Biochimica et Biophysica Acta 1844 (2014) 585–592

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Computational prediction and experimental characterization of a “size switch type repacking” during the evolution of dengue envelope protein domain III (ED3)☆ Montasir Elahi a, Monirul M. Islam a, Keiichi Noguchi b, Masafumi Yohda a, Hiroyuki Toh c, Yutaka Kuroda a,⁎ a b c

Department of Biotechnology and Life Science, Graduate School of Engineering, Tokyo University of Agriculture & Technology, Nakamachi, Koganei-shi, Tokyo 184-8588, Japan Instrumentation Analysis Center, Tokyo University of Agriculture & Technology, Nakamachi, Koganei-shi, Tokyo 184-8588, Japan Computational Biology Research Center, AIST Tokyo Waterfront Bio-IT Research Building, 2-4-7 Aomi, Koto-ku, Tokyo, 135-0064, Japan

a r t i c l e

i n f o

Article history: Received 30 October 2013 Received in revised form 17 December 2013 Accepted 19 December 2013 Available online 27 December 2013 Keywords: Ancestral sequence reconstruction Mutational analysis Crystal structure Side-chain modeling Structural compensations Thermodynamic stability

a b s t r a c t Dengue viruses (DEN) are classified into four serotypes (DEN1-DEN4) exhibiting high sequence and structural similarities, and infections by multiple serotypes can lead to the deadly dengue hemorrhagic fever. Here, we aim at characterizing the thermodynamic stability of DEN envelope protein domain III (ED3) during its evolution, and we report a structural analysis of DEN4wt ED3 combined with a systematic mutational analysis of residues 310 and 387. Molecular modeling based on our DEN3 and DEN4 ED3 structures indicated that the side-chains of residues 310/387, which are Val310/Ile387 and Met310/Leu387 in DEN3wt and DEN4wt, respectively, could be structurally compensated, and that a “size switch type repacking” might have occurred at these sites during the evolution of DEN into its four serotypes. This was experimentally confirmed by a 10 °C and 5 °C decrease in the thermal stability of, respectively, DEN3 ED3 variants with Met310/Ile387 and Val310/Leu387, whereas the variant with Met310/Leu387, which contains a double mutation, had the same stability as the wild type DEN3. Namely, the Met310Val mutation should have preceded the Leu387Ile mutation in order to maintain the tight internal packing of ED3 and thus its thermodynamic stability. This view was confirmed by a phylogenetic reconstruction indicating that a common DEN ancestor would have Met310/Leu387, and the intermediate node protein, Val310/Leu387, which then mutated to the Val310/Ile387 pair found in the present DEN3. The hypothesis was further confirmed by the observation that all of the present DEN viruses exhibit only stabilizing amino acid pairs at the 310/387 sites. © 2013 Published by Elsevier B.V.

1. Introduction The dengue (DEN) virus is a major health problem in South and South-East Asia causing 50–100 million human infections every year

Abbreviations: ASR, Ancestral Sequence Reconstruction; BSA, Bovine serum albumin; Cα-RMSD, Root Mean Square Deviation between the main-chain Cα-atoms; CD, Circular Dichroism; DEN, Dengue virus; E-protein, Envelope glycoprotein; ED3, Envelope protein Domain III; ELISA, Enzyme-Linked Immunosorbent Assay; HPLC, High Performance Liquid Chromatography; PBS, Phosphate buffer saline; MAb, Monoclonal Antibody; RNA, Ribonucleic Acid; WT (wt), Wild type ☆ Notes: We use “size switch type repacking” rather than “size switch” because, in the case of DEN ED3, the first Val310Met mutation is a small to large substitution, but the second Ile387Leu mutation does not switch the size of the side-chain whereas the bulkiness and shape complementarily are fitted ensuring the stability of the proteins (reference [14]). Similarly, in this paper we use the word “compensated” in the sense of “structurally compensated” as defined by Matthews' group (reference [14]), without the evolutionary connotation implied by “correlated mutation” as defined in reference [35]. ⁎ Corresponding author at: Department of Biotechnology and Life Science Tokyo University of Agriculture & Technology, 2-24-16, Nakamachi, Koganei-shi, Tokyo 1848588, Japan. Tel./fax: +81 42 388 7794. E-mail address: [email protected] (Y. Kuroda). 1570-9639/$ – see front matter © 2013 Published by Elsevier B.V. http://dx.doi.org/10.1016/j.bbapap.2013.12.013

[1]. Symptoms of the DEN virus infection range from mild classical dengue fever to the potentially lethal dengue hemorrhagic fever, when sequential infection by multiple serotypes occurs [1]. The DEN virus is a member of the flavivirus family with four distinct serotypes (DEN1DEN4) [2]. The DEN viral RNA is encapsulated by capsid, membrane and envelope proteins [3]. Structurally, the envelope protein (E-protein) consists of three different domains termed domain I (ED1), domain II (ED2), and domain III (ED3) [4]. The E-protein mediates DEN virus infection, and ED3 is involved in the interaction with the host cell receptor heparan sulfate, with a loop in ED2 assisting the fusion of the virus with the host cell [4,5]. Mutational studies of antibody interactions also indicate that ED3 contains most of the DEN specific epitopes, which are located in residues 305 to 311 and 379 to 391 [6,7]. ED3 thus plays an important role both in spreading and inhibiting viral infection. Due to high sequence similarities (70%–89%) ED3 retains the same native fold among all four serotypes. However, sequence variations among serotypes induce minor local structural changes, whose effects on thermal stability are not fully understood [8]. Residues buried into a protein's hydrophobic core are essential both for determining its structure and its thermodynamic stability, as demonstrated by a protein sequence simplification experiments with a small model protein, BPTI

