Structural Studies on Human Rhinovirus 14 Drug-resistant Compensation Mutants

Transcription

1 J. Mol. Biol. (1995) 253, Structural Studies on Human Rhinovirus 14 Drug-resistant Compensation Mutants Andrea T. Hadfield 1, Marcos A. Oliveira 1, Kyung H. Kim 1, Iwona Minor 1 Marcia J. Kremer 1, Beverly A. Heinz 2, Deborah Shepard 2 Daniel C. Pevear 3, Roland R. Rueckert 2 and Michael G. Rossmann 1 * 1 Department of Biological Structures have been determined of three human rhinovirus 14 (HRV14) Sciences, Purdue University compensation mutants that have resistance to the antiviral capsid binding West Lafayette compounds WIN and WIN In addition, the structure of HRV14 IN , USA is reported, with a site-directed mutation at residue 1219 in VP1. A 2 spontaneous mutation occurs at the same site in one of the compensation Institute of Molecular mutants. Some of the mutations are on the viral surface in the canyon and Virology, University of some lie within the hydrophobic binding pocket in VP1 below the ICAM Wisconsin, Madison footprint. Those mutant virus strains with mutations on the surface bind WI 53706, USA better to cells than does wild-type virus. The antiviral compounds bind to 3 ViroPharma, Inc., 1250 S. the mutant viruses in a manner similar to their binding to wild-type virus. Collegeville Road, Collegeville The receptor and WIN compound binding sites overlap, causing competition PA , USA between receptor attachment and antiviral compound binding. The compensation mutants probably function by shifting the equilibrium in favor of receptor binding. The mutations in the canyon increase the affinity of the virus for the receptor, while the mutations in the pocket probably decrease the affinity of the WIN compounds for the virus by reducing favorable hydrophobic contacts and constricting the pore through which the antiviral compounds are thought to enter the pocket. This is in contrast to the resistant exclusion mutants that block compounds from binding by increasing the bulk of residues within the hydrophobic pocket in VP Academic Press Limited *Corresponding author Keywords: picornavirus; structure; drug resistance; antiviral compound; mutant Introduction Among the problems encountered in the development of antiviral drugs is the rapid selection for drug-resistant mutants (Nanni et al., 1993). Here we examine the structures of some mutant human rhinoviruses of serotype 14 that have developed Present address: M. A. Oliveira, Department of Chemistry & Biochemistry, Welch Hall 5.262, University of Texas, Austin, TX 78712, USA. Present address: K. H. Kim, Jang Mee Apt , Shin Chun Dong, Song Pagu, Seoul 134, Korea. Present address: B. A. Heinz, Eli Lilly and Company, Lilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN 46285, USA. Present address: D. Shepard, Department of Microbiology, University of Minnesota Medical Center, Minneapolis, MN 55455, USA. Abbreviations used: HRV, human rhinovirus; MIC, minimal inhibitory concentration. resistance to capsid binding compounds that inhibit attachment and uncoating in the wild-type virus. Rhinoviruses belong to the picornavirus family and are small, icosahedral, non-enveloped particles containing one positive RNA strand, about 7.5 kb long. There are over 100 serologically distinct human rhinoviruses (Rueckert, 1990), which makes vaccine development impractical. On the other hand, capsid binding inhibitors have a wide serotype spectrum even extending to many enteroviruses (Andries et al., 1990). The three-dimensional structure of three rhinovirus serotypes (Rossmann et al., 1985; Kim et al., 1989; Oliveira et al., 1993) have been previously reported. Each of the 60 protomers that form the icosahedral protein capsid contain four viral proteins, VP1, VP2, VP3 and VP4. The three larger proteins, VP1, VP2 and VP3, have a molecular mass of roughly 30,000 daltons each and form the exterior of the protein shell (Figure 1). These three proteins have a common eight-stranded antiparallel -barrel /95/ $12.00/ Academic Press Limited

2 62 Understanding Drug Resistance in Mutants of HRV14 motif. Their amino termini intertwine in the inner surface of the protein shell forming a network of interactions that stabilize the capsid. VP4 is a small internal polypeptide. The major group of rhinovirus serotypes (Abraham & Colonno, 1984) use ICAM-1 as their cellular receptor (Greve et al., 1989; Staunton et al., 1989). The molecular interaction between some major group rhinoviruses and ICAM-1 has been characterized (Greve et al., 1989; Springer, 1990) and examined by electron microscopy for HRV16 (Olson et al., 1993) and HRV14 (P. R. Kolatkar, N. H. Olson, J. M. Greve, T. S. Baker, M. G. Rossmann, unpublished results). Rhino- and enteroviruses can be inhibited by a variety of capsid binding compounds that inhibit uncoating (McSharry et al., 1979) and in some cases attachment (Pevear et al., 1989; Shepard et al., 1993). The structure of some of these compounds complexed with various rhinoviruses (Smith et al., 1986; Badger et al., 