Correction

PHARMACOLOGYCorrection for “From in silico hit to long-acting late-stage pre-clinical candidate to combat HIV-1 infection,” by Shalley N.Kudalkar, Jagadish Beloor, Elias Quijano, Krasimir A. Spasov,Won-Gil Lee, José A. Cisneros, W. Mark Saltzman, Priti Kumar,William L. Jorgensen, and Karen S. Anderson, which was firstpublished December 26, 2017; 10.1073/pnas.1717932115 (ProcNatl Acad Sci USA 115:E802–E811).The authors would like to note that the conflict of interest

statement should state, “W.L.J. and K.S.A. are listed as the inven-tors on a patent issued to Yale University on the intellectualproperty associated with compound I [US Patent 9,382,245 (2016)].”

Published under the PNAS license.

Published online March 12, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1803199115

www.pnas.org PNAS | March 20, 2018 | vol. 115 | no. 12 | E2899

CORR

ECTION

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

22, 2

020

From in silico hit to long-acting late-stage preclinicalcandidate to combat HIV-1 infectionShalley N. Kudalkara,b,1, Jagadish Beloorc,1, Elias Quijanod, Krasimir A. Spasova,b, Won-Gil Leee, José A. Cisnerose,W. Mark Saltzmand,2, Priti Kumarc,2, William L. Jorgensene,2, and Karen S. Andersona,b,2

aDepartment of Pharmacology, Yale University School of Medicine, New Haven, CT 06520-8066; bDepartment of Molecular Biophysics and Biochemistry,Yale University School of Medicine, New Haven, CT 06520-8066; cDepartment of Internal Medicine, Section of Infectious Diseases, Yale University School ofMedicine, New Haven, CT 06520; dDepartment of Biomedical Engineering, Yale University, New Haven, CT 06511; and eDepartment of Chemistry, YaleUniversity, New Haven, CT 06520-8107

Contributed by William L. Jorgensen, November 17, 2017 (sent for review October 12, 2017; reviewed by Katherine Seley-Radtke and Nicolas Sluis-Cremer)

The HIV-1 pandemic affecting over 37 million people worldwidecontinues, with nearly one-half of the infected population on highlyactive antiretroviral therapy (HAART). Major therapeutic challengesremain because of the emergence of drug-resistant HIV-1 strains,limitations because of safety and toxicity with current HIV-1 drugs,and patient compliance for lifelong, daily treatment regimens. Non-nucleoside reverse transcriptase inhibitors (NNRTIs) that target theviral polymerase have been a key component of the current HIV-1 combination drug regimens; however, these issues hamper them.Thus, the development of novel more effective NNRTIs as anti–HIV-1agents with fewer long-term liabilities, efficacy on new drug-resistantHIV-1 strains, and less frequent dosing is crucial. Using a computa-tional and structure-based design strategy to guide lead optimization,a 5 μM virtual screening hit was transformed to a series of verypotent nanomolar to picomolar catechol diethers. One representative,compound I, was shown to have nanomolar activity in HIV-1–infectedT cells, potency on clinically relevant HIV-1 drug-resistant strains, lackof cytotoxicity and off-target effects, and excellent in vivo pharma-cokinetic behavior. In this report, we show the feasibility of com-pound I as a late-stage preclinical candidate by establishingsynergistic antiviral activity with existing HIV-1 drugs and clinical can-didates and efficacy in HIV-1–infected humanized [human peripheralblood lymphocyte (Hu-PBL)] mice by completely suppressing viralloads and preventing human CD4+ T-cell loss. Moreover, a long-acting nanoformulation of compound I [compound I nanoparticle(compound I-NP)] in poly(lactide-coglycolide) (PLGA) was developedthat shows sustained maintenance of plasma drug concentrationsand drug efficacy for almost 3 weeks after a single dose.

HIV-1 | NNRTI | nanoparticle | humanized mice | drug synergy

HIV-1/AIDS remains a major challenge as a world healthproblem, especially in developing countries, with an esti-

mated 37 million people infected with the virus according to theJoint United Nations Programme on HIV/AIDS (UNAIDS)report in 2016 (1). While there were 2.1 million new infections in2015, the number of deaths has declined, largely attributableto highly active antiretroviral therapy (HAART). HAART hasproven to be effective in suppressing viral replication for the life-time of an individual (2, 3), and the number of people on this typeregimen has risen to 17 million. The pandemic will likely continuefor many years to come in light of the number of individuals cur-rently infected, high rate of new infections, emergence of drug-resistant HIV-1 strains, and issues associated with access to therapy,chronic treatment, and patient compliance. For these reasons, thedevelopment of novel more effective anti–HIV-1 agents with fewerlong-term liabilities and efficacy on new drug-resistant HIV-1 strainsis crucial. Moreover, since therapy remains lifelong, improved thera-peutic modalities that require less frequent dosing than the currentdaily regimen would be highly desirable.The HAART regimen typically includes a combination of three

or more antiviral drugs, which target different steps of the HIV-1 replication cycle (4). Almost 20 y ago, nonnucleoside reverse

transcriptase inhibitors (NNRTIs) became part of this regimenalongside nucleoside reverse transcriptase inhibitors (NRTIs).NNRTIs significantly inhibit the catalytic function of HIV reversetranscriptase (RT) by binding in an allosteric binding pocket 10 Åaway from the RT polymerase active site and inhibiting the po-lymerization reaction (5–7). Currently, two of the five Food andDrug Administration (FDA)-approved NNRTIs, efavirenz andrilpivirine, are key components of the single-tablet regimens (STRs);Atripla and Complera/Odefsey (8–10). Although NNRTIs are keycomponents of effective HAART regimens, their effectiveness ishampered by their poor pharmacokinetic properties, dose-limitingtoxicities, and requirement of chronic treatment, which contributeto noncompliance and increasing chances of developing resistancemutations. For instance, the most recently approved NNRTI, rilpi-virine, is effective on a number of the most common drug-resistantHIV-1 strains; however, the dose-limiting cardiotoxicity caused byinhibition of the hERG (the human Ether-à-go-go-Related Gene)-ion channel (11, 12) restricts its use to patients with viral loads of<100,000 copies. The lower dose of rilpivirine necessary to avoidcardiac-related side effects resulted in higher rate of virologicfailure relative to efavirenz in several recent clinical trials (13).Over the past several years, our research has focused on de-

veloping improved NNRTIs. Our strategy combines state of the

Significance

Nonnucleoside reverse transcriptase inhibitors (NNRTIs) are es-sential components of highly active antiretroviral therapy; how-ever, concerns about poor pharmacological properties, doserestriction because of toxicity, and drug resistance have limitedtreatment options. Our computational and structure-guided de-sign studies for lead optimization have transformed a 5 μM vir-tual screening hit into a class of NNRTIs with remarkable potency,safety, drug resistance profile, and pharmacological properties.We report a representative, compound I, with marked synergywith existing HIV-1 drugs and antiviral efficacy in HIV-1–infectedhumanizedmice. A single dose of long-acting nanoformulation ofcompound I retains sustained levels and efficacy for ∼3 weeks,confirming potential as a late-stage preclinical candidate. Addi-tionally, these properties of compound I suggest that it may be apromising candidate to evaluate for preexposure prophylaxis.

Author contributions: W.M.S., P.K., W.L.J., and K.S.A. designed research; S.N.K., J.B., E.Q.,K.A.S., W.-G.L., J.A.C., and W.L.J. performed research; S.N.K., J.B., E.Q., K.A.S., W.-G.L.,J.A.C., and K.S.A. analyzed data; and S.N.K., J.B., E.Q., W.M.S., P.K., W.L.J., and K.S.A.wrote the paper.

Reviewers: K.S.-R., University of Maryland, Baltimore County; and N.S.-C., Universityof Pittsburgh.

The authors declare no conflict of interest.

Published under the PNAS license.

See Commentary on page 637.1S.N.K. and J.B. contributed equally to this work.2To whom correspondence may be addressed. Email: mark.saltzman@yale.edu, priti.kumar@yale.edu, william.jorgensen@yale.edu, or karen.anderson@yale.edu.