586

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

(bovine pancreatic trypsin inhibitor) [9]. In particular, the role of steric clashes and cavities in the hydrophobic core in modulating protein stability is well characterized [10–13]. For instance, Matthew's group created an artificially stable sequence of T4 Lysozyme by switching the sizes of two close residues (an Ala and a Leu) whose side-chains are “structurally compensated” [14]. In contrast to artificial mutations, naturally occurring “size switch” mutations should be a rare event, since two substitutions need to occur almost simultaneously for maintaining protein stability. It is thus of high interest to investigate the role of a “size switch type core repacking” during evolution, if it can happen at all. In this study, we aim at characterizing the thermodynamics of DEN ED3 during its evolution, and we report the crystal structure of DEN4wt ED3, and a mutational analysis of DEN3 and DEN4 ED3 variants. Our analysis focused on residues 310 and 387 of ED3 because side-chain modeling based on our crystal structures of DEN3 (PDB ID: 3VTT) [15] and DEN4 ED3 (PDB ID: 3WE1; this study) suggested that amino acid substitutions at these two sites might be critical for the maintenance of thermal stability. Although the calculations suggested no “size switch” mutation according to the strict original definition [14], we found a repacking that resembles a looser definition of the “size switch” mutation that occurred during the evolution of ED3. This phenomenon, which we termed a “size switch type repacking”, might have occurred between residues 310/387, as demonstrated by an experimental mutational analysis of ED3. Further, ancestral sequence reconstruction using the serotype's nucleotide information indicated that mutations at these two sites occurred in order to maintain the thermodynamic stability during the evolution of DEN. 2. Materials and methods 2.1. Construction of the plasmids and mutants Synthetic genes encoding the ED3 sequences were cloned into a pET15b vector (Novagen) at the endonuclease NdeI and BamHI sites. The sequences of DEN3wt and DEN4wt ED3 were retrieved from Uni-Prot (ID P27915.1; residues 574(294) to 678(398) and P09866; residues 575 to 679). Mutations at the 310/387 sites were introduced by site directed mutagenesis using a Quik Change protocol (Stratagene, USA). 2.2. Expression and purification of DEN ED3 The proteins were overexpressed in Escherichia coli JM109 (DE3) PLysS as inclusion bodies. Single colonies were grown at 37 °C in 50 mL Luria Broth (LB) supplemented with ampicllin (50 mg/mL) and chloramphenicol (35 mg/mL) for 10 h, and scaled up to 2 L. Protein expressions were induced at OD 590nm ~ 0.5 by the addition of 1.0 mM IPTG and the purification was carried out as previously reported [16]. In short, after harvesting, the cells were resuspended in lysis buffer (750 mM NaCl, 500 mM Tris–HCl pH 8.5, and 1% v/v NP-40). Cell lyses were carried out by sonication and the cysteines were air-oxidized for 36 h at 30 °C in 6 M Guanidine hydrochloride. His6-tagged proteins were purified by using Ni-NTA (Qiagen) chromatography, followed by dialysis against 50 mM Tris–HCl (pH 8.0) at 4 °C. The N-terminal His6-tag was cleaved by thrombin proteolysis at room temperature for 4 h and the proteins were further purified by a second passage through the Ni-NTA column. The final purification was carried out by reverse phase HPLC, and the protein's identities were confirmed by analytical HPLC and MALDITOF mass spectroscopy. The absence of free cysteines was confirmed by Ellman's assay. The purified proteins were lyophilized and stored at − 30 °C until use.

2.3. Crystallization, data collection and structure determination A stock solution was prepared by dissolving the lyophilized DEN4wt ED3 protein in 15 mM Tris–HCl pH 7.0 [15]. Crystals were developed at 4 °C in 30% PEG 3350 (Hampton Research, USA), 0.2 M ammonium sulfate, 0.1 M Tris–HCl pH8.5, and 1.0% dioxane in a sitting drop vapor diffusion setting with a drop size between 3.0 μL to 4.0 μL. The crystals were dehydrated under the same condition but with 40% PEG 3350 (Fig. S1). X-ray diffraction data were collected from a single crystal at the Photon Factory (KEK, Tsukuba, Japan) as previously reported [12]. The diffraction data were processed with the HKL2000 program package, using DENZO for the initial integration and SCALEPACK for the merging and statistical analysis of the diffraction intensities [17]. The structure was solved by molecular replacement (PHENIX) using the structure of DEN4wt E-protein (PDB ID: 3AUJ) as a molecular probe [18]. Editing and adjustment of the model were carried out by using the molecular graphic software COOT [19], and refinements were performed using PHENIX refine [18]. The structure was validated using MolProbity [20]. 2.4. CD (circular dichroism) measurements Samples for CD measurement were prepared by dissolving the lyophilized protein powders in 10 mM sodium acetate buffer pH 4.5 or phosphate buffer pH 6.5. The samples were centrifuged at 20,000 g for 30 min, and the pH of the samples was confirmed just before carrying out the experiment. CD spectra were measured using a 2 mm cuvette with a JASCO J-820 spectropolarimeter, as previously reported [16]. Thermal unfolding was monitored between 10 °C and 90 °C using the CD value at 217 nm. Experiments were conducted with 5 μM protein concentrations and a 1 °C/min scan rate, using a 1 cm optical path length cuvette. The reversibility of the thermal denaturation was confirmed by cooling the sample back to 10 °C. The melting temperatures (Tm) were computed by means of least-squares fittings of the experimental data with a two-state model using Origin 6.1 J [21]. 2.5. Indirect ELISA with monoclonal antibodies ELISA experiments were carried out using 96-well microtiter plates. The plates were coated by overnight incubation at 25 °C with 5 μg/mL of proteins (100 μL) dissolved in 50 mM phosphate buffer pH 7.4. Unbound proteins were washed out and the plates were blocked with 1% BSA in PBS. After 5 times washing with 1 × PBS, monoclonal AntiDEN3 and DEN4 antibodies (MAb 8703 and MAb 8704; Chemicon) were applied at different dilutions (a 1 mg/mL stock solution was diluted 10− 3 to 10−8 fold) and incubated at 37 °C for 2 h. After further washing with PBS for 5 times, the plates were dried and anti-mouseIgG-HRP conjugates (1:2500 dilution) were added and allowed for binding to anti-ED3 for 1 h. Finally, the unbound conjugates were removed by an extensive washing with 1X PBS and coloring was performed by adding the substrate OPD (O-phenyl Di-amine; 1 mg/mL μL) containing hydrogen peroxide (0.012%). The reaction was stopped using 20 μL of 1 M sulphuric acid, and the intensity of the color was measured at 480 nm. 2.6. Modeling of side-chain and identification of atomic clashes Structures of the mutants containing different combinations of amino acids at 310 and 387 and other selected residue pairs were modeled using the crystal structures of DEN3wt ED3 (PDB ID: 3VTT) and DEN4wt ED3 (PDB ID: 3WE1) as templates, and assuming that the overall backbone and template's side-chain structures remain unchanged. The mutated structures were further refined by using REFMAC 5.2 [22]. The side-chain structures were manually modeled by exhaustively varying the free chi angles using the rotamer conformations listed in the Richardson's penultimate rotamer library [23]. Side-