1988; Chapman et al., 1991; Zhang et al., 1992; Kim et al., 1993), polioviruses (Grant et al., 1994; Hiremath et al., 1995) and coxsackievirus (Muckelbauer et al., 1995) have been determined. The Sterling Winthrop Pharmaceutical Research Division developed a series of such compounds identified by a WIN number (Diana et al., 1985; Otto et al., 1985). All these compounds bind into a hydrophobic pocket within the -barrel of VP1. This pocket lies beneath the canyon that surrounds each of the icosahedral 5-fold axes (Figure 1 and Smith et al., 1986), which is the site of receptor attachment. The WIN compounds have been shown to stabilize the virion against denaturation by acid or heat (Heinz et al., 1990; Bibler-Muckelbauer et al., 1994), presumably by filling a hydrophobic pocket (Ericksson et al., 1992; Oliveira et al., 1993; Grant et al., 1994). When complexed to wild-type HRV14, the WIN compounds cause conformational changes of up to 4.5 Å in the C coordinates of the strand H and the GH loop, and smaller changes in the two flanking strands C and E (Figure 2(b)). This alters the structure of the floor and walls of the canyon (Smith et al., 1986; Badger et al., 1988). On the other hand, in HRV16, a serotype that, like HRV14, recognizes ICAM-1 as its receptor, there are no major conformational changes on the surface of the virus capsid when binding WIN compounds (Oliveira et al., 1993) because the antiviral agents merely replace a natural pocket factor found within the hydrophobic binding pocket. The pocket factor is an electron density feature that has been suggested to be palmitate or sphingosine in polioviruses (Hogle et al., 1985; Filman et al., 1989; Yeates et al., 1991) and a smaller fatty acid in two serotypes of rhinovirus (Kim et al., 1993; Oliveira et al., 1993). The pocket factor holds the virion capsid in a conformation similar to that of HRV14 when complexed with a WIN compound. Similar pocket factors have been found in coxsackievirus B3 (Muckelbauer et al., 1995) and bovine enterovirus (Smyth et al., 1995). Drug-resistant variants may be selected from picornaviruses grown in the presence of WIN compounds (Heinz et al., 1989; Shepard et al., 1993). Resistance in HRV14 could be traced in most cases to substitutions in the capsid of a single amino acid residue, resulting from one base change. At least two classes of mutations have been recognized (Heinz et al., 1989; Shepard et al., 1993). In the exclusion mutants, the mutations were found to be within the hydrophobic binding pocket and were invariably small residues that had been altered to residues with larger volume, thereby excluding compounds from the pocket (Heinz et al., 1989; Badger et al., 1989a). The compensation mutants mapped to sites around the hydrophobic binding pocket in the vicinity of the canyon (Figure 2). Thermal stabilization tests indicated that the antiviral compounds could bind to compensation mutants, in contrast to exclusion mutants (Heinz et al., 1990). Here we report structures for three drug-resistant compensation mutants (Table 1) of HRV14, some in the presence of the compounds WIN or WIN (Table 2). We also report the structure of HRV14 with a site-directed mutation Asn1219 Ala, which coincides with the site of the compensation mutation Asn1219 Ser. Results All the mutations studied crystallographically involved a reduction of the size of the amino acid side-chain, resulting primarily in negative peaks in the difference electron density maps (Figure 3). Difference electron density maps were calculated for the mutant viruses in the presence of WIN compounds using the data for native HRV14, and showed clearly that the compounds bind into the hydrophobic pocket (Figure 4). The conformational changes observed when the compounds bind to the mutant viruses are similar to within 0.5 Å of those observed when the compounds bind to the wild-type virus, except for the side-chains of two residues (see below). This was true even for the GH loop that undergoes the largest conformational changes relative to wild-type upon compound binding. Positive electron densities in maps calculated using coefficients (F mutant F wild-type ) indicated the presence of the antiviral compounds and conformational changes in the protein surrounding the binding site. However, in difference maps calculated using coefficients (F mutant F wild-type ) there were negative electron densities in the position where the compound was expected to bind, and in parts of the complex structure around the binding pocket, indicating a return to the uncomplexed conformation relative to the WIN dataset. This can be explained in terms of lower occupancy of the compound WIN versus WIN under Residues are numbered sequentially, starting at 1000, 2000, 3000 and 4000 for the polypeptide chains VP1, VP2, VP3 and VP4, respectively. The mutation of an asparagine to an alanine at position 219 in VP1 will be described in the form Asn1219 Ala in the text and in the abbreviated form N1219A in tables and diagrams in this paper.

