their melting temperatures were confirmed by thermal-based shift assay (Figure 7); the CD-measured melting temperatures are 53, 48, 50, and 54°C, respectively. Both variants with the highest affinity, N96W and , have melting temperatures lower than WT, so that mutation from asparagine to tryptophan at the position 96 apparently causes a decrease of IFNγR1 thermal stability. However, the CD spectra of all four proteins are highly similar (Figure 6); their analysis provided virtually identical composition of the secondary structure elements dominated by the beta-sheet fractions indicating that no global structural rearrangements were caused by the mutations and the fold of these four variants is most likely the same. Moreover, the spectra are in agreement with the spectrum measured previously [35] for WT of IFNγR1.

3.4. Analysis of Internal Dynamics of the IFNγR1 Variants

To test how a cavity-filling mutation changes the flexibility of the receptor molecule in unbound and complexed states we analyzed root-mean square fluctuations (RMSF) of the selected variants. Comparison of RMSF sorted by their values, “ranked RMSF,” for WT, N96W, and , are plotted in Figure 8 (solid lines for IFNγ/IFNγR1 complexes, dashed lines for IFNγR1 alone). These plots revealed significant differences between dynamics of the variants as is detailed below.(1)The interface residues of N96W and WT are more flexible in the free receptor than in the complex, while the flexibility of the interface residues of is similar for the free and complexed receptor (Figures 8(a) and 8(d)). This indicates entropically more favorable binding of the variant compared to the other two variants.(2)Interestingly, the origin of this behavior is different in the N-terminal and C-terminal domains of the IFNγR1 molecule: in the N-terminal domain (Figure 8(d)), the flexibility of the interface residues of all variants is similar in the bound state, while being different in unbound state; they are most flexible in N96W and the least in . In the C-terminal domain (Figure 8(e)), the flexibility of the three variants is similar in their free states, but it differs in the bound state between N96W, which has the lowest flexibility, and WT with the highest flexibility.(3)The V35L mutation stiffens the receptor nonlocally and makes especially the C-terminal interface residues more flexible in the bound state compared to the N96W mutant (Figure 8(e)).(4)To sum up, the V35L mutation brought flexibility of the free and complexed receptor closer together, indicating reduced entropy penalty of binding and resulting in the higher affinity of the double mutant compared to N96W mutant.Filling the cavity by hydrophobic groups as in the V35L mutation is stabilizing but not as much as would be implied by of the removal of the corresponding hydrophobic group to water. A compensatory effect lowering a potential increase of the protein and/or complex stability has been observed previously [13] and a comparable decrease of stabilization was also predicted here by FoldX. Filling of a cavity may stabilize the interaction by several mechanisms, for example, by reducing the entropic penalty of complexation by stiffening interacting molecules in the free state, or indirectly by destabilization of the intermediate molten globule state rather than by stabilization of the folded protein [36]. These compensatory effects further illustrate complexity of protein-protein interactions (and/or folding) and the known limits of computational approaches to increasing protein-protein affinity [37].

An important issue potentially affecting reliability of FoldX predictions is the flexibility of the receptor molecule. The first round of FoldX calculations based on the static crystal structures suggested one additional mutation, G225Y, as potentially increasing receptor affinity to IFNγ. Although further calculations using structures of snapshots from the MD simulations did not confirm this prediction, we expressed and characterized this mutation. The experimental data were in agreement with the MD-based prediction showing much lower binding affinity compared to the WT (the ratio of the respective values was 0.4), and also the double mutant had a fairly low binding affinity (compared to WT, the ratio of the respective values was 3.1, which is lower than for the N96W mutant). This observation can be explained by the structural properties of the receptor molecule. The loop region of IFNγR1 containing the G225 residue is flexible and any residue at the position 225 is thus only a fraction of time in the geometry, in which it may increase the binding affinity. An important role of flexibility at the C-terminal part of the interacting IFNγ and IFNγR1 is well illustrated by a study of IFNγ modified at its C-terminal side [38].

3.5. Sequence Conservation of Mutable Residues

We checked sequence conservation for the 12 positions selected by the FoldX calculations for potential cavity-filling mutations. Global alignment of 32 sequences of the extracellular part of IFNγR1 from various organisms by Kalign as implemented in program Ugene [39] (Figure 2(c)) shows conservation between 40 and 98% for these positions; the position V35 is well conserved (80%). The independence of sequence conservation and its potential for stabilizing mutation filling-up protein cavity (“mutability”) contrasts with previously observed tight correlation between conservation and mutability for receptor residues interacting with IFNγ [21]: we tested several mutations of the interface residues S97 and E118, which were conserved at the 90% level (Figure 2(c)), namely, S97X (X = L, N, W) and E118X (X = M, F, Y, W), and they did not bind IFNγSC at all (unpublished SPR data) despite the fact that binding of these mutants to IFNγ was predicted to be stronger than that of WT.

