HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Data Supplement Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Occhino, M. Articles by Ciccone, E. Search for Related Content PubMed PubMed Citation Articles by Occhino, M. Articles by Ciccone, E. Pubmed/NCBI databases Substance via MeSH

Dissecting the Structural Determinants of the Interaction between the Human Cytomegalovirus UL18 Protein and the CD85j Immune Receptor¹

Marzia Occhino^2,*, Fabio Ghiotto^2,3,*, Simonetta Soro, Mimosa Mortarino, Stefania Bosi, Massimo Maffei^*, Silvia Bruno^*, Marco Nardini^¶, Mariangela Figini, Anna Tramontano^, and Ermanno Ciccone^*

^* Department of Experimental Medicine, University of Genoa, Genoa; Department of Biochemical Sciences and Istituto Pasteur Fondazione Cenci Bolognetti, University "La Sapienza", Rome; Department of Experimental Oncology, Molecular Therapy Unit, Istituto Nazionale Tumori, Milan; and ^¶ Department of Biomolecular Sciences and Biotechnology and Consiglio Nazionale delle Ricerche–Istituto Nazionale per la Fisica della Materia, University of Milan, Milan, Italy

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
UL18 is a glycoprotein encoded by the human cytomegalovirus genome and is thought to play a pivotal role during human cytomegalovirus infection, although its exact function is still a matter of debate. UL18 shares structural similarity with MHC class I and binds the receptor CD85j on immune cells. Besides UL18, CD85j binds MHC class I molecules. The binding properties of CD85j to MHC class I molecules have been thoroughly studied. Conversely, very little information is available on the CD85j/UL18 complex, namely that UL18 binds CD85j through its 3 domain with an affinity that is 1000-fold higher than the MHC class I affinity for CD85j. Deeper knowledge of features of the UL18/CD85j complex would help to disclose the function of UL18 when it binds to CD85j. In this study we first demonstrated that the UL183 domain is not sufficient per se for binding and that β₂-microglobulin is necessary for UL18–CD85j interaction. We then dissected structural determinants of binding UL18 to CD85j. To this end, we constructed a three-dimensional model of the complex. The model was used to design mutants in selected regions of the putative interaction interface, the effects of which were measured on binding. Six regions in both the 2 and 3 domains and specific amino acids within them were identified that are potentially involved in the UL18–CD85j interaction. The higher affinity of UL18 to CD85j, compared with MHC class I, seems to be due not to additional interaction regions but to an overall better fit of the two molecules.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
The UL18 is a protein encoded by the human cytomegalovirus (HCMV)⁴genome (1). It is a late HCMV Ag, with its transcription occurring from 54 to at least 120 h postinfection (2), and it is not essential for HCMV replication in vitro (3). It shares structural similarity with MHC class I molecules (1). In particular, the extracellular region is formed by three domains (1, 2, and 3) with a sequence identity of 21, 20, and 22% with their class I counterparts (1), respectively. Moreover, its extracellular region associates with β₂-microglobulin (β₂m) (4) and binds peptides derived from cytoplasmic proteins (5). Similarly to class I molecules, the UL18 2 and 3 domains each contains two cysteines characteristic of the MHC Ag-recognition domain fold and of the Ig-like fold, respectively. In contrast, the Ig-like loop of the 2 domain is 10 amino acids longer than the corresponding MHC class I loop, and the CD loop of the Ig-like 3 domain contains three additional cysteines. As opposed to MHC class I molecules, which have only one to three N-glycosylation sites, the UL18 molecule bears 13 potential glycosylation sites (1), resulting in the expression of highly glycosylated forms (6). The only molecule known so far to bind UL18 is the transmembrane protein CD85j (previously reported as ILT2 or LIR-1) (7).

CD85j is a transmembrane receptor expressed on the surface of subsets of NK and T cells and of all B cells and monocytes (7, 8). It is formed by four Ig-like domains, a transmembrane region, and a cytoplasmic tail that contains four ITIM-like motifs. Cross-linking of the CD85j receptor on the surface of T cells leads to a decrease of the CD3 complex phosphorylation (9), a decrease in activatory cytokines production (10), and to an increase of cytokines that down-regulate the immune response (10). Besides UL18, CD85j binds MHC class I molecules (11) with a broad specificity. The binding properties of CD85j to MHC class I molecules have been thoroughly studied. CD85j binds MHC class I molecules primarily through the interaction of its distal Ig-like domain with class I 3 domain and β₂m (11, 12, 13) and with an affinity in the micromolar range. Most of the interactions involve residues of the β₂m domain, which might explain the broad specificity of this interaction (13).

Much less information is available about the complex between CD85j and UL18. It is known that, similar to MHC class I, UL18 binds CD85j through its 3 domain (11). Notably, UL18 binds CD85j with an affinity that is in the nanomolar range, 1000-fold higher than the MHC class I affinity for CD85j (11). Recent studies have shown that UL18-soluble molecules derived from clinical isolates all bind CD85j, although with different affinities (14, 15).

UL18 expression on cell surfaces of HCMV-infected cells is described both as an escaping mechanism from NK lysis (16, 17) and as a mechanism that leads to killing infected cells by NK cells (18) or by T lymphocytes through a non-MHC-restricted pathway (19). Furthermore, UL18 has been recently described to have an activatory role on T cells (20). Whatever the mechanism, UL18 certainly plays a pivotal role in the immune response to HCMV infection, and it is therefore important to gain an understanding of the molecular properties of its complex with CD85j and, possibly, of the high affinity of the interaction.

Our study investigates features of the UL18 molecular structure(s) that may account for its high affinity for CD85j. We analyze UL18 binding to CD85j with a mutagenesis approach and propose a model for the structure of its extracellular region.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Cell lines, transfections, and supernatant production

HeLa and Jurkat cell lines were maintained in RPMI 1640 medium with GlutaMAX (Invitrogen); HEK293T, FO-1, and Sp2/0 cell lines were maintained in DMEM medium with GlutaMAX. Both media were supplemented with penicillin, streptomycin, and 10% FCS (Invitrogen).

HeLa and FO-1 transient transfections were conducted using Lipofectamine 2000 reagent (Invitrogen) following the manufacturer’s instructions.

HEK293T transient transfections were performed as follows: HEK293T cells were plated in petri dishes and transfected by CaPO₄ coprecipitation method. Briefly, 1.5 x 10⁶ cells were transfected with 10 µg of DNA, incubated for 8 h at 37°C, washed twice with PBS, and then added with fresh OptiMEM medium (Invitrogen) without addition of FCS. Supernatants were collected after 72 h and concentrated 100x with Amicon Ultra centrifugal filter units (Millipore) (30,000 molecular weight cutoff).

The HEK293T cell line was stably transfected with vectors carrying the UL18Fc-wt molecule, the different UL18Fc mutants, and the HLA-A2Fc molecule. Transfections were performed using the Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s instructions. The transfectants were selected in DMEM medium added with 10% FBS and 500 µg/ml Geneticin (Invitrogen). For protein production, each stable cell line was cultured extensively in DMEM medium, washed with PBS, and then incubated in OptiMEM medium (Invitrogen) without FCS for 72 h. One hundred petri dishes were plated for each transfectant to produce a volume of 500 ml of supernatant.