E802–E811 | PNAS | Published online December 26, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1717932115

art technology for in silico virtual screening/structure-based drugdesign, synthetic organic chemistry, mechanistic enzymology andprotein crystallography, and pharmacological assays (10–12). A keycomponent of this computationally guided approach involves theuse of free energy perturbation (FEP) calculations to speed leadoptimization. FEP-guided lead optimization allowed a 5 μM insilico hit to be transformed into a subnanomolar lead compound.Fig. 1 shows representative steps along the way for improving thediphenylmethane substructure, A, through functionalization to B,further elaboration to a diphenyl ether scaffold C, and additionalsubstituent changes to the difluoro-substituted diphenyl ether, D,which showed potent anti–HIV-1 activity against WT HIV-1–infected T cells. Additional modifications were required to removethe potentially problematic cyanovinyl group and to increase theeffectiveness of new compounds on the most common drug-resistantHIV-1 strains as illustrated by the naphthyl catechol phenyl ethercompound I. This effort was also enhanced by measurements ofaqueous solubility and use of the program QikProp, which informsthe selection of compounds with desirable absorption, distribution,metabolism, elimination, toxicity, and drug-like properties (14–17).This effort led to the discovery of a series of catechol diethers

that have remarkable potency, efficacy on clinically relevant drug-resistant strains of HIV-1, and optimal pharmacological proper-ties, including excellent pharmacokinetics and lack of cytotoxicityand off-target effects (18–20). One of the most promising com-pounds is compound I (Fig. 1).In this study, we describe several important milestones achieved

for designating a late-stage preclinical candidate, including (i)cellular studies showing potent synergy of the candidate compoundwith a number of the FDA-approved drugs and clinical candidatesfor HIV-1, (ii) the development of a long-acting nanoformulationof the compound that produces sustained serum levels for almost3 wk after a single dose, and (iii) demonstration of the efficacy forthe free compound and long-acting nanoparticle (NP) formulationin a humanized mouse model of HIV-1 infection.

ResultsSelection of a Late-Stage Preclinical Candidate. Our multifacetedresearch focused on designing improved NNRTIs led to thediscovery of a number of potent catechol diether compoundshaving excellent antiviral activity in human T cells infected withboth WT and drug-resistant strains of HIV-1.Synergy studies. Another consideration for a late-stage preclinicalcandidate would be an assessment of that compound’s behavior incombination with other FDA-approved HIV-1 drugs. A minimumrequirement would be that the compound should not be antago-nistic with other anti–HIV-1 drugs. Recent cellular studies of dor-avirine (MK-1439), an NNRTI by Merck in phase III clinical trialsfor treating HIV-1, showed that the compound was not antagonisticwhen evaluated for antiviral activity in combination with a numberof established AIDS drugs, but no synergy was observed (21). Ide-ally, the late-stage preclinical candidate compound should havesynergistic anti–HIV-1 effects when evaluated with the currentFDA-approved drugs, such as the NRTIs Emtricitabine (FTC) andTenofovir disoproxil fumarate (TDF) and integrase strand transferinhibitors (INSTIs), such as elvitegravir (EVG) (22, 23). In ourcatechol diether series of compounds, compound I (Fig. 1) from thisclass possessed potent antiviral activity on WT and drug-resistantHIV-1 strains at low nanomolar concentrations, better solubilityprofiles, and low ClogP [logarithm of a compound’s partition co-efficient between n-octanol and water log(coctanol/cwater)] valuescompared with rilpivirine (24, 25) and efavirenz (18, 19, 26). Thepotential of compound I to exhibit synergistic antiviral effectswas determined in HIV-1–infected MT-2 cells with a number ofcurrent FDA-approved drugs that are currently used as combina-tion therapy, including FTC, TDF, and EVG. We also evaluatedtwo NRTI clinical candidates 4′-ethynyl-2-fluoro-2′-deoxyadenosine(EFdA or MK-8591) and 4′-ethynyl-2,3-didehydrothymidine (Ed4T

or festinavir) that are currently under evaluation in phase I andphase II trials, respectively (27–29).The anti–HIV-1 activity of compound I was evaluated in two-

drug combination studies with TDF, FTC, Ed4T, and EFdA inHIV-1–infected MT-2 cells using the HIV-1 IIIB virus strain asdescribed previously (30). The EC50 values observed in our MT-2 cell assay for each single drug were 2.8, 85, 100, 2.6, 920, and1.9 nM for compound I, TDF, FTC, EVG, Ed4T, and EFdA,respectively. An isobologram analysis was conducted to accessthe potential for synergy. In this analysis, isobolograms are con-structed by determining the fractional changes in EC50 (FEC50s) oftwo anti–HIV-1 agents when used in combination in the MT-2 cellular assay (30). The values on the additivity line connectingFEC50 values of one represent an additive interaction betweenthe two compounds, while points markedly below or above theline represent synergy or antagonism, respectively. The combina-tions of compound I-TDF, compound I-FTC, compound I-EVG,compound I-Ed4T, and compound I- EFdA are shown in Fig. 2.Compound I showed marked synergy in each case.The combination inhibitory data were also analyzed using

a MacSynergy II [the work by Prichard and Shipman (31)], a3D model for statistical evaluation of combination assay. Theresulting surface plots of the data reflect the difference betweenthe experimental dose–response surface and the predicted ad-ditive surface. On a 3D model, a simple additive effect will resultin a horizontal plane at 0% inhibition, whereas a synergistic orantagonistic effect will render a peak or depression above orbelow the horizontal plane. Fig. 3 shows representative examplesof the combination study of compound I with TDF, FTC, EVG,Ed4T, and EFdA. The surface plots generated for each combinationshowed that the percentage of inhibition was above the horizontalplane, which is indicative of a synergistic effect. The overall per-formance of the assays was validated by the use of positive controlfor antagonism [2′,3′-didehydro-3′-deoxythymidine (d4T) in com-bination with ribavirin] (32). The results are summarized in Table1. Hence, the data from isobologram and MacSynergy II analysesestablish that synergistic interactions were observed within theconcentration ranges examined for antiviral efficacy between com-pound I, each of the FDA-approved HIV-1 drugs, and clinical can-didates used in this study. Since a new agent would likely be used asa combination therapy with existing HIV-1 drugs, the demonstra-tion of synergistic activity for compound I meets a key criterion fora late-stage preclinical candidate.

O

NF

O N NH

O

O

F O

O

F CN

N NH

O

O

F

O

O

N NH

O

O

ClO N NH

O

OCl

O N NH

O

OH3CCl

Cl

EC50 = ~5,000 nM EC50 = 380 nM EC50 = 14 nM

EC50 = 0.320 nMEC50 = 1.9 nM

A B C

DCompound I

Cl CN

Fig. 1. FEP-guided optimization of a docking hit. The representative steps forimproving the diphenylmethane substructure (A) through functionalization to(B) further elaboration to a diphenyl ether scaffold (C) followed by additionalsubstituent changes to the potent difluoro-substituted diphenyl ether (D) andfinally, the naphthyl catechol phenyl ether compound I are presented here.EC50 values mentioned here were calculated by the MT-2 T-cell assay using theHIV-1 IIIB virus strain as described in Materials and Methods. The results arefrom three experiments involving triplicate determination.

Kudalkar et al. PNAS | Published online December 26, 2017 | E803

PHARM

ACO

LOGY

PNASPL

US

SEECO

MMEN

TARY

Pharmacological properties. Additional prioritization involved anassessment of in vivo pharmacokinetic behavior as well as thepotential for cytotoxicity and off-target effects. In prior studies,the in vitro pharmacological profiling of compound I exhibitedno adverse reaction to major targets, such as the hERG channel(20), the blockade of which is responsible for major cardiac ar-rhythmias (11), and showed no inhibition of CYP3A4, which isthe major cytochrome P450 enzyme responsible for the metabo-lism of the majority of currently approved drugs (20). Moreover,pharmacokinetic studies in BALB/c mice established that com-pound I displayed a better pharmacokinetic profile compared withefavirenz in a similar study at comparable doses (20 mg/kg) (20,33). A single 20-mg/kg dose of compound I produced sufficientlevels in serum up to 48 h (Fig. 4A), which were approximatelythreefold higher than those previously reported for efavirenz atthe same dose (33). The longer serum residence time and higherserum concentration than required for the therapeutic range makecompound I amenable to a daily dose regimen. Furthermore, ahigher area under curve (AUC0-last) value and a much lower clear-ance (CL) rate (Table 2) compared with efavirenz suggested thatcompound I can achieve a higher volume of distribution and issubjected to slower metabolism, which is beneficial to maintainoptimal control of viral load (20).In evaluating compounds in vivo to assess pharmacokinetic

behavior and efficacy using mouse models to approximate whatmay be expected in humans, dosages chosen are typically ∼10-fold higher because of the much faster rate of murine metabo-lism (34). To offset concerns with higher metabolic rates inmurine models, a dosage of 100 mg/kg has been frequently selectedfor the evaluation of efficacy of antiretroviral agents in humanizedmouse models (35–37). In our study, a 100-mg/kg dose of com-pound I administered i.p. into BALB/c mice produced a 10-d se-rum residence time and showed no morbidity or mortality inBALB/c mice over that period (Fig. 4B, Inset, black circles). Apreliminary assessment in a cell-based HIV-1 assay was conductedto establish that sufficient levels of compound I were present in theserum to show antiviral activity. An aliquot of the serum from thepharmacokinetic study was taken at the peak serum concentration

