Research ArticleAIDS/HIVImmunologyInfectious disease
Open Access |
10.1172/jci.insight.208375
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Fonseca, J. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by King, A. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Davis, K. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by O’Hagan, D. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Khoei, A. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Alves Britto Da Costa, L. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Cockerham, C. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Cottrell, M. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Ehnert, S. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by
Wood, J.
in:
PubMed
|
Google Scholar
|
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Alpert, M. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Gardner, M. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Lifson, J. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Mavigner, M. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by Martins, M. in: PubMed | Google Scholar
1Department of Pediatrics, Emory University School of Medicine, Atlanta, Georgia, USA.
2Center for Childhood Infections and Vaccines, Children’s Healthcare of Atlanta and Emory University, Atlanta, Georgia, USA.
3Department of Immunology and Microbiology, The Herbert Wertheim UF Scripps Institute for Biomedical Innovation & Technology, Jupiter, Florida, USA.
4Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
5Emory National Primate Research Center, Emory University, Atlanta, Georgia, USA.
6Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, USA.
7Emmune Inc., Cambridge, Massachusetts, USA.
8AIDS and Cancer Virus Program, Frederick National Laboratory for Cancer Research, Frederick, Maryland, USA.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Find articles by
Chahroudi, A.
in:
PubMed
|
Google Scholar
|
Published September 24, 2026 - More info
Early antiretroviral therapy (ART) limits viral reservoir establishment in infants but also constrains virus-specific immunity required to clear reactivated cells. In the early ART context, we evaluated whether combining adeno-associated virus serotype 9–vectored (AAV9-vectored) eCD4-IgG1 with pharmacological latency reversal using the second mitochondria-derived activator of caspases (SMAC) mimetic AZD5582 could promote reservoir reduction and control in SIV-infected infant macaques. AAV9 delivery achieved sustained eCD4-IgG1 expression, and AZD5582 induced on-ART viremia, but the combination did not reduce intact SIV proviral DNA relative to controls. Partial postintervention control of viremia during analytical treatment interruption (ATI) occurred in 2 of 6 infants receiving AAV9-eCD4-IgG1 plus AZD5582 with restricted reservoir expansion during recrudescence at week 13. Intact reservoir size during ATI correlated inversely with on-ART viremia during AZD5582 treatment. Animals were classified as controllers or non-controllers. Controllers did not show increased eCD4-IgG1 expression nor increased on-ART viremia during AZD5582 treatment but harbored less pre-ART intact SIV DNA in PBMCs than non-controllers. Controllers also exhibited higher frequencies and greater polyfunctionality of virus-specific CD8+ T cells prior to ATI. These findings suggest immune control of reservoir size and post-ART viral dynamics in early-treated infants and that AAV9-eCD4-IgG1 plus AZD5582 increases the likelihood of post-ART viral control.
In 2024, approximately 120,000 children under 15 years of age acquired HIV through vertical transmission (1). Early-life infection progresses rapidly, with 45%–59% mortality by age 2 in untreated children, who account for 12% of AIDS-related deaths despite comprising only 3% of HIV cases (2). Although antiretroviral therapy (ART) reduces mortality, it depends on strict lifelong adherence, often hindered by access, cost, toxicity, and comorbidities (3, 4), with treatment interruption resulting in viral rebound and disease progression (5–8). CD4+ T cells harboring replication-competent virus that persist on ART (the reservoir) drive rebound and represent the central obstacle to a cure (9, 10).
Combined latency reversal and clearance strategies seek to eliminate the reservoir by pharmacologically inducing viral expression in persistently infected CD4+ T cells, thereby exposing them to immune-mediated clearance (11). Early-treated infants are a particularly compelling target population. ART initiation within 48 hours to 12 weeks of life dramatically limits reservoir seeding, yielding up to 6-fold fewer persistently infected cells than in infants treated after 12 weeks (12, 13), a theoretically more tractable target for reactivation and clearance. Paradoxically, the same early ART that constrains the reservoir also blunts HIV-specific immunity by eliminating the antigenic stimulus required to develop and sustain effector responses capable of recognizing and clearing cells with reactivated viral expression (14), making adjunct immunotherapy essential to any cure strategy in this population.
Among candidate latency reversal agents, the second mitochondrial activator of caspases (SMAC) mimetic AZD5582 reverses HIV and SIV latency through activation of the noncanonical NF-κB pathway (ncNF-κB) (15). AZD5582 monotherapy does not reduce reservoir size (15), but combining it with SIV-specific mAbs significantly reduced total and replication-competent SIV DNA in lymph node CD4+ T cells, establishing proof-of-concept for reservoir depletion through coordinated latency reversal and immune clearance (16). Nonetheless, translation to pediatric HIV/SIV remains challenging given that infant macaques exhibit diminished AZD5582-induced latency reversal, associated with lower plasma drug exposure and reduced ncNF-κB inducibility in CD4+ T cells (17), underscoring a need for age-specific optimization.
The immunoadhesin eCD4-Ig is a promising partner for latency reversal strategies. This fusion protein combines CD4 domains 1 and 2 with an Ig Fc domain and a tyrosine-sulfated peptide mimicking the CCR5 N-terminus (18), enabling cooperative binding to conserved regions of the HIV envelope glycoprotein by mimicking both receptor and coreceptor (18). Because these regions of Env are essential for viral entry, eCD4-Ig binding makes immune escape extremely difficult (18, 19). This design confers exceptionally broad and potent neutralization of HIV-1, HIV-2, and SIV, with geometric mean IC50 values below 0.05 μg/mL (18). eCD4-Ig also binds to Env-expressing infected cells, against which it can mediate antibody-dependent cellular cytotoxicity (ADCC) (20). Adeno-associated virus (AAV)-vectored delivery of eCD4-Ig achieved sustained expression in vivo and protected adult rhesus macaques from repeated SHIV challenge for more than 40 weeks despite anti-drug antibody (ADA) formation (18). The tolerant infant immune system may further favor prolonged AAV transgene expression by limiting ADA generation (21). Together, these features support evaluation of AAV-delivered eCD4-Ig in combination with AZD5582 to promote immune-mediated reservoir clearance in early-treated infants.
We therefore combined AZD5582-mediated latency reversal with AAV9-delivered eCD4-IgG1 (Emm058) during SIV infection and ART treatment of infant rhesus macaques. Although this strategy did not reduce reservoir size, it provided a controlled framework in which to interrogate the virological and immunological factors associated with post-ART outcomes, including partial postintervention control in 2 animals. We found that postintervention controllers had smaller reservoirs during both the pre-ART and ART interruption periods, accompanied by greater and more polyfunctional virus-specific CD8+ T cell responses.
Study design and cohort characteristics. To evaluate a combined latency reversal and immune-mediated clearance strategy, we orally infected 20 infant macaques with SIVmac251 and initiated ART at week 3 after infection (Figure 1). Intervention animals (n = 10) received AAV9-delivered eCD4-IgG1, a broadly neutralizing immunoadhesin targeting the viral envelope, at ART initiation, followed by the SMAC mimetic AZD5582. Control animals (n = 10) received ART alone. eCD4-IgG1 administration timing was selected based on evidence that the first 4 weeks after birth are optimal for AAV-mediated antibody gene transfer, yielding the lowest frequency of ADA responses in rhesus macaques (21). The rh-eCD4-IgG1 transgene encoded a short-hinge variant designed to avoid misfolding from an unpaired hinge cysteine (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.208375DS1).
Figure 1Study design and intervention regimen. Twenty infant rhesus macaques were orally challenged with SIVmac251 at 4 weeks of age and randomly assigned to intervention or control groups (n = 10 per group), and then allocated to a dose optimization cohort (n = 8; 4 per group) or an ATI cohort (n = 12; 6 per group). All time points are weeks after infection. All animals initiated ART at week 3 (gray shading); plasma viral load (PVL) was monitored weekly through week 6 and biweekly thereafter. Intervention animals received AAV9 encoding eCD4-IgG1 (2.5 × 1012 genome copies kg–1) at week 3, followed by 6 weekly doses of the latency reversal agent AZD5582 beginning at week 40; control animals received ART alone. Dose optimization animals were euthanized during or after AZD5582 administration, with time-matched control animals euthanized at corresponding time points. In the ATI cohort, ART was interrupted 4 weeks after completion of the AZD5582 regimen (tan shading); animals were euthanized a mean of 18 weeks after ATI, with paired control animals euthanized at matched time points. Created with BioRender.com.