3 Understanding Drug Resistance in Mutants of HRV14 63 Figure 1. Diagrammatic view of a picornavirus with enlargement of one icosahedral asymmetric unit, showing the outline of the canyon and the entrance to the hydrophobic binding pocket. The thick line on the virion outlines one protomer, an assembly intermediate. similar conditions of concentration. WIN has a higher minimum inhibitory concentration than WIN 52084, implying a smaller affinity for binding (Fox et al., 1986). Asn1105 is situated on the C strand that separates the binding pocket and the canyon floor. The structure of the Asn1105 Ser mutant was studied in the presence of both WIN (Figure 4(a) and (b)) and WIN These compounds bind in opposite directions in the hydrophobic binding pocket of the wild-type virus (Table 2). Both compounds bound to the Asn1105 Ser mutant in the same position and orientation as in wild-type HRV14. Asn1105 faces towards the bottom of the canyon, and away from the pore (the pocket entrance) in the wild-type structure, both with and without bound WIN compound. However, when residue 1105 is a serine, the side-chain adopts an alternate rotamer (preferred conformation) that brings its OH group into the pore of the binding pocket (Figure 5). This is shown by a large negative peak and an accompanying smaller positive peak in the difference electron density calculated both with respect to the wild-type structure and with respect to the wild-type structure with bound WIN (Figure 3b). The mutant structures have a shift of up to 0.5 Å in the main chain between residues 1103 and In addition the side-chain of Asn1219 adopts a different rotomer, bringing the carboxyamide group within 4.5 Å of the hydroxyl group of Ser1105. This brings the Asn1219 side-chain closer to the C strand, increasing its distance from WIN compound and constricting the pore. This rearrangement within the hydrophobic binding pocket does not cause any significant conformational change in the floor of the canyon. The mutant Asn1219 Ser was also studied in both the presence of WIN (Figure 4(c)) and WIN These compounds bound in the hydrophobic binding pocket in the same position and orientation as observed in complexes with wild-type HRV14. In both complexes the serine at 1219 points along the top of the pocket, across the pore, 120 degrees away from the conformation of the asparagine in the wild-type complex structures. This brings the side-chain into closer proximity to the C strand, within 4.5 Å from the carbonyl group of residue The side-chain of residue 1219 probably makes a hydrogen bond with bound WIN compounds in the wild-type virus structures but not in the mutant structures (e.g. distance from atom O of Asn1219 to atom N3 of WIN is 3.6 Å in the wild-type and 5.8 Å in the mutant WIN complex). Conformational changes of less than 0.5 Å occur in the position of the main chain near Ser1219 and His1220 relative to the position of the wild-type

4 64 Understanding Drug Resistance in Mutants of HRV14 complex structure. The conformational changes cause a similar constriction of the pore and reduction in interactions with the WIN compounds to those observed for the mutant Asn1105 Ser. The site-directed mutation Asn1219 Ala was introduced to investigate the significance of the putative hydrogen bond between Asn1219 and the WIN compounds (Badger et al., 1989b; Kim et al., 1993). However, there was no change in the minimal inhibitory concentration (MIC) value upon mutation of this residue in tests against four different compounds including WIN (Table 2), which raised doubt about the existence or importance of the hydrogen bond. The mutation was easily recognizable in difference electron density maps between the mutant and wild-type data both with and without bound WIN There was no significant conformational change in the canyon floor, when either comparing the mutant structure with the wild-type or comparing the structure of the mutant/compound complex to the wild-type/compound complex. In the Asn1219 Ala mutant complex structure, the alanine side-chain in the mutant virus was at a reasonable van der Waals distance to the nearest atom in the WIN compound. The isoxazole ring of the compound, situated in the pore of the binding pocket, moved radially outwards towards the canyon floor by about 0.7 Å relative to the wild-type structure. Apart from the isoxazole group the compound appears to be in an identical position in the mutant and the wild-type. The electron density corresponding to the isoxazole ring indicated that it had rotated by about 30 relative to the wild-type structure, to form a close interaction (3.0 Å) between O1 and N2 in the isoxazole ring and S in Met1221. In the wild-type, the closest approach of any atom in WIN to the sulfur is 3.9 Å. Atom O of asparagine 1219, which makes the putative hydrogen bond, is 3.9 Å from atom N2 in the isoxazole group but only 3.4 Å from the methyl substituent of that group. The results suggest that whatever contribution the hydrogen bond at Asn1219 made to the binding energy of the compound is compensated for in the Asn1219 Ala mutant through a closer interaction with Met1221. In the Discussion we propose that resistance can be caused by reducing the binding affinity of the antiviral compounds. The MIC of WIN has not been measured for this mutant but structural evidence suggests that the site-directed mutation Asn1219 Ala maintains or possibly increases the binding affinity for this compound. It is therefore not surprising that the mutation does not confer resistance to the other related compounds for which the MIC has been measured. The conformational changes that occur on the binding of WIN to the mutant Ser1223 Gly are indistinguishable from those observed on the binding of this compound to wild-type virus. In the wild-type virus, Ser1223 does not interact with bound WIN compounds. The side-chain faces out of the binding pocket, with its hydroxyl group exposed in the floor of the canyon. There is therefore a slight change in the topology of the floor Figure 2(a) (legend on next page)

5 Understanding Drug Resistance in Mutants of HRV14 65 (b) Figure 2. Localization of the two categories of compensation mutations that confer resistance to antiviral compounds in HRV14. (a) Footprint of the receptor ICAM-1 (black outline) and a projection of the compensation escape mutations onto the viral surface within the icosahedral asymmetric unit. The position of the hydrophobic binding pocket is indicated by the superposition of the compound WIN (yellow). The icosahedral symmetry axes are indicated. (b) Localization of the compensation mutations with respect to the compound binding site. Some of the mutations lie within the pocket (pink) while others lie on the canyon floor (blue). The solvent-accessible entrance to the binding pocket or pore is marked on the left-hand side of the figure. The other end of the pocket, the toe, is enclosed and is more hydrophobic. The residues shown by thin black lines are those that undergo conformational change in the presence of a WIN compound. Table 1. Compensation mutants of HRV14 Structural studies Mutant in presence Mutant in presence Compound used for Location of mutations Phenotype Mutation Mutant of WIN of WIN mutant selection (see Figure 2) Compensation N1100S WIN On surface within ICAM footprint Compensation N1105S a WIN and WIN In WIN binding pocket entrance (pore) Compensation V1153I WIN On surface within ICAM footprint Compensation (double mutant) V1176A and WIN In WIN binding pocket N1198S Compensation N1219S WIN In WIN binding pocket entrance (pore) Compensation S1223G WIN and WIN On surface within ICAM footprint Wild-type N1219A Site-directed mutation In WIN binding pocket of wild-type entrance (pore) a Indicates that the crystallographic structure has been determined.