The Jurkat cell line and murine cell line Sp2/0 were transfected with the CD85jFc construct. Transfection was performed with a Multiporator (Eppendorf, Milan, Italy) following the manufacturer’s instructions. Transfected cells were selected with 500 and 300 µg/ml Geneticin (Invitrogen), respectively.

Constructs and mutants

UL183Fc construct. A soluble molecule formed by the CD33 leader peptide, the UL18 3 domain, and the CH₂ and CH₃ domains of the human IgG1 Ig was constructed (UL183Fc). Using PCR, the leader sequence of the CD33 protein was joined to the 3 domain of UL18 of the AD169 HCMV strain. The product was cloned with the TOPO TA cloning kit (Invitrogen), sequenced, and then subcloned into the spIg2.0 vector that carries the CH₂ and CH₃ domains of human IgG1.

UL18Fc-wt construct. A covalently bound molecule composed by a leader sequence, a peptide, β₂m, the extracellular domain of UL18 and the hinge, CH₂ and CH₃ domains of the human IgG1 Ig was constructed as described previously (21). Briefly, with a PCR strategy, the β₂m leader sequence was joined to the actin peptide ALPHAILRL. These two sequences were linked to the β₂m coding sequence with the intervening spacer (GGGS)₃ and then, with another spacer (GGGGS)₃, linked to the 1, 2, and 3 domains of the UL18 molecule derived from strain AD169. The described recombinant molecule, obtained by repeated PCR steps, was cloned with the TOPO TA cloning kit and sequenced. The construct coding for the correct amino acid sequence was digested with the restriction enzymes HindIII and NotI and subcloned into the phuPSIg2.0 expression vector, which contains a genomic portion (intron, hinge, intron, CH₂, intron, CH₃) of the human IgG1 Ig.

HLA-A2Fc construct. HLA-A2 was constructed by subcloning the HLA-A2 construct (kindly provided by T. Greten, Medizinische Hochschule Hannover, Germany) into the phuPSIg2.0 expression vector.

UL18Fc-Di cassette mutants. Cassette mutants were obtained with a two-step PCR strategy. Primers annealing to the UL18 sequence, but carrying a flanking region coding for the HLA-A2 sequence chosen to replace the corresponding UL18 sequence, were used. The product was cloned with the TOPO TA cloning kit and sequenced. A clone with the expected sequence was subcloned into the phuPSIg2.0 expression vector.

UL18Fc-Mi mutants. Single and multiple amino acid substitutions were obtained using the QuickChange II site-directed mutagenesis kit and the QuickChange multi-site-directed mutagenesis kit (Stratagene), respectively, as described by the manufacturer.

CD85jFc constructs. The extracellular portion of the CD85j molecule was amplified by PCR, cloned with the TOPO TA cloning kit, and sequenced. A clone with the expected sequence was subcloned into the phuPSIg2.0 expression vector.

β₂m construct. The coding region of the β₂m was amplified by PCR, cloned with the TOPO TA cloning kit, and sequenced. A clone with the expected sequence was subcloned into pIREShygro expression vector (Clontech Laboratories, Mountain View, CA).

HiFi Platinum Taq (Invitrogen) was used for all PCR. DNA ligation kit 2.1 (Takara Shuzo) was used in all ligation reactions. For all constructs, positive clones were analyzed by sequencing to exclude mutations and to verify the frame maintenance. Big Dye Terminator v1.1 (Applied Biosystems) was used for all sequences that were analyzed using an ABI PRISM 310 Genetic Analyzer (Applied Biosystems).

Genomic DNA extraction and UL18 amplification

Genomic DNA was extracted from HCMV seropositive donors. Briefly, 3 ml of peripheral venous blood was diluted with an equal volume of PBS, layered over 3 ml of Lymphoprep (Axis-Shield), and spun at 800 rpm for 20 min. Mononuclear cells were washed with PBS. DNA was extracted with the GeneElute Mammalian Genomic DNA miniprep kit (Sigma-Aldrich) according to the manufacturer’s instructions. Genomic DNA was amplified using the forward primer UL18–5'F2 CGCCATGATGACAATGTGGTG and the reverse primer UL18–3'R2 GCGTCGCGTGAGAAACATGAC. After 50 cycles of amplification with HiFi Platinum Taq the product was run on a 1% agarose gel. Positive samples were purified and sequenced.

Protein affinity purification

Supernatants containing recombinant proteins were incubated with rec-protein A-Sepharose 4B (Zymed Laboratories) overnight at room temperature. Beads were packed in appropriate columns and washed with abundant PBS. Bound proteins were eluted with 0.1 M glycine-HCl (pH 3.5); 10 aliquots were collected and neutralized by addition of 50 µl of 1 M Tris-HCl (pH 8.8). The absorbance at 280 nm was measured and fractions containing the protein (with an A280 > 0.05) were pooled. Pooled fractions were dialyzed against HBSS buffer (10 mM HEPES pH 7.4, 150 mM NaCl, and 3.4 mM EDTA pH 8) to perform subsequent measures with a Biacore 2000 instrument. When necessary, the concentration of purified proteins was normalized by ELISA, as described, comparing the absorbance values of samples with known concentrations of human IgG.

ELISA

The presence of proteins in every supernatant was evaluated by ELISA. A polystyrene Maxisorp ELISA plate (Nunc) was coated overnight at 4°C with 100 µl/well of goat anti-human IgG (Kirkegaard & Perry Laboratories) dissolved at 1 µg/ml in PBS (Invitrogen). The plate was washed three times in PBS containing 0.01% Tween 20 (Sigma-Aldrich) and blocked with 200 µl/well of 10 mg/ml BSA (Sigma-Aldrich) in PBS-Tween. Supernatants from transfected HEK293T (100 µl) were added to each well and incubated for 2 h at room temperature. Serial dilutions of commercial human IgG (Chemicon International) from 0.2 ng/well up to 50 ng/well were used as positive controls. After washing with PBS-Tween, the plate was incubated with HRP-conjugated goat anti-human IgG (Kirkegaard & Perry Laboratories) diluted at 0.5 µg/ml in PBS containing 1% BSA. The substrate for peroxidase, ABTS (Valeant Pharmaceuticals), was added to the plate and the absorbance at 405 nm read on a Titertek Multiskan Plus plate reader (Titertek).

Flow cytometry analysis

Cells were stained at 4°C for 30 min with the primary Ab HP-F1, specific for the CD85j molecule, then washed with cold PBS, and stained with goat anti-mouse PE-conjugated antiserum (Southern Biotechnology Associates). Staining with UL18Fc recombinant proteins was performed as above with a goat anti-human PE-conjugated antiserum (Southern Biotechnology Associates) as secondary reagent. Cells were analyzed with a FACSCalibur (BD Biosciences) using the CellQuest software.

Surface and intracellular immunofluorescence

Cells were grown on glass coverslips and stained in adherence for simultaneous surface and intracellular immunofluorescence. After surface immunofluorescence staining, by treatment with CD85jFc and Alexa 488 (green)-conjugated goat anti-human IgG (Invitrogen) at 4°C, cells were fixed with 3% paraformaldehyde and permeabilized with 0.05% Triton X-100 before being subjected to intracellular immunofluorescence staining at room temperature with CD85jFc followed by Alexa 546 (red)-conjugated goat anti-human IgG (Invitrogen). Fluorescence was imaged on a Leica SP2-AOBS confocal microscope (Leica Microsystems, Heidelberg, Germany).

