What Does It Take to Bind CAR?

8 downloads 0 Views 1MB Size Report
Aug 16, 2005 - Ad–CAR binding, taking into consideration the documented cellular tropism of other Ad serotypes. Key Words: adenovirus, serotype 30, fiber ...
REVIEW

doi:10.1016/j.ymthe.2005.05.017

What Does It Take to Bind CAR? Lane K. Law1,2 and Beverly L. Davidson2,3,4,* Program in Gene Therapy, 1Genetics Ph.D. Program, 2Department of Internal Medicine, 3Department of Neurology, and 4 Department of Physiology & Biophysics, University of Iowa, Iowa City, IA 52242, USA *To whom correspondence and reprint requests should be addressed at 200 EMRB, University of Iowa, Iowa City, IA 52242, USA. Fax: +1 (319) 353 5572. E-mail: [email protected].

Available online 16 August 2005

Recombinant adenoviruses (Ads) have been used as reagents for biological studies and therapeutic protocols for the treatment of human patients. The two most commonly used Ads, Ad2 and Ad5, infect a broad range of tissues through interaction with the coxsackie and adenovirus receptor CAR. Both mutational analyses and crystal structure data have established residues in the fiber knob and shaft critical for Ad–CAR binding. In this report we review the contributions of various residues to Ad–CAR binding, taking into consideration the documented cellular tropism of other Ad serotypes. Key Words: adenovirus, serotype 30, fiber protein, crystal structure, CAR

Contents Adenovirus . . . . . . . . . . . . . . . . . . . . . . . . Ad12 Knob–Receptor Interactions . . . . . . . . . . Ad5 Knob–Receptor Interactions . . . . . . . . . . . Ad9f– and Ad41Lf–CAR Interactions . . . . . . . . . Adenoviruses with Limited or No Affinity for CAR The Importance of the Shaft . . . . . . . . . . . . . The Hexon . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

ADENOVIRUS Adenoviruses (Ads) are nonenveloped, double-stranded DNA viruses with an icosahedral coat composed mainly of hexon, penton, and fiber. At this time, there are 51 serotypes, which fall into six subgroups (groups A–F). Adenoviruses exhibit homotrimeric fibers (320 to 587 residues) at each of the vertices of the icosahedral capsid, each of which terminates in a globular knob domain (~175 residues). Many experiments demonstrate that the 46-kDa coxsackie virus and adenovirus receptor (CAR), a member of the immunoglobulin superfamily, is the receptor for most Ads from subgroups A, D, E, and F, but not from subgroup B or the short fiber of subgroup F [1,2]. Both human and mouse CARs have been identified. CAR is expressed on the surface of many cell types and is present within the tight junctions between polarized epithelial cells [3,4]. Secondary interactions mediated by adenovirus penton base and av integrins on the cell

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy 1525-0016/$30.00

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

599 601 606 606 606 607 608 608 608

surface can be used for subsequent endocytosis of viral particles [5–9]. The fiber proteins are evenly spaced about the viral coat to allow for maximal opportunity to interact with cellular receptors. The knob domain initiates a highaffinity interaction with host cell surface proteins. Adenoviral serotypes exhibit a conservation of structure within the fiber knob domain, the portion of the Ad capsid responsible for the attachment of the adenoviral particle to receptors on the surface of a permissive cell. However, within this conserved structure, a high degree of amino acid sequence variation is observed. The crystal structures of six adenoviral fiber knobs (Ad2, Ad3, Ad5, Ad12, Ad19, and Ad37) have been solved and have been fundamental to an increased understanding of the specifics of the interaction of the fiber knob and CAR, the receptor for the majority of adenoviral serotypes studied [10–13]. Of particular importance is the contribution of Bewley et al., with

599

REVIEW

their publication of the crystal structure of the Ad12 knob–CAR D1 complex [11]. Analysis of this structure reveals the contribution of key residues that, if not critical, certainly influence knob–CAR interactions.

doi:10.1016/j.ymthe.2005.05.017

These data can help us modify the virus to promote or inhibit binding. In addition to the knob, the fiber shaft, which provides length and flexibility to the fiber, is an important factor in cellular attachment.

FIG. 1. Schematic representations of the (A) Ad12, (B) Ad37, and (C) Ad30 knob monomers. These diagrams are adapted from Xia et al. [10] by permission of the publisher. Boxes represent individual amino acid residues. Numbers correspond to the position of each amino acid in the Ad12 knob. Residues found in loops are purple. Residues found in h strands are in blue. Loops and h strands are labeled, as well as the amino and carboxy termini. Critical CAR-binding residues (for Ad12) are green.

