Avoiding Drug Resistance in HIV Reverse Transcriptase

Abstract

HIV reverse transcriptase (RT) is an enzyme that plays a major role in the replication cycle of HIV and has been a key target of anti-HIV drug development efforts. Because of the high genetic diversity of the virus, mutations in RT can impart resistance to various RT inhibitors. As the prevalence of drug resistance mutations is on the rise, it is necessary to design strategies that will lead to drugs less susceptible to resistance. Here we provide an in-depth review of HIV reverse transcriptase, current RT inhibitors, novel RT inhibitors, and mechanisms of drug resistance. We also present novel strategies that can be useful to overcome RT’s ability to escape therapies through drug resistance. While resistance may not be completely avoidable, designing drugs based on the strategies and principles discussed in this review could decrease the prevalence of drug resistance.

Graphical Abstract

1. INTRODUCTION

Human immunodeficiency virus (HIV) is the cause of acquired immunodeficiency syndrome (AIDS), which is responsible for the deaths of over 38 million people.¹ Because there is no cure for HIV, patients are subjected to life-long therapy. The standard treatment for HIV is known as highly active antiretroviral therapy (HAART), which is capable of suppressing viral replication below the limit of detection and reducing viral transmission.^2–4 HAART normally consists of three different drugs to keep HIV below detectable levels. There are six main classes of anti-HIV drugs: nucleoside reverse transcriptase inhibitors, non-nucleoside reverse transcriptase inhibitors, protease inhibitors, integrase inhibitors, fusion inhibitors, and entry inhibitors. In these classes, there are more than 27 different approved drugs.⁵ Fourteen of the approved drugs target the key viral enzyme reverse transcriptase (RT). The standard of care for most treatment naïve patients is two nucleoside reverse transcriptase inhibitors and one non-nucleoside reverse transcriptase inhibitor or integrase inhibitor.^2,4

1.1. HIV Replication Cycle

HIV infection is initiated by the envelope protein binding to a host cell CD4 receptor, followed by binding to CXCR4 or CCR5 coreceptors. In turn, these interactions result in structural changes of the envelope protein which lead to fusion of the viral membrane with the host cell membrane. The HIV virion contains two copies of single-stranded positive sense RNA (+ssRNA) genome, about 50 copies of RT, other viral enzymes such as integrase and protease and nucleocapsid chaperones.^6,7 RT has a polymerase domain able to use RNA or DNA as a template and make a DNA copy; it also has a ribonuclease H (RNase H) domain capable of degrading RNA in an RNA:DNA duplex. Using these enzymatic activities RT creates a linear double-stranded (dsDNA) product from the genomic + ssRNA. The viral dsDNA (vDNA) product is then used as substrate for the viral enzyme integrase for integration into the host genome.⁶ The integrated viral DNA or provirus serves as the template for viral genomic and viral mRNA, which is made by the host RNA polymerase and exported from the nucleus to the cytoplasm. The translation of viral RNAs (vRNA) produce the Gag polyprotein precursor, Pr55^gag, which comprises the matrix, capsid, and nucleocapsid structural proteins, the Gag-Pol polyprotein precursor, Pr160^gag–pol, with Pol comprising HIV protease, RT, and integrase enzymes. Envelope glycoprotein (Env) and regulatory and accessory viral proteins (Vif, Vpr, Vpu, Nef) are also translated. The Gag, Gag-Pol precursors, and Env are trafficked to the plasma membrane, where the vRNA is encapsidated and this structure buds off from infected cells and the virion contents mature through the action of HIV protease.^6,7 Part of the maturation process involves the conversion of the p66/p66 RT homodimer to the p66/p51 heterodimer.⁸ The p66 subunit is 560 amino acids long and consists of two active enzymatic domains: polymerase, and RNase H. The p51 subunit contains the first 440 amino acids of p66 and it lacks an RNase H domain (Figure 1).

Figure 1.

Structure of HIV-1 RT in complex with dsDNA and dTTP substrates and binding locations of FDA-approved drugs targeting RT. HIV-1 RT (PDB 1RTD)⁹ is comprised of two subunits, p66 (multicolored cartoon) and p51 (gray cartoon). The enzymatically active p66 subunit has both RNA-dependent and DNA-dependent DNA polymerization activities and contains four subdomains: fingers (blue), palm (red), thumb (green), and connection (yellow), as well as an RNase H domain (purple). Template DNA is shown in light brown, primer DNA is shown in dark brown, and dTTP is shown as gray sticks at the polymerase active site (black dashed circle). Two Mg²⁺ ions in the polymerase active site are shown as spheres. The polymerase active site is also the location where NRTIs bind. The NNRTI binding pocket (NNIBP) is located in the p66 palm subdomain, near the base of the p66 thumb subdomain and adjacent to the polymerase active site (black dashed rectangle).

1.2. Structural Components of the Polymerase Active Site of HIV Reverse Transcriptase

The p66 polymerase domain of RT is divided into four subdomains: the fingers (residues: 1−84, 119−154), palm (residues: 85−118, 155−241), thumb (residues: 242−313), and connection (residues: 314−426)^10–12 (Figure 1). The p51 subunit lacks an RNase H domain; although its structural domains have the same sequence as those in p66, they are organized differently,^10,11 and as a result p51 has no enzymatic activity and simply provides structural support for p66.¹³

The polymerase and RNase H active sites of RT are situated at opposing parts of the nucleic acid binding cleft, which accommodates ~17 base-pairs (bp) of DNA/DNA and ~18 bp of RNA/DNA.¹⁴ Near the polymerase active site DNA/DNA substrates assume an A-form,¹¹ whereas RNA/DNA substrates assume a similar H-form.¹⁵ The nucleic acid interacts with two α-helices of the p66 thumb subdomain (αH and αI) (Figure 2). The polymerase active site has three carboxylates in the palm subdomain of p66: D110, D185, and D186, which bind two Mg²⁺ in vivo (Mn²⁺ can be supported in vitro) and are essential for polymerization. D185 and D186 are part of a highly conserved YXDD motif in retroviral RTs; in HIV RT X is Met (Figure 2).^16–19 In addition to the thumb subdomain, another important RT structural element is the primer grip (β12 and β13 hairpin), which also helps position the 3′-OH of the primer at the polymerase catalytic site.^11,20,21 Important polymerase active site residues include R72²² and K65 that bind the β- and γ-phosphates of the incoming dNTP, and Y115, which contributes to the binding of the deoxyribose ring of the incoming dNTP and has been termed the “stericgate”, responsible for discriminating between deoxy- and ribonucleoside triphosphates (Figure 2).^9,23–25

Figure 2.

Structure of the HIV-1 RT polymerase active site. dTTP (gray sticks) is bound at the polymerase active site (PDB 1RTD).⁹ Conserved active site residues D110, V111, D185, and D186 chelate two Mg²⁺ ions (light-green spheres), which also bind the alpha, beta, and gamma phosphates of the dTTP. Residue Y115 is located at the bottom of the active site. dTTP sits above Y115 and is additionally stabilized through hydrogen bond interactions between the alpha and beta phosphates and R72 in the β4 fingers, and the gamma phosphate and K65 in the β3−β4 fingers loop. The conserved YMDD loop is between β9 and β10, and the primer grip is located between β12 and β13 in the p66 palm subdomain. αH and αI of the p66 thumb subdomain interact with the nucleic acid substrate to help position it correctly in the polymerase active site.

