Elicit: Role of Emtricitabine and Tenofovir Alafenamide in NRTI Inhibition

Skip to main content

Role of Emtricitabine and Tenofovir Alafenamide in NRTI Inhibition

Research reportView only

Create alertChat

May 5, 2026

What is the role of emtricitabine and tenofovir alafenamide in NRTI-mediated reverse transcription inhibition?

Emtricitabine and tenofovir alafenamide inhibit HIV-1 reverse transcription through chain termination with complementary resistance profiles that create a high genetic barrier to viral escape when used in combination.

Abstract

Emtricitabine and tenofovir alafenamide inhibit HIV-1 reverse transcription through complementary mechanisms that create a high barrier to viral resistance. Emtricitabine is phosphorylated to FTC-TP, which competes with endogenous dCTP for reverse transcriptase binding and causes chain termination upon incorporation into viral DNA. TAF functions as a prodrug that delivers tenofovir diphosphate (TFV-DP) to target cells with 4- to 7-fold higher intracellular concentrations than tenofovir disoproxil fumarate, enabling superior activity against NRTI-resistant viral populations. In viral breakthrough experiments modeling clinical concentrations, TAF suppressed 65 of 68 resistance mutants while TDF suppressed only 53 of 68.

The two drugs exhibit synergistic resistance profiles. While the M184V mutation confers >500-fold resistance to FTC, it simultaneously increases susceptibility to TAF, allowing continued virologic suppression even when FTC resistance emerges. In clinical trials of treatment-naive patients receiving elvitegravir/cobicistat/emtricitabine/TAF, only 0.7% developed NRTI resistance mutations over 48 weeks, demonstrating the combination’s high genetic barrier. TAF’s elevated intracellular TFV-DP levels provide therapeutic resilience even under conditions that compromise drug exposure, maintaining 4.21-fold higher levels than TDF when co-administered with rifampicin despite 36% reductions in TFV-DP concentrations. Together, FTC and TAF create a robust platform for reverse transcription inhibition through enhanced intracellular drug delivery, potent wild-type RT inhibition, and complementary resistance mutation interactions.

Methods

We analyzed 10 sources from an initial pool of 200, using 8 screening criteria. Each paper was reviewed for 6 key aspects that mattered most to the research question. More on methods

Records from Elicit search

n = 200

Papers screened using: Drug Specificity, Mechanism Focus, HIV Relevance, HIV Indication, TAF vs TDF, NRTI Focus, Mechanistic Data, Study Design and Quality

n = 200

Papers screened out

n = 190

Papers included for extraction

n = 10

Press enter or space to select a node.You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.

Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Paper search

We performed a semantic search across over 138 million academic papers from the Elicit search engine, which includes all of Semantic Scholar and OpenAlex.

We ran this query: “What is the role of emtricitabine and tenofovir alafenamide in NRTI-mediated reverse transcription inhibition?”

The search returned 200 total results from Elicit.

We retrieved 200 papers most relevant to the query for screening.

Screening

We screened in sources based on their abstracts that met these criteria:

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

Extract detailed mechanisms by which emtricitabine (FTC) and/or tenofovir alafenamide (TAF) inhibit HIV-1 reverse transcription, including:

Extract molecular-level details of how FTC and/or TAF interact with HIV-1 reverse transcriptase, including:

Extract information about resistance to FTC and/or TAF in the context of NRTI-mediated RT inhibition, including:

Extract comparisons of FTC and/or TAF with other NRTIs relevant to reverse transcription inhibition, including:

Extract pharmacological characteristics of FTC and/or TAF relevant to their role in NRTI-mediated RT inhibition, including:

Extract essential study details needed to interpret findings about FTC/TAF mechanisms, including:

Results

Characteristics of Included Studies

Study

Full text retrieved?

Study Type

Research Focus

Samuel DeKoven et al., 2023

No

Case series

Treatment-emergent NRTI resistance in patients with suboptimal adherence to BIC/TAF/FTC

N. Margot et al., 2020

No

In vitro study

TAF and TDF antiviral activity against HIV-1 with TAMs and M184V using multicycle and single-cycle assays

N. Margot et al., 2016

No

In vitro study

TAF resistance barrier using viral breakthrough assays with clinical isolates harboring TFV resistance mutations

N. Margot et al., 2015

Yes

In vitro study

TAF resistance characterization using MT-2 cells and patient-derived isolates over 115-147 days

M. Cerrone et al., 2019

No

Pharmacokinetic study

TAF/FTC pharmacokinetics with rifampicin in healthy volunteers

Lynne M. Bang & L. Scott, 2012

Yes

Phase I clinical trial

Drug-drug interaction between HS-10234 and emtricitabine in healthy volunteers

C. Callebaut et al., 2015

Yes

In vitro study

TAF virology profile using MT-2, MT-4 cell lines and primary CD4 T cells across HIV-1 group M subtypes A-G, group N, O, and HIV-2

N. Margot et al., 2016a

No

Phase 2/3 clinical trials

TAF resistance development in 1903 treatment-naive patients over 48 weeks

Magdeleine Hung et al., 2019

Yes

In vitro structural study

FTC and 3TC molecular interactions with HIV-1 RT using kinetic assays and crystal structures

Joy Y. Feng et al., 2006

No

In vitro study

FTC and 3TC resistance mechanisms with K65R and Q151M mutations using kinetic analysis

toof

Pageof

The included studies comprise both in vitro mechanistic investigations and clinical pharmacological evaluations. Four studies had full text available, while six were limited to abstract-only data. Research approaches included resistance selection experiments, viral breakthrough assays, kinetic analyses, crystal structure determination, and pharmacokinetic assessments in both cell culture systems and human subjects.

Mechanisms of Reverse Transcription Inhibition

Emtricitabine Mechanism of Action

Emtricitabine functions as a nucleoside reverse transcriptase inhibitor through phosphorylation by cellular enzymes to emtricitabine 5’-triphosphate (FTC-TP), which competes with the endogenous substrate deoxycytidine triphosphate (dCTP) for HIV-1 reverse transcriptase binding. Upon incorporation into viral DNA, FTC-TP causes chain termination, thereby inhibiting viral replication.

The structural basis for FTC activity involves an oxathiolane ring with unnatural (-)-stereochemistry. Crystal structure data reveal that FTC-TP binds to HIV-1 RT with the oxathiolane sulfur oriented toward the DNA primer 3’-terminus, while the triphosphate moiety exists in two distinct binding conformations. Key amino acid residues involved in binding include Y115, M184, R72, K65, and K220, which coordinate the triphosphates and interact with the oxathiolane ring. The oxathiolane ring provides greater hydrophobic contact with Y115 compared to natural dCTP.

Kinetic analyses demonstrate that FTC-TP has higher binding affinity for wild-type RT compared to dCTP but exhibits slower incorporation rates. FTC-TP was incorporated at least as efficiently as 3TC-TP across all HIV-1 RT variants and primer/template combinations tested.

Tenofovir Alafenamide Mechanism of Action

Tenofovir alafenamide serves as a prodrug of tenofovir (TFV), a nucleotide reverse transcriptase inhibitor. TAF undergoes intracellular conversion to TFV via enzymes including cathepsin A or carboxyesterase 1, depending on cell type. TFV is subsequently phosphorylated to its active metabolite, tenofovir diphosphate (TFV-DP), which inhibits HIV-1 reverse transcriptase through chain termination.

The phosphonate group in TFV mimics the first phosphate group of natural nucleotides but contains a C-O-O bond substitution, rendering it less susceptible to cellular phosphatase activities. This structural modification allows TFV to require only two phosphorylation steps for activation, compared to three steps for conventional nucleosides. TFV-DP is incorporated into viral DNA by HIV reverse transcriptase, functioning as a DNA chain terminator.

Pharmacological Properties and Drug Delivery

Intracellular Drug Concentrations

TAF achieves greater than 4-fold higher intracellular levels of TFV-DP compared to tenofovir disoproxil fumarate (TDF). In peripheral blood mononuclear cells (PBMCs), TAF increases TFV-DP concentration by 5- to 7-fold relative to TDF. This enhanced intracellular loading results from TAF’s improved plasma stability and more efficient cellular uptake compared to TDF.

