Harvesting candidate genes responsible for serious adverse drug reactions from a chemical-protein interactome

Abstract

Identifying genetic factors responsible for serious adverse drug reaction (SADR) is of critical importance to personalized medicine. However, genome-wide association studies are hampered due to the lack of case-control samples, and the selection of candidate genes is limited by the lack of understanding of the underlying mechanisms of SADRs. We hypothesize that drugs causing the same type of SADR might share a common mechanism by targeting unexpectedly the same SADR-mediating protein. Hence we propose an approach of identifying the common SADR-targets through constructing and mining an in silico chemical-protein interactome (CPI), a matrix of binding strengths among 162 drug molecules known to cause at least one type of SADR and 845 proteins. Drugs sharing the same SADR outcome were also found to possess similarities in their CPI profiles towards this 845 protein set. This methodology identified the candidate gene of sulfonamide-induced toxic epidermal necrolysis (TEN): all nine sulfonamides that cause TEN were found to bind strongly to MHC I (Cw*4), whereas none of the 17 control drugs that do not cause TEN were found to bind to it. Through an insight into the CPI, we found the Y116S substitution of MHC I (B*5703) enhances the unexpected binding of abacavir to its antigen presentation groove, which explains why B*5701, not B*5703, is the risk allele of abacavir-induced hypersensitivity. In conclusion, SADR targets and the patient-specific off-targets could be identified through a systematic investigation of the CPI, generating important hypotheses for prospective experimental validation of the candidate genes.

Figure 1. Strategy of identifying candidate genes for SADR through data-mining against the CPI and text-mining.

Step (1) Drugs retrieved from adverse events reporting system of FDA were docked into the active sites of proteins. Shown here are the binding conformations of the four forms of fluoxetine (FXT1: the parent drug; FXT2: the major metabolite norfluoxetine; FXT3 and FXT4: nitrogen-atom positively charged of FXT1 and FXT2 respectively) to the bioactive sites of the five proteins. Step (2) Visualization of clustering results against the case matrix of binding affinity between 53 drugs (columns) and 845 proteins (rows). All case drugs were known to cause SJS/TEN. A Z-score greater or less than −1.2 were represented as white or black squares respectively. Step (3) A local CPI. Interactions of 13 case drugs in the sub-CPI1 and 17 control drugs with five proteins were shown here. For each of the proteins, the number of case drugs that interact with it with a Z-score<−1.2 (dark blue square) or with Z-score>−1.2 (black) were denoted as a and c, whereas the number of control drugs that interact with it with a Z-score<−1.2 (light blue) or with a Z-score>−1.2 (gray) were denoted as b and d, respectively. Step (4) & (5) Representative proteins highlighted from sub-CPI2 and sub-CPI3. The number of case or control drugs varied because of the missing values. Step (6) A local SJS/TEN-oriented GCCN. Yellow diamonds and white circles represent the core genes and extended genes respectively. One PubMed entry (yellow bolded) referring to HLA-C mainly deals with TEN. Another entry (red bolded) describes the relationship between HLA-C and LTA (red line). Genes that are annotated with GO term “inflammatory response” (in purple rectangle), “T cell activation” (blue) and in “immune response” (red) tend to be cited more specifically in this GCCN. Step (7) If a protein highlighted from the CPI share the GO terms enriched from GCCN, its symbol is presented in the corresponding color of the GO terms in step (6).

Figure 2. The interaction of sulfonamides to MHC I (Cw4).

(A) Interaction strength among case drug molecules of SJS/TEN and MHC I (Cw*4). Drug names followed by the numbers represent the derivatives of this drug. In sub-CPI 1, 15 molecules interact strongly with MHC I (Cw*4). After trimming procedure, 13 case drugs including 9 sulfonamides (listed in C) were found binding to MHC I (Cw*4). (B) The lowest energy conformations of four sulfonamides' binding to the antigen presentation groove of MHC I (Cw*4) through hydrogen bonds between the oxygen atoms in sulfuryls to the nitrogen atoms in the two arginine residues (R97 and R156). Residues in the two α-helixes also contribute to the binding. The four sulfonamides shown here are bumetanide, celecoxib, sulfadoxine and sulfamethoxazole. See Figure S2 for the binding conformations of other sulfonamides. (C) The nine sulfonamides have different structures, but they all bind to the two stretched arginine residues through their sulfuryls as the ‘root’. Molecular structures of the 17 control drugs which do not tend to interact with MHC I (Cw*4) were shown in Figure S3.

Figure 3. Models of abacavir's interactions with the binding site of HLA-B*5703 and HLA-B*5701.

(A) The abacavir could not bind to the antigen presentation groove of MHC I (B*5703, PDB ID: 2BVP). No non-covalent interaction could be formed with N114 in the presence of Y116 due to the steric hindrance (Video S1). (B) The steric hindrance of the binding disappeared when Y116S was introduced. The binding score between abacavir and MHC I (B*5701) was much lower than with (B*5703) because of I) better compatibility of geometry shape to the binding pocket; II) hydrogen bonds (yellow dashed) were formed with three essential AA residues (D114, S116 and T143) within the antigen presentation groove (Video S2).