A Mutation in the First Ligand-Binding Repeat of t
http://www.100md.com
病菌学杂志 2005年第23期
Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, and Department of Medical Biochemistry, Medical University of Vienna, Dr. Bohr Gasse 9/3, A-1030 Vienna, Austria
Center for Anatomy and Cell Biology, Department for Nuclear Biology, Developmental Biology, and Functional Microscopy, Medical University of Vienna, Vienna, Austria
ABSTRACT
Minor group human rhinoviruses (HRVs) bind members of the low-density lipoprotein receptor family for cell entry. The ligand-binding domains of these membrane proteins are composed of various numbers of direct repeats of about 40 amino acids in length. Residues involved in binding of module 3 (V3) of the very-low-density lipoprotein receptor (VLDLR) to HRV2 have been identified by X-ray crystallography (N. Verdaguer, I. Fita, M. Reithmayer, R. Moser, and D. Blaas, Nat. Struct. Mol. Biol. 11:429-434, 2004). Sequence comparisons of the eight repeats of VLDLR with respect to the residues implicated in the interaction between V3 and HRV2 suggested that (in addition to V3) V1, V2, V5, and V6 also fulfill the requirements for interacting with the virus. Using a highly sensitive binding assay employing phage display, we demonstrate that single modules V2, V3, and V5 indeed bind HRV2. However, V1 does not. A single mutation from threonine 17 to proline converted the nonbinding wild-type form of V1 into a very strong binder. We interpret the dramatic increase in affinity by the generation of a hydrophobic patch between virus and receptor; in the presence of threonine, the contact area might be disturbed. This demonstrates that the interaction between virus and its natural receptors can be strongly enhanced by mutation.
INTRODUCTION
Human rhinoviruses (HRVs), the predominant pathogen causing common cold infections, are icosahedral particles composed of 60 copies each of the capsid proteins VP1 through VP4 and a single-stranded RNA genome of positive (messenger sense) polarity (for a review, see reference 28). Twelve serotypes, the minor group, use members of the low-density lipoprotein receptor (LDLR) family; 87 serotypes, the major group, use intercellular adhesion molecule 1 for cell entry (14, 31, 34). These membrane proteins are neither structurally nor functionally related. LDLR attaches to the highly exposed and accessible star-shaped dome at the fivefold axis of icosahedral symmetry, whereas intercellular adhesion molecule 1 binds via its first immunoglobulin-like domain within the canyon, a cleft encircling this plateau (12, 24). The geometry of the interaction between a fragment of the very-low-density lipoprotein receptor (VLDLR) and the minor group virus HRV2 has been determined to 16-? resolution by cryoelectron microscopy (22) and to 3.5-? resolution by X-ray crystallography (32); amino acid residues of the virus interacting with amino acid residues of VLDLR ligand-binding module 3 (V3) have been defined. This revealed a predominant role of K1224 in HRV2 (Lys224 in VP1) that interacts with W22 and the highly conserved acidic cluster with the sequence motive DxD/ExD (where x is any amino acid residue) in the V3 module.
VLDLR, like other members of the LDLR family, has the ligand-binding domain at the N terminus, followed by sequences exhibiting similarity to the epidermal growth factor precursor interspaced by YWTD motives. As seen in the X-ray structure of the exodomain of LDLR at low pH, these latter sequences are arranged in the form of a ?-propeller (26); due to the high sequence conservation, the same conformation is most likely present in the VLDLR. Proximal to the transmembrane sequence is an O-linked sugar domain. A sequence pattern in the cytoplasmic C terminus is responsible for internalization via the clathrin-coated pit pathway by binding to the adapter complex AP2. Other members of the family, such as LDLR-related protein (LRP), are involved not only in ligand internalization (e.g., LDLR is responsible for cholesterol homeostasis by endocytosis of apolipoprotein B [ApoB]-complexed lipids) but also in signal transduction (23).
The ligand-binding domain of VLDLR comprises eight imperfect repeats, each about 40 amino acids long (Fig. 1). Their conformation is maintained by a Ca2+ ion that is chelated by the carboxylates of Glu and Asp residues in the acidic cluster and two backbone oxygens. Further stabilization is achieved via three disulfide bonds present in each of the modules (5, 26).
We have shown previously that HeLa cells are protected against infection by minor-group HRVs by soluble forms of fragments of the ligand-binding domain of LDLR and VLDLR (15, 16, 25). Recombinant soluble VLDL minireceptors, including the first three repeats (V1 to V3) of human VLDLR, displayed inhibitory activity towards HRV2 infection, whereas a receptor fragment comprising repeats 4 to 6 (V456) failed to bind and neutralize HRV2 (25). Virus binding to a collection of receptor fragments was directly visualized by cryoelectron microscopy (12, 22). Using capillary electrophoresis methodology, it was recently demonstrated that 12 molecules of an artificial concatemer composed of five copies of repeat 3 arranged in tandem and fused to maltose-binding protein (MBP-V33333) bind per virus particle (13). This, together with the strong increase in affinity and virus-neutralizing activity with the number of concatemerized modules (20), suggests that more than one and (most probably) all five modules within the same molecule contribute to binding by establishing an oligovalent interaction with the symmetry-related binding sites on the virus. This also suggests that the natural receptors with their 7, 8, and 31 repeats that are present in LDLR, VLDLR, and LRP, respectively, might similarly attach in a multivalent manner, resulting in a 12-to-1 stoichiometry. However, as seen from our present data, it is unlikely that five consecutive modules in the native receptors participate, since some of the repeats lack binding activity, at least in VLDLR.
Analysis of a number of receptor fragments including parts of the natural receptors and several artificial concatemers by cryoelectron microscopy and image reconstruction revealed that V1, the first VLDLR repeat, does not bind HRV2; this was concluded from the absence of density attributable to V1 in reconstructed images of complexes between HRV2 and MBP-V123 or MBP-V122 and from the lack of complex formation between HRV2 and an artificial tandem concatemer of V1 (MBP-V111). Except from a report on the binding of repeat 5 of LDLR (L5) alone to LDL (7) and from data on Tva, a single-repeat receptor of avian leucosis-sarcoma virus type A (8, 35), binding of a single module to any ligand has not been observed. Furthermore, the X-ray data of a complex between HRV2 and V23 suggested attachment of individual modules but could not definitely exclude involvement of the second repeat (32). Using display of various repeats on fusion phages, we now demonstrate very weak but clearly detectable binding of individual single repeats to HRV2 and show that changing one amino acid residue in the nonbinding V1 transforms it into a strong binder.
MATERIALS AND METHODS
Production of fusion phage. DNA encoding selected VLDLR ligand-binding repeats was amplified by PCR using synthetic primers (Table 1) containing NcoI and NotI restriction sites. These sites are unique in the phagemid pCANTAB6 (Cambridge Antibody Technology, Cambridge, United Kingdom) and were thus used for directional cloning of the amplified PCR fragments. The resulting constructs encode the pIII phage protein carrying the respective minireceptors at its N terminus and a six-His tag, followed by a myc tag between the C terminus of pIII and the VLDLR modules. The phagemids were transformed into Escherichia coli TG1 (Amersham Pharmacia). Following selection on LB plates containing 100-μg/ml ampicillin, single colonies were picked and grown at 37°C overnight in 5 ml LB medium supplemented with the antibiotic. The next day, 50 ml of LB-ampicillin medium containing 1% glucose was inoculated, and the bacteria were infected with 15 μl of M13K07 helper phage (1013 particles/ml). After incubation for 30 min at 37°C without shaking to allow for infection by the helper phage, the culture was further incubated for 30 min at 37°C with shaking to allow for expression of the resistance genes. The bacteria were pelleted and resuspended in 100 ml LB medium containing 100-μg/ml ampicillin and 50-μg/ml kanamycin and grown overnight at 30°C. Bacteria were removed by low-speed centrifugation, and phage particles in the supernatant were concentrated by precipitation with polyethylene glycol and resuspended in water (29). Receptor fragments displayed on the phage were folded overnight by adjustment to 20 mM Tris-HCl (pH 8.0) and 150 mM NaCl (TBS) containing 10 mM CaCl2 (TBSC) supplemented with 1 mM cysteamine-0.1 mM cystamine. Phage was stored in this buffer at 4°C. Mutagenesis of phage-displayed V1 was carried out using a constant pair of external primers and the oligonucleotides listed in Table 1 by mutually primed synthesis, following standard protocols. Fusions of the modules with maltose-binding protein (e.g., V1pMal) (Table 1) were made essentially as described by Ronacher and colleagues (25).
Cloning of MBP-V1T17PV1T17P. DNA encoding the sequence of V1T17P was amplified by PCR using the primers V1-ConCat (Table 1). The resulting fragment was ligated into the vector V1pMal previously cut with XbaI and blunted with DNA polymerase I, large (Klenow) fragment. The fusion protein was expressed and folded as described above.