The study employed a sequential 2-cohort design. An initial dose optimization cohort (n = 4 intervention, n = 4 control) established a tolerable AZD5582 regimen, which was then applied to an analytical treatment interruption (ATI) cohort (n = 6 intervention, n = 6 control). AZD5582 dosing began at week 40 after infection in both cohorts; however, only the ATI cohort discontinued ART at week 50 to assess the impact of the interventions on post-ART viral control. Groups did not differ significantly in any baseline characteristic including age at infection, sex, Mamu-A*01 status, number of SIV inoculations, and pre-ART viral loads (Table 1 and Supplemental Table 1).
AZD5582 dose optimization. Prior studies showed that the AZD5582 dose sufficient for latency reversal in adult macaques was less effective in infants (17). Pharmacokinetic (PK) modeling predicted that an infant dose of 0.3 mg/kg would achieve a peak plasma concentration (Cmax) and total drug exposure (AUC) comparable to those observed in adults with the 0.1 mg/kg dose (Supplemental Figure 2). Because the model was built on idealized infant macaque parameters and could not fully account for biological variability in key parameters such as volume of distribution, we established a dose optimization cohort to evaluate tolerability prior to the ATI study. Weekly dosing at 0.3 mg/kg (n = 2) or 0.25 mg/kg (n = 2) produced severe cumulative toxicity, including systemic inflammation, respiratory failure, and multiorgan injury (Supplemental Table 2), leading to euthanasia (n = 3) or treatment discontinuation (n = 1) after 5–7 doses, indicating that repeated exposure-matched dosing was not tolerated in infants, possibly due to cumulative immune activation.
We therefore implemented a modified regimen consisting of a 0.25 mg/kg loading dose followed by 5 weekly maintenance doses of 0.2 mg/kg. Although this regimen was not predicted to precisely match the exposures at 0.3 mg/kg, modeled Cmax and AUC remained within the adult target range (Supplemental Figure 2). This regimen, selected for tolerability rather than precise adult PK matching, was well-tolerated in all ATI cohort animals (n = 6) without severe adverse events over the treatment course.
AAV9-mediated eCD4-IgG1 expression. To minimize the impact of preexisting maternally transferred AAV9 antibodies on eCD4-IgG1 expression, infant macaques were screened at 1–2 weeks of age for AAV9-specific neutralizing antibodies. Of 60 infants evaluated, 41 exhibited less than 50% neutralization and were eligible for enrollment (Figure 2A). Three weeks after infection, concurrent with ART initiation, intervention animals across both cohorts (n = 10) received intramuscular co-formulated AAV9-rh-eCD4-IgG1 and AAV9-TPST-2, while control animals (n = 10) received ART alone. TPST-2 promotes eCD4-IgG tyrosine sulfation required for optimal gp120 engagement (22). Plasma eCD4-IgG1 concentrations were monitored weekly for 3 weeks and biweekly thereafter to assess transgene expression kinetics and durability.
Figure 2AAV9-mediated eCD4-IgG1 expression and effects on plasma viremia. (A) Anti–AAV9-neutralizing antibody screening at birth by HEK293T transduction inhibition (1:10 serum dilution). Each symbol represents an individual animal (n = 60 screened). Animals enrolled in the study (n = 20) are shown in orange and the red dashed line denotes the acceptance threshold (<50% neutralization). (B and C) Longitudinal serum eCD4-IgG1 concentration (B) and anti-drug antibody responses (C) in intervention animals (n = 10) after AAV9-eCD4-IgG1 administration at week 3 after SIV infection. Solid green and black lines indicate the mean eCD4-IgG1 concentration and ADA response, respectively, with shaded areas indicating ± SEM with individual animal trajectories in gray. (D and E) Plasma SIV RNA kinetics on a log10 scale in intervention (D) and control (E) animals (n = 10 per group) from infection through ART-mediated viral suppression. Solid blue and orange lines indicate the mean plasma SIV RNA levels in the intervention and control groups, respectively, with corresponding shaded areas indicating ± SEM. Gray lines represent individual animal trajectories. Gray shading denotes the ART period, and the dotted horizontal line indicates the assay limit of detection (60 copies/mL). The dashed vertical line in D indicates the time of AAV9-eCD4-IgG1 administration. (F) Fold difference in plasma viral load between groups, shown on a log10 ratio scale with axis labels expressed as signed fold change (negative values denote lower viremia in the intervention group; the reference value is ± 1). Open symbols are fold differences of geometric means ± pooled SEM of the log10 difference. Mean serum eCD4-IgG1 concentration (green, ± SEM) is overlaid on the right y axis. (G) Model-estimated plasma viral load trajectories derived from mixed-effect models with natural cubic splines. Lines represent estimated marginal means of log10 plasma viral load for intervention (blue) and control (orange) animals, with shaded areas indicating 95% CIs. Yellow highlighting denotes time intervals in which viral load was significantly lower in the intervention group after FDR correction (P < 0.05).
AAV9-eCD4-IgG1 administration produced rapid transgene expression in all intervention animals (Figure 2B and Supplemental Figure 3). eCD4-IgG1 became detectable between 4 and 6 weeks after infection (mean 4.6 ± 0.84 weeks) and increased gradually thereafter, reaching a plateau by approximately 17 weeks after infection. Plateau onset was identified by fitting linear regression models to log10-transformed cohort mean plasma eCD4-IgG1 concentrations over time and by defining the earliest week at which the regression slope was no longer significantly different from zero over the subsequent observation window. Stable eCD4-IgG1 concentrations (geometric mean 9.97 μg/mL; geometric SD 1.97) were maintained from week 17 through the last follow-up (weeks 44–70 after infection). Pharmacokinetic properties at the individual animal level are summarized in Supplemental Table 3.
Anti–eCD4-IgG1 responses were detected in 7 of 10 animals (Figure 2C and Supplemental Figure 3), with substantial inter-animal variability in peak ADA magnitude (range 0–0.86; median 0.035 absorbance at 450 nm [A450]). ADA responses were predominantly low in magnitude; only 2 animals exhibited peak ADA levels greater than 0.3 A450 (RTp23 and RVp23), and only 1 animal (RTp23) showed persistent ADA expression, defined as detectable ADA at 2 or more consecutive measurements. In RTp23, plasma eCD4-IgG1 fell to undetectable levels during persistent ADA and recovered after ADA response resolution (Supplemental Figure 3). ADA responses in the remaining animals were transient or of insufficient magnitude to be associated with reduced eCD4-IgG1 levels.
Kinetics of viral suppression with ART with or without AAV9-eCD4-IgG1. Plasma viral loads peaked at 2–3 weeks after infection with similar median peak viremia between groups (intervention 7.28, control 7.23 log10 copies/mL; P = 0.943). Viral kinetics from infection until suppression are shown in Figure 2, D and E, and Supplemental Figures 4 and 5. To assess whether eCD4-IgG1 expression influenced viremia after ART initiation, fold-differences in plasma viral loads between groups from week 3 to week 36 after infection were examined (Figure 2F). At baseline prior to ART (week 3), plasma viral loads were comparable between groups (log10 difference: –0.013 ± 0.28; fold difference: 1.03). After the emergence of detectable plasma eCD4-IgG1, intervention animals exhibited lower viremia than controls, with the greatest difference observed at week 8 (log10 difference: –0.76 ± 0.17; 5.8-fold lower).
To more rigorously characterize the effect of circulating eCD4-IgG1 on viremia, we employed longitudinal mixed-effect modeling with natural cubic splines to capture nonlinear viral decay kinetics while accounting for inter-animal variability. Time was the dominant effect (F4,352.10 = 534.28, P < 0.001), reflecting progressive ART-mediated suppression in both groups. There was no evidence for a sustained overall difference in plasma viral loads between the groups across the postintervention to week 36 period (main effect of group, P = 0.369), although the group × time interaction was suggestive of diverging viral load trajectories (F4,352.10 = 2.30, P = 0.059; Supplemental Table 4). FDR-corrected post hoc comparisons of model-estimated marginal means identified a transient period of accelerated viral decay in eCD4-IgG1 recipients during weeks 5–12 after infection (Figure 2G), with the largest difference at week 10 (0.54 log10; 3.5-fold lower than controls; 95% CI 1.39–8.62; FDR-adjusted P = 0.008). Differences diminished by week 14 and remained nonsignificant thereafter (all FDR-adjusted P > 0.20), consistent with convergence of viral suppression under continued ART.
