TLR7 agonist, N6-LS and PGT121 delayed viral rebound in SHIV-infected macaques after antiretroviral therapy interruption

Toll-like receptor 7 (TLR7) agonist and PGT121 (broadly neutralizing antibody, bnAb) administration previously delayed viral rebound and induced SHIV remission. We evaluated the impact of GS-986 (TLR7 agonist) and dual bnAbs on viral rebound after antiretroviral therapy (ART) interruption. Rhesus macaques inoculated with SHIV-1157ipd3N4 were initiated on daily suppressive ART from Day 14 post SHIV inoculation. Active arm animals (n = 8) received GS-986, N6-LS and PGT121 after plasma viral suppression, starting from week 14. GS-986 induced immune activation and SHIV-specific T cell responses but not viral expression in all the active arm animals. After ART interruption, median time to viral rebound was 6 weeks in the active and 3 weeks in the control arm (p = 0.024). In this animal model, the administration of the combination of GS-986 and dual bnAbs was associated with a modest delay in viral rebound. This strategy should be further evaluated to better understand the underlying mechanisms for the induction of virus-specific immune responses and delay in viral rebound.


Introduction
The HIV reservoir, consisting of cells that harbor latent HIV-1 proviruses, is a major barrier to HIV remission [1,2]. The kick and kill strategy, where latently infected cells are stimulated to induce viral reactivation that can then be eliminated by immune responses, has been proposed as a potential strategy to achieve HIV remission [3,4]. Broadly neutralizing antibodies (bnAb) have demonstrated anti-viral activity in viremic individuals and delayed viral rebound when administered during analytical treatment interruption [5]. Recently, in the study by Borducchi et al., administration of the Toll-like receptor 7 (TLR7) agonist vesatolimod (also known as GS-9620), with the V3 glycan-dependent bnAb PGT121 during antiretroviral therapy (ART) delayed viral rebound following ART interruption in simian-human immunodeficiency virus (SHIV)-SF162P3-infected rhesus macaques [6]. Furthermore, 5/11 animals did not experience viral rebound after ART interruption. Adoptive transfer studies and CD8-depletion studies also did not reveal virus in these animals. These data suggest that the combination of innate immune stimulation with bnAb administration can target and eliminate the viral reservoir. BnAbs used in combination have shown greater antiviral activity than individual bnAbs in SHIV-infected macaques [7] and humans [8]. Thus, we evaluated the impact of the combination of TLR7 agonist GS-986, a very close analog of vesatolimod, and two bnAbs targeting different regions of the HIV envelope-CD4 binding site by N6-LS [9] and V3 glycan by PGT121 [10] in delaying viral rebound during ART interruption in rhesus macaques that were initiated on viral suppressive ART 14 days post SHIV-1157ipd3N4 infection.

Plasma and tissue levels of N6-LS and PGT121
Animals in the active arm (n = 8) received intravenous N6-LS (30 mg/kg) and PGT121 (10 mg/kg) every 2 wks from wks 24 to 32, unless anti-drug antibodies (ADA) was detected ( Fig 1A). Due to the development of ADA, animals received 7-10 doses of GS-986, 2-5 doses Plasma N6-LS and PGT121 levels were monitored weekly. Median (range) peak plasma levels ( Fig 4B). The rate of decline in plasma N6-LS and PGT121 levels varied between the animals (Fig 4C). The median time from the last administration of bnAb to plasma level <0.25 ug/mL was 7 wks (range 2-32) for N6-LS and 18.5 wks (range 4-24) for PGT121. ART was interrupted 4 wks after plasma levels of both N6-LS and PGT121 were <0.25 ug/mL. Immediately prior to ART interruption, N6-LS and PGT121 were also not detectable in samples from lymph node ( Fig 4D) and colon ( Fig 4E).

Impact of the combination of GS-986, N6-LS and PGT121 on SHIVspecific T cell responses
Interestingly, we observed an induction in SHIV-specific responses in some animals after GS-986 administration. By the time of ART interruption, IFNγ+ Gag-specific CD4 and CD8 T cells in the active arm were significantly higher when compared to wk14 (Pre GS-986, p = 0.0469) as well as to controls (p = 0.0232, p = 0.0126 respectively, Fig 5). A similar trend was observed for IFNγ+ Env-specific CD4 and CD8 T cells ( Fig 5). These data suggest that the combination of GS-986, and dual bnAbs increased SHIV-specific T cell responses.