(Cmax) and at the maximum drug concentration time point (4 h)and evaluated in a TZM-bl–based single-round infectivity cellularassay to prevent HIV-1 infection. The clinical HIV strain JR-CSFof HIV-1 was chosen, as it is commonly used to infect humanizedmice. As illustrated in Fig. 4C, in a series of dilutions, the ability toprevent HIV-1 infection is maintained with an IC50 value of ∼3.2 ±0.4 nM. Hence, compound I represented a good candidate forsustained administration in a humanized mouse model to evaluateefficacy.

A Long-Acting Nanoformulation of the Late-Stage Preclinical Candidate.As HAART remains lifelong for those individuals infected withHIV-1, decreasing dosing frequency would be another importantcriterion for a new HIV-1 therapy. Therefore, concurrent to ourstudies evaluating a free drug formulation, compound I was furtherdeveloped into a long-acting nanoformulation. The biodegradablepolymer, poly(lactide-coglycolide) (PLGA), was used to produce along-acting formulation of compound I entrapped in NPs, whichcan slowly release the compound. Particulate drug delivery systemsof PLGA have been approved by the FDA (38), leading to com-mercial products that reduce the frequency of dosage, alter bio-distribution, and improve pharmacokinetics.Development of a long-acting nanoformulation. Compound I was syn-thesized as described in Materials and Methods and used in thepreparation of compound I-loaded PLGA-based NPs (compoundI-NPs) using emulsion-solvent evaporation techniques (39). Com-pound I-NPs encapsulated in PLGA exhibited an average diameterof 370 ± 9.2 nm, a polydispersity index (PDI) of <0.2, and an av-erage negative zeta potential of −28 ± 0.1 mV (Table 3). Repre-sentative SEM images illustrated that compound I-NP exhibitedspherical shapes (Fig. 4D), and the size estimates from SEM wereconsistent with the size of compound I-NP reported by dynamiclight scattering (DLS). For our first batch, an initial drug loading(DL) of 10 wt % was used, which resulted in DL of 10.5 ± 0.9 wt %and encapsulation efficiency (EE) of 104 ± 26.2% (Table 3). Asecond batch of compound I-NPs was made with an initial DL of15 wt %, resulting in DL of 15.3 ± 1.2 wt % and EE of 95 ± 9.1%.

Fig. 2. Synergistic inhibition of HIV-1 replication by combinations of compound I with (A) TDF, (B) FTC, (C) EVG, (D) Ed4T, and (E) EFdA analyzed by theisobologram analysis. The red dotted lines from the upper left corners to the lower right corners indicate that the two drugs are additive, and the curvesbelow the lines indicate that drugs are synergistic. The results are from three experiments involving triplicate determination.

E804 | www.pnas.org/cgi/doi/10.1073/pnas.1717932115 Kudalkar et al.

Pharmacokinetics of compound I-NP in BALB/c mice. A detailed phar-macokinetic study of compound I-NP was carried out to deter-mine the dosing and sustainability of compound I-NP in BALB/cmice. A single dose of compound I-NP (20 or 125 mg/kg) fromthe first batch of NPs was injected into mice i.p., and bloodsamples were collected for a total of 35 d. Fig. 4A, Inset, solid redcircles compares the serum concentrations of compound I ad-ministered as NP with free drug (Fig. 4A, Inset, solid black circles)at a dose of 20 mg/kg. A substantially longer period of com-pound I presence for up to 35 d was observed in mice that re-ceived the nanoformulation, whereas compound I levels werebelow the limit of detection in serum by day 2 in mice receiving

free compound I. The serum drug concentration of 0.2 μg/mLfor compound I-NP–treated mice on day 35 was sub-stantially greater (>100-fold) than the EC50 for the WT HIV-1 in the MT-2 assay (18). The Cmax of compound I in NPs(3.9 ± 0.2 μg/mL, 4 h postinjection) was threefold lower thanthe Cmax of 11.8 μg/mL for free injected compound I (Fig. 4A,Inset). The pharmacokinetic parameters are summarized inTable 2. The AUC0-last (472.9 ± 20.9 μg/mL per hour) was ap-proximately two times higher and the CL (0.7 mL/min per ki-logram) was approximately two times lower than free compoundI (AUC0-last: 209.2 ± 23.1 μg/mL per hour and CL: 1.5 mL/minper kilogram).

Fig. 3. Synergistic inhibition of HIV-1 replication by combinations of compound Iwith (A) TDF, (B) FTC, (C) EVG, (D) Ed4T, and (E) EFdA analyzed byMacSynergy II3D plots. The peaks above the horizontal plane indicate that the drugs are synergistic. The results are from three experiments involving triplicate determination.

Table 1. Antiviral efficacy results for compound I in combination with FDA-approved and inclinical trial antiretroviral compounds in MT-2 cell assay

Compound class and compound*Mean synergy vol/antagonism

vol,† μM2 % (n = 3) Antiviral effects

NRTIsTDF 578.2/0 Strong synergyFTC 64.9/−0.84 Moderate synergyEFdA (MK-8591) 231.15/0 Strong synergyFestinavir (Ed4T) 481.37/−1.22 Strong synergy

INSTIEVG 318.89/0 Strong synergy

Positive controld4T/ribavirin 0/−529.8 Highly antagonistic

*The antiviral efficacy of compound I in combination with NNRTIs and INSTIs is performed in MTT {(3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyltetrazolium bromide)} assay.†The antiviral synergy plot (95%) datasets from n = 3 are combined, and MacSynergy II software is used tocalculate synergistic and antagonistic interactions.

Kudalkar et al. PNAS | Published online December 26, 2017 | E805

PHARM

ACO

LOGY

PNASPL

US

SEECO

MMEN

TARY

The serum concentration–time curve for compound I-NP atthe 125-mg/kg dose showed Cmax of 4.25 ± 0.6 μg/mL 4 h post-injection (Fig. 4B, Inset). Detectable levels of compound I wereobserved for up to 35 d with a concentration of 0.33 ± 0.12 μg/mLon day 35, which was much greater than the EC50 (19) and simi-lar to that obtained with the 20-mg/kg dose of free compoundI. Fig. 4B also compares the serum levels of mice injected withfree compound I at 100 mg/kg. Additional analyses of pharma-cokinetic data for free and nanoformulated compound I showedthat the AUC0-last (709.5 ± 43.9 μg/mL per hour) and total bodyCL (2.9 mL/min per kilogram) in mice receiving compound I-NPwere comparable with the AUC0-last (706 ± 185 μg/mL per hour)and CL (2.3 mL/min per kilogram) for mice injected with freecompound I (Table 2). Hence, the data suggest that nano-formulation of compound I provides a slow and sustained releaseof compound I in BALB/c mice. Moreover, serum concentra-tions of compound I were maintained well above the requiredtherapeutic levels for a prolonged period of time, even after a

single dose of nanoformulated compound I. These studies showthe potential of the compound I-NP as a promising, long-actingnanoformulation that should be further evaluated.Anti-HIV activity of compound I-NP in infected TZM-bl cells. Fig. 4Edepicts the inhibition of HIV-1 infection for compound I-NP. Theconcentrations of compound I-NP in serum samples collected afterdosing with the nanoformulation on days 14, 19, 22, 30, and35 were sufficient to achieve 100% inhibition of HIV-1 infection[0.1 multiplicity of infection (MOI)] in TZM-bl cells at a serumvolume of 10 μL (Fig. 4E, solid blue circles). A complete inhibitionof HIV-1 infection was also observed with free compound at 4 hpostinfection (Fig. 4E, black circle). These data suggest that asingle administration of compound I-NP may be able to providesimilar protection for up to 35 d, which was comparable with thatobserved after 4 h in mice given free compound I. Collectively, thesynergy studies, pharmacokinetic and pharmacological evaluations,and development of a long-acting nanoformulation with compoundI fulfill important criteria for a late-stage preclinical candidate and