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

chain steric clashes were identified by using Molprobity with C-atom distances less than 2.7 Å [20]. We defined an inter-residue clash as strong when ≥ 70% of the rotamers contained a steric clash, and as weak when one or two rotamers created clashes.

587

serotypes, and their side-chain conformations were very well conserved (Fig. S5), suggesting that the side-chain's choices for producing a highly interdigitzed core are actually restricted. 3.2. Identification of putative structural compensation

2.7. Phylogenetic tree and ancestral sequence reconstruction The nucleotide sequences of four serotypes of DEN were collected from GenBank [24] and the sequence alignment was carried out with CLUSTAL W [25] using the default parameters. A phylogenetic tree was constructed by maximum likelihood, and a bootstrap analysis was carried out by using PhyML (http://www.phylogeny.fr) [26]. The ancestral sequence reconstruction was carried out by using the software ASR (http://www. datamonkey.org) [27] with the tree topology obtained in PhyML. 3. Results and discussion 3.1. Structure of DEN4wt ED3 The DEN4wt ED3 crystal was tetragonal with a P41212 space group and a unit cell dimension of a = b = 91.3, c = 73.1 Å, α = β = γ = 90.0 ° (Fig. S1, and Table 1). We solved the isotropic model structure at a 2.27 Å resolution with a crystallographic R-factor of 26.5% (Fig. 1A, and Table 1). The structure was very close to both the DEN4wt ED3 structures in the full length E-protein and the ED3 fragment structure complexed with a monoclonal antibody, with CαRMSDs of 0.62 Å and 0.43 Å, respectively [28,29] (Fig. S2). A peculiar distortion in the backbone structure of isolated DEN3wt ED3 near β-1 as compared to the full length E-protein [15] was also observed in the present isolated DEN4wt ED3 structure. Among serotypes, the DEN4wt ED3 structure was closest to DEN2wt ED3 (PDB ID: 10KE) with a Cα-RMSD of 0.76 Å, and the lowest sequence similarity between DEN4wt and DEN3wt ED3 (PDB ID: 3VTT) was reflected by the largest Cα-RMSD of 1.07 Å (Fig. S3 and Table 2) [15]. The secondary structures were well conserved among the serotypes (Fig. 1A and B). Finally, the electrostatic potential around the epitope 1 region varied among the serotypes, with DEN1 being negative, DEN2 neutral to slightly positive, DEN3 positive, and DEN4 negative (Fig. S4). The hydrophobic and aromatic residues distributed throughout the sequence of ED3 and folded into a hydrophobic core that was formed by twenty-two residues. Ten core residues were not entirely conserved, suggesting some flexibility in producing the high interdigitization of the side-chains stabilizing the protein core. On the other hand, the amino acids of the remaining core residues were conserved throughout the

In order to explore the structural and thermodynamic effects of residue substitutions that occurred upon evolution into the four dengue serotypes (Fig. 2A), we first searched for putative structurally compensated amino acid pairs [14] using the DEN3 and DEN4 ED3 crystal structures, according to the following protocol. First, we selected all buried hydrophobic residues that were not fully conserved among the four serotypes, and found ten such residues (Fig. 2B). Nine of them were part of the hydrophobic core, and only one (residue 356) was located a little apart on the β6–β7 loop (Fig. S6). From this list of residues, we selected all pairs with Cα distances ≤10.0 Å, where potential clashes can occur upon amino acid substitution and found sixteen residue pairs (Table 3). Next, we modeled the side-chain structures of the sixteen pairs by replacing the original amino acid by an amino acid present in one of the four serotypes and checked for potential clashes using all rotamers in the Penultimate library [23] (see Materials and Methods for details). Steric clashes were observed in twelve out of the sixtyseven possible residue pairs (Table 3), and strong inter-residue clashes were observed only for the Val310/Ile387 (Fig. 1C and 1D) and Phe335/ Ala367 (referred later as 335/367; where the residue pair is indicated by their residue numbers) pairs in the context of DEN3, suggesting that a size switch type substitution occurred between these two sites, which should be a rare event since two mutations need to happen almost simultaneously. However, we discarded the 335/367 pair because we were interested only in pairs where structural compensation was observed among the four present serotypes, which was not the case of 335/367 (Fig. 1C and 2D). Namely, our calculations did not predict that a natural “size switch” type mutation occurred between residues 335 and 367. 3.3. Thermodynamic stability and steric clashes/cavities of DEN ED3 variants Both DEN3wt and DEN4wt ED3 exhibited cooperative thermal denaturation curves, which are indicative of a natively folded globular protein [30], and a well packed protein hydrophobic core without steric clashes or cavities (Fig. S6). The midpoint temperatures for DEN3wt and DEN4wt ED3 were 66.4 °C and 69.3 °C at pH 4.5, respectively, and limited proteolysis confirmed that both DEN3wt and DEN4wt ED3 were stable at room temperature as they were not cleaved for 2 h by