Impact of the combination of GS-986, N6-LS and PGT121 on Viral DNA levels
Median SHIV DNA at wk 2 was 10536 (IQR 4930-17275) copies/million PBMC. SHIV DNA levels reduced significantly after ART initiation, to median 323 (IQR 148-550) copies/million PBMC, p<0.0001 at wk 14 and continued to reduce between wks 14 and 24 in both arms. SHIV DNA levels were not significantly different between the arms at all time points measured ( Fig 6A). Strong correlations were seen between SHIV RNA levels at wk 2 (pre-ART) with SHIV DNA levels at wk 2 (spearman r 0.85, p<0.0001), wk 14 (spearman r 0.83, p = 0.0002), wk 24 (spearman r 0.58, p = 0.020) and at ART interruption (spearman r 0.71, p = 0.003), suggesting that the level of peak viremia is correlated with the size of the reservoir before and during ART (Fig 6B).

Delay in viral rebound following ART interruption
ART was ceased 4 wks after plasma levels of both N6-LS and PGT121 were <0.25 ug/mL, ranging from study wk 34-  animals experienced plasma viral rebound after ART interruption. The median (IQR) time to viral rebound was 3 (2.5-5.5) wks in the control arm and 6 (4.6-6.9) wks in the active arm, p = 0.024 (Fig 7A and 7B). There were no differences in post-rebound peak (Fig 7C) viremia between arms. Though more animals in the active arm had lower post-rebound set-point viral load, this did not achieve statistical significance (Fig 7D). There were no correlations between time to rebound with SHIV RNA at wk 2 (S1C

No detectable adverse effects of GS-986, N6-LS and PGT121 in the central nervous system (CNS)
Monoclonal antibodies are known to penetrate the CNS poorly due to the blood-brain barrier. Thus, we examined the potential for GS-986 to induce viral expression that is not targeted for elimination due to inadequate levels of bnAb. No increases in CSF markers of immune activation were seen at wk 24 (2 wks after the 5 th dose of GS-986) or immediately prior to ATI, when compared with pre-infection levels (Fig 8). Furthermore, CSF SHIV RNA was undetectable in

PLOS PATHOGENS
TLR7 agonist, N6-LS and PGT121 delayed viral rebound all animals at all sampling time-points after ART initiation (Fig 3). At 12 wks post rebound, CSF SHIV RNA was only detectable at low levels in 1 active and 1 control animal. These data suggest that administration of GS-986, N6-LS and PGT121 had no measurable negative impact on immune activation or reservoir size in the CNS.

Discussion
This study demonstrated that the combination of TLR7 agonist and dual bnAbs delayed viral rebound after ART interruption by 2-fold in rhesus macaques that initiated viral suppressive ART 14 days after SHIV-1157ipd3N4 inoculation. This approach may represent a potential strategy to target the HIV reservoir in HIV-infected individuals.

