JOURNAL OF VIROLOGY, July 2007, p. 7517–7528 0022-538X/07/$08.00⫹0 doi:10.1128/JVI.00605-07 Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Vol. 81, No. 14
Identification of a Ca2⫹-Binding Domain in the Rubella Virus Nonstructural Protease䌤 Yubin Zhou,1 Wen-Pin Tzeng,2 Wei Yang,1 Yumei Zhou,2 Yiming Ye,1 Hsiau-wei Lee,1 Teryl K. Frey,2* and Jenny Yang1* Department of Chemistry1 and Department of Biology,2 Georgia State University, Atlanta, Georgia 30303 Received 21 March 2007/Accepted 25 April 2007
The rubella virus (RUB) nonstructural protein (NS) open reading frame (ORF) encodes a polypeptide precursor that is proteolytically self cleaved into two replicase components involved in viral RNA replication. A putative EF-hand Ca2ⴙ-binding motif that was conserved across different genotypes of RUB was predicted within the nonstructural protease that cleaves the precursor by using bioinformatics tools. To probe the metal-binding properties of this motif, we used an established grafting approach and engineered the 12-residue Ca2ⴙ-coordinating loop into a non-Ca2ⴙ-binding scaffold protein, CD2. The grafted EF-loop bound to Ca2ⴙ and its trivalent analogs Tb3ⴙ and La3ⴙ with Kds of 214, 47, and 14 M, respectively. Mutations (D1210A and D1217A) of two of the potential Ca2ⴙ-coordinating ligands in the EF-loop led to the elimination of Tb3ⴙ binding. Inductive coupled plasma mass spectrometry was used to confirm the presence of Ca2ⴙ ([Ca2ⴙ]/ [protein] ⴝ 0.7 ⴞ 0.2) in an NS protease minimal metal-binding domain, RUBCa, that spans the EF-hand motif. Conformational studies on RUBCa revealed that Ca2ⴙ binding induced local conformational changes and increased thermal stability (⌬Tm ⴝ 4.1°C). The infectivity of an RUB infectious cDNA clone containing the mutations D1210A/D1217A was decreased by ⬃20-fold in comparison to the wild-type (wt) clone, and these mutations rapidly reverted to the wt sequence. The NS protease containing these mutations was less efficient at precursor cleavage than the wt NS protease at 35°C, and the mutant NS protease was temperature sensitive at 39°C, confirming that the Ca2ⴙ-binding loop played a structural role in the NS protease and was specifically required for optimal stability under physiological conditions. Temporal and spatial changes in the intracellular Ca2⫹ concentration in eukaryotic cells affect the regulation of numerous cellular processes by modulating the activity of Ca2⫹-binding proteins (CaBPs) with varied Ca2⫹-binding affinities in different cellular compartments. Ca2⫹-binding motifs are also present in virus proteins and function in virion assembly and stability (1, 8, 15, 16, 20, 23, 24, 31, 33, 37, 40, 43, 55) and virion-associated activities, such as cell fusion (12, 16) and neuraminidase activity (5, 7). To date, almost all of the reported viral CaBPs are structural proteins, including both capsid proteins and envelope proteins. The helix-loop-helix EF-hand moiety is one of the most common motifs in proteins of bacteria, archaea, and eukaryotes (28, 73), and by binding to Ca2⫹ this motif may undergo conformational changes enabling Ca2⫹-modulated functions, as seen in the trigger or sensor proteins calmodulin and troponin C (27, 32), or may control the concentration of Ca2⫹ to maintain local Ca2⫹ homeostasis, as reported in buffering proteins such as parvalbumin (29) and calbindin D9k (56). The coordination of Ca2⫹ in EF-hand motifs is fulfilled by adopting a pentagonal bipyramidal geometry. Though EF-hands have been found abundantly in eukaryotes and bacteria (42, 46, 47,
51, 52, 73), this Ca2⫹-binding motif has seldom been reported in viruses. Rubella virus (RUB), the only member of the genus Rubivirus, in the Togaviridae family, is the causative agent of a disease called rubella or German measles. RUB is an enveloped, single-stranded positive-polarity RNA virus with a genome size of 9,762 nucleotides (21). The RUB genome contains two long open reading frames (ORFs): a 5⬘-proximal ORF, or NSP-ORF (nucleotides [nt] 41 to 6389 of the genome; 2,116 amino acids [aa]; molecular mass, 240 kDa), encodes the nonstructural proteins (NSP) involved in viral RNA replication; a 3⬘-proximal ORF, or structural protein (SP)-ORF (nt 6512 to 9700; 1,063 aa; molecular mass, 110 kDa), encodes one capsid protein (C) and two envelope glycoproteins (E1 and E2) (18, 20). Multiple-sequence alignments of the RUB NSP-ORF against other positive-stranded RNA virus genomes revealed a number of conserved domains, namely, methyltransferase (M), protease (P), helicase (H), replicase (R), and a proline-rich region (G), as well as an X domain (X) showing high homology to the Appr-1-p processing enzyme (1, 18). Upon uncoating and release of viral mRNA after entry into host cells, the NSP-ORF is immediately translated into a polypeptide precursor (P200), which is subsequently self cleaved by its protease domain (termed the nonstructural, or NS, protease) at a cleavage site mapped between residues G1301 and G1302 into two mature products (P150 and P90) involved in viral RNA replication in association with late endosomal or lysosomal membranes (30, 38, 43). The NS protease has been reported to be a Zn2⫹-dependent papain-like cysteine protease with C1152 and H1273 constituting the catalytic dyad (6, 23, 34, 36, 37).
* Corresponding authors. Mailing address for Jenny J. Yang: Department of Chemistry, Georgia State University, 50 Decatur St., Atlanta, GA 30303. Phone: (404) 651-4620. Fax: (404) 651-2751. E-mail:
[email protected]. Mailing address for Teryl K. Frey: Department of Biology, Georgia State University, 50 Decatur St., Atlanta, GA 30303. Phone and fax: (404) 651-3105. E-mail:
[email protected]. 䌤 Published ahead of print on 2 May 2007. 7517
7518
ZHOU ET AL.
In the present study, we report our prediction of a single EF-hand Ca2⫹-binding motif within the NS protease domain of RUB. To confirm this prediction, the binding loop was grafted into a scaffolding protein, CD2, and the minimal metalbinding domain of the NS protease was bacterially expressed. Both were found to bind Ca2⫹ or its trivalent analogs Tb3⫹ and La3⫹. We also found that mutagenesis of critical Ca2⫹-binding residues negatively impacted virus infectivity and rendered the activity of the NS protease temperature sensitive. This is one of the few demonstrations of an EF-hand Ca2⫹-binding motif in a virus-encoded protein. MATERIALS AND METHODS Grafting the predicted Ca2ⴙ-binding loop into CD2. The predicted 12-residue Ca2⫹-binding loop was directly inserted between Gly53 and Ala54 by PCR using an established protocol (70). The mutant with double mutations D5A and D12A (numbered according to the loop position of the inserted Ca2⫹-binding motif) was produced using standard PCR methods. All sequences were verified by automated sequencing on an ABI PRISM-377 DNA sequencer (Applied Biosystems) in the Advanced Biotechnology Core Facilities of Georgia State University. Molecular cloning of the minimal metal-binding domain. The sequence encoding the putative Ca2⫹-binding domain (RUBCa; aa 1143 to 1252 of the NS ORF) was amplified from the RUB infectious cDNA clone Robo502 (49, 61) by using standard PCR methods. The PCR product was subsequently inserted into the pGEX-2T vector (GE Healthcare) to produce the plasmid pGEX-2TRUBCa. Expression and purification of RUBCa, CD2.RUBCa, and its mutant. RUBCa was expressed as a glutathione S-transferase (GST) fusion protein in Escherichia coli BL21(DE3) transformed with the plasmid pGEX-2T-RUBCa in Luria-Bertani medium with 100 mg/liter of ampicillin and grown at 37°C. One hundred micromolars of isopropyl--D-thiogalactopyranoside (IPTG) and 50 M of ZnCl2 were added when the optical density at 600 nm reached 0.7 to induce protein expression for another 3 to 4 h. The proteins were purified following the protocols for GST fusion protein purification (70) using glutathione Sepharose 4B beads (GE Healthcare). The protein was cleaved from its GST tag on beads by taking advantage of the thrombin cleavage site and eluted. The eluted fractions containing RUBCa were further purified using Superdex 75 and Hitrap SP columns (GE Healthcare). The molecular weight of RUBCa was confirmed by matrix-assisted laser desorption ionization–time of flight mass spectrometry in the Advanced Biotechnology Core Facilities of Georgia State University. The concentration of RUBCa was measured by its absorption at 280 nm with an extinction coefficient of 19,630 M⫺1 cm⫺1, which was calculated according to previously described methods (22). The engineered proteins CD2.RUBCa and the mutant CD2.RUBCa.D5A/D12A were expressed and purified according to methods described previously (70). The molecular weights of CD2.RUBCa and the D5A/D12A mutant were also confirmed by matrix-assisted laser desorption ionization–time of flight mass spectrometry. The concentration was determined by using the absorption at 280 nm with an extinction coefficient of 11,700 M⫺1 cm⫺1 (15). ESI-MS. The detection of metal-protein complex by electrospray ionization mass spectrometry (ESI-MS) was performed by using a Waters Micromass Q-Tof micro instrument. The data were acquired in positive-ion mode by syringe pump infusion of the protein solutions at a flow rate of 7 l/min. The protein sample stock (⬃1 mM) in 10 mM Tris, pH 7.4, was diluted 100-fold into water. Metal ions were added in 5 M excesses to the protein concentration to observe specific binding. Refolding of RUBCa. RUBCa denatured in 8 M urea was refolded by gradually diluting the sample in the presence of a 2⫻ excess molar concentration of CaCl2 until the concentration of urea reached 2 M. The mixture was incubated and stirred gently for 16 h at 4°C. The sample was then extensively dialyzed against 10 mM Tris-HCl (pH 7.4) buffer. Considering the presence of six cysteines in the protein, 1 mM reduced and 0.2 mM oxidized glutathione were added into the buffer to facilitate the proper formation of possible disulfide bonds. ICP-MS. All the glassware, plasticware, and Teflon containers used in the preparation of samples were pretreated with 2% HNO3 (optima grade; Fisher Scientific). All buffers were pretreated with Chelex-100 (Bio-Rad) to remove the background Ca2⫹ ions. The refolded proteins (20 to 30 M) were acidified with 2% HNO3 and analyzed by inductively coupled plasma mass spectrometry (ICP-
