Acquired immunodeficiency syndrome (AIDS) was reported for the first time in 1981 in a small number of patients1,2 but has developed into a major pandemic. More than 35.3 million people were reported to be living with human immunodeficiency virus (HIV), the aetiological agent of AIDS, at the end of 2012. It is estimated that 2.3 million people become infected each year.3
The pol gene of HIV encodes three key viral enzymes for the replication of this virus that have been exploited for the development of chemotherapeutic agents. Two of these enzymes, HIV reverse transcriptase and HIV protease, received early attention in terms of the development of clinically useful inhibitors.4–9 Treatment of HIV/AIDS cases worldwide has involved the use of highly active antiretroviral therapy (HAART), which utilizes a combination of different HIV drug classes to target various viral enzymes in the HIV replication cycle, particularly reverse transcriptase and protease.10,11 However, HAART is often not well tolerated by patients because of some of its harsh side-effects. This regimen also requires a high degree of compliance, is expensive and is associated with multidrug resistance.12
Zidovudine (AZT) was the first anti-HIV/AIDS drug to be approved by the Food and Drug Administration (FDA), in 1987. By late 2007, three additional classes of anti-HIV/AIDS drugs were approved. These four drug classes were (1) nucleoside reverse transcriptase inhibitors; (2) non-nucleoside reverse transcriptase inhibitors; (3) protease inhibitors; and (4) fusion inhibitors.7–10,13 The various drugs under each of these categories have been successful in controlling HIV in combination therapy, including HAART, but their adverse side-effects and the emergence of resistant strains call for new therapies.14 In this respect, a drug of a different class, maraviroc (Celsentri®, ViiV Healthcare UK Ltd, Middlesex, UK), which is an entry inhibitor and acts as a CCR5 antagonist, has also been approved.10,15
Another key enzyme in the replication cycle of HIV that has been exploited for the discovery of anti-HIV drugs is HIV integrase.16 At present, there are three FDA-approved anti-HIV drugs in clinical use in which the mechanism of action is inhibition of the HIV integrase strand transfer step. The approved drugs are raltegravir (Isentress®, Merck Sharp & Dohme Ltd, White House Station, NJ, USA), approved in 2007,17,18 elvitegravir (Truvada®, Gilead Sciences Ltd, Great Abington, UK),19–21 which was approved in 2012, and, most recently, dolutegravir (Tivicay®, GlaxoSmithKline, Brentford, UK),22 which was approved in 2013. This review article will focus on these HIV integrase inhibitors that have been approved for clinical use, as well as the current state of development of new anti-HIV active integrase inhibitors discovered in our laboratory that have favourable profiles with respect to phase I and II metabolism and, very importantly, a dual mechanism of action against HIV and multidrug-resistant tuberculosis (MDR-TB).
Human immunodeficiency virus integrase
The viral enzyme and its catalysis of the integration process
Human immunodeficiency virus-1 integrase is a viral enzyme of Mr 32,000 that is encoded at the 3′-end of the pol gene of HIV-1 (Figure 1).16,23–26 Because the structure of the catalytic domain of integrase revealed that the enzyme belongs to a superfamily of polynucleotidyl transferases, including RNase H, transposases and polymerases, these enzymes are thought to share related mechanisms during catalysis.27,28 Two metal ions are apparently involved in the catalysis, which is a property that the aforementioned enzymes share. This mechanism was first proposed for the 3′–5′ exonuclease reaction of Escherichia coli DNA polymerase I29 and subsequently for other polynucleotide polymerases.30–33
The detailed mechanism of the incorporation or integration of HIV viral DNA into human chromosomal DNA is as follows. Following reverse transcription, the viral complementary DNA (cDNA) is first tailored in the cytoplasm prior to integration in the nucleus.25,26 Initiation of integration requires fully functional HIV integrase, viral cDNA and metal ion cofactors. Integrase recognizes specific sequences in the long terminal repeats of viral DNA. In the first step of integration, referred to as 3′-processing, there is site-specific endonuclease activity and two nucleotides are removed from each 3′-end of the double helical viral DNA to produce new 3′-hydroxyl ends (CAOH-3′) that are recessed thus by two nucleotides. After 3′-processing, the viral cDNA remains bound to integrase as a multimeric preintegration complex, which is transported through the nuclear envelope to the nucleus, where integrase catalyses the insertion of the processed viral cDNA ends into host chromosomal DNA. This insertion step is a transesterification reaction, which involves a staggered cleavage of 4–6 bp in the host DNA and the joining of processed CAOH-3′ viral DNA ends to the 5′-phosphate ends of the host DNA. The joining reaction produces a gapped intermediate, and the repair of this may be accomplished by host cell enzymes. However, a role here for viral integrase is also possible.16,25,26 In the initial 3′-processing reaction, integrase activates the phosphodiester bond towards cleavage, and in the DNA strand transfer reaction integrase plays the same role as well as positioning the 3′-OH end of the viral DNA for nucleophilic attack on the phosphodiester bond in the host DNA. Both the 3′-processing and strand transfer steps require divalent metal ion cofactors (Figure 2).