1.3. Enzymology of RT

The mechanism of DNA synthesis begins with RT bound to a template/primer. The 3′-end of the primer is positioned at the polymerization site. RT incorporates dNMPs based on complementarity to the template strand.^9,11,26 The incoming dNTP initially binds at the nucleotide site (N- site or pretranslocation site), which is followed by a nucleophilic attack to form a phosphodiester bond and release of the pyrophosphate product. Following primer extension, the 3′-end of the elongated DNA moves to the post-translocation or P-site, thus vacating the N-site for the next incoming dNTP.^27,28 Insights into the molecular details of these translocation events have been provided by crystal structures of strategically designed covalent RT-nucleic acid intermediates.²⁷

The RNase H domain of RT cleaves RNA in RNA/DNA duplexes. The RNase H has a highly conserved DEDD motif (D443, E478, D498, D549), which binds the two divalent cations required for enzymatic activity.^9,29–34

1.4. HIV Reverse Transcription

While there are several viral and cellular factors that aid in reverse transcription, here we only discuss RT. Like any DNA polymerase, RT requires both a primer and a template. RT initiates DNA synthesis from a host tRNA primer: tRNA_Lys3^35,36 that is reported to be packaged in virions at a 4-fold excess with respect to the viral genomes.³⁷ The tRNA_Lys3 hybridizes to an 18-nucleotide complementary region, known as the primer binding site (PBS), which is located near the 5′ end of the viral genome.³⁸ Polymerization begins from hybridized tRNA_Lys3, which synthesizes a minus-strand DNA in complex with the plus-strand viral RNA. In vitro data have shown the addition of the first five to six nucleotides after the primer is very slow but then DNA synthesis speeds up.^39,40 As RT synthesizes DNA, it creates an RNA/DNA hybrid, which is concomitantly degraded by the RNase H enzymatic domain, leaving a single minus-strand DNA (Figure 3). The sequences of the 5′ and 3′ ends of the viral RNA genome are repeat regions.^41–44 The RNA-free minus-strand DNA can hybridize to the repeat region of the 3′ end of either of the vRNA genomes, which is known as the first jump or minus-strand transfer and is one of the reasons for viral recombination that may affect drug resistance, as discussed below.^45–48 After the minus-strand DNA hybridizes to the 3′ end of the RNA genome, synthesis of the DNA continues as the RNase H function of RT continues to degrade RNA. Additionally, there is a purine-rich sequence near the 3′ end of the vRNA, known as the 3′-polypurine tract (3′-PPT) that is resistant to cleavage by RNase H, which serves as primer for plus-strand DNA synthesis (Figure 3). There is also a central PPT (cPPT), which is known to increase plus-strand DNA synthesis but unlike the 3′-PPT is not essential.^49–57 The 3′-PPT primed DNA synthesis continues and copies the first 18 ribonucleotides of the tRNA.^51,58,59 RNase H removes the tRNA primer from the tRNA/DNA complex, leaving a ribo-A on the 3′ end of the minus-strand DNA; this sets the stage for the second jump, also known as plus-strand transfer. Once the tRNA is removed, it exposes the 3′ end of a single plus-strand DNA with PBS, at the same time, the minus strand DNA is also exposed (Figure 3).⁶⁰ The exposed minus- and plus-strands anneal to each other and create double-stranded DNA (dsDNA) that has the same sequence on both ends also known as long terminal repeats (LTRs).^{26,28,36,61,62}

Figure 3.

Reverse transcription of HIV-1. An HIV virion has two copies of the ssRNA genome. Once a host tRNA_Lys3 binds to the primer binding site (PBS), DNA synthesis starts forming the minus-strand strong-stop DNA. During DNA synthesis, RNase H digests RNA, exposing the minus-strand DNA that hybridizes to the R region of the 3′-end of either viral genome copy. The elongation of the minus strand continues until the polypurine tract (PPT), which is RNase H resistant and serves as a primer for plus-strand DNA synthesis. After plus- and minus-strand DNA syntheses are completed, the RNase H removes the tRNA primer and releases the PBS, thus facilitating the “second jump”. Strand displacement activity of RT to the PBS and PPT ends and/or DNA repair and ligation lead to dsDNA or a circular intermediate with two long terminal repeats (LTRs). Figure adapted with permission from ref 63. Copyright 2020 Elsevier. License number: 4943870547699.

2. ROLE OF RT IN HIV GENETIC DIVERSITY AND DRUG RESISTANCE

HIV patients are infected by a highly heterogeneous pool of genetic variants known as quasispecies.^64–67 The high genetic variability of HIV is thought to be the result of several factors, including a high mutation rate, high rate of viral replication that results in high viral load, the lack of 3′ to 5′ exonuclease repair function of its replicative enzyme, the RT,⁶⁸ the introduction of random mutations by the host RNA polymerase II during transcription of the nascent viral RNA from the integrated proviral DNA,⁶⁹ and by the process of viral recombination. Some of these processes are discussed below.

2.1. The Fidelity and Mutations

RT copies the ~9700 nucleotide long viral RNA genome into dsDNA through RNA- and DNA-dependent polymerization steps, totaling ~19 400 nucleotide incorporation events.^5,70,71 The in vitro misincorporation rate for RT has been reported to be ~1 × 10⁻⁴, and it is affected by factors such as type of divalent metal used.^72,73 The in vivo mutation rate of the viral replication is ~2 × 10⁻⁵.^55,74–76 The high mutation rate of HIV is critical for its survival during drug therapies. The virus can obtain drug resistance mutations, which is one of the major challenges for effective HAART.^77–86 The resistance mutations emerge as a result of poor patient adherence, which permits viral replication and mutation, increasing the chances that drug resistance mutations will occur and spread throughout the patient due to a selective growth advantage in spite of ongoing treatment. Some drug-resistant viruses are also capable of being transmitted, and consequently these strains can increase in prevalence as described below.^{77,84,87–93}

2.2. Viral Recombination

Viral recombination is a process whereby RT engages into template switching between the two viral RNA templates (~3−4 times per replication cycle) that are copackaged into the same virion.⁹⁴ Recombination of HIV increases the genetic diversity of the viral population affecting resistance to antivirals.^{47,48,95–101} The rapid turnover of 10⁸ to 10⁹ virions per day, combined with the high error rate of RT and the high rates of viral recombination, contribute significantly to the generation of a very large number of mutated HIV genomes _{per day.}^{28,45,75,102–106}

3. RT-TARGETING ANTIVIRALS

Currently there are two different types of approved RT inhibitors: nucleos(t)ide reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) (Figure 4). NRTIs approved and in use are azidothymidine (AZT), lamivudine (3TC), emtricitabine (FTC), didanosine (ddI), abacavir (ABC), and tenofovir (TFV). Two NRTIs have been discontinued due to toxic side effects: stavudine (d4T) and 2′−3-dideoxycytidine (ddC).²⁶ As mentioned above, two of these NRTIs are typically used in the current HAART regimen. In the US, treatment regimen includes a combination of the NRTIs TFV and FTC.⁴ NRTIs are also typically used in pre-exposure prophylaxis therapy (PrEP).^107–109 NNRTIs that have been approved for the treatment of HIV-1 infection: nevirapine (NVP), efavirenz (EFV), etravirine (ETR), rilpivirine (RPV), and doravirine (DOR). One NNRTI that has been discontinued is delavirdine (DLV). NNRTIs are also used in the HAART regimen, but less frequently.^5,26,110

Figure 4.

Chemical structures of approved NRTIs and NNRTIs.

3.1. Current Nucleoside RT Inhibitors

NRTIs are analogues of the natural substrate deoxynucleotide triphosphates (dNTPs). All current approved HIV NRTIs lack a 3′-OH and act as chain terminators when incorporated into the primer strand of the viral DNA and thus prevent the addition of an incoming nucleotide. NRTIs inhibit reverse transcription by mimicking dNTPs and can be incorporated by RT. Because NRTIs lack a 3′-OH, once they are incorporated into the primer, chain termination occurs, ultimately preventing further DNA synthesis. The first anti-HIV drug approved was AZT, a thymidine analogue that has an azido group in the 3′-OH position.¹¹¹ In contrast, ddC and ddI have no azido group but lack a 3′-OH,^112,113 whereas d4T has a double bond between the 2′ and 3′ positions of the ribose and ABC has a cyclopentene ring also without a 3′-OH group.^114,115 TFV is an acyclic nucleotide phosphonate¹¹⁶ delivered as a diester prodrug, earlier as a tenofovir disoproxil fumarate (TDF) and more recently (in 2015) as a tenofovir alafenamide (TAF).¹¹⁷ TAF is superior to TDF in terms of efficacy, stability, and safety.^118–122

Three factors play a major role in NRTI development: (1) efficient activation to the triphosphate form, (2) stability of the NRTI, and (3) NRTIs barrier to resistance. For NRTIs to be effective, they must be phosphorylated by host cell kinases, which convert them to their active triphosphate form.¹²³ Once activated, they must compete with the natural substrate for incorporation by RT into the elongating primer. All NRTIs must be phosphorylated three times, with the exception of TFV. Because TFV has a phosphonate, it needs an addition of only a β- and γ-phosphate, skipping the addition of the α-phosphate, the first and rate-limiting phosphorylation event for most NRTIs. However, one exception is AZT, where the rate limiting step is the addition of the second and third phosphates; due to these reasons, the cellular levels of AZT-TP are relatively low.¹²⁴ After an NRTI is converted to its triphosphate form, RT incorporates it as an NRTI monophosphate at the 3′ end of the primer end and a pyrophosphate product is released.