AZD5582-induced latency reversal. After sustained viral suppression, intervention animals (n = 10) received AZD5582 to assess SIV latency reversal in the presence of circulating eCD4-IgG1 and during continued ART. At baseline (4 weeks before AZD5582 administration), median plasma viral loads were comparable between groups (intervention: 13.0 copies/mL, IQR 11.0–22.5; control: 13.0 copies/mL, IQR 3.0–45.5; Figure 3A). Latency reversal was defined as viremia exceeding 60 copies/mL at any post-baseline time point measured by ultrasensitive assay.
Figure 3AZD5582 induces latency reversal in infant macaques during suppressive ART. (A) Longitudinal ultrasensitive plasma viremia in AZD5582-treated intervention (n = 10) and control (n = 10) infant macaques over weeks relative to the first AZD5582 dose (vertical red line). Plasma viral load is plotted on a log10 scale. Solid blue and orange lines indicate group means for intervention and control animals, respectively, with corresponding shaded areas indicating ± SEM. The dotted horizontal line denotes the assay limit of detection (3 copies/mL). (B) Per-animal geometric mean ultrasensitive plasma viremia across AZD5582-phase time points from baseline through week 7. Each symbol represents an individual animal; groups were compared using a 2-tailed Mann-Whitney U test. (C) Week-by-week fold difference in on-ART viremia between intervention and control groups. Symbols indicate ratios of geometric mean viral loads, with 95% CIs derived from Welch’s unequal-variance t test on ln-transformed values and back-transformed to the ratio scale; consequently, CIs are asymmetric around the point estimates. Values above the dotted horizontal line at y = 1 indicate higher viremia in the intervention group, consistent with latency reversal. Time points with P < 0.05 are indicated.
After AZD5582 administration, latency reversal (>60 copies/mL) was observed in 8 of 10 intervention animals (80%) compared with spontaneous viral blips above the same threshold in 7 of 10 untreated controls (70%) sampled at equivalent time points. However, latency reversal events were significantly more frequent in intervention animals, occurring at 42.1% of post-baseline time points (24/57) versus 20.3% in controls (14/69; χ2 test, P = 0.014). Notably, 36% of control blips (5/14) originated from a single animal (RQl23), suggesting that spontaneous rebound in the control group was largely driven by 1 outlier. Geometric mean on-ART viremia across the AZD5582 treatment period was significantly higher in the intervention than in the control group (48.9 vs. 12.7 copies/mL; P = 0.0147; Figure 3B). Week-by-week analysis similarly demonstrated higher geometric mean viremia in AZD5582-treated animals (Figure 3C), with marked differences at week 1 (fold increase: 9.45, 95% CI 1.51–59.24; P = 0.019), 2 (fold increase: 11.77, 95% CI 2.10–65.88; P = 0.008), and 4 (fold increase: 6.79, 95% CI 1.56–29.56; P = 0.013). Within the intervention group, the dose optimization cohort exhibited higher peak viral loads (median 1,633 copies/mL; IQR 65–3,800) and a higher incidence of latency reversal (4/4 infants) than the ATI cohort (peak viremia median 215 copies/mL; IQR 83–708; incidence of latency reversal 4/6 infants), consistent with higher AZD5582 doses leading to greater viral reactivation.
Partial postintervention control after ATI. Four weeks after the final AZD5582 dose, animals underwent ATI. Plasma SIV RNA was measured 3 times weekly during the first 5 weeks and weekly thereafter through week 20. All animals developed detectable viral rebound within 2–3 weeks (Figure 4, A and B). Pre-ATI frequencies of intact proviruses in blood, lymph node, and rectal mononuclear cells were similar between the groups (Supplemental Figure 6).
Figure 4Virological rebound and reservoir dynamics after analytical treatment interruption. (A and B) Longitudinal plasma SIV RNA after ATI in intervention (A) and control (B) animals, plotted on a log10 scale. Gray lines indicate individual animal trajectories; the 2 intervention animals exhibiting partial postintervention viral control are shown in red in A. Solid blue and orange lines indicate group means for intervention and control animals, respectively, with corresponding shaded areas indicating ± SEM. The dotted horizontal line denotes the assay limit of detection (60 copies/mL). (C) Per-animal geometric mean plasma SIV RNA across the ATI period (weeks 0–20), grouped by treatment and controller status: controllers (red), intervention non-controllers (filled blue), and control non-controllers (open orange). Groups were compared using a 2-tailed Mann-Whitney U test. (D) Kaplan-Meier estimates of time to detectable viral rebound, defined as plasma SIV RNA >60 copies/mL. Groups were compared using the log-rank (Mantel-Cox) test; the hazard ratio for intervention relative to control was estimated using the observed-minus-expected method. (E) Per-animal geometric mean plasma SIV RNA across the ATI period, grouped by controller status: controllers (red) and non-controllers (filled black, intervention; open black, control). (F) Intact SIV proviral DNA in PBMCs at week 13 after ATI, expressed as intact proviral DNA assay copies per 106 PBMCs. (G) Fold-change in intact SIV proviral DNA from pre-ATI baseline to week 13 after ATI; the dotted horizontal line indicates no change (fold change = 1). In A–D, n = 6 animals per group; in E–G, n = 2 controllers and n = 10 non-controllers. Horizontal lines indicate group geometric means in C and E and group medians in F and G.
To evaluate the group-level effect of combined AAV9-eCD4-IgG1 and AZD5582 treatment on viral burden during ATI, we compared each animal’s geometric mean plasma viral load over weeks 0–20. No difference was detected between the intervention and control groups (Figure 4C; Mann-Whitney P = 0.82). Time to detectable rebound, defined as plasma SIV RNA greater than 60 copies/mL, likewise did not differ (median 14 days in each group; log-rank P = 0.44; HR 1.42, 95% CI 0.45–4.47; Figure 4D).
Although no group-level effect was detected, individual trajectories varied. All 6 control animals maintained plasma viremia between 103 and 107 copies/mL throughout ATI, whereas 2 of 6 intervention animals had episodes of undetectable plasma viremia after initial rebound and maintained lower set points from approximately week 14. For exploratory analyses, animals were classified post hoc as controllers or non-controllers, with non-controllers defined by plasma viremia sustained above 1,000 copies/mL throughout ATI, a threshold selected after inspection of the observed trajectories. The 2 controllers were both intervention animals; the 10 non-controllers comprised 4 intervention and 6 control animals. As expected from this outcome-based classification, plasma viral burden during ATI was lower in the controllers (geometric mean, 2,746 versus 38,585 copies/mL, approximately 14-fold lower; Figure 4E). Given the small number of controllers (n = 2), no formal hypothesis testing was performed for any controller versus non-controller comparison.
We next examined whether the controller phenotype was associated with differences in the intact viral reservoir after ATI. At week 13 after ATI, the 2 controllers had 35.53 and 43.99 intact SIV proviral copies per 106 PBMCs, whereas the 10 non-controllers had a median of 740.2 (IQR 493.6–1,627; Figure 4F), approximately 17- to 21-fold higher than the individual controller values. From the pre-ATI baseline to week 13 after ATI, intact proviral DNA increased in 9 of 10 non-controllers (median 15-fold increase; IQR 5.16–25.93) but remained stable or decreased in both controllers (fold changes of 0.50 and 0.99; Figure 4G). Thus, in this exploratory post hoc analysis, secondary viral control in 2 intervention animals was accompanied by maintenance or reduction of the intact proviral reservoir after ATI, suggesting that immune mechanisms controlling viremia and reservoir size may contribute to partial postintervention control.