TAF preferentially distributes into peripheral lymphocytes and lymphatic tissues, which represent primary target cells for HIV replication. The prodrug maintains stability in blood and plasma superior to TDF. TAF consistently converts to TFV in PBMCs, demonstrating effective prodrug conversion efficiency. The active metabolite TFV-DP exhibits a long intracellular half-life exceeding 60 hours.

Drug-Drug Interactions

Rifampicin significantly affects TAF pharmacokinetics. When tenofovir alafenamide/emtricitabine was administered with rifampicin, TAF exposure decreased to 45% of levels without rifampicin (GMR 90% CI: 0.45, 0.33-0.60). Plasma tenofovir concentrations declined to 46% (GMR 90% CI: 0.46, 0.40-0.52) and intracellular tenofovir-DP concentrations decreased to 64% (GMR 90% CI: 0.64, 0.54-0.75) with rifampicin co-administration. Despite these reductions, intracellular tenofovir-DP levels with TAF plus rifampicin remained 4.21-fold higher than those achieved with TDF alone (GMR 90% CI: 4.21, 2.98-5.95). Rifampicin did not affect emtricitabine pharmacokinetics. The CYP3A4*22 rs35599367 genetic polymorphism was associated with higher plasma TAF concentrations.

Resistance Mechanisms

Primary Resistance Mutations

Mutation

Drug affected

Impact on binding/incorporation

Fold-change in susceptibility

M184V/I

FTC, 3TC

Reduced binding affinity and incorporation rate; repositioning of oxathiolane and shift of triphosphate to non-productive conformation

500-fold for FTC and 3TC

K65R

TAF, TFV

Allows RT to discriminate against TFV while recognizing natural substrates; decreased rate of incorporation

6.5-fold reduced susceptibility to TAF

toof

Pageof

The M184V mutation represents the primary resistance mechanism selected by FTC and 3TC treatment. This mutation significantly increases Kd values for both drugs, indicating reduced binding affinity. Structurally, M184V causes steric hindrance that prevents proper drug binding and shifts the triphosphate conformation into a non-productive state.

K65R confers reduced susceptibility to multiple NRTIs including 3TC, ddC, ddI, abacavir, and tenofovir. The mutation enables reverse transcriptase to discriminate against TFV while maintaining recognition of natural substrates. Resistance associated with K65R and the double mutant K65R/Q151M results primarily from decreased incorporation rates rather than altered binding affinities.

Thymidine Analog-Associated Mutations

Thymidine analog-associated mutations (TAMs), including M41L, D67N, K70R, L210W, T215Y, and K219Q, confer reduced TFV susceptibility. These mutations confer high-level resistance to zidovudine (AZT) and low-level cross-resistance to TFV and other NRTIs. However, the presence of M184V in TAM-containing HIV-1 significantly increases TAF sensitivity compared to variants without M184V.

In viral breakthrough experiments mimicking physiological drug concentrations, 15 of 68 mutants broke through with TFV (the in vitro equivalent of TDF), while only 3 of 68 broke through under TAF treatment. TAF inhibited viral breakthrough of most TAM-containing HIV-1, whereas TDF did not. These results indicate that TAF has a higher resistance threshold than TDF.

Complex Resistance Mutations

Additional resistance patterns include Q151M/K65R and T69 insertion complex mutations. At clinically relevant concentrations, TAF could inhibit HIV-1 clinical isolates harboring these TFV resistance mutations, while TFV could not. The Q151M mutation confers resistance to many NRTIs but remains sensitive to FTC, 3TC, and tenofovir.

Comparative NRTI Activity

TAF versus TDF Potency

TAF demonstrates superior antiviral activity compared to TDF across multiple parameters. The inhibitory quotient (IQ) of TAF is projected to be higher than the IQ of TDF, with potential to inhibit viruses containing TDF resistance in clinical settings. In cell culture, while TAF and TDF exhibited comparable potencies in the absence of serum pretreatment, TAF maintained activity in the presence of human serum, whereas TDF activity was significantly reduced. This difference reflects TAF’s improved plasma stability driven by distinct metabolic pathways.

Against NRTI-resistant viral strains, TAF showed reduced activity against isolates with extensive NRTI resistance mutations but retained activity against isolates with K65R mutation or multiple TAMs. TAF’s higher cellular concentrations of TFV-DP provide a higher genetic barrier compared to other NRTIs like 3TC and AZT.

FTC versus 3TC Efficacy

FTC demonstrated a 1.7 log reduction in viral HIV RNA in clinical trials. In combination therapy, emtricitabine-based regimens were as effective as lamivudine-based triple therapy and significantly more effective than stavudine or protease inhibitor-based therapy at achieving and maintaining durable suppression of HIV levels after 24-48 weeks. The M184V mutation confers high resistance to both FTC and 3TC (>500-fold reduction in potency), but this mutation increases susceptibility to TDF, stavudine, and AZT.

Combination Effects

The presence of the M184V/I substitution, associated with FTC resistance, increases susceptibility to TFV and TAF. This interaction suggests a synergistic relationship where M184V-mediated resistance to one drug class can enhance susceptibility to another. In vitro combination studies with antiretroviral drugs from different classes showed additive to synergistic interactions with TAF, consistent with ongoing clinical studies using TAF in fixed-dose combinations with multiple other antiretroviral drugs.

Clinical Resistance Development

In integrated Phase 2 and Phase 3 studies involving 1903 treatment-naive patients receiving elvitegravir/cobicistat/emtricitabine/TAF (E/C/F/TAF) for 48 weeks, resistance development was rare. Only 0.7% (7/978) of patients in the E/C/F/TAF group developed NRTI resistance-associated mutations, with 0.5% (5/978) also developing primary integrase strand transfer inhibitor resistance-associated mutations. The pattern of emergent mutations was similar between E/C/F/TAF and E/C/F/TDF groups. At screening, 7.5% of patients had pre-existing NRTI resistance-associated mutations, 18.2% had NNRTI resistance-associated mutations, and 3.4% had primary PI resistance-associated mutations, but these pre-treatment mutations did not influence treatment response at week 48.

Antiviral Spectrum

TAF displayed potent antiviral activity against all HIV-1 groups and subtypes, as well as HIV-2. Testing across HIV-1 group M subtypes A through G, group N, and group O confirmed broad-spectrum activity. TAF showed full antiviral activity in PBMCs against primary HIV-1 isolates with protease inhibitor, non-nucleoside RT inhibitor, or integrase strand transfer inhibitor resistance, but reduced activity against isolates with extensive NRTI resistance amino acid substitutions.

The activity of TAF is specific for HIV. TAF lacked activity against a large panel of human viruses, with the exception of herpes simplex virus 2, where weak antiviral activity was observed, consistent with previous observations for TFV.

Synthesis

The evidence demonstrates that TAF provides multiple pharmacological advantages over TDF for NRTI-mediated reverse transcription inhibition, despite sharing the same active metabolite. These advantages arise from distinct prodrug properties rather than fundamental differences in RT inhibition mechanisms.

The 4- to 7-fold higher intracellular TFV-DP concentrations achieved by TAF compared to TDF translate directly into superior activity against resistant viral populations. In viral breakthrough experiments modeling clinical drug concentrations, TAF suppressed 65 of 68 resistance mutants while TDF suppressed only 53 of 68. This difference is particularly pronounced for viruses harboring TAMs, which confer reduced TFV susceptibility. The higher resistance threshold of TAF versus TDF reflects pharmacokinetic superiority rather than altered resistance mutation patterns, as both prodrugs select for identical resistance mutations when used at equivalent intracellular concentrations.

The interaction between FTC/3TC resistance (M184V) and tenofovir susceptibility represents a clinically relevant synergistic relationship. M184V confers >500-fold resistance to FTC and 3TC but simultaneously increases susceptibility to TFV, stavudine, and AZT. In TAM-containing viruses, the addition of M184V significantly improved TAF sensitivity. This antagonistic pleiotropy between resistance mutations explains why FTC/TAF combinations maintain virologic suppression even when M184V emerges, as the M184V mutation that confers FTC resistance paradoxically enhances TAF activity.