3.2. Non-nucleoside RT Inhibitors

Unlike NRTIs, NNRTIs do not compete for the natural substrate; instead, they block RT as allosteric inhibitors. NNRTIs bind in a hydrophobic pocket ~10 Å away from the polymerase active site known as the NNRTI binding pocket (NNIBP). The NNIBP consists of the following residues: L100, K101, K103, V106, T107, V108, V179, Y181, Y188, V189, G190, F227, W229, L234, and Y318 of p66 and E138 of p51 (Figure 5).^10,12

Figure 5.

Structure of the HIV-1 RT NNIBP. Select residues of the NNIBP of HIV-1 RT (PDB 1VRT)¹²⁵ are shown as sticks, with NVP shown as transparent orange sticks. These include L100, K101, K103, V106, V179, Y181, Y188, F227, W229, L234, and P236 of the p66 palm subdomain (pink sticks), Y318 of the p66 thumb subdomain (green sticks), and E138 of the p51 subunit (gray sticks). Additional residues in the NNIBP include T107, V108, V189, and G190, which are not shown for clarity.

Crystallographic studies have shown that the NNIBP does not exist in the absence of NNRTIs.^{5,10,125–131} For an NNRTI to bind in the NNIBP, structural changes must occur, particularly in residues Y181 and Y188 at the “floor” and W229 at the “ceiling” of NNIBP. The latter is formed in part by the “primer grip”, a structural element of RT that positions the nucleic acid substrate at the polymerase active site. These changes perturb the “primer grip”, and as a result, NNRTI binding affects the alignment of the primer terminus in the active site (Figure 6), which ultimately inhibits chemical synthesis.^{125,132–136} Of note, because the primer directly interacts with the base of the thumb (Figure 6), NNRTI binding and primer grip repositioning also affect the position of the thumb, which ends up in an overextended conformation (pale green vs green in Figure 6).^26,137

Figure 6.

Structural changes occur in HIV-1 RT upon NNRTI binding, which affect the position of nucleic acid binding. HIV-1 RT covalently cross-linked to dsDNA (p66 palm is shown in red cartoon, p66 thumb is shown in green cartoon, primer is shown as dark-gray sticks; template has been removed for clarity) with NVP (purple sticks) bound at the NNRTI binding pocket and AZT-MP (dark gray sticks) at the 3′-end of the DNA primer strand in the P-site (PDB 3V81).¹³⁸ The crystal structure of HIV-1 RT (p66 palm and p66 thumb in pink and light-green cartoon, respectively) covalently cross-linked to dsDNA (primer in light gray sticks; template not shown for clarity) with AZT-MP at the P-site and AZT-TP in the N-site (PDB 3V4I)¹³⁸ was aligned to structure 3V81 by p66 residues 100−210. Binding of NVP at the NNIBP causes Y181 and Y188 in the NNIBP to flip 180° and also a shift in the position of W229 (pink vs red sticks before and after NVP binding). These changes lead to a cascade of structural rearrangements that include repositioning of the conserved YMDD loop at the polymerase active site, the primer grip, and the p66 thumb subdomain, which are important for nucleic acid binding (movements noted with black dashed arrows). The nucleic acid repositioning is noted by the black dashed arrow between the two AZT-MP molecules at the P-site.

NNRTIs can potently inhibit the replication of WT HIV-1; however, because the RT enzymatic activity does not directly involve the NNIBP, this pocket can be relatively easily mutated to impart NNRTI resistance (low barrier to resistance) without significantly affecting the ability of the virus to efficiently replicate, known as viral fitness. The early NNRTIs NVP, EFV, and DLV bind at the NNIBP in a rigid “butterfly-like conformation”.^{126,128,139,140} A major drawback of early NNRTIs is that, typically, a single NNIBP resistance mutation (such as L100I, K103N, V106A, E138K, Y181C, Y188L, G190A, or H221Y) can result in significant loss of their efficacy.¹²⁶ In contrast, the newer NNRTIs, ETR and RPV, were developed to have higher conformational flexibility (Figure 7) and inhibit the common NNRTI resistance mutants.^129,141,142 Furthermore, this flexibility trait helped maintain the efficacy of ETR and RPV in the presence of K103N and Y181C.¹⁴¹ RPV was designed to efficiently bind in the NNIBP and inhibit effectively using the following interactions: (1) hydrophobic interactions within the NNIBP, (2) hydrogen bonding with the main chain of the K101 residue,¹⁴³ and (3) water-mediated hydrogen bonds.^5,144 Structural evidence demonstrated that RPV could adapt multiple conformations in the presence of various NNRTI resistance mutations. This trait is commonly known as “wiggling” and “jiggling”.¹⁴³ As of 2018, DOR has been approved and was also shown to be effective against common NNRTI resistance mutations, including K103N, Y181C, and G190A.^{110,145–147}

Figure 7.

Second-generation NNRTIs have more intrinsic flexibility that enables binding and inhibition in the presence of several NNRTI resistance mutations. (A) Chemical structure of second-generation NNRTI rilpivirine (RPV), with labeled torsion angles demonstrating the flexibility of this molecule. (B) The terms “wiggling” and “jiggling” are used to describe how newer NNRTIs position themselves to adopt multiple conformations and adjust for the changing side chains in mutated NNIBPs. (C) Although structurally similar, diarylpyrimidine (DAPY) analogues R120393 (a), TMC120-R147681 (dapivirine) (b), TMC125-R165335 (etravirine) (c), and R185545 (d) are able to bind at the NNIBP assuming diverse conformations, thus avoiding mutated residues in the NNIBP. (B,C) Reproduced unmodified from ref 141. Copyright 2004 American Chemical Society.

3.3. Novel RT Inhibitors

4′-Ethynyl-2-fluoro-2′-deoxyadenosine (EFdA, MK-8591, or islatravir) is an exceptionally potent and promising long-acting RT inhibitor in phase III clinical trials.^148–150 EFdA has the following three structural attributes that contribute to the stability, long-acting potential, and potency: (1) 2-fluoro, (2) 4′-ethynyl (4′-E), and (3) 3′-OH.^151–153 The 2-fluoro group contributes to the remarkable stability of EFdA by making EFdA resistant to adenosine deaminase, a key metabolizing enzyme of adenosine and adenosine-based antivirals.^149,154,155 Unlike any other approved HIV NRTI, EFdA retains a 3′-OH; it is also efficiently phosphorylated to EFdA-MP by the cellular deoxycytidine kinase.^149,154 The 4′-E group stabilizes EFdA into a conserved hydrophobic pocket at the polymerization active site.¹⁵⁶ EFdA inhibits reverse transcription primarily by blocking translocation of RT on the template-primer. For this reason, EFdA is known as a nucleoside reverse transcriptase translocation inhibitor (NRTTI). EFdA can block translocation through two mechanisms: (1) immediate chain termination (ICT) and (2) delayed chain termination (DCT) (Figure 8).¹⁵⁰ During ICT, RT is arrested immediately following incorporation of EFdA-monophosphate (EFdA-MP), and the 3′-end of the terminated primer remains bound at the pretranslocation or N-site. DCT occurs when EFdA-MP is incorporated into the 3′-end of the primer and translocates to the primer- or P-site, allowing the addition of a single nucleotide before chain termination. Crystallographic studies of various EFdA-inhibition intermediates of RT provide glimpses on how EFdA affects RT translocation through two mechanisms.¹⁵⁶ Because of the high potency and stability of EFdA, it is being tested as a once-yearly therapy instead of the current once daily dosing regimens (see below).^157,158

Figure 8.

Immediate and delayed chain termination mechanisms of EFdA inhibition. Initial binding of EFdA-TP in the N-site (N, orange box), prior to incorporation (left). Once EFdA-MP is incorporated into the 3′-end of the primer, it may cause immediate chain termination (ICT) (middle) or delayed chain termination (DCT) (right). During DCT, after EFdA-MP incorporation at the 3′-end of the primer and translocation, which moves the EFdA-MP at the P-site (P, blue box), RT adds a single additional nucleotide before inhibition of further DNA synthesis. Figure based on ref 156.