J. VIROL. MS) (Finnigan Element 2). The dialyzed buffer was used as a blank, and the residual Ca2⫹ background of 2 to 5 M was subtracted from the measurement of protein samples. CD spectroscopy. The circular dichroism (CD) spectra of proteins were recorded in a Jasco-810 spectropolarimeter at ambient temperature using a quartz cell of 10-mm path length, with the protein concentrations ranging from 2 to 5 M. All spectra were obtained as the average of at least eight scans with a scan rate of 100 nm/min. The calculation of secondary-structure elements was performed by using DICHROWEB, an online server for protein secondary structure analyses (67). The thermal denaturation was studied using a 1-mm quartz cell with protein concentrations of 15 to 30 M in 10 mM Tris-HCl (pH 7.4), 100 mM KCl with 1 mM EGTA or 1 mM CaCl2. The ellipticity was measured from 190 to 260 nm and converted to mean residue molar ellipticity (degrees [deg] cm2 dmol⫺1 residue [res]⫺1). To obtain the thermal transition point, the signal changes at 222 nm were fitted using the equation described previously (39). Fluorescence spectroscopy. Fluorescence emission spectra were measured using a 1-cm-path-length cell on a PTI lifetime fluorimeter at ambient temperature. For intrinsic tryptophan fluorescence, spectral measurements were carried out at protein concentrations of 2 to 4 M in 20 mM piperazine-N,N⬘-bis(2-ethanesulfonic acid) (PIPES)–10 mM KCl at pH 6.8 with slit widths of 4 and 8 nm for excitation and emission, respectively. The emission spectra were collected from 300 to 400 nm with an excitation wavelength at 282 nm. 8-Anilino-1-naphthalene sulfonic acid (ANS) fluorescence emission spectra were recorded from 400 to 600 nm, with the excitation wavelength at 390 nm. Protein samples (5 M) with 1 mM CaCl2 or 1 mM EGTA were added into the solution containing 40 M ANS, 20 mM PIPES–10 mM KCl at pH 6.8. Tyr/Trp-sensitized Tb3⫹ fluorescence energy transfer (Tb3⫹-FRET) experiments were conducted as described previously (69, 71). For the metal competition studies, the solution containing 40 M of Tb3⫹ and 1.5 M of protein was set as the starting point. The stock solution of metal ions with the same concentration of Tb3⫹ and protein was gradually added into the initial mixture. The fluorescence intensity was normalized by subtracting the contribution of the baseline slope using logarithmic fitting. The Tb3⫹-binding affinity of protein was obtained by fitting normalized fluorescence intensity data using the equation f⫽
共关P兴T ⫹ 关M兴T ⫹ Kd兲 ⫺
冑共关P兴T ⫹ 关M兴T ⫹ Kd兲2 ⫺ 4关P兴T关M兴T 2关P兴T
(1)
where f is the fractional change, Kd is the dissociation constant for Tb3⫹, and [P]T and [M]T are the total concentrations of protein and Tb3⫹, respectively. The Ca2⫹ competition data were first analyzed using the apparent dissociation constant obtained with equation 1. By assuming that the sample is almost saturated with Tb3⫹ at the starting point of the competition, the Ca2⫹-binding affinity is further obtained by using the equation Kd,Ca ⫽ Kapp ⫻
Kd,Tb Kd,Tb ⫹ 关MTb兴
(2)
where Kd,Ca and Kd,Tb are the dissociation constants of Ca2⫹ and Tb3⫹, respectively. Kapp is the apparent dissociation constant. All the measurements were conducted in triplicate. 1 H NMR spectroscopy. One-dimensional 1H nuclear magnetic resonance (1D 1 H NMR) spectra were recorded on a Varian 500 MHz NMR spectrometer with a spectral width of 6,600 Hz. Samples of 200 to 250 M were placed in 20 mM PIPES-10 mM KCl, 10% D2O at pH 6.8. La3⫹ stock solution was gradually added into the NMR sample tube. The program FELIX98 (MSI) was used to process NMR data, with an exponential line broadening of a 2-Hz window function and the suppression of water signal with a Gaussian deconvolution function with a width of 20. Cells, infectious clone, and site-directed mutagenesis. The infectious genomic cDNA clone Robo502 was previously described (49, 61). To generate the mutated construct Robo502AA (D1210A and D1217A), a two-round asymmetric PCR strategy was employed to create the mutations in an amplified fragment between unique BsmI and RsrII sites at nt 3243 and 3897 of the RUB genome. The doubly digested fragment was used to replace the corresponding fragment in Robo502. The mutations were also transferred to the replicons RUBrep-FLAG/ GFP and RUBrep-HA/GFP (63), which express a P150 tagged with the FLAG or the hemagglutinin (HA) epitope and the reporter protein, green fluorescent protein (GFP), in place of the SP-ORF. In vitro transcription, transfection, Northern blotting, and Western blotting. Vero cells were obtained from the American Type Culture Collection and maintained at 35°C under 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) supplemented with 5% fetal bovine serum and 10 g/ml gentamicin. CsCl density gradient-purified plasmids were linearized with EcoRI (New England
Ca2⫹-BINDING DOMAIN IN THE RUB NONSTRUCTURAL PROTEASE
VOL. 81, 2007
Biolabs), followed by phenol-chloroform extraction and ethanol precipitation. RNA transcripts were synthesized at 37°C in a 25-l reaction mixture containing 40 mM Tris-HCl (pH 7.5), 1 g DNA, 6 mM MgCl2, 2 mM spermidine, 10 mM dithiothreitol, 1 mM of each nucleoside triphosphate, 1 U of RNasin (Roche Applied Science), 2 mM cap analog m7G(5⬘)ppp(5⬘)G (New England Biolabs), and 25 U of SP6 DNA-dependent RNA polymerase (Epicenter Technologies). The transcription reaction mixtures were directly used for cell transfection without DNase treatment. The transcripts were analyzed by electrophoresis of 2-l aliquots of the reaction mixture in 0.8% agarose gel. The precise yield was determined using a UV/vis spectrometer (Shimadzu). Transfection of the RNA transcripts was done using Vero cells at ⬃80% confluence in 60-mm2 plates. Vero cells were washed twice with 3 ml of phosphate-buffered saline and once with 3 ml of Opti-MEMI (Gibco). Ten microliters of the transcripts was subsequently mixed with 500 l Opti-MEMI and 7 l Lipofectamine 200 (1 mg/ml). The mixture and 1.5 ml of Opti-MEMI were added to the plates to cover the monolayer of Vero cells. After 4 h of incubation, the transfection mixture was removed and 4 ml of DMEM supplemented with 2% fetal bovine serum was added (transfection is designated passage 0, or P0). Following development of significant cytopathic effect, the medium was harvested and passaged twice in Vero cells (P1 and P2). Virus titers were determined by plaque assay as previously described (48, 65). To determine if the D1210A and D1217A mutations were maintained following transfection and passage, plaques were picked from terminal dilution plaque assay plates of P1 culture fluid and amplified once in Vero cells. Following amplification, total infected cell RNA was extracted and the NS protease region of the genome was amplified by reverse transcriptionPCR, and the reverse transcription-PCR product was sequenced (62). The expression of reporter gene GFP in living, transfected cells was monitored by a Zeiss Axioplan upright microscope with epifluorescence capability and photographed with a Zeiss Axiocam. Northern blotting was used to detect replicon-specific RNA species by following previously established protocols (63). To detect in vivo NS protease activity, Vero cells transfected with RUBrep-HA/ GFP, RUBrepAA-HA/GFP, or RUBrepNSⴱ-HA/GFP (containing a C1152A mutation of the catalytic site) were incubated at 35°C or 39°C. Six hours posttransfection, the cells were lysed and resolved by 8% sodium dodecyl sulfate–8% polyacrylamide gel electrophoresis (SDS–8% PAGE), electroblotted onto nitrocellulose membranes, and probed with anti-HA monoclonal antibodies (Sigma) according to a previously established procedure (63). Computer-aided sequence and structural analyses. The Ca2⫹-binding motif signatures developed in our laboratory (http://www.chemistry.gsu.edu/faculty /Yang/Calciomics.htm) as well as the pattern PS00018 from PROSITE (http://us .expasy.org/cgi-bin/nicedoc.pl?PDOC00018) were used to scan through the genome of RUB for any potential EF-hand Ca2⫹-binding site (73). Sequence alignments were conducted by using the program ClustalW. The secondary structure was predicted based on the consensus prediction results using programs PSIPRED, JPRED, and PHD (10, 41, 54). The homology modeling of the engineered protein and the protease was constructed using the comparative structure modeling server SWISS-MODEL (57). CD2 (Protein Databank [pdb] code 1hng) and the leader protease of foot-and-mouth disease virus (FMDV) (pdb code 1qmy) were used as the templates for structure modeling based on their available high-resolution structures (15, 24). Prediction of Ca2⫹-binding sites in the modeled structure was conducted by using the program GG, a computational algorithm developed in our laboratory on the basis of the geometric description, graph theory, and key structural and chemical features associated with the calcium binding in proteins (11).