Binding assays

Binding experiments were performed with a Biacore 2000 equipped with research-grade CM5 sensor chips (Biacore). A standard amine-coupling protocol (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride, normal human serum, and sodium ethanolamine HCl, pH 8.5) was used to immobilize CD85jFc protein. The immobilization was conducted at a concentration of 58 ng/µl using HBSP (10 mM HEPES, 150 mM NaCl, pH 7.4, 0.005% P-20) as the running buffer. Flowcell 1 was used as a reference cell, and 3800 resonance units of CD85jFc was immobilized in flowcell 2. UL18Fc-wt, its mutants, and HLA-A2Fc were injected at a flow rate of 30 µl/min for 3 min. All the sample concentrations were previously normalized by ELISA. Sample dilutions were prepared in HBS buffer, and binding analyses were performed at a concentration of 200 nM.

Structural modeling

The database entry of the UL18 (human herpesvirus 5 strain AD169) sequence is gi 9625703 ref NP_039952.1.

Blast (22), HHPred (23), and M-coffee (24) with default parameters were used for database searches and multiple sequence alignment, respectively. The database used for searching was the nonredundant protein sequence database.

Secondary structure prediction for UL18 was obtained by the PredictProtein server (25); the secondary structure of the template was obtained by the DSSP program (26). Modeler9v1 (27) was used for model building.

Side chains were modeled using the program Scwrl (28) and were successively analyzed using the program Scit (29), which allows the identification of the structural neighbors of each side chain on the basis of interatomic distances. The model of UL18 in complex with the CD85j receptor was built by superimposition of the conserved main chain region of UL18 and the crystal structure of HLA-A2 bound to CD85j (Brookhaven Protein Data Bank (PDB) code: 1p7q; Ref. 13). The structural alignment was performed using the program FATCAT (30) in its rigid body comparison mode.

The model is available at www.caspur.it/PDMDB with identifier PM0074961.

	Results

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Role of β₂m in the binding of UL18 to CD85j and construction of a stable UL18-soluble molecule

UL18 binds CD85j primarily through its 3 domain (11). We first investigated whether the 3 domain of UL18 is sufficient per se for binding to CD85j or whether the presence of β₂m is required. To ascertain this, we constructed a soluble molecule formed by the UL18 3 domain and by the CH₂ and CH₃ domains of the human IgG1 Ig. The expression of the construct inside the endoplasmic reticulum was restored by adding the leader peptide of the CD33 protein at the 5' end of the molecule. Because this molecule only contains the 3 domain of UL18, its folding is not expected to be significantly influenced by β₂m. The resulting protein (UL183Fc) was transiently expressed in HeLa cells, and its presence in the supernatant was verified by ELISA. The supernatant was concentrated and used to stain a Jurkat-transfected cell line stably expressing CD85j on the surface (Jurkat/CD85j). The UL183Fc recombinant protein was able to bind CD85j on the surface of the Jurkat/CD85j cell line (Fig. 1A). Next, to investigate the role of β₂m in the UL18–CD85j interaction, we transiently transfected UL183Fc both in the wild-type, β₂m-deficient cell line FO-1 and in a FO-1 cell line stably transfected with a vector carrying the β₂m cDNA. Supernatants from both transfections were tested for UL183Fc presence by ELISA, concentrated and used to stain the Jurkat/CD85j cell line. The UL183Fc protein did not bind CD85j when produced in FO-1 cells, whereas the UL183Fc protein produced in FO-1 cells stably expressing β₂m was able to bind the CD85j molecule (Fig. 1B).

View larger version (32K): [in this window] [in a new window]   FIGURE 1. β2m is required for UL18 binding to CD85j. A, Staining of the Jurkat/CD85j cell line with the anti-CD85j HP-F1 mAb and with the UL183Fc protein (thick lines). Negative controls (secondary mAb alone) are shown (thin lines). B, Staining of the Jurkat/CD85j cell line with the HP-F1 mAb, with the supernatant obtained from FO-1 cells (FO-1/β2m–) transiently transfected with the UL183Fc construct and with the supernatant obtained from FO-1 cells stably expressing the β2m cDNA (FO-1/β2m+) transiently transfected with the UL183Fc construct (thick lines). Negative controls (secondary mAb alone) are shown (thin lines). C, Staining of the Jurkat/CD85j cell line with the anti-UL18 HP-F1 mAb, with the UL18Fc protein (not covalently linked to β2m) and with the UL18Fc-wt single chain protein (thick lines). Negative controls (secondary mAb alone) are shown (light gray hystograms). D, Staining of the Jurkat/CD85j cell line with the HP-F1 mAb and with UL18Fc-mutβ2m mutant protein in which Glu2, Arg3, Thr86, and Ser88 of the β2m were mutated to Ala (thick lines). Negative controls (secondary mAb alone) are shown (thin lines).

View larger version (28K): [in this window] [in a new window]   FIGURE 6. Effect of cassette replacement and site-specific mutations on the UL18/CD85j interaction. A, Staining of the CD85j-positive cell line Jurkat/CD85j with anti-UL18 HP-F1 mAb and with the UL18 single-chain soluble molecule UL18Fc-wt (thick lines). B, Staining of the Jurkat/CD85j cell line with the UL18Fc-Di cassette mutants obtained by replacing the D1–D6 UL18 regions with the corresponding HLA-A2 regions (thick lines). C, Staining of the Jurkat/CD85j cell line with mutants in which specific residues were mutated to Ala (thick lines). In all panels negative controls (secondary mAb alone) are shown (thin and dotted lines).

Structural prediction of the UL18 extracellular portion and identification of conserved sequence regions

The overall architecture of the UL18 molecule is similar to that of MHC class I molecules. UL18 is composed of three domains (1, 2, and 3): the 1 and 2 domains are predicted to be the peptide-loading domains, and the 3 domain has an Ig-like fold. Comparative modeling techniques can be useful in producing a reliable model of a protein structure provided that a homologous protein of known structure (template) can be identified. The quality and reliability of the model, and therefore its appropriate use, depends on the evolutionary distance between the target and template (31, 32) and on the availability of a good alignment between the sequences of the target protein and its structural template. It is well established that, whenever possible, the pairwise alignment should be derived from a multiple sequence alignment of the protein family including as many sequences as possible, (33) and that these sequences should be evenly distributed in terms of their similarities (34).

We performed a Blast search (22) and used the HHpred publicly available server (23) based on hidden Markov models to identify proteins homologous to UL18. The HHpred server has been shown to be among the best in the last critical assessment of techniques for protein structure prediction experiments (35). Among the retrieved proteins we found, as expected, HLA molecules. In particular, the closest protein of known structure is HLA-A2, sharing 26% sequence identity with UL18.