The distinct resistance barriers of TAF and FTC complement each other in fixed-dose combinations. TAF’s primary resistance mutation K65R reduces susceptibility 6.5-fold, while FTC’s M184V reduces activity >500-fold. However, the clinical impact of these mutations differs substantially. K65R resistance to TAF is partially overcome by TAF’s elevated intracellular TFV-DP levels, whereas M184V completely ablates FTC activity but enhances TAF susceptibility. The extremely low rates of resistance development in clinical trials (0.7% developing NRTI mutations over 48 weeks) suggest that the combination presents a high barrier to resistance evolution, requiring simultaneous mutations that have antagonistic fitness effects.

Even in the presence of strong enzyme inducers like rifampicin, which reduce TAF exposure by 55% and intracellular TFV-DP by 36%, TAF maintains 4.21-fold higher intracellular TFV-DP levels than TDF. This pharmacological resilience suggests that TAF’s therapeutic margin is substantially wider than TDF’s, providing activity even under conditions that compromise drug exposure. The combination of TAF’s enhanced intracellular delivery with FTC’s potent wild-type RT inhibition and favorable resistance interactions creates a robust therapeutic platform for reverse transcription inhibition across diverse clinical scenarios.

References

Samuel DeKoven, Mark Naccarato, C. Brumme, D. Tan\ (2023).Treatment‐emergent reverse transcriptase resistance during antiretroviral therapy with bictegravir, tenofovir alafenamide, and emtricitabine: A case series. HIV Medicine

N. Margot, Audun Johnson, Michael D. Miller, C. Callebaut\ (2015).Characterization of HIV-1 Resistance to Tenofovir Alafenamide In Vitro. Antimicrobial Agents and Chemotherapy

M. Cerrone, O. Alfarisi, M. Neary, M. Marzinke, T. Parsons, and 5 more\ (2019).Rifampicin effect on intracellular and plasma pharmacokinetics of tenofovir alafenamide. Journal of Antimicrobial Chemotherapy

Lynne M. Bang, L. Scott\ (2012).Emtricitabine. Drugs

C. Callebaut, G. Stepan, Ya-Wen Tian, Michael D. Miller\ (2015).In Vitro Virology Profile of Tenofovir Alafenamide, a Novel Oral Prodrug of Tenofovir with Improved Antiviral Activity Compared to That of Tenofovir Disoproxil Fumarate. Antimicrobial Agents and Chemotherapy

N. Margot, Renee R. Ram, M. Abram, R. Haubrich, C. Callebaut\ (2020).Antiviral Activity of Tenofovir Alafenamide against HIV-1 with Thymidine Analog-Associated Mutations and M184V. Antimicrobial Agents and Chemotherapy

N. Margot, K. Kitrinos, M. Fordyce, S. McCallister, Michael D. Miller, and 1 more\ (2016).Rare emergence of drug resistance in HIV-1 treatment -naïve patients after 48 weeks of treatment with elvitegravir/cobicistat/emtricitabine/tenofovir alafenamide. HIV Clinical Trials

Magdeleine Hung, E. J. Tokarsky, Leanna L. Lagpacan, Lijun Zhang, Z. Suo, and 1 more\ (2019).Elucidating molecular interactions of L-nucleotides with HIV-1 reverse transcriptase and mechanism of M184V-caused drug resistance. Communications Biology

N. Margot, Yang Liu, Michael D. Miller, C. Callebaut\ (2016).High resistance barrier to tenofovir alafenamide is driven by higher loading of tenofovir diphosphate into target cells compared to tenofovir disoproxil fumarate. Antiviral Research

Joy Y. Feng, Florence T. Myrick, N. Margot, G. Mulamba, L. Rimsky, and 3 more\ (2006).Virologic and Enzymatic Studies Revealing the Mechanism of K65R- and Q151M-Associated HIV-1 Drug Resistance Towards Emtricitabine and Lamivudine. Nucleosides, Nucleotides & Nucleic Acids

Download BIBDownload RISDownload TXT

Report

Status

Gather sources

200 sources found

Details

Screen sources

10 sources included

Details

Extract data

60 data points extracted

Details

Generate report

Save PDF

BIBLaTeX, ZoteroRISZotero, MendeleyTXTAPA BibliographyPDFPDFDOCXMicrosoft Word

Chat

Got some follow-up questions?

Sign up or sign in to chat with this report.

Back

Elucidating molecular interactions of L-nucleotides with HIV-1 reverse transcriptase and mechanism of M184V-caused drug resistance

Magdeleine Hung, E. J. Tokarsky, Leanna L. Lagpacan, Lijun Zhang, Z. Suo, E. Lansdon

Communications Biology·

2019·

13 citations

SourceDOI

Plain textPDF

Searching for PDF

Unable to find PDF from source

Back

Drug Mechanisms

- Mechanism of action: Chain termination - Active metabolite form: FTC-TP - Binding interactions: Coordination with Mg2+, R72, K65, K220; hydrophobic contact with Y115 - Structural features: Oxathiolane ring with unnatural (-)-stereochemistry - Incorporation kinetics and binding affinities: Higher binding affinity than dCTP but slower incorporation rate; M184V mutation increases Kd and shifts triphosphates into non-productive conformation

Molecular Interactions

- Crystal structure data: FTC and 3TC interact with HIV-1 RT with their oxathiolane sulfur oriented towards the DNA primer 3′-terminus, and their triphosphates exist in two binding conformations. - Key amino acid residues: Y115, M184, R72, K65, and K220 are involved in binding and coordinating triphosphates. - Differences from natural nucleotides: Oxathiolane ring provides greater hydrophobic contact with Y115 compared to dCTP. - Structural basis for selectivity and specificity: Unique stereochemistry of the oxathiolane ring interacts with key residues like M184. - Phosphorylation pathways and cellular processing: Not explicitly mentioned in the provided text.

Resistance Mechanisms

- Specific resistance mutations: M184V - Impact of mutations on drug binding and incorporation: Reduced binding affinity and incorporation rate - Cross-resistance patterns with other NRTIs: Increased susceptibility to TDF, stavudine, and AZT - Structural basis for resistance: Steric hindrance and conformational changes - Fold-change in susceptibility for resistant variants: >500-fold for FTC and 3TC

Comparative NRTI Activity

- Relative potency and efficacy: FTC demonstrated a 1.7 log reduction in viral HIV RNA in clinical trials. - Differences in resistance barriers and genetic barriers: M184V mutation confers high resistance to FTC and 3TC, with a >500-fold reduction in potency. - Activity against NRTI-resistant viral strains: M184V increases susceptibility to TDF, stavudine, and AZT. - Synergistic or antagonistic effects with other NRTIs: Not mentioned.

Pharmacological Properties

- Intracellular half-lives of active metabolites: Not mentioned - Cellular uptake and phosphorylation efficiency: FTC and 3TC are converted by cellular kinases into their active triphosphate forms - Tissue distribution and target cell penetration: Not mentioned - Drug-drug interactions affecting RT inhibition: FTC is combined with TAF, but specific interactions are not detailed - Prodrug conversion efficiency (for TAF vs TDF): Not mentioned

Study Context

- Experimental system: In vitro - Cell lines or patient populations: Not mentioned - Experimental methods: Pre-steady-state kinetic assays, crystal structure determination - HIV-1 strains or subtypes: M184V mutation in HIV-1 RT - Time points and duration of observations: Not explicitly detailed

Emtricitabine (FTC) and lamivudine (3TC), containing an oxathiolane ring with unnatural (−)-stereochemistry, are widely used nucleoside reverse transcriptase inhibitors (NRTIs) in anti-HIV therapy. Treatment with FTC or 3TC primarily selects for the HIV-1 RT M184V/I resistance mutations. Here we provide a comprehensive kinetic and structural basis for inhibiting HIV-1 RT by (−)-FTC-TP and (−)-3TC-TP and drug resistance by M184V. (−)-FTC-TP and (−)-3TC-TP have higher binding affinities (1/Kd) for wild-type RT but slower incorporation rates than dCTP. HIV-1 RT ternary crystal structures with (−)-FTC-TP and (−)-3TC-TP corroborate kinetic results demonstrating that their oxathiolane sulfur orients toward the DNA primer 3′-terminus and their triphosphate exists in two different binding conformations. M184V RT displays greater (>200-fold) Kd for the L-nucleotides and moderately higher (>9-fold) Kd for the D-isomers compared to dCTP. The M184V RT structure illustrates how the mutation repositions the oxathiolane of (−)-FTC-TP and shifts its triphosphate into a non-productive conformation. Magdeleine Hung et al. report the comprehensive kinetic and structural basis for the inhibition of HIV-1 Reverse Transcriptase by FTC and 3TC, as well as the drug resistant M184V variant. Structural analysis of HIV-1 RT and inhibitors provide insights into the mechanism of binding and resistance.