600

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

REVIEW

doi:10.1016/j.ymthe.2005.05.017

FIG. 1 (continued).

A thorough understanding of proposed knob–CAR interactions, or the lack of interactions, should facilitate vector development. This review summarizes the amino acid residues required for knob–CAR interactions in several serotypes. In addition we use modeling programs to understand better why some serotypes do not bind CAR. The influence of the shaft domain is also addressed.

AD12 KNOB–RECEPTOR INTERACTIONS The crystal structure of the Ad12 knob alone and complexed with the NH2-terminal domain of human CAR (CAR D1) was determined to gain further insight into the mechanism of Ad infection [11]. Similar to the previously crystallized Ad5 knob monomers [10], Ad12 knob monomers have an eight-stranded anti-parallel hsandwich fold. A schematic diagram of the Ad12 monomer is shown in Fig. 1A. Indicated in these diagrams is the convergence of strands and loops forming the V (closer to the virion) and R (defined originally as receptor binding) sheets. Also indicated are the positions of the amino acids within the loops and those residues critical for CAR interaction. The homotrimeric structure can be described as a three-bladed propeller with a surface depression around the threefold molecular symmetry axis. The R sheets are the faces of the blades. Extensive interactions between

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

monomers, including hydrogen bonds, salt bridges, van der Waals contacts, and water-mediated charge interactions, maintain the trimeric structure. The h strands within the V sheet provide contacts between monomers. Strands J, C, B, and A (which contribute to the V sheet) of adjacent monomers form the tightly packed trimer interface. This interface creates a large cavity located along the central threefold axis of symmetry. Xia et al. proposed that this cavity was the receptor binding site [10]. Strands G, H, D, and I (the R sheet) form a solvent-exposed h sheet, also identified as a putative receptor-binding site [10]. The only major difference observed between the crystal structure of the Ad5 and Ad12 knobs is in the HI loop. In Ad12 (residues 549 to 556) this loop is well ordered. In Ad5 (residues 537 to 549) the HI loop is extended by 5 residues and is disordered [11]. Interestingly, a deletion of 2 amino acids in the Ad12 knob HI loop, a site of insertion for targeting of Ad5 [14,15], can reduce Ad knob–CAR affinity by improving FG loop flexibility and decreasing water-mediated hydrogen bonding [11]. Bewley et al. crystallized the Ad12 knob–human CAR D1 complex to discern the Ad12 knob CAR-binding sites [11] (Fig. 2A). The Ad12 knob–CAR D1 complex is described as triskelion (a figure consisting of three branches radiating from a common center) with three CAR D1 monomers bound per knob trimer. The ability of the knob to bind three CAR molecules may account in part

601

REVIEW

for the high efficiency of Ad infection. The crystallization studies show that CAR D1 does not bind the previously predicted sites on knob (i.e., the cavity at the interface located along the central threefold axis of symmetry or the R sheet) [10], rather each CAR D1 molecule binds at the interface between two adjacent Ad12 knob monomers. This observation is consistent with previous findings in that most neutralizing antibodies to knob are directed against the trimer, rather than the monomer [11,16,17]. Knob–CAR complex formation does not induce notable changes in the structure of the knob protein. The orientation/structure of CAR D1, but not CAR D2, was ascertained in the work by Freimuth, Flanagan, and colleagues [11]. Comparisons of CAR D1 with structures of homologous proteins, solved with both D1 and D2 domains, would predict that CAR D2 does not make extensive contacts with knob [11]. These findings are expected as soluble CAR D1 alone is sufficient for Ad2and Ad12-knob binding [18]. The conformation of the complex is suggestive of the local specificity of Ad12 knob for CAR. The interface between knob and CAR D1 is particularly informative as it is formed by four regions of the Ad12 knob in close proximity to a single face of the CAR D1 sandwich [11]. The regions forming the interface are the AB loop, the carboxyl end of the DE loop, the F strand, and the FG loop of an adjacent knob monomer. The convergence of the loops at the interface is depicted in Fig. 2A. The majority of amino acid residues contributing to the knob–CAR interaction are found in the AB loop, which spans the width of CAR D1 (Fig. 2B). The interaction between the AB loop of Ad12 and CAR includes three hydrogen bonds formed by residues Ad12f-Asp415 with CAR-Lys125 and CAR-Lys123, Ad12f-Leu426 with CAR-Tyr83 and CAR-Tyr85, and Ad12f-Lys429 with CAR-Glu58. These hydrogen bonds may be a key anchor for the complex [11]. The residue Ad12f-Pro418 is conserved among many Ad serotypes and found in the middle of the AB loop. [See Table 1 for a listing of Ad12 amino acid residues involved in CAR binding, the type of bond (when known), the corre-

doi:10.1016/j.ymthe.2005.05.017

sponding residues in other Ad serotypes (2, 3, 5, 9, 19, 30, 37, and 41), and the binding residues in CAR D1.] Alignment of the Ad12 knob sequence with that of other non-CAR-binding serotypes (Ad3, Ad7, and Ad35)