The retrieved sequences, as well as sequences of UL18 that we derived from nine HCMV-positive donors, were all aligned with UL18 from AD169 strain.⁵ The alignment was obtained by T-coffee and M-coffee (24). The resulting multiple-sequence alignment contains 44 sequences, 3 of which come from nonhuman viruses: the chimpanzee CMV sequence, the red squirrel poxvirus, and the murine CMV viruses, which share 50, 23, and 22% of sequence identity with human CMV UL18 sequence, respectively. The sequence identity of entries in our alignment ranges between 99 and 22%. The sequence logo derived from the multiple-sequence alignment is shown in Fig. 2.

View larger version (82K): [in this window] [in a new window]   FIGURE 2. Logo representation of the multiple sequence alignment used to build the UL18 model (see also Supplemental Material).5 The image has been obtained using the WebLogo server (38 ).

We extracted the pairwise alignment between UL18 and HLA-A2 from the multiple-sequence alignment (Fig. 3) and used it together with the structure recorded in the 1akj entry (36) of the PDB database (37) to build a model using Modeler (27). There are two different determinations of the structure of HLA-A2 (PDB = 1akj, resolution = 2.65; PDB = 1p7q, resolution = 3.40). We selected the 1akj structure because it has a significantly better resolution. The 1p7q structure (13), although at lower resolution, contains a complex between the HLA-A2 molecule and CD85j, so it was used as a template for reconstructing the complex. To achieve this, we first superimposed the main chain atoms of our model with the corresponding regions of 1p7q (267 equivalent positions; root mean square deviation of the C_ atoms = 2.95) and merged the coordinates of the CD85j molecule with those of our model (Fig. 4).

View larger version (56K): [in this window] [in a new window]   FIGURE 3. Structural alignment of the HLA-A2 protein (structure with PDB = 1akj) and the UL18 protein. UL18 and template secondary structures (SS), obtained by the PredictProtein server (5 ) and the DSSP program (1 ), are indicated; s = β-sheet, h = -helix. Arginines (N) of potential glycosylation sites are in light gray. Conserved cysteines are in bold and underlined. Cassette mutation regions (from D1 to D6) are framed in light gray boxes, while mutated amino acids are in bold and the position is indicated.

View larger version (62K): [in this window] [in a new window]   FIGURE 4. Structural modeling of the UL18-CD85j complex. A, Front view of the modeled complex between UL18 (metallic orange), β2m (transparent green), and CD85j (transparent gray). In the model, the cassette mutation regions are highlighted in color: D1 in orange, D2 in blue, D3 in red, D4 in green; D5 in purple, D6 in ice-blue. The image was obtained with the VMD software (39 ). B, Side view of the complex, which has been rotated by –60° with respect to the y-axis. To highlight the regions of interaction between UL18 and β2m, CD85j has been removed. Regions involved in the interaction are opaque. C, Enlargement of the 3 domain of the front view of the UL18 model. Residues that, when mutated, are responsible of the loss of binding of UL18 to the CD85j are indicated by colored balls. D, Superposition between UL18 (in golden orange) and the 1akj template used to build the model (in light blue). The regions that mainly differ between the two structures are highlighted (opaque areas).

The first region along the sequence is named D1 (Gly¹³⁶ to Thr¹⁴³). This is a Gly-rich region expected to have a high degree of flexibility and predicted by our model to be involved in the interaction between UL18 and β₂m.

The second region, named D2 (Gly¹⁵⁶ to Lys¹⁶⁸), contains a long insertion (nine residues long) when compared with the template. Interestingly, this region, predicted to form an -helix, could provide additional interactions with CD85j because, in our model, it lies very close to it. However, as mentioned before, D2 is not structurally conserved, and therefore the reliability of our model in this region is necessarily limited.

The next two putative regions of interaction are called D3 (Phe²⁰⁴ to His²¹⁰) and D4 (Asn²¹⁹ to Arg²²⁴). D3 is the linker region between domains 2 and 3 and is rather different from the corresponding region in HLA-A2 because of the presence of a four-residue gap. This deletion could cause a different orientation of the domains and a higher rigidity of the linker fragment, which could be further increased by the presence of two Pro residues in wild-type UL18, which are absent in HLA-A2. In our model, D4 is directly involved in the interaction with CD85j; however, because of the presence of a deletion, it cannot assume the same conformation as the corresponding region in HLA-A2, and the model predicts that the residues of UL18 involved in the interaction in this region might differ from those of HLA-A2. The UL18 D4 residue Asp²²², although conserved between the viral and human proteins, is in a different relative position than the corresponding Asp¹⁹⁶ of HLA-A2. The HLA-A2 Asp¹⁹⁶ interacts with Tyr⁷⁶ of CD85j, while the UL18 Asp²²² is rather far from this residue. Furthermore, in our model, UL18 Glu²²⁶ interacts with CD85j Lys⁴², and UL18 Arg²²⁴ forms an ion pair with CD85j Glu⁴⁰ (distance 3.46 Å). UL18 Arg²²⁴ structurally corresponds to HLA-A2 Glu¹⁹⁸, which interacts with CD85j Lys⁴¹ (distance 3.95 Å). This region is very well conserved among the UL18 variants, with the only exception being the substitution of Gln²²⁰ which is a His in 11 of 30 UL18 sequences (Fig. 5).

View larger version (51K): [in this window] [in a new window]   FIGURE 5. ClustalW alignment of the amino acid sequences of the 2 and 3 domains of the UL18 gene from AD169, Merlin, Toledo, and Towne strains, from HCMV-seropositive donors (A03, A10, B14, B16, B25, C01, C06, C10, K01, accession numbers AJ972608–AJ972614, AJ972616, AJ972617), from the literature (14 15 ), and from GenBank. Conservative substitutions are highlighted in gray. The consensus shows the positions that are identical in all the sequences. Cassette mutants are shown (boxes) and the mutated amino acids are indicated (arrows).

The D6 region contains three additional cysteine residues. From a visual inspection of the model, two of these, Cys²⁵⁸ and Cys²⁷³, belong to the same β-sheet (filaments D and E of the β-sandwich domain) and may form a disulfide bridge. Consistently with this hypothesis, these residues are highly conserved among the UL18 sequences (Fig. 5). The entire D5–D6 region is a proline-rich fragment, containing four Pro residues (Pro²⁶⁰ and Pro²⁶³ in D6, Pro²⁵¹ and Pro²⁵⁶ in D5) that are highly conserved among the HCMV sequences (Pro²⁶³ also among CMV of other species and throughout the whole HLA family). They could confer significant rigidity to this domain and could affect the binding between the D6 fragment and β₂m. The D6 region contains another cysteine (Cys²⁷⁹) that is exposed to the solvent and conserved only in HCMV.

Mutational analysis of the 3 domain of the UL18 molecule

Analyses of both the model and the multiple-sequence alignment led us to design a set of cassette mutants with the aim of obtaining a more detailed picture of the role of the D1–D6 regions in the UL18–CD85j interaction. We chose to replace these regions of the UL18Fc-wt molecule with the corresponding regions of the HLA-A2 sequence. The mutations were selected by taking into account the three-dimensional model of the protein, and they are therefore located in regions unlikely to be part of the conserved core or to be important for maintaining the fold of the protein. Because of this and because of the structural similarity between UL18 and HLA-A2, we expect the mutations not to change the overall fold of the chimeric molecules.