RESULTS Predicting an EF-hand Ca2ⴙ-binding motif and homology modeling of the RUB NS protease. By taking advantage of the sequence alignment of currently available EF-hand proteins and considering the structural context of the Ca2⫹-binding loop, we generated a series of patterns for the prediction of EF-hand proteins (73). Using this prediction method (http: //www.chemistry.gsu.edu/faculty/Yang/Calciomics.htm), we detected one putative EF-hand Ca2⫹-binding motif (aa 1197 to 1225) within the RUB NS protease intertwined with the partially characterized cysteine-rich Zn2⫹-binding motif (Fig. 1A) (37). This putative Ca2⫹-binding motif shared high homology with well-known EF-hand proteins such as calmodulin and calcyphosine (Fig. 1B). In canonical EF-hand Ca2⫹-binding
7519
FIG. 1. Genomic organization of RUB and identification of a potential EF-hand Ca2⫹-binding motif. (A) Schematic representation of the RUB genome organization and domain locations. M, methyltransferase; G, proline-rich hinge; P, protease; H, helicase; R, RNA-dependent RNA polymerase; C, capsid protein; E1 and E2, envelope glycoproteins. The X domain represents a conserved domain with unknown function that is also found in NSP3 from alphaviruses and belongs to the Appr-1-processing enzyme family. The protease contains at least one Zn2⫹-binding site as well as a putative EF-hand Ca2⫹-binding motif (box with diagonal pattern). It has been proposed that the residues C1152 and H1273 constitute the catalytic dyad (stars). The minimal metal-binding domain RUBCa (aa 1143 to 1252) is bacterially expressed and used in this study. (B) Sequence alignment results of putative EF-hand Ca2⫹-binding motif RUBCa in the protease domain with calmodulin and calcyphosine. The motif remained conserved in both clades I (genotypes RV1a, 1B, 1C, 1D, and 1E) and II (genotypes RV2A, 2B, and 2c) of RUB. Boldface residues represent the potential Ca2⫹-coordinating residues. NCBI or GenBank accession numbers are the following: CaM_EF1 and CaM_EF2 (calmodulin, P62158); CAYP_hum and CAYP_rab (calcyphosine from human, Q13938; rabbit, P41150). The genomic sequence of the eight different genotypes of RUB was sequenced in our laboratory (unpublished data). (C) Homology modeling of RUB NS protease. The leader protease of FMDV (pdb code 1qmy, chain A), a papain-like cysteine protease with a high-resolution structure available, is chosen as the template for homology modeling. The active site consists of C1152 and H1273 (shown as sticks). The predicted EF-hand Ca2⫹-binding motif, located at the opposite side, is highlighted in blue with the calcium ion shown as a cyan sphere. The cysteine (green) and tryptophan (magenta) residues in the protease domain are shown as sticks. The cleavage site G1301G1302 is highlighted in black.
7520
ZHOU ET AL.
J. VIROL.
FIG. 2. Grafting the predicted EF-hand Ca2⫹-binding loop into CD2.D1 and formation of metal-protein complex. (A) Model structure of the engineered protein CD2.RUBCa. The Ca2⫹-binding loop (black) from the RUB NS protease (aa 1206 to 1217), with two glycines on the left and three glycines on the right, rendering flexibility, is grafted to the loop which connects strands C⬙ and D. The model structure was built using the automated comparative protein-modeling server SWISS-MODEL. Aromatic residues Trp (black) are shown as sticks. The Ca2⫹ ion is shown as a sphere. Intrinsic Trp fluorescence emission spectra (B) and far-UV CD spectra (C) of CD2 (open circle) and engineered protein CD2.RUBCa (closed circle) are compared to examine the perturbation of insertion on the scaffold protein CD2.D1. Buffers consists of 10 mM Tris, 10 mM KCl, pH 7.4. (D) Electrospray mass spectra of CD2 with a grafted EF-loop from RUB NS protease (P) in the presence of a fivefold molar excess of TbCl3 (⫹156).
motifs, the bidentate ligand residue at loop position 12 is either a Glu (95%) or an Asp (5%). Considering its longer side chain and closer distance to the metal ion, a Glu at this position (as seen in calmodulin) ensures a stronger binding affinity for Ca2⫹ than does an Asp. The predicted motif in the RUB NS protease contained an Asp at loop position 12, similar to the first EF-hand in calcyphosine, regulatory light chain of myosin (3, 66), and sarcoplasmic CaBP (9) that exhibit relatively weaker Ca2⫹-binding affinities. Prediction of the secondary structure elements revealed that the RUB NS protease Ca2⫹binding loop was flanked by two helices, as observed in most EF-hand CaBPs (47). The predicted Ca2⫹-binding motif is highly conserved among the eight genotypes of RUB for which sequence of this region is available (72), including all of the Ca2⫹-binding coordination ligands (Fig. 1B). Since the three-dimensional (3D) structure of a viral papainlike cysteine protease, the leader protease of FMDV, has been determined (25, 53), it provides an excellent template for homology modeling of the NS protease of RUB. Pairwise sequence alignment of the leader protease of FMDV (pdb code 1qmy) with the RUB NS protease revealed 12.1% sequence identity and highly similar secondary-structure arrangement.
Based on the sequence alignment results, a homology model of the RUB NS protease has been generated using the program SWISS-MODEL (Fig. 1C). With the program GG (11), we further detected a potential Ca2⫹-binding pocket around the predicted EF-hand Ca2⫹-binding motif in this model structure, indicating perfect agreement between the prediction made on the basis of primary sequence and the 3D structure of RUB NS protease. Grafting the EF-hand Ca2ⴙ-binding loop into CD2 domain 1 (CD2.D1). Although the RUB NS protease was previously expressed in active form as a maltose-binding protein (MBP) fusion (37), it was never purified free from the MBP domain despite numerous attempts. We have likewise expressed it as a GST fusion protein but were unable to purify it following cleavage from the GST domain, because it was insoluble (likely because it is highly hydrophobic in nature). Therefore, to study the Ca2⫹-binding capability of the putative Ca2⫹-binding motif, we grafted the 12-residue Ca2⫹-binding loop (aa 1206 to 1217) into a non-Ca2⫹-binding scaffold protein, CD2.D1. As seen in Fig. 2A, CD2 is a cellular adhesion molecule composed of nine -strands with an immunoglobulin-like fold, and the putative Ca2⫹-binding loop from the RUB NS protease was
VOL. 81, 2007
Ca2⫹-BINDING DOMAIN IN THE RUB NONSTRUCTURAL PROTEASE
7521
FIG. 3. Obtaining metal-binding affinity using aromatic residue-sensitized Tb3⫹-FRET and 1D 1H NMR. (A) Tb3⫹ fluorescence enhancement at 545 nm due to resonance energy transfer (excited at 282 nm) as a function of protein concentrations of CD2 (open diamond), CD2.RUBCa (closed circle), and its mutant D5A/D12A (open circle). (B) Tb3⫹ titration of CD2.RUBCa. Normalized fluorescence intensity was plotted as a function of the Tb3⫹ concentration. The inset shows the Ca2⫹ competition titration curve of CD2.RUBCa (1.5 M) preincubated with 40 M Tb3⫹ in 20 mM PIPES, 10 mM KCl, pH 6.8. (C) La3⫹ titration of CD2.RUBCa monitored by 1D 1H NMR. Some resonances (arrows) at amide regions of 1D 1H NMR spectra of CD2.RUBCa (0.2 mM) shift with the increased concentrations of La3⫹ (from bottom to top: 0, 39.2, 113.2, 214.3, 442.2, 360.7, 605.1, 929.1, and 1,411.2 M, respectively) in 20 mM PIPES, 10 mM KCl, at pH 7.4. (D) The chemical shift change at different resonant regions as a function of the concentration of La3⫹. ppm, parts per million.
