Analysis of three structurally related antiviral compounds in complex with human rhinovirus 16.

Rhinoviruses are a frequent cause of the common cold. A series of antirhinoviral compounds have been developed that bind into a hydrophobic pocket in the viral capsid, stabilizing the capsid and interfering with cell attachment. The structures of a variety of such compounds, complexed with rhinovirus serotypes 14, 16, 1A, and 3, previously have been examined. Three chemically similar compounds, closely related to a drug that is undergoing phase III clinical trials, were chosen to determine the structural impact of the heteroatoms in one of the three rings. The compounds were found to have binding modes that depend on their electronic distribution. In the compound with the lowest efficacy, the terminal ring is displaced by 1 A and rotated by 180 degrees relative to the structure of the other two. The greater polarity of the terminal ring in one of the three compounds leads to a small displacement of its position relative to the other compounds in the hydrophobic end of the antiviral compound binding pocket to a site where it makes fewer interactions. Its lower efficacy is likely to be the result of the reduced number of interactions. A region of conserved residues has been identified near the entrance to the binding pocket where there is a corresponding conservation of the mode of binding of these compounds to different serotypes. Thus, variations in residues lining the more hydrophobic end of the pocket are primarily responsible for the differences in drug efficacies.

Structures of four methyltetrazole-containing antiviral compounds in human rhinovirus serotype 14.

Four novel antiviral WIN compounds, that contain a methyl tetrazole ring as well as isoxazole, pyridazine or acetylfuran rings, have had their structures determined in human rhinovirus serotype 14 at 2.9 A resolution. These compounds bind in the VP1 hydrophobic pocket, but are shifted significantly towards the pocket pore when compared to previously examined WIN compounds. A putative water network at the pocket pore is positioned to hydrogen bond with these four WIN compounds, and this network can account for potency differences seen in structurally similar WIN compounds.

[(Biaryloxy)alkyl]isoxazoles: picornavirus inhibitors.

A series of biphenyl analogs, 6, of 5-[5-(2,6-dichloro-5-oxazolylphenoxy)pentyl]-3-methylisoxazole (2) have been synthesized and tested in vitro against 10 human rhinovirus serotypes in a TCID50 assay. The most potent compound in the series 6s, 3-[3-[2,6-dimethyl-4-(4-fluorophenyl)-phenoxy]propyl]-3-methylisoxazole , was screened against an additional 84 serotypes. It was found to be active against 64 of the serotypes, while 87 serotypes were sensitive to 2 at < 3 micrograms/mL. On comparison of the active serotypes, 6s exhibited greater potency versus 2. Analogs 6a-c,s were examined for in vitro metabolic stability by monkey liver microsomal assay. These analogs exhibited a greater than 7-fold improvement (t1/2 > 200 min) in metabolic stability compared with 2 (t1/2 > 27 min).

Picornavirus inhibitors: trifluoromethyl substitution provides a global protective effect against hepatic metabolism.

Several modifications of the oxazoline ring of WIN 54954, a broad spectrum antipicornavirus compound, have been prepared in order to address the acid lability and metabolic instability of this compound. We have previously shown that the oxadiazole analogue 3 displayed comparable activity against a variety of rhinoviruses and appeared to be stable to acid. A monkey liver microsomal assay was developed to examine the metabolic stability in vitro of both compounds, and it was determined that WIN 54954 displayed 18 metabolic products while 3 was converted to 8 products. Two major products of 3 were determined by LC-MS/MS to be monohydroxylated at each of the terminal methyl groups. Replacement of the methyl on the isoxazole ring with a trifluoromethyl group, while preventing hydroxylation at this position, did not reduce the sensitivity of the molecule to microsomal metabolism at other sites. However, the (trifluoromethyl)oxadiazole 9 not only prevented hydroxylation at this position but also provided protection at the isoxazole end of the molecule, resulting in only two minor products to the extent of 4%. The major product was identified as the monohydroxylated compound 23. The global metabolic protective effect of trifluoromethyl group on the oxadiazole ring was further demonstrated by examining a variety of analogues including heterocyclic replacements of the isoxazole ring. In each case, the trifluoromethyl analogue displayed a protective effect when compared to the corresponding methyl analogue.

An evaluation of the antirhinoviral activity of acetylfuran replacements for 3-methylisoxazoles. Are 2-acetylfurans bioisosteres for 3-methylisoxazoles?

