The HIV-1 Dimerization Initiation Site (DIS) is an intriguing, yet underutilized, viral RNA target for potential antiretroviral therapy. To study the recognition features of this target and to provide a quantitative, rapid, and real-time tool for the discovery of new binders, a fluorescence-based assay has been constructed. It relies on strategic incorporation of 2-aminopurine, an isosteric fluorescent adenosine analogue, into short hairpin RNA constructs. These oligomers self-associate to form a kissing loop that thermally rearranges into a more stable extended duplex, thereby mimicking the association and structural features of the native RNA sequence. We demonstrate the ability of two fluorescent DIS constructs, DIS272(2AP) and DIS273(2AP), to report the binding of known DIS binders via changes in their emission intensity. Binding of aminoglycosides such as paromomycin to DIS272(2AP) results in significant fluorescence enhancement, while ligand binding to DIS273(2AP) results in fluorescence quenching. These observations are rationalized by comparison to the sequence-analogous bacterial A-site, where the relative emission of the fluorescent probe is dependent on the placement of the flexible purine residues inside or outside the helical domain. Analysis of binding isotherms generated using DIS272(2AP) yields submicromolar EC50 values for paromomycin (0.5 +/- 0.2 microM) and neomycin B (0.6 +/- 0.2 microM). Other neomycin-family aminoglycosides are less potent binders with neamine, the core pharmacophore, displaying the lowest affinity of 21 +/- 1 microM. Screening of additional aminoglycosides and their derivatives led to the discovery of new, previously unreported, aminoglycoside binders of the HIV DIS RNA, among them butirosin A (5.5 +/- 0.6 microM) and apramycin (7.6 +/- 1.0 microM). A conformationally constrained neomycin B analogue displays a rather high affinity to the DIS (1.9 +/- 0.2 microM). Among a series of nucleobase aminoglycoside conjugates, only the uracil derivatives display a measurable affinity using this assay with EC50 values in the 2 microM range. In addition, similarity between the solution behavior of HIV-1 DIS and the bacterial decoding A-site has been observed, particularly with respect to the intra- and extra-helical residence of the conformationally flexible A residues within the bulge. Taken together, the observations reported here shed light on the solution behavior of this important RNA target and are likely to facilitate the design of new DIS selective ligands as potential antiretroviral agents.

In an extension of the previous paper, we describe the force dependence of break-up of doublets of latex spheres cross-linked by protein G-IgG bonds via the Fc region of the antibody. The receptor, the monoclonal Bear-1 antibody, was either covalently linked to 4.75-microns aldehyde/sulfate (A/S) latex spheres in a one-step reaction, or physically adsorbed to the 4.63-microns carboxyl-modified latex spheres used in Part I of this paper. The spheres were suspended in 19% buffered Dextran 40 containing the ligand, the bivalent recombinant protein G (Gamma-Bind G), and observed in the counter-rotating cone and plate Rheoscope. Break-up of doublets, tracked individually under the microscope, as well as in populations of 50-150 particles, was studied over a range of normal force from 20 to 260 pN. In individual particle studies, the fraction of doublets of spheres with covalently linked IgG breaking up in the first 10 rotations, increased from 16% in the low-force to 63% in the high-force range. In population studies, the fraction broken up increased with duration and magnitude of the applied force, and decreased with increasing ligand concentration. Moreover, doublets of physically adsorbed IgG spheres required significantly lower force than doublets of covalently linked IgG spheres for the same degree of break-up, possibly because of surface detachment of IgG molecules rather than rupture of receptor-ligand bonds. Computer simulation, using the Bell stochastic model of break-up and a Poisson distribution for the number of bonds, described in Part I, showed that the parameters of the protein-protein bond differed significantly from those of the carbohydrate-protein bond studied in Part I of this paper, the former being much more responsive to force than the latter.

No abstract