We constructed the following mutants: UL18Fc-D4, UL18Fc-D5, and UL18Fc-D6 (Fig. 3). These mutated UL18Fc molecules were produced by transiently transfecting HEK293T cells. The presence of the recombinant protein was verified by ELISA, and the supernatant was then concentrated 100x and used to stain the Jurkat/CD85j cell line. This procedure was used for the production of all recombinant proteins and for all cytofluorimetric analyses except where noted. Consistent with our expectations, the results show that all mutants lose their capacity to bind CD85j (Fig. 6B).

To have a more sensitive measurement method for evaluating the binding capacity of the UL18Fc mutants to CD85j, we performed binding measurements using Biacore. We consequently obtained HEK293T cell lines stably transfected with UL18Fc-wt, UL18Fc mutants, and HLA-A2Fc. These transfectants were used to produce high amounts of recombinant proteins. The supernatants were collected, purified, and dialyzed, and their recombinant protein concentrations were normalized by ELISA as described in Materials and Methods. Furthermore, we produced and purified soluble CD85j protein (CD85jFc) from the supernatants obtained by culturing Sp2/0 cells stably transfected with the CD85jFc construct. The CD85jFc molecule was immobilized on a biosensor chip where the recombinant purified proteins were injected and Biacore analysis was performed. This procedure was used for all Biacore analyses. With Biacore measurements no binding of UL18Fc-D4, UL18Fc-D5, and UL18Fc-D6 to CD85jFc was observed (Fig. 7, B and D). This result is in agreement with our model that predicts D4 to directly mediate the interaction with CD85j, an indirect role for the D5 region in binding β₂m, and the presence of a potentially important disulfide bridge in D6.

View larger version (39K): [in this window] [in a new window]   FIGURE 7. A, Cys279 is not involved in the UL18–CD85j interaction. HeLa cells transfected with a construct carrying a UL18 in which Cys279 was mutated to Ser, and Tyr362 and Lys374 were mutated to Ala (UL18-M10), were stained with CD85jFc and analyzed with a confocal microscope. B–E, Biacore analysis of the binding of UL18 mutants to CD85jFc. B, Binding measurements obtained with Biacore binding assays. Two independent experiments are shown. The mean value among the two experiments is shown in the fourth column as relative binding of HLA-A2 and mutants with respect to UL18Fc-wt binding, which was operationally normalized to 100. C, Biacore sensogram showing the relative binding of UL18Fc-wt and UL18Fc-D2. D, Biacore sensogram showing the relative binding of selected mutants. E, Biacore sensogram showing the relative binding of the UL18Fc-wt molecule (green line, 200 nM) and of serial dilutions of the UL18Fc-D2 mutant (200, 100, 50, 25, 12.5, and 6.2 nM) (purple line).

The results of the cytofluorimetric analysis of the above mutants are as follows (Fig. 6C): 1) Mutants UL18Fc-D6M1, UL18Fc- D6M6, and UL18Fc- D6M7 do not bind CD85j. The result obtained with the UL18Fc-D6M1 mutant strongly supports the presence of a disulfide bridge between Cys²⁵⁸ and Cys²⁷³. The result obtained with the UL18Fc-D6M6 emphasizes the structural role of Pro²⁶⁰ and Pro²⁶³ (the latter is very well conserved also in the HLA family), which can significantly contribute to the rigidity of the domain. Furthermore, according to our model, Pro²⁶³ may specifically interact with β₂m. Finally, loss of binding of the UL18Fc-D6M7 mutant is consistent with our prediction that Asp²⁶⁶ is involved in the interaction with β₂m, which, in turn, is necessary for the interaction with CD85j. 2) Mutants UL18Fc-D5M4 and UL18Fc- D6M5 retained binding capacity. The Pro²⁵¹ mutated in UL18Fc-D5M4 is located, in our alignment, inside an insertion with respect to HLA-A2, and it is situated in the corresponding CD85j-binding area. It probably has a structural role (contributing to the rigidity of the local structure), but, as predicted by our model, does not seem to play a major role in the interaction with CD85j or β₂m. Similarly, the two leucine residues mutated in UL18Fc-D6M5 (Leu²⁶¹ and Leu²⁶²) are predicted to not directly interact with the β₂m, differing from what is observed for the corresponding region in HLA-A2. In our model, Leu²⁶¹ is oriented toward the core of the UL18 molecule, while Leu²⁶² is more exposed. 3) Mutants UL18Fc-D5M8 and UL18Fc-D6M9 bind CD85j better than UL18Fc-wt. As mentioned earlier, in our model the D5 region does not seem to be involved in the interaction between UL18 and CD85j. The data suggest that the increased flexibility of this region upon mutation of Pro²⁵⁶ might favor the formation of novel interactions with CD85j. The result for the UL18Fc-D5M8 mutant confirms our model, as the D5 region is not essential for the interaction between UL18 and CD85j.

All of the above mutants, except for UL18Fc-D6M7, were used for Biacore analyses of their binding to CD85jFc (Fig. 7, B and D). Unfortunately, we were not able to obtain a stable transfectant producing the UL18Fc-D6M7 protein. The Biacore analysis showed that UL18, as well as all the mutants, bind to CD85j with a good binding stability and a dissociation rate much slower than that of the natural binder HLA-A2 (see slopes of the dissociation curves in Fig. 7D). The only mutant that shows a different dissociation profile is UL18Fc-DM4, which has a faster dissociation compared with the UL18Fc-wt, but it is still slower when compared with HLA-A2. This faster dissociation may explain why its binding to CD85j is higher by Biacore analysis, although it is lower than UL18Fc-wt by flow cytometry. Altogether, taking into account the differences in the protein production method and sensitivity in the two systems, the data obtained by Biacore for these mutants are in good agreement with those obtained with the cytofluorimetric analysis.

We also constructed a mutant of the wild-type transmembrane UL18 molecule in which Cys²⁷⁹ was mutated to Ser and Tyr³⁶² and Lys³⁷⁴ were mutated to Ala (UL18-M10). The Cys²⁷⁹ was predicted by our model to be exposed on the surface of the molecule and therefore available for intermolecular disulphide bonds, whereas Tyr³⁶² and Lys³⁶⁴ are involved in regulation of the intracellular trafficking of UL18 (41). Basically, Tyr³⁶² and Lys³⁶⁴ were mutated to allow the UL18-M10 mutant to be expressed on the surface of the cell. The UL18-M10 mutant construct was used to transiently transfect HeLa cells. After 24 h the cells were stained with the recombinant protein CD85jFc and analyzed by confocal microscopy (Fig. 7A). The figure shows that CD85jFc bound to the UL18-M10 mutant, thus revealing that Cys²⁷⁹ is not directly involved in binding of the two molecules.

Mutational analysis the 2 domain of the UL18 molecule

To further investigate the role of the 2 domain regions identified by our model in the UL18–CD85j interaction, we designed three cassette mutants in this domain: UL18Fc-D1, UL18Fc-D2, and UL18Fc-D3 (Fig. 3). The D1 region is predicted to interact with β₂m. It is rich in Gly residues and is therefore expected to be very flexible. In UL18Fc-D1, we replaced it with the corresponding less flexible HLA-A2 sequence. In UL18Fc-D2, we substituted a region predicted to have an -helix structure with the shorter HLA-A2 corresponding sequence. In UL18Fc-D3, we replaced the short (7 residues) UL18 loop, located at the boundary between the 2 and 3 domains, with the longer and more structured corresponding portion of HLA-A2 (11 residues, 4 of which are in -helical conformation).

