Skip to main content
NCBI home page
Journal of Virology logoLink to Journal of Virology
. 2008 Jun 4;82(16):7923–7931. doi: 10.1128/JVI.00754-08

Structural Variations in Species B Adenovirus Fibers Impact CD46 Association ▿ †

Lars Pache 1,3, Sangita Venkataraman 2, Vijay S Reddy 2, Glen R Nemerow 1,*
PMCID: PMC2519554  PMID: 18524830

Abstract

A majority of species B adenoviruses (Ads) use CD46 as their primary receptor; however, the precise mechanisms involved in the binding of different Ad types to CD46 have not been resolved. Although previous studies indicate close similarities between two members of species B2 Ads in their usage of CD46, our current investigations revealed a surprisingly low CD46 binding affinity of the species B1 Ad16 fiber knob (equilibrium dissociation constant of 437 nM). We determined the crystal structure of the Ad16 fiber knob and constructed a model of this protein in complex with CD46. A comparison of this model to that of the CD46-Ad11 complex revealed structural differences in the FG and IJ loops that are part of the CD46 binding site. An analysis of a panel of recombinant fiber knobs with mutations targeting these regions in Ad16 and Ad11 uncovered a major contribution of the FG loop on CD46 binding. Two extra residues in the FG loop of the Ad16 fiber significantly reduce receptor interaction. Although avidity effects permit the use of CD46 on host cells by Ad16, virus binding occurs with lower efficiency than with B2 Ad types. The longer FG loop of the Ad16 fiber knob also is shared by other species B1 Ad fibers and, thus, may contribute to the low CD46 binding efficiencies observed for these Ad types. Our findings provide a better understanding of how different Ad types associate with CD46 and could aid in the selection of specific Ad fibers for more efficient Ad gene delivery vectors.

Species B adenoviruses (Ads) cause infections of the upper respiratory tract (primarily species B1) as well as the kidney and urinary tract (primarily species B2). This division of species B correlates more with tissue tropism than with receptor usage. The majority of species B Ads use CD46 as their primary cellular receptor (9, 29, 31), with the two N-terminal extracellular domains of CD46 mediating fiber association (7, 10). In addition to CD46, a poorly characterized cell surface molecule (designated receptor X) has been proposed to mediate the attachment and infection of certain species B viruses. A new classification based on the usage of either one or both of these receptors has been proposed (35).

In addition to their disease associations, Ad vectors are being used in clinical gene transfer trials. The majority of these vectors is based on Ad type 5 (Ad5), a member of subgroup C that utilizes coxsackie adenovirus receptor (CAR) as its primary receptor (2). A challenge for the use of Ad5-based vectors is the limited expression of CAR in certain target cell types, such as hematopoietic cells (30) or certain malignant tumor isolates (1, 8, 13). Moreover, Ad5-based vectors often are prone to neutralization by preexisting anti-capsid antibodies. Considering these obstacles, vectors based on species B may prove to be useful alternatives (28). In this regard, CD46 is ubiquitously expressed, allowing gene transfer to a broader range of target cells. Further, neutralizing antibodies to species B Ad types are present at lower levels compared to those directed against Ad5 (36). To date, most studies have focused on the mechanism of receptor binding by two species B2 Ads, Ad11 and Ad35. Recent investigations found that both viruses engage CD46 in a similar fashion and with similar efficiencies (23, 24, 37). Based on the cocrystal structure of the Ad11 fiber knob in complex with CD46, a large continuous binding region in the virus fiber was shown to encompass the DG, HI, and IJ loops of the fiber knob. Moreover, a highly conserved arginine residue in the HI loop was found to form a critical salt bridge with CD46, and mutations targeting this amino acid profoundly reduced receptor association (37). The degree to which other members of species B, especially B1, use CD46 has not been extensively investigated.

In this study, we compared the association of the Ad16 fiber knob, a member of species B1, with CD46 to that of the Ad11 knob, and we provide a detailed kinetic analysis for the binding sites of both fiber types. In addition, we solved the crystal structure of the Ad16 fiber knob and generated a comparative model of this protein in complex with CD46. Based on the model of this complex as well as targeted mutagenesis studies of recombinant fiber knobs, we identified a critical region in the FG loop of different species B Ad fibers that plays a key role in the interaction with CD46. These studies aid our understanding of Ad-CD46 association and may help guide the future development of improved viral vectors with enhanced receptor binding.

MATERIALS AND METHODS

Cells and viruses.

All cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (Omega Scientific, Tarzana, CA), 10 mM HEPES, pH 7.55, 4 mM l-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 100 U/ml penicillin, and 100 μg/ml streptomycin. Tissue culture reagents were obtained from Invitrogen (Carlsbad, CA). CHO-K1 cells were stably transfected with an expression plasmid containing the C2 isoform of CD46 (25) or with an empty vector as a control and were maintained in DMEM supplemented as described above, with the addition of 0.5 mg/ml G418 sulfate. 293EBNA cells were stably transfected with an expression plasmid containing the extracellular domain of the C isoform of CD46 (sCD46) (38) and maintained in DMEM that was supplemented as described above with the addition of 5 μg/ml puromycin.

Ad11p (Slobitsky strain) was obtained from the ATCC (Manassas, VA) and propagated in A549 cells. Ad5.16F (21) was propagated in 293 cells. The propagation and purification of the viruses have been described previously (38). The concentrations of purified viruses were determined by UV absorbance measurements. Viruses were fluorescently conjugated as previously described (32), using an Alexa Fluor 488 labeling kit (Invitrogen, Carlsbad, CA) under conditions that maintained at least 90% of the original infectivity.

Protein expression and purification.