Table 2. Pharmacokinetic parameters of compound I-NP after i.p. injection in BALB/c mice

Pharmacokineticparameters Free compound I Compound I-NP Free compound I Compound I-NP

Dose, mg/kg 20 20 100 125AUC0-last, μg h/mL 209.2 ± 23.18 472.9 ± 20.99 706 ± 185 709.5 ± 43.9CL, mL/min per kilogram 1.5 0.7 2.3 2.9

Fig. 4. Pharmacokinetics of free compound I and compound I-NP in BALB/c mice. (A) Comparison of serum levels of compound I after i.p. administration of20 mg/kg free (black circles) and NP (red circles) compound I. The Inset depicts the enlarged view of the low serum concentrations observed for compound I-NP(red circle) and compound I (black circle). (B) Comparison of serum levels of compound I after i.p. administration of 100 mg/kg free (black circles) and 190 mg/kg NP(blue circles) compound I. The Inset depicts the enlarged view of the low serum concentrations observed for compound I-NP (blue circle) and compound I(black circle). The blood samples were collected at different intervals and analyzed as described in Materials and Methods. (C) HIV-1 inhibition by serumcontaining free compound I in TZM-bl cells. The values are mean ± SD from three different experiments involving triplicate measurements. (D) SEM image ofcompound I-NP. (Scale bar, 1 μM.) (E) HIV-1 inhibition by serum containing free compound I (black) and compound I-NP (blue) in TZM-bl cells. Acc. V, acceleratingvoltage; Det, detector; Exp, exposure number; LOD, limit of detection of HPLC used to quantify serum concentrations of compound I/compound I-NP; Magn,magnification; Spot, technical term that describes the width of the SEM beam, usually defined in reference to the center point of the beam (reference point andvalue may change based on specific SEM unit); WD, working distance.

E806 | www.pnas.org/cgi/doi/10.1073/pnas.1717932115 Kudalkar et al.

provide a firm basis for the efficacy studies in humanized mice de-scribed below.

Establishing Efficacy of the Late-Stage Preclinical Candidate inHumanized Mouse Models of HIV-1 Infection. Several humanizedmouse models are now available to assess various HIV-1 therapeuticmodalities (40). For our proof-of-concept studies to evaluate anti–HIV-1 therapeutic efficacy in vivo, NOD.Cg-PrkdcSCID Il2rgtm1Wjl/SzJmice that are reconstituted with human peripheral blood lymphocyte(NSG-Hu-PBL) were chosen. These NSG-Hu-PBL mice are recon-stituted with activated T cells and can, therefore, be readily infectedwith HIV-1JR-CSF strain to produce an infection causing a rapid lossof CD4+ T cells within a few days of infection (41, 42).On the basis of the data described above, we selected com-

pound I (free drug) at the 100-mg/kg dose for in vivo studies inHIV-1JR-CSF–infected NSG-Hu-PBL mice to assess efficacy ofthe late-stage preclinical candidate. Using this humanized mousemodel, we compared in vivo efficacy of free compound I in-jected daily with a single dose of compound I-NP given i.p. toHIV-1JR-CSF–infected NSG-Hu-PBL mice. The antiretroviralactivity of free and nanoformulated compound I was evaluatedby looking at the preservation of CD4+ T cells and determinationof plasma viral loads (PVLs) in NSG-Hu-PBL mice.Pharmacokinetics of compound I and compound I-NP in JR-CSF–infectedNSG-Hu-PBL mice. Experiments with free compound I and com-pound I-NP were performed in four groups of NSG-Hu-PBLmice infected with HIV-1JR-CSF (30,000 pfu) as shown in thescheme in Fig. 5A. Starting 1 d before HIV-1 infection (day −1),one group was treated with free compound I at 100 mg/kg per dayfor up to 19 d, after which the drug was withdrawn to assess theviral rebound and thus, productive infection (withdrawal group).The second group was treated with free compound I at 100 mg/kgper day for 32 d (continuous group). Mice in the third group re-ceived a single dose of 190 mg/kg compound I-NP from the secondbatch of NPs, where EE was higher. Several recent reports haveevaluated HAART in nanoformulation at dosing ranging from200 to 250 mg/kg (43, 44). Therefore, for the efficacy studies,we chose a dose of 190 mg/kg for compound I-NP. A fourth

group of mice received no treatment and was marked as thecontrol group.The serum concentrations of free and nanoformulated com-

pound I in mice were determined on days 8, 16, 25, and 32 andwere compared with the control group (Fig. 5B). Fig. 5B showsthat comparable levels of compound I were observed in serum ofall treated groups until day 16. On day 25, there was a greaterthan fourfold increase in serum levels of compound I in micefrom the continuous group in contrast to the withdrawal groupand compound I-NP group (Fig. 5B, solid red circles and solidblue circles, respectively). High serum levels of compound Iwere maintained until day 32 in the mice from continuousgroup. Accordingly, the serum concentrations of compound Ibecame undetectable on day 32 in the withdrawal group, butthese mice maintained serum levels comparable (14.6 μg/mL)with those of the continuous group on day 16. The compoundI-NP group maintained 14.5 μg/mL of compound I in serumuntil day 16, which was similar to the continuous group. Therewas a threefold reduction in compound I serum concentrationson day 25 in compound I-NP–treated mice. These data suggestthat one single injection of compound I in nanoformulationis able to maintain comparable serum levels of compound Icompared with the group that received free compound dailyfor 32 d.Antiviral efficacy of compound I and compound I-NP in HIV-1–infectedhumanized NOD.Cg-PrkdcSCID Il2rgtm1Wjl/SzJ mice. The antiviral effica-cies of free compound I and compound I-NP were evaluated infour groups of NSG-Hu-PBL mice that were infected with HIV-1JR-CSF (30,000 pfu) as shown in the scheme in Fig. 5A. Theserum samples from these mice were collected on days 0 (pre-infection), 8, 16, 19, 25, and 32 for levels of peripheral CD4+

T cells and HIV-1–RNA copies (PVLs) in comparison with un-treated mice (Fig. 6). The loss of human CD4+ T cells causedby the HIV-1 infection in reconstituted immunodeficient miceprovides an indicator of viral infection (41, 45). Thus, levels ofhuman CD4+ and CD8+ cells as a percentage of total humanCD3+ T cells and CD4+/CD8+ T-cell ratio were determined fromFACS analyses of blood samples (Fig. 6 A and B), which compares

Fig. 5. Experimental protocols used in efficacy studies to determine the antiviral activity of free compound I and compound I-NP in HIV-1–infected NSG-Hu-PBLmice. (A) Treatment paradigm to determine preinfection antiviral activity of 100 mg/kg free compound I and 190 mg/kg compound I-NP administered i.p. (B) Serumdrug levels in HIV-1–infected NSG-Hu-PBL mice treated with 100 mg/kg of free compound I daily and 190 mg/kg of compound I-NP on day −1. Mice receiving freecompound Iwere further divided into two groups. Mice in group I (green circles) received 100mg/kg of free compound I daily until day 32, whereas mice in group II(red circles) stopped receiving compound I on day 19. For both the groups, the serum drug levels were determined on days 8, 16, 25, and 32. After i.p. administrationof 190 mg/kg of compound I-NP (blue circles) on day −1, serum levels of compound I were determined on days 8, 16, 25, and 32. Data points represent ±SD.