Elsulfavirine or VM-1500 is an another potent NNRTI (EC₅₀=1.2 nM) that is active against a broad range of NNRTI resistant viruses (VM-1500 is the name of the prodrug and VM-1500A is the active compound).^146,159 Elsulfavirine has been approved for HIV-1 treatment in Russia.¹⁶⁰

4. MECHANISMS OF DRUG RESISTANCE

4.1. NRTI Resistance

Under selective drug pressure, drug resistant mutants can emerge and gain a competitive advantage over NRTI-treated WT virus, thus becoming the dominant quasispecies. In principle, mutations should affect drug binding or function but should still allow viral replication. HIV-1 resistance to NRTIs usually involves two general mechanisms: NRTI discrimination and NRTI excision.

4.1.1. (a) NRTI Discrimination.

NRTI discrimination occurs when RT reduces the NRTI incorporation efficiency compared to the natural dNTP. This tends to appear along the dNTP binding pocket from the β3-β4 fingers and the conserved YMDD loop (Figure 2). K65R emerges in response to treatments with TFV, ABC, ddI, and d4T.^161–163 Structural studies have revealed that K65R becomes resistant to NRTIs, because R65 together with R72 forms a platform that helps discriminate TFV-DP from dATP (Figure 9).¹⁶⁴ The resistance mutation Y115F emerged in response to treatment with ABC; as the Y115 residue supports dNTP binding, mutation to phenylalanine maintains most structural interactions but increases the chance of NRTI discrimination.¹⁶⁵ Together, K65R and Y115F further decrease TFV susceptibility.¹⁶⁶ The L74V mutation has been shown to confer ddI resistance; the L74 residue supports base-pair formation with the incoming dNTP, and the L74V mutation weakens the template interaction with the dNTP substrate or reduces the template support.^167,168 M184V/I RT mutations cause resistance to FTC and 3TC by directly interfering with the binding of the triphosphate form of the drugs at the polymerase active site.^9,169–171 The mutated residues at the 184 position, isoleucine or valine, are β-branched amino acids, and sterically clash with the L-oxathiolane ring of 3TC-TP/FTC-TP.^9,169–171 The Q151M mutation causes resistance to dideoxynucloside analogues, especially in combination of other associated mutations that are known as the Q151M complex (Q151M/A62V/V75I/F116Y). These mutations emerged in response to combination therapy of AZT with dideoxynucleoside drugs.^172,173 The Q151M complex causes resistance to many NRTIs, but not to 3TC/FTC and TFV.

Figure 9.

Mutation K65R in HIV-1 RT imparts resistance to NRTIs through discrimination. Structural studies revealed that K65R becomes resistant to NRTIs because 65R together with R72 form a platform (light-blue sticks and spheres) that helps discriminate dATP (A, magenta sticks and spheres, PDB 3JYT) from TFV-DP (B, gold sticks and spheres, PDB 3JSM) at the polymerase active site (red cartoon and pink sticks and spheres). Figure based on ref 164.

4.1.2. (b) NRTI Excision.

NRTI excision was initially discovered as a resistance mechanism during treatment with AZT.¹⁷⁴ Hence, the NRTI resistance mutations were termed as thymidine analogue resistance mutations (TAMs) (Table 1). These mutant RTs are capable of incorporating AZT as efficiently as WT RT. However, RTs containing TAMs can unblock NRTI-terminated primers allowing DNA synthesis to resume.^175,176 For an NRTI to be excised, it must be located in the N-site; as the metal binding machinery is required for excision, RT cannot excise from the P-site. AZT-MP can be excised more frequently and better than other NRTIs because of the long azido group that interferes with the binding an incoming dNTP, thus keeping AZT-MP in the N-site more frequently.²⁷ Also, excision occurs because two main mutants K70R and T215Y create an ATP-binding pocket next to the dNTP binding cleft, where the aromatic ring of the tyrosine (T215Y) stacks with the ATP base, and the arginine (K70R) forms a polar interaction with the 3′-OH and the α-phosphate of ATP.^177,178 The ATP binding pocket allows binding of ATP and positioning its β- and γ-phosphates in a manner similar to a pyrophosphate group. In turn, this enabling chelation of the two Mg²⁺ ions, and ultimately promoting pyrophosphorolysis (Figure 10), so RT can remove the AZT-MP from the end of the DNA primer.¹⁷⁹ The product of the excision reaction when AZT-terminated primers are unblocked by ATP is an AZTppppA tetraphosphate and an unblocked primer.^177,178 The most common mutations responsible for excision are K70R and T215Y, but other mutations are known to enhance excision and the virus’ replication capabilities.^177,180,181 D67N and K219Q are found to be associated with K70R,^{177,182–184} while M41L and L210W are associated with T215Y.^185,186 The deletion of codon 67 also increases the excision efficiency of AZT.^187,188 Interestingly, a T69 insertion into the β3- and β4-loops of RT was found to help excise other NRTIs.^189–192

Table 1.

NRTI Resistance Mutations

NRTI resistance mutation(s)	mechanism of resistance	drug(s) affected by resistance mutation(s)
K65R	discrimination	TFV, ABC, ddI, and d4T
Y115F	discrimination	ABC
L74V	discrimination	ddI
M184V/I	discrimination	FTC and 3TC
Q151M	discrimination	ddI, AZT, d4T, and ABC
Q151M/A62V/V75I/F116Y	discrimination	ddI, AZT, d4T, and ABC
Type 1: M41L/L210W/T215Y	excision	AZT and d4T
Type 2: D67N/K70R/T215Y/K219Q	excision	AZT and d4T

Figure 10.

Structural basis for excision-based AZT resistance: interactions of excision product with TAMs RT. HIV-1 RT is shown as cartoon (colored as in Figure 1) bound to dsDNA (template in light brown, primer in dark brown) with key residues shown as pink sticks. The product of AZT-MP excision, AZTppppA, is shown as cyan sticks at the N-site. AZT-resistance mutations (M41L, D67N, K70R, T215Y, K219Q; thymidine-associated mutations or TAMs; shown as magenta sticks) facilitate excision by forming the ATP substrate binding site (PDB 3KLE). Figure based on ref 177.

4.2. NNRTI Resistance

As mentioned in section 3.2, NNIBP residues are not directly involved in the polymerization mechanism of RT, and as such they can mutate without a significant cost in replication fitness (Table 2). Thus, the barrier to NNRTI resistance is low and NNRTI resistance mutations appear relatively fast. There are three main mechanisms by which NNIBP mutations impart NNRTI resistance: (1) loss of key hydrophobic interactions (V106A, V179D, Y181C, Y188L, F227C/L), (2) steric hindrance (L100I or G190A/S), and (3) pocket entrance mutations (K101E/P, K103N, and E138K (of p51)).

Table 2.

Fold-Changes of NNRTI Resistance compared to WT EC₅₀s

resistance mutation	NVP	ETR	EFV	RPV	DOR	ref
L100I	7.3	1.3	20.3	0.9	<3	218–220
K101E	NA	2.7	3.8	2.4	4.5	110,219
K103N	81	0.9	32.5	0.9	1	218–220
V106A	>42	0.5	2.0	0.6	∼45	110,219,220
V179D	4.6	1.9	2.7	1.7	NA	219
Y181C	317	4.0	2.1	2.7	1.1	218–220
Y188L	>42	1.1	42.6	2.8	>100	110,219,220
G190A	128	1.1	8.1	1.1	<5	218–220
F227C	17	3.6	5.1	4.1	>10	110,220,221
E138K (of p51)	1.1	2.6	2.0	2.8	1	219,220