The generation of cDNA encoding recombinant fiber knob proteins corresponding to residues 118 to 325 of the Ad11p fiber (Slobitzki strain) and residues 151 to 353 of the Ad16 fiber has been described previously (21, 23). Mutations in the fiber knob constructs were introduced using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Primers were designed under the consideration of guidelines by Zheng et al. (40). All fiber knob constructs were expressed in Escherichia coli BL21(DE3) cells (Invitrogen, Carlsbad, CA) and purified by metal affinity chromatography. The removal of the 6× His tag and ion exchange chromatography were done as previously described (23). To verify expression and trimerization, fiber knobs diluted in loading buffer (80 mM Tris, pH 6.8, 2% sodium dodecyl sulfate, 10% glycerol, 0.2% bromophenol blue, 5% β-mercaptoethanol) were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis with or without prior boiling. Protein bands were stained with SimplyBlue protein stain (Invitrogen, Carlsbad, CA).

The extracellular domain of the C isoform of CD46 (sCD46) was expressed in mammalian 293EBNA cells and purified as described previously (38).

Surface plasmon resonance analysis of CD46-fiber binding.

Surface plasmon resonance experiments were performed on a Biacore 2000 system (GE Healthcare, Piscataway, NJ) at 25°C. CM5 sensor chips, N-hydroxysuccinimide (NHS), 1-ethyl 3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC), ethanolamine hydrochloride, Surfectant P20, and HSB-EP buffer were obtained from GE Healthcare (Piscataway, NJ). Carboxymethyl-dextran sodium salt (CM-dextran) was obtained from Sigma-Aldrich (St. Louis, MO). Between 50 and 80 resonance units of purified fiber knob protein diluted in 20 mM sodium acetate, pH 4.0, was covalently immobilized on CM5 sensor chips using standard amine coupling chemistry in HSB-EP running buffer as recommended by the manufacturer. A reference flow cell was generated by activation and subsequent blocking without the immobilization of the fiber knob. Binding studies were carried out by injecting sCD46 at concentrations of 49.6, 24.8, 12.4, 6.2, 3.1, 1.6, and 0.8 nM at flow rates of 20 and 40 μl/min (for the Ad11 fiber knob) or at concentrations of 827, 621, 414, 207, 103, 52, and 26 nM at flow rates of 40 and 50 μl/min (for the Ad16 fiber knob) in running buffer consisting of 10 mM Tris, pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% Surfectant P20, and 100 μg/ml CM-dextran. All concentrations were analyzed in two replicate injections. Additional injections at flow rates ranging from 20 to 75 μl/min were used to verify that binding was not influenced by mass transport effects (19). The response from a reference flow cell was subtracted from the sensograms. Kinetic constants were obtained by a global fitting of the data to a 1:1 interaction model. Kinetic analysis included double referencing (20). Equilibrium data were extracted from sensograms to calculate the equilibrium dissociation constants of Ad16 fiber knob and the mutant Ad11 fiber knob, FK11 246AA247, independently of the kinetic analysis. Data analysis was performed using the software Scrubber 2.0 (BioLogic Software, Campbell, Australia).

Crystallization, data collection, and processing.

The purified Ad16 fiber knob was concentrated to 10 mg/ml in 20 mM Tris, pH 8.0, 100 mM NaCl in Ultrafree-0.5 PBTK centrifugal filter units (Millipore, Billerica, MA). Equal volumes (1.5 μl) of the fiber knob sample and the reservoir solution containing 100 mM HEPES (pH 7.5), 0.7 M NaH2PO4, and 0.85 M KH2PO4 were used in the crystallization setups. Sitting drop vapor diffusion yielded crystals after 1 week at 22.5°C. X-ray diffraction data for Ad16 crystals were collected at the GM/CA CAT beamline at Advanced Photon Source, Chicago, IL. The crystals were presoaked in the mother liquor containing 25% glycerol for 5 min and flash-frozen prior to data collection. The X-ray diffraction data consisted of a set of 176 images collected from a single crystal on a MAR300 charge-coupled device detector with an oscillation angle of 1°, a wavelength of 0.979 Å, and a crystal-to-detector distance of 350 mm. The diffraction data was processed and scaled using the HKL2000 suite of programs (22) to a resolution of 2.4 Å in the space group P6522. The overall completeness of the data was 99%, with an Rmerge of 11.2%. The final data reduction statistics are shown in Table 1.

TABLE 1.

Data statistics

Parametera Datab
Space group P6522
Resolution range (Å) 50-2.4 (2.5-2.4)
Unit cell
    a, b (Å) 167.830
    c (Å) 262.161
No. of unique reflections 84,274
I/σ(I) 45.4 (2)
Completeness (%) 99.3 (98.5)
Rmergeb (%) 11.2 (88.1)
R factor/free R (%) 24.5/27.2
No. of protein atoms/asymmetric unit 9,168
No. of solvent atoms/asymmetric unit 100
RMSDc bond (Å) 0.007
RMSD angles (°) 1.5
Average B factor for all atoms (Å2) 65.3
Ramachandran statistics
    Most favored region (%) 83.2
    Additionally allowed region (%) 14.3
    Generally allowed region (%) 1.0
    Disallowed region (%) 1.5
Open in a new tab
a

Rmerge = Σhkl |I − <I>|/ΣI; hkl, indices of the reflections; I, intensity; RMSD, root mean square deviation.

b

Data in parentheses refer to the last resolution shell.

Data analysis and structure determination.

The Ad16 fiber knob structure was solved using the molecular replacement program Phaser (26, 34), with a trimer of the Ad11 fiber knob (Protein Data Bank no. 2O39) (24) as the search model. Due to the presence of two trimers in the asymmetric unit, the phases were improved by twofold NCS averaging using CCP4 (5) and the RAVE suite of programs (15). The model was built using the program O (14) and later refined using CNS (3). The data and the final refinement statistics are shown in Table 1. The program TOP3d from the CCP4 suite (5) was used for structure alignments and to obtain the NCS matrix that was used in the density modification. Structure visualization, analysis, and figure generation was done using PyMOL (W. L. DeLano [http://pymol.sourceforge.net/]).

Sequence alignments.

The amino acid sequences of fiber knobs were aligned using the program ClustalW2 (4). Structure-based alignments were performed using the program TM-align (39). A phylogenetic tree diagram was created using the PHYLIP program package (version 3.6; J. Felsenstein, University of Washington, Seattle [http://evolution.genetics.Washington.edu/phylip.html]).