Recent work on a prototype foamy virus (PFV) integrase, which involved co-crystallization of the PFV integrase with approved HIV-1 integrase inhibitors, has provided more light on the details of the binding of these inhibitors to PFV integrase and therefore HIV integrase.36,37 The drug ligand chelation of the two divalent metals in the catalytic site can be discerned explicitly from the crystal structures, which is a key aspect of the inhibition of integrase.
Dinucleotide inhibitors of human immunodeficiency virus integrase
In our early quest for inhibitors of integrase, we designed small DNA nucleotide systems that not only possessed nuclease-stable internucleotide phosphate bonds, but which also carried critical structural features for recognition and inhibition of HIV integrase. The molecular design led to the discovery of novel, nuclease-resistant, non-natural dinucleotides with defined base sequences that inhibited both the 3′-processing (3′P) and strand transfer steps of the HIV-1 integrase mechanism of action.38
Inhibition studies with recombinant human immunodeficiency virus integrase
Compound 1 (Figure 3), which was designed to be nuclease stable and to mimic the CA terminus of the viral DNA, was found to have inhibitory activity against recombinant wild-type HIV integrase in reproducible assays (IC50 19 μM for 3′P and 25 μM for strand transfer). The inhibitory activity of compound 1 is much better than that of dideoxynucleoside monophosphates [e.g. (S,S)-2-iso-2,3-dideoxyadenosine-5′-monophosphate (> 300 μM 3′P and > 300 μM for strand transfer), 3′-azido-3′-deoxy-thymidine-5′-monophosphate (> 110 μM 3′P and 140 μM for strand transfer)],23,39 but the IC50 values are a little higher than that observed for the corresponding natural dinucleotide, pdApdC (6 μM 3′P and 3 μM for strand transfer).40 However, dinucleotides of natural origin are substrates for nucleases. The anti-HIV integrase activity of the non-natural, dinucleotide, compound 1 and its natural analogue suggested that the base sequence selectivity is consistent with the catalytic mechanism of 3′-processing in which endonuclease activity produces truncated viral DNA with a terminal CA dinucleotide. Molecular recognition by the integrase of the ultimate and penultimate bases at the 5′-end of the minus strand of non-cleaved viral DNA may result in stable complex formation before the strand transfer reaction.41 Thus, the inhibitory activity of compound 1 may reflect the affinity that HIV integrase has for this dinucleotide sequence.38 The discovery of compound 1 was the first example of an active-site directed, nuclease-stable, dinucleotide inhibitor of HIV-1 integrase. Recognition of the heteroaromatic base systems by integrase discerned from these studies and related studies16,35,36,38 provided the background for the scaffolds used in the discovery of the potent anti-HIV integrase inhibitors described in the section entitled ‘New generation antihuman immunodeficiency virus-1 integrase drugs’.