FIG. 2. The Ad12 knob–CAR-binding interface. (A) A view of the fiber trimer complexed with CAR. Knob monomers were yellow, blue, and gray in DeepView. The CAR D1 molecule was green. The AB loop of monomer A is red. The FG loop of the adjacent monomer B is pink. A white arrow indicates the interface between the Aad12knob and the CAR D1 molecule. (B) Key contact residues between the Ad12 AB loop and CAR D1. The AB loop of monomer A is red. The CAR D1 molecule is green. The side chains of Asp415, Pro417, Pro418, Leu426, and Lys429 of the Ad12 monomer A AB loop are visible, as are the side chains of Glu58, Val72, Leu75, Tyr82, Tyr85, Lys123, and Lys125 of the CAR D1 molecule. Carbon atoms are white. Oxygen atoms are red. Nitrogen atoms are blue. (C) Residues from the second monomer’s FG loop (pink): Pro517, Pro519, Asn520, and Glu523. In particular Asn520 forms a hydrogen bond with Asp70 of CAR D1 (adapted from Bewley et al. [11] using DeepView, reproduced by permission of the publisher).

602

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

603

REVIEW

Adenoviral fibers exhibiting CAR binding are listed in white with teal background. Ad12 residues in the primary (yellow monomer, Fig. 2A) binding monomer are shown above in orange. Ad12 residues in the secondary (blue) binding monomer are in blue. Residues with green background are critical for Ad5 knob-CAR interaction. Residues with gray background are conserved in the majority of the serotypes shown. Residues in brown are potential inhibitors of CAR binding. w A Leu75Ala substitution in the CAR D1 molecule reduces adenoviral knob binding. *The salt bridge formed between Lys429 (Ad12) and Glu58 (CAR D1) is important for cellular attachment.

doi:10.1016/j.ymthe.2005.05.017

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

TABLE 1: Listing of the critical amino acids for CAR binding in Ad12 and Ad5, their partner residues in CAR D1, and the corresponding residues in Ad5, Ad2, Ad9, Ad41L, Ad30, and Ad3

REVIEW

demonstrates the conservation of the AB loop (Fig. 3) [19,20]. Ad3, Ad7, and Ad35 have either insertions or substitutions in this loop, for example, in positions corresponding to the conserved residues Ad12f-Pro418 and Ad12f-Asn419. Ad12 sequences have been mutated to those found in non-CAR-binding knobs (serotypes 3, 7, and 35) to confirm their contribution to CAR binding (depicted in Fig. 3). The substitutions Ad12f-Pro417Glu and Ad12fPro418Ala converted the Ad12 AB loop to an Ad3-like AB loop. In addition, a Thr-Ile insertion before Ad12f-Ser421 (lengthens the AB loop), deletion of Ad12f-Glu425Leu426 (shortens the AB loop), and an Ad12f-Glu425Ser

doi:10.1016/j.ymthe.2005.05.017

substitution was studied [11]. Each change abolished CAR D1 binding. These data support the structural findings that sequences in the AB loop are required for Ad12 knob–CAR complex formation. A fluorescence polarization method of measuring knob binding to CAR D1 has also been used and indicates that Ad2 knob has a fivefold greater affinity for CAR D1 than Ad12 knob [21]. Comparisons of the Ad2 knob with the Ad12–CAR D1 complex suggest that Ad2f-Ser408 may participate in hydrogen bond stabilization with CAR [21]. The Ad12 knob has Pro, rather than Ser, at this position (Table 1). Mutating Ad12fPro417 to Ser caused a threefold increase in Ad12 knob

FIG. 3. Primary amino acid alignment of the knob region of adenoviral serotypes 30, 37, 19, 15, 9, 2, 5, 12, 35, 3, 7, and 41L. Adenoviral serotypes 9, 2, 5, 12, and 41L can bind CAR (boxed in teal). Amino acids conserved among serotypes are shaded in gray. Residues critical for CAR binding have a green background. Residues boxed in brown are potentially inhibitory for CAR binding. Ad37 Lys240 (orange background) is required for binding to conjunctival cells. Ad30 sequences that may contribute to impaired CAR binding are indicated (orange boxes). The loops and strands are indicated.