Western blotting and virus overlay blot. The concentration of the phage in the resuspended polyethylene glycol pellet was determined spectrophotometrically and as the number of CFU (29). The presence of the foreign protein fused to the pIII phage protein was assessed by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of the phage proteins, followed by electrotransfer onto a polyvinylidene difluoride (PVDF) membrane. To conserve the native structure of the receptors, nonreducing sample buffer was used throughout; under these conditions, LDLR can even be boiled in SDS without the loss of binding activity (3). After being blocked with TBS containing 2% nonfat dried milk and 2% bovine serum albumin (BSA) (blocking buffer), the membrane was incubated with a rabbit antiserum against the myc tag (Abcam, Ltd., Cambridge, United Kingdom) diluted 1:7,000, followed by incubation with goat anti-rabbit immunoglobulin G coupled to horseradish peroxidase (HRP; Bio-Rad Laboratories) diluted 1:10,000. Bound antibody was revealed by chemiluminescence using the ECL detection kit from Pierce, following the manufacturer's instructions. Binding of virus to the electrotransferred proteins was assessed by incubation of the blot with 100,000 cpm of [35S]methionine-cysteine-labeled HRV2 (17, 21) in TBSC supplemented with 1% BSA and 1% nonfat dried milk (incubation buffer) overnight at 4°C or, alternatively, for 2 h at room temperature. For competition experiments, the radiolabeled virus was preincubated with 1 μg MBP-V33333, an artificial concatemer of five copies of repeat 3 arranged in tandem and fused to the maltose-binding protein (20, 32) in 100 μl PBS prior to incubation of the membrane. The membrane was washed three times with TBSC containing 0.1% Tween 20 (wash buffer), air dried, and exposed to X-ray film.
Determination of binding of fusion phage to HRV2. Enzyme-linked immunosorbent assay (ELISA) vinyl assay plates from Costar or Maxisorb plates from Nunc were coated overnight at 4°C with 10-μg/ml MBP-V33333 in TBS. Wells were blocked with blocking buffer for 1 h and incubated with purified HRV2 at 10 μg/ml for 3 h at room temperature. This procedure minimizes structural changes of the virion that might occur upon direct coating to the plastic surface. After being washed with blocking buffer, phage (1013 particles/ml as determined by spectrophotometry) was added, and incubation was continued overnight at 4°C. Note that for V123 and the strongly binding mutant of V1, final concentrations of 2 x 1012 and 4 x 1012 particles/ml, respectively, were employed to not exceed an absorbance value (A450) of 1.8 upon photometric measurement. Following 10 washes with TBS containing 0.1% Tween 20, the wells were incubated for 90 min at room temperature with monoclonal antibody against M13 (anti-M13, diluted 1:6,000; Amersham Pharmacia), followed by secondary HRP-conjugated antibody against mouse immunoglobulin G (Bio-Rad) diluted 1:10,000. Bound phage was finally revealed by incubation for 15 min at room temperature with 0.1 mg/ml trimethylbenzidine and 0.03% H2O2 in 0.1 M Na acetate, pH 6.0. Color development was halted by the addition of 1 M H2SO4; A450 was determined with a microplate reader. Negative controls included use of 1-mg/ml BSA instead of phage or the addition of 20 mM EDTA to the phage suspension during the attachment reaction.
Alternatively to the ELISA determination, a biological test was carried out. Exponentially growing strain TG1 (100 μl) was added to the wells and incubated for 30 min at 37°C without shaking, followed by another 30 min at 37°C with shaking. Then, bacteria were removed and plated on ampicillin plates. The next day, colonies arising from infection with phage that had been retained in the wells were counted.
RESULTS
Sequence alignment of VLDLR modules suggests that V1, V2, V3, V5, and V6, but not V4, V7, and V8, can bind HRV2. During cryoelectron microscopy analysis of complexes between HRV2 and various soluble VLDLR fragments or module concatemers, we observed that V1 does not contribute to HRV2 binding. Whereas concatenation of several copies of V3 greatly enhanced its avidity toward HRV2, three copies of V1 fused in tandem to maltose-binding protein in the concatemer MBP-V111 were inactive with respect to virus attachment. Proper folding of the receptor was shown by its binding radioactive 45Ca2+ (19, 22). To understand the basis of this deficiency, we examined sequence alignments of all VLDLR repeats with respect to eventual conservation of binding residues, as seen in the X-ray structure of the complex between V23 and HRV2 (32). As depicted in Fig. 1, V1, V2, V3, V5, and V6 possess the tryptophan and the acidic cluster DxD/ExD, known to be key players in the binding reaction; a lysine strictly conserved in all minor group HRVs (33, 34) interacts with the receptor module via its aliphatic side chain with W22 and via its -amino group with the acidic cluster around the Ca2+ ion (see Fig. 4A) (32). This latter binding reaction has an electrostatic component contributed by the carboxylates that are involved in Ca2+ complexation but also an ionic component resulting from a salt bridge with the central aspartate-glutamate. In addition, W22 exhibits hydrophobic interactions with L1132 and with I1226 from the symmetry-related VP1 molecule. The tryptophan is present in all modules except V4, V7, and V8, where it is replaced by phenylalanine, arginine, and lysine, respectively. In addition, the acidic cluster is incomplete in V7 and V8. Therefore, the latter three modules should not be able to bind the virus. For V1, however, we found no reasonable explanation for the lack of binding from the sequence alone.
Phage display allows detection of binding of single repeats. Single repeats bind HRV2 extremely weakly; in virus overlay blots, as utilized to demonstrate affinity of a given receptor for the virus, single repeats only occasionally gave rise to barely visible bands. Similarly, in a cell protection assay only very high concentrations of the recombinant soluble single modules fused to MBP were able to marginally inhibit infection of HeLa cells with HRV2 (32). This is most probably related to avidity effects, since binding, as well as viral neutralization, is enormously increased by oligomerization of the modules. This is seen for native VLDLR and for concatenated identical repeats in artificial receptor derivatives (see above) (13, 20). We therefore sought a method that would allow easy detection of virus binding by single modules. V1, V2, V3, and V123 were cloned in the phagemid pCANTAB6 N terminal to the phage attachment protein pIII, as detailed in Materials and Methods. Recombinant phage was propagated in E. coli TG1 and concentrated by precipitation with polyethylene glycol. Microtiter plates, to which HRV2 was bound via MBP-V33333, a concatemer of V3 that exhibits high-affinity binding (32), were challenged with recombinant phage; binding was assessed either by an ELISA format by sequential incubation with anti-phage antibody, secondary HRP-conjugated antibody, and substrate (Fig. 2A) or by a biological assay in which attached phage was allowed to infect bacteria. The phage titer was subsequently determined by a colony-forming assay (Fig. 2B). Binding of the recombinant phage carrying the different receptor modules was weak but clearly detectable by both methods and showed, in agreement with earlier data, that V123 bound strongly, followed by V3 and V2. V1 did not detectably attach to HRV2 (not shown, but see below and Fig. 3). As expected from the strict requirement for Ca2+ ions for the structural integrity of the receptor, binding of the recombinant phage to HRV2 was abolished in the presence of 20 mM EDTA. Helper phage without a foreign protein fused to its pIII protein and replacement of phage with BSA were used as additional controls and showed only background binding. This latter control was also included in the biological assay to test for eventual carryover of infectious phage. The grading of the binding affinity of the modules was similar in both assays (compare Fig. 2A and B). Thus, the phage system was well suited to detect the interaction of very weakly binding single receptor repeats to HRV2.
A single mutation in V1 transforms the nonbinding wild-type module into a strongly binding module. Comparison of the sequences of the virus-binding modules V2 and V3 with the nonbinding module V1 drew our attention to the presence of R14 instead of Q (which is involved in the interaction between V3 and virus), of T17 instead of P, and of V28 instead of an acidic residue (E in V2 and D in V3). The reasons to change the arginine and the valine are obvious; threonine was replaced by proline, based on the assumption that this might have an effect on the local conformation of the binding epitope. Two V1 mutants were constructed by site-directed mutagenesis in which residues were exchanged for those present in V2. In V1T17P,V28E, T17 and V28 were replaced by P and E, respectively. In V1R14Q, R14 was replaced by Q. The constructs were cloned into the phagemid, and fusion phage was produced. As controls, V4 and V5 were also cloned into the phagemid and expressed. The result of an ELISA (Fig. 3A) revealed that wild-type V1 and V1R14Q, as well as V4, as expected from the absence of the essential W22 in this latter module, did not bind virus. As already demonstrated, V2 and V3 bound weakly and V123 bound strongly. In accordance with the presence of W22 and the complete acidic cluster, V5 also recognized HRV2 and bound even more strongly than V3. Most interestingly, however, the double mutant V1T17P,V28E bound much more strongly than V3 and V5. Again, using controls with BSA and helper phage and conducting the attachment reaction in the presence of EDTA showed very low levels of background binding, indicative of the specificity of the assay.