6 Table 2. Description of antiviral compounds Minimal inhibitory Compound concentration Relative resistance b (MIC) a Pore end Toe (enclosed) in wild-type HRV14 Asn1105 Ser Ser1223 Gly Asn1219 Ser Asn1219 Ala of pocket end of pocket ( g/ml) (site-directed mutation) WIN c WIN NA d a Concentration of compound required to inhibit viral plaque titer by 50%. b Relative resistance measured as MIC of mutant divided by MIC of wild-type virus (Heinz et al. 1989). c The S enantiomer, WIN (S), illustrated here, was used in all of the studies presented in this paper. d Measurement not available.

7 Understanding Drug Resistance in Mutants of HRV14 67 Figure 3. Difference electron density maps with positive density shown in blue and negative density in red. The density in this Figure and those following is contoured at approximately the level of the highest noise peak. (a) Difference electron density map around the Ser1223 Gly mutation, calculated using terms (F mutant F wild-type)wexp(i wild-type).the orientation is roughly perpendicular to that of Figure 2(b) with the floor of the canyon at the top of the Figure. The wildtype structure is shown in black. There is no evidence for changes in the virus structure due to the replacement of the serine by glycine. (b) Difference electron density for the mutation Asn1105 Ser, calculated using terms (F mutant F wild-type)wexp(i wild-type). The wild-type structure is shown in gold, and the mutant/compound complex structure is shown in black. The difference in the size of the positive and negative peaks illustrates both the presence of the mutation and the slight shift of this side-chain in the presence of the compound. WIN has been included for orientation, but density is not included for this portion of the structure. Part of the strand H (around Met1221) in the native structure can also be seen at the bottom of the Figure. This strand has an altered conformation compared to the native upon binding of a WIN compound in both wild-type and mutant HRV14. of the canyon in the mutant where the hydroxyl group is missing. Difference maps show a conformational change in the DE loop near the 5-fold axis for the mutants Asn1105 Ser in the presence of WIN and of Asn1219 Ser in the presence of WIN or WIN (Figure 6; see Table 3). The wild-type virus binds a cation, Ca 2+, on the 5-fold axis, liganded by the five carbonyl groups of residue Asn1141. The conformational change observed in the DE loop occurred only where there was a significant negative electron density peak on the 5-axis relative to the wild-type virus (see Table 3). Thus this change appears to be related to the occupancy of the Ca 2+ site, which reflects different soaking times and possibly slight variations in ionic concentration in the dimethyl sulfoxide-compound solution. A similar change in conformation of the DE loop is observed in a difference map calculated using data from wild-type crystals of HRV14 soaked in a buffer solution from which the calcium had been omitted (Zhao et al., unpublished results). The variations could also depend on the ph environment of the crystal, which is likely to vary under different conditions of temperature and different soak times, particularly when using Tris as a buffer. In low ph studies on HRV14 some changes were also observed on the 5-fold axis (Giranda et al., 1992). Thus the observed changes in the vicinity of the Ca 2+ site are probably unrelated to the resistance of the mutant viruses to the antiviral compounds. Discussion The most significant observation made in the various studies (Table 3) is that the antiviral compounds bind into the WIN pocket of compensation mutant viruses (Figure 4), in contrast to the exclusion mutants (Badger et al., 1989a). This is true for both the canyon lining and the pocket lining compensation mutants, and is consistent with the observation that the mutants are stabilized at elevated temperatures when complexed with WIN compounds (Heinz et al., 1990). The compensation mutants fall into two categories with respect to their position in the viral capsid. Asn1100 Ser, Val1153 Ile and Ser1223 Gly all lie on the floor of the canyon within the footprint of the cellular receptor ICAM-1 (Figure 2; and Olson et al., 1993; Oliveira et al., 1993). In contrast, Asn1105 Ser, Asn1219 Ser and the double mutant Val1176 Ala and Asn1198 Ser all lie around the binding pocket of the compound, at least 6 Å from the viral surface. In Ser1223 Gly, there is a structural change within the footprint of the ICAM-1 receptor caused directly by the mutation. This is consistent with the observation that the compensation mutants Asn1100 Ser, Val1153 Ile and Ser1223 Gly have increased affinity for HeLa cells (Shepard et al., 1993) or cell membranes (Pevear et al., unpublished results). There is no evidence that the surface of the virus is altered by the WIN-pocket lining compensation

8 or 68 Understanding Drug Resistance in Mutants of HRV14 Figure 4. Compound WIN bound to compensation mutants. (a) Structures in the binding pocket with and without the compound present. The structure of the mutant Asn1105 Ser complexed with WIN is shown by thick lines and the wild-type native structure for HRV14 is shown by thin lines. The view is similar to that in Figure 2(b). (b) Electron density for the Asn1105 Ser mutant in the presence of WIN calculated using terms (F mutant F wild-type)wexp(i wild-type). (c) Electron density for the Asn1219 Ser mutant in the presence of WIN calculated using terms (F mutant F wild-type)wexp(i wild-type). mutants. However the change to smaller residues in every case should result in a decrease in binding affinity for the compound because of reduced van der Waals contacts. These mutations bring about an increase in the volume of the pocket, a reduction in interactions between the bound compound and the viral protein and a slight constriction of the pore through which the compounds are likely to enter. Oliveira et al. (1993) pointed out that the binding sites of ICAM-1 and the antiviral compounds overlap, causing competition between binding of the receptor in the canyon and binding of a compound or pocket factor in the hydrophobic pocket. Thus, increasing the affinity between virus and receptor