Approved human immunodeficiency virus-1 integrase inhibitors
Raltegravir was approved by the FDA in late 2007 as an integrase antiretroviral agent for the treatment of HIV-1 infections (Figure 4)18,42 and is a HIV-1 integrase strand transfer inhibitor. This drug, which is orally administered (twice daily), has been shown to be effective in combination therapy, with viral loads being reduced in both naive and experienced antiretroviral patients.43 For those treatment-experienced patients who were facing virological failure resulting from multidrug resistance issues, raltegravir has been co-administered with etravirine (Intelence®, Janssen-Cilag Ltd, High Wycombe, UK) (a non-nucleoside reverse transcriptase inhibitor) and darunavir (Prezista®, Janssen-Cilag Ltd, High Wycombe, UK) or ritonavir (Novir®, AbbVie Ltd, Maidenhead, UK) (protease inhibitors) as salvage therapy. This co-administered treatment was apparently able to achieve viral suppression similar to that observed in HIV treatment-naive patients.44–47 Thus, after its FDA approval, raltegravir contributed to providing a second option to patients who were left with almost no treatment alternatives following the failure of HAART owing to resistance problems. Clearance of raltegravir is through glucuronidation involving uridine 5′-diphospho-glucuronosyl-transferases (UGTs), primarily UGT1A1, and to a lesser extent by UGT1A9 and UGT1A3.48 However, polymorphism affecting the function of UGTs has been shown to increase the plasma levels of raltegravir among HIV-infected patients.49
Elvitegravir (see Figure 4) was designed structurally to possess a mono-keto-acid motif.19,20,50 Elvitegravir, another integrase strand transfer inhibitor, demonstrated an in vitro IC50 of 7 nM against integrase and an EC90 of 1.7 nM in cell culture-based assays. However, elvitegravir is rapidly metabolized by cytochrome P450 3A4 (CYP3A4), which requires boosting with cobicistat (Tybost®, Gilead Sciences Ltd, Great Abington, UK), a potent CYP3A4 inhibitor, in combination therapeutic applications.51 In addition, elvitegravir shares a moderate genetic barrier with raltegravir, involving integrase resistance. There is extensive cross-resistance between these two drugs, and this factor appears to be a significant drawback to clinical use. The mutations N155H, Q148H/R/K and G140A/C/S were selective for elvitegravir both in cell culture and in patients.19,52,53 Because these mutations are typical for raltegravir, elvitegravir may be unsuitable the treatment of most raltegravir-resistant viruses. The only major raltegravir-associated mutations not selective for elvitegravir are Y143C/R/H. However, subsequent studies have shown that viruses containing Y143C/R/H remain susceptible to elvitegravir.54 In addition to the raltegravir-associated resistance mutations, elvitegravir developed other resistance mutations, such as T66I, and, although the T66I mutation did not confer a high level of resistance to raltegravir,52 it gave a fold change (FC) of > 10 for elvitegravir. Furthermore, a T66R mutation showed a FC of > 10 for raltegravir but > 80 for elvitegravir.19,55 The T66I mutation is associated with a series of accessory mutations, which include F121Y, S153Y and R263K. However, the S153Y and R263K mutations have not been associated with raltegravir.56
Elvitegravir displayed moderate bioavailability in dogs and rats, with maximal plasma concentrations being achieved rapidly after administration. It was found to be well tolerated and efficacious in clinical trials.50 Elvitegravir is metabolized primarily by CYP3A4, with some metabolism through glucuronidation.57 The half-life can be significantly increased by co-administration with a CYP3A inhibitor. Thus, boosting of elvitegravir with either ritonavir or cobicistat (GS-9350, a non-therapeutic CYP3A inhibitor) allows for once-daily dosing.58,59
Dolutegravir (see Figure 4) is also a strand transfer integrase inhibitor exhibiting an IC50 value of 2.7 nM with recombinant integrase in enzyme assays. Dolutegravir is highly active against HIV replication in infected peripheral blood mononuclear cells (PBMCs) with an EC50 value of 0.5 nM. This compound also demonstrated efficacy against most viral clones that were resistant to raltegravir and elvitegravir, as well as against clinical isolates of HIV-1 and HIV-2.60,61 However, some viruses that carry E138K, G140S or Q148H mutations were not as susceptible to dolutegravir. Double mutants, containing combinations of E138K, G140S and R148H, had FCs of > 10 for dolutegravir, but this was favourable when compared with raltegravir, which produced FCs of > 330 and > 140, respectively. In vitro resistance selection studies identified, in order of appearance, mutant viruses carrying