Table 1 Data collection and refinement statistics for DEN3wt and DEN4wt ED3.

Data collection statistics

Refinement statistics

Ramachandran plot statistics

Parameters

DEN3wt ED3

DEN4wt ED3

Crystal system

Monoclinic

Tetragonal

Space group

P1211

P41212

Unit cell parameters (Å, °)

a = 51.0 b = 37.6 c = 52.5, α = 90.0, β = 105.0, γ = 90.0 30.20–1.70 (1.76–1.70)a 59,322 21,277 (2084) 99.1 (98.8) 3.8 (31.3) 18.9 (2.03) 2.8 (2.6) 22.7 20.6/25.5 0.006 1.14 96.5 3.5 3VTT

a = b = 91.3, c = 73.1, α = β = γ = 90.0 9.96–2.27 (2.36–2.27) 196,786 14,490 (1429) 100 (100) 6.7 (27.0) 29.1 (12.25) 13.6 (13.8) 34.8 26.5/28.5 0.008 1.13 95.3 3.6 3WE1

Resolution range (Å) Number of measured reflections Number of unique reflections Completeness (%) Rmerge (%) I/σ(I) Redundancy Wilson plot B (Å2) Rwork/Rfree (%) RMSD bonds (Å) RMSD angles (°) Most favored (%) Allowed (%)

PDB accession code a

Values for the highest resolution shell are given in parentheses.

588

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

Fig. 1. Structure and sequence of DEN ED3 (A). Ribbon model of the crystal structure of isolated DEN4wt ED3 (PDB ID: 3WE1). Residues 310/387 are shown with spheres. (B). Multiple sequence alignment of ED3 sequences of the four DEN serotypes using CLUSTAL W. The secondary structure elements are shown on the top of the sequences. The bold arrow and curved line indicate a β-strand and a β-turn, respectively. Conserved residues are shaded in gray. (C). Side-chain structures of the residues Met310/Leu387 as observed in DEN4wt ED3 crystal structure. (D). Side-chain structures of the residues Val310/Ile387 as observed in the isolated DEN3wt ED3 crystal structure [15]. In both panels, the van der Waal's radii are presented by a dotted surface and the minimum distances between the side-chains are indicated by dotted lines.

trypsin (1000:1) (Fig. S7). Mutational analysis indicated that Val310Met, which created a strong clash in DEN3_MI, and Ile387Leu, which created a small cavity in the hydrophobic core of DEN3_VL, destabilized DEN3 ED3 by as much as 10.2 °C and 5.5 °C, respectively (Fig. 3, and Table S1). However, the double mutations Val310Met/ Ile387Leu restored the stability to the level of DEN3wt (65.7 °C; Fig. 3, and Table S1). The choice of the template had some influence on our analysis, and our modeling predicted that the above mutations were destabilizing in the context of DEN1 and DEN3 ED3, but should barely affect the stability of DEN2, and DEN4 ED3 (Figs. 2D, 4, and Fig. S8). This difference between the DEN1/DEN3 and DEN2/DEN4 was related to a slight displacement of their backbones at these two residues. For example, the Cα distance between 310/387 was 8.4 Å in DEN3wt ED3 and 9.2 Å in DEN4wt ED3 (Figs. 2C, D, and S9). This Angstrom level backbone displacement was sufficient to mitigate the steric clash in DEN4_MI. We previously observed a similar situation when predicting mutations that significantly affected BPTI stability from clashes between two residues located in a surface exposed loop [12]. Table 2 Sequence and structural similarities among DEN ED3s.

DEN1 DEN2 DEN3 DEN4 Node 0 Node 1

DEN1

DEN2

DEN3

DEN4

– 79.0% 88.6% 71.4% 72.8% 86.0%

0.87 Åa – 78.0% 77.1% 87.7% 72.3%

0.57 Å 0.74 Å – 69.5% 63.8% 75.9%

1.01 Å 0.76 Å 1.07 Å – 74.9% 65.3%

a The values in the right top half and bottom left half represent, respectively, the Cα-RMSD and percentage identity between two serotypes ED3s.

