Long‐acting and extended‐release implant and nanoformulations with a synergistic antiretroviral two‐drug combination controls HIV‐1 infection in a humanized mouse model

Abstract The HIV pandemic has affected over 38 million people worldwide with close to 26 million currently accessing antiretroviral therapy (ART). A major challenge in the long‐term treatment of HIV‐1 infection is nonadherence to ART. Long‐acting antiretroviral (LA‐ARV) formulations, that reduce dosing frequency to less than once a day, are an urgent need that could tackle the adherence issue. Here, we have developed two LA‐ART interventions, one an injectable nanoformulation, and the other, a removable implant, for the delivery of a synergistic two‐drug ARV combination comprising a pre‐clinical nonnucleoside reverse transcriptase inhibitor (NNRTI), Compound I, and the nucleoside reverse transcriptase inhibitor (NRTI), 4′‐ethynyl‐2‐fluoro‐2′‐deoxyadenosine. The nanoformulation is poly(lactide‐co‐glycolide)‐based and the implant is a copolymer of ω‐pentadecalactone and p‐dioxanone, poly(PDL‐co‐DO), a novel class of biocompatible, biodegradable materials. Both the interventions, packaged independently with each ARV, released sustained levels of the drugs, maintaining plasma therapeutic indices for over a month, and suppressed viremia in HIV‐1‐infected humanized mice for up to 42 days with maintenance of CD4+ T cells. These data suggest promise in the use of these new drugs as LA‐ART formulations in subdermal implant and injectable mode.