Cell binding and competition assays.

Fiber knob competition studies to evaluate CD46 interactions were performed as previously described (23). Briefly, CHO cells expressing CD46 were detached from culture flasks with Versene (Invitrogen, Carlsbad, CA) and rinsed with serum-free DMEM. Cells (105) were deposited into individual wells of a 96-well plate and resuspended in 50 μl serum-free medium containing various amounts of fiber knob. After 30 min of incubation on ice, 3 × 109 fluorescently labeled Ad particles in 50 μl medium were added to each well for 45 min on ice. After being pelleted, the cells were fixed in phosphate-buffered saline (PBS) containing 4% paraformaldehyde (EMS, Ft. Washington, PA) and diluted in PBS containing 0.2% bovine serum albumin and 0.2% sodium azide (Sigma-Aldrich, St. Louis, MO). A total of 10,000 cells were acquired on a FACScan flow cytometer and analyzed using CellQuest software (BD Biosciences, Billerica, MA).

Structure accession number.

The Ad16 fiber knob coordinates and structure factors have been deposited in the Protein Data Bank with accession code 3CNC.

RESULTS

Kinetic analysis of fiber knob binding to CD46.

Although most species B Ads utilize CD46 as their primary receptor, little is known of their relative binding efficiencies. Therefore, surface plasmon resonance (Biacore) was used to measure the affinity and kinetics of the interaction between the Ad11 or Ad16 fiber knob and a recombinant CD46 protein consisting of the entire extracellular domain of CD46 (sCD46). Previous studies indicated that fiber knob binding to CD46 immobilized on a biosensor surface is complex and cannot easily be modeled (37). Thus, we covalently coupled Ad fiber knobs to the biosensor surface and measured the association of soluble sCD46. These experiments measured the binding of a CD46 molecule to a single binding site on the fiber knob without being influenced by potential avidity effects. The association rate (ka) determined for sCD46 binding to the Ad11 fiber knob (Fig. 1A) was 1.865 × 106 M−1 s−1, and the dissociation constant (Kd) was found to be 0.0108 s−1, resulting in an equilibrium dissociation constant (KD) of 5.77 nM. The ka for sCD46 binding to the Ad16 fiber knob (Fig. 1B) was similar to that of the Ad11 fiber knob, with a measured ka of 1.315 × 106 M−1 s−1. In striking contrast, however, the Kd obtained for the Ad16 fiber knob was more than 50-fold higher than for the Ad11 fiber, with a Kd of 0.575 s−1. These kas and Kds resulted in a calculated KD of 437 nM for the binding of sCD46 to the Ad16 fiber knob, a more than 70-fold lower affinity than that of sCD46 with the Ad11 fiber knob. As the Kd determined for the Ad16 fiber knob-sCD46 interaction is at the extreme edge of the recommended range for the biosensor used for these studies, we used equilibrium binding analysis (18) to verify the KD independently of the on- and off-rates (Fig. 1C). Nonlinear regression analysis revealed a KD of 431 nM, confirming the overall affinity (KD) calculated from the kas and Kds.

FIG. 1.

FIG. 1.

Open in a new tab

Biacore analysis of sCD46 binding to Ad fiber knob protein. sCD46 was injected at various concentrations over a sensor surface containing the immobilized Ad11 (A) or Ad16 (B) fiber knob. Measured response curves were overlaid with a global fit to a 1:1 interaction model to obtain kinetic constants. Sensograms show one of duplicate sets of injections used for curve fitting. (C) The amount of sCD46 bound to the Ad16 fiber knob at equilibrium was plotted against the concentration of injected sCD46. The KD was obtained by nonlinear curve fitting of the Langmuir binding model. Values in parentheses represent the error of the fitting procedure in the final digit.

Comparison of Ad11 and Ad16 binding to CD46-expressing cells.

The Biacore studies described above were designed to measure the binding of soluble CD46 to a single binding site on the fiber knob. Under physiological conditions, however, the binding of the fiber knob to the cell surface likely is influenced by avidity effects, as revealed by the crystal structure of the Ad11 fiber knob complexed to CD46 (24). To assess receptor binding under more physiologic conditions, we analyzed fiber knob binding to CHO cells exogenously expressing CD46. In initial studies, we asked whether CD46 is required for the efficient attachment of Ad11 or of an Ad5 vector equipped with the type 16 fiber (Ad5.16F). Both viruses efficiently bound to CHO-CD46 cells but not to CD46-negative CHO-NULL cells, demonstrating that CD46 is indeed required for the cell attachment of these Ad serotypes (Fig. 2A). Next we compared the relative binding efficiency of fluorescently labeled Ad11 and Ad5.16F virions to CD46-expressing CHO cells in the presence of increasing amounts of purified Ad11 or Ad16 fiber knobs (Fig. 2B, C). Flow cytometric analyses showed that the Ad11 fiber knob was significantly better at inhibiting Ad11 as well as Ad5.16F attachment to CD46-CHO cells. The CAR binding Ad5 fiber knob did not block virus attachment, confirming the specific interaction of these species B Ads with CD46 (data not shown). These findings indicated that the Ad11 and Ad16 fibers possess distinct binding efficiencies to CD46 on the cell surface.

FIG. 2.

FIG. 2.

Open in a new tab

Functional analysis of fiber knob binding to CD46 on cells. (A) Stably transfected CHO-K1 cells expressing either CD46 or an empty vector (NULL) were incubated in the presence of 30,000 Ad11 or Ad5.16F virus particles per cell. The mean fluorescence intensity, measured by flow cytometry, was normalized to 100 for CD46-expressing cells incubated with virus. (B and C) CD46-expressing CHO cells were preincubated with different amounts of the Ad11 (FK11) or Ad16 (FK16) fiber knob prior to incubation with Ad11 (B) or Ad5.16F (C) virus. After subtracting the background fluorescence of cells in the absence of virus, the fluorescence of cells incubated with virus in the absence of fiber knob was defined as 100%. Data points represent the means from triplicate samples. Error bars represent standard deviations.