604

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

REVIEW

doi:10.1016/j.ymthe.2005.05.017

FIG. 3 (continued).

affinity for CAR D1. The formation of an additional hydrogen bond in Ad12f-Ser417 relative to the parental Pro-containing Ad12 knob was confirmed by crystal structure and may be partially responsible for the increased affinity [11,21]. Glu58 within CAR D1, which participates in Ad12f-Ser417 hydrogen bond interaction, also interacts with Ad12-Lys429, a highly conserved residue (Table 1, Figs. 2B and 3). In addition to Ad2f-Ser408, Tyr477 within the DG loop may participate in hydrogen bonding; the corresponding residue in the Ad12–CAR D1 complex is serine [21]. Interestingly, a Tyr489Ser substitution in Ad12 knob increased CAR binding affinity approximately eightfold [21]. Ad12 residues Val450 and Lys451 of the CD loop, Gln487 (DE loop), Gln494 (E h strand), and Ser497 and Val498 (F h strand) are also required for CAR binding. The

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

crystal structure of Ad12 knob–CAR demonstrates that the adjacent monomer also contributes to the complex. Residues from the second monomerTs FG loop, Ad12fPro517, Ad12f-Pro519, Ad12f-Asn520, and Ad12f-Glu523 contact CAR D1 directly, and in particular Ad12f-Asn520 forms a hydrogen bond with Asp70 of CAR D1 (Fig. 2C) [11]. While Ad12f-Gln494, Ad12f-Pro519, and Ad12fAsn520 are all contact residues in Ad12, substitution of the spatially similar residues in Ad5 fiber knob (Ad5fAsn482, Ad5f-Ser507, and Ad5f-His508) had no effect on CAR binding [22], suggesting that the adjacent monomer may not contribute to CAR interactions in Ad5 and, by similarity, Ad2. The higher affinity for CAR exhibited by Ad5 and Ad2 may preclude the necessity of these additional contacts. Thus, the binding interface with CAR varies among the different CAR-binding serotypes.

605

REVIEW AD5 KNOB–RECEPTOR INTERACTIONS The amino acid sequences of the serotypesT fiber knobs vary widely (Fig. 3). The variance is from 29 to 66% relative to the Ad5 fiber knob. Interestingly, most adenoviruses recognize CAR [23]. Based on the hypothesis that amino acids or amino acid motifs involved in CAR binding are necessarily conserved, Roelvink et al. aligned sequences of 14 adenoviral knobs from subgroups exhibiting CAR-binding serotypes, scored amino acid identity and similarity, and tested how substitutions of amino acids in the Ad5 knob AB loop impacted their binding properties [23]. The changes Ad5f-Ser408Glu or Gly, Ad5fPro409Ala, and Ad5f-Lys417Gly or Leu (AB loop) eliminated or inhibited binding to A549 cells as assessed by competition with Ad.CMV-hgal (recombinant adenovirus serotype 5 containing an Escherichia coli h-galactosidase expression cassette) [23]. Similarly, Kirby et al. found that Ad5f-Ser408Glu and Ad5f-Pro409Lys substitutions ablated binding to soluble CAR D1 and to receptorbearing cells in competition-based assays [22]. The molecular basis for these results was confirmed by the Ad12f– CAR D1 crystal structure; Ad5f-Ser408 likely interacts with CAR-Glu58, and an Ad5f-Pro409Lys substitution would impact AB loop contacts with CAR D1. Moreover, an Ad5fLys417Gly would impair interactions with CAR-Tyr 83 and CAR-Tyr85 (Table 1). Molecular evidence for Ad5fSer408–CAR-Glu58 interactions has also been shown [21]. These findings are supportive of the conservation of critical residues for CAR binding and the importance of the AB loop [23]. AD9F– AND AD41LF–CAR INTERACTIONS On the flip side, CAR mutations have also been made to test if residues CAR-Leu73, -Lys121, and -Lys123 are important (see Fig. 2B), among others. Experiments show that Leu73 and Lys121/123 are important for CAR–Ad12, –Ad5, –Ad9, and –Ad41L knob interactions [24]. Acting on the prediction that the amino acids corresponding to critical amino acids in the Ad5 knob might also be critical in other serotypes, Roelvink et al. generated point mutants in the Ad9 and the Ad41L fiber knobs, both of which bind CAR [23], the former with lower affinity. Residues critical for Ad5 binding are conserved in Ad9 (Ad9f-Ser189, Ad9f-Pro190, Ad9fLys198, Ad9f-Lys201, and Ad9f-Tyr262) (Fig. 3, Table 1). Ad41L has 3 of 5 conserved (Ad41Lf-Ser395, Ad41LfPro396, and Ad41Lf-Lys407). To assess their role, 5 residues in the Ad9 fiber knob and 4 in the F41L fiber knob were switched. All residues critical for Ad5–CAR binding were also critical for A9–CAR and Ad41L–CAR binding [19]. Kinetics of binding for Ad9f-Asp223Lys revealed an increased CAR affinity for Ad9, while the converse substitution Ad5f-Lys442Asp decreased binding affinity, suggesting that it may be at least partially responsible for the reduced affinity of Ad9 [24]. These