3.6. Relationship Between FoldX Values and Naturally Occurring IFNγR1 Variants

Interesting, albeit indirect, validation of the present FoldX predictions of of mutations can be found among naturally occurring IFNγR1 single-point mutations collected in the database of single nucleotide polymorphism (dbSNP) [40]. The database contains 25 nucleotide mutations at 22 unique positions of the extracellular part of the IFNγ receptor, which is studied here; these 22 positions are marked blue in Figure 2(c). Most of the predictions for these natural mutants show neutral effect on the stability of free IFNγR1 and on its complex with IFNγ. This is in agreement with the fact that only two of the natural mutants exhibit deleterious effects or are represented by a pathological phenotype.

4. Conclusions

We present a new computational strategy for designing higher affinity variants of a binding protein and show that it is possible to increase the affinity of a protein-protein interaction by mutations not at the interface, but in the interior cavities of a binding partner. The mutations were selected at positions lining internal cavities of one binding partner, and an in silico protocol identified mutations that would fill the protein cavities and increase the stability of the complex. We showed that the selection of such cavity mutations in interferon-γ receptor 1 (IFNγR1) could be performed based on a combination of simple empirical force-field calculations and MD simulations. The mechanism by which the cavity mutations cause affinity increase is shown to be restriction of molecular fluctuations, which can be related to reduced entropy penalty upon binding [6, 7]. IFNγR1 WT and all computationally designed receptor mutants were expressed, purified, and refolded, and the affinity towards the cognate protein, IFNγSC, was measured by SPR. While single mutants showed roughly the same affinity as WT, double mutants combining cavity mutations with the best interface mutation obtained previously [21] were successful in further increasing the binding affinity.

The results demonstrate that mutating cavity residues is a viable strategy for designing protein variants with increased binding affinity. The comparison of computational data and experiments helped to further improve our understanding of forces governing protein-protein interactions. The newly obtained high-affinity binders of IFNγ could be developed into a new diagnostic tool. The significance of the present work can be seen in the fact that small gains of cavity mutants led to significant increase of affinity when combined with more conventional mutations influencing the interface.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgments

Support from Grant P305/10/2184 from the Czech Science Foundation is greatly acknowledged. This study was supported by BIOCEV CZ.1.05/1.1.00/02.0109 from the ERDF, Biotechnological expert CZ.1.07/2.3.00/30.0020, and by institutional Grant RVO 86 652 036.