As a probe of the 3-methylisoxazole portion of our broad-spectrum antipicornaviral series, a panel of 2-acetylfuran analogues was prepared as replacements for the 3-methylisoxazole ring. Comparison of the two series showed remarkable similarity in potency, spectrum of activity, logP, and electrostatic parameters. X-ray studies of 21b bound to human rhinovirus-14 showed that the 2-acetyl group adopted a syn conformation and the carbonyl oxygen acts as a hydrogen bond acceptor with ASN219 in much the same way as the nitrogen of the isoxazole. The importance of the syn conformation and the hydrogen-bonding capability was confirmed by the reduced antiviral activity of the 2-methylfuran and 2-formylfuran analogues. From the results of this study, it is apparent that the syn-2-acetylfuran ring is acting as a bioisostere for the 3-methylisoxazole.

Oxadiazoles as ester bioisosteric replacements in compounds related to disoxaril. Antirhinovirus activity.

A series of 1,2,4-oxadiazoles has been prepared as ester bioisosteres and tested against 15 human rhinovirus serotypes, and the MIC80, the concentration which inhibits 80% or 12 of the serotypes tested, was determined. Homologation of the alkyl group attached to the oxadiazole ring resulted in a reduction in activity with increased chain length. Introduction of hydrophilic groups in this position rendered the compounds inactive. Increasing the length of the side chain attached to the isoxazole ring resulted in an increase in activity. Replacement of the methyl with alkoxyalkyl substituents retained activity; however, introduction of a hydroxyl group on to the side chain reduced activity. Compound 8a, where both the isoxazole and oxadiazole rings were substituted with methyl groups, was one of the most active compounds in the series. A comparison was made between 8a and the two isomeric oxadiazoles 41 and 46, and an attempt was made to explain the difference in activity by examining electrostatic potential maps and by an energy profiling study. No conclusive results were obtained from these studies.

Human rhinovirus 14 complexed with fragments of active antiviral compounds.

Crystallographic studies of human rhinovirus 14 (HRV14) crystals soaked with fragments of antiviral WIN compounds, at high concentrations (82-200 micrograms/ml), show the compounds bind into the hydrophobic beta-barrel (WIN pocket) of VP1. Two of these short compounds (5-[3,5-dimethyl-4-hydroxyphenyl]-2-methyltetrazole and phenol oxazoline) cause conformational changes in the virus similar to the active, longer WIN compounds. In addition, thermostabilization studies suggest these short WIN compounds provide some stability to the HRV14 capsid. We conclude that the short compounds appear to mimic the cellular cofactors observed in the hydrophobic pocket of VP1 for some picornaviruses. Both cofactors and short WIN compounds bind into the pocket, cause conformational changes in VP1, and provide a small degree of virion stabilization but are unlikely to inhibit attachment. Three specific binding sites for dimethyl sulfoxide (DMSO), used as solvent, were also identified. One of the DMSO molecules binds into the drug binding pocket near the pocket opening, while the other two bind in the canyon near the VP1 protomer-protomer interface.

Antipicornavirus activity of tetrazole analogues related to disoxaril.

A series of tetrazole analogues of Win 54954, a broad-spectrum antipicornavirus compound, has been synthesized to address the acid lability of the oxazoline ring of this series of compounds. The results of X-ray crystallography studies of several members of the oxazoline series bound to human rhinovirus type 1A and 14 have been used to design compounds in the tetrazole series with a broad spectrum of activity. Compound 16b, which has a three-carbon linkage between the isoxazole and phenyl rings and a propyl chain extending from the isoxazole ring, exhibiting an MIC80 for 15 rhinovirus serotypes of 0.20 microM as compared to 0.40 microM for Win 54954. X-ray studies of 16b bound to human rhinovirus-14 show that the propyl side chain extends into a pore in the binding site with the possibility of hydrophobic interactions with a pocket formed by Leu106 and a portion of Ser107.

The structure of human rhinovirus 16.

BACKGROUND: Rhinoviruses and the homologous polioviruses have hydrophobic pockets below their receptor-binding sites, which often contain unidentified electron density ('pocket factors'). Certain antiviral compounds also bind in the pocket, displacing the pocket factor and inhibiting uncoating. However, human rhinovirus (HRV)14, which belongs to the major group of rhinoviruses that use intercellular adhesion molecule-1 (ICAM-1) as a receptor, has an empty pocket. When antiviral compounds bind into the empty pocket of HRV14, the roof of the pocket, which is also the floor of the receptor binding site (the canyon), is deformed, preventing receptor attachment. The role of the pocket in viral infectivity is not known. RESULTS: We have determined the structure of HRV16, another major receptor group rhinovirus serotype, to atomic resolution. Unlike HRV14, the pockets contain electron density resembling a fatty acid, eight or more carbon atoms long. Binding of the antiviral compound WIN 56291 does not cause deformation of the pocket, although it does prevent receptor attachment. CONCLUSIONS: We conjecture that the binding of the receptor to HRV16 can occur only when the pocket is temporarily empty, when it is possible for the canyon floor to be deformed downwards into the pocket. We further propose that the role of the pocket factor is to stabilize virus in transit from one host cell to the next, and that binding of ICAM-1 traps the pocket in the empty state, destabilizing the virus as required for uncoating.

