A look inside HIV resistance through retroviral protease interaction maps

Abstract

Retroviruses affect a large number of species, from fish and birds to mammals and humans, with global socioeconomic negative impacts. Here the authors report and experimentally validate a novel approach for the analysis of the molecular networks that are involved in the recognition of substrates by retroviral proteases. Using multivariate analysis of the sequence-based physiochemical descriptions of 61 retroviral proteases comprising wild-type proteases, natural mutants, and drug-resistant forms of proteases from nine different viral species in relation to their ability to cleave 299 substrates, the authors mapped the physicochemical properties and cross-dependencies of the amino acids of the proteases and their substrates, which revealed a complex molecular interaction network of substrate recognition and cleavage. The approach allowed a detailed analysis of the molecular-chemical mechanisms involved in substrate cleavage by retroviral proteases.

Figure 1. Goodness of Fit and Experimental Validation for the CRM

(A) Observed versus predicted rate of cleavage of 299 substrates by 61 retroviral proteases (in total, 760 protease–substrate combinations) for the CRM, in which all predictions relate to model-building data (R² = 0.77; Q² = 0.52). Each bullet represents naturally occurring and artificially mutated protease forms of HIV-1 (gray), HIV-2 (magenta), AMV (light green), RSV (blue), HTLV-1 (orange), BLV (red), Mo-MuLV (yellow), EIAV (green), and FIV (light blue). (B) A priori prediction of cleavage rates of 15 novel peptides with diverse structures by the CRM (Table 3, numbers 4–18). Shown is the predicted versus experimentally determined cleavage rates by HXB2 (red) and mutant HIV-1 proteases, I84V (blue), L90M (magenta), and I84 + L90M (green). The prediction error for the cleavage rates was less than one log(k_cat/K_m) unit for 68% of the protease–substrate pairs; the correlation for the a priori predicted rates versus the experimentally determined rates yielding a correlation coefficient r = 0.47 (p < 0.0001), as indicated on the panel. The data in (A) is also shown in (B) (gray).

Figure 2. External Predictions for Retroviral Proteases by CRMs

Each panel represents the predictions of a model created on the data collected herein (Table S1), but excluded all data for the proteases of one retroviral strain, one at a time, and uses the model to predict the excluded data. Blue bullets correspond to prediction of cleavage activity for new proteases and new substrates (i.e., the cases in which neither the protease nor the peptide was represented in the dataset used in creation of the model). Red bullets correspond to prediction of the cleavage rates for the new proteases only (i.e., the cases in which the peptide, but not the protease, was represented in the dataset used for model creation). Gray dots represent the observed versus computed cleavage rates for each of the respective models (i.e., the data used for model creation). (A–H) represent external predictions for wild-type, naturally occurring, and artificially mutated proteases as follows: (A) AMV, (B) Mo-MuLV, (C) HIV-2, (D) RSV, (E) EIAV, (F) FIV, (G) HTLV-1, and (H) HIV-1 proteases. (For [H] the data excluded were for 23 mutant HIV-1 proteases, with mutations associated with drug resistance.) The percentages in each panel indicate the fraction of predicted observations with a prediction error less than one log(k_cat/K_m) unit. RMSEP, the correlation coefficients (showing correlation between the observed versus the externally predicted rates), and their significances are shown on each panel.

Figure 3. Map of Physicochemical Properties of Retroviral Protease Substrates Based on the Regression Coefficients of Substrate and Substrate–Substrate Cross-Terms of the CRM

The figure summarizes the physicochemical requirements that a substrate should possess to be efficiently cleaved by a “swarm” of viral variants, notwithstanding that individual variants may favor or disfavor particular substrates. Spheres correspond to the five principal properties (z₁–z₅) for each amino acid [16]. Red spheres denote that a position favors an amino acid with a positive value of its z-scale for high cleavage rate. Blue spheres indicate that an amino acid with negative value of the z-scale is favored. For example, Ala, Asn, Asp, Pro, and Ser have positive values for z₃ and z₅ and are thus amino acids preferred for the P₄ position for affording a substrate with higher cleavage activity. Lines indicate the most important substrate–substrate cross-terms. Red lines denote that when the z-scales of both amino acids show large positive or large negative values, higher cleavage activity is favored. Blue lines indicate that higher cleavage rates are favored when both the z-scales have large values with different tokens. Black spheres denote the remaining substrate amino acid properties, which have a smaller effect on efficient substrate cleavage (see Materials and Methods).

Figure 4. The Ten Most Important Nonconserved Residues in Retroviral Protease for Substrate Recognition and the Most Important Cross-Dependences of Retroviral Protease and Substrate Amino Acids Identified by Use of the CRM

(A) The amino acids are shown in red on the 3-D structure of the HXB2 HIV-1 protease as a template. Because the retroviral proteases are homodimers, the modeling does not allow a distinction between cases where only one or both of the amino acids of the homoprotein should be assigned as important. (B) The retroviral protease amino acid residues most important for P₄ substrate position (shown in blue). (C) The retroviral protease amino acid residues most important for the P₃ (light blue), P₂ (yellow), P₁ (red), and P₁′ (magenta) substrate positions. (D) The retroviral protease amino acid residues most important for the P₃′ (white) and P₄′ (orange) substrate positions (see Materials and Methods for details).