9 Understanding Drug Resistance in Mutants of HRV14 69 Figure 5. Conformational changes in the pore of the binding pocket. The structure of the mutant Asn1105 Ser complexed with WIN is shown by thick lines, with the native structure of wildtype HRV14 shown by broken lines and the structure of the wild-type HRV14 complexed with the same compound shown by thin lines. or decreasing the affinity of the compound for the binding pocket could result in displacement of the compounds from the pocket in the presence of the receptor (Figure 7). Competition between the receptor and antiviral compound should be more in favor of the receptor for smaller, less well binding compounds (Bibler-Muckelbauer et al., 1994). Some of the compensation mutants identified by Heinz et al. (1989) were selected only with respect to WIN The escape mutations selected against this smaller compound were mostly on the canyon surface. The larger compound WIN binds better to wild-type HRV14 than the smaller compounds and 52035, and has a correspondingly smaller MIC 50 (Table 2). The escape mutations selected against this compound were mostly within the pocket. Thus for the less well binding compound the preferred method of compensation is to increase the binding affinity of ICAM for the virus, while for the better binding compound the preferred method of compensation is to decrease the binding affinity of the compound for the virus. The mutation Ser1223 Gly is one exception to this generalization. It has been shown previously that the binding of the receptor to the virus is extremely sensitive to site-directed mutations of this residue (Colonno et al., 1988). This mutation is selected against both compounds, implying that the increase in binding affinity of receptor for virus in the presence of this mutation is sufficient to displace even a strongly binding compound. Filman et al. (1989) suggested that the pocket factor may have a biological role in stabilizing the virion. Oliveira et al. (1993) went on to propose that the competition between the viral receptor and the pocket factor may be functionally important for destabilizing the virus to encourage uncoating caused by displacement of the pocket factor upon cell entry. The present results on compensation mutants support this hypothesis, as do the observations of Smyth et al. (1995) with regard to an enterovirus. Coordinates have been deposited with the Protein Data Bank (1RUC, 1RUD, 1RUE, 1RUF, 1RUG, 1RUH, 1RUI and 1RUJ) and are available directly from the authors on request, until they have been processed and released. Experimental Methods The mutants were selected as previously described (Heinz et al., 1989), in the presence of low concentrations of WIN or WIN (Table 1), with the exception of Asn1219 Ala, which was engineered using standard protocols into an infectious clone of HRV14 (D. C. Pevear, unpublished data). The mutant Ser1223 Gly was the one most commonly selected. The compensation mutants Figure 6. Conformational changes at the 5-fold axis. Difference electron density map for the mutant Asn1219 Ser in the presence of WIN with positive density shown in blue and negative density shown in red, calculated using terms (F mutant F wild-type)wexp(i wild-type). A negative peak lies in the position where the cation usually resides. The wild-type structure for the DE loop is shown in black, and the structure for the mutant model is shown in green. For ease of interpretation, the structure is shown for only a single icosahedral asymmetric unit. However, additional electron density for other symmetry-related copies is included.

10 70 Understanding Drug Resistance in Mutants of HRV14 Table 3. Summary of results of the structural investigations L2170V Density in Comparison of mutant structure to Five-fold Mutant Data set height a pocket wild-type cation a Five-fold axis N1105S WIN Yes Change in pore of binding pocket 1.8 Shift in DE loop for around residues 1105 and 1219 residues WIN Yes Change in pore of binding pocket 0.0 around residues 1105 and 1219 S1223G Uncomplexed 2.5 No Change on viral surface at residue 1.2 Slight loss of ion 1223 relative to wild-type. No other change WIN Yes Change on viral surface at residue N1219S WIN Yes Change in binding pocket at 2.2 Shift in DE loop for residue Possible shift residues upwards of main chain by <0.5 Å at WIN Yes Change in binding pocket at 2.9 Shift in DE loop for residue Possible shift residues upwards of main chain by <0.5 Å at N1219A Uncomplexed 2.5 No Change in binding pocket at +3.2 Increase in cation residue 1219 occupancy relative to wild-type WIN Yes Change in binding pocket at +3.2 residue 1219 and in isoxazole orientation of WIN compound a Height of peak in map calculated using coefficients Fmutant(+compound) Fwild-type and phases wild-type normalized with respect to background. Figure 7. Schematic representation of the competition between receptor binding and binding of a pocket factor within the hydrophobic pocket in VP1. The structures represented in white with a heavy outline have been determined crystallographically, and the structure represented in grey with a heavy outline has been determined by electron microscopy (Olson et al. 1993). (a) In wild-type HRV14, it is presumed that the pocket factor binds weakly and hence is not observed in crystallographic studies. When WIN compounds bind into the pocket, they deform its roof, which is also the floor of the canyon. Even in the presence of compound the pocket is alternately occupied and empty, the degree of occupancy decreasing with decreasing binding affinity of the compound. We have proposed (Oliveira et al. 1993) that the receptor binds to the canyon only when the pocket is empty. (b) Some of the compensation escape mutations are found on the floor of the canyon and increase the affinity of ICAM-1 for the virus. Although WIN compounds can still bind, the mutant viruses retain much of their infectivity. Thus, the binding affinity of ICAM-1 for the mutant virus must now be greater than that of the WIN compounds, shifting the equilibrium in favor of the receptor virus complex. A consequence of increased receptor binding will be a larger number of empty pockets, and therefore presumably less stabilization of the capsid and more ready uncoating even in the presence of antiviral compound. (c) Other compensation escape mutations are found lining the hydrophobic binding pocket. They have less bulk than the wild-type residues, causing small changes that should reduce the affinity of the WIN compounds for the virus. Since the affinity of the receptor ICAM-1 for the virus is unchanged the equilibrium again moves in favor of receptor binding rather than WIN compound binding.