inserted between Gly53 and Ala54, which lie within the loop between strands C⬙ and D of CD2; the grafted construct was termed CD2.RUBCa. Additionally, three glycines were placed on each side of the Ca2⫹-binding loop to serve as linkers to provide sufficient conformational freedom for the grafted Ca2⫹-binding loop and to minimize perturbation on the host protein while allowing the inserted loop to retain its capability of chelating Ca2⫹ (31, 69–71). The spectroscopic silence of Ca2⫹ makes it extremely challenging to investigate Ca2⫹-binding properties directly. Therefore, lanthanides are commonly used to probe Ca2⫹-binding sites due to their similar ionic radii (Ca2⫹, 1.00 Å; Tb3⫹, 0.92 Å; La3⫹, 1.03 Å) and metal coordination chemistry. In CD2.RUBCa, four aromatic residues in the host protein spaced closely (⬍15 Å) to the grafted loop made it possible to monitor the binding of Tb3⫹ using FRET (Fig. 2A). A variety of spectroscopic methods confirmed that the structure of the CD2.D1 had not been perturbed by the inserted loop either in the absence or presence of metal ions. The Trp intrinsic fluorescence spectrum of CD2 overlapped with that of CD2.RUBCa, both of which exhibited the emission maximum at 325 nm and a shoulder peak at 313 nm (Fig. 2B). The
far-UV CD spectra of CD2.RUBCa remained similar to that of CD2.D1 alone with a deep trough at 216 nm, characteristic of typical -sheet elements (Fig. 2C). The differences observed could be primarily from the inserted 12-residue loop and flanking triple-glycine linkers. The 2D 1H NMR further showed that the majority of the dispersed signals in the fingerprint regions of the total correlated spectroscopy and nuclear overhauser enhancement spectroscopy spectra of CD2.RUBCa were located at the same positions as that of CD2, suggesting that the integrity and packing of the host protein frame are maintained after the insertion of the Ca2⫹-binding loop from RUB NS protease (unpublished data). In general, the insertion of the foreign acidic sequence did not cause major conformational changes of the host protein, and the overall fold of CD2.RUBCa remained similar to that of CD2.D1. The ability of the grafted protein (calculated molecular mass, 12,712.9 Da) to form 1:1 metal/protein complexes was confirmed by ESIMS. As shown in Fig. 2D, the molecular mass of the engineered protein CD2.RUBCa agreed well with theoretical mass. The presence of a fivefold excess of Tb3⫹ led to the emergence of a new peak and additional mass of 156 Da. Similar results were also obtained for La3⫹. This provides the
7522
ZHOU ET AL.
J. VIROL.
FIG. 4. Metal selectivity of the engineered protein CD2.RUBCa. (A) Metal competition assay. The addition of 1.5 M CD2.RUBCa to free Tb3⫹ (40 M) solution resulted in an increase of fluorescence intensity at 545 nm by over 20-fold due to the binding of Tb3⫹ to the protein and the resultant FRET. K⫹ (100 mM), 10 mM Mg2⫹, 1 mM Ca2⫹, and 0.1 mM La3⫹ were subsequently added to individually prepared solutions containing 40 M Tb3⫹ and 1.5 M CD2.RUBCa. (B) Amide region of 1D 1H NMR spectrum of CD2.RUBCa with sequential addition of 100 mM K⫹, 10 mM Mg2⫹, and 1 mM Ca2⫹. Resonances exhibiting changes are indicated by arrows. ppm, parts per million.
foundation for measuring the intrinsic metal-binding affinity with a minimized contribution from protein conformational change and minimal influence of the host protein environment on the grafted EF-hand loop (69). Obtaining the metal-binding affinities of grafted Ca2ⴙ-binding loop by FRET and NMR. As shown in Fig. 3A, the addition of increasing amounts of CD2.RUBCa to 10 M Tb3⫹ resulted in an increase of Tb3⫹ emission fluorescence intensity at 545 nm when excited at 282 nm, suggesting the formation of a metal-protein complex. However, the addition of both CD2.D1 and CD2.RUBCa.AA, a mutated construct with glutamic acidto-alanine changes at two of the ligands within the grafted loop (D5A/D12A), had substantially weaker enhancement than that of CD2.RUBCa, suggesting that the mutant lost its specific Tb3⫹-binding capability and that Tb3⫹ bound to the grafted loop through the ligands D5 and D12. By monitoring the changes of Tb3⫹ emission enhancement of a fixed amount of proteins as a function of Tb3⫹ concentration (Fig. 3B) and by assuming a 1:1 (Tb3⫹ to protein) binding ratio, a dissociation constant (Kd) of 47 ⫾ 4 M was obtained for CD2.RUBCa. In addition, the dissociation constant of Ca2⫹ was indirectly estimated by a metal ion competition assay in which 1.5 M of CD2.RUBCa with 40 M Tb3⫹ was titrated with increasing amounts of Ca2⫹ in excess. As shown in the inset of Fig. 3B, the Tb3⫹ fluorescence emission maximum at 545 nm decreased due to the competition between Tb3⫹ and Ca2⫹ for the grafted metal-binding site, yielding an apparent dissociation constant of 395 ⫾ 135 M and a real dissociation constant of 214 ⫾ 73 M for Ca2⫹ according to equation 2. 1D 1H NMR was conducted to monitor the La3⫹-induced chemical shift changes of CD2.RUBCa as a function of La3⫹ concentration. Several resonances, such as those at 7.97, 6.98, 6.79, and 2.67 ppm (Fig. 3C), gradually shifted or sharpened as the La3⫹ concentration increased, indicating a single binding process. The La3⫹-binding affinity, with an average of 14 ⫾ 7
M, was obtained by plotting the chemical shift of these peaks as a function of La3⫹ concentration (Fig. 3D). Metal selectivity. Competition assays based on Tb3⫹-FRET have been used as a convenient method to probe the binding capacity of physiologically competing metal ions (e.g., 100 mM K⫹ and 10 mM Mg2⫹) to EF-hand CaBPs, such as calmodulin (64) and galactose-binding protein (14). As seen in Fig. 4A, a significant decrease in Tb3⫹-FRET elicited from Tb3⫹-loaded CD2.RUBca was only observed following the addition of 1 mM Ca2⫹ or 0.1 mM La3⫹ (another Ca2⫹ analog, as discussed above). However, the addition of 100 mM K⫹ or 10 mM Mg2⫹ led to only a slight decrease (⬍10%) in the intensity of Tb3⫹ fluorescence. To further clarify whether the engineered protein CD2.RUBCa exhibited selectivity for Ca2⫹, we monitored the 1D 1H NMR spectrum with sequential additions of K⫹, Mg2⫹, and Ca2⫹. As shown in Fig. 4B, while excess K⫹ and Mg2⫹ did not result in any changes in the chemical shifts, Ca2⫹ was able to specifically induce changes of chemical shifts in the main-chain amide proton region even in the presence of 100 mM K⫹ and 10 mM Mg2⫹. These results clearly demonstrate that CD2.RUBCa is capable of binding Ca2⫹ with selectivity over Mg2⫹ or K⫹, similar to other known EF-hand proteins, such as calmodulin (44, 50). Using the minimal metal-binding domain RUBCa for metalbinding and conformational studies. While our attempts to express the entire NS protease (aa 1000 to 1300) of P150 were fruitless, we successfully expressed a minimal metal-binding domain as a GST fusion protein and purified the minimal metal-binding domain (RUBCa, aa 1143 to 1252; see Fig. 1), which contains the Ca2⫹-binding loop, both catalytic residues, and the ligands thought to be involved in Zn2⫹ binding. Purified RUBCa was used to characterize the predicted EF-hand Ca2⫹-binding motif in its native protein environment. When purified RUBCa was reconstituted in the presence of Ca2⫹ and subjected to ICP-MS to measure its metal content, it was found
VOL. 81, 2007
Ca2⫹-BINDING DOMAIN IN THE RUB NONSTRUCTURAL PROTEASE
7523
FIG. 5. Metal ion titration of the minimal metal-binding domain RUBCa monitored by aromatic residue-sensitized Tb3⫹ fluorescence (A) and intrinsic Trp fluorescence (B). (A) Normalized Tb3⫹ fluorescence spectra of RUBCa with increasing concentrations of Tb3⫹ (from bottom to top: 0, 1.0, 4.0, 9.9, 14.8, 19.6, and 14.4 M, respectively). The inset shows the Tb3⫹ fluorescence enhancement at 545 nm due to energy transfer as a function of the concentration of Tb3⫹. (B) Intrinsic Trp fluorescence emission spectra of RUBCa (2.5 M) with increasing concentrations of Ca2⫹ (from top to bottom: 0, 49.8, 291.3, 566.0, 740.7, and 909.1 M, respectively). The inset shows the intrinsic Trp fluorescence intensity plotted as a function of the concentration of Ca2⫹. An average dissociation constant of 316 M was obtained by assuming a 1:1 binding model. The excitation wavelength was set at 282 nm. All the buffers used in metal titration consist of 20 mM PIPES, 10 mM KCl, pH 6.8.