A comparison of the anti-rhinoviral drug binding pocket in HRV14 and HRV1A.

The three-dimensional structures of two human rhinovirus serotypes (HRV14 and HRV1A) are compared when complexed with various antiviral agents. Although these agents all bind into the same hydrophobic pocket, the exact viral-drug interactions differ. In the absence of drugs, the pocket is occupied by a fatty acid in HRV1A, but is empty in HRV14 except for two water molecules. The conformation of each drug is dependent upon the shape of the hydrophobic pocket. In HRV14 the major residues determining the shape of the binding site are Y1128, P1174 and M1224, corresponding to I1125, M1169 and I1220 in HRV1A. When there is no cofactor or a drug in the pocket, the entrance to the pocket is open. However, the entrance is closed when the pocket is occupied by a cofactor or a drug. There are relatively small conformational changes when the agents displace the natural cofactor in HRV1A. In contrast, there are much larger conformational changes on binding a drug in HRV14. These differences cause an inhibition of viral attachment in HRV14 but not in HRV1A. Binding of the drugs results in three additional interprotomer hydrogen bonds in HRV14 and one in HRV1A. These hydrogen bonds and a potential loss of flexibility upon efficient packing of the pocket may contribute to the inhibition of uncoating in both serotypes.

Conformationally restricted analogues of disoxaril: a comparison of the activity against human rhinovirus types 14 and 1A.

A series of conformationally restricted analogs of disoxaril has been synthesized and evaluated against human rhinovirus types (HRV) 14 and 1A. The sensitivity of these serotypes to this series varied and was dependent upon the length of the molecule as well as upon the flexibility of the aliphatic chain. Minimum energy conformations of these compounds were overlaid with the X-ray structure of a closely related analog 9 bound to the capsid protein of both HRV-14 and -1A and then modeled in the compound-binding site of both serotypes. A comparative sweep volume of these compounds about the isoxazole ring revealed an inaccessible region of space for the cis-olefin 8b, which is not the case for either the trans-olefin 8a or the acetylene 5. This region may be important to the binding of the compounds to the HRV-14 site particularly during entry into the pocket.

Antirhinoviral activity of heterocyclic analogs of Win 54954.

A variety of heterocyclic analogs of Win 54954 have been synthesized and tested in vitro against human rhinovirus type 14 (HRV-14) in a plaque reduction assay. The more active compounds were tested against 14 additional serotypes, and the concentration which inhibited 80% of the serotypes tested (MIC80) was measured. One compound, 37, exhibited activity comparable to Win 59454. Physicochemical as well as electrostatic parameters were calculated and the results subjected to a QSAR analysis in an effort to explain differences in activity observed between these compounds; however, no meaningful correlation with biological activity was found with any of these parameters.

CoMFA analysis of the interactions of antipicornavirus compounds in the binding pocket of human rhinovirus-14.

A CoMFA analysis of eight compounds related to disoxaril whose X-ray structures bound to HRV-14 had been determined resulted in a strong positive correlation of activity with steric effects of the compounds, particularly toward the pore end of the compound binding site, and no correlation with electrostatic effects. These results confirm what had been previously found, that the activity of these compounds was highly dependent upon their hydrophobic nature as expressed by log p. The CoMFA study also confirmed the results from the comparison of a series of active and inactive compounds using volume maps which showed that bulk at the pore end of the molecule was conducive to high levels of antiviral activity while excessive bulk around the ring led to poor activity.

Binding affinities of structurally related human rhinovirus capsid-binding compounds are related to their activities against human rhinovirus type 14.