We also assessed the involvement of surrounding residues in the predicted stabilization/destabilization of ED3. For example, residue 310 is very close to 320 in the tertiary structure with a Cα distance of 5.5 Å. Structural modeling suggested that three of the thirteen rotamers in DEN3_MI created clashes with Ile320 (Fig. S10). Further, this was experimentally confirmed, as the Ile320Val mutation could not overcome the destabilization of DEN3_MI, but rather showed an additive destabilizing effect (Fig. S11). 3.4. Ancestral sequence reconstruction supports maintenance of thermodynamic stability during evolution The nucleotide sequences of four serotypes of DEN were used for the reconstruction of the phylogenetic tree and inference of the ancestral sequences. The tree topology was obtained by the maximum likelihood method, and a bootstrap analysis strongly supported the tree topology with 97% reliability [26] (Fig. 5). The tree topology was consistent with those reported by other groups [31]. The close sequence relationship between DEN1 and DEN3 was also supported by the structural similarity (Fig. 4B, and Table 2). We reconstructed the ancestral nucleotide sequences for the nodes 0 and 1 based on this tree topology [27,32]. For some sites multiple nucleotides were inferred as ancestral states, but the ancestral nucleotides corresponding to amino acid residues 310/ 387 were predicted with strong support (Table S2). Thus, hereafter, only the nucleotides with the strongest support were used as the ancestral nucleotides. The nucleotides at node 0 for residues 310/387 were inferred as ATG (Met) and CTC (Leu), which were the same as those of DEN4, and at node 1 as GTG (Val) and CTC (Leu) (Fig. 5). As described in the previous section, the Val310/Leu387 pair created a cavity in a hypothetical DEN3_VL, which is responsible for a 5 °C thermal stability

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

589

Fig. 2. Structurally compensated pairs and “size switch type repacking” during the evolution of DEN. (A). Schematics of structural compensation pairs and “size switch type repacking” at residues 310/387 during the evolution of DEN. The figure is adapted for Dengue ED3 from Fig. 2a in reference [14]. The most probable evolutionary path is shown by the curved arrow. Experimental stability changes upon mutation are indicated by arrows with length proportional to the destabilization/stabilization effect. (B). Flow chart for detecting structurally compensated residue pairs. (C). Side-chains of DEN3wt ED3 (DEN3_VI) in the crystal structure, DEN3_MI, and DEN3_ML in the models from left to right in the upper panel. Side-chains of DEN3_VL, DEN3_IL, and DEN3_II in the model structures are presented in the lower panel. (D). Side-chains of DEN4wt (DEN4_ML) in the crystal structure, DEN4_MI, and DEN4_VI in the models from left to right in the upper panel. Side-chains of DEN4_VL, DEN4_IL, and DEN4_II in the model structures are presented in the lower panel. The minimum distance observed between two side-chains and the Cα distances between residues 310 and 387 are shown for all mutants. The strong clashes and cavities are indicated and the distances are shown by dotted lines. Nine of the thirteen rotamers created clashes with Ile387 and free rotations of the side-chains were not sufficient to overcome the clashes in DEN3_MI (Fig. S10). However, in the DEN4_MI model structure eleven of the thirteen rotamers could overcome steric clashes by the additional free rotations of the side-chains chi angles (Fig. S14).

decrease. However, this destabilization is presumably not detrimental for the survival of DEN3_VL virus, since DEN1wt does accommodate a stability drop arising from a similar cavity created by the Val310/Leu387

of its wild type structure (see Figs. S12 and S13 for a detailed consideration). With respect to the template dependency of the thermal stability of node 1's ED3 caused by the amino acid substitutions at sites 310/387,

Table 3 List of putative structural compensation residue pairs in DEN3wt ED3 based on conservation, burial, hydrophobicity, and Cα distances. Residue pair

Distance between residues (Å)

Respective amino acids in DEN3wt

Tested amino acid pairs for structural compensation (Represent all four serotypes and hypothetical combinations)

Accessible surface area (%) (Residue/Residue)

306/310 306/320 306/387 310/318 310/320 310/387 318/320 318/367 318/387 320/387 333/335 333/356 333/363 335/337 335/367 356/363

8.64 7.49 9.33 6.83 5.46 8.35 6.45 5.36 10.00 9.34 6.24 6.79 9.46 7.16 9.11 6.80

Leu/Val Leu/Ile Leu/Ile Val/Ile Val/Ile Val/Ile Ile//Ile Ile/Al Ile/Ile Ile/Ile Ile/Phe Ile/Val Ile/Val Phe/Thr Phe/Ala Val/Val

Val/Val, Ile/Val, Ile/Meta, Ile/Ile, Leu/Meta, Leu/Ile Val/Ile, Ile/Ilea, Leu/Val, Ile/Val, Val/Val Val/Ile, Ile/Ilea, Leu/Leu, Val/Leu, Ile/Leu Val/Val, Val/Thr, Ile/Ilea, Met/Ilea, Ile/Thr, Met/Thr Ile/Ilea, Met/Ilea, Val/Val, Ile/Val, Met/Val Ile/Ilea, Met/Ileb, Val/Leu, Ile/Leua, Met/Leu Ile//Val, Val//Ile, Val/Val, Thr/Ile, Thr/Val Ile/Leu, Val/Ala, Val/Leu, Thr/Ala, Thr/Leu Val/Ile, Thr/Ile, Ile/Leu, Val/Leu, Thr/Leu Val/Ile, Ile/Leu, Val/Leu Ile/Ile, Val/Phe, Val/Ile Ile/Ala, VaI/Val Ile/Thr, Val/Val, Val/Thr Phe/Ile, Ile/Ile, Ile/Thr Phe/Leub,Ile/Leub,Ile/Ala Val/Thr, Ala/Val, Ala/Thr

30.0/12.6 30.0/0.6 30.0/8.1 12.6/1.0 12.6/0.6 12.6/8.1 1.0/0.6 1.0/0.9 1.0/8.1 0.6/8.1 17.8/3.8 17.8/21.8 17.8/1.6 3.8/20.2 3.8/2.1 21.8/1.6

a b

Pairs which represent a weak inter-residue clash, defined when one or two rotamers contained a clash. Pairs which represent a strong inter-residue clash according to our definition, when ≥70% of the rotamers contained a steric clash.