H

uman immunodeficiency virus (HIV) and acquired immunodeficiency syndrome (AIDS) afflict approximately 36.9 million people worldwide according to the WHO 1 . The most effective treatment regimen is highly active antiretroviral therapy (HAART) which consists of a backbone of nucleoside or nucleotide reverse transcriptase inhibitors (NRTIs) combined with either a protease inhibitor, a non-nucleoside reverse transcriptase inhibitor (NNRTI), or an integrase strand transfer inhibitor (INSTI). Of the available NRTIs as treatment options, tenofovir disoproxil fumarate (TDF) and emtricitabine (FTC) (Fig. 1 ) are part of the recommended treatment backbone 2 . These two drugs are available in a single pill (Truvada®) which in addition to being used in HIV therapy is approved for preexposure prophylaxis to prevent HIV infection. Single tablet regimens combining TDF and FTC with an additional active anti-HIV agent are also available (Atripla®, Complera®, and Stribild®). These treatment options allow patients to take just one pill a day to control their HIV infection. Recently, FTC has been combined with tenofovir alafenamide, a next generation prodrug of tenofovir approved for HIV treatment (Descovy®), along with elvitegravir and cobicistat (Genvoya®) as well as rilpivirine (Odefsey®).

NRTIs mimic natural deoxynucleosides or deoxynucleotides and target the reverse transcriptase (RT) enzyme. RT incorporates active metabolites of NRTIs into the growing DNA chain which act as chain terminators due to their lack of 3ʹ-OH. The NRTIs FTC (2ʹ,3ʹ-dideoxy-5-fluoro-3ʹ-thiacytidine) and 3TC (lamivudine; 2ʹ,3ʹ-dideoxy-3ʹ-thiacytidine) are cytidine analogs. FTC and 3TC are both administered in their 5ʹ-OH form and once absorbed into cells are converted by cellular kinases into their active metabolites FTC-triphosphate ((-)-FTC-TP) and 3TCtriphosphate ((-)-3TC-TP), respectively 3 . The triphosphate form is recognized by HIV-1 RT and is incorporated into the DNA primer strand, leading to the termination of viral replication. In initial cell-based screening assays, FTC was found to be more potent and less cytotoxic than its corresponding (+)-stereoisomer against the HIV virus, thus leading to further development of the (-)-enantiomer 3 . Most notably, in place of the deoxyribose ring, these drugs contain an oxathiolane ring in a (-)-β-L configuration, with the opposite stereochemistry from normal D-ribose (Fig. 1 ). In clinical trials as a monotherapy, FTC demonstrated a 1.7 log reduction in viral HIV RNA 4 . In addition to being approved to treat HIV, FTC and 3TC are also effective inhibitors of hepatitis B virus (HBV) polymerase. Currently, 3TC is an approved antiviral drug for HBV treatment and FTC has been tested in clinical trials 5 .

Both in vitro and in vivo experiments have shown that FTC and 3TC primarily select for resistance mutations, M184V or M184I [6][7][8] . M184 is part of the conserved active site YMDD motif found in several viral polymerases, including HIV-1 RT and HBV polymerase. The M184I mutation will often emerge first, possibly because it results from a single nucleotide change 9 . M184V arises from a two nucleotide change but it outcompetes M184I and is observed in most patients with virologic failure resulting from FTC or 3TC treatment 8,9 . The M184V mutation has been shown to have multiple effects on RT activity and resistance 10 . As a single amino acid residue mutation, M184V confers very high resistance to FTC and 3TC as compared to WT (>500-fold) 6 . The mutation has been associated with lower polymerase processivity; however, it confers higher fidelity for correct dNTPs [11][12][13] and is connected with reduced viral replication 14 . While M184V provides resistance to FTC and 3TC, it can concurrently lead to an increased susceptibility to TDF, stavudine, and azidothymidine (AZT), and can slow the resistance development against these NRTIs 15 .

The mechanism of M184V resistance to FTC and 3TC has been debated in the literature. Several groups have explored whether M184V confers NRTI resistance by reducing binding affinities of these L-nucleotide analogs by the HIV-1 RT or whether it affects the rates of L-nucleotide analog incorporation. Steady-state and pre-steady-state kinetic data collected by Krebs et al. suggested that the M184V mutation primarily affects the rate of incorporation of the L-nucleotide analogs 16 . Additionally, Gao et al. using gel shift assays, demonstrated reduced rates of Lnucleotide analog incorporation by the M184V mutant with no effect on their binding 17 . On the other hand, kinetic experiments Fig. 1 Chemical structures of (-)-FTC-TP, (+)-FTC-TP, (-)-3TC-TP, (+)-3TC-TP, and the natural substrate deoxycytidine triphosphate (dCTP). The stereochemistry of the oxathiolane ring has an unnatural L-conformation in (-)-FTC-TP and (-)-3TC-TP as opposed to the natural D-conformation found with dCTP, (+)-3TC-TP, and (+)-FTC-TP.

from Wilson et al. and Feng et al. suggest that the mutation primarily affects L-nucleotide analog binding to HIV-1 RT 18,19 .

Crystal structures of (-)-FTC-TP and (-)-3TC-TP have been reported with various DNA polymerases 20,21 but not HIV-1 RT. We sought to establish the features of WT HIV-1 RT that recognize the unique stereochemistry of (-)-β-L oxathiolane analogs. Herein, using the same DNA construct, we report the pre-steady-state kinetic analysis with WT RT and M184V to determine the incorporation efficiency of both (+)-and (-)-enantiomers of FTC-TP and 3TC-TP compared to dCTP in addition to crystal structures of WT RT in complex with dsDNA (RT-DNA) bound to either, (-)-FTC-TP, (-)-3TC-TP, (+)-FTC-TP, or dCTP. Furthermore, binary (M184V-DNA), and ternary structures with (-)-FTC-TP and dCTP were determined to better understand the mechanism by which the M184V mutation confers resistance to NRTIs. These comprehensive kinetic and structural studies provide the most definitive insights into the binding modes of oxathiolane analogs and the mechanism of NRTI resistance achieved by the M184V mutation of HIV-1 RT.