As mentioned above, strong binding of virus to a receptor can be assessed via a virus overlay blot in which receptor proteins are separated on a SDS-PAGE gel under nonreducing conditions and electrotransferred to a PVDF membrane. The membrane is subsequently incubated with radiolabeled virus, and bands are revealed by autoradiography. We previously used this method extensively to identify the minor group HRV receptor (17, 18) and to demonstrate binding of minor-group HRVs to various LDL receptor derivatives (15). However, we could only occasionally detect extremely weak virus binding to a single repeat upon heavily overloading the gel. Therefore, we wondered whether the mutations in V1 might have increased the binding affinity to such an extent as to allow for its detection in virus overlay blots. Fusion phage was dissociated into its components by being boiled in sample buffer without reducing agent. Phage carrying the single repeats V1, V2, V3, V4, and V5 were also included in the analysis. The material was separated on a SDS-PAGE gel and blotted, and fusion proteins were detected with antiserum against the myc tag. Figure 3B, top, shows that all fusion proteins were present in similar amounts. Proteins eventually recognized by the virus were revealed by incubation with radiolabeled HRV2 on an identical blot prepared in parallel (Fig. 3B, bottom). Whereas no virus binding to V1, V2, V4, and V5 was seen in this assay, a strong band appeared in the lane containing V1T17P,V28E, largely exceeding the intensity of V3 that was also visible at this high load. In disagreement with the data from the ELISA (Fig. 3A), V5 binding was not detected. This might be due to a large proportion of the protein present in a misfolded form in this sample, as suggested from the presence of a double band (Fig. 3B, top). When 20 mM EDTA was added to the incubation medium, no binding was detected (results not shown).
Replacement of threonine 17 by proline is responsible for the increase in affinity of V1 for HRV2. To cut down the change in the properties of V1 to a single amino acid, a mutant was constructed in which only T17 was replaced by a proline. Similar tests as described above showed that the single mutant essentially displayed the same affinity for the virus as the double mutant (not shown). This demonstrates that the V28E mutation has no effect on the affinity and that replacement of threonine 17 by proline is the reason for the enormous increase in affinity for HRV2.
MBP-V33333 competes with V1T17P for virus binding. The amount of recombinant protein expressed on phage was small. To obtain sufficient material for further experiments, the single mutant V1T17P was cloned into the pMal-2c vector downstream of MBP. As in other MBP fusion proteins (25), the presence of a His6 tag in the vector allowed for easy purification. MBP-V1T17P was expressed, purified, folded, run on a SDS-PAGE gel, and blotted onto a PVDF membrane as in the results shown in Fig. 3B. Binding of radiolabeled virus was then assessed in the absence and in the presence of MBP-V33333. As depicted in Fig. 3C, virus bound well to MBP-V1T17P (left), but its attachment was abolished upon addition of MBP-V33333. From this, and the loss of binding in the presence of EDTA (Fig. 2), we conclude that attachment of V1T17P to HRV2 is specific and occurs to a site overlapping with or identical to the binding site of V3 on the viral surface.
A dimer of V1T17P protects HeLa cells against infection with HRV2. We have demonstrated previously that soluble receptor derivatives fused to MBP inhibit viral infection and protect HeLa cells against lysis to various degrees (25, 32). However, for single repeats, the concentrations required for noticeable protection are very high compared to that of artificial concatemers. Thus, cell protection assays were carried out using MBP-V1T17P, essentially as described in a previous paper (32). However, despite of displaying much higher affinity for HRV2 than MBP-V3, MBP-V1T17P failed to protect HeLa cells against infection with HRV2 (Fig. 3D). Since concatemerization of even two V3 modules resulted in a strong increase in neutralization capacity, a dimer of V1T17P was constructed and expressed as a fusion with MBP (MBP-V1T17PV1T17P). This protein exhibited cell protection efficiency exceeding that of MBP-V3; whereas MBP-V3 protected to some extent at 25 μg/ml, MBP-V1T17PV1T17P displayed a clear effect at 1.6 μg/ml under the specified conditions (Fig. 3D).
DISCUSSION
Single repeats of the members of the LDL receptor family exhibit very low affinity for minor group HRVs and other ligands such as ApoE complexed with lipid. It was thus assumed that at least two consecutive modules are required for binding (6, 15). This was corroborated by structure determination of a complex between HRV2 and a receptor fragment with two consecutive receptor modules (V23) by cryoelectron microscopy (22) and by X-ray crystallography (32). Although the resolution of 3.5 ? allowed identification of the interacting amino acid residues, it was not possible to definitely decide whether only single repeats were bound or whether, at least to a minor extent, two repeats were also attached simultaneously to two of the five symmetry-related binding sites at the vertex of the viral icosahedron. At least V3 alone binds very weakly to HRV2, since it is able at high concentration to marginally protect HeLa cells against infection, but the concatemer MBP-V33333 is about 10,000 times more effective (32). This increased neutralization capacity indeed arises from multimodule attachment (13, 20).
To first compare the binding affinities of the various repeats, we developed an assay utilizing fusion phage. Filamentous phage such as M13 contains 5 copies of the attachment protein pIII at the tip and about 2,000 copies of the coat protein pVIII among several other minor proteins (1). Presence of no more than two copies of a pIII fusion protein within a phage particle is compatible with attachment to the host bacteria and allows for propagation in the presence of helper phage that provides the necessary copies of wild-type pIII. We reasoned that interaction of the receptor moiety with immobilized virus might be detected with high sensitivity via antibodies recognizing pVIII or via a biological assay that, at least in theory, can detect a single infectious phage particle. Therefore, we constructed fusion phage displaying various single-receptor repeats and combinations and showed that low-affinity interactions could indeed be measured by this assay. We found that V3 bound more strongly than V2 and that V1 did not bind at all (Fig. 2 and 3).
We then addressed the question of why V1 failed to bind despite possessing all specific residues believed to be necessary for interaction (Fig. 1). Comparison of the sequences of the modules drew our attention to three residues that were subsequently exchanged for those that are at least functionally conserved at the equivalent positions in V2 and V3 (Fig. 1). ELISA and a colony-forming assay with fusion phage carrying the mutated modules demonstrated that a change from threonine 17 to proline and from valine 28 to glutamic acid resulted in a module that had acquired binding activity by much exceeding that of V3 (Fig. 3). Finally, a mutant with the single change from threonine 17 to proline was constructed and expressed as a fusion with maltose-binding protein. This mutant was as active as the double mutant in virus binding.
Usually, one or two copies of the fusion protein are displayed on phages. Therefore, phage binding might have occurred due to the presence of two copies of the module in spatial vicinity that could attach simultaneously to symmetry-related sites on the virion. Binding of MBP-V1T17P excludes bivalent attachment and demonstrates that a single copy of a binding module can indeed attach to the virus. V1T17P and V3 bind to the same or at least to a closely overlapping site, since they compete (Fig. 3C).
The particular proline in question is rather conserved and is present in Tva, in human LDLR (L2 and L4), VLDLR (V2, V3, and V5) and LRP (R2, R4, R5, R7, R8, R10, R15, R17, R18, R21 through R26, R28, and R29). However, its absence from L5 suggests that it is not essential, since L5 makes a strong contribution to binding of ApoB and ApoE to LDLR (4) and is clearly involved in binding HRV1A, another minor-group HRV (10). Therefore, the presence of the proline, together with W22 and the acidic cluster, is still not sufficient to form a binding epitope.
What is the basis of the affinity increase upon exchange of threonine 17 for proline? Inspection of residues in the complex of V3 and HRV2 within 6 ? of the proline revealed that only leucine 1226 and isoleucine 1132, contributed from the next symmetry related virus subunit, are in its vicinity (Fig. 4B). Together with the neighboring tryptophan 22, this results in a hydrophobic cluster that might be interrupted upon exchange of the proline with threonine. We thus believe that the hydrophobic core is at least in part responsible for HRV2 binding to V2, V3, V5, and most probably V1T17P. A model of V1 and of V1T17P calculated with SWISS-MODEL (27) showed that W22 and P17 might adopt an arrangement very similar to that in V3 (data not shown). Conformational effects might contribute to the affinity increase upon mutation of T17 as well. Nevertheless, in the absence of real structural data this interpretation must remain speculative. Proline 17 (or proline 19 in V1) is absent from LDLR modules L4 and L5. When fused to phage, binding of these single LDLR repeats to virus could not be detected, suggesting that the proline might indeed be responsible, at least in part, to strong binding (data not shown). It is noteworthy that the proline is replaced by a histidine in LDLR module L5. According to Rudenko and colleagues, this histidine forms a salt bridge with a glutamic acid of the YWTD ?-propeller at pH 5.3 and thus participates in intramolecular interactions (26). In the case of the virus receptor complex, at pH 7.4, the histidine is most probably not protonated and thus fits into the hydrophobic cluster. Upon exposure to the low-pH environment of the endosome, protonation might therefore be in part responsible for dissociation of HRV2 from LDLR at low pH (2).
Cell protection assays showed that the increase in binding affinity of the single repeat was not correlated with neutralization efficiency, since the weakly binding MBP-V3 neutralized very weakly and the strongly binding MBP-V1T17P did not neutralize at all. Taken together with other binding and neutralization data (13, 20, 25), this indicates that viral neutralization strongly depends on simultaneous binding of more than one module to the viral surface. Therefore, it most probably results from stabilization of the viral capsid similar to that exerted by bivalent antibody binding, which inhibits uncoating (11, 30). A receptor module bridges two symmetry related sites on the viral surface (22). Subtle differences in the binding forces directed towards the two sites might be responsible for the difference in neutralization efficiency.