606

doi:10.1016/j.ymthe.2005.05.017

data confirm that the CAR-binding site consists of residues from the AB loop and the DE loop (Ad5 residues Ser408, Pro409, Lys417, Lys420, and Tyr477) [23], and they converge on the side of the knob (see Fig. 1).

ADENOVIRUSES CAR

WITH

LIMITED

OR

NO AFFINITY

FOR

The crystal structures of Ad3, Ad19, and Ad37 knobs have been solved [12,25]. The overall structure of Ad3 knob shows two anti-parallel h sheets formed by h strands A to J [12], similar to Ad5 and Ad12. Minor differences in comparison with other knobs include the absence of h strand F, a fusion of the canonical EF and FG loops (called EG loop for purposes of discussion below), and the insertion of two short helical regions inserted into the CD and EG loop regions [12]. Based on the revealed structure, Durmort et al. provided some suggestions as to why Ad3 knob does not bind CAR. The presence of Ad3f-Glu140 in the position corresponding to serine, threonine, or proline residues in CAR-binding serotypes (Fig. 3) is significant as mutations at the corresponding residue in Ad5 fiber, Ser408, reduce CAR binding [26]. Also, Ad3f-Lys138 is likely to have a negative impact on CAR binding as the corresponding oppositely charged residue in Ad12 (Asp415) forms salt bridges with Lys123 and Lys125 of CAR (see Fig. 3). The Ad3 conformation of the DE and EG loops may also inhibit CAR binding; a markedly different conformation is evident when aligned to the Ad12 knob– CAR complex within DeepView. In particular the Cterminal portion of the EG loop (flexible in the Ad3 structure) would result in a clear steric clash with CAR bound in the same position as is found in the Ad12–CAR complex. The crystal structures of Ads 19 and 37 differ only at Lys240Glu and Asn340Asp [25]. They are also very similar to the structures of Ad5 and Ad12 with the exception of the FG loop. In the carboxyl portion of the FG loop, a consensus insertion of SKKY relative to Ads 2, 5, and 12 is present in Ad19 and Ad37. By overlapping the structures of Ad19 and Ad37 with the crystal structure of the Ad12 knob–CAR D1 complex, the SKKY insertion bulges into the knob–CAR interface. As shown pictorially for Ad37 knob, the bulge extends toward the CAR binding interface of the adjacent monomer (Fig. 4A). Increased flexibility within this loop might impair CAR binding [25]. As discussed below, the impediment to CAR–Ad37 knob binding may be relieved by changing the Ad37 fiber shaft length. However, no Ad knob crystal structures with fiber shafts present have been reported to test this hypothesis directly. Ad30 does not bind CAR although direct comparison of amino acids important for CAR binding would suggest otherwise [27] (Figs. 1C and 3). However, sequence

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

REVIEW

doi:10.1016/j.ymthe.2005.05.017

based on available structures for Ad19 and Ad37. The Swiss Model Web site (http://www.expasy.org/swissmod/ SWISS_MODEL.html) and the DeepView program were used to generate the theoretical Ad30 knob model. Finally, we used the Ad12–CAR model to approximate Ad30 knob–CAR interactions (Fig. 4B). In Ad30 the FG loop is expanded and includes Gln304, Ser305, Lys306, Ile307, and Tyr308. The Q304-I308 span (Ad30 numbering) creates a bulge greater than that found in Ad19 and Ad37 and forces the side chain of Asp303 into very close proximity to CAR D1 (Fig. 4B). Similar to the Lys138 and Glu140 present in the Ad3 knob [12], Asp303 in the Ad30 knob likely contributes charge and/or steric hindrance. It is interesting to note that in the crystal structure of the Ad37 knob the side chains of Asn298 and Ser299 (the residues corresponding closest to Ad30 Asp303) do not project toward CAR D1, but rather off to the side (Fig. 4A).