Structure determination of the Ad16 fiber knob.

In order to uncover the underlying molecular basis of the relatively low CD46 binding efficiency of the Ad16 fiber knob, we determined the crystal structure of the Ad16 fiber knob by molecular replacement using the Ad11 fiber knob as a search model (Table 1). Similar to the structure of the Ad11 fiber knob, the Ad16 fiber knob is a trimer (Fig. 3A, B), and there are two trimers in the crystallographic asymmetric unit. The structure of Ad16FK showed significant deviation from that of Ad11 at the surface loops (Fig. 3C). In order to gain a better understanding of the mode of CD46 binding, the structure of the Ad16 fiber knob was superimposed onto the structure of the Ad11-CD46 complex (24). In the Ad11-CD46 complex, three loops of Ad11, designated HI, DG, and IJ, were found to interact with the SCR domains of CD46. The DG loop in the Ad11 fiber knob lacks the E and F β-sheets. Ad16, in contrast, contains these β-sheets, forming the corresponding FG loop. The superimposition of the Ad16 structure onto the structure of the Ad11-CD46 complex revealed that the HI loop adopts similar conformations in both structures, while the FG and IJ loops of Ad16 differ significantly from their counterparts in Ad11. The FG loop is longer and the IJ loop is shorter in the Ad16 structure than the corresponding DG and IJ loops in the Ad11 structure. The differences in conformation and the length of these loops were considered potential causes for the reduced CD46 binding to the Ad16 fiber knob.

FIG. 3.

FIG. 3.

Open in a new tab

Structural comparison of the Ad16 and Ad11 fiber knobs. (A) Ribbon diagram showing the crystal structure of the Ad16 fiber knob trimer viewed down the threefold axis. (B) Stereo diagram showing the quality of 2Fo−Fc electron density for residues 289 to 293 of the Ad16 fiber contoured at 1.0σ. (C) A ribbon diagram showing two subunits of the Ad16 fiber knob (red) superimposed on the complex of the Ad11 fiber knob (green) and SCR1-SCR2 of CD46 (blue). FG, HI, and IJ loops of Ad16 are indicated. (D) Structure-based alignment of the amino acids comprising the exposed regions of three surface loops that likely are involved in CD46 binding.

Localization of CD46 binding regions in the Ad16 fiber knob.

To ascertain the relative contribution of different regions in Ad16 and Ad11 fibers to CD46 interaction, we generated a panel of Ad16 fiber knob mutants targeting the HI, FG, and IJ loops (Fig. 4), which, based on the structure model, could affect CD46 binding. Each of the expressed knobs, with few exceptions, exhibited proper folding and assembly (trimerization), as determined by polyacrylamide gel electrophoresis. Our structural model of the Ad16-CD46 complex suggests that the HI loop of the fiber knob makes contact with SCR1 and SCR2 of CD46. The structural features of this receptor-interacting region are highly conserved between Ad11 and Ad16 (Fig. 5A). The HI loop of the Ad11 fiber knob contains two key arginine residues (R279 and R280) that mediate CD46 binding (24), and these residues also are conserved in the Ad16 fiber knob. Arginine 310 (R310) of the Ad16 fiber knob, corresponding to R280 in the Ad11 fiber knob, likely makes direct contact with CD46, possibly forming a salt bridge with glutamic acid 63 (E63). To test this model, we mutated R310 to asparagine, thereby eliminating the positive charge of the side chain. The R310N mutation had a pronounced effect on the ability of the Ad16 fiber knob to block virus attachment, essentially eliminating the ability of the fiber knob to compete in Ad5.16F virus attachment (Fig. 5B). The flanking amino acid (R309) in the Ad16 fiber knob that corresponds to R279 in the Ad11 fiber knob is a residue that also has been identified as critical for CD46 interaction (12). R279 does not interact directly with CD46 but is thought to orient the HI and DG loops via association with N247 and R280 (24). We therefore replaced R309 in Ad16 with a glutamine residue to eliminate the positive charge while retaining the approximate size of the side chain. The R309Q mutation resulted in a substantial loss of CD46 binding, as indicated in virus-binding competition assays, although the effect of this particular mutation was somewhat less pronounced than the effect of the R310N mutation. Our results demonstrate that the conserved arginine residues in the HI loop retain their critical function in CD46 binding, suggesting that Ad16 engages the receptor in a fashion similar to that of Ad11.

FIG. 4.

FIG. 4.

Open in a new tab

Sequences of recombinant fiber knobs containing mutations. Wild-type and mutant recombinant fiber knob proteins used in the virus binding competition experiments are shown in a structure-based alignment. The expression and trimerization of fiber knob proteins was verified by seminative gel electrophoresis. Mutated residues are highlighted. F16K, Ad16 fiber knob; F11K, Ad11 fiber knob.

FIG. 5.

FIG. 5.

Open in a new tab

Comparison of Ad16 fiber knobs binding to CD46. (A, C, and E) Structural comparison of the CD46 binding loops with associated residues in Ad16 and Ad11 fiber knobs (FK16 and FK11, respectively). Corresponding Ad11 residues are indicated in parentheses. CD46 is shown in dark gray, Ad16 in black, and Ad11 in light gray. (A) Parts of the HI loop in complex with E63 of CD46, R309 and R310 of Ad16, and the corresponding arginines of Ad11, R279 and R280, shown as stick representations. (C) IJ loop with R69 and L72 of CD46; T212 of Ad16; and V304, Q305, and M184 of Ad11 shown as stick representations. (E) FG loop with T64 of CD46 and Y275 and A276 of Ad16 shown as stick representations. (B, D, and F) Functional analysis of Ad16 fiber knob mutants. CHO cells expressing CD46 were preincubated with various concentrations of fiber knob mutants targeting the HI (B), IJ (D), or FG (F) loop of the Ad16 fiber knob prior to the addition of 30,000 Ad5.16F virus particles per cell. The mean fluorescence intensity was measured by flow cytometry. The background fluorescence of cells in the absence of virus was subtracted, and the fluorescence of cells incubated in the presence of virus without preincubation with a fiber knob was defined as 100%. Data points represent the means from triplicate experiments, with standard deviations shown by error bars. The plots of FG loop mutants were overlaid with a curve (dashed line) showing inhibition by the Ad11 fiber knob for comparison.