References

  1. P. L. Kastritis and A. M. J. J. Bonvin, “Molecular origins of binding affinity: seeking the Archimedean point,” Current Opinion in Structural Biology, vol. 23, no. 6, pp. 868–877, 2013. View at: Publisher Site | Google Scholar
  2. R. Grünberg, M. Nilges, and J. Leckner, “Flexibility and conformational entropy in protein-protein binding,” Structure, vol. 14, no. 4, pp. 683–693, 2006. View at: Publisher Site | Google Scholar
  3. T. N. Bhat, G. A. Bentley, G. Boulot et al., “Bound water molecules and conformational stabilization help mediate an antigen-antibody association,” Proceedings of the National Academy of Sciences of the United States of America, vol. 91, no. 3, pp. 1089–1093, 1994. View at: Publisher Site | Google Scholar
  4. Y. Urakubo, T. Ikura, and N. Ito, “Crystal structural analysis of protein-protein interactions drastically destabilized by a single mutation,” Protein Science, vol. 17, no. 6, pp. 1055–1065, 2008. View at: Publisher Site | Google Scholar
  5. K. K. Frederick, M. S. Marlow, K. G. Valentine, and A. J. Wand, “Conformational entropy in molecular recognition by proteins,” Nature, vol. 448, no. 7151, pp. 325–329, 2007. View at: Publisher Site | Google Scholar
  6. M. S. Marlow, J. Dogan, K. K. Frederick, K. G. Valentine, and A. J. Wand, “The role of conformational entropy in molecular recognition by calmodulin,” Nature Chemical Biology, vol. 6, no. 5, pp. 352–358, 2010. View at: Publisher Site | Google Scholar
  7. A. J. Wand, “The dark energy of proteins comes to light: conformational entropy and its role in protein function revealed by NMR relaxation,” Current Opinion in Structural Biology, vol. 23, no. 1, pp. 75–81, 2013. View at: Publisher Site | Google Scholar
  8. B. Schneider, J. C. Gelly, A. G. de Brevern, and J. Cerny, “Local dynamics of proteins and DNA evaluated from crystallographic B factors,” Acta Crystallographica D: Biological Crystallography, vol. 70, part 9, pp. 2413–2419, 2014. View at: Publisher Site | Google Scholar
  9. C. Wang, O. Schueler-Furman, and D. Baker, “Improved side-chain modeling for protein-protein docking,” Protein Science, vol. 14, no. 5, pp. 1328–1339, 2005. View at: Publisher Site | Google Scholar
  10. C. Cole and J. Warwicker, “Side-chain conformational entropy at protein-protein interfaces,” Protein Science, vol. 11, no. 12, pp. 2860–2870, 2002. View at: Publisher Site | Google Scholar
  11. M. Bueno, N. Cremades, J. L. Neira, and J. Sancho, “Filling small, empty protein cavities: structural and energetic consequences,” Journal of Molecular Biology, vol. 358, no. 3, pp. 701–712, 2006. View at: Publisher Site | Google Scholar
  12. T. Ohmura, T. Ueda, K. Ootsuka, M. Saito, and T. Imoto, “Stabilization of hen egg white lysozyme by a cavity-filling mutation,” Protein Science, vol. 10, no. 2, pp. 313–320, 2001. View at: Publisher Site | Google Scholar
  13. M. Tanaka, H. Chon, C. Angkawidjaja, Y. Koga, K. Takano, and S. Kanaya, “Protein core adaptability: crystal structures of the cavity-filling variants of Escherichia coli rnase HI,” Protein and Peptide Letters, vol. 17, no. 9, pp. 1163–1169, 2010. View at: Publisher Site | Google Scholar
  14. T. Koudelakova, R. Chaloupkova, J. Brezovsky et al., “Engineering enzyme stability and resistance to an organic cosolvent by modification of residues in the access tunnel,” Angewandte Chemie—International Edition, vol. 52, no. 7, pp. 1959–1963, 2013. View at: Publisher Site | Google Scholar
  15. S. Atwell, M. Ultsch, A. M. de Vos, and J. A. Wells, “Structural plasticity in a remodeled protein-protein interface,” Science, vol. 278, no. 5340, pp. 1125–1128, 1997. View at: Publisher Site | Google Scholar
  16. Y. Kawasaki, E. E. Chufan, V. Lafont et al., “How much binding affinity can be gained by filling a cavity?” Chemical Biology and Drug Design, vol. 75, no. 2, pp. 143–151, 2010. View at: Publisher Site | Google Scholar
  17. L. Morellato-Castillo, P. Acharya, O. Combes et al., “Interfacial cavity filling to optimize CD4-mimetic miniprotein interactions with HIV-1 surface glycoprotein,” Journal of Medicinal Chemistry, vol. 56, no. 12, pp. 5033–5047, 2013. View at: Publisher Site | Google Scholar
  18. J. Černý, J. Vondrášek, and P. Hobza, “Loss of dispersion energy changes the stability and folding/unfolding equilibrium of the trp-cage protein,” The Journal of Physical Chemistry B, vol. 113, no. 16, pp. 5657–5660, 2009. View at: Publisher Site | Google Scholar
  19. D. J. Thiel, M.-H. Le Du, R. L. Walter et al., “Observation of an unexpected third receptor-molecule in the crystal structure of human interferon-γ receptor complex,” Structure, vol. 8, no. 9, pp. 927–936, 2000. View at: Publisher Site | Google Scholar
  20. M. Randal and A. A. Kossiakoff, “Crystallization and preliminary X-ray analysis of a 1 : 1 complex between a designed monomeric interferon-gamma and its soluble receptor,” Protein Science, vol. 7, no. 4, pp. 1057–1060, 1998. View at: Google Scholar