Results of the cytofluorimetric analysis show that mutants UL18Fc-D1 and UL18Fc-D3 did not bind CD85j. Surprisingly, mutant UL18Fc-D2 showed a higher binding capacity compared with UL18Fc-wt (Fig. 6B).

For mutants UL18Fc-D2 and UL18Fc-D3, we also perfomed Biacore analysis (Fig. 7, B and C). No binding of UL18Fc-D3 was observed, whereas UL18Fc-D2 bound CD85j better than did UL18Fc-wt, as shown by cytofluorimetric analysis. Different UL18Fc-D2 dilutions were used for Biacore analysis showing that an eight-fold diluted solution of UL18Fc-D2 binds CD85j with the same capability of undiluted UL18Fc-wt (Fig. 7E).

In the first two cases, the results are consistent with our model. In the UL18Fc-D1 mutant, the binding with β₂m is expected to be disrupted. The UL18Fc-D3 mutant, where a linker region is replaced with a longer and more flexible region, is likely to cause a different orientation of the 3 and 1–2 domains. The results obtained with the UL18Fc-D2 mutant cannot be easily explained on the basis of our three-dimensional model. This implies that this region plays a different function in UL18 with respect to its counterpart in HLA-A2 (e.g., a regulatory role).

We next focused on the role played by selected amino acid residues inside the UL18Fc-D3 cassette. We constructed the following mutants: UL18Fc-D3M2 (Pro²⁰⁷ to Ala) and UL18Fc-D3M3 (His²¹⁰ to Ala) (Fig. 3). The results of the cytofluorimetric analysis showed that the UL18Fc-D3M2 mutant weakly bound CD85j whereas UL18Fc-D3M3 was not able to bind CD85j (Fig. 6C).

The Biacore analysis confirmed the weak binding of UL18Fc-D3M2 and showed a very low level of binding of UL18Fc-D3M3 to CD85j (Fig. 7, B and D). These results point to an important role played by Pro²⁰⁷ and His²¹⁰ in the UL18–CD85j interaction. Our model suggests that the role played by Pro²⁰⁷ is likely to be structural and related to the rigidity of the linker region, while His²¹⁰ (which forms hydrogen bonds with UL18 3 domain CD85j residues) is predicted to be important for the correct orientation of the 3 domain with respect to CD85j.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
It is well known how MHC class I molecules (i.e., HLA-A2) interact with CD85j (13), whereas much less information is available on the UL18–CD85j interaction. It has been shown that UL18 binds CD85j with an affinity 1000-fold higher than MHC class I molecules (11). The class I-like molecule HFE, in which the 3 domain was replaced with the 3 domain of the UL18 molecule, is able to bind CD85j, indicating that the primary sites of interaction are in the 3 domain and/or in the β₂m; also, both the extensive UL18 glycosylation and the presence or absence of a loaded peptide do not influence its binding to CD85j (11). Furthermore, CD85j bind both class I MHC molecules and UL18 through the same D1 and D2 domains (12).

In the work presented here, we attempted to dissect the structural determinants of binding of UL18 to CD85j. We constructed a three-dimensional model of the complex and used it to design mutants in selected regions of the putative interaction interface and to interpret their effects on binding. The model was built by comparative modeling and therefore can be effectively used to predict the role of structurally conserved regions and to rationally design mutants unlikely to affect the overall structure of the protein. The detailed structural prediction of regions where evolutionary changes are expected to modify the structure more substantially (e.g., where insertions and deletions are present) is unfortunately still beyond our present capabilities. Nevertheless, in these regions, if used in conjunction with experimental data, a model can still be used to derive general conclusions about the putative functional effect of the structural differences.

Using this strategy, we identified six regions (named D1 to D6) in both the 2 and 3 domains and specific amino acids within them, potentially involved in the UL18–CD85j function and interaction and unlikely to be involved in the overall stabilization of the fold. These regions were replaced with the equivalent regions of HLA-A2, with the rationale that the structural similarity of this latter molecule with UL18 would reduce the probability of mutations causing major structural effects.

We first investigated whether β₂m is necessary for binding to CD85j. Our results indicate that the UL18 3 domain is not sufficient per se for binding and that the presence of the β₂m is required. This suggests a direct role of β₂m in the interaction. Additionally, mutations in β₂m residues described to interact with CD85j in the CD85j/HLA-A2 complex structure (13) almost completely abolish the UL18–CD85j interaction.

The complete replacement of each of the D1–D6 regions, except for UL18Fc-D2, led to mutants unable to bind the CD85j molecule, suggesting a key role of these regions in the UL18–CD85j interaction. Surprisingly, the UL18Fc-D2 mutant showed a significantly increased binding affinity for CD85j. Because the D2 region is not structurally conserved between the CD85j molecule and the structural template HLA-A2, our model must be speculative in this region. We speculate that the substitution reduces the rigidity of the segment, predicted to be structured as an -helix in CD85j but not in HLA-A2, and therefore allows more interactions to take place. This aspect requires further investigation, as it might be important for understanding specific aspects of the UL18 interactions and, possibly, for its regulation.

The D4 region is the only one predicted to directly interact with CD85j, although its conformation is different from that of its HLA-A2 counterpart. Regions D1 and D6 seem to play a role in interaction with β₂m, probably affecting its direct interaction with CD85j. Region D3, situated at the boundary between the 2 and 3 domains, may affect the relative orientation of the two domains, whereas region D5 may play a structural role.

We do not have direct biophysical evidence that our cassette mutants preserve a protein folding closely similar to native UL18, except for UL18Fc-D2 mutant. The fact that this mutant binds UL18 with a higher affinity than does UL18Fc-wt indicates that, at least in the regions interacting with UL18, the folding of the protein is maintained as fully functional. UL18Fc-D1 and especially UL18Fc-D4 are mutants in which a short cassette is replaced and no insertions or deletions are present, making it very unlikely that the fold is different from the wild-type molecule. This might not be true for mutants UL18Fc-D3, UL18Fc-D5, and UL18Fc-D6 where, in fact, the mutated regions were investigated in more detail.

To further dissect the role of these regions in the interaction with CD85j, we mutated single or a few amino acids in some of them. Amino acids with peculiar characteristics (e.g., disulfide bridge formation, structure rigidity, steric hindrance, and electric charge) within regions shown to be potentially important by the model were chosen to infer the importance of such regions. Most mutations were likely to modify the local folding of the protein. Conversely, mutants that were likely not to mutate the local folding (i.e., UL18Fc-D5M4 and UL18Fc-D6M9) were observed not to significantly modify the UL18–CD85j interaction. This analysis showed that there are positions that play a pivotal role in the UL18–CD85j interaction. The mutation of His²¹⁰ alone (UL18Fc-D3M3 mutant) almost completely abolished binding to CD85j. The mutation of Cys²⁵⁸ and Cys²⁷³ (UL18Fc-D6M1) also had a dramatic effect on the binding of UL18 to CD85j, suggesting that they may form a disulfide bridge, as predicted by our model. Furthermore, the loop formed by Cys²⁵⁸ and Cys²⁷³ contains two Pro (Pro²⁶⁰ and Pro²⁶³) that, when mutated (UL18Fc-D6M6), dramatically reduced the binding to CD85j, strongly suggesting that both the structure and flexibility of this region are important for binding. Of note, a very recent study investigated the structural elements that underlie the UL18 high affinity for CD85j. The results pointed out, in accordance with our study, the importance of the D4 region as well as the pivotal role of the disulfide bridge formed by Cys²⁵⁸ and Cys²⁷³ (40).