PLOS PATHOGENS
TLR7 agonist, N6-LS and PGT121 delayed viral rebound post GS-986 dosing or in the tissues, including lymph node, colon and CSF 2 wks after the 5 th as well as final administration of GS-986. These results are consistent with recent studies performed in both SHIV [6] or SIV-infected [11,13,14] macaques on ART or in HIV-infected individuals [15]. The absence of detectable plasma viral reactivation does not preclude transient viral reactivation in tissue sites outside of sampling time points.
The induction of Gag-specific T cells is of importance as these cells are essential in viral control [16][17][18][19] and are thought to be key to HIV remission [20,21]. The mechanisms underlying this could not be clearly delineated in this study. A potential postulation is that GS-986 administrations induced transient low-level viral expression, stimulating immune responses either directly or through binding to bnAbs and the generation of a vaccinal effect. These mechanisms may have contributed to the delayed viral rebound in the active arm in this study.
To avoid confounding anti-viral effects from bnAbs on time to viral rebound, ART was interrupted 4 wks after plasma levels of both N6-LS and PGT121 were <0.25 ug/mL. This resulted in variations in the timing of ART interruption between the animals. Though no statistically significant differences in duration of viral suppression or timing of ART interruption were identified, the variation in ART duration may pose as a limitation of this study. Importantly, our data showed that the administration of a combination of TLR7 agonist and dual bnAbs to animals that initiated ART early in the course of SHIV infection was associated with a delay in viral rebound after ART interruption, corroborating findings by Borducchi et al. [6]. In this study, ART was initiated 14 days post SHIV inoculation. The MHRP RV217 prospective cohort involving seronegative high-risk individuals who underwent twice-weekly HIV-1 RNA testing estimated the eclipse phase (the time between HIV-1 infection and a diagnosable infection by nucleic acid testing) to be a week [22]. Thus ART initiation at day 14 more closely mirrors what is feasible in HIV-infected individuals, including our current clinical RV254 cohort in Thailand, where individuals are diagnosed and initiated on ART at the earliest stages of HIV infection [23]. This one week difference in ART initiation resulted in a 1.5 log 10 copies/mL increase in median pre-ART SHIV RNA levels: 5.7 log 10 copies/mL in this study vs 4.2 log 10 copies/mL when ART was initiated at 7 days post infection and pre-ART SHIV RNA level was identified as a correlate of time to viral rebound [6]. In addition, the timing of administration of TLR7 agonist was also different (12 wks after ART initiation in this study, before the viral reservoir reached a steady-state vs 95 wks). The earlier initiation and longer duration of ART in the Borducchi et al study likely translated to a smaller reservoir, with viral DNA being largely undetectable in PBMC (vs median of 323 copies/million PBMC in this study) at the time the interventions were administered [6]. The smaller reservoir may thus be more amenable to elimination, allowing for some NHP to achieve remission when initiating ART 7 days post infection.
The number of doses of bnAbs administered (ranging from 2-5) were limited by the development of ADA. The development of anti-N6-LS antibody is not uncommon based on previous NHP studies (Dr. John Mascola, personal communication) and may have been further exacerbated by the high dose (30 mg/kg) administered. We hypothesize that induction of anti-N6-LS antibody led to the development of cross reactive anti-PGT121 antibody, likely to the common Fc regions. The shorter duration of bnAb exposure may also have reduced the efficacy of the TLR7 agonist and dual bnAb strategy. In a recent study by Barouch et al., in macaques that initiated ART 1 year after SHIV-SF162P3 infection and were virologically suppressed for 2.5 years, the administration of 10 infusions of PGT121 or GS-9721 (Fc-modified version of PGT121) with TLR7 agonist prevented viral rebound in 41% (7 of 17) of animals following ART interruption [24]. Finally, intrinsic differences between SHIV-1157ipd3N4 and SHIV-SF162P3 may also have contributed to differences in impact on viral rebound between the 3 studies.
In summary, the administration of the combination of TLR7 agonist (GS-986) with N6-LS and PGT121 to macaques that initiated ART 14 days after SHIV-1157ipd3N4 infection was associated with a modest delay in viral rebound, despite limited dosing of bnAbs due to ADA. The evaluation of this strategy in humans, where the development of xenogeneic anti-drug antibodies will not be an issue, would allow the assessment of the impact of extended dosing.

Ethics statement
Animals were housed at the AAALAC International-accredited, Armed Forces Research Institute of Medical Science (AFRIMS; Bangkok, Thailand). The study was approved by the AFRIMS Institutional Animal Care and Use Committee under protocol number PN17-01. Research was conducted in compliance with Thai laws, the Animal Welfare Act and other U.S. federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, 2011 edition [25].

Study design
Sixteen male, adult, Indian-origin rhesus macaque (Macaca mulatta), pre-screened to exclude protective MHC alleles (MamuA � 01, B � 08 and B � 17), were inoculated intrarectally with SHIV-1157ipd3N4 (an R5-tropic, mucosally transmissible virus, constructed using SIVmac239 backbone, encoding an HIV subtype C env derived from a Zambian infant [26], at 3.9 x10e 7 RNA copies, at week 0. Animals were randomly assigned into active or control arms whilst balancing Trim 5 alpha status, day 14 plasma SHIV RNA, weight and age. Daily ART (9-(2-phosphonomethoxypropyl) adenine, PMPA 20 mg/kg, emtricitabine 40 mg/kg and dolutegravir 2.5 mg/kg), was administered subcutaneously, from Day 14, as previously described [27]. Animals in the active arm (n = 8) received GS-986 (0.1 mg/kg) via oral gavage every 2 wks from wks 14 to 32 and intravenous N6-LS (30 mg/kg) and PGT121 (10 mg/kg) every 2 wks from wks 24 to 32, unless anti-drug antibodies (ADA) were detected. Animals were monitored for the development of ADA to N6-LS and PGT121 seven days after each dose. The administration of the respective bnAb was suspended after detection of ADA. ART was ceased when plasma levels of N6-LS and PGT121 were <0.25 ug/mL for 4 wks. Animals in the control arm (n = 8) received intravenous normal saline infusions at wk 24-32 and ceased ART at wk 40. During ART interruption, plasma SHIV RNA was assessed twice weekly to monitor for viral rebound. Animals were humanely euthanized 12 wks post viral rebound.

Quantitation of SHIV RNA
Plasma, CSF and tissue SHIV-RNA levels were measured using real-time quantitative PCR as previously described [28,29]. Limit of quantification of the assay was 10 copies/mL.