Results

Pre-steady-state kinetics of L-nucleotide incorporation. Previous kinetic studies 16,18,20,[22][23][24][25][26] have aimed at understanding the efficiency of incorporation for (-)-FTC-TP or (-)-3TC-TP, how they compete against dCTP, and the effect of resistance mutations, such as M184V. As in our previous work 20,25,26 , single-turnover kinetic assays were performed by rapidly mixing a pre-incubating solution of 120 nM HIV-1 RT (WT or M184V) and 30 nM 5ʹ-[ 32 P]-labeled 18/26-mer dsDNA substrate (Methods) with varying concentrations of dCTP, or an active metabolite of NRTIs for different times, before being quenched by 0.37 M EDTA. Representative kinetic plots displaying M184Vcatalyzed incorporation of (-)-FTC-TP onto 18/26-mer dsDNA substrate are shown in Fig. 2 . Notably, the same 18/26-mer dsDNA substrate was used in both the single-turnover kinetic assays and crystal structure determination (see below). We determined the pre-steady-state kinetic parameters of maximal nucleotide incorporation rate constants (k p ) and apparent equilibrium dissociation constants (K d ) for dCTP, (+)-FTC-TP, (-)-FTC-TP, (+)-3TC-TP and (-)-3TC-TP with WT or M184V HIV-1 RT (Table 1 ). Comparing the measured K d values for WT RT, both (-)-FTC-TP (0.10 ± 0.01 µM) and (-)-3TC-TP (0.25 ± 0.01 µM) exhibit substantially tighter binding compared to natural dCTP (6.3 ± 0.2 µM), a 63-and 25-fold increase, respectively. Conversely, the k p was 119-and 175-fold slower for (-)-FTC-TP (0.100 ± 0.001 s -1 ) and (-)-3TC-TP The k obs values were then plotted against respective concentrations of (-)-FTC-TP and the plot was fit to a hyperbolic equation (Materials and Methods) to yield a k p of 0.073 ± 0.001 s -1 and a K d of 28 ± 2 μM (Table 1 ). (0.068 ± 0.001 s -1 ), respectively, than dCTP (11.9 ± 0.1 s -1 ). The resulting selectivity factors ((k p /K d ) dCTP /(k p /K d ) NRTI ) were determined to be only 1.9 and 7.0-fold higher for dCTP over (-)-FTC-TP and (-)-3TC-TP, respectively. These data suggest that these chain-terminating L-nucleotide analogs are excellent competitive inhibitors as to dCTP during WT HIV-1 RT-catalyzed viral genome replication. Kinetic analysis of the (+)-analogs show similar K d values to the (-)-analogs; however, the k p values were slightly higher resulting in selectivity factors of 2.4 and 2.5 for (+)-FTC-TP and (+)-3TC-TP, respectively. Furthermore, (-)-FTC-TP (1.0 µM -1 s -1 ) was incorporated by WT HIV-1 RT with a 3.7-fold higher efficiency (k p /K d ) than (-)-3TC-TP (0.27 µM -1 s -1 ), a result driven by the 2.5-fold binding affinity increase (Table 1 ).

As expected, M184V incorporated dCTP (k p /K d = 2.3 µM -1 s -1 ) with a similar efficiency to WT (k p /K d = 1.9 µM -1 s -1 ) (Table 1 ). However, the binding affinities of the M184V mutant for (-)-FTC-TP and (-)-3TC-TP were reduced (280-fold and 212-fold, respectively), compared to WT, whereas the k p values remain relatively unchanged. Conjointly, the selectivity factors increased to 880 for (-)-FTC-TP and 2400 for (-)-3TC-TP. Interestingly, (+)-FTC-TP (K d = 1.8 ± 0.3 µM) and (+)-3TC-TP (K d = 5.4 ± 1.2 µM) were bound more tightly to M184V than the (-)-analogs leading to relatively lower selectivity factors of 30 and 69, respectively. Of note, differences in observed kinetic parameters in this study versus previous reports 16,18,19 may be related to different reaction conditions and dsDNA substrates containing different DNA sequences and ends (blunt-end versus staggered-end).

L-nucleotide structures show two triphosphate conformations.

To understand the interaction of cytidine analog drugs with WT HIV-1 RT, crystal structures were determined by covalently cross-linking purified RT to an 18/26-mer dsDNA substrate (Methods) via an N 2 -cystamine-deoxyguanosine to Q258C present in the p66 subunit of RT 27 . The ternary complex determined with RT, DNA, and (-)-FTC-TP (RT-DNA•(-)-FTC-TP) displays good electron density for the NRTI and clearly shows the position of the oxathiolane ring and 5-fluoro cytidine base. Normal Watson-Crick base pairing is observed with the template guanine (Fig. 3a ). The (-)-β-L oxathiolane ring sits above Y115 (Fig. 4a ) and faces towards the last nucleobase of the primer strand. The sulfur atom within the ring points towards M184 and is within van der Waals distance to the side chain of M184 (4.0 Å). The phosphates are coordinated by one Mg 2+ ion and the side chains of R72, K65, and K220 (Fig. 4a ). The Mg 2+ interacts in a typical octahedral coordination with oxygen atoms from all three phosphates as well as the side chains of D110, D185, and the backbone carbonyl of V111. After the initial refinement of the structure it became apparent that the triphosphates were adopting an alternate conformation in addition to the canonical conformation associated with nucleotide binding (Fig. 4b ). A second conformation was simultaneously refined where the α-phosphate is flipped down into a nearly identical position as the β-phosphate in the first conformation described above (Fig. 3a ). This brings the α-phosphate further away from the primer 3′-terminal nucleotide, likely preventing proper orientation for incorporation into DNA (Fig. 4c ). The β-phosphate is flipped up (relative to the first conformation) and the γ-phosphorous atom is in a comparable position in both conformations. We refer to the first as a catalytically competent and productive conformation primed for incorporation and the second conformation as catalytically ineffective, or non-productive. After structure refinement and occupancy optimization, a split of 55%/45% for productive/non-productive conformation resulted in lower Rvalues (Table 2 ).

The net effect of the non-productive conformation is that the α-phosphate is shifted away from the primer strand, compared to the productive conformation (Fig. 4c ). The pivot point for this rotation is through the C5′ carbon that connects the oxathiolane and triphosphates. In the productive conformation, the distance from the C3ʹ in the deoxyribose ring of the last priming nucleotide and the α-phosphate of (-)-FTC-TP is 4.8 Å, while in the second conformation the distance is 7.9 Å. This shift of 3.1 Å of the α-phosphate would likely prevent the proper alignment for nucleophilic attack by the primer 3ʹ-OH and thus, slow the incorporation of (-)-FTC-TP into the DNA primer strand.

Residue R72 has been noted to be involved with proper positioning of the α-phosphate for incorporation and stabilizing the transition state 28 . In this ternary structure, R72 lies across the face of the cytidine base in a near parallel fashion and forms a hydrogen bond with an oxygen directly connected to the αphosphate (Fig. 4a ). The guanidinium group forms a hydrogen bond with Q151 through Nε to help position the side chain of Q151. Due to the positioning of the R72 side chain in the active site, it appears to form a π-stacking interaction with the face of the cytidine base. For (-)-FTC-TP, the 5-fluoro atom in the cytidine ring comes relatively close (3.2 Å) to the R72 nitrogen Nη (Fig. 4a ). Since the angle between the fluorine and the nitrogen is not ideal for a hydrogen bond and 5-fluoro is a weak hydrogen bond acceptor, the 5-fluoro likely interacts with R72 through an ion-dipole.

The ternary structure RT, DNA, and (-)-3TC-TP (RT-DNA• (-)-3TC-TP) was also crystallized and solved (Fig. 3b ). Overall, the binding mode is very similar to the one in RT-DNA•(-)-FTC-TP (Fig. 5a ). There is a normal Watson-Crick base pair formed with the template guanine and two conformations are observed for the triphosphates. R72 lies across the cytosine base in the same position as with RT-DNA•(-)-FTC-TP, forming a π-stacking interaction with the base. Since (-)-3TC-TP lacks a 5-fluoro atom in the cytidine base, there is no possibility of forming an ion-dipole with R72. The lack of this interaction is potentially why there is a 2.5-fold higher K d value (or a 2.5-fold lower binding affinity) of (-)-3TC-TP over (-)-FTC-TP and 3.7-fold greater incorporation efficiency observed for (-)-FTC-TP over (-)-3TC-TP (Table 1 ).

(+)-FTC-TP possesses a similar binding conformation as dCTP. In order to determine the effect of the oxathiolane ring ribose mimic in the context of a D-nucleoside, the crystal structure of RT, DNA, and (+)-FTC-TP (RT-DNA•(+)-FTC-TP) was determined. The defined electron density clearly showed the different positions of the oxathiolane atoms between (+)-FTC-TP and (-)-FTC-TP (Figs. 3c and 5b ). The position of the oxygen in the oxathiolane ring of the (+) isomer points towards the primer 3′terminus and the sulfur is positioned near the phosphates (Fig. 3c ). There was only one conformation of the triphosphates apparent in the electron density, similar to dCTP (see below) and therefore a productive conformation for incorporation (Fig. 3c, d ). The correct positioning of the phosphates appears to be aided by the oxathiolane sulfur mimicking the position of C3ʹ of deoxyribose. An apparent effect of the 3ʹ-sulfur is steric which prevents the α-phosphate from flipping away from the DNA primer into a non-production conformation. Additionally, the 3ʹ-sulfur sits closely (3.3 Å) to the bridging oxygen between αand β-phosphates (Fig. 3c ) which is less than the van der Waals radii of a sulfur and oxygen (1.8 Å + 1.52 Å = 3.32 Å). Our pre-steady-state kinetic parameters for the incorporation of (+)-FTC-TP and (-)-FTC-TP WT RT show similar efficiency (k p /K d ) values for the two analogs (Table 1 ). However, the k p for (+)-FTC-TP incorporation is improved by 2.4-fold relative to (-)-FTC-TP, possibly due to the phosphates in (+)-FTC-TP being locked into the productive conformation.

