Bioinformatics Advance Access originally published online on September 5, 2007
Bioinformatics 2007 23(20):2797-2799; doi:10.1093/bioinformatics/btm424
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Identification of potential HIV-1 targets of minocycline
1National Center for Genetic Engineering and Biotechnology, National Science and Technology Development Agency, 113 Thailand Science Park, Pahonyothin Road, Klong 1, Klongluang, Pathumtani 12120, Thailand and 2Department of Microbiology, University of Washington School of Medicine, Seattle, WA 98195, USA
*To whom correspondence should be addressed.
 |
ABSTRACT |
|---|
 TOP ABSTRACT 1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
Summary: Minocycline, a broad spectrum antibiotic, has been discovered to have inhibitory activity against HIV-1 in vitro, but the targets inhibited are unknown. We used a docking with dynamics protocol developed by us to predict the binding affinities of minocycline against seven active sites of five HIV-1 proteins to putatively identify the potential target(s) of minocycline. The results indicate that minocycline has the highest predicted binding affinity against HIV-1 integrase.
Contact: ram{at}compbio.washington.edu
|
1 INTRODUCTION |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
Minocycline is a broad spectrum antibiotic approved by the US Food and Drug Administration for treatment of bacterial infections. Minocycline has been discovered to substantially inhibit replication of HIV-1 in primary cultures of microglia (Si et al., 2004), macrophages and lymphocytes (Zink et al., 2005). Experimental evidence has shown that the p38 mitogen-activated protein kinase cascade is critical for HIV-1 replication (Darcissac et al., 2000), and that blockade of this signal transduction pathway may be a primary mechanism of minocycline for suppression of HIV-1 infection (Darcissac et al., 2000; Zink et al., 2005). Previous studies have pointed to evidence that minocycline is unlikely to have classic antiviral activity against HIV-1, such as reverse transcriptase and protease inhibition, since minocycline was not engineered to target a specific viral protein (Zink et al., 2005). However, there is no clear evidence to support this conclusion.
|
2 METHODS |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
We hypothesized that minocycline inhibits HIV-1 proteins that are essential for viral replication. To test this hypothesis, we used a computational docking with molecular dynamics (MD) simulation method developed by us (Jenwitheesuk and Samudrala, 2005) to dock minocycline into seven active sites of five HIV-1 proteins to identify the putative targets of minocycline. We downloaded the X-ray diffraction structures of HIV-1 capsid, gp41 HR1 (theoretical structure), integrase, protease, and reverse transcriptase from the Protein Data Bank (PDB). For each structure, the inhibitor was first removed from the active site and mutant side chains in these structures were then substituted to the wild-type side chain using SCWRL version 3.0. Protein structures from this step were used as a template for protein–inhibitor binding energy and cutoff calculations.
Protein–inhibitor docking calculations were carried out using AutoDock version 3.0.5 with a Lamarckian genetic algorithm as described in (Goodsell et al., 1996). We performed preliminary docking experiments to locate the potential binding sites of minocycline by generating a grid box that is large enough to cover the entire surface of each protein active site, as described by the experimentalists who solved the structures. The protein–minocycline complex derived from the preliminary docking procedure was consequently allowed to relax using MD simulation. The MD simulations are carried out using the X-PLOR software version 3.851. The topology and parameters for all inhibitors were obtained from the PRODRG server (Schuttelkopf and van Aalten, 2004). One hundred steps of energy minimization of the protein–inhibitor complex were initially performed, followed by MD simulation at 300 K. The trajectories at 0.1 ps were recorded and processed in a second docking step using similar docking parameters as used in the preliminary docking procedure.
We determined the binding affinity between a protein and its inhibitor by calculating the inhibitory constant (Ki) of the protein–inhibitor complex, which is correlated to the concentration at which 50% of the protein is inhibited (IC50). AutoDock generates three binding energy terms: intermolecular energy, internal energy of the ligand, and torsional free energy. The final docked energy was calculated from the sum of the intermolecular energy and the internal energy of the ligand. The free energy of binding was calculated from the sum of the intermolecular and the torsional free energies, and consequently converted into a Ki according to Hess's law. The lowest- energy solution was accepted as the calculated binding energy and its Ki value was used to define the binding affinity of the inhibitors. Further details of the MD simulation and docking protocols are given elsewhere (Jenwitheesuk and Samudrala, 2005).