590

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

Fig. 3. Thermal denaturation studies of DEN3 variants (A). Normalized thermal denaturation curves of DEN3wt ED3 and related mutants in 10 mM acetate buffer pH 4.5. Thermal denaturation was monitored by far-UV CD at 217 nm, and the raw data were corrected using linear baselines for the native (20 °C to 30 °C) and denatured states (80 °C to 90 °C). The DEN3 (filled squares; ■) and the related mutants DEN3_MI (open triangles; △), DEN3_VL (filled reverse triangles; ▼) and DEN3_ML (open circles; ○) are presented. Continuous lines represent the theoretical melting curves obtained by fitting the corresponding experimental data. (B). Normalized thermal denaturation curves of DEN3wt ED3 and related mutants in 10 mM phosphate buffer pH 6.5. Data analysis and representation are the same as Fig. 3A. (C). Surface representation of the area around residues 310 and 387 of DEN3wt (VI), and (D). DEN3_MI. The dots show the overlap of the van der Waals radius of Cδ of Met310 overlapped with the Cγ2 of Ile387. (E). Cavity created by an Ile387Leu substitution in the DEN3_VL model structure. The cavity was observed in all visual molecular graphic software including PyMol, CCP4MG, and JMol. In the DEN3wt (VI), this cavity is filled by the Cγ2 of Ile387. (F). The cavity created by the Ile387Leu substitution was filled by the Cδ of Met310 in the DEN3_ML model structure (ancestral pair).

it is reasonable to use the modeled DEN1 or DEN3 ED3's structure, since DEN1 and DEN3 are the direct descendants of the ancestral virus corresponding to node 1 (Fig. 5) and therefore their ED3 sequences are closest to that of the node 1 ED3 (Table 2, and Fig. 4B). Taken together, the reconstructed ancestral sequence agrees with the conclusion that ED3 evolved so that the node proteins on the evolutionary path had a minimal stability decrease. This hypothesis is in line with previous theoretical considerations predicting that the stability of globular proteins is maintained during molecular evolution [33,34]. Both phylogenetic analyses and structural modeling suggested the same order of substitution for residues 310/387 during the evolution of DEN into its four serotypes: the ancestral Met310 first mutated to Val310 and formed a Val310/Leu387 pair which was slightly destabilizing. A second Leu387Ile mutation then occurred, producing the Val310/Ile387 pair, which is observed in the present DEN3wt serotype. The alternative order, where the Leu387Ile mutation appears first was not sustained by the phylogenetic analysis and would produce Met310/Ile387, which was strongly (N10 °C) destabilizing in the context of DEN3 suggesting that the possibility of such an event is limited (Fig. 2A, and Table S3). This hypothesis is further corroborated by a simple analysis of amino acid variations at positions 310/387, which shows that no sequence in the current viral population contains the destabilizing Met310/Ile387 pair (Table 4).

Finally, we note that mutations at residues 310/387 are not “compensatory or correlated mutation” in the evolutionary sense. To date, we predicted, using the structures of DEN3 and DEN4 ED3, that the side-chains of residues 310/387 were “structurally compensated” [14] and that a size-switch type repacking had occurred at these sites, and we experimentally confirmed that the thermal stability of the single mutation variants decreased. However, the ancestral sequence reconstruction did not suggest that mutations at these two positions were correlated. That is, the mutations at the positions did not occur simultaneously, and the changes are not “compensatory” in the evolutionary sense [35]. This is because the observed evolutionary path (Met310Val and then Leu387Ile; Figs. 2A, 5, and Table S3) preserves thermal stability at each step of the evolution, and mutations can thus occur in a sequential manner without destabilization (Fig. 2A, and Table S3). The mutations would be identified as correlated (in an evolutionary sense) if only destabilizing paths were available, requiring simultaneous mutations at the two residues (Table S4). 3.5. Thermal stability and antibody interaction at physiological temperatures Despite the sharp drop in the stability of DEN3_MI, the melting temperatures of ED3 were all above 56 °C indicating that all molecules are folded at a physiological temperature. However, the 10 °C drop in the stability of DEN3_MI clearly correlated to a lower ELISA response

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

591

Fig. 4. (A). Thermal stability (Tm) and antibody interactions (ELISA) of DEN3 and DEN4 ED3 variants. The bars represent the midpoint temperature (Tm), and filled squares represent the ELISA reading at OD480. The error bars indicate standard deviations calculated from three successive experiments (see Table S1). ELISA was carried out by using commercially available anti DEN3-Mab against DEN3wt ED3 and its variants; and anti DEN4-Mab against DEN4wt ED3 and its variants. (B). Sequence similarities and Cα-RMSDs DEN ED3s from different serotypes. The bars and the filled triangles represent respectively, the percent similarities and Cα-RMSDs.

measured at room temperature (Fig. 4A, and Table S1). The weaker binding of the anti DEN3 MAb, might be related to a local unfolding in the region around residue 310, which includes the epitope as well as the heparan sulfate binding residues [5,6]. The node 1 sequence was closer to DEN3 than DEN4, and we speculate that the node 1 structure will be closer to that of DEN3 than DEN4 (Table 2 and Fig. 4B). Thus a hypothetical node 1 ED3 with Met310/Ile387, which was not predicted by ASR [27], would have a lower binding affinity to the heparan sulfate receptor and thus would be less infectious and evolutionary less fit than a node 1 ED3 with Val310/Leu387. Further, both the melting temperature and the ELISA response of all of the DEN4 variants remained

unchanged corroborating the hypothesis that the thermal stability of ED3 correlates with antibody interaction strength at physiological temperatures (Fig. 4A). 4. Conclusion Sequence comparison has long provided a powerful approach to explore evolutionary events [36], but some recent experiments have started to complement this traditional view [33,37,38]. Hypotheses based on experimental data need to be carefully interpreted in light of the evolutionary context. For example, artificial “size switch” mutations