Viral features of postintervention control. Before examining antiviral immune responses, we assessed whether post-ATI viral control was associated with virological or pharmacological variables that preceded or were independent of the intervention. The 2 controllers did not exhibit lower systemic viral burden early in infection; both had pre-ART plasma viremia near the upper end of the non-controller range (Figure 5A). During suppressive ART from weeks 4–36 after infection, however, both controllers had cumulative plasma viremia among the lowest in the cohort (Figure 5B).
Figure 5Virological predictors of posttreatment control after analytical treatment interruption. (A–D) Comparison of controllers (n = 2) and non-controllers (n = 10) for geometric mean plasma viral load during early infection (A; weeks 1–3 after infection) and suppressive ART (B; weeks 4–36 after infection), and for intact SIV proviral DNA at ART initiation in PBMCs (C) and lymph node mononuclear cells (LNMCs) (D). Data are presented descriptively. (E–G) Linear regression of geometric mean plasma viral load during suppressive ART against intact SIV proviral DNA in PBMCs at week 13 after ATI, across all ATI cohort animals (E; n = 12) and separately in intervention (F; n = 6) and control (G; n = 6) animals. Solid lines indicate linear regression fits, and dashed lines indicate 95% CI bounds. (H and I) Partial Pearson’s correlations between geometric mean plasma viremia during the AZD5582 dosing phase in intervention animals or matched time points in control animals and intact SIV proviral DNA in PBMCs at week 13 after ATI, adjusted for geometric mean plasma viral load during suppressive ART, in intervention (H; n = 6) and control (I; n = 6) animals. Pearson’s correlation coefficients (r), coefficients of determination (R2), and exact P values are shown in E–G; partial correlation coefficients (r) and exact P values are shown in H and I. Each symbol represents an individual animal identified in the figure key. In A–E, controllers are shown in red, intervention non-controllers as filled black symbols, and control non-controllers as open black symbols. In F–I, controllers are shown in red, intervention non-controllers as filled blue symbols, and control non-controllers as open orange symbols.
Within the intervention group, AZD5582-associated viremia and cumulative eCD4-IgG1 exposure were similar in controllers and non-controllers (Supplemental Table 5). Intact SIV proviral DNA measured after intervention but before ATI in peripheral blood, lymph nodes, and rectal mononuclear cells also showed no clear separation (Supplemental Figure 7). These observations provide no evidence that the controller phenotype resulted from lower early plasma viremia, greater eCD4-IgG1 exposure, or a smaller intact reservoir immediately before ATI.
One preintervention measurement showed a distinct pattern. At week 3 after infection, before ART initiation, both controllers had fewer intact SIV proviruses in peripheral blood than any non-controller (4,653 and 5,019 vs. a non-controller median of 9,937 copies per 106 PBMCs; Figure 5C). In contrast, intact proviral DNA in lymph node mononuclear cells at the same time point was comparable between the controllers and non-controllers (Figure 5D). Thus, lower peripheral-blood intact proviral burden before ART was observed in both animals that later developed secondary viral control, although the small, post hoc–defined controller group precludes conclusions regarding prediction or causality.
To generate hypotheses regarding the association of these virological characteristics with post-ART viremia, we performed linear regression analyses across the entire cohort. We evaluated baseline predictors of 2 complementary measures of postintervention control: the geometric mean of plasma viremia during ATI and the levels of intact SIV proviral DNA in PBMCs at week 13 of ATI (Supplemental Table 6). Geometric mean viremia during initial suppression (weeks 4–36 after infection) was a strong predictor of intact proviral DNA during ATI (Figure 5E). By contrast, no virological variables predicted plasma viremia during ATI; however, intact SIV proviral DNA during ATI was strongly rank-correlated with the geometric mean of ATI viremia (Spearman’s ρ = 0.78, P = 0.005), indicating concordance between these measures. Accordingly, intact SIV proviral DNA was used as the primary outcome measure in subsequent analyses to identify determinants of postintervention control. Using this approach, we again found the geometric mean of viremia during initial suppression to be a robust predictor of intact SIV DNA during ATI in both the intervention (Figure 5F) and control groups (Figure 5G).
Given this strong influence of initial viral decay on ATI outcomes, we used partial correlation analyses to determine whether AZD5582-induced viremia was associated with postintervention reservoir size, independently of suppression-phase viremia, motivated by a possible association in the initial linear regression analyses, which suggested a possible association between AZD5582-induced viremia and ATI reservoir size (Supplemental Table 6; R2 = 0.32, P = 0.055). After adjusting for suppression-phase viremia, AZD5582-induced viremia was strongly inversely correlated with intact proviral reservoir size in intervention animals (r = –0.76, R2 = 0.58, P = 0.061; Figure 5H). No such correlation was observed in control animals when considering residual viremia over the matched timeframe (r = 0.48, R2 = 0.23, P = 0.338; Figure 5I), revealing a divergent treatment-dependent pattern in reservoir dynamics.
Collectively, these analyses identified viremia during initial ART suppression as the strongest predictor of intact proviral reservoir size during ATI, potentially implicating antiviral immune responses in viral control both at ART initiation and after ART withdrawal. Despite the limited cohort size, partial correlation analyses adjusting for baseline viremia suggested a treatment-dependent effect of AZD5582-induced on-ART viremia and smaller ATI reservoir size in the intervention group. However, comparisons of controllers versus non-controllers within the intervention group (n = 2 vs. n = 4) were underpowered to detect differences in AZD5582-induced latency reversal.
Immunological correlates of postintervention control. Because the lower pre-ART intact SIV proviral burden in the controllers could reflect an antiviral T cell response that also contributed to post-ATI control, we next examined SIV-specific T cell responses in relation to virological outcome. All animals were negative for Mamu-B*08 and Mamu-B*17. Mamu-A*01 positivity was similarly distributed across outcome groups, occurring in 1 of 2 controllers (RLq23) and 5 of 10 non-controllers (RIv23, RAl23, ROm23, RQl23, and RTk23). Thus, the MHC-I alleles examined did not segregate with post-ATI outcome, although the small number of controllers limits this interpretation. Both controllers received AZD5582 plus AAV9-eCD4-IgG1, whereas the non-controller group included 4 intervention animals and 6 ART-only controls. Therefore, intervention assignment and post-ATI outcome were not independent in this cohort. Because only 2 animals met the criteria for control, comparisons between outcome groups were considered descriptive, and no statistical tests were performed. SIV-specific CD4+ and CD8+ T cell responses were evaluated immediately before ATI, 4 weeks after completion of AZD5582 treatment, and at week 13 after ATI. Samples with poststimulation viability below 40% were excluded from the analyses; therefore, the number of evaluable animals varied by assay and time point.
We first quantified Gag-specific CD8+ T cells by the activation-induced marker (AIM) assay. Before ATI, both controllers had higher frequencies of Gag-specific CD8+ T cells than all evaluable non-controllers (3.17% and 5.07% vs. 0% to 1.12% in non-controllers, n = 6; Figure 6A). The lower controller value was approximately 2.8-fold greater than the highest non-controller value. This separation was no longer apparent at week 13 after ATI, when controller frequencies of 1.67% and 5.46% fell within the non-controller range of 0% to 8.83% (median 2.82%, non-controller n = 7). At this time point, non-controllers were viremic, whereas controllers had undetectable plasma viremia. The increased Gag-specific responses in several non-controllers were therefore consistent with stimulation by ongoing antigen exposure during ATI.
Figure 6SIV-specific T cell frequency and function in relation to postintervention control. (A) Frequencies of SIVmac239 Gag-specific CD8+ T cells measured by AIM assay before and 13 weeks after ATI. (B) Mean distribution of 15 Boolean combinations of IFN-γ, TNF-α, CD107a, and granzyme K among function-positive Gag-specific CD8+ T cells; proportions sum to 100%. (C) Proportions of Gag-specific CD8+ T cells expressing 1–4 functions per animal before ATI and 13 weeks after ATI. (D) CD8+ T cell polyfunctionality index calculated as described by Larsen et al. (55). (q = 1; range, 0.250–1.0). (E) Frequencies of SIVmac239 Gag-specific CD4+ T cells measured by AIM assay at the indicated time points. (F) Mean distribution of 7 Boolean combinations of IFN-γ, TNF-α, and CD107a among function-positive Gag-specific CD4+ T cells; proportions sum to 100%. (G) Proportions of Gag-specific CD4+ T cells expressing 1–3 functions per animal before ATI and 13 weeks after ATI. (H) CD4+ T cell polyfunctionality index (range, 0.333–1.0). Because the index is normalized to the number of functions, values in D and H are not directly comparable. In A, C–E, G, and H, controllers are red, intervention non-controllers are filled black, and ART-only non-controllers are open black. Symbols represent individual animals. Horizontal lines indicate non-controller medians. No statistical comparisons were performed because only 2 animals met criteria for postintervention control.