  21. P. Mikulecký, J. Černý, L. Biedermannová et al., “Increasing affinity of interferon-γ receptor 1 to interferon-γ by computer-aided design,” BioMed Research International, vol. 2013, Article ID 752514, 12 pages, 2013. View at: Publisher Site | Google Scholar
  22. K. Schroder, P. J. Hertzog, T. Ravasi, and D. A. Hume, “Interferon-gamma: an overview of signals, mechanisms and functions,” Journal of Leukocyte Biology, vol. 75, no. 2, pp. 163–189, 2004. View at: Publisher Site | Google Scholar
  23. E. C. Borden, G. C. Sen, G. Uze et al., “Interferons at age 50: past, current and future impact on biomedicine,” Nature Reviews Drug Discovery, vol. 6, no. 12, pp. 975–990, 2007. View at: Publisher Site | Google Scholar
  24. J. Schymkowitz, J. Borg, F. Stricher, R. Nys, F. Rousseau, and L. Serrano, “The FoldX web server: an online force field,” Nucleic Acids Research, vol. 33, no. 2, pp. W382–W388, 2005. View at: Publisher Site | Google Scholar
  25. A. Landar, B. Curry, M. H. Parker et al., “Design, characterization, and structure of a biologically active single-chain mutant of human IFN-γ,” Journal of Molecular Biology, vol. 299, no. 1, pp. 169–179, 2000. View at: Publisher Site | Google Scholar
  26. N. R. Voss and M. Gerstein, “3V: cavity, channel and cleft volume calculator and extractor,” Nucleic Acids Research, vol. 38, no. 2, pp. W555–W562, 2010. View at: Publisher Site | Google Scholar
  27. W. Humphrey, A. Dalke, and K. Schulten, “VMD: visual molecular dynamics,” Journal of Molecular Graphics, vol. 14, no. 1, pp. 33–38, 1996. View at: Publisher Site | Google Scholar
  28. B. Webb and A. Sali, “Protein structure modeling with MODELLER,” Methods in Molecular Biology, vol. 1137, pp. 1–15, 2014. View at: Google Scholar
  29. B. Hess, C. Kutzner, D. van der Spoel, and E. Lindahl, “GROMACS 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation,” Journal of Chemical Theory and Computation, vol. 4, no. 3, pp. 435–447, 2008. View at: Publisher Site | Google Scholar
  30. P. Eastman and V. S. Pande, “OpenMM: a hardware-independent framework for molecular simulations,” Computing in Science & Engineering, vol. 12, no. 4, pp. 34–39, 2010. View at: Google Scholar
  31. M. S. Friedrichs, P. Eastman, V. Vaidyanathan et al., “Accelerating molecular dynamic simulation on graphics processing units,” Journal of Computational Chemistry, vol. 30, no. 6, pp. 864–872, 2009. View at: Publisher Site | Google Scholar
  32. P. A. Kollman, “Advances and continuing challenges in achieving realistic and predictive simulations of the properties of organic and biological molecules,” Accounts of Chemical Research, vol. 29, no. 10, pp. 461–469, 1996. View at: Publisher Site | Google Scholar
  33. G. Bohm, R. Muhr, and R. Jaenicke, “Quantitative analysis of protein far UV circular dichroism spectra by neural networks,” Protein Engineering, vol. 5, no. 3, pp. 191–195, 1992. View at: Google Scholar
  34. E. Sviridova, L. Bumba, P. Rezacova et al., “Crystallization and preliminary crystallographic characterization of the iron-regulated outer membrane lipoprotein FrpD from Neisseria meningitidis,” Acta Crystallographica Section F: Structural Biology and Crystallization Communications, vol. 66, part 9, pp. 1119–1123, 2010. View at: Publisher Site | Google Scholar
  35. M. Fountoulakis and R. Gentz, “Effect of glycosylation on properties of soluble interferon gamma receptors produced in prokaryotic and eukaryotic expression systems,” Nature Biotechnology, vol. 10, no. 10, pp. 1143–1147, 1992. View at: Publisher Site | Google Scholar
  36. T. Sengupta, Y. Tsutsui, and P. L. Wintrode, “Local and global effects of a cavity filling mutation in a metastable serpin,” Biochemistry, vol. 48, no. 34, pp. 8233–8240, 2009. View at: Publisher Site | Google Scholar
  37. T. S. Chen and A. E. Keating, “Designing specific protein-protein interactions using computation, experimental library screening, or integrated methods,” Protein Science, vol. 21, no. 7, pp. 949–963, 2012. View at: Publisher Site | Google Scholar
  38. E. Saesen, S. Sarrazin, C. Laguri et al., “Insights into the mechanism by which interferon-γ basic amino acid clusters mediate protein binding to heparan sulfate,” Journal of the American Chemical Society, vol. 135, no. 25, pp. 9384–9390, 2013. View at: Publisher Site | Google Scholar
  39. K. Okonechnikov, O. Golosova, M. Fursov, and UGENE Team, “Unipro ugene: a unified bioinformatics toolkit,” Bioinformatics, vol. 28, no. 8, pp. 1166–1167, 2012. View at: Google Scholar
  40. S. T. Sherry, M.-H. Ward, M. Kholodov et al., “DbSNP: the NCBI database of genetic variation,” Nucleic Acids Research, vol. 29, no. 1, pp. 308–311, 2001. View at: Publisher Site | Google Scholar

Copyright © 2015 Jiří Černý et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1304 Views | 530 Downloads | 2 Citations
 PDF  Download Citation  Citation
 Download other formatsMore
 Order printed copiesOrder