Fig. 5. Phylogenetic tree topology of DEN viruses based on the ED3 sequences analyzed by maximum likelihood using PhyML [26]. The amino acids and their respective nucleotides at positions 310/387 are shown within the brackets. The nodes are named according to ASR analysis [32]. Nucleotide sequences of the four serotypes were used to reconstruct the phylogenetic tree and ancestral sequences. To date, the Leu codon in DEN1wt is a CTA, which is an exception, the Leu in all of the other serotypes being encoded with a CTC.

592

M. Elahi et al. / Biochimica et Biophysica Acta 1844 (2014) 585–592

Table 4 Distribution of residue pairs at 310 and 387 in natural isolates of DEN sequences. Serotypes

DEN1

DEN2

DEN3 DEN4

Amino acid residues at 310

387

Val Met Leu Ile Ile Val Met Val Met

Leu Leu Leu Leu Ile Leu Leu Ile Leu

Relative percentage

99.0% b1.0% b1.0% 98.0% 1.0% b1.0% b1.0% 100.0% 100.0%

The sequences were collected from Uni-prot, and multiple alignments using CLUSTAL W, and examined the frequency of amino acid pairs at positions 310/387.

are easily designed and introduced in a protein sequence [14], but it should be a rare natural event since two substitutions need to occur almost simultaneously for maintaining protein stability. “Size switch type repacking” between two residues is less rare, though it is not as common as demonstrated by our identification of a mere single event from the over 10,000 residue pairs of ED3. However, our study suggests that “size switch type repacking” can occur in any protein whenever an alternative evolutionary path with a non-detrimentally destabilizing intermediate is available. Author contributions Y.K. and M.E. designed research; M.E. performed research; M.Y. and K.N. contributed new reagents/analytic tools; M.E. and M.M.I. analyzed data; and M.E., H.T., and Y.K. wrote the paper. Funding sources This research was supported by a Japan Society for the Promotion of Science (JSPS) postdoctoral fellowship to M.M.I., and M.E. was supported by a MEXT PhD scholarship. Data deposition The coordinates and structure factors of DEN4 ED3 have been deposited in the Protein Data Bank (PDB) under the accession number 3WE1. Acknowledgement We thank Dr. Tetsuya Kamioka for help with expression vector preparation, Dr. Masafumi Odaka for advice on X-ray crystallography, Yuki Umezawa and Ryo Suzuki for computational help, and Patricia McGahan for English proofreading. X-ray Diffraction data were collected at the Photon Factory, KEK (Tsukuba, Japan). Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bbapap.2013.12.013. References [1] S. Bhatt, S.I. Hay, The global distribution and burden of dengue, Nature 496 (2013) 504–507. [2] G. Kuno, G.J. Chang, K.R. Tsuchiya, N. Karabatsos, C.B. Cropp, Phylogeny of the genus Flavivirus, J. Virol. 72 (1998) 73–83. [3] R.J. Kuhn, W. Zhang, T.S. Baker, J.H. Strauss, Structure of dengue virus: implications for flavivirus organization, maturation, and fusion, Cell 108 (2002) 717–725. [4] Y. Modis, S. Ogata, D. Clements, S.C. Harrison, Structure of the dengue virus envelope protein after membrane fusion, Nature 427 (2004) 313–319. [5] D. Watterson, B. Kobe, P.R. Young, Residues in domain III of the dengue virus envelope glycoprotein involved in cell-surface glycosaminoglycan binding, J. Gen. Virol. 93 (2012) 72–82.