to bind Ca2⫹ with a molar ratio (Ca2⫹/protein) of 0.7 ⫾ 0.2 to 1 (n ⫽ 2). RUBCa contains three Trp residues, and according to the model of the structure of the RUB NS protease (Fig. 1C), one of these aromatic residues, W1153, is in close proximity to the predicted Ca2⫹-binding loop (3.5 Å) (Fig. 1C). This feature enabled us to probe the metal-binding properties of RUBCa using Tb3⫹-FRET. As shown in Fig. 5A, a significant enhancement of Tb3⫹ fluorescence intensity was observed when Tb3⫹ was added to RUBCa, indicating Tb3⫹ binding to the predicted loop. A Kd for Tb3⫹ of 3 ⫾ 1 M was obtained (Fig. 5A, inset). In comparison with the 355 nm for free Trp, the emission maximum of the purified RUBCa blue-shifted to 339 nm, suggesting that at least some Trp in RUBCa was partly shielded from the solvent, though not fully buried inside the hydrophobic core (Fig. 5B). The shoulder at 355 nm suggests that some Trp residues were exposed to the solvent. No Ca2⫹-induced emission peak position change was observed. However, the addition of Ca2⫹ led to the decrease of the emission intensity, indicating local changes of the chemical environment around aromatic residues (Fig. 5B). By monitoring this intensity change, a dissociation constant of 316 ⫾ 4 M was obtained (Fig. 5B, inset), which was in agreement with the Ca2⫹-binding affinity obtained using the grafted protein CD2.RUBCa (Table 1). Far-UV CD was performed to reveal any possible changes of secondary structure of RUBCa induced by Ca2⫹. As seen in Fig. 6A, the spectra of both EGTA-treated and Ca2⫹-loaded RUBCa had two troughs at 222 and 206 nm, indicating the existence of ␣-helical secondary structures. Using the program
DICHROWEB (67), the best fit of the CD spectrum of RUBCa indicated that 17.8% was ␣-helix and 19.5% was -sheet, whereas the remainder was random coil. In excess Ca2⫹, the CD signal of RUBCa at 208 nm and 222 nm was 5% more negative than that of RUBCa in 1 mM EGTA, and the DICHROWEB-predicted ␣-helix and -sheet contents were 23.8% and 14.9%, respectively. Thus, the observed gain in negative ellipticity could be attributed to the formation of a higher degree of ␣-helical content induced by Ca2⫹ binding. The anionic amphiphile ANS was further used as a hydrophobic probe to examine the conformational properties of RUBCa. As shown in Fig. 6B, upon addition of RUBCa the emission peak of ANS fluorescence blue-shifted from 510 nm to 490 nm and the maximal emission intensity increased by 30%, suggesting that part of the hydrophobic regions of the purified RUBCa were exposed to the solvent and thus were
TABLE 1. The metal-binding affinities of the engineered protein CD2.RUBCa and the minimal metal-binding domain RUBCa Protein
Metal ion
Kd (M)
CD2.RUBCa
Tb Ca2⫹b La3⫹c
47 ⫾ 4 214 ⫾ 73 14 ⫾ 7
RUBCa
Tb3⫹a Ca2⫹d
3⫾1 316 ⫾ 4
a
3⫹a
Results were obtained from the Tb3⫹-FRET assay. Results were obtained from the Ca2⫹ competition assay. c Results were obtained from 1D 1H NMR with La3⫹ titration. d Results were obtained from metal titration monitored by intrinsic Trp fluorescence. b
7524
ZHOU ET AL.
J. VIROL.
FIG. 6. Ca2⫹-induced conformational changes and thermal unfolding of the putative Ca2⫹-binding domain RUBCa. (A) Far-UV CD spectra of RUBCa with 1 mM EGTA (open circle) or 1 mM Ca2⫹ (closed circle) in 10 mM Tris-HCl, 100 mM KCl. The inset shows normalized CD signal at 222 nm plotted as a function of increasing temperature (5 to 90°C) in the presence of 1 mM EGTA (open circle) or 1 mM Ca2⫹ (closed circle). (B) Fluorescence emission spectra of 40 M ANS (open square) and ANS-RUBCa complex with 1 mM EGTA (open circle) or 1 mM Ca2⫹ (closed circle). The excitation wavelength was set at 390 nm. The buffer consisted of 10 mM Tris-HCl, 10 mM KCl (pH 7.4).
accessible to ANS. The addition of excess Ca2⫹ did not cause significant conformational changes in these hydrophobic regions, considering the overlapping emission fluorescence spectra of RUBCa in the presence of 1 mM EGTA or 1 mM Ca2⫹. Taken together, these data indicated that the binding of Ca2⫹ induced a local conformational change, whereas the secondary structure and hydrophobic surface were not significantly altered. In order to gain more insight into the possible role of Ca2⫹ binding, thermal unfolding was carried out by monitoring the CD signal change at 222 nm as a function of temperature under Ca2⫹-depleted or Ca2⫹-loaded conditions. With increasing temperature from 5 to 90°C, RUBCa gradually underwent thermal denaturation, leading to the decrease of CD signals. Compared with 1 mM EGTA, the melting temperature (Tm) of RUBCa with Ca2⫹ increased from 37.7°C ⫾ 0.8°C to 41.8°C ⫾ 0.4°C (Fig. 6A). The observed increase of melting temperature suggested that the binding of Ca2⫹ stabilized the overall structure of RUBCa under physiological conditions (100 mM KCl). Mutating the potential Ca2ⴙ-binding ligands in infectious cDNA clones and replicons. To determine if the sequence that could potentially bind Ca2⫹ is of physiological significance, the D5A/D12A double mutant shown to abrogate binding in CD2.RUBCa was introduced into the Robo502 infectious cDNA clone of RUB (the mutant was termed Robo502AA). Following transfection, the virus titer obtained from Robo502AA was ⬃20-fold lower than that from the Robo502 parent (Fig. 7A), indicating that these residues with their potential Ca2⫹ binding were necessary for optimal RUB replication. Following two subsequent passages of the transfected culture fluids, the Robo502AA titer rose to within threefold of that of Robo502 (Fig. 7A and B). When four individual plaques from the first passage of Robo502AA were amplified and the sequence of the NS protease determined, three were found to
have reverted to the wild-type (wt) sequence at both sites, while the fourth had reverted to the wt sequence at one site (Fig. 7C). Thus, the increase in titer was due to reversion of the mutant sequences. To prevent the accumulation of revertants, the AA mutation was also introduced into the replicon, RUBrep/GFP, in which the SP-ORF is replaced with the GFP reporter gene. Following transfection, RUBrep/GFP replicates and expresses GFP but does not spread from cell to cell. As shown in Fig. 8A, RNA synthesis by RUBrep-AA/GFP was delayed by ⬃1 day postinfection in comparison with the wt replicon. Concomitantly, GFP expression was similarly delayed (data not shown). We also assayed P200 NSP precursor cleavage in cells transfected with a RUBrep-HA/GFP derivative, replicons expressing a P150 tagged with the HA epitope (63). These experiments were done at 6 h posttransfection, when translation from the input transcripts is detectable but replication has not yet started. As shown in Fig. 8B, at 35°C P200 to P150 cleavage was efficient in wt RUBrep-HA-transfected cells but only ⬃50% efficient in RUBrepAA-HA/GFP-transfected cells. RUBrep-NSⴱ-HA/GFP, a construct with a C1152S substitution at the catalytic site unable to undertake cleavage, served as an uncleaved control. Given our biophysical measurements on RUBCa that indicate that Ca2⫹ binding increased the Tm of RUBCa from 37.7oC to 41.8°C (Fig. 6A), we also did the experiment in cells transfected at 39°C. As shown in Fig. 8B, wt replicon cleavage of the P200 precursor was more efficient than that at 35°C; however, cleavage in AA mutant replicontransfected cells could not be detected, indicating that the mutant protease is temperature sensitive. Thus, not only is Ca2⫹ binding important for NS protease activity, in the absence of the Ca2⫹ binding the protease is rendered temperature sensitive.
Ca2⫹-BINDING DOMAIN IN THE RUB NONSTRUCTURAL PROTEASE
VOL. 81, 2007
7525
FIG. 7. Effects of mutations of the potential Ca2⫹ coordination ligands on RUB replication. Transcripts from the wt infectious cDNA clone, Robo502, or Robo502AA containing the D1210A and D1217A mutations in the Ca2⫹-binding loop were used to transfect Vero cells. Culture fluid from the transfection plate (P0) was harvested on day 7 posttransfection and passaged twice in Vero cells (P1 and P2). The virus titer in the P0, P1, and P2 culture fluids was determined by plaque assay in triplicate (A). Open bar, Robo502; black bar, Robo502AA. (B) Representative plaques at each passage. (C) To check for the generation of revertants in the Robo502AA population, four plaques were picked from terminal plaque assay dilution plates from P1 (left) culture fluid, and after one round of amplification in Vero cells, the sequence of the metal-binding domain in the NSP was determined, as shown in comparison to the wt sequence.