We next evaluated the functional composition of Gag-specific CD8+ T cell responses by intracellular cytokine staining. Boolean gates were used to define the proportion of total cytokine-producing CD8+ T cells expressing combinations of IFN-γ, TNF-α, CD107a, and granzyme K (Figure 6B). Before ATI, multifunctional subsets were generally enriched in the controllers (n = 2) compared with non-controllers (n = 10). The average proportions in the 2 controllers and the non-controller group were 8.9% and 2.4%, respectively, for IFN-γ+TNF-α+ cells; 15.9% and 8.9% for TNF-α+CD107a+ cells; and 6.0% and 2.1% for IFN-γ+TNF-α+CD107a+ cells. TNF-α+CD107a+granzyme K+ cells accounted for 0.5% of the response in controllers and 0.1% in non-controllers. Conversely, monofunctional CD107a+ cells represented a smaller proportion of the response in controllers than in non-controllers, with averages of 14.5% and 26.4%, respectively. At week 13 after ATI (n = 2 and n = 6), no functional subset showed consistent separation by outcome.
Grouping responding cells by the number of simultaneously expressed functions showed a similar pattern at the individual-animal level (Figure 6C). Before ATI, single-function cells comprised 54.4% and 65.9% of the responses in the 2 controllers, lower than in 9 of 10 non-controllers (median 76.6%; range 62.3%–100%). Two-function cells accounted for 37.4% and 25.8% of controller responses, versus a non-controller range of 0% to 29.1%, and cells expressing 3 or more functions accounted for 8.2% and 8.3% compared with a non-controller median of 2.6%. Notably, 1 intervention-arm non-controller, RJv23, fell within the controller range across all functional categories, indicating that pre-ATI CD8+ T cell polyfunctionality alone was insufficient for post-ATI control.
Polyfunctionality indices were used to summarize the response quality for each animal (Figure 6D). Before ATI, the 2 controllers had indices of 0.356 and 0.385, exceeding 9 of 10 non-controllers (non-controller median 0.317; range 0.250–0.378), with RJv23 being the exception with an index of 0.378 falling between the 2 controllers. At week 13 after ATI, controller values encompassed much of the non-controller range. RLq23 had the highest index among 8 animals evaluated (0.429), whereas RHq23 had an index of 0.322, below the non-controller median (0.325) Thus, within-controller variability exceeded the apparent difference between outcome groups.
We next assessed whether Gag-specific CD4+ T cell responses differed by post-ATI outcome. By AIM assay, response magnitude did not clearly distinguish controllers from non-controllers at either time point (Figure 6E). Before ATI, controller frequencies (4.31% and 1.76%) overlapped those of non-controllers (median 1.34%; range 0.58%–3.51%; n = 6). At week 13 after ATI, controller frequencies were 0% and 1.24%, both at or below the non-controller median of 2.34% (range 0%–6.20%; n = 7). As in the CD8+ T cell analysis, ongoing viremia in non-controllers that was absent in the controllers may have contributed to differences in antigen-driven responses during ATI.
Beyond response magnitude, we characterized the Gag-specific CD4+ T cell responses by intracellular cytokine staining for IFN-γ, TNF-α, and CD107a (Figure 6F). Dual-function subsets constituted a greater average proportion of the pre-ATI response in controllers than in non-controllers: 40.4% vs. 10.8% for IFN-γ+TNF-α+ and 10.7% versus 3.3% for TNF-α+CD107a+ cells. Conversely, monofunctional CD107a+ cells represented an average of 5.8% of the controller response versus 34.0% in the non-controllers. Triple-function cells showed less separation (4.9% controllers vs. 2.4% non-controllers). At week 13 after ATI, no functional subset consistently differed by virological outcome. Grouping by number of functions showed a similar pattern (Figure 6G): before ATI, single-function cells averaged 41.8% in controllers versus 78.5% in non-controllers and dual-function cells averaged 53.4% versus 19.1%, respectively. Controller polyfunctionality indices were 0.652 and 0.435 compared with a non-controller median of 0.417 (range 0.334–0.486; Figure 6H). At week 13 after ATI, the controller indices of 0.593 and 0.436 fell within the non-controller range of 0.366 to 0.651.
The pre-ATI differences in CD4+ T cell functionality were not consistent between the 2 controllers. RLq23 had the most functionally diverse CD4+ T cell response in the cohort, with single-function cells comprising 11.5% of the response and a polyfunctionality index of 0.652, representing the most extreme values among all 12 animals. In contrast, single-function cells comprised 72.0% of the response in RHq23, whose polyfunctionality index of 0.435 was within the non-controller range (Figure 6, G and H). Thus, the apparent outcome-associated differences in average CD4+ T cell functionality were driven predominantly by 1 controller, and neither the magnitude nor the functional composition of the CD4+ T cell responses consistently distinguished both controllers from the non-controllers.
Finally, we compared T cell responses by treatment group rather than post-ATI outcome. Before ATI, triple-function CD4+ T cell responses were higher in animals that received AZD5582 plus AAV9-eCD4-IgG1 than in ART-only controls, whereas no corresponding difference was observed for CD8+ T cells (Supplemental Figure 8).
Taken together, these data indicate that the animals that went on to control viral replication had mounted the largest and most functionally diverse Gag-specific CD8+ T cell responses in the cohort at the time of ART interruption, and these distinctions were not maintained once viral replication resumed. The corresponding CD4+ T cell responses did not differ in magnitude and differed in quality only in 1 of the 2 controllers, suggesting that the immunological features associated with postintervention control in this study were largely restricted to the CD8+ T cell compartment. Given that only 2 animals met criteria for control, these associations are hypothesis generating.
Early ART–treated infants are a compelling yet immunologically challenging population for HIV cure strategies — early therapy restricts reservoir seeding but also limits antigenic exposure and the development of HIV-specific immunity (14, 23, 24), making combined latency reversal and immune-mediated clearance an important but complex objective in this population. We tested this strategy using AAV9-delivered eCD4-IgG1 and AZD5582 in SIVmac251-infected infant rhesus macaques. Although the original goal to decrease reservoir size was not achieved, partial post-ART viral control occurred in 2 of 6 intervention animals and none of the controls. The observation of undetectable viremia by week 10 after ATI and reduced set point thereafter in the 2 controllers provided a framework to identify virological and immunological features distinguishing postintervention control in early ART–treated infants.
The initial challenge was determining a pharmacologically adequate AZD5582 regimen for infant macaques, as our prior work demonstrated that standard adult dosing yields subtherapeutic plasma concentrations in infants (17). A distinct infant-specific safety profile emerged: despite improved initial efficacy with the highest dose tested (0.3 mg/kg), cumulative toxicity after the fifth dose suggested a narrower therapeutic window in the developing host, potentially driven by developmental immaturity of phase I/II metabolic enzymes or renal clearance (25, 26). A modified regimen consisting of a 0.25 mg/kg loading dose followed by 0.2 mg/kg maintenance dosing resolved this toxicity while preserving latency reversal activity, establishing a pediatric-optimized AZD5582 regimen balancing efficacy and tolerability. These findings indicate that ncNF-κB–mediated latency reversal is mechanistically conserved across the lifespan but requires age-specific dosing to navigate a narrower therapeutic window.
To complement AZD5582, we selected eCD4-IgG1 (Emm058) as the immune effector based on its broad viral neutralization through CD4 and CCR5 binding site mimicry and its intact IgG1 Fc domain, which promotes ADCC against HIV and SIV (18). AAV9-mediated delivery achieved stable transgene expression in all intervention animals throughout the follow-up period, although circulating eCD4-IgG1 concentrations rarely exceeded 10 μg/mL. The transient acceleration of viral suppression during weeks 5–12 of ART, coinciding with rising eCD4-IgG1 concentrations, is consistent with in vivo biological activity.