THE IMPORTANCE

FIG. 4. Ad37 and Ad30 knob interactions with CAR. (A) Illustration of the bulge toward CAR present in the Ad37 FG loop. Amino acids Asp70 (CAR molecule) and Asn298 and Ser299 (Ad37 knob) are shown. A schematic diagram of the Ad37 monomer is depicted in Fig. 1B. A comparison of (A) with Fig. 2C illustrates the extent of the bulge. (B) Illustration of the bulge toward CAR present in the Ad30 FG loop. Amino acids Asp70 (CAR molecule) and Asp303 (Ad30 knob) are shown.

alignments do show that conserved amino acids adjacent to important contact residues differ in Ad30. For example, in Ad30f, a Leu271Tyr272 pair is present in place of a conserved Trp-Tyr pair, with the tryptophan conserved in all known CAR-binding Ads (Ad2, Ad5, Ad9, Ad12, Ad17, Ad41L) as well as Ad19 and Ad37 (Fig. 3). Other potential inhibitory changes are evident in the Ad30 FG loop. In this region, Ad12-Pro517 and -Pro519 are important for CAR binding. These residues are conserved in the majority of serotypes, with Pro517 invariant among all serotypes—even non-CAR-binding ones, except for Ad30 (Ala298 and Ala300; Fig. 3 and Table 1). To gain further insight into physical impediments that are not evident from primary sequence alignment (Figs. 1C and 3), we modeled the tertiary structure of Ad30 knob

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

OF THE

SHAFT

It has become increasingly apparent that the shaft of the adenovirus fiber influences binding affinity. The shaft domain gives the fiber protein its length and contains pseudo-repeats of up to 15+ amino acids, which form h sheets. The structural motif proposed for the shaft region is that of a triple h spiral [28]. The repetitive nature of the shaft domain fosters the formation of the triple h spiral, once the fibers have assembled into homotrimers. Different adenoviral subgroups exhibit different shaft lengths, and the number of repeats ranges from 6 (in Ad3, Ad11, and Ad35) to 23 (in Ad12) [29]. Wu and colleagues created fiber shaft–fiber knob chimeras and showed that short Ad37 shafts impaired Ad5 knob binding to CAR. Also, Ad37, which shows low affinity for CAR+ cells, could bind CAR when the short shaft was replaced with that of Ad5. It is possible that the rigid short Ad37 fiber would preclude binding to CAR due to steric hindrance [30] and by affecting knob–CAR interactions as described above. Similar studies also showed that shorter shafts negatively impact CAR binding and secondary interactions with av integrins [31]. In addition to length, flexibility of the fiber may influence knob–CAR interactions. Negative stain images of Ad2 fiber demonstrate a kink in the N-terminal portion of the protein [29], indicative of movement at that position. The kink corresponds to a nonconsensus motif, KLGXGLXFD/N, apparent in the final repeat of the shaft [30]. Interestingly, the D serotype fibers 8, 9, 15, and 30 have this sequence with the exception of the initial KL, while the other D serotypes have the final 5 residues (Law and Davidson, unpublished observations). Chroboczek et al. suggested that bending at the kink can promote penton base–integrin interaction [29]. In contrast to Ad2, cryo-EM image reconstructions of the Ad37 fiber,

607

REVIEW

including the distal knob domain, support the conclusion that D serotype viruses have rigid fibers [30,32]. These studies, and those evaluating Ad5 shaft mutants in vitro and in vivo [33], indicate that shaft–knob interactions are complex. Thus, data regarding receptor– fiber interactions obtained using only knobs may not reflect the interaction of an intact virion. We would predict that even in the context of the Ad5 shaft, Ad30 knob would be incapable of directing transduction via CAR. While Ad9 and Ad37 exhibit higher infectivity in the context of the Ad5 shaft, Ad30 knob has a number of amino acid differences not present in the Ad37 and Ad9 knobs that could preclude binding to CAR. These amino acids include Leu271 (which for the majority of serotypes is Trp), Ala298 (Pro in all other serotypes), Ala300 (Pro in D serotypes and Ad12), Gln304, and the Asp303 hindrance discussed above.