To predict the inhibitory constant cutoff, we downloaded the X-ray diffraction structural complexes of 89 HIV-1 inhibitors (1 integrase inhibitor, 32 non-nucleoside reverse transcriptase inhibitors (NNRTIs) and 56 protease inhibitors) from the PDB. For each wild-type structure generated in the first step, we docked its inhibitor into the active site and predicted its Ki value. The mean predicted Ki of each group of HIV-1 proteins was used as the cutoff to identify potential target(s) of minocycline. Although there were several nucleoside reverse transcriptase inhibitors (NRTIs) available in the PDB, we did not predict the cutoff value for this target due to its different inhibitory mechanism; i.e. NRTI is a chain terminator of the newly forming viral DNA, whereas minocycline tends to occupy the NRTI active site and does not directly interact with the viral DNA. Therefore, their predicted Ki values are not comparable.
|
3 RESULTS |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
We previously showed that our docking with dynamics protocol, that integrates protein flexibility using molecular dynamics simulations with an inhibitor flexible-docking technique, has higher accuracy in predicting binding energies of HIV-1 protease–inhibitor complexes than that of the original inhibitor flexible-docking protocol. The calculated binding energies and the calculated Ki predicted by docking with dynamics were highly correlated with the experimental binding energies and the IC50, with correlations ranging from 0.35 when no protein dynamics was incorporated to 0.88 when dynamics was incorporated for a set of 25 HIV-1 protease–inhibitor complexes (Jenwitheesuk and Samudrala, 2003; Jenwitheesuk and Samudrala, 2005). This higher correlation enables us to apply this technique to identify the potential HIV-1 targets of minocycline.
In this study, our docking results indicate that minocycline has moderately high binding affinity against HIV-1 integrase and protease with a predicted Ki of 2.29 x 10–7 and 3.74 x 10–6, respectively (Table 1). Minocycline was ranked 38th among 56 protease inhibitors and its predicted Ki fell within 0.5 SD above the cutoff (of 7.76 x 10–6). In contrast, minocycline had no affinity with the HR1 groove of gp41, and bound to the other four active sites (C-terminal and N-terminal of capsid, and the active sites of NRTI and NNRTI) with low affinity, concurring with previous experimental studies that minocycline had low inhibitory activity against HIV-1 reverse transcriptase (IC50 = 1200 µM) (Wondrak et al., 1988).
|
We then inspected the binding modes of minocycline on the active sites of integrase and protease to determine the most likely target. Minocycline has four connecting hexagonal rings making its structure too rigid and too short to fit in the four subpockets of protease. Therefore, it is less likely that minocycline would have high inhibitory activity against HIV-1 protease, though the structural flexibility and the molecular size of minocycline may be increased by dimer formation. Minocycline, on the other hand, has similar shape and size to that of 1-(5-chloroindol-3-yl)-3-hydroxy-3-(2H-tetrazol-5-yl)-propenone (5ClTEP), an inhibitor designed to target HIV-1 integrase (Goldgur et al., 1999). We predicted that minocycline binds to the integrase active site with high binding affinity, and interacts with the Mg2+ ion in a similar manner to 5ClTEP (Fig. 1).
|
Previous docking studies on target identification of curcumin against HIV-1 integrase and protease using the same docking software showed a high correlation between the predicted binding energies and the experimental IC50s, indicating the accuracy of the software in predicting protein–inhibitor binding energies (Vajragupta et al., 2005). Curcumin was predicted to have high binding affinity against HIV-1 integrase (predicted binding energy was –8.79 kcal/mol) which was comparable to that of 5ClTEP against HIV-1 integrase (–8.25 kcal/mol). In contrast, the predicted binding energy of curcumin against protease (–9.77 kcal/mol) was much weaker than that of L745,524, a known protease inhibitor (–18.40 kcal/mol). By using the predicted binding energies of HIV-1 inhibitors as a reference, curcumin was predicted to target HIV-1 integrase. The experimental (IC50) results indicate that curcumin had an IC50 of 30–40 µM against HIV-1 integrase (Artico et al., 1998), which was more potent than that of curcumin against HIV-1 protease (100 µM) (Sui et al., 1993).