DISCUSSION 2ⴙ
Identification of an EF-hand Ca -binding motif in a viral NSP. In this report, we have predicted and further characterized a Ca2⫹-binding motif in a virus NSP. Most other viral Ca2⫹-binding motifs occur in virion structural proteins, and of these, only the rotavirus VP7 outer capsid protein (DITADPT TAPQTE) (19), the human immunodeficiency virus type 1 transmembrane protein gp41 (16), and the polyomavirus VP1 (DENGVGPLCKGE) (26) contain EF-hand Ca2⫹-binding motifs. In a canonical EF-hand Ca2⫹-binding motif, the coor-
dination geometry in the loop is formed by oxygen atoms from the side chain carboxyl or hydroxyl groups (loop positions 1, 3, 5, and 12), the main chain carbonyls (position 7), and a bridged water (position 9), and the flanking sequences around the loop preferentially form a helical secondary structure. The negatively charged residue pairs at the Z axis (loop positions 5 and 12) are essential for the binding of cations (50). Examination of the primary sequence of the motifs in the above-mentioned virion structural proteins reveals that none of these are genuine canonical EF-hand Ca2⫹-binding moieties due to devia-
7526
ZHOU ET AL.
FIG. 8. Replicon RNA synthesis and P200 cleavage in transfected Vero cells. Vero cells were transfected with transcripts from RUBrep/ GFP or RUBrepAA/GFP containing mutations D5A and D12A (A) or from RUBrep-HA/GFP or RUBrepAA-HA/GFP, which express an HA-epitope-tagged P150 (B). (A) Total cell RNA was extracted 1 to 4 days posttransfection, and replicon plus-strand RNA species (G, genomic; SG, subgenomic) were resolved by Northern blotting following agarose gel electrophoresis. (B) Transfections were performed at 35°C or 39°C. Six hours posttransfection, cells were lysed and the P200 precursor and P150 product were resolved by Western blotting probed with anti-HA antibodies following SDS-PAGE (the other product, P90, does not appear because it does not contain the HA epitope). Cells transfected with a replicon containing a C1152S catalytic site mutation (RUBrep-NSⴱ-HA/GFP) that cannot mediate P200 cleavage served as an uncleaved control.
tions in flanking sequence structural requirements (HIV gp41), the length of the loop (13 residues instead of 12 in rotavirus VP7), and the coordinating residues (V instead of D/E/N/Q/ T/S at loop position 5 in polyomavirus VP1). Thus, the Ca2⫹binding motif in the RUB NS protease is the first to be identified with a canonical EF-hand structure. Interestingly, using our prediction method we also detected a putative EF-hand Ca2⫹-binding motif close to the methyltransferase domain in nonstructural protein nsP1 that is conserved among the alphaviruses, the other Togavirus genus (unpublished data). Because of difficulty in expressing the entire NS protease, the metal-binding properties of this Ca2⫹-binding motif were probed by grafting the loop into the scaffold protein CD2.D1 and by expressing the entire motif within the native minimal
J. VIROL.
metal-binding domain of the NS protease. Table 1 provides a summary of the metal-binding affinities of the Ca2⫹-binding motif in both the engineered CD2.RUBCa and the minimal metal-binding domain, RUBCa. It is worth noting that numerous studies of EF-hand calcium-binding motifs have been made at the peptide or subdomain level (17, 58, 68). The Ca2⫹-binding affinity of the RUBCa loop grafted into CD2 (214 M) is weaker than the grafted EF-loop I from calmodulin (34 M) (69), likely due to the presence of an Asp at the bidentate position 12 of the loop rather than a Glu, which ensures a high Ca2⫹-binding affinity due to its larger side chain and stronger interaction with Ca2⫹. Replacement of Glu at position 12 in CaM with Lys or Gln reduces Ca2⫹-binding affinity by 10- to 100-fold (2, 40, 68, 74), while the replacement of Asp with Glu at position 12 in the regulatory light chain of myosin leads to a 15-fold increase in binding affinity (3). The Ca2⫹-binding motif in the RUB NS protease also exhibits selectivity for Ca2⫹ and La3⫹ in the presence of excess Mg2⫹ and K⫹, similar to paired EF-hand Ca2⫹ motifs in CaBPs such as calmodulin and troponin C (35). Ca2ⴙ-induced local conformational change in the RUB NS protease and its potential stabilizing role. With the inability to express and purify the entire NS protease, conformational studies were done using the RUBCa subdomain which contained both catalytic residues, the putative Zn2⫹-binding ligands, and the EF-hand Ca2⫹-binding motif. Boundaries of the RUBCa subdomain were chosen near unstructured or loop regions according to our model (Fig. 1C), and within this model the RUBCa subdomain had a well-folded, globular structure. We found that the RUBCa subdomain bound stoichiometric amounts of Ca2⫹ and underwent metal-induced local changes in conformation (Fig. 5 to 6). Ca2⫹ binding to RUBCa resulted in a modest decrease in its intrinsic Trp fluorescence intensity without emission peak position shift. However, the addition of Ca2⫹ did not significantly alter the secondary structure of RUBCa as revealed by far-UV CD. In addition, ANS binding suggested that no hydrophobic surface was perturbed upon Ca2⫹ binding. These results indicate that Ca2⫹ binding does not lead to a global conformational change of RUBCa. In the RUB NS protease model structure, the predicted helix-loop-helix EF-hand Ca2⫹-binding site is located at the exposed surface on the opposite side of the activesite C1152 and H1273 (Fig. 1C), consistent with our finding that Ca2⫹ binding to RUBCa did not induce global conformational changes. In the model, the W1153 residue is 3.5 Å, on average, away from the predicted EF-hand motif (magenta in Fig. 1C). This proximity could lead to perturbation in the chemical environment around W1153 upon metal binding and thus the model is also consistent with our observation of significant enhancement of Tb3⫹ fluorescence intensity in RUBCa due to Tb3⫹-FRET (Fig. 5A). Finally, we found that the binding of Ca2⫹ increased the Tm of RUBCA from 37.7 to 41.8°C, indicating that Ca2⫹ binding serves to stabilize the NS protease. Significantly, we found that a replicon with two mutations in the EF-hand Ca2⫹-binding loop that abrogate metal binding by the CD2.RUBCa chimera expressed a protease that was temperature sensitive at 39°C, confirming this prediction. Such a stabilizing role of Ca2⫹ has been observed in nonviral protease, such as subtilisin (59) and thermolysin (60). It will be of interest to determine if the putative EF-hand Ca2⫹-binding
Ca2⫹-BINDING DOMAIN IN THE RUB NONSTRUCTURAL PROTEASE
VOL. 81, 2007