| INTRODUCTION
Antiretroviral therapy (ART) has revolutionized the treatment of HIV-AIDS and the disease is no longer a death sentence. However, as ART does not cure, treatment must be life-long. Nonadherence to daily oral ART medication remains a significant barrier to achieve long-term suppression of HIV replication and prevention of the emergence of drug-resistant virus. Numerous studies show direct correlation between ART adherence and reduction in viral loads and elevation in CD4 T cell counts. [1][2][3] Side-effects and psychological reactions to taking ART also leads to nonadherence in HIV-positive subjects, particularly younger individuals, 4,5 who account for more than half of all new HIV infections. 6 With the exception of the recently approved Cabenuva, an injectable extended-release nanoformulation of the integrase strand transfer inhibitor (INSTI), cabotegravir, and the nonnucleoside reverse transcriptase inhibitor (NNRTI) rilpivirine, there are no other approved ART formulations with a dosing frequency of less than once daily. Thus, new drugs with improved pharmacological properties, and a long-acting (LA) antiretroviral (ARV) drug delivery system that can reduce the dosing frequency to weekly, monthly, or even longer periods of time could represent a significant advance in HIV treatment, especially in high-risk populations.
The majority of approved and investigational ARVs with sufficient antiviral potency are difficult to adapt for LA formulations due to suboptimal physicochemical properties as well as short plasma half-lives, necessitating high drug loading (DL). Cabenuva, approved by FDA in January 2021, comprises once-monthly intramuscular injections of rilpivirine and cabotegravir each, in a large (3 ml) volume, to achieve the desired pharmacokinetic profile. 7,8 An additional critical drawback of injectable LA nanoformulations is that they cannot be removed from the body in case of a medical emergency and can magnify adverse effects because of their intrinsic property of extending the half-life of plasma drug concentrations. 9 Some of these limitations may be addressed by removable biodegradable implants, which can also provide extended drug release over many months. 10 Currently, 4 0 -ethynyl-2-fluoro-2 0 -deoxyadenosine (EFdA; Figure 1(a)) (also called MK-8591 and Islatravir), an investigational nucleoside reverse transcriptase inhibitor (NRTI) and dolutegravir (DTG), an INSTI, have been tested after formulating in removable, LA implants. 10,11 For EFdA implants, pharmacokinetic studies in rodents showed sustained therapeutic levels of the drug for up to 6 months although antiviral efficacy was not reported. 11 The DTG implants delivered up to 9 months and showed some viral suppression and protection after vaginal challenge in animal models; however, drug resistance and viral breakthrough were noted as early as 19 days post-therapy. 10 These studies indicate promising results for removable HIV drug implants but reveal the limitations of monotherapy and the critical need for a multidrug combination, as in oral ART regimens.
The long-term experience and success with three-drug HIV therapeutic regimens, all of which contain two NRTIs in combination with an NNRTI, INSTI or a protease inhibitor, is well documented. 12 While the only two approved two-drug regimens do not include NRTIs, 12 a new two-drug NRTI/NNRTI regimen is currently being evaluated by Merck in Phase III clinical trials (ClinicalTrials.gov Identifier: NCT04233879). Many of these agents have shown synergistic antiviral potency in cell culture. 13 Our previous studies showed potent synergy between a preclinical candidate NNRTI, Compound I, which is a catechol diether and the NRTI, EFdA (Figure 1(a,b)). [14][15][16][17] Development of Compound I was guided by mechanistic studies and computational design leading to enhanced pharmacological properties, drug resistance profiles, and a wide margin of safety relative to the current FDA approved NNRTIs such as efavirenz and rilpivirine. Compound I is active in the nanomolar range against wild-type HIV-1 strains and common drug resistant variants including Y181C and K103N, and demonstrates synergistic antiviral activity with existing HIV-1 drugs and clinical candidates, notably EFdA. [13][14][15][16][17] Compound I also F I G U R E 1 Chemical structures of (a) 4 0 -ethynyl-2-fluoro-2 0 -deoxyadenosine (EFdA) and (b) Compound I demonstrates favorable pharmacokinetics and absorption, distribution, metabolism, excretion, toxicity (ADME-Tox) profile as well as antiviral efficacy in a hu-mouse model for HIV-1. 13,18 Notably, Compound I showed no inhibition of the HERG ion channel which might prolong the Q-T interval leading to cardiotoxicity which has limited the dosing of the NNRTI rilpivirine. 18 Furthermore, a poly(lactide-coglycolide) (PLGA) nanoformulation of Compound I enabled sustained maintenance of plasma drug concentrations and antiviral efficacy for 3-5 weeks after a single dose. 13 In this report, we extend on these observations using the PLGA-based nanoformulation to deliver the synergistic two-drug combination of Compound 1 and EFdA for an extended release profile in hu-mice. Compound I and EFdA were chosen as the NNRTI and NRTI combination, based upon their optimal pharmacological ADME-Tox properties, potent and synergistic antiviral efficacy in suppressing viral replication. EFdA has been one of the most potent NRTIs evaluated and the clinical trials conducted by Merck have shown very promising results. 19,20 We further advanced the formulation to implants with copolymers of ω-pentadecalactone and p-dioxanone, poly(PDL-co-DO), a novel class of biocompatible, biodegradable materials that maintain structural integrity during degradation and also allow for ready removal should issues with toxicity arise. Poly(PDL-co-DO) has physical properties particularly well suited for providing long-term sustained release not achievable with other polymers that degrade via hydrolysis such as PLGA and other polyesters. [21][22][23] Poly(PDL-co-DO) is semicrystalline over the entire range of copolymer compositions; varying the copolymer composition alters its degradation rate, drug release rate. Importantly, degradation is slow enough to allow production of a degradable device that remains mechanically strong for at least 12 months. Previously, we showed that films of poly(PDL-co-DO) are well-tolerated after subcutaneous implantation and that poly(PDL-co-DO) can be formulated into particles that slowly release doxorubicin or siRNA. 23 We have also demonstrated that poly(PDLco-DO) can be engineered into a degradable contraceptive implant that provides consistent release of levonorgestrel for periods of over 2 years. 24 The current report describes the efficacy of Compound I/EFdA combination, independently formulated as LA PLGA-based nanoformulations as well as poly(PDL-co-DO) implants in terms of pharmacokinetics and antiviral efficacy in a hu-mouse model of HIV infection.

| Fabrication of Compound I and EFdA-NPs
Compound I-loaded PLGA NPs was formulated by a single emulsionsolvent evaporation method. 13,25 Due to the hydrophilic properties of EFdA, EFdA-loaded PLGA NPs were formulated using a water-in-oil-in-water (W-O-W) technique. 26  were analyzed using HPLC as described. 18 The limit of detection (LOD) for Compound I and EFdA was 0.1 and 0.25 μg/ml, respectively.
NP size, PDI, and zeta (ζ) potential were measured by dynamic light scattering by resuspending 0.05 mg NP in 1 ml deionized water using a Zetasizer Nano ZS90 (Malvern Instruments). SEM images were obtained on a Hitachi SU7000 scanning electron microscope. To measure surface charge (ζ), NPs were diluted in deionized water at a concentration of 0.5 mg/ml; 750 μl of solution was loaded into a disposable capillary cell (Malvern Instruments), and the charge measured using a Malvern Nano-ZS.