Table 3. Characterization of compound I-NP

NPsParticle size,

nm PDIZeta potential,

mV

DL, % EE, %

Batch 1 Batch 2 Batch 1 Batch 2

Compound I-NP 370 ± 9.2 <0.2 −28 ± 0.1 10.5 ± 0.9 15.3 ± 1.2 104 ± 26.2 95 ± 9.1

Kudalkar et al. PNAS | Published online December 26, 2017 | E807

PHARM

ACO

LOGY

PNASPL

US

SEECO

MMEN

TARY

the no treatment, 100 mg/kg per day of compound I, and 190-mg/kgsingle-dose compound I-NPs over a period of 32 d. On the dayof infection, CD4+ T-cells levels were comparable in all fourgroups of mice (Fig. 6 A and B, open black circles). Low levels ofCD4+ T cells (∼2%) were observed in the control group by 2 wkpostinfection. CD4+ T-cells levels were preserved in all treat-ment groups at 2 wk post–HIV-1JR-CSF inoculation (69.3 ± 2,71.6 ± 0.8, and 70.6 ± 0.8% of total cells in continuous, with-drawal, and compound I-NP groups, respectively). While CD4+

T cells in mice from the continuous group were maintained to60.6 ± 4.7 and 50 ± 4.3% at weeks 3 and 4, respectively, mice in thewithdrawal group had reduced CD4+ T cells (32.1 ± 2.5%) by week4. In mice receiving compound I-NP, the CD4+ T cells weremaintained at 47.7 ± 11.8 and 25.1 ± 8.3% through weeks 3 and4, respectively.In all treatment groups, we found that compound I reduced

HIV-1 RNA by 3 log10 (1,000-fold; from a mean of 107 HIV-1RNA copies in untreated mice to 104 copies in treated mice) byday 8 and by 4 log10 (10,000-fold from a mean of 107 HIV-1 RNA

copies in untreated mice to 103 copies in treated mice) by day16 post–HIV-1JR-CSF inoculation. The continuous daily adminis-tration of free compound I at 100 mg/kg per day after HIV-1JR-CSFinoculation reduced viremia to undetectable levels (<150 copies) inall mice within 19 d of compound I treatment, and the viral loadsremained undetectable after 4 wk (Fig. 6C, solid green circles).Accordingly, in the withdrawal group, when treatment was stoppedat day 19 (Fig. 6C, solid red circles and withdrawal denoted bystar), viremia rebounded at week 3 to 104 copies (similar to that onday 8 posttreatment) and increased to 105 copies on withdrawal offree compound I by days 25 and 32. This is reflected in the drop ofCD4+ T cells in Fig. 6B. PVL of mice treated with a single doseof compound I-NP (Fig. 6C, solid blue circles) was below detectionlimit until day 19 and increased to 104 and 105 copies only by days25 and 32, respectively. In each of the latter two cases, the viremiarebound was accompanied by a drop in the plasma concentrationof compound I. These data indicate that continual exposure of freeor a single dose of long-acting compound I effectively controls virusreplication in NSG-Hu-PBL mice.

Fig. 6. Antiretroviral activities of free compound I and compound I-NP in NSG Hu PBL mice infected intraperitoneally with HIV-1 (JR-CSF) on day 0, 24 h afterdrug administration. One hundred milligrams per kilogram of free compound I was administered daily (days -1 to 32) by i.p. route, whereas 190 mg/kg ofcompound I-NP was administered by i.p. route on day -1. (A) FACS analyses compared the ratio of CD4+/CD8+ cells in untreated (Top), compound I-NP–treated(Middle), and free compound I–treated (Bottom) mice. (B) Quantification of FACS analysis shows the percentage of human CD4+ among total CD3+ T cells ofperipheral blood. (C) Total PVL determined by RT-PCR in HIV-1JR-CSF–infected NSG mice. Data points represent ±SD. LOD, limit of detection of assay used toquantify PVLs.

E808 | www.pnas.org/cgi/doi/10.1073/pnas.1717932115 Kudalkar et al.

DiscussionHAART as a combination therapy is the current standard of careto prevent HIV-1 infection or resistance development. The HIV/AIDS pandemic will continue for many years, underscoring theneed for developing novel more effective therapies. Consider-ations for improved anti–HIV-1 agents include fewer issues withlong-term toxicity, ability to counteract emergence of new drug-resistant variants, and a long-acting formulation that requiresless frequent dosing. Moreover, synergistic activity with existingHIV-1 drugs would offer added value as components of combi-nation therapy. Our computational and structure-guided strategyfocused on designing NNRTIs has yielded a potent series of cat-echol diether compounds with favorable pharmacokinetic andtoxicological profiles (18–20). In this study, compound I (Fig. 1)was selected for additional consideration as a late-stage preclinicalcandidate. Important criteria to meet for advancement were (i)synergy with existing FDA-approved drugs and clinical candidates,(ii) suitability as a long-acting formulation, and (iii) efficacy in adisease-relevant animal model for HIV-1 infection. This studyrepresents the culmination of an effort to design enhanced NNRTIsthat provide advantages over the current FDA-approved NNRTIs.A major benefit for any NNRTI, as part of a combination

HAART regimen, would be the ability to exhibit antiviral synergywith existing NRTIs or other classes, such as INSTIs. Therefore,the first criterion examined for compound I was whether it couldpotentiate the antiviral potency of NRTIs in HIV-1–infected hu-man T cells. The NRTIs chosen were FTC or TDF, as they arecomponents of STRs, Atripla and Complera. The INSTI EVG wasselected to examine another class of AIDS drugs, as it is currentlybeing evaluated in a long-acting formulation with rilpivirine in theLong-Acting Antiretroviral Treatment Enabling (LATTE) Trial(46). In addition, Ed4T and EFdA were examined. Ed4T (festinavir),a more potent, less toxic derivative of D4T (Zerit), was chosen, asit has shown promising results in phases I and II clinical trials (27,47). EFdA (MK-8591), a very potent new NRTI being developedby Merck, was selected, as it has also shown promising resultsin late-stage preclinical and early phase I clinical trials (28, 29,47). As illustrated in the isobologram and MacSynergy II plotsin Figs. 2 and 3, compound I showed strong synergy with all ofthe compounds tested representing current STR regimens(FTC and TDF) and another HIV-1 drug class, INSTIs (EVG).These characteristics of compound I confer advantages overthe NNRTI doravirine (MK-1439) currently being evaluated inphase III trials, which only showed nonantagonistic behaviorwith other HIV-1 agents (21).A second criterion taken into account was the suitability of

compound I for development of a long-acting formulation.Biodegradable PLGA NPs were used to encapsulate compound Ias a strategy to reduce the daily dosage by prolonged and sus-tained release. Characterization of compound I-NPs showed thatit was encapsulated in PLGA NPs with the appropriate particlesize (370 ± 9.2 nm) and zeta potential (−28 ± 0.1). The negativezeta potential suggests that compound I-NP could have pro-longed circulation time in vivo by escaping uptake in cells andtissues (48, 49). A prolonged circulation time for compoundI-NP was confirmed by preliminary pharmacokinetic studies.When a single 20- or 125-mg/kg dose of compound I-NP wasadministered i.p. to BALB/c mice, detectable levels of compoundI in mouse serum were observed for 35 d (Fig. 4 A and B), con-firming a substantially longer residence time. Additionally, theserum levels of ∼0.5 μg/mL of compound I-NP on day 35 weresignificantly higher than the concentrations required to inhibitHIV-1 replication in cells in our MT-2 (20) and TZM-bl assays(Fig. 4C). The encapsulation of compound I prevented the burst orrapid release of the drug from NP (50) as seen by 3- and 10-foldlower Cmax levels at 20- and 100-mg/kg doses, respectively, comparedwith free injected compound I. Additional analysis of pharmaco-

kinetic data at the 20-mg/kg dose indicated that the AUC0-last ofcompound I-NP was twofold higher and that the total body CL wastwofold lower compared with free injected compound I. The ob-served AUC0-last value (472.9 ± 20.9) of compound I-NP at 20 mg/kgwas also twofold higher compared with the AUC0-last value (230.6 ±80.6 μg h/mL) of efavirenz observed in a similar pharmacokineticstudy in BALB/c mice (33). The larger AUC0-last value and thelower CL value could suggest that a higher volume of distributioncan be achieved by compound I-NP because of prolonged releaseof compound I and also, that encapsulation of compound I leads toa slower metabolism. In contrast, the AUC0-last and CL of com-pound I-NP at higher dose were comparable with those of freeinjected compound I. Hence, comparison of pharmacokineticprofiles of free and nanoformulated compound I suggests that asingle dose of compound I-NP displays a longer residence time ofover a month when administered in an equivalent dose as freecompound I in BALB/c mice. Additionally, the serum levels ob-served in pharmacokinetic study were well above the therapeuticconcentrations, which could provide good protection against acuteloss of CD4+ T cells and could also decrease viral loads to eradi-cate HIV-1 from the anatomically privileged sites, where free drugconcentrations can be subtherapeutic (51, 52). An initial assess-ment of the antiviral efficacy ex vivo for the compound I-NPs wasprovided by examining the ability of the serum aliquots from eachtime point to inhibit HIV-1 infection in TZM-bl cells infected withHIV-1JR-CSF strain (Fig. 4E). The prolonged residence time, serumlevels above the therapeutic concentration for compound I-NP,and confirmation of antiviral activity in HIV-1–infected cellsretained even at the 35-d collection suggest that single dosing ofcompound I-NP should suffice for maintaining a good virus controland that it has an excellent potential for use as a long-acting anti–HIV-1 NNRTI.The third criterion, perhaps the most crucial for compound I