A few interesting speculations can be derived from our analysis. Both the family sequence alignment and the fact that regions selected on the basis of the HLA-A2/CD85j/β₂m template structure do affect binding of UL18 strongly suggest that the architecture of the complex is conserved. Furthermore, no large insertions or regions expected to have substantially different local structure are present in the regions involved in the interaction, and both experimental and modeling results indicate that the binding of UL18 to CD85j is mediated by several regions that correspond to those known to interact in the HLA-A2 complex. Taken together, our results suggest that the higher affinity of UL18 compared with class I MHC molecule is unlikely to be brought about by the presence of additional interaction regions, but rather by an overall better fit of the two molecules, possibly brought about by a combination of changes in the flexibility of key parts of the interface and/or by subtle effects that could lead to more effective exploitation of the involvement of the β₂m moiety in complex formation.

The results presented here, together with the three-dimensional model of the complex now publicly available, are important for understanding details of the complex interactions between UL18 and its host partner, and may contribute to comprehension of the molecular basis of UL18 function.

	Acknowledgments

 
This study is dedicated to the memory of Carlo Enrico Grossi.

	Disclosures

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
The authors have no financial conflict of interest.

	Footnotes

 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

¹ This work was supported by grants from Fondazione Compagnia di San Paolo, Ministero per l’Istruzione, l’Università e la Ricerca Scientifica, and Progetto Finalizzato Ministero della Salute (to E.C.), and was partially supported by the European Commission within its FP6 Programme contract number LSHG-CT-2003-503265, and by the Istituto Pasteur Fondazione Cenci Bolognetti (to A.T.).

² M.O. and F.G. contributed equally to this work.

³ Address correspondence and reprint requests to Dr. Fabio Ghiotto, Department of Experimental Medicine, University of Genoa, Via De Toni 14, 16132 Genoa, Italy. E-mail address: fghiotto{at}unige.it

⁴ Abbreviations used in this paper: HCMV, human cytomegalovirus; β₂m, β₂-microglobulin; PDB, Brookhaven Protein Data Bank.

⁵ The online version of this article contains supplemental material.

Received for publication July 24, 2007. Accepted for publication October 30, 2007.