In order to compare how the cytosine based NRTIs bind in relation to the natural substrate, the structure of the ternary complex of RT, DNA and dCTP (RT-DNA•dCTP) was determined (Fig. 3d ). The binding conformation of dCTP is equivalent to previously reported crystal structures with natural substrates, dTTP (PDB code 1RTD) 29 and dATP (PDB code 3KK2) 27 . The triphosphates interact with a Mg 2+ ion with octahedral coordination, as well as R72, K65, and K220. To further help position the phosphates, dCTP forms an intermolecular hydrogen bond (2.8 Å) between its 3ʹ-OH and an oxygen atom from the β-phosphate, thus stabilizing the triphosphate conformation (Fig. 5c ). Comparing the central ring, it is striking that the C1ʹ and C4ʹ have nearly perfect alignment between (-)-FTC-TP and dCTP (Fig. 5c ). The productive triphosphate binding conformation for RT-DNA•(-)-FTC-TP matches well with the triphosphate conformation in RT-DNA•dCTP. Strikingly, the M184 side chain exists as different rotamers in the WT structures. Although M184 is not in direct contact with the central ring of dCTP or NRTI substrates described here, the orientation of the Cγ atom shifts depending on the incoming nucleotide. Cγ is positioned closer to dCTP and is shifted away from the oxathiolane ring indicating its sensitivity to dCTP vs. a NRTI in WT RT (Fig. 5c ).

Mechanistic basis of M184V resistance to L-nucleotide analogs. The M184V mutation was introduced into both the p66 and p51 subunits of RT (Methods) and the crystal structure of the binary complex of RT M184V and DNA (M184V-DNA) was solved (Fig. 6a ). The electron density for the valine residue was clearly visible and the atom positions were assigned for the side chain in both subunits. Comparing M184V-DNA to RT-DNA (PDB code 3KJV) in the same crystallography system, i.e. space group and DNA sequence, the V184 is orientated in the P-site under the deoxyribose ring of the primer-terminal base and shows no apparent change to the conformation of other amino acids in the polymerase active site (Fig. 6a ). Alignment of RT-DNA and M184V-DNA structures resulted in an RMSD of 0.39 Å. There appears to be a small shift in the primer 3′-terminal nucleotide in response to the mutation which was also observed in the previous binary structure of M184I RT bound to DNA 30 .

As with the above-mentioned WT structures, the same procedure (Methods) was used to generate the ternary structure of M184V, DNA, and (-)-FTC-TP (M184V-DNA•(-)-FTC-TP) (Fig. 3e ). The resolution (2.75 Å, Table 2 ) and quality are comparable between the structures in Fig. 3 . Overall, the positions of the cytidine and oxathiolane are similar as in the RT-DNA• (-)-FTC-TP structure (Fig. 7b ). There is only one conformation of the triphosphates observed (Fig. 7a ) which matches the nonproductive conformation observed in RT-DNA•(-)-FTC-TP and RT-DNA•(-)-3TC-TP. As with the RT-DNA•(-)-FTC-TP, there is a Mg 2+ ion with an octahedral ligand coordination sphere observed; however the Mg 2+ interacts with the αand γ-phosphates and not the β-phosphate (Fig. 7a ). Because the βphosphate is shifted and not interacting with the Mg 2+ , a water molecule was observed completing the coordination sphere (Fig. 7a ). Even though Mg 2+ has very stringent requirements for a coordination sphere (i.e. lengths and angles), it is still able coordinate the triphosphates in this non-productive conformation. The lack of change in the Mg 2+ location is perhaps due to the protein atoms not moving, which provides an anchor point to coordinate the flexible triphosphates.

Examining the van der Waals radius of V184, it is clear that there is a direct interaction with the branched side chain and the sulfur in the oxathiolane ring of (-)-FTC-TP (Fig. 7d ). For the RT-DNA•(-)-FTC-TP complex, the closest distance of the oxathiolane ring to M184 was 4.1 Å, compared to 3.4 Å for the valine side chain (Fig. 7c ). Structure alignment through residues 107-112 and 151-215 in the p66 subunit between the WT and M184V structures with (-)-FTC-TP shows that the oxathiolane sulfur is shifted away from residue V184 by 0.3 Å compared to M184. A concerted movement is observed in (-)-FTC-TP that results in the C5´carbon, which is the pivot point of the different triphosphate orientations, further shifting by 0.6 Å. This perhaps has a larger effect on the triphosphate orientation. Although the calculated maximum likelihood coordinate error reported in Phenix refinement is 0.33 Å for the WT and M184V (-)-FTC-TP structures, this shift is propagated to cause other atoms of (-)-FTC-TP to move more than 0.33 Å from the WT structure. The combination of a steric shift and altered triphosphate orientation likely results in reduced binding affinity of (-)-FTC-TP (280-fold higher K d in Table 1 ) with M184V than with WT RT.

As with WT RT, the crystal structure of M184V-DNA bound to dCTP (M184V-DNA•dCTP) was determined (Fig. 3f ). The overall binding conformation is nearly identical for RT-DNA•dCTP and M184V-DNA•dCTP (Fig. 6b ), and there is no apparent steric clash with the valine side chain. Our pre-steady-state kinetic data support that M184V does not discriminate against dCTP based on similar kinetic parameters measured for both WT and M184V (Table 1 ). Only the productive triphosphate binding conformation of (-)-FTC-TP is shown for simplicity. The unnatural (-)-β-L oxathiolane ring points back towards the last base in the DNA primer strand. There is a Mg 2+ ion that forms the typical octahedral coordination (yellow dashed lines) with the triphosphates and active site residues of RT. b Fo-Fc map (green mesh) drawn at +4.5σ. Calculated by refining either the productive or non-productive triphosphate conformation independently. The lines show how the alternate conformation fills out the extra e -density. c The productive conformation brings the αphosphate to 4.8 Å (black dashed lines) from the 3ʹ-carbon of the primer 3′-terminal nucleotide while the α-phosphate in the non-productive conformation is 7.9 Å away.

Analysis of various RT-DNA ternary complexes in the PDB 27,29,31,32 reveals a wide range of M184 conformations. Similarly, with our solved WT structures of (-)-FTC-TP, (-)-3TC-TP, (+)-FTC-TP, the methionine shifts away from the oxathiolane sulfur compared to dCTP (Fig. 5c ). Remarkably, the γC of M184 in the RT-DNA•dCTP structure mimics the branched carbon of valine, suggesting that it has some role in stabilizing dCTP binding (Fig. 6b ). M184 shifts upon NRTI binding, whereas V184 remains static and keeps (-)-FTC-TP or (-)-3TC-TP in a non-productive conformation, thus likely preventing their tight binding and efficient incorporation as quantified through our presteady-state kinetic parameters (Table 1 ).

Discussion

Often the best candidate for drug development balances important properties such as potency, metabolic stability, and limited offtarget toxicity. FTC and 3TC represent a distinct class of nucleoside antiviral drugs which contain an unnatural (-)-β-L stereochemistry for a deoxyribose sugar analog. The fact that HIV-1 RT and HBV polymerase can recognize these drugs and effectively incorporate them is remarkable. Likewise, the fact that human kinases can utilize the free hydroxyl (prodrug) form to convert them into the active triphosphate metabolite is equally surprising. The crystal structures of RT-DNA•(-)-FTC-TP and RT-DNA•(-)-3TC-TP presented here provide a structural basis for binding of the (-)-β-L oxathiolane ring to HIV-1 RT. The L configuration is well tolerated by RT with no steric clash from protein residues or DNA to effectively discriminate against it. There are two distinct conformations of the incoming triphosphates of (-)-FTC-TP and (-)-3TC-TP. One conformation placed the triphosphate chain in an orientation consistent with RT-DNA•dCTP and was presumably a productive state for incorporation. The other nonproductive conformation of the triphosphate chain is possibly due to a lack of steric hindrance afforded by the part of the oxathiolane ring furthest from the primer, or the removal of an intramolecular 3´OH to β-phosphate bond that is found in RT-bound natural dNTPs. The tighter binding of the oxathiolane ring can also be attributed to greater hydrophobic contact with Y115 due to the larger sulfur atom (Fig. 4a ).