as a late-stage preclinical candidate, was an assessment of effi-cacy in HIV-1–infected NSG-Hu-PBL mice. The design of thetreatment groups to evaluate efficacy is illustrated in Fig. 5A.Earlier pharmacokinetic studies with compound I given as an i.p.dose of 100 mg/kg to BALB/c mice indicated that sufficient levelsof compound remained after 24 h, prompting consideration as aonce daily dose (20, 37). The efficacy for the treatment groupagainst HIV-1 infection was assessed by examining two key pa-rameters: loss of CD4+ T cells as an indicator of HIV-1 infectionand the PVL, which reflects the efficiency of viral replication.The i.p. administration of compound I-NP and free compound Iin HIV-1–infected Hu-PBL-NSG mice clearly showed the strik-ing antiviral potency of this catechol diether NNRTI (Fig. 6). Adaily i.p. administration of compound I at 100 mg/kg preventedthe acute loss of CD4+ T cells in HIV-1–infected NSG-Hu-PBLmice as determined from FACS analysis (Fig. 6A), preserved thispopulation of cells by almost 90% after 2 wk, and protectedHIV-1 in these mice by reducing PVLs to undetectable levels(Fig. 6 B and C, solid green and red circles). These mice showedplasma viral levels with 3- to 4-log10 decreases within 8–16 d ofinitiation of treatment. A similar preservation of CD4+ T cellsand decrease in plasma viral levels were observed with a singledose of compound I-NPs in the 2-wk period (Fig. 6 B and C, solidblue circles). As anticipated, for the withdrawal group, viremiarebounded in 10 d after treatment cessation (Fig. 6C, solid redcircles). However, the viral rebound was slower, and the PVLswere similar to ones seen within 8 d posttreatment initiation. Thetreatment with a single dose of compound I-NP halted acuteCD4+ T-cell loss and decreased PVL until day 25 posttreat-ment in HIV-1–infected NSG-Hu-PBL mice, while in thecontinuous group, the viral load remained undetectable for theentire 32 d of the study. Hence, these studies have shown thattreatment of HIV-1–infected NSG-Hu-PBL mice with a singledose of compound I-NP a day before infection suppressesplasma HIV-1 viral load and prevents the acute loss of CD4+

Kudalkar et al. PNAS | Published online December 26, 2017 | E809

PHARM

ACO

LOGY

PNASPL

US

SEECO

MMEN

TARY

T cells for almost 3 wk, which is comparable with mice receivingfree compound I daily.Taken together, the data presented in this study show the potent

antiviral activity of compound I in free and nanoformulated formsin HIV-1–infected Hu-PBL-NSG mice. The experiments reportedhere established the feasibility of using PLGA compound I-NPs asa long-acting drug delivery system, which was able to provide asustained and prolonged release of compound I in mice. Addi-tionally, given the potential to be combined with other anti–HIV-1drugs with synergistic effect, lack of toxicity and effects on hERGion channel and CYP3A4 (20), and effectiveness against clinicallyrelevant drug-resistant HIV-1 strains (i.e., Y181C, K103N, K101P/E,E138K, M184V) (53), compound I exhibits exceptional propertiesas an NNRTI for the treatment of HIV-1 infection that couldprovide durable clinical antiviral efficacy. Considering that ex-tended duration anti–HIV-1 agents in combination with otherdrugs are the main focus of emerging HIV-1 therapies, com-pound I is a promising late-stage preclinical candidate thatwarrants full development. One of the emerging HIV-1 therapies,preexposure prophylaxis (PrEP), holds the promise of eliminatingnew infections along with the associated risks of HIV infection.Hence, the qualities of compound I, synergistic effects with otherHIV-1 drugs, and ease of formulation to a long-acting drug withpotent antiviral activity make it a good candidate to additionalstudy as a PrEP formulation.

Materials and MethodsChemicals and Synthesis of Compound I. Poly(d,l lactic-coglycolic-acid), 50:50with inherent viscosity 0.95–1.20 dL/g, was purchased from DURECT Corpo-ration. Compound I was synthesized as detailed previously (19).

Preparation of Humanized Mice. NOD.Cg-PrkdcSCID Il2rgtm1Wjl/SzJ (NSG) micewere purchased from the Jackson Laboratory. All experiments were per-formed according to protocols approved by the Institutional Review Boardand the Institutional Animal Care and Use Committee of Yale University.NSG-Hu-PBL mice were prepared by engrafting peripheral blood mono-nuclear cells (PBMCs) as described (42, 54). Briefly, 5–7 × 106 PBMCs puri-fied by Ficoll density gradient centrifugation of healthy donor blood(obtained from the New York Blood Bank) were injected i.p. in a 200-μLvolume into 7- to 10-wk-old NSG mice. Cell engraftment was checked 14 dpostinjection as described previously; 50–100 μL of blood was collected byretroorbital bleeding. PBMCs were isolated by Ficoll density gradientcentrifugation; stained with Fluor-conjugated anti-human CD45, CD3,CD4, and CD8 antibodies; and analyzed by flow cytometry to confirmengraftment.

Fabrication of Compound I-NPs. Briefly, 100mg of PLGA polymer was dissolvedin dichloromethane at a concentration of 50 mg/mL. Compound I was dis-solved separately in 200 μL of DMSO and added to the polymer solution. Theaqueous phase consisted of 4 mL of 5% polyvinyl alcohol. An emulsion ofthe solutions was created through dropwise addition of the polymer/drugsolution to the aqueous solution under vortex followed by three cycles (10 seach) of sonication using a probe sonicator (amplitude 38%). NPs werecollected by centrifugation at 16,100×g. Trehalose was added as a cryo-protectant at 100% (wt/wt) before lyophilization.

Characterization of Compound I-NPs. NP size, PDI, and zeta (ζ) potential weremeasured by DLS by resuspending 0.05 mg NPs in 1 mL deionized waterusing a Zetasizer Nano ZS90 (Malvern Instruments). SEM images wereobtained on an XL-30 scanning electron microscope (FEI; Hilsboro). Tomeasure surface charge (zeta potential), NPs were diluted in deionizedwater at a concentration of 0.5 mg/mL; 750 μL of solution was loaded intoa disposable capillary cell (Malvern Instruments), and the charge wasmeasured using a Malvern Nano-ZS. The abbreviations used in SEM imageare: Acc. V, accelerating voltage; Det, detector; Exp, exposure number;Magn, magnification; Spot, technical term that describes the width ofthe SEM beam, usually defined in reference to the center point of thebeam (reference point and value may change based on specific SEM unit);and WD, working distance.

Reverse-phase HPLC was used to measure DL (%), which is defined as themeasured mass of compound I per mass of PLGA NP, and EE (%), which is

defined as the ratio of the compounds loaded to the total drugs used forfabricating the NPs:

DLð%Þ=mass  of  compound  I  in NPstotal mass  of NPs

×100

EEð%Þ= mass  of  drug  in  NPsmass  of  polymer  used  in  formulation

×100.

Briefly, DL was determined by dissolving a known mass of lyophilizedcompound I-NP in DMSO. The samples were filtered by using an Acrodisc25-mm syringe filter with a 0.45-μm HT Tuffryn membrane (Pall Life Sciences)followed by additional dilution in acetonitrile. The samples were analyzedusing the HPLC system as previously described in detail (20).

Pharmacokinetic Studies. To determine dose-dependent serum drug con-centrations, in vivo pharmacokinetics of free compound I and compoundI-NP was investigated in BALB/c mice (20 mg), which were obtained fromthe Jackson Laboratory. BALB/c mice were randomly divided into fourgroups containing three mice each. Two groups of mice were injected i.p.with 20 or100 mg/kg of free compound I suspended in sterile saline so-lution containing 10% Tween 80. The other groups was injected i.p.with 20 and 125 mg/kg compound I-NP suspended in sterile saline solu-tion. These doses corresponded to human doses of 1.6–10.2 mg/kg basedon an interspecies scaling factor of 12.3 (36). The blood samples werecollected at predetermined time points from the ocular venous plexusby retroorbital venipuncture and used for subsequent analysis asdetailed previously.