In the context of these experimental studies, our own study then suggests that minocycline is likely to bind and display inhibitory activity against HIV-1 integrase. We propose that, in addition to the inhibitory mechanism on p38 mitogen-activated protein kinase cascade, minocycline may suppress HIV-1 infection by inhibiting HIV-1 integrase. However, there are several other HIV-1 proteins whose 3D structures were not included in this study and they might be minocycline targets as well. Finally, there is also the consideration that minocycline indirectly inhibits replication by interfering with host mechanisms that HIV relies on, for example by reducing concentrations of inflammatory cytokines that facilitate HIV-1 replication.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
This work was partly supported by NIH grant GM068152, NSF grant DBI-0217241, a NSF CAREER award, a Searle Scholar Award, and the Puget Sound Partners in Global Health. The authors would like to thank Brady Bernard and members of the Samudrala group for valuable discussions and comments.
Conflict of Interest: none declared.
|
FOOTNOTES |
|---|
Associate Editor: Dmitrij Frishman
Received on June 5, 2007; revised on July 28, 2007; accepted on August 14, 2007
|
REFERENCES |
|---|
|
TOPABSTRACT1 INTRODUCTION2 METHODS3 RESULTSACKNOWLEDGEMENTSREFERENCES |
|---|
Artico M, et al. Geometrically and conformationally restrained cinnamoyl compounds as inhibitors of HIV-1 integrase: synthesis, biological evaluation, and molecular modeling. J. Med. Chem (1998) 41:3948–3960.[CrossRef][Web of Science][Medline]
Darcissac EC, et al. The synthetic immunomodulator murabutide controls human immunodeficiency virus type 1 replication at multiple levels in macrophages and dendritic cells. J. Virol (2000) 74:7794–7802.
Goldgur Y, et al. Structure of the HIV-1 integrase catalytic domain complexed with an inhibitor: a platform for antiviral drug design. Proc. Natl. Acad. Sci. USA (1999) 96:13040–13043.
Goodsell DS, et al. Automated docking of flexible ligands: applications of AutoDock. J. Mol. Recognit (1996) 9:1–5.[CrossRef][Web of Science][Medline]
Jenwitheesuk E, Samudrala R. Improved prediction of HIV-1 protease-inhibitor binding energies by molecular dynamics simulations. BMC. Struct. Biol (2003) 3:2.[CrossRef][Medline]
Jenwitheesuk E, Samudrala R. Prediction of HIV-1 protease inhibitor resistance using a protein-inhibitor flexible docking approach. Antivir. Ther (2005) 10:157–166.[Web of Science][Medline]
Schuttelkopf AW, van Aalten DM. PRODRG: a tool for high-throughput crystallography of protein-ligand complexes. Acta. Crystallogr. D Biol. Crystallogr (2004) 60:1355–1363.[CrossRef][Medline]
Si Q, et al. A novel action of minocycline: inhibition of human immunodeficiency virus type 1 infection in microglia. J. Neurovirol (2004) 10:284–292.[CrossRef][Web of Science][Medline]
Sui Z, et al. Inhibition of the HIV-1 and HIV-2 proteases by curcumin and curcumin boron complexes. Bioorg. Med. Chem (1993) 1:415–422.[CrossRef][Medline]
Vajragupta O, et al. Active site binding modes of curcumin in HIV-1 protease and integrase. Bioorg. Med. Chem. Lett (2005) 15:3364–3368.[CrossRef][Medline]
Wondrak EM, et al. Inhibition of HIV-1 RNA-dependent DNA polymerase and cellular DNA polymerases alpha, beta and gamma by phosphonoformic acid and other drugs. J. Antimicrob. Chemother (1988) 21:151–161.
Zink MC, et al. Neuroprotective and anti-human immunodeficiency virus activity of minocycline. J. Am. Med. Assoc (2005) 293:2003–2011.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||