Although the data on competition with V33333 indicate that V1T17P and V3 attach to the same epitope, the absence of noticeable cell protection by V1T17P is puzzling. To definitely exclude binding of V1T17P to a site different from that used by V3, a dimer of V1T17P was also constructed. This dimer indeed protected HeLa cells against infection with HRV2, lending definite proof to its binding being specific. However, due to the cloning procedure, the linkers between the modules differ from those between the V3 modules in MBP-V33; instead of PG in V33, the linker in V1T17PV1T17P is VKSR. The different composition and distance might affect binding strength. A comparison of the respective concentrations resulting in neutralization is thus not possible.
In summary, we have demonstrated that mutation of a single amino acid residue in module V1 of VLDLR transforms a nonbinding ligand-binding repeat into a strongly binding module. The amino acid residue changed is to some extent involved in interaction with the virus surface by contributing to a hydrophobic cluster. However, it might also influence the overall conformation of the module, thereby modifying the interacting surface. A dimer of V1T17P had the capacity to neutralize virus. Comparison of the kinetics of virus binding of V1T17P and V3 might allow unraveling the parameters underlying viral binding and neutralization. Experiments to determine the kinetic on and off rates by using surface plasmon resonance and nuclear magnetic resonance techniques are currently in progress.
ACKNOWLEDGMENTS
This work was funded by the Austrian Science Foundation, grant no. P16699.
We acknowledge the expert technical assistance of Irene Goesler.
REFERENCES
Barbas, C. F., D. R. Burton, J. D. Scott, and G. J. Silverman. 2001. Phage display: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Brabec, M., G. Baravalle, D. Blaas, and R. Fuchs. 2003. Conformational changes, plasma membrane penetration, and infection by human rhinovirus type 2: role of receptors and low pH. J. Virol. 77:5370-5377.
Daniel, T. O., W. J. Schneider, J. L. Goldstein, and M. S. Brown. 1983. Visualization of lipoprotein receptors by ligand blotting. J. Biol. Chem. 258:4606-4611.
Esser, V., L. E. Limbird, M. S. Brown, J. L. Goldstein, and D. W. Russell. 1988. Mutational analysis of the ligand binding domain of the low density lipoprotein receptor. J. Biol. Chem. 263:13282-13290.
Fass, D., S. Blacklow, P. S. Kim, and J. M. Berger. 1997. Molecular basis of familial hypercholesterolaemia from structure of LDL receptor module. Nature 388:691-693.
Fisher, C., D. Abdul-Aziz, and S. C. Blacklow. 2004. A two-module region of the low-density lipoprotein receptor sufficient for formation of complexes with apolipoprotein E ligands. Biochemistry 43:1037-1044.
Gaus, K., A. Basran, and E. A. Hall. 2001. Assessment of the fifth ligand-binding repeat (LR5) of the LDL receptor as an analytical reagent for LDL binding. Analyst 126:329-336.
Gilbert, J. M., P. Bates, H. E. Varmus, and J. M. White. 1994. The receptor for the subgroup A avian leukosis-sarcoma viruses binds to subgroup A but not to subgroup C envelope glycoprotein. J. Virol. 68:5623-5628.
Guex, N., and M. C. Peitsch. 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18:2714-2723.
Herdy, B., L. Snyers, M. Reithmayer, P. Hinterdorfer, and D. Blaas. 2004. Identification of the human rhinovirus serotype 1A binding site on the murine low-density lipoprotein receptor by using human-mouse receptor chimeras. J. Virol. 78:6766-6774.
Hewat, E. A., and D. Blaas. 1996. Structure of a neutralizing antibody bound bivalently to human rhinovirus 2. EMBO J. 15:1515-1523.
Hewat, E. A., E. Neumann, J. F. Conway, R. Moser, B. Ronacher, T. C. Marlovits, and D. Blaas. 2000. The cellular receptor to human rhinovirus 2 binds around the 5-fold axis and not in the canyon: a structural view. EMBO J. 19:6317-6325.
Konecsni, T., L. Kremser, L. Snyers, C. Rankl, F. Kilar, E. Kenndler, and D. Blaas. 2004. Twelve receptor molecules attach per viral particle of human rhinovirus serotype 2 via multiple modules. FEBS Lett. 568:99-104.
Ledford, R. M., N. R. Patel, T. M. Demenczuk, A. Watanyar, T. Herbertz, M. S. Collett, and D. C. Pevear. 2004. VP1 sequencing of all human rhinovirus serotypes: insights into genus phylogeny and susceptibility to antiviral capsid-binding compounds. J. Virol. 78:3663-3674.
Marlovits, T. C., C. Abrahamsberg, and D. Blaas. 1998. Soluble LDL minireceptors—minimal structure requirements for recognition of minor group human rhinovirus. J. Biol. Chem. 273:33835-33840.
Marlovits, T. C., C. Abrahamsberg, and D. Blaas. 1998. Very-low-density lipoprotein receptor fragment shed from HeLa cells inhibits human rhinovirus infection. J. Virol. 72:10246-10250.
Mischak, H., C. Neubauer, B. Berger, E. Kuechler, and D. Blaas. 1988. Detection of the human rhinovirus minor group receptor on renaturing Western blots. J. Gen. Virol. 69:2653-2656.
Mischak, H., C. Neubauer, E. Kuechler, and D. Blaas. 1988. Characteristics of the minor group receptor of human rhinoviruses. Virology 163:19-25.
Moestrup, S. K., K. Kaltoft, L. Sottrup-Jensen, and J. Gliemann. 1990. The human alpha 2-macroglobulin receptor contains high affinity calcium binding sites important for receptor conformation and ligand recognition. J. Biol. Chem. 265:12623-12628.
Moser, R., L. Snyers, J. Wruss, J. Angulo, H. Peters, T. Peters, and D. Blaas. 2005. Neutralization of a common cold virus by concatemers of the third ligand binding module of the VLDL-receptor strongly depends on the number of modules. Virology 338:259-269.
Neubauer, C., L. Frasel, E. Kuechler, and D. Blaas. 1987. Mechanism of entry of human rhinovirus 2 into HeLa cells. Virology 158:255-258.
Neumann, E., R. Moser, L. Snyers, D. Blaas, and E. A. Hewat. 2003. A cellular receptor of human rhinovirus type 2, the very-low-density lipoprotein receptor, binds to two neighboring proteins of the viral capsid. J. Virol. 77:8504-8511.
Nykjaer, A., and T. E. Willnow. 2002. The low-density lipoprotein receptor gene family: a cellular Swiss army knife? Trends Cell Biol. 12:273-280.
Olson, N. H., P. R. Kolatkar, M. A. Oliveira, R. H. Cheng, J. M. Greve, A. McClelland, T. S. Baker, and M. G. Rossmann. 1993. Structure of a human rhinovirus complexed with its receptor molecule. Proc. Natl. Acad. Sci. USA 90:507-511.
Ronacher, B., T. C. Marlovits, R. Moser, and D. Blaas. 2000. Expression and folding of human very-low-density lipoprotein receptor fragments: neutralization capacity toward human rhinovirus HRV2. Virology 278:541-550.
Rudenko, G., L. Henry, K. Henderson, K. Ichtchenko, M. S. Brown, J. L. Goldstein, and J. Deisenhofer. 2002. Structure of the LDL receptor extracellular domain at endosomal pH. Science 298:2353-2358.
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.
Semler, B. L., and E. Wimmer. 2002. Molecular biology of picornaviruses. ASM Press, Washington, D.C.
Smith, G. P., and J. K. Scott. 1993. Libraries of peptides and proteins displayed on filamentous phage. Methods Enzymol. 217:228-257.
Smith, T. J., A. G. Mosser, and T. S. Baker. 1995. Structural studies on the mechanisms of antibody-mediated neutralization of human rhinovirus. Semin. Virol. 6:233-242.
Uncapher, C. R., C. M. Dewitt, and R. J. Colonno. 1991. The major and minor group receptor families contain all but one human rhinovirus serotype. Virology 180:814-817.
Verdaguer, N., I. Fita, M. Reithmayer, R. Moser, and D. Blaas. 2004. X-ray structure of a minor group human rhinovirus bound to a fragment of its cellular receptor protein. Nat. Struct. Mol. Biol. 11:429-434.
Vlasak, M., S. Blomqvist, T. Hovi, E. Hewat, and D. Blaas. 2003. Sequence and structure of human rhinoviruses reveal the basis of receptor discrimination. J. Virol. 77:6923-6930.
Vlasak, M., M. Roivainen, M. Reithmayer, I. Goesler, P. Laine, L. Snyers, T. Hovi, and D. Blaas. 2005. The minor receptor group of human rhinovirus (HRV) includes HRV23 and HRV25, but the presence of a lysine in the VP1 HI loop is not sufficient for receptor binding. J. Virol. 79:7389-7395.