Anti–eCD4-IgG1 antibodies were detected in 7 of 10 intervention animals but were generally low and transient. Only 1 animal developed persistent, high-level ADA, with circulating eCD4-IgG1 declining to an undetectable level and recovering after ADA resolution. Several mechanisms may explain this transient response. ADA risk after AAV-vectored antibody delivery increases with age at dosing (21), and our animals were treated at approximately 7 weeks, beyond the neonatal window of maximal tolerance but during ongoing immune maturation. Sustained antigen exposure may also attenuate humoral immunity, as repeated dosing of biologics in humans reduces ADA relative to a single dose (27). Continuous transgene expression similarly lacks the peak-trough kinetics and inflammatory context of intermittent protein administration, which may favor nondurable responses. Antigen-specific regulatory mechanisms may also contribute, although evidence for Treg-mediated tolerance after AAV delivery derives largely from liver-directed gene transfer (28). Because only 1 animal developed persistent ADA and relevant immune subsets were not profiled, these mechanisms remain speculative.
Despite stable transgene expression and minimal immunogenicity, with only 1 animal developing ADA that reduced eCD4-IgG1 concentrations, the combination strategy did not measurably reduce the latent reservoir. This may reflect a critical exposure-response threshold. Although circulating eCD4-IgG1 concentrations exceeded the in vitro neutralization IC50 (<0.05 μg/mL), the geometric mean plateau concentration in infants (9.97 μg/mL) remained below levels reported in prior adult and infant macaque studies (18, 29) and below the estimated in vivo eCD4-Ig neutralization IC50 (~25 μg/mL) (30). Since an in vivo ADCC-relevant threshold has not been defined, it remains uncertain whether the concentrations achieved were sufficient to engage the Fc-mediated effector mechanisms most pertinent to reservoir reduction. Higher circulating concentrations may therefore be necessary for effective reservoir targeting, though overwhelming virus reactivation in this highly pathogenic SIV model or other incompletely characterized mechanisms affecting biological activity over time cannot be excluded. Consistent with studies identifying early reservoir size as a predictor of ATI outcome (31), the 2 animals that showed secondary viral suppression also had the smallest week 3 pre-ART intact reservoir size in peripheral blood. However, in contrast with prior reports (31–33), longitudinal analysis revealed no significant reservoir differences across tissues between the controllers and non-controllers during ART-mediated suppression. Divergence reemerged only during ATI, with non-controllers exhibiting a median 15-fold expansion of intact proviral DNA by week 13, whereas the 2 controllers maintained stable or declining levels. This finding mirrors observations in adult macaque and human posttreatment controllers, where reservoir expansion is restricted despite ART withdrawal (32, 34), and implicates active immune-mediated control as the mechanistic driver.
Consistent with this interpretation, controllers exhibited a distinct pre-ATI immunological profile characterized by higher Gag-specific CD8+ T cell frequencies and greater CD8+ T cell polyfunctionality than non-controllers, while the higher polyfunctionality observed in the CD4+ compartment was driven by mostly 1 controller. Antiviral CD8+ T cells have been implicated in post-ART control both without additional interventions and after broadly neutralizing antibody (bNAb) therapy (35–37), with polyfunctionality, ex vivo viral suppressive and proliferative capacity, and recall cytotoxicity linked to viral control in humans and adult nonhuman primates (38–41). Notably, in a prior therapeutic vaccine study in SIVmac251-infected, ART-suppressed infant macaques, the animal with the greatest induction of SIV-specific IFN-γ ELISPOT responses showed evidence of posttreatment control, although this pattern was not observed across the vaccinated group (42).
Although pre-ART immunophenotyping and functional studies were not feasible due to prioritization of infected cell quantification, multiple lines of evidence indicate superior immunological control in controllers throughout the study. Their lower intact proviral DNA in PBMCs at week 3 likely reflects more effective early immune responses in animals predisposed to viral control. Reservoir size during suppressive ART did not distinguish controllers from non-controllers, consistent with attenuation of T cell responses under reduced antigen exposure on ART (43, 44). Rather, the controllers’ defining feature was their ability to limit reservoir expansion during viral recrudescence. In this context, the association between suppression-phase viremia and post-ATI reservoir size most plausibly reflects an underlying favorable host phenotype, namely the capacity to contain viral replication more efficiently throughout ART, rather than a direct causal relationship.
Within this framework, partial correlation analyses suggested a potential contribution of AZD5582. After adjusting for suppression-phase viremia, AZD5582-induced viremia was inversely correlated with intact proviral reservoir size in intervention animals, whereas the relationship in controls was directionally opposite (Fisher’s r to z: P = 0.063). These findings support a broader role for SMAC mimetics in potentiating effector immunity beyond direct latency reversal, consistent with preclinical evidence that AZD5582 enhances antigen-specific CD8+ T cell responses in tumor-bearing mice (45). We have also previously found increased Gag-specific IFN-γ+ CD8+ T cells after treatment of SIV-infected, ART-suppressed adult macaques with AZD5582 and SIV Env mAbs (16). Although underpowered, the divergent correlation directionality between groups is consistent with pharmacological reactivation amplifying virus-specific effector responses in animals already immunologically primed for control, rather than functioning as an independent reservoir-clearing mechanism.
This study has several limitations. It was not powered to assess efficacy definitively, and the small number of posttreatment controllers (n = 2) constrained statistical inference and precluded multivariable analysis. Reservoir and immune characterization were also incomplete: intact proviral DNA does not fully capture the replication-competent reservoir or tissue-level heterogeneity, particularly in lymphoid compartments, where sampling was limited; and the absence of pre-ART immunophenotyping prevented the assessment of baseline immune features associated with control. In vivo eCD4-IgG1 concentrations remained below estimated therapeutic thresholds, limiting the evaluation of the full potential of this approach. The post-ATI follow-up period was limited to 20 weeks, precluding assessment of delayed control. Finally, species-specific differences in immune development and pharmacokinetics may limit translation from infant macaques to human infants.
Because intervention animals received AAV9-eCD4-IgG1 and AZD5582 concurrently, the contribution of each agent cannot be resolved. An AAV9-eCD4-IgG1-only arm was not included, although eCD4-Ig alone did not alter reservoir size or post-ATI rebound kinetics in a prior adult macaque study (46). Historical comparisons are similarly constrained: prophylactic studies assessed neutralization at challenge, and the concentrations associated with sterilizing protection (18, 29, 47) may not reflect those required for Fc-mediated clearance of reactivated reservoir cells. Completed therapeutic studies in adult macaques differ from those in infants due to numerous biologic and experimental variables: immune ontogeny, reservoir seeding under early ART, AZD5582 pharmacology and ncNF-κB inducibility (17), immunoadhesin Fc isotype, dosing and delivery strategy, challenge virus and route, ART timing, duration of suppression, and reservoir assays.
Pediatric HIV cure strategies have unique challenges imposed by early ART initiation, limited antigenic exposure, and age-specific pharmacology. This study makes 2 complementary contributions. First, we establish a pediatric-optimized AZD5582 regimen that achieves latency reversal while maintaining tolerability in infant macaques, providing a foundation for future combination cure studies. Second, despite lack of reservoir clearance on ART or delayed virus rebound after ATI, we describe posttreatment control associated with superior virus-specific T cell polyfunctionality and the capacity to restrict reservoir expansion during viral recrudescence. Partial correlation analyses further suggest that AZD5582-induced viremia may potentiate immune-mediated control in immunologically primed animals, pointing to a broader immunomodulatory role for SMAC mimetics beyond direct latency reversal. These findings define concentration thresholds required for reservoir attrition, establish immunological correlates of posttreatment control in early-treated infants, and provide a translational framework for next-generation pediatric HIV cure strategies.
Sex as a biological variable. Both male and female animals were included in this study. Animals were distributed across groups without significant differences in sex composition. Formal comparisons between sexes were not performed.