The amino acid residues that are located in the hydrophobic core of the NNIBP are Y181, Y188, and F227 (Figure 5).^10,125–127 Mutations in these residues can cause high resistance through loss of hydrophobic interactions with the incoming NNRTIs.^{131,193–196} Resistance mutation Y181C is primarily selected with treatment to NVP (>50-fold reduced susceptibility).^197,198 Because of their conformational flexibility, the newer generation NNRTIs, ETR and RPV, are able to maintain antiviral activity better than the first-generation NVP, enduring antiviral potency losses of only 5-fold and 4-fold, respectively.^198–201 Resistance mutation Y188L confers high-level resistance to both NVP and EFV.^{197,202–204} In contrast, RPV has only a 5-fold reduced activity against Y188L.^205–207 F227C is frequently selected during treatment with DOR or with ETR and RPV.²⁰¹ It is also selected during NVP or EFV-based therapies, although in these cases it usually appears in combination with V106A, conferring high-levels of resistance.^203,204,208

L100I and G190A mutations cause steric hindrance at the center of the NNIBP, leading to resistance by changing the shape of the pocket. For instance, L100I mutates from a γ-branched amino acid to a β-branched amino acid.¹⁹⁶ L100I has high-level resistance to EFV, ETR, and RPV. G190A introduces a bulge, mutating from a small residue, glycine, to an alanine.²⁰⁹ Other mutations can also occur at G190 such as a serine, glutamic acid, or a glutamine, which also introduce a bulge and distort the NNIBP. DOR remains effective at inhibiting G190A and has mildly reduced efficacy against the other mutants at this position.²¹⁰ RPV and ETR are effective at inhibiting G190A/S (unless it is in combination with other NNRTI-resistance mutations) and have reduced susceptibility to G190E/Q.²⁰⁶

Pocket entrance mutations interfere with an NNRTI entering the NNIBP. Mutant K103N does not specifically interact with an NNRTI but rather restricts access to the NNIBP.^135,211 However, ETR, RPV, and DOR remain effective against K103N, while NVP and EFV do not.^203,206,212 Other pocket entrance mutations K101E/P and E138K (of p51) seem to act similarly to K103N, with their side chains protruding out of the pocket and preventing NNRTI entry. K101P causes resistance to all current NNRTIs, except DOR, unless in combination with other mutations.^199,213 K101E causes resistance to RPV and NVP.^206,214 In addition, residue 138 can mutate to glycine, glutamine, arginine, or most commonly to lysine (E138K) that imparts RPV resistance.²⁰¹ E138K is clinically relevant because it was selected together with M184V/I in patients that experienced virological failure while being treated with RPV/TDF/FTC. Virological assays showed that E138K/Q/R mutations can compensate for M184I in both enzymatic fitness and viral replication capacity assays.^215–217

5. STRATEGIES TO AVOID DRUG RESISTANCE

HIV RT has been a key target for anti-HIV drugs since 1986.¹¹¹ HAART has been successful in suppressing HIV. However, due to the high genetic diversity of HIV, prolonged treatments lead to the emergence of drug resistance mutations. Thus, there is a need for novel antivirals with excellent affinity, strong potency, high barrier to resistance, and able to overcome the known mechanisms of resistance using a combination of strategies discussed below.

5.1. Using the “Substrate Envelope” Strategy to Design Inhibitors with Improved Resistance Profile

The substrate envelope is based on the premise that inhibitors that fit into the substrate envelope have a better resistance profile, as they stay protected from binding site mutations that cannot reach into the substrate envelope without significantly affecting the binding of the native substrate.^222–224 Hence, inhibitors that protrude from the substrate envelope are more likely to elicit resistance.

The substrate envelope in the case of RT comprises the binding site of dNTP (N-site, Figure 2) and nucleic acid. The resistance profiles of TFV, 3TC, and FTC, which are major NRTIs used in today’s HAART, are relevant to this principle. Specifically, unlike 3TC/FTC that protrude outside the substrate envelope and sterically clash with V/I184 leading to >100-fold resistance, the acyclic TFV is fully active against M184V/I because it binds within the substrate envelope (Figure 11).^{9,169–171,225} However, RT eventually develops resistance to TFV, by acquiring the K65R mutation, which does not cause a steric clash (Figure 9). Instead, K65R decreases the rate of TFV incorporation and causes mild resistance (3−5-fold).¹⁶⁴ Importantly, this mutation also affects the viral fitness by decreasing the efficiency of dNMP incorporation.^226,227 In that respect, TFV is an example of a successful design that avoids the steric clash mechanism of drug resistance and leads to a mutation (K65R) that forces escape through a costly decrease in viral replication.

Figure 11.

TFV avoids resistance relationship between NRTI resistance and the substrate envelope. The natural substrate dATP (gray sticks; PDB 5TXL)²²⁸ defines the substrate envelope. FTC-TP (blue sticks; PDB 6UIR)²²⁹ protrudes from the substrate envelope, thus causing a steric clash with M184V (red dashed lines and explosion cartoon), resulting in >100-fold resistance. TFV-DP (light-cyan sticks; PDB 3JSM)¹⁶⁴ binds within the substrate envelope and thus remains active against M184V/I. Structural alignments were based on RT residues 50−200.

5.2. Targeting Evolutionarily Conserved Regions in RT

Targeting conserved regions in a highly genetically diverse virus is essential to drug design because such regions are much less likely to mutate. Evolutionary conservation is expected to have functional significance that is indispensable for virus survival. Ideally, a drug should interact with key residues that are involved in the mechanism of reverse transcription. Thus, mutations of such residues should result in loss of function for RT and decreased viral fitness.

Because NRTIs mimic natural dNTPs and bind at the dNTP or N-site they may interact, at least in part, with highly conserved residues in the polymerase active site. For example, K65 and R72 are highly conserved residues that interact with the γ- and β-phosphates of the incoming dNTP or NRTItriphosphates.^9,24 As mentioned above, the K65R mutation may allow narrow escape of TFV with a modest drug resistance, however, even the slight rearrangement of the 65/72 residues observed in the structure of K65R RT (Figure 9) results in a fitness cost that is direct result of mutation of the highly conserved function of precise alignment of the triphosphate component of incoming nucleotide or NRTI triphosphates.¹⁶⁴

Another example of targeting conserved residues is the case of novel NRTTI EFdA. This compound has a 4′-E that binds into a conserved hydrophobic pocket defined by these residues: A114, Y115, F160, and M184 and the aliphatic part of D185¹⁵⁶ (Figure 12). After incorporation of EFdA-MP into the viral DNA, the 4′-E of the elongated primer is stabilized by the interactions with residues of the conserved pocket leading to inhibition of the DNA synthesis (Figure 12). Three (A114, Y115, F160) out of the four residues that define the 4′-E binding pocket are highly conserved, leading to a high barrier to resistance for EFdA.

Figure 12.

Interactions of EFdA-TP at the N-site of the HIV-1 RT polymerase active site. HIV-1 RT (PDB 5J2M) is colored as in Figures 1 and 2. The α-, β-, and γ-phosphates of EFdA-TP (gold sticks) participate in chelation of the Mg²⁺ ion along with conserved active site residues D110, the main chain of V111, and D185 (interactions shown as blue dashed lines). The 4′-ethynyl (4′-E) group of EFdA-TP sits in a hydrophobic pocket at the base of the polymerase active site composed of residues A114, Y115, F160, M184, and D185 (interactions shown as black dashed lines). The 3′-OH participates in hydrogen bond interactions with its β-phosphate (interaction shown as green dashed line). For clarity, we do not show additional interactions between the 3′-OH and the main chain of Y115 and water-mediated hydrogen bond interactions with A114 and F116. The 2-fluoro (2-F) group of EFdA-TP participates in water-mediated interactions with Y115 and the main chain nitrogen of G152 (interactions shown as red dashed lines).

Another example of an RT-targeting antiviral that interacts with conserved residues to block translocation of RT is foscarnet or phosphonoformic acid, which is an analogue of the pyrophosphate product of DNA synthesis (Figure 13). Foscarnet has not been approved for the treatment of HIV infection because of its side effects. However, it is an RT inhibitor that stabilizes the pretranslocation state of RT by interacting with conserved residue K65. Of note, RT escapes foscarnet inhibition through the E89G mutation that affects primer translocation, but also viral fitness.²³⁰

Figure 13.

Chemical structures of pyrophosphate (PPi) and foscarnet (phosphonoformic acid, PFA).