integrase mutations T124S/S153F, T124A/S153Y, L101I/T124A/S153F and S153Y. Although these mutations persisted throughout serial passaging, they did not confer a high-level resistance to dolutegravir.61 Despite an apparent genetic barrier for resistance selection, there is evidence from tissue culture and biochemical studies that an R263K mutation in integrase may confer some resistance to dolutegravir.62 The favourable pharmacokinetic profile, anti-HIV effect and potency of dolutegravir have been well documented.63,64 However, dolutegravir is an inhibitor of organic cation transporter 2, which may require averting its use with certain drug combinations. The principal route for the metabolism of dolutegravir is through UGT-catalysed glucuronidation.63,64
New-generation antihuman immunodeficiency virus-1 integrase drugs
Compound 2 and its antihuman immunodeficiency virus data
In order to explore the possibility of finding HIV-1 integrase inhibitors that would exhibit potent anti-HIV efficacy, low toxicity and a favourable profile with respect to human CYP and UGT isozymes/isoforms, we carried out structure–activity studies on derivatives of compounds discovered in our laboratory that were low nanomolar inhibitors of the strand transfer step of HIV-1 integrase.16,65–67 These drug discovery studies, combined with computational biology, molecular modelling and phase I and II metabolism enzymology and enzyme kinetics, led to novel and potent anti-HIV-active integrase inhibitors. An example is 4-[5-(2,6-difluorobenzyl)-1-(2-fluorobenzyl)-2-oxo-1,2-dihydropyridin-3-yl]-4-hydroxy-2-oxo-N-(2-oxopyrrolidin-1-yl)but-3-enamide (hereafter referred to as compound 2) (Figure 5).68,70,71 This integrase inhibitor is relatively easily produced in eight synthetic steps and with high overall yields from readily available starting materials.72 Its structure was confirmed by single-crystal X-ray,69 ultraviolet, high-resolution mass spectroscopy (HRMS) and 1H/13C nuclear magnetic resonance data, including gradient-selected correlation spectroscopy, heteronuclear single quantum coherence spectroscopy and heteronuclear multiple bond correlation spectroscopy correlations. The purity of the compound used in these studies was determined by HPLC and other methods to be > 99.6%.
The collective data from in vitro anti-HIV activity studies in PBMC cultures revealed that the compound had significant anti-HIV activity against a broad and diverse set of HIV-1 subtypes of major group M, as well as against HIV-2 and Simian immunodeficiency virus.68 Major group M of HIV-1 and its subtypes are responsible for most HIV infections.73 For group M subtypes A, B, C and F, which cover HIV infections in most of the world, the mean EC50 was 18.9 nM (SD ± 9.4) and the EC90 was 83.0 nM (SD ± 11.3 nM). Therapeutic indices for the isolates were > 5000, the highest being about 14 000. Cytotoxicity data (CC90 173 000 ± 3700 nM) gave strong evidence that the compound possessed low toxicity in human PBMC cultures. Control compounds used in these assays were integrase inhibitors, raltegravir and elvitegravir. AZT was also used as a positive control. Molecular docking studies (Figure 6), utilizing X-ray crystallographic data,36 suggest that the mechanism of anti-HIV activity was the inhibition of HIV integrase, and this was supported by the fact that the inhibitor was discovered in our laboratory from structure–activity studies of closely related prototypes and precursors, which showed IC50 data for integrase strand transfer inhibition at low nanomolar levels.65 Further validation for integrase inhibition came from integrase inhibition enzymology assays and from the observed T66 mutation in the integrase coding region of the HIV-1 genome, as well as from the cross-resistance data.
Resistant mutants and mutagenesis attributed to compound 2
In dose escalation studies employing MT-4 cells (human T-cell leukaemia cells) infected with HIV-1, the identification of HIV-1 isolates resistant to compound 2 was investigated.68 The selection of a single amino acid mutation from threonine to isoleucine at amino acid 66 (T66I) of integrase began to emerge following passage 4 with 600 nM of compound 2 and became a complete change following passage 9 (at 19.2 µM). Continued passaging with 20 µM of the compound (up to passage 15) did not result in the emergence of any additional mutations in integrase (Figure 7). The T66I mutation is in the catalytic core domain of the integrase coding region. In drug susceptibility studies in MT-4 cells (Figure 8) the fold change in the EC50 of compound 2 against resistant viruses with clinically relevant integrase mutations were found to be favourable.68 All integrase mutant viruses retained complete susceptibility to AZT.