motif in the MT domain of the alphaviruses plays a similar stabilizing role. It should be pointed out that the effect of Ca2⫹ on the RUB NS protease cannot be studied in an in vitro translation system, because such systems (i) include EGTA to chelate Ca2⫹ to inactivate the micrococcal nuclease employed to remove endogenous mRNA and (ii) are incubated at 30°C, below the 39°C required to observe an effect in vivo. Role of Ca2ⴙ in the replication cycle of RUB. When the double Ca2⫹-binding mutant was introduced into the Robo502 RUB infectious cDNA clone, titers produced were ⬃20-fold lower than those produced by wt Robo502; however, during subsequent passaging, the mutant virus titer rose by over 10fold, and sequencing of isolated virus plaques indicated that this was due to reversion of the mutants to the wt sequence. In a replicon incapable of cell-to-cell spread, the double Ca2⫹binding mutant exhibited delayed RNA synthesis. Using the replicon constructs, we finally showed that cleavage of the P200 nonstructural precursor was impaired at 35°C and temperature sensitive at 39°C. While minus-strand RNA synthesis is catalyzed by the uncleaved precursor, the precursor must be cleaved before plus-strand RNA synthesis can occur (33). Thus, an impaired cleavage is consistent with delayed plusstrand RNA synthesis, as observed with the replicon, and lower titers, as observed with the infectious cDNA clone. However, ongoing (though delayed) RNA synthesis would lead to the generation of revertants, as observed with the infectious cDNA clone. A submillimolar Kd for Ca2⫹ binding (⬃200 to 300 M) was derived from both CD2.RUBCa and RUBCa. In the cell, cytosolic Ca2⫹ concentrations range from 0.1 to 10 M, while the Ca2⫹ concentration in endosomes or lysosomes is as high as 400 to 600 M and is maintained in part by the proton gradient across lysosomal membranes (8, 21). Presumably, in the replication complex the protease resides on the cytosolic side of the lysosomal membrane; however, release of Ca2⫹ from the lysosome could result in local Ca2⫹ concentrations high enough for binding. For example, in acidifying endosomes the uptake of extracellular Ca2⫹ (in millimolars) is immediately accompanied by rapid release of Ca2⫹ to the cytoplasm, and with an external [Ca2⫹] of 2 mM the endocytosed Ca2⫹ may be sufficient to increase the total cellular Ca2⫹ concentration by 2 M per min (21). It is also possible that association of the viral replicase proteins with late endosomal or lysosomal membranes alters the Ca2⫹ gradient, leading to a localized increase in Ca2⫹ concentration. Nevertheless, given the weak Ca2⫹binding affinity of the Ca2⫹-binding loop, the protease is possibly inactive in the cytosol and must be associated with the late endosomal or lysosomal membrane before encountering Ca2⫹ concentrations sufficient for binding. Thus, Ca2⫹ may play a role in regulating the virus replication cycle by delaying the nonstructural precursor cleavage until it is associated with late endosomal or lysosomal membranes and thus modulating the synthesis of plus strand RNA. A similar situation has been shown to be the case with rotaviruses whose NSP4 glycoprotein mediates an increase in intracellular Ca2⫹ concentration, which modulates transcriptase activity (4, 13, 45, 55). ACKNOWLEDGMENTS We appreciate the critical review by Dan Adams and other members in J.J.Y.’s group. We are thankful to Shunyi Li and Siming Wang for
7527
help in ESI-MS, to Yun Huang and Yanyi Chen for protein expression, and to Suganthi Suppiah for helpful discussion. This work was supported in part by grant R01 AI21389 from the NIAID to T.K.F. and J.J.Y. and in part by the MCB-0092486 (NSF) and GM62999 and GM070555 (NIH) grants for J.J.Y. Y.Z. is a fellow of the Molecular Basis of Disease Area of Focus at Georgia State University. REFERENCES 1. Allen, M. D., A. M. Buckle, S. C. Cordell, J. Lowe, and M. Bycroft. 2003. The crystal structure of AF1521 a protein from Archaeoglobus fulgidus with homology to the non-histone domain of macroH2A. J. Mol. Biol. 330:503– 511. 2. Beckingham, K. 1991. Use of site-directed mutations in the individual Ca2(⫹)-binding sites of calmodulin to examine Ca2(⫹)-induced conformational changes. J. Biol. Chem. 266:6027–6030. 3. Blumenschein, T. M., and F. C. Reinach. 2000. Analysis of affinity and specificity in an EF-hand site using double mutant cycles. Biochemistry 39:3603–3610. 4. Bowman, G. D., I. M. Nodelman, O. Levy, S. L. Lin, P. Tian, T. J. Zamb, S. A. Udem, B. Venkataraghavan, and C. E. Schutt. 2000. Crystal structure of the oligomerization domain of NSP4 from rotavirus reveals a core metal-binding site. J. Mol. Biol. 304:861–871. 5. Burmeister, W. P., S. Cusack, and R. W. Ruigrok. 1994. Calcium is needed for the thermostability of influenza B virus neuraminidase. J. Gen. Virol. 75:381–388. 6. Chen, J. P., J. H. Strauss, E. G. Strauss, and T. K. Frey. 1996. Characterization of the rubella virus nonstructural protease domain and its cleavage site. J. Virol. 70:4707–4713. 7. Chong, A. K., M. S. Pegg, and M. von Itzstein. 1991. Influenza virus sialidase: effect of calcium on steady-state kinetic parameters. Biochim. Biophys. Acta 1077:65–71. 8. Christensen, K. A., J. T. Myers, and J. A. Swanson. 2002. pH-dependent regulation of lysosomal calcium in macrophages. J. Cell Sci. 115:599–607. 9. Cox, J. A., and E. A. Stein. 1981. Characterization of a new sarcoplasmic calcium-binding protein with magnesium-induced cooperativity in the binding of calcium. Biochemistry 20:5430–5436. 10. Cuff, J. A., M. E. Clamp, A. S. Siddiqui, M. Finlay, and G. J. Barton. 1998. JPred: a consensus secondary structure prediction server. Bioinformatics 14:892–893. 11. Deng, H., G. Chen, W. Yang, and J. J. Yang. 2006. Predicting calciumbinding sites in proteins—a graph theory and geometry approach. Proteins 64:34–42. 12. Dimitrov, D. S., C. C. Broder, E. A. Berger, and R. Blumenthal. 1993. Calcium ions are required for cell fusion mediated by the CD4-human immunodeficiency virus type 1 envelope glycoprotein interaction. J. Virol. 67:1647–1652. 13. Dormitzer, P. R., H. B. Greenberg, and S. C. Harrison. 2000. Purified recombinant rotavirus VP7 forms soluble, calcium-dependent trimers. Virology 277:420–428. 14. Drake, S. K., K. L. Lee, and J. J. Falke. 1996. Tuning the equilibrium ion affinity and selectivity of the EF-hand calcium binding motif: substitutions at the gateway position. Biochemistry 35:6697–6705. 15. Driscoll, P. C., J. G. Cyster, I. D. Campbell, and A. F. Williams. 1991. Structure of domain 1 of rat T lymphocyte CD2 antigen. Nature 353:762– 765. 16. Ebenbichler, C. F., H. Stoiber, R. Schneider, J. R. Patsch, and M. P. Dierich. 1996. The human immunodeficiency virus type 1 transmembrane gp41 protein is a calcium-binding protein and interacts with the putative secondreceptor molecules in a calcium-dependent manner. J. Virol. 70:1723–1728. 17. Finn, B. E., J. Evenas, T. Drakenberg, J. P. Waltho, E. Thulin, and S. Forsen. 1995. Calcium-induced structural changes and domain autonomy in calmodulin. Nat. Struct. Biol. 2:777–783. 18. Frey, T. K. 1994. Molecular biology of rubella virus. Adv. Virus Res. 44:69– 160. 19. Gajardo, R., P. Vende, D. Poncet, and J. Cohen. 1997. Two proline residues are essential in the calcium-binding activity of rotavirus VP7 outer capsid protein. J. Virol. 71:2211–2216. 20. Garbutt, M., L. M. Law, H. Chan, and T. C. Hobman. 1999. Role of rubella virus glycoprotein domains in assembly of virus-like particles. J. Virol. 73: 3524–3533. 21. Gerasimenko, J. V., A. V. Tepikin, O. H. Petersen, and O. V. Gerasimenko. 1998. Calcium uptake via endocytosis with rapid release from acidifying endosomes. Curr. Biol. 8:1335–1338. 22. Gill, S. C., and P. H. von Hippel. 1989. Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182:319–326. 23. Gorbalenya, A. E., E. V. Koonin, and M. M. Lai. 1991. Putative papainrelated thiol proteases of positive-strand RNA viruses. Identification of rubiand aphthovirus proteases and delineation of a novel conserved domain associated with proteases of rubi-, alpha- and coronaviruses. FEBS Lett. 288:201–205.
7528
ZHOU ET AL.