THE HEXON The most prevalent capsid protein is the hexon capsomere. This protein, like the fiber protein, is a homotrimer. Hexon exhibits three surface loops (L1, L2, and L4) facing outward. The hypervariable region 1 (HVR1) in L1 embodies the greatest variability among serotypes [34,35]. An alignment of the hexon sequence from different serotypes indicated that Ad5 has a markedly more acidic HVR1 loop compared to Ad9 hexon [35]. It is also of interest that for other CAR-recognizing serotypes (Ad12) as well as those with short fibers (Ad8, Ad9, and Ad37), the negatively charged region within HVR1 seen in Ad2 and Ad5 is absent (Ad9, Ad12, and Ad37) or positively charged (Ad8) [35]. Like the interdependence of knob–shaft interactions on receptor binding, it is clear that hexon can impart major influences on fiber–CAR interactions. In summary, we reviewed the structural requirements for fiber–CAR interactions among adenoviral serotypes exhibiting varying transduction characteristics. Specific amino acid residues confer high-affinity CAR receptor binding. Alteration of these residues or changes to shaft length, either as occur in nature or through manipulations at the bench, can inhibit or enhance adenovirus binding. Current and future work to specify the tropism of adenoviruses to a cell type of interest will broaden the application of adenoviral vectors for biological research and therapy. ACKNOWLEDGMENTS We thank Christine McLennan for assistance in manuscript preparation. This work was supported in part by the NIH (HL07638, HD33531) and the Roy J. Carver Trust (B.L.D.). RECEIVED FOR PUBLICATION NOVEMBER 16, 2004; REVISED APRIL 26, 2005; ACCEPTED MAY 9, 2005.

REFERENCES 1. Bergelson, J. M., et al. (1997). Isolation of a common receptor for coxsackie B viruses and adenoviruses 2 and 5. Science 275: 1320 – 1323.

608

doi:10.1016/j.ymthe.2005.05.017

2. Tomko, R. P., Xu, R., and Philipson, L. (1997). HCAR and MCAR: the human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses. Proc. Natl. Acad. Sci. USA 94: 3352 – 3356. 3. Cohen, C. J., Shieh, J. T., Pickles, R. J., Okegawa, T., Hsieh, J. T., and Bergelson, J. M. (2001). The coxsackievirus and adenovirus receptor is a transmembrane component of the tight junction. Proc. Natl. Acad. Sci. USA 98: 15191 – 15196. 4. Walters, R. W., Freimuth, P., Moninger, T. O., Ganske, I., Zabner, J., and Welsh, M. J. (2002). Adenovirus fiber disrupts CAR-mediated intercellular adhesion allowing virus escape. Cell 110: 789 – 799. 5. Bai, M., Campisi, L., and Freimuth, P. (1994). Vitronectin receptor antibodies inhibit infection of HeLa and A549 cells by adenovirus type 12 but not by adenovirus type 2. J. Virol. 68: 5925 – 5932. 6. Belin, M. T., and Boulanger, P. (1993). Involvement of cellular adhesion sequences in the attachment of adenovirus to the HeLa cell surface. J. Gen. Virol. 74: 1485 – 1497. 7. Davison, E., Diaz, R. M., Hart, I. R., Santis, G., and Marshall, J. F. (1997). Integrin alpha5beta1-mediated adenovirus infection is enhanced by the integrin-activating antibody TS2/16. J. Virol. 71: 6204 – 6207. 8. Mathias, P., Galleno, M., and Nemerow, G. R. (1998). Interactions of soluble recombinant integrin alphav beta5 with human adenoviruses. J. Virol. 72: 8669 – 8675. 9. Wickham, T. J., Mathias, P., Cheresh, D. A., and Nemerow, G. R. (1993). Integrins alpha v beta 3 and alpha v beta 5 promote adenovirus internalization but not virus attachment. Cell 73: 309 – 319. 10. Xia, D., Henry, L. J., Gerard, R. D., and Deisenhofer, J. (1994). Crystal structure of the receptor-binding domain of adenovirus type 5 fiber protein at 1.7 2 resolution. Structure 2: 1259 – 1270. 11. Bewley, M. C., Springer, K., Zhang, Y.-B., Freimuth, P., and Flanagan, J. M. (1999). Structural analysis of the mechanism of adenovirus binding to its human cellular receptor, CAR. Science 286: 1579 – 1583. 12. Durmort, C., et al. (2001). Structure of the fiber head of Ad3, a non-CAR-binding serotype of adenovirus. Virology 285: 302 – 312. 13. van Raaij, M. J., Louis, N., Chroboczek, J., and Cusack, S. (1999). Structure of the human adenovirus serotype 2 fiber head domain at 1.5-2 resolution. Virology 262: 333 – 343. 14. Xia, H., Anderson, B., Mao, Q., and Davidson, B. L. (2000). Recombinant human adenovirus: targeting to the human transferrin receptor improves gene transfer to brain microcapillary endothelium. J. Virol. 74: 11359 – 11366. 15. Krasnykh, V., Dmitriev, I., Mikheeva, G., Miller, C. R., Belousova, N., and Curiel, D. T. (1998). Characterization of an adenovirus vector containing a heterologous peptide epitope in the HI loop of the fiber knob. J. Virol. 72: 1844 – 1852. 16. Fender, P., Kidd, A. H., Brebant, R., Oberg, M., Drouet, E., and Chroboczek, J. (1995). Antigenic sites on the receptor-binding domain of human adenovirus type 2 fiber. Virology 214: 110 – 117. 17. Henry, L. J., Xia, D., Wilke, M. E., Deisenhofer, J., and Gerard, R. D. (1994). Characterization of the knob domain of the adenovirus type 5 fiber protein expressed in Escherichia coli. J. Virol. 68: 5239 – 5246. 18. Freimuth, P., Springer, K., Berard, C., Hainfeld, J., Bewley, M., and Flanagan, J. (1999). Coxsackievirus and adenovirus receptor amino-terminal immunoglobulin V-related domain binds adenovirus type 2 and fiber knob from adenovirus type 12. J. Virol. 73: 1392 – 1398. 19. Roelvink, P. W., et al. (1998). The coxsackivirus-adenovirus receptor protein can function as a cellular attachment protein for adenovirus serotypes from subgroups A, C, D, E, and F. J. Virol. 72: 7909 – 7915. 20. Yeh, H. Y., Pieniazek, N., Pieniazek, D., Gelderblom, H., and Luftig, R. B. (1994). Human adenovirus type 41 contains two fibers. Virus Res. 33: 179 – 198. 21. Howitt, J., Bewley, M. C., Graziano, V., Flanagan, J. M., and Freimuth, P. (2003). Structural basis for variation in adenovirus affinity for the cellular coxsackievirus and adenovirus receptor. J. Biol. Chem. 278: 26208 – 26215. 22. Kirby, I., et al. (2000). Identification of contact residues and definition of the CAR_binding site of adenovirus type 5 fiber protein. J. Virol. 74: 2804 – 2813. 23. Roelvink, P. W., Lee, G. M., Einfeld, D. A., Kovesdi, I., and Wickham, T. J. (1999). Identification of a conserved receptor-binding site on the fiber proteins of CAR_recognizing adenoviridae. Science 286: 1568 – 1571. 24. Kirby, I., et al. (2001). Adenovirus type 9 fiber knob binds to the coxsackie B virusadenovirus receptor (CAR) with lower affinity than fiber knobs of other CAR-binding adenovirus serotypes. J. Virol. 75: 7210 – 7214. 25. Burmeister, W. P., Guilligay, D., Cusack, S., Wadell, G., and Arnberg, N. (2004). Crystal structure of species D adenovirus fiber knobs and their sialic acid binding sites. J. Virol. 78: 7727 – 7736. 26. Leissner, P., et al. (2001). Influence of adenoviral fiber mutations on viral encapsidation, infectivity and in vivo tropism. Gene Ther. 8: 49 – 57. 27. Law, L., and Davidson, B. L. (2002). Adenovirus serotype 30 fiber does not mediate transduction via the coxsackie adenovirus receptor. J. Virol. 76: 656 – 661. 28. van Raaij, M. J., Mitraki, A., Lavigne, G., and Cusack, S. (1999). A triple beta-spiral in the adenovirus fibre shaft reveals a new structural motif for a fibrous protein. Nature 401: 935 – 938. 29. Chroboczek, J., Ruigrok, R. W., and Cusack, S. (1995). Adenovirus fiber. Curr. Top. Microbiol. Immunol. 199: 163 – 200.