Phase I and II isozyme/isoform metabolism profile of compound 2
A major focus of this investigation was determination of the profile of our integrase inhibitor towards key human CYP and UGT isozymes.77–79 The CYP isozymes used in this study are known to be involved in the clearance mechanisms of approximately 90% of known therapeutic drugs. The results showed compound 2 to be relatively stable in pooled human liver microsomes. Two key CYP-mediated metabolites of our drug were formed from mono-oxidation of the phenyl rings and their structures were confirmed by bioanalytical data, including HRMS. CYP isozyme kinetic data revealed that the IC50 values for inhibition of CYP isozymes (3A4, 2D6, 2C8, 2C9, 2C19) were all > 200 µM (Figure 9). In addition, our compound was not an activator of these CYP isozymes.68,70,71
Uridine 5′-diphospho-glucuronosyl-transferases belong to a superfamily of human phase II metabolizing isozymes that are involved in the glucuronidation and subsequent clearance through bile or urine of a significant number of drugs, including raltegravir.48 Thus, in this investigation, the UGT profile of compound 2 was considered to be of high significance, in terms of both HIV co-infection therapeutics80,81 and the issue of utilization of UGT-cleared integrase inhibitors for HIV/AIDS during fetal development and early infancy, given the low UGT activity during these phases.82 Glucuronidation studies of compound 2 and, for comparison, raltegravir were determined in pooled human liver microsomes verified to contain UGT 1A1, 1A4, 1A6, 1A9 and 2B7 (Figure 10). Compound 2 was not a substrate for these key UGTs in human liver microsomes or for specific cDNA-expressed UGT isozymes, UGT1A1 and UGT1A3.68 Furthermore, in the kinetic studies in human liver microsomes, there was no indication of the activation of UGT isozymes. In contrast, raltegravir was a good substrate for UGT, which is consistent with previously reported data.48 The two other FDA-approved integrase inhibitors, elvitegravir and dolutegravir, are also substrates for UGTs.22,61,64,83 In addition, no evidence for significant competitive inhibition of the key UGT isozymes, 1A1, 1A4, 1A6, 1A9 and 2B7, was found for compound 2 (IC50 > 300 µM).
Profile of compound 2 involving human permeability glycoprotein
In the transport of therapeutic drugs, two major superfamilies of transporters, ATP-binding cassette (ABC) and solute carrier, are among the most important. Among the ABC transporters which are relevant in the case of compound 2, the best recognized transporters are permeability glycoproteins (Pgps), which play a significant role in the distribution and elimination of many therapeutic drugs.84 Pgps act as ‘gatekeepers’ for the action of CYP-450 isozymes. Madin–Darby canine kidney II cells expressing human Pgp (expressing ABC transporter MDR1) were utilized for determination of substrate activity associated with human Pgp-mediated transport. Efflux ratio data were calculated as follows: PappB – A/PappA – B. An efflux ratio of > 2 indicates that there is transport by Pgp. The efflux ratio of compound 2 (10 µM) was 8.27 ± 2.78, indicating significant Pgp-mediated transport. Studies with ketoconazole (Nizoral®, Janssen-Cilag Ltd, High Wycombe, UK), a known Pgp inhibitor, suggested that Pgp is likely to be a major transporter for compound 2.76
Protein-binding studies on compound 2
Protein binding of compound 2 involving human plasma was also studied and revealed that protein binding is high (98.3%).76 The range of protein binding of compound 2 is in the comparable to that of the recently approved drug, dolutegravir (> 98.7%).85 Competitive binding experiments show that human serum albumin is the major binding protein for compound 2, which appears to bind to drug site subdomains IB and IIB on human serum albumin, as indicated from kinetic data for binding in the presence of indomethacin and ibuprofen, respectively.76,86
Preclinical studies compound 2 in animals
Pharmacokinetic studies in Sprague–Dawley rats (three dose groups of 30 mg/kg, 100 mg/kg and 300 mg/kg p.o.) revealed rapid absorption, a reasonably long plasma half-life and moderate bioavailability.76 Neither absorption nor first-pass metabolism appeared to be saturable up to doses of 300 mg/kg. Drug exposure increased with increasing drug concentration, indicative of appropriate dose-dependent correlation. The apparent volume of distribution was high and ranged from 12.4 l/kg to 17.7 l/kg for the p.o. dose groups. The plasma mean residence time ranged from 4.1 hours to 7.3 hours with a clearance rate of 2675 ml/h/kg. The compound exhibited significant extravascular tissue distribution and the bilirubin and urobilinogen levels in serum and urine were not affected. There were no adverse treatment-related findings from the clinical pathology or from the clinical chemistry parameters. There were no changes in erythropoietic, white blood cell or platelet parameters. The preclinical studies also revealed that the no observable adverse effect level and maximum tolerated dose were > 500 mg/kg/day, providing further evidence of low potential for toxicity for compound 2 with oral administration.