Wang, Q. Y., W. Huang, K. Dolmer, P. G. Gettins, and L. Rong. 2002. Solution structure of the viral receptor domain of Tva and its implications in viral entry. J. Virol. 76:2848-2856.(Stephane Nizet, Juergen W)
Center for Anatomy and Cell Biology, Department for Nuclear Biology, Developmental Biology, and Functional Microscopy, Medical University of Vienna, Vienna, Austria
ABSTRACT
Minor group human rhinoviruses (HRVs) bind members of the low-density lipoprotein receptor family for cell entry. The ligand-binding domains of these membrane proteins are composed of various numbers of direct repeats of about 40 amino acids in length. Residues involved in binding of module 3 (V3) of the very-low-density lipoprotein receptor (VLDLR) to HRV2 have been identified by X-ray crystallography (N. Verdaguer, I. Fita, M. Reithmayer, R. Moser, and D. Blaas, Nat. Struct. Mol. Biol. 11:429-434, 2004). Sequence comparisons of the eight repeats of VLDLR with respect to the residues implicated in the interaction between V3 and HRV2 suggested that (in addition to V3) V1, V2, V5, and V6 also fulfill the requirements for interacting with the virus. Using a highly sensitive binding assay employing phage display, we demonstrate that single modules V2, V3, and V5 indeed bind HRV2. However, V1 does not. A single mutation from threonine 17 to proline converted the nonbinding wild-type form of V1 into a very strong binder. We interpret the dramatic increase in affinity by the generation of a hydrophobic patch between virus and receptor; in the presence of threonine, the contact area might be disturbed. This demonstrates that the interaction between virus and its natural receptors can be strongly enhanced by mutation.
INTRODUCTION
Human rhinoviruses (HRVs), the predominant pathogen causing common cold infections, are icosahedral particles composed of 60 copies each of the capsid proteins VP1 through VP4 and a single-stranded RNA genome of positive (messenger sense) polarity (for a review, see reference 28). Twelve serotypes, the minor group, use members of the low-density lipoprotein receptor (LDLR) family; 87 serotypes, the major group, use intercellular adhesion molecule 1 for cell entry (14, 31, 34). These membrane proteins are neither structurally nor functionally related. LDLR attaches to the highly exposed and accessible star-shaped dome at the fivefold axis of icosahedral symmetry, whereas intercellular adhesion molecule 1 binds via its first immunoglobulin-like domain within the canyon, a cleft encircling this plateau (12, 24). The geometry of the interaction between a fragment of the very-low-density lipoprotein receptor (VLDLR) and the minor group virus HRV2 has been determined to 16-? resolution by cryoelectron microscopy (22) and to 3.5-? resolution by X-ray crystallography (32); amino acid residues of the virus interacting with amino acid residues of VLDLR ligand-binding module 3 (V3) have been defined. This revealed a predominant role of K1224 in HRV2 (Lys224 in VP1) that interacts with W22 and the highly conserved acidic cluster with the sequence motive DxD/ExD (where x is any amino acid residue) in the V3 module.
VLDLR, like other members of the LDLR family, has the ligand-binding domain at the N terminus, followed by sequences exhibiting similarity to the epidermal growth factor precursor interspaced by YWTD motives. As seen in the X-ray structure of the exodomain of LDLR at low pH, these latter sequences are arranged in the form of a ?-propeller (26); due to the high sequence conservation, the same conformation is most likely present in the VLDLR. Proximal to the transmembrane sequence is an O-linked sugar domain. A sequence pattern in the cytoplasmic C terminus is responsible for internalization via the clathrin-coated pit pathway by binding to the adapter complex AP2. Other members of the family, such as LDLR-related protein (LRP), are involved not only in ligand internalization (e.g., LDLR is responsible for cholesterol homeostasis by endocytosis of apolipoprotein B [ApoB]-complexed lipids) but also in signal transduction (23).
The ligand-binding domain of VLDLR comprises eight imperfect repeats, each about 40 amino acids long (Fig. 1). Their conformation is maintained by a Ca2+ ion that is chelated by the carboxylates of Glu and Asp residues in the acidic cluster and two backbone oxygens. Further stabilization is achieved via three disulfide bonds present in each of the modules (5, 26).
We have shown previously that HeLa cells are protected against infection by minor-group HRVs by soluble forms of fragments of the ligand-binding domain of LDLR and VLDLR (15, 16, 25). Recombinant soluble VLDL minireceptors, including the first three repeats (V1 to V3) of human VLDLR, displayed inhibitory activity towards HRV2 infection, whereas a receptor fragment comprising repeats 4 to 6 (V456) failed to bind and neutralize HRV2 (25). Virus binding to a collection of receptor fragments was directly visualized by cryoelectron microscopy (12, 22). Using capillary electrophoresis methodology, it was recently demonstrated that 12 molecules of an artificial concatemer composed of five copies of repeat 3 arranged in tandem and fused to maltose-binding protein (MBP-V33333) bind per virus particle (13). This, together with the strong increase in affinity and virus-neutralizing activity with the number of concatemerized modules (20), suggests that more than one and (most probably) all five modules within the same molecule contribute to binding by establishing an oligovalent interaction with the symmetry-related binding sites on the virus. This also suggests that the natural receptors with their 7, 8, and 31 repeats that are present in LDLR, VLDLR, and LRP, respectively, might similarly attach in a multivalent manner, resulting in a 12-to-1 stoichiometry. However, as seen from our present data, it is unlikely that five consecutive modules in the native receptors participate, since some of the repeats lack binding activity, at least in VLDLR.
Analysis of a number of receptor fragments including parts of the natural receptors and several artificial concatemers by cryoelectron microscopy and image reconstruction revealed that V1, the first VLDLR repeat, does not bind HRV2; this was concluded from the absence of density attributable to V1 in reconstructed images of complexes between HRV2 and MBP-V123 or MBP-V122 and from the lack of complex formation between HRV2 and an artificial tandem concatemer of V1 (MBP-V111). Except from a report on the binding of repeat 5 of LDLR (L5) alone to LDL (7) and from data on Tva, a single-repeat receptor of avian leucosis-sarcoma virus type A (8, 35), binding of a single module to any ligand has not been observed. Furthermore, the X-ray data of a complex between HRV2 and V23 suggested attachment of individual modules but could not definitely exclude involvement of the second repeat (32). Using display of various repeats on fusion phages, we now demonstrate very weak but clearly detectable binding of individual single repeats to HRV2 and show that changing one amino acid residue in the nonbinding V1 transforms it into a strong binder.
MATERIALS AND METHODS
Production of fusion phage. DNA encoding selected VLDLR ligand-binding repeats was amplified by PCR using synthetic primers (Table 1) containing NcoI and NotI restriction sites. These sites are unique in the phagemid pCANTAB6 (Cambridge Antibody Technology, Cambridge, United Kingdom) and were thus used for directional cloning of the amplified PCR fragments. The resulting constructs encode the pIII phage protein carrying the respective minireceptors at its N terminus and a six-His tag, followed by a myc tag between the C terminus of pIII and the VLDLR modules. The phagemids were transformed into Escherichia coli TG1 (Amersham Pharmacia). Following selection on LB plates containing 100-μg/ml ampicillin, single colonies were picked and grown at 37°C overnight in 5 ml LB medium supplemented with the antibiotic. The next day, 50 ml of LB-ampicillin medium containing 1% glucose was inoculated, and the bacteria were infected with 15 μl of M13K07 helper phage (1013 particles/ml). After incubation for 30 min at 37°C without shaking to allow for infection by the helper phage, the culture was further incubated for 30 min at 37°C with shaking to allow for expression of the resistance genes. The bacteria were pelleted and resuspended in 100 ml LB medium containing 100-μg/ml ampicillin and 50-μg/ml kanamycin and grown overnight at 30°C. Bacteria were removed by low-speed centrifugation, and phage particles in the supernatant were concentrated by precipitation with polyethylene glycol and resuspended in water (29). Receptor fragments displayed on the phage were folded overnight by adjustment to 20 mM Tris-HCl (pH 8.0) and 150 mM NaCl (TBS) containing 10 mM CaCl2 (TBSC) supplemented with 1 mM cysteamine-0.1 mM cystamine. Phage was stored in this buffer at 4°C. Mutagenesis of phage-displayed V1 was carried out using a constant pair of external primers and the oligonucleotides listed in Table 1 by mutually primed synthesis, following standard protocols. Fusions of the modules with maltose-binding protein (e.g., V1pMal) (Table 1) were made essentially as described by Ronacher and colleagues (25).
Cloning of MBP-V1T17PV1T17P. DNA encoding the sequence of V1T17P was amplified by PCR using the primers V1-ConCat (Table 1). The resulting fragment was ligated into the vector V1pMal previously cut with XbaI and blunted with DNA polymerase I, large (Klenow) fragment. The fusion protein was expressed and folded as described above.