| Formulation of Compound I and EFdA containing LA subdermal implants
Poly(PDL-co-DO) with 40% p-dioxanone (DO) content (mol%) and a molecular weight of 51,178 Da was used. For Compound I implants, poly(PDL-co-DO) (180 mg) and Compound I (120 mg) were dissolved in 10 ml of a 50:50 DCM:chloroform mixture. For EFdA implants, poly(PDL-co-DO) (180 mg) and EFdA (120 mg) were dissolved in 10 ml of DCM in a glass vial. A rotary evaporator was used to completely evaporate the DCM and chloroform over~20 min. The resulting drug/polymer film was lyophilized to remove excess water.
The film (~120 mg) was then loaded into a Teflon mold and baked for 1 h at 90 C under argon protection and atmospheric pressure to form implants. Immediately after baking, the implants were compressed overnight using a stainless steel plunger. The resulting implants were 2 cm in length and approximately 100 mg in weight. The implants were then cut to 1 cm in length and weighed.

| Characterization of drug-loaded implants
For the Compound I implant, the sample was dissolved in 1 ml of DCM, and DCM was evaporated under a steady stream of nitrogen.
For the EFdA implant, the sample was dissolved in 1 ml of chloroform, and 1 ml water was added to it. The mixture was vortexed and let it sit to extract the drug into the water phase. Centrifugation was carried out to separate layer of chloroform and water. The upper water layer containing the sample was removed and lyophilized to dry out the water. The dried pellets were dissolved in ACN for HPLC analysis.
The LODs have been detailed above. Reverse-phase HPLC was used to measure DL (%), defined as the measured mass of Compound I per mass of PLGA NP/implants, and EE (%), which is defined as the ratio of the compounds loaded to the total drugs used for fabricating the NPs/implants as described earlier. 18 Compound I and EFdA implants were also evaluated under an ultra-high-resolution Hitachi scanning electron microscopy (SU7000).
Implants were flash frozen in liquid nitrogen. Implants were then broken in half using tweezers and cross-sectioned using a razor blade to about 1 mm thickness. Samples were placed on a stub using carbon tape, with razor blade edge facing down. Samples were coated with platinum to a thickness of 5 nm using a high resolution sputter coater  University. The animal model used in these studies is the NSG-Hu-PBL as described. 28
The i.p. route was chosen based on our previous studies with Compound I as a free drug solubilized in 10% DMSO/detergent. 18 The dosing was based on in vivo efficacy with free drug and calculated taking into account the amount of each ARV loaded in the NP formulation.

| Implantation procedure
Hu-mice were anesthetized by i.p. injection of ketamine (80 mg/kg) and xylazine (12 mg/kg) in PBS (10 ml/kg) based on individual mouse body weight. After anesthetizing, approximately 1-2 cm square of the dorsal skin was shaved using an electric hair clipper, the area was cleaned with ethanol and disinfected with Betadine. Using a sterile disposable surgical blade, an incision of 4-5 mm was made through the skin. Gently, 2 cm Â 3 cm subcutaneous pockets were created with forceps and implants loaded with Compound I and EFdA were co-inserted at the same site. The opening was sutured using absorbable sutures followed by subcutaneous administration of 0.05 mg/kg buprenorphine. The animals were kept warm using the temperaturecontrolled heating pads until they regained consciousness. Toxicity was evaluated by clinical observations, cage-side observations (twice daily), and body weight (at least weekly).

| In vivo pharmacokinetic studies for implants and NPs
Blood samples were collected at predetermined time points from the ocular venous plexus by retro-orbital venipuncture and serum was used for subsequent HPLC analysis as detailed previously. 18

| HIV-1 challenge experiments in hu-mice
Hu-PBL mice that were i.p. injected with NPs or surgically inserted with implants were i.p. challenged with 30,000 FFU of HIV-1 JRCSF .
The kinetics of virus infection were monitored by weekly measurements of plasma viral loads (PVLs) and peripheral blood CD4 T cell in longitudinal bleeds as described previously. 13 2.11 | Analysis of ARV activity in serum from drug administered mice

| Statistical analysis
The data were plotted as a concentration-time curve using PRISM 9.0 (GraphPad Software Inc., La Jolla, CA). EC 50 was calculated using Graph pad prism where EC 50 of each compound was used along with Hill slope of 1. The predicted area under the curve (AUC) for drug concentration in blood serum against time (AUC predicted) was calculated based on the linear trapezoid method. 29 (Figure 2(a)) and EFdA-NP (Figure 2(b)) exhibited fairly uniform spherical shapes. We undertook to achieve a DL of 10 wt% for each drug in the NP formulation; we attained 9.7 ± 0.8 wt% for Compound I; however, due to the hydrophilicity of EFdA, the resultant DL achieved was 0.8 ± 0.2 wt% (Figure 2(c)).  (Figure 3(d)). While productive infection did occur in the test cohort of mice, the PVLs were 3 log units lower (median 0.72 Â 10 3 copies/ml, range 0.71-5.19 Â 10 3 copies/ml, Figure 3(d)). PVLs fell to below LOQ and were maintained at these levels for 3-4 weeks ( Figure 3(d)). PVLs rebounded in one mouse at Day 35 and translated to a drop in CD4+ T cells by Day 49 (Figure 3(d)). On an average, a reduction of four log 10 was observed in the test cohort compared to the highest PVLs attained in the control group at D14 indicating continued HIV-1 inhibition in these mice. In both mice, >85% of CD4+ T cells were protected throughout the study relative to D0 (Figure 3(e)).