Ethics statement and animal care. All animal procedures were conducted in accordance with the USDA Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals (National Academies Press, 2011) and approved by the Emory University IACUC (protocol PROTO202100117). Studies were performed at the Emory National Primate Research Center (ENPRC), which is accredited by AAALAC International. Infant rhesus macaques (Macaca mulatta, Indian origin), male and female, were obtained from the ENPRC breeding colony and enrolled at 4 weeks of age. Husbandry and clinical monitoring are described in the Supplemental Methods.
Study design, SIV infection, and experimental interventions. Twenty infant rhesus macaques negative for Mamu B*08 and B*17 alleles were enrolled in a sequential 2-cohort design: a dose optimization cohort (n = 4 intervention, n = 4 control) and an ATI cohort (n = 6 intervention, n = 6 control). Infants were screened for AAV9-neutralizing antibodies, and of 41 meeting the acceptance threshold, 20 were enrolled based on availability and randomly assigned to intervention or control groups. Beginning at 4 weeks of age, animals were orally challenged weekly with 105 TCID50 SIVmac251 (provided by Koen Van Rompay, University of California, Davis, California National Primate Research Center (CNPRC), Davis, California, USA) until infection was confirmed.
At week 3 after infection, all animals initiated suppressive ART administered as a combined daily s.c. formulation of tenofovir disoproxil fumarate (5.1 mg/kg), emtricitabine (40 mg/kg), and dolutegravir (2.5 mg/kg). Viral suppression was defined as 2 consecutive plasma viral load measurements below the limit of detection (60 copies/mL). Sustained suppression was defined as 8 consecutive undetectable measurements over at least 16 weeks.
Concurrent with ART initiation, intervention animals (n = 10) received i.m. coformulated AAV9-eCD4-IgG1 and AAV9-TPST-2 at doses of 2.5 × 1012 and 6.3 × 1011 GC/kg, respectively, distributed across thigh and arm sites. AAV9-TPST-2 encodes rhesus tyrosylprotein sulfotransferase 2, which sulfates tyrosine residues within the eCD4-Ig sulfopeptide, a modification required for optimal gp120 engagement (22). Control animals received no AAV vectors, and no repeat dosing was performed.
Latency reversal was initiated at week 40 after infection with the SMAC mimetic AZD5582 by slow i.v. infusion (see Supplemental Methods). In the dose optimization cohort, animals received weekly infusions at 0.3 mg/kg (n = 2) or 0.25 mg/kg (n = 2) for up to 10 planned doses. Cumulative toxicity after 5–7 doses resulted in euthanasia of 3 animals and treatment discontinuation in 1 animal, per predefined humane endpoints; none underwent ATI. Based on observed toxicity and pharmacokinetic modeling, the ATI cohort received a modified regimen consisting of a 0.25 mg/kg loading dose followed by 5 weekly maintenance doses of 0.2 mg/kg, which was well-tolerated. Control animals in the ATI cohort remained on ART without AZD5582.
Four weeks after completing AZD5582 dosing, ATI cohort animals (n = 12) underwent structured ATI. Plasma viral load was measured 3 times weekly for the first 5 weeks, and then weekly to assess viral rebound kinetics. Animals were classified post hoc as controllers or non-controllers, with non-controllers defined by plasma viremia sustained above 1,000 copies/mL throughout the ATI period.
Pharmacokinetic modeling. Blood was collected from adult and infant rhesus macaques into EDTA tubes before and 2, 4, 6, 24, and 48 hours after AZD5582 infusion. Plasma was separated by centrifugation (800g, 15 minutes) and stored at –80°C until AZD5582 quantification by liquid chromatography–tandem mass spectrometry (15).
A population pharmacokinetic model was developed in NONMEM from infant macaque plasma AZD5582 concentrations after 0.1 mg/kg i.v. dosing (46 observations across 14 animals) (17). A 3-compartment model was fit using the FOCE algorithm with proportional residual error and interindividual variability on clearance; the 6 structural parameters (clearance [CL], volume of the central compartment [V1], intercompartmental clearance between the central and first peripheral compartment [Q2], volume of the first peripheral compartment [V2], intercompartmental clearance between the central and second peripheral compartment [Q3], and volume of the second peripheral compartment [V3]) were estimated with a relative standard error (percent coefficient of variation, CV%) of less than 58%. Because infants achieved lower Cmax and AUC than adults at the same 0.1 mg/kg dose, the model was simulated for infant plasma profiles for doses of 0.2, 0.25, 0.3, 0.4, and 1.0 mg/kg across infusion durations of 0.25–1.5 hours. Simulated infant profiles were compared with observed adult concentrations at 0.1 mg/kg (15) to identify the infant regimen best matching adult Cmax and AUC (Supplemental Figure 2).
Sample collection and processing. Peripheral blood, lymph node, and rectal biopsy specimens were collected longitudinally at the time points indicated in Figure 1. Plasma was isolated from EDTA whole blood and stored at –80°C. Mononuclear cells were isolated from blood and lymph nodes by density-gradient centrifugation and from enzymatically digested rectal biopsies on a discontinuous Percoll gradient, and then cryopreserved in liquid nitrogen (see Supplemental Methods).
Plasma viral load quantification and intact proviral DNA measurement. Plasma SIVmac251 RNA was quantified at each time point by the Translational Virology Core at the Emory Center for AIDS Research (CFAR) using validated qRT-PCR (lower limit of detection 60 copies/mL). During the AZD5582 window, high-sensitivity viral load testing (lower limit of detection 3 copies/mL) was performed in the Quantitative Molecular Diagnostics Core of the AIDS and Cancer Virus Program of the Frederick National Laboratory for Cancer Research, essentially as previously described (48, 49). Intact proviral DNA was quantified in mononuclear cells using an SIVmac251-adapted intact proviral DNA assay (Accelevir Diagnostics), as previously described (50–52). Results were expressed as intact proviral copies per 1 × 106 mononuclear cells from peripheral blood, lymph node biopsies, or rectal biopsies.
AAV vector production. Recombinant AAV9 vectors were produced by triple transfection of HEK293T cells with plasmids encoding AAV2 rep/AAV9 cap, the transgene cassette, and an adenoviral helper. The eCD4-IgG1 vector comprised the chicken β-actin promoter driving rhesus eCD4-IgG1 (rhesus CD4 ectodomain fused to a short-hinge rhesus IgG1 Fc domain and gp120-binding sulfopeptide), followed by woodchuck hepatitis virus posttranscriptional regulatory element and SV40 polyA signal (53). Because eCD4-Ig lacks a light chain, the canonical IgG1 hinge cysteine that would pair with the light chain constant domain is unpaired, which can promote aberrant intermolecular disulfide bonding and misfolding of the secreted protein. The rhesus IgG1 hinge was therefore truncated to remove this residue, yielding a short-hinge variant (Emm058; Supplemental Figure 1). TPST2 was expressed from a CMV promoter (22, 29). Both constructs were flanked by ssAAV inverted terminal repeats. Vectors were purified by cesium chloride gradient centrifugation, purity was assessed by silver-stained SDS-PAGE, and vector genomes were quantified by digital droplet PCR. Full-to-empty capsid ratios were periodically evaluated by electron microscopy and routinely exceeded 90%.
Quantification of plasma eCD4-IgG1 and anti-eCD4-IgG1 antibodies. Plasma eCD4-IgG1 concentrations were measured by sandwich ELISA using an anti-CD4 capture antibody and an HRP-conjugated anti-human Fcγ detection reagent against a standard curve of recombinant eCD4-IgG1 spiked into macaque serum. Anti–eCD4-IgG1 antibodies (ADAs) were measured using the same ELISA format with plates coated with recombinant eCD4-IgG1 (1 μg/mL). Background-subtracted absorbance values are reported. All ADA samples were assayed on the same plate to permit direct comparisons across time points. Full assay conditions are presented in the Supplemental Methods.
AAV9-neutralizing antibody screening. Preexisting AAV9-neutralizing antibodies were assessed by transduction inhibition assay. Macaque serum (1:10 in DMEM/5% FBS) was mixed with an AAV9:CAG-GFP-P2A-luciferase-WPRE-SV40 reporter vector (4 × 107 to 8 × 107 vector genomes per 40 μL), incubated 1 hour at 37°C, then transferred in triplicate onto HEK293T cells (MOI 500–1,000 VG/cell) for 24 hours. Luciferase activity was measured and percentage neutralization calculated relative to no-serum controls. Animals with less than 50% neutralization at 1:10 dilution were eligible for AAV9 administration.