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

doi:10.1016/j.ymthe.2005.05.017

30. Wu, E., et al. (2003). Flexibility of the adenovirus fiber is required for efficient receptor interaction. J. Virol. 77: 7225 – 7235. 31. Shayakhmetov, D. M., and Lieber, A. (2000). Dependence of adenovirus infectivity on length of the fiber shaft domain. J. Virol. 74: 10274 – 10286. 32. Chiu, C. Y., Wu, E., Brown, S. L., Von Seggern, D. J., Nemerow, G. R., and Stewart, P. L. (2001). Structural analysis of a fiber-pseudotyped adenovirus with ocular tropism suggests differential modes of cell receptor interactions. J. Virol. 75: 5375 – 5380.

MOLECULAR THERAPY Vol. 12, No. 4, October 2005 Copyright C The American Society of Gene Therapy

REVIEW

33. Smith, T. A., et al. (2003). Adenovirus serotype 5 fiber shaft influences in vivo gene transfer in mice. Hum. Gene Ther. 14: 777 – 787. 34. Bailey, A., and Mautner, V. (1994). Phylogenetic relationships among adenovirus serotypes. Virology 205: 438 – 452. 35. Crawford-Miksza, L., and Schnurr, D. P. (1996). Analysis of 15 adenovirus hexon proteins reveals the location and structure of seven hypervariable regions containing serotype-specific residues. J. Virol. 70: 1836 – 1844.

609