11 Understanding Drug Resistance in Mutants of HRV14 71 Table 4. X-ray diffraction data collected from compensation mutants of HRV14 Mutation N1105S N1105S N1219S N1219S S1223G S1223G N1219A N1219A compound WIN WIN WIN WIN None WIN None WIN Number of films (0.3 oscillation) Total number of reflections 102,231 49,305 44,671 50,793 19,491 96,067 72,027 35,325 I > n (I), where n is: Unique reflections a 89,386 46,561 42,372 48,084 19,125 b 85,028 60,802 33,721 Rm (%) c Resolution (Å) Difference coefficient (k) Synchrotron source CHESS d CHESS d CHESS d CHESS d SRS e SRS e CHESS d CHESS d Wavelength (Å) Exposure time (min, s) 3, 30 3, 30 3, 30 3, 30 7, 30 3, 16 4, 40 0, 30 1, 15 1, 40 a Most of the unique reflections were matched by observations of uncomplexed wild-type data. b Although these data represent about 4% of the computer data set to 3 Å resolution, experience has shown that the 20-fold redundancy and the high quality of the native phases assures interpretable electron density from such a small percentage of the unique data. c ( Ih siihi) h Rm = i Ih where Ih is the normalized intensity, Ihi is the intensity and si is the scale factor for the film on which reflection Ihi was recorded. d Beam line A1, Cornell High Energy Synchrotron Source, Ithaca. e Beam line 9.6, Synchrotron Radiation Source, Daresbury. h i, Ser1223 Gly, Asn1105 Ser and Asn1219 Ser were studied crystallographically. These mutant viruses were propagated, purified and crystallized using an adaptation of the wild-type HRV14 procedure (Arnold et al., 1984). The resultant crystals are cubic and belong to space group P2 13 with the cell dimension a = Å. The antiviral compounds were dissolved in dimethyl sulfoxide and soaked into the mutant crystals as described by Smith et al. (1986). The final concentration of the drug in the stabilizing solution containing the crystals was 10 g/ml. X-ray diffraction data were collected on Kodak DEF film at various synchrotrons (Table 4) using a 0.3 oscillation angle. The films were indexed with Kim s (1989) autoindexing program, and then processed and scaled using the Purdue system (Rossmann, 1979; Rossmann et al., 1979) (Table 4). Each data set was placed onto the same relative scale as the wild-type data set (Arnold et al., 1987), using local scaling in shells of resolution and then in ranges of intensity. Electron density maps were calculated with terms (F mutant kf wild-type)wexp(i mr), and averaged according to the 20-fold non-crystallographic symmetry. F mutant and F wild-type are the structure factor amplitudes for mutant and wild-type data with or without bound drug as appropriate. The molecular replacement phases, mr, and weights, w, were previously determined by non-crystallographic symmetry averaging for either wild-type HRV14 in the absence of any bound drug wild-type (Rossmann et al., 1985; Arnold et al., 1987) or for wild-type HRV14 complexed with WIN wild-type (Badger et al., 1988, 1989b). The phase determination depended only on the non-crystallographic symmetry and observed structure amplitudes and was therefore independent of the assumption of any molecular model. The weights are a measure of the quality of the phases (Arnold et al., 1987). The value of k was set to 1.0 for computation of difference maps. Difference maps for mutant virus in the presence of either WIN or WIN were calculated in relation to both wild-type HRV14 and wild-type complexed with (for which a moderately complete data set was available). These maps were used to identify significant differences between the mutant and wild-type structures. The uncomplexed wild-type data were collected prior to the spontaneous mutation of residue 2170 (Leu Val), which occurred in the course of passaging the virus at the end of All the remaining data were collected after Thus, all difference maps with respect to the wild-type uncomplexed data have a negative peak at this site, which acted as a convenient control for the quality of the difference maps. A putative cation site on the 5-fold axis associated with Asn1141 has been described previously (Giranda et al., 1992) and more recently has been shown to be a Ca 2+ (Zhao, unpublished results). The occupancy of this cation site varies among crystals soaked with different compounds for different lengths of time. The relative height of this peak above the background on difference maps has therefore been recorded in Table 3. The relative heights were determined by measuring the level to which background peaks are just present in an arbitrarily chosen 25 Å 3 box, and then dividing the height of a peak by this background measurement. Maps were also calculated with the value of k such that the heights of the electron density of the compound and the protein were about equal (Smith et al., 1986). A suitable choice of k compensates for incomplete substitution of compound and phase bias toward the wild-type structure (Kim et al., 1993). These maps were displayed using the programs FRODO (Jones, 1978) and O (Jones et al., 1991) and used, along with previously described difference maps, to interpret the conformational changes induced by compound binding. Acknowledgements We are grateful for help with data collection at synchrotrons by many members of our laboratory, and the help of the staffs of the CHESS and Daresbury synchrotrons. The work was supported by grants from NIH and from Sterling Winthrop to M.G.R.