Western blotting and virus overlay blot. The concentration of the phage in the resuspended polyethylene glycol pellet was determined spectrophotometrically and as the number of CFU (29). The presence of the foreign protein fused to the pIII phage protein was assessed by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of the phage proteins, followed by electrotransfer onto a polyvinylidene difluoride (PVDF) membrane. To conserve the native structure of the receptors, nonreducing sample buffer was used throughout; under these conditions, LDLR can even be boiled in SDS without the loss of binding activity (3). After being blocked with TBS containing 2% nonfat dried milk and 2% bovine serum albumin (BSA) (blocking buffer), the membrane was incubated with a rabbit antiserum against the myc tag (Abcam, Ltd., Cambridge, United Kingdom) diluted 1:7,000, followed by incubation with goat anti-rabbit immunoglobulin G coupled to horseradish peroxidase (HRP; Bio-Rad Laboratories) diluted 1:10,000. Bound antibody was revealed by chemiluminescence using the ECL detection kit from Pierce, following the manufacturer's instructions. Binding of virus to the electrotransferred proteins was assessed by incubation of the blot with 100,000 cpm of [35S]methionine-cysteine-labeled HRV2 (17, 21) in TBSC supplemented with 1% BSA and 1% nonfat dried milk (incubation buffer) overnight at 4°C or, alternatively, for 2 h at room temperature. For competition experiments, the radiolabeled virus was preincubated with 1 μg MBP-V33333, an artificial concatemer of five copies of repeat 3 arranged in tandem and fused to the maltose-binding protein (20, 32) in 100 μl PBS prior to incubation of the membrane. The membrane was washed three times with TBSC containing 0.1% Tween 20 (wash buffer), air dried, and exposed to X-ray film.
Determination of binding of fusion phage to HRV2. Enzyme-linked immunosorbent assay (ELISA) vinyl assay plates from Costar or Maxisorb plates from Nunc were coated overnight at 4°C with 10-μg/ml MBP-V33333 in TBS. Wells were blocked with blocking buffer for 1 h and incubated with purified HRV2 at 10 μg/ml for 3 h at room temperature. This procedure minimizes structural changes of the virion that might occur upon direct coating to the plastic surface. After being washed with blocking buffer, phage (1013 particles/ml as determined by spectrophotometry) was added, and incubation was continued overnight at 4°C. Note that for V123 and the strongly binding mutant of V1, final concentrations of 2 x 1012 and 4 x 1012 particles/ml, respectively, were employed to not exceed an absorbance value (A450) of 1.8 upon photometric measurement. Following 10 washes with TBS containing 0.1% Tween 20, the wells were incubated for 90 min at room temperature with monoclonal antibody against M13 (anti-M13, diluted 1:6,000; Amersham Pharmacia), followed by secondary HRP-conjugated antibody against mouse immunoglobulin G (Bio-Rad) diluted 1:10,000. Bound phage was finally revealed by incubation for 15 min at room temperature with 0.1 mg/ml trimethylbenzidine and 0.03% H2O2 in 0.1 M Na acetate, pH 6.0. Color development was halted by the addition of 1 M H2SO4; A450 was determined with a microplate reader. Negative controls included use of 1-mg/ml BSA instead of phage or the addition of 20 mM EDTA to the phage suspension during the attachment reaction.
Alternatively to the ELISA determination, a biological test was carried out. Exponentially growing strain TG1 (100 μl) was added to the wells and incubated for 30 min at 37°C without shaking, followed by another 30 min at 37°C with shaking. Then, bacteria were removed and plated on ampicillin plates. The next day, colonies arising from infection with phage that had been retained in the wells were counted.
RESULTS
Sequence alignment of VLDLR modules suggests that V1, V2, V3, V5, and V6, but not V4, V7, and V8, can bind HRV2. During cryoelectron microscopy analysis of complexes between HRV2 and various soluble VLDLR fragments or module concatemers, we observed that V1 does not contribute to HRV2 binding. Whereas concatenation of several copies of V3 greatly enhanced its avidity toward HRV2, three copies of V1 fused in tandem to maltose-binding protein in the concatemer MBP-V111 were inactive with respect to virus attachment. Proper folding of the receptor was shown by its binding radioactive 45Ca2+ (19, 22). To understand the basis of this deficiency, we examined sequence alignments of all VLDLR repeats with respect to eventual conservation of binding residues, as seen in the X-ray structure of the complex between V23 and HRV2 (32). As depicted in Fig. 1, V1, V2, V3, V5, and V6 possess the tryptophan and the acidic cluster DxD/ExD, known to be key players in the binding reaction; a lysine strictly conserved in all minor group HRVs (33, 34) interacts with the receptor module via its aliphatic side chain with W22 and via its -amino group with the acidic cluster around the Ca2+ ion (see Fig. 4A) (32). This latter binding reaction has an electrostatic component contributed by the carboxylates that are involved in Ca2+ complexation but also an ionic component resulting from a salt bridge with the central aspartate-glutamate. In addition, W22 exhibits hydrophobic interactions with L1132 and with I1226 from the symmetry-related VP1 molecule. The tryptophan is present in all modules except V4, V7, and V8, where it is replaced by phenylalanine, arginine, and lysine, respectively. In addition, the acidic cluster is incomplete in V7 and V8. Therefore, the latter three modules should not be able to bind the virus. For V1, however, we found no reasonable explanation for the lack of binding from the sequence alone.
Phage display allows detection of binding of single repeats. Single repeats bind HRV2 extremely weakly; in virus overlay blots, as utilized to demonstrate affinity of a given receptor for the virus, single repeats only occasionally gave rise to barely visible bands. Similarly, in a cell protection assay only very high concentrations of the recombinant soluble single modules fused to MBP were able to marginally inhibit infection of HeLa cells with HRV2 (32). This is most probably related to avidity effects, since binding, as well as viral neutralization, is enormously increased by oligomerization of the modules. This is seen for native VLDLR and for concatenated identical repeats in artificial receptor derivatives (see above) (13, 20). We therefore sought a method that would allow easy detection of virus binding by single modules. V1, V2, V3, and V123 were cloned in the phagemid pCANTAB6 N terminal to the phage attachment protein pIII, as detailed in Materials and Methods. Recombinant phage was propagated in E. coli TG1 and concentrated by precipitation with polyethylene glycol. Microtiter plates, to which HRV2 was bound via MBP-V33333, a concatemer of V3 that exhibits high-affinity binding (32), were challenged with recombinant phage; binding was assessed either by an ELISA format by sequential incubation with anti-phage antibody, secondary HRP-conjugated antibody, and substrate (Fig. 2A) or by a biological assay in which attached phage was allowed to infect bacteria. The phage titer was subsequently determined by a colony-forming assay (Fig. 2B). Binding of the recombinant phage carrying the different receptor modules was weak but clearly detectable by both methods and showed, in agreement with earlier data, that V123 bound strongly, followed by V3 and V2. V1 did not detectably attach to HRV2 (not shown, but see below and Fig. 3). As expected from the strict requirement for Ca2+ ions for the structural integrity of the receptor, binding of the recombinant phage to HRV2 was abolished in the presence of 20 mM EDTA. Helper phage without a foreign protein fused to its pIII protein and replacement of phage with BSA were used as additional controls and showed only background binding. This latter control was also included in the biological assay to test for eventual carryover of infectious phage. The grading of the binding affinity of the modules was similar in both assays (compare Fig. 2A and B). Thus, the phage system was well suited to detect the interaction of very weakly binding single receptor repeats to HRV2.
A single mutation in V1 transforms the nonbinding wild-type module into a strongly binding module. Comparison of the sequences of the virus-binding modules V2 and V3 with the nonbinding module V1 drew our attention to the presence of R14 instead of Q (which is involved in the interaction between V3 and virus), of T17 instead of P, and of V28 instead of an acidic residue (E in V2 and D in V3). The reasons to change the arginine and the valine are obvious; threonine was replaced by proline, based on the assumption that this might have an effect on the local conformation of the binding epitope. Two V1 mutants were constructed by site-directed mutagenesis in which residues were exchanged for those present in V2. In V1T17P,V28E, T17 and V28 were replaced by P and E, respectively. In V1R14Q, R14 was replaced by Q. The constructs were cloned into the phagemid, and fusion phage was produced. As controls, V4 and V5 were also cloned into the phagemid and expressed. The result of an ELISA (Fig. 3A) revealed that wild-type V1 and V1R14Q, as well as V4, as expected from the absence of the essential W22 in this latter module, did not bind virus. As already demonstrated, V2 and V3 bound weakly and V123 bound strongly. In accordance with the presence of W22 and the complete acidic cluster, V5 also recognized HRV2 and bound even more strongly than V3. Most interestingly, however, the double mutant V1T17P,V28E bound much more strongly than V3 and V5. Again, using controls with BSA and helper phage and conducting the attachment reaction in the presence of EDTA showed very low levels of background binding, indicative of the specificity of the assay.
