Inhibitors of urokinase-type plasminogen activator (uPA) were selected in vitro from two ecotin phage-display libraries to study the effect on binding of amino acid substitutions at critical positions 108, 110, 112, and 113 within the 100s loop (RNKL, respectively, in wild type ecotin). The first, a focused library, was the result of a computation-assisted approach using the three-dimensional structure of the ecotin-trypsin complex to guide the modeling of amino acid substitutions predicted to increase affinity for uPA. The second, a complete library, allowed for all substitutions at the above identified positions. The consensus sequences selected from the focused, and complete libraries were RRWS and R(R/N)QL, respectively. Inhibition constant determinations showed ecotin variants containing these sequences to be similarly potent (K i = 1–2 nm). These substitutions were combined with previously identified substitutions in another critical region of ecotin. One of these combinations (D70R/M84R/RRQL) is the tightest (K i = 50 pm) ecotin variant inhibitor of uPA. The blending of combinatorial methods and computer algorithms designed to predict stronger binders has allowed us to obtain protein derivatives that exhibit greatly increased affinity for a predetermined target. This technology can be applied to select for enhanced binding interactions at protein-protein interfaces and accelerate the process of protease inhibitor development.

The dimeric aspartyl protease of HIV has been the subject of intense research for almost a decade. Knowledge of the substrate specificity and catalytic mechanism of this enzyme initially guided the development of several potent peptidomimetic small molecule inhibitors. More recently, the solution of the HIV protease structure led to the structure-based design of improved peptidomimetic and non-peptidomimetic antiviral compounds. Despite the qualified success of these inhibitors, the high mutation rate associated with RNA viruses continues to hamper the long-term clinical efficacy of HIV protease inhibitors. The dimeric nature of the viral protease has been conducive to the investigation of dominant-negative inhibitors of the enzyme. Some of these inhibitors are defective protease monomers that interact with functional monomers to form inactive protease heterodimers. An advantage of macromolecular inhibitors as compared to small-molecule inhibitors is the increased surface area of interaction between the inhibitor and the target gene product. Point mutations that preserve enzyme activity but confer resistance to small-molecule inhibitors are less likely to have an adverse effect on macromolecular interactions. The use of efficient retrovirus vectors has facilitated the delivery of these macromolecular inhibitors to primary human lymphocytes. The vector-transduced cells were less susceptible to HIV infection in vitro, and showed similar levels of protection compared to other macromolecular inhibitors of HIV replication, such as RevM10. These preliminary results encourage the further development of dominant-negative HIV protease inhibitors as a gene therapy-based antiviral strategy.

The yeast two-hybrid assay was used to study the dimerization of engineered and naturally occurring variants of human immunodeficiency virus (HIV) protease (PR) monomers. Defective monomers that were previously shown to exhibit a dominant-negative (D-N) effect in cultured mammalian cells were tested for their ability to interact in the two-hybrid assay. Similarly, monomers with dimer-interface substitutions and monomers harboring in vivo selected mutations that confer multidrug resistance (mdr) in an AIDS patient were tested for interaction in yeast. Dimer formation between wt monomers with catalytic aspartates was not detected in yeast, whereas the dimerization of PR monomers harboring the acid active site substitution D25N was readily demonstrated. The use of inactive monomers harboring the D25N substitution as a genetic background for studying additional HIV PR mutations allowed for the probing of interactions between monomers with mdr-associated mutations with those based on the HIV-1 HXB2R sequence. The HTLVIII/HIV-1 HXB2R clone has been the basis for a large number of HIV-related plasmids, primers, antibodies, and other specific reagents throughout the HIV research community. The results of our assay suggest that HXB2R-based D-N PR inhibitors associate with variant monomers based on the recently obtained nucleotide sequence from an AIDS patient with a multidrug-resistant virus. These results further encourage the use of D-N PR inhibitors as antiviral agents which may complement existing small-molecule combination therapies.

Defective variants of human immunodeficiency virus type 1 (HIV-1) protease (HIV PR) have been engineered to inhibit wild-type (wt) PW PR activity. These variants were designed to promote the formation of heterodimers and to destabilize the formation of inactive variant homodimers of HIV-1 protease through substitutions at Asp-25, Ile-49, and Gly-50 (Babe, L, M., Rose, J., and Craik, C, S. (1995) Proc. Natl, Acad Sci. U.S.A. 92, 10069-10073; McPhee, F., Good, A. C., Kuntz, I. D., and Craik, C. S. (1996) Proc. Natl. Acad Sci. U.S.A. 93, 11477-11481). The mechanism of action of these dominant-negative inhibitors was established using recombinantly expressed defective monomers, The defective monomers were refolded in vitro in the presence of wt HIV PR and showed dose-dependent inhibition of proteolytic activity. This inhibition was shown to result from the formation of inactive heterodimers between defective and wt HIV PR monomers, Heterodimer formation was detected by (i) isolating refolded, inactive heterodimers using histidine-tagged defective monomers and (ii) isolating heterodimers from bacteria coexpressing both wt and defective variants of HIV PR. Single-chain variants of HIV PR, in which the C terminus of the wt HIV PR monomer was covalently tethered to the N terminus of the defective monomer, were also expressed and analyzed. Thermal denaturation of these single-chain heterodimers using differential scanning calorimetry revealed a 1.5-7.2 degrees C greater thermal stability than single-chain wt HIV PR, The thermodynamic trend shown by these three variants mirrors their relative inhibition in provirus transfection assays. These data support the model that the effects seen both in tissue culture and in vitro arise from an increase in stability conferred on these heterodimers by interface mutations and identifies heterodimer formation as their mechanism of inhibition.