Assessment of SIV-specific T cell responses. SIV-specific T cell responses were assessed by AIM and intracellular cytokine staining assays using cryopreserved PBMCs collected before ATI and at week 13 after ATI. Cells were thawed, rested overnight at 37°C, and stimulated at 0.5 × 106 cells per condition with SIVmac239 Gag overlapping peptides (15-mers overlapping by 11; 2.5 μg/mL/peptide; BEI Resources, HRP-12364) plus anti-CD28 and anti-CD49d (1 μg/mL each; BioLegend). DMSO and staphylococcal enterotoxin B (1 μg/mL; Sigma-Aldrich) served as negative and positive controls, respectively.
The multiplexed AIM assay was adapted from Lemieux et al. (54). After 24-hour stimulation, cells were stained with viability dye, a PE-Cy5 dump channel (CD14, CD16, and CD20), and antibodies against CD3, CD4, CD8α, CD95, CD25, CD69, OX40, and 4-1BB, and then fixed with 1% PFA. Antigen-specific cells were defined as CD4+CD95+ or CD8+CD95+ T cells expressing at least 2 of 4 AIMs (CD69, CD25, OX40, and 4-1BB), using Boolean OR gating across all pairwise double-positive combinations to count each responding cell once.
For intracellular cytokine staining, cells were stimulated for 6 hours with anti-CD107a added at initiation and brefeldin A (10 μg/mL) and monensin (BD GolgiStop) added after 1 hour. Cells were surface-stained with the same viability dye; dump channel; and CD3, CD4, CD8α, and CD95 antibodies used for the AIM assay, and then fixed and permeabilized with BD Cytofix/Cytoperm, and stained intracellularly for IFN-γ, TNF-α, and granzyme K. Responses among CD4+CD95+ and CD8+CD95+ T cells were quantified by Boolean gating after subtraction of paired DMSO-control responses.
Samples were acquired on a Cytek Aurora spectral flow cytometer using single-color controls and autofluorescence correction and analyzed in FlowJo v11.1. Antibody clones, fluorochromes, and suppliers are listed in the Supplemental Methods; gating strategies are shown in Supplemental Figures 9 and 10.
Statistics. All statistical analyses were performed in R (version 4.5.2; R Core Team) using RStudio (version 2026.01.0) or GraphPad Prism (version 11.0.0). Figures were generated using GraphPad Prism. Statistical significance was defined as P less than 0.05 (2-sided). Data are presented as median with IQR for nonparametric comparisons and as mean ± SEM for parametric analyses, unless otherwise indicated. Plasma viral load values below the limit of detection were assigned that value prior to log10 transformation, using 60 copies/mL for the standard assay and 3 copies/mL for the ultrasensitive assay.
Longitudinal viral load trajectories (weeks 3–36) were analyzed using linear mixed-effect models (lme4/lmerTest packages), with time modeled as a natural cubic spline (4 df; splines package) to capture nonlinear suppression dynamics. Treatment group, spline-transformed time, and their interaction were fixed effects; animal identifier was a random intercept. Post hoc comparisons of estimated marginal means (emmeans package) were Benjamini-Hochberg FDR-corrected. Fold differences in plasma viral load between groups were calculated as the difference in group mean log10 viral load at each time point, with SEM pooled across groups, and were back-transformed to the ratio scale; ratios below 1 are displayed as negative signed fold-changes on a log10 axis.
Between-group comparisons of continuous variables were performed using 2-sided Mann-Whitney U tests. Categorical variables were compared using the χ2 test or Fisher’s exact test when expected cell counts were less than 5. Week-level fold-differences in on-ART viremia during AZD5582 dosing were estimated by Welch’s unequal-variance t test on ln-transformed viral loads, with 95% CIs back-transformed to the ratio scale.
Time to viral rebound was defined as the interval between initiation of ATI and the first plasma sample containing more than 60 SIV RNA copies/mL. All 12 animals experienced viral rebound, and no observations were censored. Rebound-free survival distributions were estimated using the Kaplan-Meier method and compared between groups using a 2-sided log-rank (Mantel-Cox) test. The hazard ratio and 95% CI were estimated using the observed-minus-expected log-rank method, with the control group as the reference.
Associations were assessed by Pearson’s correlation; partial correlations adjusting for geometric mean post-ART viremia were computed with pcor.test (ppcor package), and group differences in correlation strength were assessed by Fisher’s r to z transformation.
T cell polyfunctionality was assessed using Boolean gating to enumerate all effector function combinations for CD8+ T cells (IFN-γ, TNF-α, CD107a, granzyme K; 15 combinations) and CD4+ T cells (IFN-γ, TNF-α, CD107a; 7 combinations). A polyfunctionality index (PI) was calculated as described by Larsen et al. (55), in which PI = Σᵢ Fᵢ × (i/n)q, where Fᵢ is the proportion of the cytokine-producing T cell population expressing exactly i of the n measured functions and q = 1. Frequencies were calculated relative to the total responding population rather than the parent CD4+ or CD8+ T cell population, such that the index reflects response quality independently of response magnitude. Because the index is normalized to the number of measured functions, values are not directly comparable between lineages; the minimum attainable index is 0.250 for CD8+ (n = 4) and 0.333 for CD4+ (n = 3) T cells. Comparisons of PI and functional category proportions between controllers and non-controllers were descriptive, and no statistical tests were applied, given that only 2 animals met the criteria for postintervention control.
Study approval. All animal procedures were approved by the Emory University IACUC (protocol PROTO202100117).
Data availability. All source data are provided in the Supporting Data Values file. Requests for additional data or materials described in this manuscript will be promptly reviewed by the corresponding author to determine whether these requests are subject to intellectual property, confidentiality, or ethical obligations. If not subject to these restrictions, data will be shared with the requester in a timely manner. Materials that can be shared will be released via a materials transfer agreement. Inquiries regarding data or material availability should be directed to the corresponding author.
Author contributions. AC and MAM conceptualized the study, developed methodology, analyzed data, and supervised the research; AC additionally acquired funding. JAF developed methodology, conducted experiments, acquired and curated data, performed data analysis, prepared figures, and managed project administration. ACK conducted experiments, acquired and curated data, and managed project administration. DOH, AK, and LABDC developed methodology, acquired data, and prepared figures; DOH additionally reviewed the manuscript. KSD and CC conducted experiments and acquired data. MC, MDA, MRG, and JDL developed methodology and provided reagents. SE and JSW provided resources and managed project administration. MM designed research studies and supervised the research. JAF and AC wrote the manuscript. All authors reviewed and approved the final manuscript.
AI disclosure. The authors declare that no artificial intelligence tools were used in the design, conduct, analysis, or reporting of this study.
Conflict of interest. MRG, MDA, and MAM have financial interests in Emmune Inc., which develops HIV immunotherapies based on eCD4-Ig. MDA and MDG are confounders and have equity stakes in Emmune Inc.; MM has a consulting financial interest in Emmune, Inc. AC serves on a Scientific Advisory Board for ViiV Healthcare, Inc.; these conflicts are managed by their institutions.
Funding. This study was supported by the following:
This work is the result of NIH funding, in whole or in part, and is subject to the NIH Public Access Policy. Through acceptance of this federal funding, the NIH has been given a right to make the work publicly available in PubMed Central.
The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US government.
We thank Koen Van Rompay for providing SIVmac251, Gilead Sciences and ViiV Healthcare for antiretroviral drugs, and the Emory Center for AIDS Research Virology & Molecular Biomarkers Core and the Research Resources Team at the Emory National Primate Research Center for their support.
Address for correspondence. Ann Chahroudi, 1750 Haygood Dr. NE, Atlanta, Georgia 30322, USA. Phone: 404.727.9209; Email: ann.m.chahroudi@emory.edu.
Copyright: © 2026, Fonseca et al. This is an open access article published under the terms of the Creative Commons Attribution 4.0 International License.
Reference information: JCI Insight. 2026;11(21):e208375.https://doi.org/10.1172/jci.insight.208375.