Efficacy of our antihuman immunodeficiency virus integrase inhibitors against multidrug-resistant tuberculosis
World problem on human immunodeficiency virus and tuberculosis co-infection
About one-third of the world’s population (about 2 billion people) are infected with the causative agent of tuberculosis (TB), Mycobacterium tuberculosis, a Gram-positive bacterium. Of those with latent TB, but who are not infected with HIV, some 5–10% become sick or are infectious at some period during their lifetime. People with HIV are much more likely to develop TB. Estimates from the World Health Organization suggest that there are over 9 million new cases of TB each year, and about 2 million people die from TB each year.3,80,87,88 Further complicating this huge global problem is the emergence of MDR-TB. The current pipeline for approved and investigational drugs against MDR-TB is very narrow.89–91 In addition, co-infection of HIV with MDR-TB is a very serious and deadly liaison.80,92–94
Antimultidrug-resistant tuberculosis activity of our integrase inhibitors
Perhaps the most significant ramification of our work on anti-HIV integrase inhibitors is the efficacy of some of these compounds against MDR-TB, i.e. they possess a dual mechanism of action against both HIV and MDR-TB.95 Evidence for this comes from the agar drug susceptibility assay data (Clinical and Laboratory Standards Institute Guidelines) from the Southern Research Institute (Birmingham, AL, USA) TB Programme obtained through a contractual research agreement with the University of Georgia (Athens, GA, USA). This agar method is viewed as the ‘gold’ standard for studies to measure minimum inhibitory concentrations (MIC, i.e. the lowest concentration at which there is no visible bacterial growth). Streptomycin was used as one of the standards. The most active compound exhibited an MIC of < 0.4 μg/ml against MDR-TB. A drug cytotoxicity MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] proliferation assay for 7 days using uninfected macrophages at the MIC and fourfold greater concentration confirmed that the compound was relatively non-toxic (percentage viability > 90% after 7 days).
The mechanism of action of these integrase inhibitors appears to be unique in that it seems to be unlike that of any of the known TB drugs, i.e. they are not gyrase or topoisomerase inhibitors, they do not target the 30S or 23S ribosomal subunits, they are not cell wall or cell membrane inhibitors, they are not ATPase inhibitors, they are not peptidyl transferase inhibitors and they do not act at the same RNA polymerase subunit as rifampin. Our inhibitors appear to bind to the magnesium ion within the catalytic site as well as to key amino acid residues in the β′-subunit of the TB RNA polymerase holoenzyme, prior to the binding of the latter to promoter DNA. The outcome of this inhibitor binding is the failure of TB RNA polymerase to even begin to initiate bacterial transcription. Interestingly, this proposed mechanism has an unmistakeable resemblance to the binding of these novel compounds to HIV integrase.
Although progress has been made in the field of anti-HIV active-integrase inhibitors and their use in combination therapeutics against HIV and AIDS, some very significant challenges still remain to be surmounted with next-generation integrase inhibitors for HIV/AIDS. These include appropriate profiles with respect to phase I and II metabolism, minimizing adverse drug–drug interactions, cross-resistance issues and, very importantly, HIV co-infections involving drug-resistant infections. We have discovered conceptually new anti-HIV integrase inhibitors with low toxicity and favourable metabolism profiles with respect to CYP and UGT, and these inhibitors possess the uniquely inherent therapeutic property of having a dual mechanism of action against both HIV and MDR-TB infections. This research breakthrough on new drugs has the potential to have major global health impact.