HIV-1 Challenge in Humanized Mice, Pharmacokinetic Studies, and AntiretroviralActivity of Free Compound I and Compound I-NP. HIV-1 infection and main-tenance of infected mice were done in the BSL-3 animal facility understandard caging conditions. Twelve NSGmice were divided into four groupsof n = 3: control or no treatment, free compound I injected at 100 mg/kgper day for a total of 35 d, free compound I withdrawn on day 19, andcompound I-NP (single dose at 190 mg/kg on day 0). One day after thetreatment, animals were challenged i.p. with HIV-1JR-CSF [30,000 TissueCulture Infectious Dose (TCID50)] and were bled retroorbitally to check thelevels of CD4+ T cells and the PVLs. Serum levels of compound I were an-alyzed on days 8, 16, 19, 25, and 32 after retroorbital venipuncture usingHPLC as described earlier. On day 19, free compound I was withdrawnfrom one group (n = 3), and the mice were observed for viral rebound.Additionally, pharmacokinetic studies in these mice were carried out aspreviously described for BALB/c mice.

Quantification of PVLs and FACS Analyses of T-Cell Subsets. Quantification ofPVLs was done using the method described previously (55). Briefly, plasmaviral RNA was extracted using the QIAamp viral RNA mini kit (QIAGEN)following the manufacturer’s protocol. SYBR green real-time PCR assay wascarried out in a 20-μL PCR mixture volume consisting of 10 μL of 2× Quan-titect SYBR green RT-PCR Master Mix (Qiagen) containing HotStarTaq DNApolymerase, 0.5 μL of 500 nM each oligonucleotide primer, 0.2 μL of 100×QuantiTect RT Mix (containing Omniscript and Sensiscript RTs), and 8 μL ofRNA extracted from plasma samples or standard HIV-1 RNA (from 5 × 105 to5 copies per 1 mL). Highly conserved sequences on the gag region of HIV-1 were chosen, and specific HIV-1 gag primers were selected (56). The se-quences of HIV-1 gag primers are 5′ TGCTATGTCAGTTCCCCTTGGTTCTCT3′ and 5′ AGTTGGAGGACATCAAGCAGCCATGCAAAT 3′. Amplification wasdone in an Applied Biosystems 7500 real-time PCR system, and it involvedactivation at 45 °C for 15 min and 95 °C for 15 min followed by 40 amplifi-cation cycles of 95 °C for 15 s, 60 °C for 15 s, and 72 °C for 30 s. For thedetection and quantification of viral RNA, the real-time PCR of each samplewas compared with threshold cycle value of standard curve. For FACSanalysis, 50–100 μL of blood was collected by retroorbital bleeding on days8, 16, 19, 25, and 32. PBMCs were isolated by Ficoll density gradient cen-trifugation; stained with Fluor-conjugated anti-human CD45, CD3, CD4,and CD8 antibodies; and analyzed by flow cytometry to assess CD4+

T-cell levels.

Measurement of the Inhibitory Activity of Compound I-NP in Infected TZM-bl Cells.The in vitro anti–HIV-1 activity of free compound I and compound I-NPwas determined using a luciferase reporter gene assay as described pre-viously (20).

E810 | www.pnas.org/cgi/doi/10.1073/pnas.1717932115 Kudalkar et al.

Combination Study. The combination inhibitory effects of compound I anddifferent drugs (TDF, FTC, EVG, Ed4T, and EFdA) were tested in a two-drugcombination as described previously (30, 57). The combination inhibitory datawere analyzed using isobologram analysis and MacSynergy II 3D plots (31). Forisobologram analysis, isobolograms were constructed by measuring the FEC50values of two anti–HIV-1 agents when used in combination in the MT-2 assay(30). The values on the additivity line connecting FEC50 values of one representan additive interaction between the two compounds, while points markedlyabove or below the line represent antagonism or synergy, respectively. Thecombination inhibitory effect of compound I was also analyzed using theMacSynergy II 3D model by Prichard and Shipman (31). Ribavirin and d4T wereused as the positive controls as an antagonism pair (32, 58). The resulting sur-face plots of the data reflect the difference between the experimental dose–

response surface and the predicted additive surface. On a 3D model, a simpleadditive effect will result in a horizontal plane at 0% inhibition, whereas asynergistic or antagonistic effect will render a peak or depression above orbelow the horizontal plane. The volumes of the peaks/depressions were thencalculated to quantify the effect of the drug combination on antiviral activity[synergy/antagonism volumes (μM2%)]. For these studies, synergy volumes aredivided into the following categories: minor synergy: 25–50 μM2 %; moderatesynergy: 50–100 μM2 %; strong synergy: over 100 μM2 %; minor antagonistic:−50 and −100 μM2 %; and strong antagonistic: less than −100 μM2 %.

ACKNOWLEDGMENTS. This work was supported by NIH Grants AI44616,GM49551, AI112443, AI122384, and EB000487.

1. UNAIDS (2016) Global AIDS Update 2016. Available at http://www.unaids.org/sites/default/files/media_asset/global-AIDS-update-2016_en.pdf. Accessed December 4, 2017.

2. Schneider MF, et al. (2005) Patterns of the hazard of death after AIDS through theevolution of antiretroviral therapy: 1984-2004. AIDS 19:2009–2018.

3. Wood E, et al. (2008) Highly active antiretroviral therapy and survival in HIV-infectedinjection drug users. JAMA 300:550–554.

4. Palella FJ, Jr, et al.; HIV Outpatient Study Investigators (1998) Declining morbidity andmortality among patients with advanced human immunodeficiency virus infection. NEngl J Med 338:853–860.

5. Spence RA, Kati WM, Anderson KS, Johnson KA (1995) Mechanism of inhibition ofHIV-1 reverse transcriptase by nonnucleoside inhibitors. Science 267:988–993.

6. Hopkins S, Furman PA, Painter GR (1989) Investigation of the stereochemical course ofDNA synthesis catalysed by human immunodeficiency virus type 1 reverse transcrip-tase. Biochem Biophys Res Commun 163:106–110.

7. Nissley DV, et al. (2007) Characterization of novel non-nucleoside reverse transcrip-tase (RT) inhibitor resistance mutations at residues 132 and 135 in the 51 kDa subunitof HIV-1 RT. Biochem J 404:151–157.

8. Fellner C (2016) Pharmaceutical approval update. P T 41:220–221.9. Permpalung N, Putcharoen O, Avihingsanon A, Ruxrungtham K (2012) Treatment of

HIV infection with once-daily regimens. Expert Opin Pharmacother 13:2301–2317.10. Sebaaly JC, Kelley D (2017) Single-tablet regimens for the treatment of HIV-1 in-

fection. Ann Pharmacother 51:332–344.11. Bowes J, et al. (2012) Reducing safety-related drug attrition: The use of in vitro

pharmacological profiling. Nat Rev Drug Discov 11:909–922.12. Health Canada (2011) Summary Basis Decision (SBD) EDURANT, Rilpivirine hydro-

chloride, 25 mg tablet (Janssen, Toronto), Technical Report 205395.13. Molina JM, et al.; ECHO study group (2011) Rilpivirine versus efavirenz with tenofovir

and emtricitabine in treatment-naive adults infected with HIV-1 (ECHO): A phase3 randomised double-blind active-controlled trial. Lancet 378:238–246.

14. Jorgensen WL (2006) QikProp (Schrodinger, New York), Version 3.0.15. Jorgensen WL, Duffy EM (2002) Prediction of drug solubility from structure. Adv Drug

Deliv Rev 54:355–366.16. Lipinski CA, Lombardo F, Dominy BW, Feeney PJ (2001) Experimental and computa-

tional approaches to estimate solubility and permeability in drug discovery and de-velopment settings. Adv Drug Deliv Rev 46:3–26.

17. Norinder U, Bergström CA (2006) Prediction of ADMET properties. ChemMedChem 1:920–937.

18. Lee WG, et al. (2013) Picomolar inhibitors of HIV reverse transcriptase featuring bi-cyclic replacement of a cyanovinylphenyl group. J Am Chem Soc 135:16705–16713.

19. Lee WG, et al. (2014) Picomolar inhibitors of HIV-1 reverse transcriptase: Design andcrystallography of naphthyl phenyl ethers. ACS Med Chem Lett 5:1259–1262.