Another CD46 binding region of the fiber knob is comprised of the IJ loop. In Ad11, this loop contains two additional residues relative to the loop of Ad16, resulting in a longer IJ loop, a situation that could allow greater contact with CD46 based on the structure model (Fig. 5C). Thus, we investigated whether Ad16 fiber knob binding to CD46 could be enhanced by inserting two residues from the Ad11 fiber knob into the IJ loop of the Ad16 fiber knob. However, a mutant protein with the missing residues Val and Gln inserted after Glu333 of the Ad16 fiber knob (Ad16FK 333VQ334) could not be expressed, suggesting that the insertion of these residues in the context of the Ad16 fiber knob interfered with the proper folding and/or trimerization of the molecule. In support of this, we noted that the extended IJ loop of the Ad11 fiber knob likely is supported by the large side chain of the buried amino acid, Met184. This residue corresponds to Thr212, with its shorter side chain, in the Ad16 fiber knob. To investigate whether this longer side chain is needed to support the extended IJ loop, we created an additional T212M mutation in the Ad16 fiber knob. Both constructs, the T212M mutant as well as the combined 333VQ334/T212M mutant, could be expressed and retained a trimeric organization. Interestingly, however, the 333VQ334/T212M mutation did not improve, and actually decreased, the ability to compete in virus binding, while the T212M mutation alone inhibited virus attachment to CD46 at levels similar to those of the wild-type Ad16 fiber knob (Fig. 5D). Thus, we conclude that the two extra residues in the IJ loop are not necessary and/or are not beneficial for CD46 association in the context of the Ad16 fiber knob, potentially due to unfavorable interactions with side chains of other residues that are in close proximity.

One of the most obvious structural differences in the species B Ad fibers is in the FG loop (Fig. 5E). Two additional residues in Ad16 result in a longer FG loop that extends further than the corresponding DG loop of the Ad11 fiber. Our structure model (Fig. 3C) suggests that this extended loop may need to bend out of the way in order to allow the efficient interaction of the Ad16 fiber knob with CD46. To investigate the possibility that the longer FG loop of Ad16 fiber interferes with CD46 association, we created two Ad16 fiber knob mutants with either one or two residues removed from the FG loop. We deleted either Tyr275 or Ala276 or both residues, since we anticipated that the side chains of these amino acids likely would not be involved in direct critical interactions with CD46. The ΔY275 as well as the ΔA276 mutant each showed a small improvement in CD46 association compared to that of the wild-type Ad16 fiber knob. In contrast, the double deletion mutant ΔY275/A276 showed a substantial improvement in CD46 binding, as measured in virus competition assays (Fig. 5F). These findings suggest that the larger FG loop in the Ad16 fiber knob creates steric hindrance for CD46 association, although the side chains in the extended loop do not appear to contribute to this situation.

To investigate further whether the longer FG loop in Ad16 contributes to a lower CD46 binding efficiency, we extended the corresponding loop of the Ad11 fiber knob so that it resembled that of Ad16 by inserting one or two Ala residues in the DG loop (246A247 and 246AA247). The insertion of a single Ala resulted in a substantial loss in binding efficiency (Fig. 6A), and the addition of a second Ala reduced the binding of the Ad11 fiber knob close to the level of the Ad16 fiber knob. A third mutant, consisting of an insertion of a Tyr and an Ala residue (246YA247), exhibited almost equivalent binding to that of the 246AA247 mutant, further indicating that the length of the loop, rather than composition of the amino acid side chains, modulates CD46 binding. To investigate whether the decreased CD46 binding efficiency on cells also was reflected in a loss of intrinsic receptor association, we determined the kinetic constants of the FK11 246AA247 mutant (Fig. 6B). This analysis revealed that due to a substantial increase in the dissociation constant (Kd), the mutant fiber displayed a significantly reduced binding constant, a KD of ∼379 nM, further indicating that the overall length of the FG loop plays a role in regulating the efficiency of CD46 association.

FIG. 6.

FIG. 6.

Open in a new tab

Effect of mutations in the DG loop on Ad11 binding to CD46. (A) Virus binding competition experiments with insertion mutants targeting the DG loop of the Ad11 fiber knob were performed as described in the legend to Fig. 5. Data points represent the means from triplicate experiments, with standard deviations shown by error bars. (B) Comparison of the kinetic constants of the Ad11 and Ad16 fiber knobs (FK11 and FK16, respectively) to those of FK11 246AA247. The overall affinity (KD) and kinetics (ka and Kd) of FK11 246AA247 were analyzed by Biacore, as described for the Ad16 fiber knob. Kinetic constants were obtained by curve fitting using a 1:1 interaction model. The KD calculated from association and dissociation rates was verified by equilibrium binding analysis. Values in parentheses represent the errors of the fitting procedure in the final digit.

DISCUSSION

Recent studies comparing the CD46 binding efficiencies of two species B2 Ads, Ad11 and Ad35, showed that despite subtle structural differences in their CD46 binding sites, these fiber knobs exhibit comparable binding efficiencies (23, 24, 37). However, it had not previously been determined whether B1 and B2 species utilize CD46 with similar efficiency.