	References

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 

1. Beck, S., B. G. Barrell. 1988. Human cytomegalovirus encodes a glycoprotein homologous to MHC class-I antigens. Nature 331: 269-272. [Medline]
2. Hassan-Walker, A. F., I. M. Kidd, C. Sabin, P. Sweny, P. D. Griffiths, V. C. Emery. 1999. Quantity of human cytomegalovirus (CMV) DNAemia as a risk factor for CMV disease in renal allograft recipients: relationship with donor/recipient CMV serostatus, receipt of augmented methylprednisolone, and antithymocyte globulin (ATG). J. Med. Virol. 58: 182-187. [Medline]
3. Browne, H., M. Churcher, T. Minson. 1992. Construction and characterization of a human cytomegalovirus mutant with the UL18 (class I homolog) gene deleted. J. Virol. 66: 6784-6787. [Abstract/Free Full Text]
4. Browne, H., G. Smith, S. Beck, T. Minson. 1990. A complex between the MHC class I homologue encoded by human cytomegalovirus and β 2 microglobulin. Nature 347: 770-772. [Medline]
5. Fahnestock, M. L., J. L. Johnson, R. M. Feldman, J. M. Neveu, W. S. Lane, P. J. Bjorkman. 1995. The MHC class I homolog encoded by human cytomegalovirus binds endogenous peptides. Immunity 3: 583-590. [Medline]
6. Griffin, C., E. C. Wang, B. P. McSharry, C. Rickards, H. Browne, G. W. Wilkinson, P. Tomasec. 2005. Characterization of a highly glycosylated form of the human cytomegalovirus HLA class I homologue gpUL18. J. Gen. Virol. 86: 2999-3008. [Abstract/Free Full Text]
7. Cosman, D., N. Fanger, L. Borges, M. Kubin, W. Chin, L. Peterson, M. L. Hsu. 1997. A novel immunoglobulin superfamily receptor for cellular and viral MHC class I molecules. Immunity 7: 273-282. [Medline]
8. Colonna, M., F. Navarro, T. Bellon, M. Llano, P. Garcia, J. Samaridis, L. Angman, M. Cella, M. Lopez-Botet. 1997. A common inhibitory receptor for major histocompatibility complex class I molecules on human lymphoid and myelomonocytic cells. J. Exp. Med. 186: 1809-1818. [Abstract/Free Full Text]
9. Dietrich, J., M. Cella, M. Colonna. 2001. Ig-like transcript 2 (ILT2)/leukocyte Ig-like receptor 1 (LIR1) inhibits TCR signaling and actin cytoskeleton reorganization. J. Immunol. 166: 2514-2521. [Abstract/Free Full Text]
10. Saverino, D., A. Merlo, S. Bruno, V. Pistoia, C. E. Grossi, E. Ciccone. 2002. Dual effect of CD85/leukocyte Ig-like receptor-1/Ig-like transcript 2 and CD152 (CTLA-4) on cytokine production by antigen-stimulated human T cells. J. Immunol. 168: 207-215. [Abstract/Free Full Text]
11. Chapman, T. L., A. P. Heikeman, P. J. Bjorkman. 1999. The inhibitory receptor LIR-1 uses a common binding interaction to recognize class I MHC molecules and the viral homolog UL18. Immunity 11: 603-613. [Medline]
12. Chapman, T. L., A. P. Heikema, A. P. West, P. J. Bjorkman. 2000. Crystal structure and ligand binding properties of the D1D2 region of the inhibitory receptor LIR-1 (ILT2). Immunity 13: 727-736. [Medline]
13. Willcox, B. E., L. M. Thomas, P. J. Bjorkman. 2003. Crystal structure of HLA-A2 bound to LIR-1, a host and viral major histocompatibility complex receptor. Nat. Immunol. 4: 913-919. [Medline]
14. Cerboni, C., A. Achour, A. Warnmark, M. Mousavi-Jazi, T. Sandalova, M. L. Hsu, D. Cosman, K. Karre, E. Carbone. 2006. Spontaneous mutations in the human CMV HLA class I homologue UL18 affect its binding to the inhibitory receptor LIR-1/ILT2/CD85j. Eur. J. Immunol. 36: 732-741. [Medline]
15. Vales-Gomez, M., M. Shiroishi, K. Maenaka, H. T. Reyburn. 2005. Genetic variability of the major histocompatibility complex class I homologue encoded by human cytomegalovirus leads to differential binding to the inhibitory receptor ILT2. J. Virol. 79: 2251-2260. [Abstract/Free Full Text]
16. Farrell, H. E., H. Vally, D. M. Lynch, P. Fleming, G. R. Shellam, A. A. Scalzo, N. J. Davis-Poynter. 1997. Inhibition of natural killer cells by a cytomegalovirus MHC class I homologue in vivo. Nature 386: 510-514. [Medline]
17. Reyburn, H. T., O. Mandelboim, M. Vales-Gomez, D. M. Davis, L. Pazmany, J. L. Strominger. 1997. The class I MHC homologue of human cytomegalovirus inhibits attack by natural killer cells. Nature 386: 514-517. [Medline]
18. Leong, C. C., T. L. Chapman, P. J. Bjorkman, D. Formankova, E. S. Mocarski, J. H. Phillips, L. L. Lanier. 1998. Modulation of natural killer cell cytotoxicity in human cytomegalovirus infection: the role of endogenous class I major histocompatibility complex and a viral class I homolog. J. Exp. Med. 187: 1681-1687. [Abstract/Free Full Text]
19. Saverino, D., F. Ghiotto, A. Merlo, S. Bruno, L. Battini, M. Occhino, M. Maffei, C. Tenca, S. Pileri, L. Baldi, et al 2004. Specific recognition of the viral protein UL18 by CD85j/LIR-1/ILT2 on CD8⁺ T cells mediates the non-MHC-restricted lysis of human cytomegalovirus-infected cells. J. Immunol. 172: 5629-5637. [Abstract/Free Full Text]
20. Wagner, C. S., G. C. Riise, T. Bergstrom, K. Karre, E. Carbone, L. Berg. 2007. Increased expression of leukocyte Ig-like receptor-1 and activating role of UL18 in the response to cytomegalovirus infection. J. Immunol. 178: 3536-3543. [Abstract/Free Full Text]
21. Greten, T. F., F. Korangy, G. Neumann, H. Wedemeyer, K. Schlote, A. Heller, S. Scheffer, D. M. Pardoll, A. I. Garbe, J. P. Schneck, M. P. Manns. 2002. Peptide-β2-microglobulin-MHC fusion molecules bind antigen-specific T cells and can be used for multivalent MHC-Ig complexes. J. Immunol. Methods 271: 125-135. [Medline]
22. Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25: 3389-3402. [Abstract/Free Full Text]
23. Soding, J., A. Biegert, A. N. Lupas. 2005. The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Res. 33: W244-W248. [Abstract/Free Full Text]
24. Wallace, I. M., O. O’Sullivan, D. G. Higgins, C. Notredame. 2006. M-coffee: combining multiple sequence alignment methods with T-coffee. Nucleic Acids Res. 34: 1692-1699. [Abstract/Free Full Text]
25. Rost, B., G. Yachdav, J. Liu. 2004. The PredictProtein server. Nucleic Acids Res. 32: W321-W326. [Abstract/Free Full Text]
26. Kabsch, W., C. Sander. 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22: 2577-2637. [Medline]
27. Sali, A., T. L. Blundell. 1993. Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234: 779-815. [Medline]
28. Canutescu, A. A., A. A. Shelenkov, R. L. Dunbrack. 2003. A graph-theory algorithm for rapid protein side-chain prediction. Protein Sci. 12: 2001-2014. [Medline]
29. Gautier, R., A. C. Camproux, P. Tuffery. 2001. SCit: web tools for protein side chain conformation analysis. Nucleic Acids Res. 32: W508-W511.
30. Ye, Y., A. Godzik. 2003. Flexible structure alignment by chaining aligned fragment pairs allowing twists. Bioinformatics 19: (Suppl. 2):ii246-ii255. [Abstract]
31. Chothia, C., A. M. Lesk. 1986. The relation between the divergence of sequence and structure in proteins. EMBO J. 5: 823-826. [Medline]
32. Lesk, A. M., M. Levitt, C. Chothia. 1986. Alignment of the amino acid sequences of distantly related proteins using variable gap penalties. Protein Eng. 1: 77-78. [Free Full Text]
33. Tramontano, A.. 1998. Homology modeling with low sequence identity. Methods 14: 293-300. [Medline]
34. Cozzetto, D., A. Tramontano. 2005. Relationship between multiple sequence alignments and quality of protein comparative models. Proteins 58: 151-157. [Medline]
35. Berman, H. M., S. K. Burley, W. Chiu, A. Sali, A. Adzhubei, P. E. Bourne, S. H. Bryant, R. L. Dunbrack, K. Fidelis, J. Frank, et al 2006. Outcome of a workshop on archiving structural models of biological macromolecules. Structure 14: 1211-1217. [Medline]
36. Gao, G. F., J. Tormo, U. C. Gerth, J. R. Wyer, A. J. McMichael, D. I. Stuart, J. I. Bell, E. Y. Jones, B. K. Jakobsen. 1997. Crystal structure of the complex between human CD8() and HLA-A2. Nature 387: 630-634. [Medline]
37. Berman, H., K. Henrick, H. Nakamura. 2003. Announcing the worldwide Protein Data Bank. Nat. Struct. Biol. 10: 980[Medline]
38. Crooks, G. E., G. Hon, J. M. Chandonia, S. E. Brenner. 2004. WebLogo: a sequence logo generator. Genome Res. 14: 1188-1190. [Abstract/Free Full Text]
39. Humphrey, W., A. Dalke, K. Schulten. 1996. VMD: visual molecular dynamics. J. Mol. Graphics 14: 33-38. [Medline]
40. Wagner, C. S., A. Rölle, D. Cosman, H. G. Ljunggren, K. D. Berndt, A. Achour. 2007. Structural elements underlying the high binding affinity of human cytomegalovirus UL18 to leukocyte immunoglobulin-like receptor-1. J. Mol. Biol. 373: 695-705. [Medline]
41. Maffei, M., F. Ghiotto, M. Occhino, M. Bono, A. De Santanna, L. Battini, G. L. Gusella, F. Fais, S. Bruno, E. Ciccone. 2008. Human cytomegalovirus regulates surface expression of the viral protein UL18 by means of two motifs present in the cytoplasmic tail. J. Immunol. 180: 969-979. [Abstract/Free Full Text]

This article has been cited by other articles:

				Z. Yang and P. J. Bjorkman Structure of UL18, a peptide-binding viral MHC mimic, bound to a host inhibitory receptor PNAS, July 22, 2008; 105(29): 10095 - 10100. [Abstract] [Full Text] [PDF]

				M. Maffei, F. Ghiotto, M. Occhino, M. Bono, A. De Santanna, L. Battini, G. L. Gusella, F. Fais, S. Bruno, and E. Ciccone Human Cytomegalovirus Regulates Surface Expression of the Viral Protein UL18 by Means of Two Motifs Present in the Cytoplasmic Tail J. Immunol., January 15, 2008; 180(2): 969 - 979. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Data Supplement Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Occhino, M. Articles by Ciccone, E. Search for Related Content PubMed PubMed Citation Articles by Occhino, M. Articles by Ciccone, E. Pubmed/NCBI databases Substance via MeSH