Quantitation of total SHIV DNA
Total cellular DNA was isolated from~5x10 6 PBMCs by proteinase K (Invitrogen, Carlsbad, California, USA) lysis. Viral DNA was measured by real-time qPCR in triplicate, and the number of copies was calculated based on parallel quantitation of a standard dilution of 3D8 cells containing a single copy of integrated SIV genomic DNA. qPCR was performed using TaqMan Fast Advanced Master Mix (Applied Biosystems, Vilnius, Lithuania), according to the manufacturer's instructions, with SIV gag qPCR primer and probe sequences as previously described [30][31][32]. SIV gag copy numbers were normalized to rhesus albumin gene (alb) copy numbers detected by qPCR using a 10-fold dilution of PBMC from a healthy control rhesus macaque as the standard, as follows: (gag copies)/(alb copies) X 2(alb copies)/cell.

Measurement of plasma bNAb levels
Plasma N6-LS and PGT121 concentrations were measured by ELISA as previously described [33]. Maxisorp ELISA plates (Thermo Fisher Scientific, Roskilde, Denmark) were coated with anti-idiotype Ab for N6-LS or PGT121 at 2 ug/mL, 100 uL/well, overnight at 4˚C. Plates were then washed with PBS (Sigma Aldrich, St. Louis, Missouri, USA) -0.05% Tween 20 (Amresco, Solon, Ohio, USA) and blocked for 1 h at room temperature with Tris-Buffered Saline (TBS, Sigma Aldrich, St. Louis, Missouri, USA) with 5% nonfat dried milk powder (AppliChem, Damstadt, Germany) and 2% bovine serum albumin (BSA, Sigma Aldrich, St. Louis, Missouri, USA) and then washed. Standard curve calibrators (8 serial 2.5-fold dilutions of N6-LS or PGT121), spiked plasma control (10 ug/mL) and heat-inactivated plasma samples (3 serial 5-fold dilutions) were plated in duplicate and incubated for 1 h at room temperature. Plates were washed and incubated with donkey anti-human IgG (Fcγ-fragment specific)-HRP (Jackson Immunoresearch, Chester, Pennsylvania, USA) diluted to 1:10000 in blocking buffer for 30 mins at room temperature. Plates were washed again and developed with KPL SureBlue TMB Microwell Peroxidase Substrate (Seracare, Gaithersburg, Maryland, USA) for 20 mins followed by the addition of 1N Sulfuric acid (Fisher Chemical, Bergen County, New Jersey, USA). Plates were read at 450 nm on a SpectraMax microplate reader (Molecular Device, Sunnyvale, California, USA) using Softmax Pro version 4.3.1 software. The SoftMax Pro software calculated 4-Parameter curve fits for the standard calibrators and the test sample concentrations were determined by interpolation into the calibration curves.

Measurement of lymph node and gut bNAb levels
Sigmoid colon and lymph node samples were thawed on ice, homogenized for 30 seconds in 200 μl elution buffer (PBS and EDTA-free complete protease inhibitor, Roche Diagnostics GmbH, Mannheim, Germany). The homogenate was cleared by centrifugation for 15 min at 4˚C and the supernatant was filtered (0.22 μm; Corning, NY, USA). Levels of N6-LS and PGT121 antibodies were measured by ELISA, as above. The main difference was that tissue homogenate was diluted 2-fold with blocking buffer.

Detection of ADA
Plasma anti-N6-LS and anti-PGT121 antibodies were measured by ELISA as previously described [34]. Maxisorp ELISA plates (Thermo Fisher Scientific, Roskilde, Denmark) were coated with N6-LS or PGT121 at 2 ug/mL, 100 uL/well, overnight at 4˚C. Plates were then washed with PBS-0.05% Tween 20 and blocked for 1 h at room temperature with TBS with 5% nonfat dried milk power and 2% BSA and then washed. Heat-inactivated plasma samples (7 serial 5-fold dilutions) were plated in duplicate and incubated for 1 h at room temperature.
Plates were washed. Wells were incubated with mouse anti-monkey IgG-HRP (Southern Biotechnology Associates, Birmingham, Alabama, USA) diluted to 1:8000 in blocking buffer for 30 mins at room temperature. Plates were washed again and developed with KPL SureBlue TMB Microwell Peroxidase Substrate for 20 mins followed by the addition of 1N Sulfuric acid. Plates were read at 450 nm on a SpectraMax microplate reader. The presence of ADA was defined to be >10-fold increase in end-point titer.
performed using Wilcoxon matched-pairs signed rank test. Comparisons between animals in the 2 arms were performed using two-sided Mann-Whitney tests. Correlations were assessed by two-sided Spearman rank-correlation tests. Analysis of time to viral rebound was performed using Log-rank (Mentel-Cox) test.