Finally, another strategy of targeting evolutionarily conserved regions is the design of compounds that affect protein−protein interactions (PPI). Such compounds have been useful in the design of anti-HIV inhibitors that block interactions of LEDGF75 with HIV integrase²³¹ as well gp120 interactions with CD4.^232,233 Moreover, inhibitors can also interfere with dimerization of the HIV-1 protease dimerization,^234,235 interactions between integrase subunits leading to aggregation,^236,237 and misdirection of viral maturation, as well as antivirals that target the interface of the p66/p51 RT heterodimer. There are hydrophobic interactions at p66/p51 interface involving a cluster of tryptophans (W398, W401, W402, W406, W410, and W414), which are fundamental for RT dimerization as demonstrated by mutational studies.^238,239 An example of this type of inhibitors that target RT dimerization is MAS0.^240,241 Also, as the NNIBP comprises not only residues of the p66 subunit but also of p51 (E138), some NNRTIs, such as EFV, ETR, and dapivirine, have been reported to affect RT dimerization²⁴² and intracellular processing of Gag and Gag-Pol polyproteins.²⁴³ Therefore, targeting various conserved regions of RT has been a main strategy of drug design for RT inhibitors.

5.3. Targeting the Conserved Divalent Metal Binding Sites

RT uses divalent metals for both its polymerase and RNase H enzymatic functions. The polymerase site YMDD loop that includes the three highly conserved catalytic carboxylates (D110, D185, and D186) are critical for positioning the two Mg²⁺ ions that are required for DNA synthesis (Figure 2).^16–19 Disrupting their divalent metal binding function would be expected to be an efficient strategy for designing antivirals with high barrier to resistance. As many cellular polymerases have related divalent metal binding motifs, it would be important that such designed inhibitors avoid binding to related motifs of cellular polymerases or kinases and other phosphotransferases that often use divalent metal-binding mechanism for their catalytic functions. A similar approach has been successful in the design of compounds that bind the active site metals of the integrase enzyme of HIV.^244,245 In such cases, the inhibitor would be composed of one part that binds the two divalent metals bound to the conserved catalytic carboxylates, and another part of the inhibitor would impart specificity for the viral active site, rather than the cellular two-metal binding enzymes.

Another example of this principle is the design of antivirals that bind the RNase H domain of RT. Various RNase H inhibitors that have been previously explored in the context of HIV-1 have chelating groups, which interact with the two divalent metals of the RNase H domain and inhibit the RNase H activity of RT (Figure 14).^246–256

Figure 14.

Interactions of RNase H inhibitors (RNHIs) at the RNase H active site of HIV-1 RT. (A) A 2-hydroxyisoquinoline-1,3-dione RNHI, YLC2–155 (orange sticks), chelates two Mn²⁺ ions (light-blue spheres) that are bound by conserved RNase H active site residues D443, E478, E498, and D549. YLC2–155 also forms hydrogen bond interactions with Q500 and H539 (PDB 5UV5). YLC2–155 can also bind to the RNase H active site in a different conformation, with the furan ring pointing toward H539 (not shown). Based on ref 252. (B) A hydroxypyridonecarboxylic acid RNHI, 10y (yellow sticks), chelates two Mg²⁺ ions (light-green spheres) that are bound by the conserved RNase H active site residues. 10y also forms hydrogen bond interactions with H539 and K540 (PDB 5J1E). Based on ref 256.

5.4. Designing Drugs That Interact with Main Chain Residues of RT

Designing antivirals that target main chain polyamide backbone atoms is based on the premise that regardless of the residue mutation, the inhibitor will still maintain its interaction with the main chain of the residue. This approach can be particularly valuable when targeting a virus that has a high mutation rate like HIV. Some examples of such interactions of drugs that target RT are discussed below.

For example, the 2-fluoro substituent of EFdA-TP at the N-site is an important contributor to the success of EFdA because it interacts with the G152 main chain amide (Figure 12A).¹⁵⁶ Maintaining these main chain residue interactions helps minimize the potential for drug resistance to occur. Similarly, the inhibitory effect of EFdA during delayed chain termination (DCT) is based on a negative interaction of the main chain of Y183 and the 4′-E of EFdA-MP after its incorporation into the primer strand and elongation by a single nucleotide (DNA_EFdA-MP^P_•dTMP^N) (Figure 15).¹⁵⁶ This disruption of the interactions of RT with the nucleic acid prevents further DNA synthesis resulting in the DCT mechanism of inhibition. Any mutation at the 183 position would not enable escape and drug resistance, as the steric interactions would be present regardless the type of amino acid present at this position.

Figure 15.

RT interactions during the delayed chain termination (DCT) inhibition mechanism. Superposition of two RT structures with EFdA variants before and after translocation. Transparent light-brown sticks show the position of the DNA primer with ddG at the post-translocation P-site and EFdA-TP at the pretranslocation N-site (PDB 5J2M). After EFdA-MP is incorporated into the 3′-end of the primer and further elongated by a single nucleotide (RT/DNA_EFdA-MP^P_{• dTMP}^N), EFdA-MP is located at the P-site (PDB 5J2N). In this position, the 4′-ethynyl (4′-E) sterically clashes with the main chain carbonyl of Y183, and a large shift occurs only at the P-site of this primer to alleviate this negative interaction, as shown by the distance between the 4′ carbon atoms and the dihedral angles of the two structures. Based on ref 156.

There are several examples of NNRTIs that interact with main chain residues of RT. EFV forms hydrogen bonds with the main chain C=O and N−H groups of residue K101 (Figure 16A).¹³⁰ Similarly, NVP forms water-mediated hydrogen bonds with the main chain amino and carboxyl groups of the K101 residue (Figure 16B).¹²⁵ Also, hydrogen bonds between a nitrogen from RPV and the main chain C=O and N−H of K101, and water-mediated hydrogen bonds with between an additional nitrogen and the main chain of E138 in p51 contribute to binding of RPV (Figure 16C).^143,257 Finally, DOR interacts with V106 and L100 and the methyl triazolone stacks with P236 and interacts with the main chain atoms of K103 (Figure 16D).¹⁴⁵

Figure 16.

Several NNRTIs interact with residues in the NNIBP through main chain interactions. (A) EFV (orange sticks, PDB 1FK9)¹³⁰ forms hydrogen bond interactions with the main chain C=O and N−H groups of K101. (B) NVP (green sticks, PDB 1VRT)¹²⁵ interacts with the main chain C=O and N−H groups of K101 through water-mediated hydrogen bonds. (C) RPV (yellow sticks, PDB 2ZD1)¹⁴³ forms hydrogen bonds with the C=O and N−H groups on the main chain of K101 and interacts with E138 in p51 through water-mediated hydrogen bonds with the main chain C=O group. (D) DOR (blue sticks, PDB 4NCG)¹⁴⁵ has hydrophobic interactions with L100 and V106 and forms hydrogen bond interactions with the C=O and N−H groups on the main chain of K103.

Therefore, interactions between hydrogen bond donors or acceptors from a potential RT inhibitor could significantly contribute to its binding potential and decrease its susceptibility to drug resistance.

5.5. Designing Covalent Inhibitors of RT

Covalent inhibitor drugs are typically small molecules that bind irreversibly their target. They have functional groups that can react and bind covalently to an amino acid side chain of the target protein, leading to its inactivation.^258–261 They include drugs that have been used for many decades; aspirin and penicillin are some of the early and well-known members of this class of drugs. Among the advantages of covalent inhibitors is that their potency is usually higher than that of the noncovalent version of the inhibitor, as by definition, the k_off of an irreversible covalent inhibitor is zero. Also, covalent inhibitors offer long duration of action if the target protein has slow turnover.²⁶² Covalent inhibitors bind a specific region of the target protein and have a reactive functional group known as “warhead” that covalently links to a specific amino acid, leading to a loss of function. Designing covalent inhibitors begins with identifying a conserved nucleophilic amino acid (a cysteine, lysine, serine, threonine, tyrosine, methionine, glutamate, or aspartate), at a site where the inhibitor binding can affect the functionality of the enzyme.^263–266 The residue selection is important because it should be near the active site or at a pocket that is important for the function of the target protein.²⁵⁹ Targeting conserved residues is also important, as it should result in compounds that have relatively high barrier to resistance. Furthermore, the binding site needs to be unique or rare, to minimize the chances of nonspecific binding and off-target effects. There are now computational tools that can be used for the design of covalent inhibitors, including AutoDock.^267,268

There have been increasing efforts toward the discovery of covalent inhibitors that target HIV. A recent study focused on covalent RT inhibitors (CRTIs) (Figure 17) that specifically target RT that carries the Y181C mutation that renders several NNRTIs ineffective.^269,270 The compounds used in this study were based on an originally noncovalent inhibitor that was modified by adding an electrophilic warhead on the carbonchloride bond (Figure 17); this change enabled covalent binding at the C181 residue after the specific initial binding at the NNIBP (Figure 17). Such compounds could be useful for patients that have failed some NNRTI-based therapies that lead to the Y181C mutation. Hence, inhibitors that have the potential to specifically inactivate wild-type or mutant RTs can be a useful strategy in combating HIV RT drug resistance. In this specific case, a compound that specifically targets the sulfydryl group of Y181C NNRTI-resistant viruses could be used in combination with NNRTIs that target the WT version of HIV.