The binding affinities (Kds) and the rates of association and dissociation of members of a chemical class of antiviral compounds at their active sites in human rhinovirus type 14 (HRV-14) were determined. On the basis of analysis by LIGAND, a nonlinear curve-fitting program, of saturation binding experiments with HRV-14, the Kds for Win 52084, Win 56590, disoxaril (Win 51711), and Win 54954 were found to be 0.02, 0.02, 0.08, and 0.22 microM, respectively. The independently determined kinetic rates of association and dissociation resulted in calculated Kd values which were in agreement with the Kd values determined in saturation binding experiments. Scatchard plots of each of four compounds for the binding data indicated that approximately 40 to 60 molecules were bound per HRV-14 virion. Hill plots showed no evidence of cooperativity in binding. Furthermore, the antiviral activities (MICs in plaque reduction assays with HRV-14) for this limited series of compounds (n = 4) correlated well (r = 0.997) with the observed Kds. Likewise, the absence of detectable binding of Win 54954 to the drug-resistant mutant HRV-14 (Leu-1188) corresponded to a lack of antiviral activity. The positive relationship between the antiviral activities and the Kds that were determined may have implications for the molecular design of capsid-binding antirhinovirus drugs.

Stabilization of human rhinovirus serotype 2 against pH-induced conformational change by antiviral compounds.

Four WIN compounds with anti-picornavirus activities were tested for their ability to stabilize human rhinovirus serotype 2 (HRV-2) against low pH-induced conformational changes in vitro, as determined by specific immunoprecipitation. These results were compared to the minimal inhibitory concentration (MIC) as measured in a plaque reduction assay. A direct relationship was observed between the concentration of the compound that prevented the low pH-induced conformational changes and the MIC, indicating that stabilization is an important element in the mode of action of these drugs against HRV-2.

Human rhinovirus 14 complexed with antiviral compound R 61837.

The binding of the antirhinoviral agent R 61837 to human rhinovirus 14 has been examined by X-ray crystallographic methods. The compound R 61837 binds in the same pocket (underneath the canyon floor) as the "WIN" antirhinoviral agents. It does not penetrate as far into the pocket but causes similar conformational changes in the virus capsid. The movement of residues 1217 to 1221 of viral protein 1 (in the "FMDV loop") is more pronounced for R 61837 than for WIN compounds. Although both R 61837 and WIN antiviral agents partially fill the same hydrophobic pocket, atomic binding interactions differ, showing that considerable diversity in the nature of antiviral agents is possible.

A model for compounds active against human rhinovirus-14 based on X-ray crystallography data.

A number of (oxazolinylphenyl)isoxazoles have been synthesized and tested against human rhinovirus-14 (HRV-14). Several of the more active compounds have been examined by X-ray crystallography and their orientation in the compound binding site on the capsid protein of HRV-14 has been determined. Based on the minimum inhibitory concentration against HRV-14 and the X-ray conformation of the compounds, a model has been developed which distinguishes between the space-filling properties of the active and inactive compounds in this series. The model was generated by overlaying composite structures and comparing the van der Waals generated volume maps. The results of this study indicate that inactive compounds display areas of excessive bulk particularly around the phenyl ring, while the active compounds occupy space below the pore area of the compound binding site.

In vitro and in vivo activities of WIN 54954, a new broad-spectrum antipicornavirus drug.

WIN 54954 (5-[5-[2,6-dichloro-4-(4,5-dihydro-2-oxazolyl)phenoxy]pentyl]-3- methylisoxazole) is a new member of the class of broad-spectrum antipicornavirus compounds known to bind in a hydrophobic pocket within virion capsid protein VP1. In plaque reduction assays, WIN 54954 reduced plaque formation of 50 of 52 rhinovirus serotypes (MICs ranged from 0.007 to 2.2 micrograms/ml). A concentration of 0.28 microgram/ml was effective in inhibiting 80% of the 52 serotypes tested (EC80). WIN 54954 was also effective in inhibiting 15 commonly isolated enteroviruses, with an EC80 of 0.06 microgram/ml. Furthermore, WIN 54954 was effective in reducing the yield of two selected enteroviruses in cell culture by 90% at concentrations approximately equal to their MICs. The therapeutic efficacy of intragastrically administered WIN 54954 was assessed in suckling mice infected with coxsackievirus A-9 or echovirus type 9 (Barty) 2.5 days prior to initiation of therapy. Single daily doses of 2 and 100 mg/kg protected 50% of the mice from developing paralysis (PD50) following infection with coxsackievirus A-9 and echovirus-9, respectively. At the PD50 doses for these two viruses, levels of WIN 54954 in serum were maintained above the in vitro MICs for a significant portion of the dosing interval. The dose-dependent reduction in viral titers observed in coxsackievirus A-9-infected mice correlated well with the therapeutic dose response. The potency and spectrum of WIN 54954 make it a potentially useful compound for the treatment of human enterovirus and rhinovirus infections.