As mentioned above, strong binding of virus to a receptor can be assessed via a virus overlay blot in which receptor proteins are separated on a SDS-PAGE gel under nonreducing conditions and electrotransferred to a PVDF membrane. The membrane is subsequently incubated with radiolabeled virus, and bands are revealed by autoradiography. We previously used this method extensively to identify the minor group HRV receptor (17, 18) and to demonstrate binding of minor-group HRVs to various LDL receptor derivatives (15). However, we could only occasionally detect extremely weak virus binding to a single repeat upon heavily overloading the gel. Therefore, we wondered whether the mutations in V1 might have increased the binding affinity to such an extent as to allow for its detection in virus overlay blots. Fusion phage was dissociated into its components by being boiled in sample buffer without reducing agent. Phage carrying the single repeats V1, V2, V3, V4, and V5 were also included in the analysis. The material was separated on a SDS-PAGE gel and blotted, and fusion proteins were detected with antiserum against the myc tag. Figure 3B, top, shows that all fusion proteins were present in similar amounts. Proteins eventually recognized by the virus were revealed by incubation with radiolabeled HRV2 on an identical blot prepared in parallel (Fig. 3B, bottom). Whereas no virus binding to V1, V2, V4, and V5 was seen in this assay, a strong band appeared in the lane containing V1T17P,V28E, largely exceeding the intensity of V3 that was also visible at this high load. In disagreement with the data from the ELISA (Fig. 3A), V5 binding was not detected. This might be due to a large proportion of the protein present in a misfolded form in this sample, as suggested from the presence of a double band (Fig. 3B, top). When 20 mM EDTA was added to the incubation medium, no binding was detected (results not shown).
Replacement of threonine 17 by proline is responsible for the increase in affinity of V1 for HRV2. To cut down the change in the properties of V1 to a single amino acid, a mutant was constructed in which only T17 was replaced by a proline. Similar tests as described above showed that the single mutant essentially displayed the same affinity for the virus as the double mutant (not shown). This demonstrates that the V28E mutation has no effect on the affinity and that replacement of threonine 17 by proline is the reason for the enormous increase in affinity for HRV2.
MBP-V33333 competes with V1T17P for virus binding. The amount of recombinant protein expressed on phage was small. To obtain sufficient material for further experiments, the single mutant V1T17P was cloned into the pMal-2c vector downstream of MBP. As in other MBP fusion proteins (25), the presence of a His6 tag in the vector allowed for easy purification. MBP-V1T17P was expressed, purified, folded, run on a SDS-PAGE gel, and blotted onto a PVDF membrane as in the results shown in Fig. 3B. Binding of radiolabeled virus was then assessed in the absence and in the presence of MBP-V33333. As depicted in Fig. 3C, virus bound well to MBP-V1T17P (left), but its attachment was abolished upon addition of MBP-V33333. From this, and the loss of binding in the presence of EDTA (Fig. 2), we conclude that attachment of V1T17P to HRV2 is specific and occurs to a site overlapping with or identical to the binding site of V3 on the viral surface.
A dimer of V1T17P protects HeLa cells against infection with HRV2. We have demonstrated previously that soluble receptor derivatives fused to MBP inhibit viral infection and protect HeLa cells against lysis to various degrees (25, 32). However, for single repeats, the concentrations required for noticeable protection are very high compared to that of artificial concatemers. Thus, cell protection assays were carried out using MBP-V1T17P, essentially as described in a previous paper (32). However, despite of displaying much higher affinity for HRV2 than MBP-V3, MBP-V1T17P failed to protect HeLa cells against infection with HRV2 (Fig. 3D). Since concatemerization of even two V3 modules resulted in a strong increase in neutralization capacity, a dimer of V1T17P was constructed and expressed as a fusion with MBP (MBP-V1T17PV1T17P). This protein exhibited cell protection efficiency exceeding that of MBP-V3; whereas MBP-V3 protected to some extent at 25 μg/ml, MBP-V1T17PV1T17P displayed a clear effect at 1.6 μg/ml under the specified conditions (Fig. 3D).
DISCUSSION
Single repeats of the members of the LDL receptor family exhibit very low affinity for minor group HRVs and other ligands such as ApoE complexed with lipid. It was thus assumed that at least two consecutive modules are required for binding (6, 15). This was corroborated by structure determination of a complex between HRV2 and a receptor fragment with two consecutive receptor modules (V23) by cryoelectron microscopy (22) and by X-ray crystallography (32). Although the resolution of 3.5 ? allowed identification of the interacting amino acid residues, it was not possible to definitely decide whether only single repeats were bound or whether, at least to a minor extent, two repeats were also attached simultaneously to two of the five symmetry-related binding sites at the vertex of the viral icosahedron. At least V3 alone binds very weakly to HRV2, since it is able at high concentration to marginally protect HeLa cells against infection, but the concatemer MBP-V33333 is about 10,000 times more effective (32). This increased neutralization capacity indeed arises from multimodule attachment (13, 20).
To first compare the binding affinities of the various repeats, we developed an assay utilizing fusion phage. Filamentous phage such as M13 contains 5 copies of the attachment protein pIII at the tip and about 2,000 copies of the coat protein pVIII among several other minor proteins (1). Presence of no more than two copies of a pIII fusion protein within a phage particle is compatible with attachment to the host bacteria and allows for propagation in the presence of helper phage that provides the necessary copies of wild-type pIII. We reasoned that interaction of the receptor moiety with immobilized virus might be detected with high sensitivity via antibodies recognizing pVIII or via a biological assay that, at least in theory, can detect a single infectious phage particle. Therefore, we constructed fusion phage displaying various single-receptor repeats and combinations and showed that low-affinity interactions could indeed be measured by this assay. We found that V3 bound more strongly than V2 and that V1 did not bind at all (Fig. 2 and 3).
We then addressed the question of why V1 failed to bind despite possessing all specific residues believed to be necessary for interaction (Fig. 1). Comparison of the sequences of the modules drew our attention to three residues that were subsequently exchanged for those that are at least functionally conserved at the equivalent positions in V2 and V3 (Fig. 1). ELISA and a colony-forming assay with fusion phage carrying the mutated modules demonstrated that a change from threonine 17 to proline and from valine 28 to glutamic acid resulted in a module that had acquired binding activity by much exceeding that of V3 (Fig. 3). Finally, a mutant with the single change from threonine 17 to proline was constructed and expressed as a fusion with maltose-binding protein. This mutant was as active as the double mutant in virus binding.
Usually, one or two copies of the fusion protein are displayed on phages. Therefore, phage binding might have occurred due to the presence of two copies of the module in spatial vicinity that could attach simultaneously to symmetry-related sites on the virion. Binding of MBP-V1T17P excludes bivalent attachment and demonstrates that a single copy of a binding module can indeed attach to the virus. V1T17P and V3 bind to the same or at least to a closely overlapping site, since they compete (Fig. 3C).
The particular proline in question is rather conserved and is present in Tva, in human LDLR (L2 and L4), VLDLR (V2, V3, and V5) and LRP (R2, R4, R5, R7, R8, R10, R15, R17, R18, R21 through R26, R28, and R29). However, its absence from L5 suggests that it is not essential, since L5 makes a strong contribution to binding of ApoB and ApoE to LDLR (4) and is clearly involved in binding HRV1A, another minor-group HRV (10). Therefore, the presence of the proline, together with W22 and the acidic cluster, is still not sufficient to form a binding epitope.
What is the basis of the affinity increase upon exchange of threonine 17 for proline? Inspection of residues in the complex of V3 and HRV2 within 6 ? of the proline revealed that only leucine 1226 and isoleucine 1132, contributed from the next symmetry related virus subunit, are in its vicinity (Fig. 4B). Together with the neighboring tryptophan 22, this results in a hydrophobic cluster that might be interrupted upon exchange of the proline with threonine. We thus believe that the hydrophobic core is at least in part responsible for HRV2 binding to V2, V3, V5, and most probably V1T17P. A model of V1 and of V1T17P calculated with SWISS-MODEL (27) showed that W22 and P17 might adopt an arrangement very similar to that in V3 (data not shown). Conformational effects might contribute to the affinity increase upon mutation of T17 as well. Nevertheless, in the absence of real structural data this interpretation must remain speculative. Proline 17 (or proline 19 in V1) is absent from LDLR modules L4 and L5. When fused to phage, binding of these single LDLR repeats to virus could not be detected, suggesting that the proline might indeed be responsible, at least in part, to strong binding (data not shown). It is noteworthy that the proline is replaced by a histidine in LDLR module L5. According to Rudenko and colleagues, this histidine forms a salt bridge with a glutamic acid of the YWTD ?-propeller at pH 5.3 and thus participates in intramolecular interactions (26). In the case of the virus receptor complex, at pH 7.4, the histidine is most probably not protonated and thus fits into the hydrophobic cluster. Upon exposure to the low-pH environment of the endosome, protonation might therefore be in part responsible for dissociation of HRV2 from LDLR at low pH (2).
Cell protection assays showed that the increase in binding affinity of the single repeat was not correlated with neutralization efficiency, since the weakly binding MBP-V3 neutralized very weakly and the strongly binding MBP-V1T17P did not neutralize at all. Taken together with other binding and neutralization data (13, 20, 25), this indicates that viral neutralization strongly depends on simultaneous binding of more than one module to the viral surface. Therefore, it most probably results from stabilization of the viral capsid similar to that exerted by bivalent antibody binding, which inhibits uncoating (11, 30). A receptor module bridges two symmetry related sites on the viral surface (22). Subtle differences in the binding forces directed towards the two sites might be responsible for the difference in neutralization efficiency.