We used surface plasmon resonance (Biacore) to study the kinetics and affinities of CD46 binding to B1 and B2 Ad fiber knobs. The overall affinity of CD46 binding to the Ad11 fiber knob measured by Biacore compared well to that previously reported using isothermal titration calorimetry (24). The modest difference (∼2 nM versus 5.77 nM) between the two values likely results from the different methods used: isothermal titration calorimetry detects the interactions of two molecules in solution, while Biacore studies are performed on a solid support. The kinetic constants of Ad11 determined by Biacore in our studies also are similar to the published constants of the Ad35 fiber knob (37), with a less than threefold difference in overall affinity (KD). In striking contrast, the Ad16 fiber knob exhibited a more than 70-fold lower affinity for soluble CD46 than the Ad11 fiber knob. As both Ad11 and Ad16 are known to use CD46 as a receptor, we were surprised to observe such a pronounced difference in CD46 binding. Given the relatively weak intrinsic binding affinity of individual CD46 binding sites on the Ad16 fiber knob, we wondered whether this might significantly impact virus binding to CD46 on host cells. Fiber-based competition experiments demonstrated that despite the low intrinsic binding affinity for CD46, the Ad16 fiber knob still was capable of mediating virus attachment to CD46 on the cell surface, albeit at a significantly reduced level compared to that of the Ad11 fiber knob.

To further investigate the underlying structural basis of CD46 binding, we solved the crystal structure of the Ad16 fiber knob, which allowed us to generate a comparative model of its complex with CD46. While the structure of the Ad16 fiber knob generally is very similar to that of other members of subgroup B, distinguishing features were apparent in the surface loops involved in CD46 binding. In mutagenesis studies, we specifically targeted these loops to shed light on the structure-function relationships between the Ad11 and Ad16 fiber knobs. The HI loop is the most structurally conserved of all three loops involved in CD46 binding and has a crucial impact in receptor binding that is retained in Ad16. In contrast, the FG and IJ loops show notable structural differences compared to those of the Ad11 fiber knob. Two additional residues extend the FG loop of the Ad16 fiber knob, while the IJ loop lacks two residues present in Ad11. We were not able to conclusively determine the impact of the shortened IJ loop in the Ad16 fiber knob by mutating it to match that of the Ad11 fiber. We interpret this to mean that either the insertion resulted in additional (structural) changes by interfering with other residues in the fiber knob or that the elongated IJ loop might not be necessary/beneficial in the context of the Ad16 fiber knob. However, as the IJ loop of Ad16 fiber is identical to the IJ loop of Ad35 fiber, the binding efficiency of which closely resembles that of Ad11, this structural difference probably is not sufficient to explain the striking difference in overall receptor affinities.

The most notable structural difference between the fiber knobs was located in the elongated FG loop of Ad16. Interestingly, the protruding loop of Ad16 is not shared by any members of B2 Ad species. In our studies involving mutants targeting the FG loop of the Ad16 fiber knob as well as the corresponding loop of the Ad11 fiber knob, we demonstrated that the length of the FG loop is a critical structural feature in CD46 binding. In particular, as few as two amino acid residues inserted into the corresponding loop (DG) of the Ad11 fiber greatly reduced CD46 association, while the absence of these residues from the wild-type fibers of B1 Ad types favored receptor binding. It should be noted, however, that besides the structural differences in the FG loop, other individual residues in the fiber knobs also are likely to significantly influence the binding efficiency of Ad16, as the shortening of the FG loop of Ad16 to match that of Ad11 did not completely restore optimal CD46 association.

Our findings raise the question of whether other B1 Ads that have been reported to have low or negligible CD46 binding activity are influenced by their FG loop sequence. Based on sequence similarity, the fiber knobs most similar to the Ad16 fiber knob are those of Ad3 and Ad7h (Fig. 7A). Interestingly, there has not been complete agreement on the ability of Ad3 to utilize CD46. While Marttila et al. (17) suggested that CD46 is not a receptor for Ad3, other reports have indicated that Ad3 does bind to CD46 at a level lower than that of certain other serotypes (7, 31). In comparing the sequences of the predicted CD46 binding regions, we noted that the fiber knobs of Ad3 and Ad7h, like that of Ad16, have an FG loop that is two residues longer than that of species B2 and certain species B1 Ads (Fig. 7B). The crystal structure of Ad3 (6) shows that the FG loop is indeed extended further than that of B2 Ads and, because of a proline close to the apex of the loop, is oriented in a way that its binding to CD46 might be even more reduced than that of Ad16. It also is worth noting that while Ad3 retains the arginine in the HI loop that is critical for forming a salt bridge with CD46 and has a lysine in place of the adjacent arginine, retaining the positive charge of this residue, the side chains of these residues seem to be positioned away from the potential binding site due to the bent FG loop. Based on our findings with Ad16 and these potential structural similarities with Ad3, we suggest that certain species B1 fibers share reduced CD46 binding due to a protruding FG loop.

FIG. 7.

FIG. 7.

Open in a new tab

Sequence comparison of species B Ad fiber knobs. (A) Phylogenetic tree diagram of the amino acid sequence of species B Ad fiber knob domains. Sequences were aligned from the TLWT sequence that marks the start of the knob domain to the stop codon. (B) Sequence alignment of the partial FG loop of the Ad16 fiber knob (residues 275 to 281) with the corresponding sequences of other species B Ads. Ad16, Ad3, and Ad7h differ from the other Ad serotypes by having two extra residues in the FG loop.

One implication of our findings is that the evolutionary pressures that impact sequence variation in external Ad fiber loops (e.g., escaping antibody neutralization) nonetheless can preserve receptor usage due to avidity effects. The fiber knob has been shown to be an important target for the humoral immune response (16, 33). During the immune response, anti-fiber antibodies appear early, before anti-penton base or anti-hexon antibodies, and are thought to play an important role in the neutralization of Ad in combination with anti-penton antibodies. Antibodies against fibers recognize conformational epitopes on the knob domain of the trimeric fiber protein (11). Thus, conformational changes caused by the variations of exposed surface loops in the fiber knob might be an efficient way for the virus to escape antibody neutralization. Avidity effects provide a relatively large window for effective CD46 association on host cells, albeit with distinct efficiencies, thereby allowing a considerable level of sequence variation on exposed loops of the fibers. As previous studies indicated that preexisting neutralizing antibodies against epitopes on capsid proteins may negatively impact the expression of transgene products in gene therapy applications (27), our results could provide insights for the development of novel gene therapy vectors that preserve receptor binding while evading the host immune response.

