Kamlendra Singh [1,2,3], Leonard C. Rogers [1], Jacqueline A. Flores [1], Anders Sönnerborg [3,4], Ujjwal Neogi [3]
Affiliates: [1] Department of Molecular Microbiology and Immunology, University of Missouri, Columbia, Missouri, USA. [2] Bond Life Sciences Center, University of Missouri, Columbia, Missouri, USA. [3]Division of Clinical Microbiology, Department of Laboratory Medicine, Karolinska Institutet, Huddinge, Stockholm, Sweden. [4] Department of Medicine Huddinge, Unit of Infectious Diseases, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden.

Polymorphisms (PMs) are known to affect therapy outcome. The PMs in the integrase gene of HIV-1 are not well studied. The objective was to determine if there are significant differences in the polymorphisms of HIV-1B and non-B subtypes. Moreover, the binding-affinities of first- and second-generation INSTIs in non-B HIV subtypes are not well-studied. The binding affinities of drugs to their targets are good predictors of their efficacy. Hence, the other objective was to determine in vitro and in silico subtype-specific differences in the binding affinities of first (raltegravir and elvitegravir) and second generation (dolutegravir) integrase strand transfer inhibitors (INSTIs).

HIV-1 integrase sequences (n=8,120) from patients with HIV-1 A1/A2 (n=483), HIV-1B (n=4,379), HIV-1C (n=1,155), CRF01_AE (n=1,581) and 02_AG (n=522) were analyzed for IN polymorphisms. Collectively these subtypes represent >90% of the global infections. Multiple sequence alignment was performed against HIV-1 HXB2. Variant-calling of each residue was performed using an in-house R script and a CIRCOS plot generated to visualise variant-calling. Subtype-specific consensus sequences were generated by an alignment. Naturally occurring polymorphisms were defined as mutations that were present in >50% of sequences. Using consensus sequences, the structures of HIV-1 IN for the representative subtypes (HIV-1B, HIV-1C, HIV-1A1/1A2, CRF02_AG and CRF01_AE) were generated. These structures were used to dock first- and second-generation INSTIs (raltegravir, RAL; elvitegravir, EVG; dolutegravir, DTG). Moreover, we cloned, expressed and purified IN proteins from HIV-1B, HIV-1C, 01_AE and 02_AG patient isolates and biochemically assessed the binding affinity of first- and second-generation INSTIs.

Among the non-B subtypes, 11 to 13 naturally occurring polymorphisms were observed in the integrase protein coding sequences. Among the mutations, V31I, T112V, T124A, T125A, V201I, L234I and S283G were observed in all of the non-B subtypes tested. In HIV-1A1/1A2, 01_AE and 02_AG, the K14R mutation was observed. Some subtype specific naturally occurring polymorphisms were also observed. The CIRCOS data revealed that HIV-1B had the least number of polymorphism residues (9 in HIV-1B versus 19 in non-B). The binding affinity of second-generation inhibitors to non-B INs was 1.8-fold (p<0.04) greater than to HIV-1B. On average, the binding affinity of DTG was better (2.1-fold, p<0.03) than those of the first-generation INSTIs in the strand-transfer reaction of non-B subtypes. However, the binding affinity of first-generation INSTIs was better (5.6 – fold, p<0.024) than DTG binding in the 3’-end processing reaction. The molecular modeling results showed that DTG’s average binding energy for non-B subtypes was -22.6 kcal/mol compared to -8.7 and 6.6 kcal/mol for RAL and EVG, respectively. For HIV-1B, the binding energies of DTG, RAL and EVG was comparable (-13.7, -11.2 and -10.4, respectively).

Our molecular modeling results, suggest that the second-generation INSTI DTG is a better inhibitor for non-B subtypes than the RAL and EVF, despite the presence of more polymorphisms in non-B. DTG is therefore an attractive alternative from a virological point of view for the use in countries where non-B subtypes dominate.