|
PVL AUC was significantly smaller in the test cohort compared to the control group (p = .05; Wilcoxon signed rank test).

| Development of LA implants
The poly(PDL-co-DO) implants were again formulated independently with the EFdA and Compound I because of their distinct physiochemical properties. Both implants were~1 cm long and offwhite in color (Figure 4(a)). Figure 4(b) depicts the physical characteristics of each implant. EFdA loading efficacies in the implant were significantly better than in the NP formulations and the average calculated percent loading per implant was 16.4% for Compound I and 30% for EFdA. The implants were characterized using SEM to assess their morphology and uniformity (Figure 4(c-e)). Compared to the empty polymer implants, which displayed a continuous gray region by SEM, the Compound I-and EFdA-loaded implants had regions that appeared brighter, presumably due to the presence of drug or drug crystals.

| Pharmacokinetics of drug release and antiviral efficacy of Compound I-and EFdA-loaded implants in Hu-mice
Implants loaded with Compound I and EFdA were surgically inserted at 14 days prior to HIV-1 JRCSF challenge (D-14, Figure 5 (a)). The profile of serum drug concentrations in hu-mice harboring the two drug-loaded implants are shown in Figure 5 to maintain viral suppression ( Figure 5(b,c)). For instance, levels  (Table 1).
Overall, there were no significant abnormalities due to the implant, and the majority of observations noted were considered to be incidental, procedure related, or common findings for mice undergoing a surgical procedure. The incision sites healed normally within a few days after surgery. There was no evidence of inflammation, toxicity, or poor tolerability at the implantation sites throughout the duration of the study.
On Day 14 postimplant, the mice were infected with HIV-1 JRCSF .
HIV-RNA was detected in plasma of all exposed mice at the first sampling, a week after infection ( Figure 5(d)). The control cohort implanted displayed high levels of plasma viremia (median 3.24 Â 10 6 copies/ml, range 0.73-19.2 Â 10 6 copies/ml). In contrast, PVLs in the implant group were 3 log units lower (median 3.31 Â 10 3 copies/ml,  Figure 6).

| DISCUSSION
Patient adherence to lifelong HIV therapeutic regimens is limited by the necessity of daily dosing of medications. 34 Therefore, a number of recent research efforts have focused on long-acting, extended release formulations summarized in a recent review. 9 Cabenuva has currently been approved by the FDA. 8 Further advances have been reported in preclinical studies with AIDS drugs formulated as LA, extended release implants. 9 These include an in situ generated implant with DTG as a monotherapy for HIV treatment and prevention. 10  In the current study, we describe the pharmacokinetics and antiviral efficacy of a synergistic combination of an NRTI and NNRTI formulated as LA formulations. The NRTI EFdA was selected due to its exceptional potency and lack of toxicity including mitochondrial toxicity observed with other NRTIs and promising Phase III clinical trials.
The NNRTI, Compound I, is a computationally designed preclinical candidate chosen because of its excellent antiviral efficacy as a LA NP formulation, ADME-Tox and drug resistance profiles as well as synergistic behavior with EFdA as a two-drug combination in cell culture. Nevertheless, the disparate PK parameters observed for LA-NP and LA-implant did not affect the efficacy of these two compounds.  [31][32][33] This is despite plasma drug levels reaching EC50 well within a week, whether this is because it takes longer to achieve stable intracellular and tissue EC50 drug concentrations needs to be investigated. It also needs to be investigated if the rebounding virus was due to viral escape, and the small serum sample volumes did not permit these studies.

| CONCLUSION
The current study using poly(PDL-co-DO) implants or PLGA nanoformulations to deliver a two-drug therapeutic regimen demonstrates the value of two synergistic antiviral compounds to completely suppress viral loads and protect CD4+ T cells. This approach significantly extends previous findings with a long-acting implant containing DTG as monotherapy where the onset of drug resistance was found as early as Day 19 postinfection. 10 Future considerations will include the use of hu-mouse models that will permit evaluation of antiviral efficacy over even longer time periods.

ACKNOWLEDGMENTS
Gratitude is expressed to the National Institutes of Health (R56AI141572 to P. K. and K. S. A., R33AI122384, R01AI145164, and P50AI150464 to P. K.; R01GM49551 and R01AI155072 to K. S. A.; R01AI44616 to W. L. J.) for research support.