Acknowledgments

We thank Samia N. Naccache for comments and editorial suggestions.

This work was supported by NIH grant R24 EY017540, by NIH grant R56 AI070771 to V.S.R., and by NIH grant RO1 EY011431 to G.R.N.

Footnotes

▿

Published ahead of print on 4 June 2008.

†

This work is done as manuscript #19477 of The Scripps Research Institute.

REFERENCES

  • 1.Anders, M., C. Christian, M. McMahon, F. McCormick, and W. M. Korn. 2003. Inhibition of the Raf/MEK/ERK pathway up-regulates expression of the coxsackievirus and adenovirus receptor in cancer cells. Cancer Res. 632088-2095. [PubMed] [Google Scholar]
  • 2.Bergelson, J. M., J. A. Cunningham, G. Droguett, E. A. Kurt-Jones, A. Krithivas, J. S. Hong, M. S. Horwitz, R. L. Crowell, and R. W. Finberg. 1997. Isolation of a common receptor for Coxsackie B viruses and adenoviruses 2 and 5. Science 2751320-1323. [DOI] [PubMed] [Google Scholar]
  • 3.Brünger, A. T., P. D. Adams, G. M. Clore, W. L. DeLano, P. Gros, R. W. Grosse-Kunstleve, J. S. Jiang, J. Kuszewski, M. Nilges, N. S. Pannu, R. J. Read, L. M. Rice, T. Simonson, and G. L. Warren. 1998. Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54905-921. [DOI] [PubMed] [Google Scholar]
  • 4.Chenna, R., H. Sugawara, T. Koike, R. Lopez, T. J. Gibson, D. G. Higgins, and J. D. Thompson. 2003. Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res. 313497-3500. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Collaborative Computational Project, Number 4. 1994. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50760-763. [DOI] [PubMed] [Google Scholar]
  • 6.Durmort, C., C. Stehlin, G. Schoehn, A. Mitraki, E. Drouet, S. Cusack, and W. P. Burmeister. 2001. Structure of the fiber head of Ad3, a non-CAR-binding serotype of adenovirus. Virology 285302-312. [DOI] [PubMed] [Google Scholar]
  • 7.Fleischli, C., S. Verhaagh, M. Havenga, D. Sirena, W. Schaffner, R. Cattaneo, U. F. Greber, and S. Hemmi. 2005. The distal short consensus repeats 1 and 2 of the membrane cofactor protein CD46 and their distance from the cell membrane determine productive entry of species B adenovirus serotype 35. J. Virol. 7910013-10022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Fuxe, J., L. Liu, S. Malin, L. Philipson, V. P. Collins, and R. F. Pettersson. 2003. Expression of the coxsackie and adenovirus receptor in human astrocytic tumors and xenografts. Int. J. Cancer 103723-729. [DOI] [PubMed] [Google Scholar]
  • 9.Gaggar, A., D. M. Shayakhmetov, and A. Lieber. 2003. CD46 is a cellular receptor for group B adenoviruses. Nat. Med. 91408-1412. [DOI] [PubMed] [Google Scholar]
  • 10.Gaggar, A., D. M. Shayakhmetov, M. K. Liszewski, J. P. Atkinson, and A. Lieber. 2005. Localization of regions in CD46 that interact with adenovirus. J. Virol. 797503-7513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Gahéry-Ségard, H., F. Farace, D. Godfrin, J. Gaston, R. Lengagne, T. Tursz, P. Boulanger, and J. G. Guillet. 1998. Immune response to recombinant capsid proteins of adenovirus in humans: antifiber and anti-penton base antibodies have a synergistic effect on neutralizing activity. J. Virol. 722388-2397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Gustafsson, D. J., A. Segerman, K. Lindman, Y. F. Mei, and G. Wadell. 2006. The Arg279Gln substitution in the adenovirus type 11p (Ad11p) fiber knob abolishes EDTA-resistant binding to A549 and CHO-CD46 cells, converting the phenotype to that of Ad7p. J. Virol. 801897-1905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Hemminki, A., A. Kanerva, B. Liu, M. Wang, R. D. Alvarez, G. P. Siegal, and D. T. Curiel. 2003. Modulation of coxsackie-adenovirus receptor expression for increased adenoviral transgene expression. Cancer Res. 63847-853. [PubMed] [Google Scholar]
  • 14.Jones, T. A., J. Y. Zou, S. W. Cowan, and M. Kjeldgaard. 1991. Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47110-119. [DOI] [PubMed] [Google Scholar]
  • 15.Kleywegt, G. J., and T. A. Jones. 1994. Halloween…masks and bones, p. 59-66. In S. Bailey, R. Hubbard, and D. Waller (ed.), From first map to final model. SERC Daresbury Laboratory, Warrington, United Kingdom.
  • 16.Krause, A., J. H. Joh, N. R. Hackett, P. W. Roelvink, J. T. Bruder, T. J. Wickham, I. Kovesdi, R. G. Crystal, and S. Worgall. 2006. Epitopes expressed in different adenovirus capsid proteins induce different levels of epitope-specific immunity. J. Virol. 805523-5530. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Marttila, M., D. Persson, D. Gustafsson, M. K. Liszewski, J. P. Atkinson, G. Wadell, and N. Arnberg. 2005. CD46 is a cellular receptor for all species B adenoviruses except types 3 and 7. J. Virol. 7914429-14436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Morelock, M. M., R. H. Ingraham, R. Betageri, and S. Jakes. 1995. Determination of receptor-ligand kinetic and equilibrium binding constants using surface plasmon resonance: application to the lck SH2 domain and phosphotyrosyl peptides. J. Med. Chem. 381309-1318. [DOI] [PubMed] [Google Scholar]
  • 19.Morton, T. A., and D. G. Myszka. 1998. Kinetic analysis of macromolecular interactions using surface plasmon resonance biosensors. Methods Enzymol. 295268-294. [DOI] [PubMed] [Google Scholar]