Figure 17.

Covalent inhibitors targeting NNRTI-resistant HIV-1 RT. (A) Chemical structure of compound 1, which served as the basis for the design of covalent RT inhibitors (CRTIs). A chloride bound to a carbon (circled in green) in this compound was replaced with an electrophilic warhead to create various CRTIs that were designed to inhibit Y181C HIV-1 RT. (B) Crystal structure of compound 3 covalently bound to 181C in the NNIBP of Y181C HIV-1 RT (PDB 5VQX).²⁶⁹ Compound 3 (beige sticks) forms additional interactions with the main chain amide of K103.

5.6. Using Conformational Flexibility to Design Inhibitors That Assume Multiple Conformations and Successfully Target Ever-Changing Binding Pockets

Drug molecules that are rigid may be susceptible to multiple drug resistance mutations. This is because they may not adjust efficiently to minor perturbations introduced by mutations that otherwise do not affect the target protein’s function. This may be highly relevant for HIV, a virus with high genetic diversity that can acquire resistance mutations.

It has been proposed that first-generation NNRTIs NVP and DLV have a relatively low barrier to resistance due to their rigid structure that causes loss of interactions and potency through even a single resistance mutation.¹⁹⁸ In contrast, the newer generation NNRTI RPV was developed through a strategic approach to have more intrinsic flexibility that enables binding and inhibition even in the presence of NVP and DLV resistance mutations. This approach, known as “wiggling” and “jiggling”, explains the success of newer generation of drugs that orient and position themselves to adopt multiple conformations and adjust for the changing side chains of the NNIBP (Figure 7B).¹⁴³ Thus, the intrinsic flexibility of these compounds (Figure 7A) allows them to retain their potency in the presence of several common NNRTI resistance mutations. RPV and related diarylpyrimidine (DAPY) analogues are able to bind at the NNIBP using multiple conformations (Figure 7C) and thus angle away from changes at residue I100 and move toward residue 103.¹⁴³ Hence, designing more flexible inhibitors should result to drugs that adapt multiple conformations and have a better chance to overcome drug resistance.

5.7. Design Inhibitors That Allosterically Interact at Nonactive Site Binding Pockets

Allosteric inhibitors bind to a nonactive site of the enzyme, which alters the confirmation of the active site and ultimately disrupts the substrate bound to the enzyme. An example of HIV allosteric inhibitors are NNRTIs; however, NNRTIs have a relatively low barrier to resistance. The NNRTIs bind in the NNIBP, where this pocket is not directly essential for the function of RT and resistance mutations are more likely to occur. However, work from the Arnold and Tachedjian laboratories have used a strategy known as fragment screening by X-ray crystallography to identify novel potential sites in RT that could provide an allosteric site with a higher barrier to resistance (Figure 18).^271–274

Figure 18.

Fragment screening to identify potential binding sites for novel antivirals targeting HIV-1 RT. Fragment screening can be used to identify new binding sites for allosteric inhibitors. Shown are fragment binding sites on HIV-1 RT (orange spheres). Reproduced with permission from ref 273. Copyright 2013 American Chemical Society.

5.8. Designing Combinations of Drugs That Act Synergistically

On several occasions, drugs that belong to the same family, such as NRTIs, are not cross-resistant to each other’s escape mutations. In fact, there are reported cases where resistance to one drug renders RT hypersensitive to the other. Combinations of such drugs exhibit synergy and can be powerful tools in antiviral therapies.

For example, 3TC (or FTC)/AZT (combivir) was one of the relatively successful early combination therapies, because not only FTC (and 3TC) maintains potency against the AZT-resistant TAMs, but also FTC (and 3TC)-resistant M184V/I RT becomes hypersensitive to AZT. This is because M184V/I RT is unable to unblock AZT-terminated primers, thus leading to increased susceptibility to AZT and synergistic action.^{9,169,170,177–179,275} Another example of successful combination is TFV and FTC (truvada). In this case, the FTC-resistant M184V/I mutant RT is hypersensitive to TFV, while K65R reduces the susceptibility of TFV and FTC. K65R alone is not usually sufficient to suppress the activity of FTC in vivo.^{9,169,170,276–279} There are also examples of synergy in combinations of NRTIs and NNRTIs that use different mechanisms of RT inhibition. For example, RPV and TFV have been shown to inhibit RT synergistically because RPV stabilized complexes of TFV-terminated DNA primer/templates and RT.²⁸⁰ Similarly, it has also been reported that combination of RPV and EFdA have synergistic effects.²⁸¹ Finally, another study showed that NNRTIs DOR and RPV had nonoverlapping resistance profiles and it was thus proposed that they could be used in combination.²⁸² Hence, understanding in depth the molecular, biochemical, and virological basis of synergy between drugs is necessary for designing optimal combinations that will maximize suppression of a genetically diverse virus such as HIV.

5.9. Designing Drugs That Block Specific Resistance Mechanisms

In section 5.1, we discussed strategies to avoid the discrimination mechanism of RT resistance to FTC/3TC. Here, we discuss strategies to avoid the excision mechanism of RT. Excision based resistance mutations M41L/D67N/K70R/L210W/T215Y/K219Q form a novel pocket that enables binding of ATP that is used for a phosphorolytic reaction to unblock the chain terminated primer.¹⁷⁷ The products of the excision reaction is an unblocked DNA primer that can now continue to be extended by the polymerase action of RT as well as a tetraphosphate (AZTppppA). The structure of this tetraphosphate product in complex with RT carrying TAMs has been solved and the specific molecular interactions are known¹⁷⁷ (Figure 10). Hence, it is possible that an antiviral simultaneously targeting these sites can be made. Such a compound would serve both as a polymerase inhibitor, as it will occupy the dNTP binding site, but also as an inhibitor of the excision reaction, as it will also occupy the binding site of the ATP excision substrate. In designing such dual inhibitors, it is important to consider that the excision reaction is to a certain extent reversible, and the AZTppppA can serve as RT substrate, resulting into an AZT-terminated primer and ATP. Thus, nonhydrolyzable versions of the tetraphosphate moiety were made^283,284 (AZTpSpCX2ppSA in Figure 19), designed to block the reverse reaction. Therefore, compounds that block both the polymerase and the excision reaction that causes NRTI resistance can be rationally designed.

Figure 19.

Chemical structures of dinucleoside tetraphosphate analogues. ATP-based excision reaction of AZT-chain terminated template/primers leads to the production of AZTppppA dinucleoside tetraphosphate products. Analogues of such products are shown with substitutions of oxygen with sulfur atoms in the tetraphosphate moiety. These modifications ensure that the compounds cannot support DNA synthesis by RT. The design strategy aims for compounds that can be used as inhibitors of the polymerase and excision activities of RTs, especially those carrying excision mutations (TAMs). This is accomplished by binding at the N- and the ATP binding sites. Modified from 284. Copyright 2007 American Chemical Society.

5.10. Targeting DNA Synthesis by Multiple Mechanisms of Action

EFdA is a nucleoside analogue that blocks reverse transcription by multiple mechanism of action: (a) as discussed in section 5.3, it can block DNA synthesis either as an immediate or delayed chain terminator (ICT or DCT) (Figure 8).¹⁵⁰ Although the nontranslocated EFdA-MP-terminated primers in ICT can be unblocked, they are often converted back to the EFdA-MP-terminated form. However, when a primer with EFdA-MP at its 3′-end is further extended by a single nucleotide during the DCT mode of inhibition, the EFdA-MP-terminated primers are protected from excision. Thus, compounds that can act by delayed chain termination are less likely to be unblocked by the excision mechanism of drug resistance.