[6] J.J. Hung, M.T. Hsieh, M.J. Young, C.L. Kao, W. Chang, An external loop region of domain III of dengue virus type 2 envelope protein is involved in serotype-specific binding to mosquito but not mammalian cells, J. Virol. 78 (2004) 378–388. [7] W.D. Crill, J.T. Roehrig, Monoclonal antibodies that bind to domain III of dengue virus E glycoprotein are the most efficient blockers of virus adsorption to Vero cells, J. Virol. 75 (2001) 7769–7773. [8] E.A. Franzosa, Y. Xia, Structural determinants of protein evolution are contextsensitive at the residue level, Mol. Biol. Evol. 26 (2009) 2387–2395. [9] M.M. Islam, S. Sohya, K. Noguchi, M. Yohda, Y. Kuroda, Crystal structure of an extensively simplified variant of bovine pancreatic trypsin inhibitor in which over one-third of the residues are alanines, Proc. Natl. Acad. Sci. U. S. A. 105 (2008) 15334–15339. [10] A.E. Eriksson, W.A. Baase, M. Blaber, E.P. Baldwin, B.W. Matthews, Response of a protein structure to cavity-creating mutations and its relation to the hydrophobic effect, Science 255 (1992) 178–183. [11] K. Ishikawa, H. Nakamura, K. Morikawa, S. Kanaya, Stabilization of Escherichia coli ribonuclease HI by cavity-filling mutations within a hydrophobic core, Biochemistry 32 (1993) 6171–6178. [12] M.M. Islam, S. Sohya, K. Noguchi, S. Kidokoro, M. Yohda, Y. Kuroda, Thermodynamic and structural analysis of highly stabilized BPTIs by single and double mutations, Proteins 77 (2009) 962–970. [13] J. Roche, M. Dellarole, C.A. Royer, B.E. Garcia-Moreno, C. Roumestand, Structural, energetic, and dynamic responses of the native state ensemble of staphylococcal nuclease to cavity-creating mutations, Proteins 81 (2013) 1069–1080. [14] E. Baldwin, J. Xu, O. Hajiseyedjavadi, W.A. Baase, B.W. Matthews, Thermodynamic and structural compensation in “size-switch” core repacking variants of bacteriophage T4 lysozyme, J. Mol. Biol. 259 (1996) 542–559. [15] M. Elahi, M.M. Islam, K. Noguchi, M. Yohda, Y. Kuroda, High resolution crystal structure of dengue-3 envelope protein domain III suggests possible molecular mechanisms for serospecific antibody recognition, Proteins 81 (2013) 1090–1095. [16] A. Kato, M. Yamada, S. Nakamura, S. Kidokoro, Y. Kuroda, Thermodynamic properties of BPTI variants with highly simplified amino acid sequences, J. Mol. Biol. 372 (2007) 737–746. [17] W.M. Zbyszek Otwinowski, Processing of X-ray diffraction data collected in oscillation mode, Methods Enzymol. 276 (1997) 307–326. [18] P.D. Adams, P.V. Afonine, D.C. Richardson, J.S. Richardson, T.C. Terwilliger, P.H. Zwart, PHENIX: a comprehensive Python-based system for macromolecular structure solution, Acta Crystallogr. D Biol. Crystallogr. 66 (2010) 213–221. [19] P. Emsley, K. Cowtan, Coot: model-building tools for molecular graphics, Acta Crystallogr. D Biol. Crystallogr. 60 (2004) 2126–2132. [20] I.W. Davis, J.S. Richardson, D.C. Richardson, MolProbity: all-atom contacts and structure validation for proteins and nucleic acids, Nucleic Acids Res. 35 (2007) 375–383. [21] S. Akiyama, A. Suenaga, T. Kobayashi, T. Kamioka, M. Taiji, Y. Kuroda, Experimental identification and theoretical analysis of a thermally stabilized green fluorescent protein variant, Biochemistry 51 (2012) 7974–7982. [22] E. Potterton, P. Briggs, M. Turkenburg, E. Dodson, A graphical user interface to the CCP4 program suite, Acta Crystallogr. D Biol. Crystallogr. 59 (2003) 1131–1137. [23] S.C. Lovell, J.M. Word, J.S. Richardson, D.C. Richardson, The penultimate rotamer library, Proteins 40 (2000) 389–408. [24] D.A. Benson, I. Karsch-Mizrachi, D.J. Lipman, J. Ostell, D.L. Wheeler, GenBank: update, Nucleic Acids Res. 32 (2004) 23–26. [25] J.D. Thompson, D.G. Higgins, T.J. Gibson, CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positionspecific gap penalties and weight matrix choice, Nucleic Acids Res. 22 (1994) 4673–4680. [26] A. Dereeper, V. Guignon, M. Lescot, J.M. Claverie, O. Gascuel, Phylogeny.fr: robust phylogenetic analysis for the non-specialist, Nucleic Acids Res. 36 (2008) 465–469. [27] S.L. Pond, S.D. Frost, Datamonkey: rapid detection of selective pressure on individual sites of codon alignments, Bioinformatics 21 (2005) 2531–2533. [28] C.M. Midgley, A. Flanagan, J.M. Grimes, G.R. Screaton, Structural analysis of a dengue cross-reactive antibody complexed with envelope domain III reveals the molecular basis of cross-reactivity, J. Immunol. 188 (2012) 4971–4979. [29] J.J.B. Cockburn, L.V. Chernomordik, C.J. Lai, F.A. Rey, Structural insights into the neutralization mechanism of a higher primate antibody against dengue virus, EMBO J. 31 (2012) 767–779. [30] K. Fukushima, J. Kikuchi, S. Koshiba, T. Kigawa, Y. Kuroda, S. Yokoyama, Solution structure of the DFF-C domain of DFF45/ICAD. A structural basis for the regulation of apoptotic DNA fragmentation, J. Mol. Biol. 321 (2002) 317–327. [31] E.C. Holmes, S.S. Twiddy, The origin, emergence and evolutionary genetics of dengue virus, Infect. Genet. Evol. 3 (2003) 19–28. [32] Z. Yang, S. Kumar, M. Nei, A new method of inference of ancestral nucleotide and amino acid sequences, Genetics 141 (1995) 1641–1650. [33] J.D. Bloom, S.T. Labthavikul, C.R. Otey, F.H. Arnold, Protein stability promotes evolvability, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 5869–5874. [34] N. Tokuriki, D.S. Tawfik, Stability effects of mutations and protein evolvability, Curr. Opin. Struct. Biol. 19 (2009) 596–604. [35] U. Göbel, C. Sander, R. Schneider, A. Valencia, Correlated mutations and residue contacts in proteins, Proteins 18 (1994) 309–317. [36] C.A. Orengo, J.M. Thornton, Protein families and their evolution—a structural perspective, Annu. Rev. Biochem. 74 (2005) 867–900. [37] S. Akanuma, Y. Nakajima, K. Miyazono, M. Tanokura, A. Yamagishi, Experimental evidence for the thermophilicity of ancestral life, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 11067–11072. [38] J. Lee, M. Blaber, Experimental support for the evolution of symmetric protein architecture from a simple peptide motif, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 126–130.