Although the data on competition with V33333 indicate that V1T17P and V3 attach to the same epitope, the absence of noticeable cell protection by V1T17P is puzzling. To definitely exclude binding of V1T17P to a site different from that used by V3, a dimer of V1T17P was also constructed. This dimer indeed protected HeLa cells against infection with HRV2, lending definite proof to its binding being specific. However, due to the cloning procedure, the linkers between the modules differ from those between the V3 modules in MBP-V33; instead of PG in V33, the linker in V1T17PV1T17P is VKSR. The different composition and distance might affect binding strength. A comparison of the respective concentrations resulting in neutralization is thus not possible.
In summary, we have demonstrated that mutation of a single amino acid residue in module V1 of VLDLR transforms a nonbinding ligand-binding repeat into a strongly binding module. The amino acid residue changed is to some extent involved in interaction with the virus surface by contributing to a hydrophobic cluster. However, it might also influence the overall conformation of the module, thereby modifying the interacting surface. A dimer of V1T17P had the capacity to neutralize virus. Comparison of the kinetics of virus binding of V1T17P and V3 might allow unraveling the parameters underlying viral binding and neutralization. Experiments to determine the kinetic on and off rates by using surface plasmon resonance and nuclear magnetic resonance techniques are currently in progress.
ACKNOWLEDGMENTS
This work was funded by the Austrian Science Foundation, grant no. P16699.
We acknowledge the expert technical assistance of Irene Goesler.
REFERENCES
Barbas, C. F., D. R. Burton, J. D. Scott, and G. J. Silverman. 2001. Phage display: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Brabec, M., G. Baravalle, D. Blaas, and R. Fuchs. 2003. Conformational changes, plasma membrane penetration, and infection by human rhinovirus type 2: role of receptors and low pH. J. Virol. 77:5370-5377.
Daniel, T. O., W. J. Schneider, J. L. Goldstein, and M. S. Brown. 1983. Visualization of lipoprotein receptors by ligand blotting. J. Biol. Chem. 258:4606-4611.
Esser, V., L. E. Limbird, M. S. Brown, J. L. Goldstein, and D. W. Russell. 1988. Mutational analysis of the ligand binding domain of the low density lipoprotein receptor. J. Biol. Chem. 263:13282-13290.
Fass, D., S. Blacklow, P. S. Kim, and J. M. Berger. 1997. Molecular basis of familial hypercholesterolaemia from structure of LDL receptor module. Nature 388:691-693.
Fisher, C., D. Abdul-Aziz, and S. C. Blacklow. 2004. A two-module region of the low-density lipoprotein receptor sufficient for formation of complexes with apolipoprotein E ligands. Biochemistry 43:1037-1044.
Gaus, K., A. Basran, and E. A. Hall. 2001. Assessment of the fifth ligand-binding repeat (LR5) of the LDL receptor as an analytical reagent for LDL binding. Analyst 126:329-336.
Gilbert, J. M., P. Bates, H. E. Varmus, and J. M. White. 1994. The receptor for the subgroup A avian leukosis-sarcoma viruses binds to subgroup A but not to subgroup C envelope glycoprotein. J. Virol. 68:5623-5628.
Guex, N., and M. C. Peitsch. 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18:2714-2723.
Herdy, B., L. Snyers, M. Reithmayer, P. Hinterdorfer, and D. Blaas. 2004. Identification of the human rhinovirus serotype 1A binding site on the murine low-density lipoprotein receptor by using human-mouse receptor chimeras. J. Virol. 78:6766-6774.
Hewat, E. A., and D. Blaas. 1996. Structure of a neutralizing antibody bound bivalently to human rhinovirus 2. EMBO J. 15:1515-1523.
Hewat, E. A., E. Neumann, J. F. Conway, R. Moser, B. Ronacher, T. C. Marlovits, and D. Blaas. 2000. The cellular receptor to human rhinovirus 2 binds around the 5-fold axis and not in the canyon: a structural view. EMBO J. 19:6317-6325.
Konecsni, T., L. Kremser, L. Snyers, C. Rankl, F. Kilar, E. Kenndler, and D. Blaas. 2004. Twelve receptor molecules attach per viral particle of human rhinovirus serotype 2 via multiple modules. FEBS Lett. 568:99-104.
Ledford, R. M., N. R. Patel, T. M. Demenczuk, A. Watanyar, T. Herbertz, M. S. Collett, and D. C. Pevear. 2004. VP1 sequencing of all human rhinovirus serotypes: insights into genus phylogeny and susceptibility to antiviral capsid-binding compounds. J. Virol. 78:3663-3674.
Marlovits, T. C., C. Abrahamsberg, and D. Blaas. 1998. Soluble LDL minireceptors—minimal structure requirements for recognition of minor group human rhinovirus. J. Biol. Chem. 273:33835-33840.
Marlovits, T. C., C. Abrahamsberg, and D. Blaas. 1998. Very-low-density lipoprotein receptor fragment shed from HeLa cells inhibits human rhinovirus infection. J. Virol. 72:10246-10250.
Mischak, H., C. Neubauer, B. Berger, E. Kuechler, and D. Blaas. 1988. Detection of the human rhinovirus minor group receptor on renaturing Western blots. J. Gen. Virol. 69:2653-2656.
Mischak, H., C. Neubauer, E. Kuechler, and D. Blaas. 1988. Characteristics of the minor group receptor of human rhinoviruses. Virology 163:19-25.
Moestrup, S. K., K. Kaltoft, L. Sottrup-Jensen, and J. Gliemann. 1990. The human alpha 2-macroglobulin receptor contains high affinity calcium binding sites important for receptor conformation and ligand recognition. J. Biol. Chem. 265:12623-12628.
Moser, R., L. Snyers, J. Wruss, J. Angulo, H. Peters, T. Peters, and D. Blaas. 2005. Neutralization of a common cold virus by concatemers of the third ligand binding module of the VLDL-receptor strongly depends on the number of modules. Virology 338:259-269.
Neubauer, C., L. Frasel, E. Kuechler, and D. Blaas. 1987. Mechanism of entry of human rhinovirus 2 into HeLa cells. Virology 158:255-258.
Neumann, E., R. Moser, L. Snyers, D. Blaas, and E. A. Hewat. 2003. A cellular receptor of human rhinovirus type 2, the very-low-density lipoprotein receptor, binds to two neighboring proteins of the viral capsid. J. Virol. 77:8504-8511.
Nykjaer, A., and T. E. Willnow. 2002. The low-density lipoprotein receptor gene family: a cellular Swiss army knife? Trends Cell Biol. 12:273-280.
Olson, N. H., P. R. Kolatkar, M. A. Oliveira, R. H. Cheng, J. M. Greve, A. McClelland, T. S. Baker, and M. G. Rossmann. 1993. Structure of a human rhinovirus complexed with its receptor molecule. Proc. Natl. Acad. Sci. USA 90:507-511.
Ronacher, B., T. C. Marlovits, R. Moser, and D. Blaas. 2000. Expression and folding of human very-low-density lipoprotein receptor fragments: neutralization capacity toward human rhinovirus HRV2. Virology 278:541-550.
Rudenko, G., L. Henry, K. Henderson, K. Ichtchenko, M. S. Brown, J. L. Goldstein, and J. Deisenhofer. 2002. Structure of the LDL receptor extracellular domain at endosomal pH. Science 298:2353-2358.
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.
Semler, B. L., and E. Wimmer. 2002. Molecular biology of picornaviruses. ASM Press, Washington, D.C.
Smith, G. P., and J. K. Scott. 1993. Libraries of peptides and proteins displayed on filamentous phage. Methods Enzymol. 217:228-257.
Smith, T. J., A. G. Mosser, and T. S. Baker. 1995. Structural studies on the mechanisms of antibody-mediated neutralization of human rhinovirus. Semin. Virol. 6:233-242.
Uncapher, C. R., C. M. Dewitt, and R. J. Colonno. 1991. The major and minor group receptor families contain all but one human rhinovirus serotype. Virology 180:814-817.
Verdaguer, N., I. Fita, M. Reithmayer, R. Moser, and D. Blaas. 2004. X-ray structure of a minor group human rhinovirus bound to a fragment of its cellular receptor protein. Nat. Struct. Mol. Biol. 11:429-434.
Vlasak, M., S. Blomqvist, T. Hovi, E. Hewat, and D. Blaas. 2003. Sequence and structure of human rhinoviruses reveal the basis of receptor discrimination. J. Virol. 77:6923-6930.
Vlasak, M., M. Roivainen, M. Reithmayer, I. Goesler, P. Laine, L. Snyers, T. Hovi, and D. Blaas. 2005. The minor receptor group of human rhinovirus (HRV) includes HRV23 and HRV25, but the presence of a lysine in the VP1 HI loop is not sufficient for receptor binding. J. Virol. 79:7389-7395.
Wang, Q. Y., W. Huang, K. Dolmer, P. G. Gettins, and L. Rong. 2002. Solution structure of the viral receptor domain of Tva and its implications in viral entry. J. Virol. 76:2848-2856.(Stephane Nizet, Juergen W)