20. Kudalkar SN, et al. (2017) Structural and preclinical studies of computationally de-signed non-nucleoside reverse transcriptase inhibitors for treating HIV infection. MolPharmacol 91:383–391.

21. Lai MT, et al. (2014) In vitro characterization of MK-1439, a novel HIV-1 nonnucleosidereverse transcriptase inhibitor. Antimicrob Agents Chemother 58:1652–1663.

22. Richman D, et al. (1991) BI-RG-587 is active against zidovudine-resistant human im-munodeficiency virus type 1 and synergistic with zidovudine. Antimicrob AgentsChemother 35:305–308.

23. Chong K-T, Pagano PJ (1997) Inhibition of human immunodeficiency virus type 1 in-fection in vitro by combination of delavirdine, zidovudine and didanosine. AntiviralRes 34:51–63.

24. Janssen PA, et al. (2005) In search of a novel anti-HIV drug: Multidisciplinary coordinationin the discovery of 4-[[4-[[4-[(1E)-2-cyanoethenyl]-2,6-dimethylphenyl]amino]-2- pyrimidinyl]amino]benzonitrile (R278474, rilpivirine). J Med Chem 48:1901–1909.

25. Sun LQ, et al. (2012) Optimization of 2,4-diarylanilines as non-nucleoside HIV-1 re-verse transcriptase inhibitors. Bioorg Med Chem Lett 22:2376–2379.

26. Frey KM, et al. (2014) Structure-based evaluation of C5 derivatives in the catecholdiether series targeting HIV-1 reverse transcriptase. Chem Biol Drug Des 83:541–549.

27. Haraguchi K, et al. (2013) From the chemistry of epoxy-sugar nucleosides to the dis-covery of anti-HIV agent 4′-ethynylstavudine-Festinavir. Curr Pharm Des 19:1880–1897.

28. Wu VH, et al.; University of Washington–Dakar HIV-2 Study Group (2017) MK-8591 (4′-ethynyl-2-fluoro-2′-deoxyadenosine) exhibits potent activity against HIV-2 isolates anddrug-resistant HIV-2 mutants in culture. Antimicrob Agents Chemother 61:e00744-17.

29. Stoddart CA, et al. (2015) Oral administration of the nucleoside EFdA (4′-ethynyl-2-fluoro-2′-deoxyadenosine) provides rapid suppression of HIV viremia in humanizedmice and favorable pharmacokinetic properties in mice and the rhesus macaque.Antimicrob Agents Chemother 59:4190–4198.

30. Ray AS, et al. (2005) Mechanism of anti-human immunodeficiency virus activity ofbeta-D-6-cyclopropylamino-2′,3′-didehydro-2′,3′-dideoxyguanosine. Antimicrob AgentsChemother 49:1994–2001.

31. Prichard MN, Shipman C, Jr (1990) A three-dimensional model to analyze drug-druginteractions. Antiviral Res 14:181–205.

32. Margot NA, Miller MD (2005) In vitro combination studies of tenofovir and othernucleoside analogues with ribavirin against HIV-1. Antivir Ther 10:343–348.

33. Destache CJ, Belgum T, Goede M, Shibata A, Belshan MA (2010) Antiretroviral release frompoly(DL-lactide-co-glycolide) nanoparticles in mice. J Antimicrob Chemother 65:2183–2187.

34. Demetrius L (2005) Of mice and men. When it comes to studying ageing and themeans to slow it down, mice are not just small humans. EMBO Rep 6:S39–S44.

35. Rabin L, et al. (1996) Use of standardized SCID-hu Thy/Liv mouse model for preclinicalefficacy testing of anti-human immunodeficiency virus type 1 compounds. AntimicrobAgents Chemother 40:755–762.

36. Stoddart CA, et al. (2007) Validation of the SCID-hu Thy/Liv mouse model with fourclasses of licensed antiretrovirals. PLoS One 2:e655.

37. Speck RF (2014) Humanized Mice for HIV Research (Springer, New York).38. Yan Wang WQ, Choi SH (2016) FDA’s regulatory science program for generic PLA/

PLGA-based drug products. Am Pharm Rev 19:__.39. Le Thi Mai H, Nguyen Tai C, Nguyen Minh T, Le Ngoc Thanh N, Dang Mau C (2009) Prep-

aration of drug nanoparticles by emulsion evaporation method. J Phys Conf Ser 187:012047.40. Akkina R, et al. (2016) Improvements and limitations of humanized mouse models for

HIV research: NIH/NIAID “meet the experts” 2015 workshop summary. AIDS Res HumRetroviruses 32:109–119.

41. Mosier DE, et al. (1991) Human immunodeficiency virus infection of human-PBL-SCIDmice. Science 251:791–794.

42. Kumar P, et al. (2008) T cell-specific siRNA delivery suppresses HIV-1 infection in hu-manized mice. Cell 134:577–586.

43. Dash PK, et al. (2012) Long-acting nanoformulated antiretroviral therapy elicits po-tent antiretroviral and neuroprotective responses in HIV-1-infected humanized mice.AIDS 26:2135–2144.

44. Roy U, et al. (2012) Pharmacodynamic and antiretroviral activities of combinationnanoformulated antiretrovirals in HIV-1-infected human peripheral blood lymphocyte-reconstituted mice. J Infect Dis 206:1577–1588.

45. Mosier DE (1995) Distinct rate and patterns of human CD4+ T-cell depletion in hu-PBL-SCID mice infected with different isolates of the human immunodeficiency virus.J Clin Immunol 15(6 Suppl):130S–133S.

46. Margolis DA, et al. (2017) Long-acting intramuscular cabotegravir and rilpivirine inadults with HIV-1 infection (LATTE-2): 96-Week results of a randomised, open-label,phase 2b, non-inferiority trial. Lancet 390:1499–1510.

47. Sohl CD, et al. (2012) Balancing antiviral potency and host toxicity: Identifying anucleotide inhibitor with an optimal kinetic phenotype for HIV-1 reverse transcrip-tase. Mol Pharmacol 82:125–133.

48. Jiang Y, et al. (2015) Nanoparticle-based ARV drug combinations for synergistic in-hibition of cell-free and cell-cell HIV transmission. Mol Pharm 12:4363–4374.

49. Zhang L, et al. (2008) Function analysis of sequences in human APOBEC3G involved inVif-mediated degradation. Virology 370:113–121.

50. Brazel CS, Huang X (2004) The cost of optimal drug delivery: Reducing and preventingthe burst effect in matrix systems. Carrier-based drug delivery. ACS Symp Ser AmChem Soc 879:267–282.

51. Wong JK, et al. (1997) Reduction of HIV-1 in blood and lymph nodes following potentantiretroviral therapy and the virologic correlates of treatment failure. Proc Natl AcadSci USA 94:12574–12579.

52. Weissman D, et al. (2000) Interleukin-2 up-regulates expression of the human immunode-ficiency virus fusion coreceptor CCR5 by CD4+ lymphocytes in vivo. J Infect Dis 181:933–938.

53. Gray WT, et al. (2015) Potent inhibitors active against HIV reverse transcriptase withK101P, a mutation conferring rilpivirine resistance. ACS Med Chem Lett 6:1075–1079.

54. Kim SS, Kumar P, Ye C, Shankar P (2012) Humanized mice for studying human leu-kocyte integrins in vivo. Methods Mol Biol 757:509–521.

55. Gibellini D, Vitone F, Gori E, La Placa M, Re MC (2004) Quantitative detection ofhuman immunodeficiency virus type 1 (HIV-1) viral load by SYBR green real-time RT-PCR technique in HIV-1 seropositive patients. J Virol Methods 115:183–189.

56. Christopherson C, Sninsky J, Kwok S (1997) The effects of internal primer-templatemismatches on RT-PCR: HIV-1 model studies. Nucleic Acids Res 25:654–658.

57. Feng JY, et al. (2009) The triple combination of tenofovir, emtricitabine and efavirenz showssynergistic anti-HIV-1 activity in vitro: A mechanism of action study. Retrovirology 6:44.

58. Lai MT, et al. (2009) Antiviral activity of MK-4965, a novel nonnucleoside reversetranscriptase inhibitor. Antimicrob Agents Chemother 53:2424–2431.

Kudalkar et al. PNAS | Published online December 26, 2017 | E811

PHARM

ACO

LOGY

PNASPL

US

SEECO

MMEN

TARY