The poor translation efficiency of genome-length human rhinovirus RNA in vitro using HeLa cell extract-supplemented rabbit reticulocyte lysate has hampered the study of rhinovirus IRES-mediated translation and polyprotein synthesis in a cell-free system. In contrast, the efficient in vitro translation characteristics of poliovirus RNAs have ultimately allowed the programming of cell-free coupled translation/replication extracts which are able to produce infectious poliovirus particles in vitro. A possible explanation for the decreased burst size observed during the course of a rhinovirus infection, compared to poliovirus infection, is reduced levels of polyprotein synthesis in vivo. In order to test this hypothesis and extend in vitro translation/replication technology to the study of human rhinoviruses, a chimeric cDNA construct was engineered which allowed the in vitro synthesis of T7 transcripts containing the intact poliovirus type 1 (PV1) 5' noncoding region (5' NCR) and initiation codon upstream of the human rhinovirus 14 (HRV14) polyprotein-coding region and 3'-terminal sequences. These chimeric RNAs translated efficiently in vitro and were used successfully to program a cell-free replication extract. Unexpectedly, parental HRV14 RNAs also translated efficiently in the HeLa cell-free translation/replication extract but replicated less efficiently than the chimera in vitro. The chimeric HRV14/PV1 RNAs were infectious and gave rise to a virus with a growth phenotype similar to that of parental HRV14. Preliminary characterization of this chimeric virus suggests that the biological properties characteristic of rhinovirus in vivo are determined primarily by the rhinovirus gene products. Although the translation efficiency of the HRV14 5' NCR may be a limitation in rabbit reticulocyte lysate-based in vitro translation extracts, it does not appear to be a major limiting determinant for growth of rhinovirus in vivo or replication in the HeLa cell-free extract.

The genomic RNA 3' noncoding region is believed to be a major cis-acting molecular genetic determinant for regulating picornavirus negative-strand RNA synthesis by promoting replication complex recognition. We report the replication of two picornavirus RNAs harboring complete deletions of the genomic RNA 3' noncoding regions. Our results suggest that while specific 3'-terminal RNA sequences and/or secondary structures may have evolved to promote or regulate negative-strand RNA synthesis, the basic mechanism of replication initiation is not strictly template specific and may rely primarily upon the proximity of newly translated viral replication proteins to the 3' terminus of template RNAs within tight membranous replication complexes.

The specific recognition of genomic positive strand RNAS as templates for the synthesis of intermediate negative strands by the picornavirus replication machinery is presumably mediated by cis-acting sequences within the genomic RNA 3' non-coding region (NCR). A structure-infectivity analysis was conducted on the 44 nt human rhinovirus 14 (HRV14) 3' NCR to identify the primary sequence and/or secondary structure determinants required for viral replication. Using biochemical RNA secondary structure probing techniques, we have demonstrated the existence of a single stem-loop structure contained entirely within the 3' NCR, which appears to be phylogenetically conserved within the rhinovirus genus. We also report the in vivo analysis of a number of 3' NCR deletion mutations engineered into infectious cDNA clones which were designed to disrupt the stem-loop secondary structure to varying degrees. Large deletions (up to 37 nt) resulted in defective growth phenotypes, although they were not lethal. We propose that the absolute requirements for initiation of negative strand synthesis are less stringent than previously postulated, even though defined RNA secondary structure determinants may have evolved to facilitate and/or regulate the process of viral RNA replication.

The replication of a picornavirus genomic RNA is a template-specific process involving the recognition of viral RNAs as target replication templates for the membrane-bound viral replication initiation complex. The virus-encoded RNA-dependent RNA polymerase, 3Dpol, is a major component of the replication complex; however, when supplied with a primed template, 3Dpol is capable of copying polyadenylated RNAs which are not of viral origin. Therefore, there must be some other molecular mechanism to direct the specific assembly of the replication initiation complex at the 3' end of viral genomic RNAs, presumably involving cis-acting binding determinants within the 3' noncoding region (3' NCR). This report describes the use of an in vitro UV cross-linking assay to identify proteins which interact with the 3' NCR of human rhinovirus 14 RNA. A cellular protein(s) was identified in cytoplasmic extracts from human rhinovirus 14-infected cells which had a marked binding preference for RNAs containing the rhinovirus 3' NCR sequence. This protein(s) showed reduced cross-linking efficiency for a 3' NCR with an engineered deletion. Virus recovered from RNA transfections with in vitro transcribed RNA containing the same 3' NCR deletion demonstrated a defective replication phenotype in vivo. Cross-linking experiments with RNAs containing the poliovirus 3' NCR and cytoplasmic extracts from poliovirus-infected cells produced an RNA-protein complex with indistinguishable electrophoretic properties, suggesting that the appearance of the cellular protein(s) may be a common phenomenon of picornavirus infection. We suggest that the observed cellular protein(s) is sequestered or modified as a result of rhinovirus or poliovirus infection and is utilized in viral RNA replication, perhaps by binding to the 3' NCR as a prerequisite for replication complex assembly at the 3' end of the viral genomic RNA.

Cells infected with poliovirus express a virally encoded polyprotein which undergoes self-mediated cleavage into structural and nonstructural viral proteins. Most of these cleavages are catalyzed by the 3C proteolytic domain of the polyprotein. Polyprotein synthesized in vitro from an RNA template containing a three-nucleotide insertion in 3C underwent proteolytic processing at all but one of the 3C-dependent cleavage sites. When transfected into HeLa cells, this RNA template displayed a lethal phenotype. We report here the isolation of two pseudorevertant progeny strains with restored protein-processing phenotypes, one of which appears to have arisen by transduction of a stretch of nucleotides from human 28S rRNA.