M184V is the primary resistance mutation generated during HIV treatment with FTC and 3TC [6][7][8] . Generally, M184I precedes the M184V mutation and can arise through a single nucleotide change from G to A. Hypermutation at this position has been reported for HIV 33 . However, M184V requires a two nucleotide change and in some cases has been shown to mutate directly from WT and not through a progression of I to V 10,33 . M184V eventually outcompetes M184I based on higher RT polymerase activity and better processivity 34 . The M184V mutant confers some of the highest resistance as a single mutation against an NRTI with up to >300-fold reduction in potency in enzyme-based assays and >500fold reduction in cell-based assays for FTC and 3TC 7,18 .

A prior binary crystal structure of M184I RT-DNA predicted a steric clash with the oxathiolane ring. Additionally it was noted that there was movement of the primer terminus for which M184I or V is directly below 30 . Thus, it was hypothesized that the movement of the primer terminus to a less reactive position was the cause of the reduced reactivity of the mutant toward (-)-FTC-TP and the cause of the resistance. In the M184V binary structure reported here, the position of the primer terminus does not appear to have a large effect on the mechanism of resistance. In the binary structure of M184V-DNA, there is a shift in the DNA primer compared to RT-DNA (Fig. 6a ). However, in both M184V-DNA•(-)-FTC-TP and M184V-DNA•dCTP the primer shifts back to a position comparable to WT (Fig. 3 ). Therefore it appears unlikely that improper positioning of the primer terminus has an effect in resistance to NRTIs.

The WT structures showed two distinct positions of the Cγ atom of M184. For RT-DNA•dCTP, Cγ is positioned closer to the ribose ring. For RT-DNA•(-)-FTC-TP, RT-DNA•(-)-3TC-TP, and RT-DNA•(+)-FTC-TP the Cγ atom is flipped away from the oxathiolane ring (Fig. 5a-c ). Comparing the position of Cγ in RT-DNA•(-)-FTC-TP to RT-DNA•dCTP, the atom moves 1.8 Å (Fig. 5c ). Comparatively, the Cγ1 and Cγ2 atoms of V184 occupy very similar positions to the two rotamer positions of M184 Cγ in the WT structures (Fig. 6b ). Valine therefore reduces the conformational flexibility apparent with M184 which aids in the binding of these NRTIs to WT RT. The net effect of the M184V mutation produces a steric hindrance by directly contacting the sulfur atom in the oxathiolane ring (Fig. 7d ). During (-)-FTC-TP and (-)-3TC-TP incorporation by WT RT, their triphosphates are found in productive and nonproductive conformations (Fig. 5a ); however, the M184V mutation shifts the conformation equilibrium towards the catalytically ineffective state (Fig. 7c ). This is consistent with measured K d of nucleotide binding which increased from 0.10 and 0.25 µM with WT HIV-1 RT to 28 and 53 µM with M184V for (-)-FTC-TP and (-)-3TC-TP, respectively, and with measured substrate specificity (k p /K d ) which decreased from 1.0 and 0.27 µM -1 s -1 to 2.6 × 10 -3 and 9.4 × 10 -4 µM -1 s -1 for (-)-FTC-TP and (-)-3TC-TP, respectively (Table 1 ). Taken together, our pre-steady-state kinetic data in combination with our crystal structures support the mechanism whereby the M184V mutation confers resistance via primarily affecting the binding and proper orientation of (-)-FTC-TP and (-)-3TC-TP within the active site of HIV-1 RT. Recently, ternary crystal structures were reported of human mitochondrial DNA polymerase γ, DNA, and FTC-TP or 3TC-TP 21 . Based on a structural comparison to HIV-1 RT, it was postulated that there could be a hydrogen bond formed between R72 in RT and the 5-fluoro of (-)-FTC-TP. The structures reported here, demonstrate a likely interaction between the 5fluoro and R72; however, the contact appears to be best characterized as an ion-dipole interaction. The angle between the guanidinium nitrogen of R72 and 5-fluoro is not ideal for hydrogen bond formation. Consistently, our pre-steady-state kinetic data indicate a 2.5-fold tighter binding for (-)-FTC-TP (K d = 0.10 µM) than (-)-3TC-TP (K d = 0.25 µM, Table 1 ), likely caused by the ion-dipole interaction between R72 and the 5fluoro of (-)-FTC-TP. Interestingly, R72 is one of the few invariant residues found in RT which does not mutate in response to drug treatment 35 . In the case of HIV-1 RT, R72 plays a role in tighter binding of (-)-FTC-TP over (-)-3TC-TP.

Other crystal structures of (-)-FTC-TP have been reported in ternary complexes with human DNA polymerases λ (Polλ) 20 and β (Polβ) 26 as well as S. solfataricus Dpo4 25 . Both Polλ and Polβ are involved with DNA repair and are implicated in off-target toxicity by NRTIs 23 . Polλ was crystallized with a single nucleotide gapped DNA substrate and (-)-FTC-TP (PDB code 4K4I). Interestingly, when bound in the active site of Polλ (Complexes A and E in the crystal structure), (-)-FTC-TP exhibits a similar binding conformation as when bound to RT. For the other two complexes of Polλ in the asymmetric unit, R517 surprisingly used its side chain to coordinate the cytidine base through two hydrogen bonds 20 . No comparable residue exists in RT and only normal Watson-Crick base pairing of cytosine to guanine was observed with RT (Fig. 3 ). Unexpectedly, the pre-catalytic ternary structure of Polβ, a single nucleotide gapped DNA substrate, and (-)-FTC-TP (PDB code 5U2T) shows that Polβ, a sequence and structure homolog of Polλ, bound (-)-FTC-TP with Watson-Crick base pairs and productive triphosphate conformations, but with accumulation of several active site rearrangements that led to decreased nucleotide binding affinity and incorporation rate 26 . We also reported the ternary structure of Dpo4, dsDNA, and (-)-FTC-PPNP, a non-hydrolyzable triphosphate analog where a nitrogen replaced the bridging oxygen between the βand γ-phosphates (PDB code 4QW9). Multiple conformations of the phosphates were also observed in this Dpo4 ternary structure 25 which is consistent with the observed triphosphate conformations observed in our RT-DNA•(-)-FTC-TP and RT-DNA•(-)-3TC-TP structures reported in Fig. 5a . Taken together, the aforementioned DNA polymerases and HIV-1 RT bind and incorporate (-)-FTC-PPNP through unique structural mechanisms 26 .

In addition to FTC and 3TC, other drugs with similar stereoisomer chemistry have progressed into human clinical trials. Racivir was developed as a 50/50 racemic mixture of (+)-FTC and (-)-FTC and was tested in Phase II clinical trials (NCT00121979). It was reported that (+)-FTC selects for T215Y as opposed to M184V in cell culture 36 . Although M184V still conferred 30-fold selection against (+)-FTC-TP (Table 1 ), it seems likely that (+)-FTC may not choose for this mutation since the oxathiolane ring is turned away from M184 (Fig. 3c ). Telbivudine, the L-nucleoside analog of thymidine, has been successfully developed to treat HBV 37 . Reportedly, this drug does not inhibit HIV 37 . Apricitabine (AVX754, SPD754) is also a cytidine analog containing an oxathiolane ring. This oxathiolane has natural D stereochemistry but the position of the oxygen and sulfur atoms are switched in comparison to (+)-FTC and (+)-3TC 38 . This drug reportedly entered Phase IIb/III clinical trials but further development was halted (NCT00612898).