24. Guarne, A., B. Hampoelz, W. Glaser, X. Carpena, J. Tormo, I. Fita, and T. Skern. 2000. Structural and biochemical features distinguish the foot-andmouth disease virus leader proteinase from other papain-like enzymes. J. Mol. Biol. 302:1227–1240. 25. Guarne, A., J. Tormo, R. Kirchweger, D. Pfistermueller, I. Fita, and T. Skern. 1998. Structure of the foot-and-mouth disease virus leader protease: a papain-like fold adapted for self-processing and eIF4G recognition. EMBO J. 17:7469–7479. 26. Haynes, J. I., Jr., D. Chang, and R. A. Consigli. 1993. Mutations in the putative calcium-binding domain of polyomavirus VP1 affect capsid assembly. J. Virol. 67:2486–2495. 27. Ikura, M., and J. B. Ames. 2006. Genetic polymorphism and protein conformational plasticity in the calmodulin superfamily: two ways to promote multifunctionality. Proc. Natl. Acad. Sci. USA 103:1159–1164. 28. Kawasaki, H., S. Nakayama, and R. H. Kretsinger. 1998. Classification and evolution of EF-hand proteins. Biometals 11:277–295. 29. Kretsinger, R. H., and C. E. Nockolds. 1973. Carp muscle calcium-binding protein. II. Structure determination and general description. J. Biol. Chem. 248:3313–3326. 30. Kujala, P., T. Ahola, N. Ehsani, P. Auvinen, H. Vihinen, and L. Kaariainen. 1999. Intracellular distribution of rubella virus nonstructural protein P150. J. Virol. 73:7805–7811. 31. Lee, H. W., W. Yang, Y. Ye, Z. R. Liu, J. Glushka, and J. J. Yang. 2002. Isolated EF-loop III of calmodulin in a scaffold protein remains unpaired in solution using pulsed-field-gradient NMR spectroscopy. Biochim. Biophys. Acta 1598:80–87. 32. Levine, B. A., D. C. Dalgarno, M. P. Esnouf, R. E. Klevit, G. M. Scott, and R. J. Williams. 1983. The mobility of calcium-trigger proteins and its function. Ciba Found. Symp. 93:72–97. 33. Liang, Y., and S. Gillam. 2001. Rubella virus RNA replication is cis-preferential and synthesis of negative- and positive-strand RNAs is regulated by the processing of nonstructural protein. Virology 282:307–319. 34. Liang, Y., J. Yao, and S. Gillam. 2000. Rubella virus nonstructural protein protease domains involved in trans- and cis-cleavage activities. J. Virol. 74:5412–5423. 35. Linse, S., and S. Forsen. 1995. Determinants that govern high-affinity calcium binding. Adv. Second Messenger Phosphoprotein Res. 30:89–151. 36. Liu, X., S. L. Ropp, R. J. Jackson, and T. K. Frey. 1998. The rubella virus nonstructural protease requires divalent cations for activity and functions in trans. J. Virol. 72:4463–4466. 37. Liu, X., J. Yang, A. M. Ghazi, and T. K. Frey. 2000. Characterization of the zinc binding activity of the rubella virus nonstructural protease. J. Virol. 74:5949–5956. 38. Magliano, D., J. A. Marshall, D. S. Bowden, N. Vardaxis, J. Meanger, and J. Y. Lee. 1998. Rubella virus replication complexes are virus-modified lysosomes. Virology 240:57–63. 39. Maniccia, A. W., W. Yang, S. Y. Li, J. A. Johnson, and J. J. Yang. 2006. Using protein design to dissect the effect of charged residues on metal binding and protein stability. Biochemistry 45:5848–5856. 40. Maune, J. F., K. Beckingham, S. R. Martin, and P. M. Bayley. 1992. Circular dichroism studies on calcium binding to two series of Ca2⫹ binding site mutants of Drosophila melanogaster calmodulin. Biochemistry 31:7779–7786. 41. McGuffin, L. J., K. Bryson, and D. T. Jones. 2000. The PSIPRED protein structure prediction server. Bioinformatics 16:404–405. 42. Michiels, J., C. Xi, J. Verhaert, and J. Vanderleyden. 2002. The functions of Ca(2⫹) in bacteria: a role for EF-hand proteins? Trends Microbiol. 10:87–93. 43. Novoa, R. R., G. Calderita, R. Arranz, J. Fontana, H. Granzow, and C. Risco. 2005. Virus factories: associations of cell organelles for viral replication and morphogenesis. Biol. Cell 97:147–172. 44. Ogawa, Y., and M. Tanokura. 1984. Calcium binding to calmodulin: effects of ionic strength, Mg2⫹, pH and temperature. J. Biochem. (Tokyo) 95:19–28. 45. Pando, V., P. Isa, C. F. Arias, and S. Lopez. 2002. Influence of calcium on the early steps of rotavirus infection. Virology 295:190–200. 46. Perret, C., N. Lomri, and M. Thomasset. 1990. Evolution of the “EF-hand” family of calcium-binding proteins. Adv. Exp. Med. Biol. 269:17–20. 47. Persechini, A., N. D. Moncrief, and R. H. Kretsinger. 1989. The EF-hand family of calcium-modulated proteins. Trends Neurosci. 12:462–467. 48. Pugachev, K. V., E. S. Abernathy, and T. K. Frey. 1997. Improvement of the specific infectivity of the rubella virus (RUB) infectious clone: determinants of cytopathogenicity induced by RUB map to the nonstructural proteins. J. Virol. 71:562–568. 49. Pugachev, K. V., W. P. Tzeng, and T. K. Frey. 2000. Development of a rubella
J. VIROL.
50. 51. 52. 53. 54. 55. 56. 57. 58.
59.
60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70.
71. 72. 73. 74.
virus vaccine expression vector: use of a picornavirus internal ribosome entry site increases stability of expression. J. Virol. 74:10811–10815. Reid, R. E., and R. M. Procyshyn. 1995. Engineering magnesium selectivity in the helix-loop-helix calcium-binding motif. Arch. Biochem. Biophys. 323: 115–119. Rigden, D. J., and M. Y. Galperin. 2004. The DxDxDG motif for calcium binding: multiple structural contexts and implications for evolution. J. Mol. Biol. 343:971–984. Rigden, D. J., M. J. Jedrzejas, O. V. Moroz, and M. Y. Galperin. 2003. Structural diversity of calcium-binding proteins in bacteria: single-handed EF-hands? Trends Microbiol. 11:295–297. Roberts, P. J., and G. J. Belsham. 1995. Identification of critical amino acids within the foot-and-mouth disease virus leader protein, a cysteine protease. Virology 213:140–146. Rost, B., G. Yachdav, and J. Liu. 2004. The PredictProtein server. Nucleic Acids Res. 32:W321—W326. Ruiz, M. C., J. Cohen, and F. Michelangeli. 2000. Role of Ca2⫹ in the replication and pathogenesis of rotavirus and other viral infections. Cell Calcium 28:137–149. Schroder, B., C. Schlumbohm, R. Kaune, and G. Breves. 1996. Role of calbindin-D9k in buffering cytosolic free Ca2⫹ ions in pig duodenal enterocytes. J. Physiol. 492:715–722. Schwede, T., J. Kopp, N. Guex, and M. C. Peitsch. 2003. SWISS-MODEL: an automated protein homology-modeling server. Nucleic Acids Res. 31:3381– 3385. Sharma, Y., S. Chandani, M. B. Sukhaswami, L. Uma, D. Balasubramanian, and T. Fairwell. 1997. Modified helix-loop-helix motifs of calmodulin–the influence of the exchange of helical regions on calcium-binding affinity. Eur. J. Biochem. 243:42–48. Smith, C. A., H. S. Toogood, H. M. Baker, R. M. Daniel, and E. N. Baker. 1999. Calcium-mediated thermostability in the subtilisin superfamily: the crystal structure of Bacillus Ak. 1 protease at 1.8 A resolution. J. Mol. Biol. 294:1027–1040. Tajima, M., I. Urabe, K. Yutani, and H. Okada. 1976. Role of calcium ions in the thermostability of thermolysin and Bacillus subtilis var. amylosacchariticus neutral protease. Eur. J. Biochem. 64:243–247. Tzeng, W. P., and T. K. Frey. 2003. Complementation of a deletion in the rubella virus p150 nonstructural protein by the viral capsid protein. J. Virol. 77:9502–9510. Tzeng, W. P., and T. K. Frey. 2002. Mapping the rubella virus subgenomic promoter. J. Virol. 76:3189–3201. Tzeng, W. P., J. D. Matthews, and T. K. Frey. 2006. Analysis of rubella virus capsid protein-mediated enhancement of replicon replication and mutant rescue. J. Virol. 80:3966–3974. Wang, C. L., R. R. Aquaron, P. C. Leavis, and J. Gergely. 1982. Metalbinding properties of calmodulin. Eur. J. Biochem. 124:7–12. Wang, C. Y., G. Dominguez, and T. K. Frey. 1994. Construction of rubella virus genome-length cDNA clones and synthesis of infectious RNA transcripts. J. Virol. 68:3550–3557. Werber, M. M., and A. Oplatka. 1974. Physico-chemical studies on the light chains of myosin. 3. Evidence for a regulatory role of a rabbit myosin light chain. Biochem. Biophys. Res. Commun. 57:823–830. Whitmore, L., and B. A. Wallace. 2004. DICHROWEB, an online server for protein secondary structure analyses from circular dichroism spectroscopic data. Nucleic Acids Res. 32:W668–W673. Yang, J. J., A. Gawthrop, and Y. Ye. 2003. Obtaining site-specific calciumbinding affinities of calmodulin. Protein Pept. Lett. 10:331–345. Ye, Y., H. W. Lee, W. Yang, S. Shealy, and J. J. Yang. 2005. Probing site-specific calmodulin calcium and lanthanide affinity by grafting. J Am. Chem. Soc. 127:3743–3750. Ye, Y., H. W. Lee, W. Yang, S. J. Shealy, A. L. Wilkins, Z. R. Liu, I. Torshin, R. Harrison, R. Wohlhueter, and J. J. Yang. 2001. Metal binding affinity and structural properties of an isolated EF-loop in a scaffold protein. Protein Eng. 14:1001–1013. Ye, Y., S. Shealy, H. W. Lee, I. Torshin, R. Harrison, and J. J. Yang. 2003. A grafting approach to obtain site-specific metal-binding properties of EFhand proteins. Protein Eng. 16:429–434. Zhou, Y., H. Ushijima, and T. K. Frey. 2007. Genomic analysis of diverse rubella virus genotypes. J. Gen. Virol. 88:932–941. Zhou, Y., W. Yang, M. Kirberger, H. W. Lee, G. Ayalasomayajula, and J. J. Yang. 2006. Prediction of EF-hand calcium-binding proteins and analysis of bacterial EF-hand proteins. Proteins 65:643–655. Zhu, T., K. Beckingham, and M. Ikebe. 1998. High affinity Ca2⫹-binding sites of calmodulin are critical for the regulation of myosin Ibeta motor function. J. Biol. Chem. 273:20481–20486.