12 72 Understanding Drug Resistance in Mutants of HRV14 References Abraham, G. & Colonno, R. J. (1984). Many rhinovirus serotypes share the same cellular receptor. J. Virol. 51, Andries, K., Dewindt, B., Snoeks, J., Wouters, L., Moereels, H., Lewi, P. J. & Janssen, P. A. J. (1990). Two groups of rhinoviruses revealed by a panel of antiviral compounds prevent sequence divergence and differential pathogenicity. J. Virol. 64, Arnold, E., Erickson, J. W., Fout, G. S., Frankenberger, E. A., Hecht, H. J., Luo, M., Rossmann, M. G. & Rueckert, R. R. (1984). Virion orientation in cubic crystals of the human common cold virus HRV14. J. Mol. Biol. 177, Arnold, E., Vriend, G., Luo, M., Griffith, J. P., Kamer, G., Erickson, J. W., Johnson, J. E. & Rossmann, M. G. (1987). The structure determination of a common cold virus, human rhinovirus 14. Acta Crystallog. sect. A, 43, Badger, J., Minor, I., Kremer, M. J., Oliveira, M. A., Smith, T. J., Griffith, J. P., Guerin, D. M. A., Krishnaswamy, S., Luo, M., Rossmann, M. G., McKinlay, M. A., Diana, G. D., Dutko, F. J., Fancher, M., Rueckert, R. R. & Heinz, B. A. (1988). Structural analysis of a series of antiviral agents complexed with human rhinovirus 14. Proc. Natl Acad. Sci. USA, 85, Badger, J., Krishnaswamy, S., Kremer, M. J., Oliveira, M. A., Rossmann, M. G., Heinz, B. A., Rueckert, R. R., Dutko, F. J. & McKinlay, M. A. (1989a). Three-dimensional structures of drug-resistant mutants of human rhinovirus 14. J. Mol. Biol. 207, Badger, J., Minor, I., Oliveira, M. A., Smith, T. J. & Rossmann, M. G. (1989b). Structural analysis of antiviral agents that interact with the capsid of human rhinoviruses. Proteins, 6, Bibler-Muckelbauer, J. K., Kremer, M. J., Rossmann, M. G., Diana, G. D., Dutko, F. J., Pevear, D. C. & McKinlay, M. A. (1994). Human rhinovirus 14 complexed with fragments of active antiviral compounds. Virology, 201, Chapman, M. S., Minor, I., Rossmann, M. G., Diana, G. D. & Andries, K. (1991). Human rhinovirus 14 complexed with antiviral compound R J. Mol. Biol. 217, Colonno, R. J., Condra, J. H., Mizutani, S., Callahan, P. L., Davies, M. E. & Murcko, M. A. (1988). Evidence for the direct involvement of the rhinovirus canyon in receptor binding. Proc. Natl Acad. Sci. USA, 85, Diana, G., McKinlay, M. A., Otto, M. J., Akullian, V. & Oglesby, C. (1985). [ (4,5-Dihydro-2-oxazolyl)phenoxy alkyl]isoxazoles. Inhibitors of picornavirus uncoating. J. Med. Chem. 28, Eriksson, A. E., Baase, W. A., Wozniak, J. A. & Matthews, B. A. (1992). A cavity-containing mutant of T4 lysozyme is stabilized by buried benzene. Nature, 355, Filman, D. J., Syed, R., Chow, M., Macadam, A. J., Minor, P. D. & Hogle, J. M. (1989). Structural factors that control conformational transition and serotype specificity in type 3 poliovirus. EMBO J. 8, Fox, M. P., Otto, M. J. & McKinlay, M. A. (1986). The prevention of rhinovirus and poliovirus uncoating by WIN 51711: A new antiviral drug. Antimicrob. Agents Chemother. 30, Giranda, V. L., Heinz, B. A., Oliveira, M. A., Minor, I., Kim, K. H., Kolatkar, P. R., Rossmann, M. G. & Rueckert, R. R. (1992). Acid-induced structural changes in human rhinovirus 14: Possible role in uncoating. Proc. Natl Acad. Sci. USA, 89, Grant, R. A., Hiremath, C. N., Filman, D. J., Syed, R., Andries, K. & Hogle, J. M. (1994). Structures of poliovirus complexes with anti-viral drugs: Implications for viral stability and drug design. Curr. Biol. 4, Greve, J. M., Davis, G., Meyer, A. M., Forte, C. P., Yost, S. C., Marlor, C. W., Kamarck, M. E. & McClelland, A. (1989). The human rhinovirus receptor is ICAM-1. Cell, 56, Heinz, B. A., Rueckert, R. R., Shepard, D. A., Dutko, F. J., McKinlay, M. A., Fancher, M., Badger, J. & Rossmann, M. G. (1989). Genetics and molecular basis for resistance of human rhinovirus 14 to an antiviral drug. J. Virol. 63, Heinz, B. A., Shepard, D. A. & Rueckert, R. R. (1990). Escape mutant analysis of a drug-binding site can be used to map functions in the rhinovirus capsid. In Use of X-ray Crystallography in the Design of Antiviral Agents (Laver, W. G. & Air, G. M., eds), pp , Academic Press, San Diego. Hiremath, C. N., Grant, R. A., Filman, D. J. & Hogle, J. M. (1995). Binding of the antiviral drug WIN51711 to the Sabin strain of type 3 poliovirus: Structural comparison with drug binding in rhinovirus 14. Acta Crystallog. sect. D, 51, Hogle, J. M., Chow, M. & Filman, D. J. (1985). Threedimensional structure of poliovirus at 2.9 Å resolution. Science, 229, Jones, T. A. (1978). A graphics model building and refinement system for macromolecules. J. Appl. Crystallog. 11, Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the localisation of errors in these models. Acta Crystallog. sect. A, 47, Kim, K. H., Willingmann, P., Gong, Z. K., Kremer, M. J., Chapman, M. S., Minor, I., Oliveira, M. A., Rossmann, M. G., Andries, K., Diana, G., Dutko, F. J., McKinlay, M. A. & Pevear, D. C. (1993). A comparison of the antirhinoviral drug binding pocket in HRV14 and HRV1A. J. Mol. Biol. 230, Kim, S. (1989). Auto-indexing oscillation photographs. J. Appl. Crystallog. 22, Kim, S., Smith, T. J., Chapman, M. S., Rossmann, M. G., Pevear, D. C., Dutko, F. J., Felock, P. J., Diana, G. D. & McKinlay, M. A. (1989). The crystal structure of human rhinovirus 1A (HRV1A). J. Mol. Biol. 210, McSharry, J. J., Caliguiri, L. A. & Eggers, H. J. (1979). Inhibition of uncoating of poliovirus by arildone, a new anti-viral drug. Virology, 97, Muckelbauer, J. K., Kremer, M., Minor, I., Tong, L., Zlotnick, A., Johnson, J. E. & Rossmann, M. G. (1995). The structure determination of coxsackievirus B3 to 3.5 Å resolution. Acta Crystallog. sect. D, in the press. Nanni, R. G., Ding, J., Jacobo-Molina, A. J., Hughes, S. H. & Arnold, E. (1993). Review of HIV-1 reverse transcriptase three dimensional structure: Implications for drug design. J. Comp.-aided Mol. Design, Suppl. 1, Oliveira, M. A., Zhao, R., Lee, W.-M., Kremer, M. J., Minor, I., Rueckert, R. R., Diana, G. D., Pevear, D. C., Dutko, F. J., McKinlay, M. A. & Rossmann, M. G. (1993). The structure of human rhinovirus 16. Structure, 1,

13 Understanding Drug Resistance in Mutants of HRV14 73 Olson, N. H., Kolatkar, P. R., Oliveira, M. A., Cheng, R. H., Greve, J. M., McClelland, A., Baker, T. S. & Rossmann, M. G. (1993). Structure of a human rhinovirus complexed with its receptor molecule. Proc. Natl Acad. Sci. USA, 90, Otto, M. J., Fox, M. P., Fancher, M. J., Kuhrt, M. F., Diana, G. D. & McKinlay, M. A. (1985). In vitro activity of WIN 51711, a new broad-spectrum antipicornavirus drug. Antimicrob. Agents Chemother. 27, Pevear, D. C., Fancher, M. J., Felock, P. J., Rossmann, M. G., Miller, M. S., Diana, G., Treasurywala, A. M., McKinlay, M. A. & Dutko, F. J. (1989). Conformational change in the floor of the human rhinovirus canyon blocks adsorption to HeLa cell membranes. J. Virol. 63, Rossmann, M. G. (1979). Processing oscillation diffraction data for very large unit cells with an automatic convolution technique and profile fitting. J. Appl. Crystallog. 12, Rossmann, M. G., Leslie, A. G. W., Abdel-Meguid, S. S. & Tsukihara, T. (1979). Processing and post-refinement of oscillation camera data. J. Appl. Crystallog. 12, Rossmann, M. G., Arnold, E., Erickson, J. W., Frankenberger, E., Griffith, J. P., Hecht, H. J., Johnson, J. E., Kamer, G., Luo, M., Mosser, A. G., Rueckert, R. R., Sherry, B. & Vriend, G. (1985). Structure of a human common cold virus and functional relationship to other picornaviruses. Nature, 317, Rueckert, R. R. (1990). Picornaviridae and their replication. In Virology (Fields, B. N. & Knipe, D. M., eds), pp , Raven Press, New York. Shepard, D. A., Heinz, B. A. & Rueckert, R. R. (1993). WIN inhibits both attachment and eclipse of human rhinovirus 14. J. Virol. 67, Smith, T. J., Kremer, M. J., Luo, M., Vriend, G., Arnold, E., Kamer, G., Rossmann, M. G., McKinlay, M. A., Diana, G. D. & Otto, M. J. (1986). The site of attachment in human rhinovirus 14 for antiviral agents that inhibit uncoating. Science, 233, Smyth, M., Tate, J., Hoey, E., Lyons, C., Martin, S. & Stuart, D. (1995). Implications for viral uncoating from the structure of bovine enterovirus. Struct. Biol. 2, Springer, T. A. (1990). Adhesion receptors of the immune system. Nature, 340, Staunton, D. E., Merluzzi, V. J., Rothlein, R., Barton, R., Marlin, S. D. & Springer, T. A. (1989). A cell adhesion molecule, ICAM-1, is the major surface receptor for rhinoviruses. Cell, 56, Yeates, T. O., Jacobson, D. H., Martin, A., Wychowski, C., Girard, M., Filman, D. J. & Hogle, J. M. (1991). Three-dimensional structure of a mouse adapted type2/type1 poliovirus chimera. EMBO J. 10, Zhang, A., Nanni, R. G., Oren, D. A., Rozhon, E. J. & Arnold, E. (1992). Three-dimensional structure-activity relationships for antiviral agents that interact with picornavirus capsids. Semin. Virol. 3, Edited by I. A. Wilson (Received 19 May 1995; accepted 25 July 1995)