Nucleoside analogs are powerful antiviral drugs and the development of new NRTIs are constantly needed to combat the emergence of drug resistance in addition to the threat of emerging epidemics such as the Zika virus outbreak in 2016. Understanding the basis of resistance that arises from viral mutation is paramount for developing efficacious drugs therapies. In the case of (-)-FTC-TP and (-)-3TC-TP, M184V causes resistance by sterically hindering the oxathiolane ring and pushing the phosphates away from the primer strand DNA, thus locking them in a non-productive conformation (Fig. 7 ). One design strategy to avoid improper phosphate positioning would be to adjust the steric bulk around the central ring to prevent the phosphates from flipping to a non-productive state, thus maintaining the flexibility of the methionine at position 184. This is akin to the (+)-β-D stereochemistry which only allows the phosphates to exist in a productive conformation (Fig. 3c ). This could potentially be achieved with substitutions to the oxathiolane ring that rigidify the ring but would still be accepted in the NRTI binding pocket. In addition, further optimization of NRTI interactions with R72 could increase the potency against HIV-1 RT as this residue is essential for polymerase function and not readily mutable. Identifying analogous residues to R72 in other viral polymerases could also be a fruitful avenue for improving the efficacy of drug therapies against selected viruses.

FTC and 3TC have been key components of HAART and HIV treatment. The structures reported here reveal how HIV-1 RT recognizes the unnatural (-)-β-L stereochemistry. In addition, our crystal structures of HIV-1 RT M184V show how this mutation could potentially confer resistance to (-)-FTC-TP and (-)-3TC-TP. Furthermore, our combined structural and kinetic analysis will aid in rational drug design to combat the HIV pandemic.

Methods

Expression and purification of the RT-DNA complex. HIV-1 RT p66 and p51 subunits were expressed and purified as described previously 27,39 . Site-directed mutagenesis to generate the M184V the mutation in both p66 and p51 subunits was performed using the QuickChange II kit (Stratagene, La Jolla, CA) according to manufacturer's protocol. The primer and template oligonucleotides were purchased from TriLink Biotechnologies (San Diego, CA) and resuspended in 1 mM HEPES pH 7.5. The 26-mer DNA template (5ʹ-ATGGGGGGCGCCCGAACAGGGACT GT-3ʹ) was annealed to a dideoxy terminated 18-mer nucleotide primer (5ʹ-GTCC CTGTTCGGXCGCCC dd -3ʹ). The X in the primer sequence represents N 2 -cystamine-deoxyguanosine which covalently cross-links to Q258C in the p66 subunit. The method of covalently tethering HIV-1 RT dsDNA to generate the RT-DNA complex has been formally described 27,29,40 . To form the covalent complex, RT and dsDNA substrate were incubated together at 25 µM and 50 µM, respectively at room temperature for approximately 18 h. The progress of the tethering reaction was monitored by the observation of an increase in the molecular weight of the p66 subunit as judged by non-reducing SDS-PAGE. Covalently linked RT-DNA complexes were purified 27 and mass spectrometry performed to analyze the purified sample confirming that a homogeneous heterodimer consisting of p66 tethered to DNA and p51 subunit existed.

Measurement of pre-steady-state kinetic parameters. HIV-1 RT (120 nM) and 5ʹ-[ 32 P]-labeled 18/26-mer DNA (30 nM; DNA sequence shown above the 18mer primer without chemical modification) were pre-incubated at 37 °C for 5 min in reaction buffer containing 50 mM Tris-HCl (pH, 7.9), 10 mM MgCl 2 , 50 mM NaCl, 0.1 mM EDTA, 10% glycerol, 5 mM DTT, and 0.1 µg/mL BSA (all concentrations are final upon mixing). The reactions were initiated by rapidly mixing with varying concentrations of dCTP, FTC-TP, or 3TC-TP on a rapid chemicalquench flow apparatus (KinTek) and quenched with 0.37 M EDTA at various times. The DNA products were analyzed by sequencing gel electrophoresis (17% polyacrylamide, 8 M Urea), and quantitated using a Typhoon Trio (GE Healthcare) and ImageQuant software (Molecular Dynamics). The kinetic data were fit using non-linear regression software KaleidaGraph (Synergy) to a singleexponential equation

where A and k obs represent the reaction amplitude and observed nucleotide incorporation rate constant, respectively. Values for k obs were then plotted against the respective dNTP (or an NRTI triphosphate) concentration and each plot was fit to a hyperbolic equation

where k p is the maximal nucleotide incorporation rate constant and K d is the apparent equilibrium dissociation constant for the binding of dNTP (or an NRTI triphosphate) to the RT and DNA complex. Each set of kinetic parameters (k p , K d , k p /K d ) are derived from 8 individual experiments where the concentration of nucleotide is variable (i.e. 8 distinct concentrations of dCTP or NRTI). All reported error values are derived from data fitting using non-linear regression software KaleidaGraph (Synergy). For kinetic experiments, we did not cross-link RT and DNA.

Crystallization and data collection. Crystals of the binary WT and M184V RT-DNA complex were grown by hanging drop vapor diffusion over a mother liquor containing 2-4% PEG 4000, 100 mM MES (pH = 6.0), and 10 mM MgSO 4 at either 4 or 20 °C. Equal parts of protein and reservoir solution were mixed to produce 4 µl drops. To obtain ternary complexes with FTC-TP, 3TC-TP or dCTP, binary RT-DNA crystals were placed into a mother liquor of 36% PEG 4000, 6% glycerol, 100 mM MES (pH = 6.0), and 10 mM MgSO 4 with 0.5 mM of the NRTI triphopshate, or dCTP for 5-18 h. Prior to data collection, crystals were moved from the above buffer and flash-cooled in a bath of liquid nitrogen. All X-ray diffraction data were collected at The Advanced Light Source (Table 2 ) at a temperature of 100 K and processed with HKL2000 41 or XDS 42 .

Structure determination and refinement. Molecular replacement was performed by the refinement package Phenix 43 using the starting model PDB code 3KK1 27 (ternary complex) for WT and M184V structures. The molecular replacement for the structure of binary M184V RT-DNA was determined using PDB code 3KJV 27 (binary complex) as the starting model. Rigid body refinement, simulated annealing, energy minimization, and B-factor refinement were performed with Phenix. Model building was carried out by the molecular graphics program Coot 44 .

Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

Atomic coordinates and structure factors for the reported crystal structures have been deposited with the Protein Data Bank. Accession numbers are 6UJX, 6UJY, 6UJZ, 6UIT, 6UKO, 6UIR, and 6UIS for WT RT-DNA with (-)-FTC-TP, (-)-3TC-TP, (+)-FTC-TP, and dCTP, binary M184V RT-DNA, M184V RT-DNA with (-)-FTC-TP, and M184V RT-DNA with dCTP, respectively.

Author contributions

M.H. and L.L. cloned, expressed, and purified RT proteins. E.J.T. and Z.S. designed presteady state kinetic experiments while E.J.T. performed them. E.B.L. crystallized, collected X-ray data, and refined the crystal structures. L.Z. synthetically prepared (+)-FTC-TP. E.B.L, Z.S., M.H. and E.J.T. contributed to writing the paper.

Competing interests

M.H., L.L., L.Z. and E.B.L. are or were employed by Gilead Sciences at the time of data generation. E.J.T. and Z.S. declare no competing interests.

Additional information

Supplementary information is available for this paper at https://doi.org/10.1038/s42003-019-0706-x.

Correspondence and requests for materials should be addressed to Z.S. or E.B.L.

Reprints and permission information is available at http://www.nature.com/reprints Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Acknowledgements

AcknowledgementsThis work was supported by Gilead Sciences, Inc. and a grant from the National Science Foundation (grant number MCB-1716168) to Z.S.The authors would like to thank the staff at The Advanced Light Source for help in data collection.The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.The authors would also like to thank Richard Mackman and Jason Perry (Gilead Sciences) with critical review of the manuscript.

Failed to load PDF:

StripeM-Inner

Paper sources

Abstract screening pilot

Abstract screening results

Extraction pilot

Extraction results

Research report

Modify setup