Another strength of EFdA is that it works by an additional mechanism of action; it can be efficiently misincorporated by RT, leading to mismatched primers that are extremely hard to extend and are also protected from excision.

The use of substituent that has an ethynyl (−C≡CH) or cyano (−C≡N) group added to the ribose ring of an NRTI has been used in compounds other than EFdA (Figure 19): a similar approach inspired the design of HBV NRTIs that have a 4′-cyano and are expected to act by a similar mechanism of inhibition (Figure 20).^285,286 In addition, remdesivir, a related compound with a 1-ethynyl substitution on the ribose ring, has been introduced as a potent RNA-dependent RNA polymerase inhibitor of Ebola and recently approved as the first antiviral against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, COVID-19) (Figure 20).^287–291 Hence, compounds that act by unique and multiple mechanisms of action may be able to maintain antiviral activity if only a single of their mechanism of action is reversed.

Figure 20.

Chemical structures of EFdA, CdG, and remdesivir.

5.11. Targeting the RNase H Function of RT

The RNase H function of RT degrades RNA in an RNA/DNA duplex. It is essential for HIV replication, and as such it could be a target for the development of antivirals that work by a different mechanism of action. Because the RNase H active site is distinct from the polymerase (Figure 1 and 13), it is expected that NRTI- and NNRTI-resistant RTs would still be susceptible to RNase H inhibitors. The topic of RNase H inhibitors has been studied extensively.^{247,249–256,292–296} Although there are currently no approved RNase H inhibitors, introduction of antivirals that targets this activity and their use as combinations with NRTIs and/or NNRTIs should be an efficient method to combat drug resistance, as there would be a low likelihood for cross-resistance among the various inhibitors that bind different active sites.

5.12. Long-Acting Regimens

Since the introduction of the first anti-HIV drugs, patient adherence has been a key challenge. Patients’ noncompliance leads to resistance mutations that in turn can lead to viral breakthrough and therapy failure. Daily regimens have dramatically improved over the years, and current HAART is based primarily on once-daily dosing regimens. Newer inhibitors that are more potent and pharmacologically stable could decrease the dosing frequency for patients. Consequently, patient compliance would increase, thus decreasing the incidents of drug resistance. Long-acting regimens are also important in pre-exposure prophylaxis (PrEP), as adherence to PrEP can decrease the overall prevalence and transmission of HIV (Table 3).

Table 3.

Half-Lives of Potential Long-Acting Drugs

	median terminal half-lives	ref
rilpivirine (RPV)	50 h	302
		https://www.accessdata.fda.gov/drugsatfda_docs/label/2015/202022s008lbledt.pdf
tenofovir alafenamide (TAF)	92 min*	*in dogs
		119
TFV-DP (active metabolite of TAF)	50−180 h	303,304
		https://www.accessdata.fda.gov/drugsatfda_docs/label/2016/207561s002lbl.pdf
islatravir (EFdA)	78.5−128 h	299

The most promising potential long-acting RT-targeting nucleoside analogue is EFdA, an NRTTI. Studies have shown that a single-injection extended-release form of EFdA maintained efficient inhibition activity for more than 180 days.²⁹⁷ With such promising results, EFdA is being considered for a variety of long-dosing regimens.^297–299 Currently, EFdA is in phase 2a trials for once-monthly doses (ClinicalTrials.gov identifier: NCT04003103). Also, EFdA is being considered for a once-yearly implant for PrEP regimens.^157,158

Another potential long-acting RT inhibitor is TAF, which has better potency than its predecessor, TDF, and is used at approximately 1/10th lower dose than TDF. A recent study found that a subdermal polyvinyl alcohol implant maintained inhibitory TFV concentration for more than six weeks.^118,300

NNRTIs such as RPV and elsulfavirine are also being studied for their long-acting potential. The HPTN 076 clinical trial evaluated the safety and efficacy of a long-acting form of RPV. The trial administered intramuscular injections every eight weeks and found that it was well-tolerated and effective and RPV could be used as a potential drug during PrEP.³⁰¹ An elsulfavirine preclinical study showed plasma levels of the active form to be above 50 ng/mL for a minimum of 4 weeks, thus providing promise for use as a long-acting drug.¹⁶⁰ Hence, introduction of long-acting RT-targeting antivirals will likely help overcome the challenge of patient compliance that leads to drug resistance.

5.13. Leading to a Dead-End Reaction Intermediate

INDOPY-1 is a non-nucleoside analogue RT inhibitor that acts by a distinct mechanism of action: it binds at the N- or dNTP-binding site of the polymerase active site and is thus known as nucleotide-competing RT inhibitor (NcRTI). When bound to RT, INDOPY-1 stacks with the terminal DNA base pair while intercalating with the first overhang template base (Figure 21). In doing so, the inhibitor forms a dead-end reaction intermediate that blocks further DNA synthesis. Notably, interactions of INDOPY-1 at the active site involve some conserved residues, including K65, R72, D110, D185, as well as the main chain N of G152.^274,305,306 Resistance to INDOPY-1 has been reported through the FTC/3TC resistance mutation M184V and through Y115F that is known to cause resistance to multiple NRTIs.³⁰⁷ The TFV resistance mutation K65R confers hypersusceptibility to INDOPY-1.³⁰⁸ Hence, compounds that block RT by leading to dead-end reaction intermediates while engaging in interactions with conserved residues and main chain atoms may offer another strategy to minimize drug resistance mutations.

Figure 21.

Structural basis of HIV-1 RT inhibition by nucleotide competing RT inhibitor (NcRTI) INDOPY-1. (A) Chemical structure of INDOPY-1. (B) Structure of INDOPY-1 (orange sticks) in complex with HIV-1 RT (yellow sticks) and template/primer (pink sticks). (B) Reproduced modified from ref 305. Copyright 2019 American Chemical Society.

6. CONCLUSIONS

HIV RT is a key enzyme in the viral replication cycle, and it still remains a major drug target for the development of anti-HIV drugs. Great efforts have been made since the discovery of the first RT inhibitor. However, a major challenge and the reason for therapy failure is the emergence of drug resistance mutations. Developing new inhibitors based on the various strategies mentioned in this review may help improve the ability of compounds to perform better in terms of drug resistance. While it is important to note that fully avoiding drug resistance is likely not possible, using these strategies may help the field move in the right direction. Furthermore, these strategies can be applied to not just RT inhibitors but other viral enzymes as well.

Biographies

Maria E. Cilento is currently a Ph.D. candidate under the supervision of Dr. Stefan Sarafianos in the Division of Laboratory of Biochemical Pharmacology in the Department of Pediatrics at Emory University School of Medicine. She received her B.S. from the Department of Biology at Stockton University, and her Master of Science in Microbiology & Immunology at Thomas Jefferson University, where she conducted research on HIV entry. Her current research focuses on studying the mechanism of inhibition and resistance of HIV from various subtypes to 4′-ethynyl-2-fluoro-2′-deoxyadenosine (EFdA/Islatravir).

Dr. Karen A. Kirby is an Assistant Professor in the Division of Laboratory of Biochemical Pharmacology in the Department of Pediatrics at Emory University School of Medicine. She received her B.S. in Chemistry from the University of Minnesota and her Ph.D. in Chemistry from the University of Missouri, studying small molecule crystallography. She was trained in structural biology related to HIV and other viral proteins, biochemistry, and biophysics (Dr. Stefan Sarafianos, University of Missouri). Her current research focuses on structural mechanisms of inhibition and resistance to HIV and other viral pathogens.

Dr. Stefan G. Sarafianos is a Professor and Associate Director of the Division of Laboratory of Biochemical Pharmacology in the Department of Pediatrics at Emory University School of Medicine. He received his Ph.D. at the Department of Chemistry at Georgetown University. He was trained in molecular biology of HIV (Dr. Mukund Modak, UMDNJ), in structural biology and virology by Dr. Eddy Arnold (Center for Advanced Biotechnology and Medicine, Rutgers University,) and Dr. Stephen Hughes (NCI, NIH), working on structural mechanisms of inhibition and resistance in HIV reverse transcriptase. He is currently the Nahmias Schinazi Distinguished Chair in Research, and his research focuses on HIV, HBV, SARSCoV-2, and other pathogens.