  • 20.Myszka, D. G. 1999. Improving biosensor analysis. J. Mol. Recognit. 12279-284. [DOI] [PubMed] [Google Scholar]
  • 21.Nepomuceno, R. R., L. Pache, and G. R. Nemerow. 2007. Enhancement of gene transfer to human myeloid cells by adenovirus-fiber complexes. Mol. Ther. 15571-578. [DOI] [PubMed] [Google Scholar]
  • 22.Otwinowski, Z., and W. Minor. 1997. Processing X-ray diffraction data collected in oscillation mode, p. 307-326. In C. W. J. Carter and R. M. Sweet (ed.), Methods in enzymology, vol. 276. Academic Press, San Diego, CA. [DOI] [PubMed] [Google Scholar]
  • 23.Pache, L., S. Venkataraman, G. R. Nemerow, and V. S. Reddy. 2008. Conservation of fiber structure and CD46 usage by subgroup B2 adenoviruses. Virology 375573-579. [DOI] [PubMed] [Google Scholar]
  • 24.Persson, B. D., D. M. Reiter, M. Marttila, Y. F. Mei, J. M. Casasnovas, N. Arnberg, and T. Stehle. 2007. Adenovirus type 11 binding alters the conformation of its receptor CD46. Nat. Struct. Mol. Biol. 14164-166. [DOI] [PubMed] [Google Scholar]
  • 25.Post, T. W., M. K. Liszewski, E. M. Adams, I. Tedja, E. A. Miller, and J. P. Atkinson. 1991. Membrane cofactor protein of the complement system: alternative splicing of serine/threonine/proline-rich exons and cytoplasmic tails produces multiple isoforms that correlate with protein phenotype. J. Exp. Med. 17493-102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Read, R. J. 2001. Pushing the boundaries of molecular replacement with maximum likelihood. Acta Crystallogr. D Biol. Crystallogr. 571373-1382. [DOI] [PubMed] [Google Scholar]
  • 27.Rosenecker, J., K. H. Harms, R. M. Bertele, A. Pohl-Koppe, E. v. Mutius, D. Adam, and T. Nicolai. 1996. Adenovirus infection in cystic fibrosis patients: implications for the use of adenoviral vectors for gene transfer. Infection 245-8. [DOI] [PubMed] [Google Scholar]
  • 28.Sakurai, F., K. Kawabata, and H. Mizuguchi. 2007. Adenovirus vectors composed of subgroup B adenoviruses. Curr. Gene Ther. 7229-238. [DOI] [PubMed] [Google Scholar]
  • 29.Segerman, A., J. P. Atkinson, M. Marttila, V. Dennerquist, G. Wadell, and N. Arnberg. 2003. Adenovirus type 11 uses CD46 as a cellular receptor. J. Virol. 779183-9191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Segerman, A., Y. F. Mei, and G. Wadell. 2000. Adenovirus types 11p and 35p show high binding efficiencies for committed hematopoietic cell lines and are infective to these cell lines. J. Virol. 741457-1467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Sirena, D., B. Lilienfeld, M. Eisenhut, S. Kalin, K. Boucke, R. R. Beerli, L. Vogt, C. Ruedl, M. F. Bachmann, U. F. Greber, and S. Hemmi. 2004. The human membrane cofactor CD46 is a receptor for species B adenovirus serotype 3. J. Virol. 784454-4462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Smith, J. G., and G. R. Nemerow. 2008. Mechanism of adenovirus neutralization by Human alpha-defensins. Cell Host Microbe 311-19. [DOI] [PubMed] [Google Scholar]
  • 33.Stallwood, Y., K. D. Fisher, P. H. Gallimore, and V. Mautner. 2000. Neutralisation of adenovirus infectivity by ascitic fluid from ovarian cancer patients. Gene Ther. 7637-643. [DOI] [PubMed] [Google Scholar]
  • 34.Storoni, L. C., A. J. McCoy, and R. J. Read. 2004. Likelihood-enhanced fast rotation functions. Acta Crystallogr. D Biol. Crystallogr. 60432-438. [DOI] [PubMed] [Google Scholar]
  • 35.Tuve, S., H. Wang, C. Ware, Y. Liu, A. Gaggar, K. Bernt, D. Shayakhmetov, Z. Li, R. Strauss, D. Stone, and A. Lieber. 2006. A new group B adenovirus receptor is expressed at high levels on human stem and tumor cells. J. Virol. 8012109-12120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Vogels, R., D. Zuijdgeest, R. van Rijnsoever, E. Hartkoorn, I. Damen, M. P. de Bethune, S. Kostense, G. Penders, N. Helmus, W. Koudstaal, M. Cecchini, A. Wetterwald, M. Sprangers, A. Lemckert, O. Ophorst, B. Koel, M. van Meerendonk, P. Quax, L. Panitti, J. Grimbergen, A. Bout, J. Goudsmit, and M. Havenga. 2003. Replication-deficient human adenovirus type 35 vectors for gene transfer and vaccination: efficient human cell infection and bypass of preexisting adenovirus immunity. J. Virol. 778263-8271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Wang, H., Y. C. Liaw, D. Stone, O. Kalyuzhniy, I. Amiraslanov, S. Tuve, C. L. Verlinde, D. Shayakhmetov, T. Stehle, S. Roffler, and A. Lieber. 2007. Identification of CD46 binding sites within the adenovirus serotype 35 fiber knob. J. Virol. 8112785-12792. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Wu, E., S. A. Trauger, L. Pache, T. M. Mullen, D. J. von Seggern, G. Siuzdak, and G. R. Nemerow. 2004. Membrane cofactor protein is a receptor for adenoviruses associated with epidemic keratoconjunctivitis. J. Virol. 783897-3905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Zhang, Y., and J. Skolnick. 2005. TM-align: a protein structure alignment algorithm based on the TM-score. Nucleic Acids Res. 332302-2309. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Zheng, L., U. Baumann, and J. L. Reymond. 2004. An efficient one-step site-directed and site-saturation mutagenesis protocol. Nucleic Acids Res. 32e115. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Virology are provided here courtesy of American Society for Microbiology (ASM)

RESOURCES