Immune Activation Profiles Elicited by Distinct, Repeated TLR Agonist Infusions in Rhesus Macaques

Immunotherapy, the treatment of disease by modulating the immune system through administration of stimulatory, suppressive, or disease-specific agents, is a rapidly growing area of research. Commonly used in the cancer field, immunotherapy treatments include immune checkpoint inhibitors, adoptive transfer of Ag-specific T cells, mAbs, and immune adjuvants or cytokines (1). An emerging field of immunotherapy is the treatment of chronic viral diseases, such as hepatitis C with IFN and EBV-associated and human papilloma virus–associated cancers with virus-specific T cells (2).

TLR agonists are a promising class of immune system modulators. TLRs are pattern recognition receptors, expressed on cell surface or internal membranes, composed of an ectodomain consisting of leucine-rich repeats, a transmembrane domain, and a cytoplasmic Toll/IL-1R domain responsible for initiation of downstream signaling (3). The extracellular TLRs (TLR1, TLR2, TLR4, TLR5, TLR6, and TLR10) primarily recognize microbial membrane components, including lipids and proteins. Intracellular TLRs (TLR3, TLR7, TLR8, and TLR9), located on intracellular vesicles including the endoplasmic reticulum and endosomes, recognize viral and bacterial nucleic acid (4). TLR signaling occurs through either the MyD88 or TRIF (Toll/IL-1R domain-containing adapter inducing IFN-β) pathways, resulting in the expression of inflammatory cytokines, type I IFN, and other TLR-inducible proteins (4, 5).

TLR agonists have received approval from the U.S. Food and Drug Administration for use as immunotherapy, including the bacillus Calmette–Guérin (BCG) vaccine given intravesically for treatment of noninvasive transitional cell carcinoma in the bladder (6, 7), and a TLR7 agonist, imiquimod, for treatment of warts and basal cell carcinoma (8). Many other TLRs are under preclinical or clinical investigation, either as standalone treatments, adjuvants for cancer vaccines, as vaccine adjuvants for infectious diseases (9), or in combination with chemotherapy or other immunotherapeutic treatments (10–12). TLR agonists have also been trialed for chronic viral diseases, such as hepatitis B virus (HBV). TLR3 and TLR7 ligands can suppress HBV replication in in vitro models, and TLR9 ligands have been demonstrated to enhance HBV-specific T cell responses and clearance (13).

TLR agonists also show promise as a component of HIV “cure” strategies. These efforts aim to stimulate latently infected T cells to express virus and promote their elimination, or to enhance antiviral immune responses, leading to durable control in the absence of antiretroviral therapy (ART). TLR agonists have been hypothesized to contribute to both of these cure strategies (reviewed in Ref. 14). Multiple TLR agonists can reactivate latent HIV, with TLR7 or TLR7/8 (15–17), TLR8 (18, 19), TLR3 (20), TLR4 (21, 22), TLR5 (23), TLR2 (17, 24–26), and TLR9 (21, 27, 28) ligands able to stimulate HIV replication in in vitro models. Additionally, CD8⁺ T cells (29, 30) and NK cells (28) display enhanced killing of HIV-infected cells after reactivation by TLR ligands. The TLR7 agonists GS-986 and GS-9620 combined with broadly neutralizing mAb treatment have shown promise in preclinical HIV cure studies in rhesus macaques (31–33), although their ability to reactivate latent virus in vivo is controversial (34, 35). In clinical trials, TLR7 and TLR9 agonists did not modulate plasma viremia or cell-associated virus (36, 37).

In this study, we investigated three different TLR agonists to thoroughly define their immunomodulatory properties in vivo in rhesus macaques (Macaca mulatta), a widely used model for infectious disease research. Rhesus macaques are particularly valuable for evaluating immunotherapeutic strategies due to their close genetic relatedness with humans, and consequent similar immunobiology and TLR expression patterns (38). They also recapitulate disease phenotypes for chronic viral infections such as HIV. We compared three agonists targeting different TLRs:LPS, a TLR4 agonist (39), 2BXy, a TLR7/8 agonist (40, 41), and a live attenuated BCG vaccine, which contains components that are TLR2/4/9 agonists (42). We characterized immune activation in plasma and in blood cells longitudinally following administration of each agonist alone, at escalating doses and following repeat administrations. These data provide a foundational understanding of TLR agonist immunostimulatory behavior in primates that will facilitate application of these and related TLR agonists for immunotherapeutic interventions.

Twelve Chinese-origin rhesus macaques (six male and six female) 6–9 y of age and weighing between 5 and 10 kg at the study start were colony-bred in the United States and distributed between treatment groups by sex and weight. Animals were administered one of three TLR agonists. 2BXy (provided by Avidea Technologies) was administered either i.v. at a dose of 5–125 μg/kg or i.m. into the left and right quadriceps at doses spanning 25–125 μg/kg. LPS (ultrapure LPS from Escherichia coli K12; InvivoGen) was administered i.v. at doses ranging from 10 to 100 μg/kg. BCG i.v. immunization consisted of 4.8 × 10⁷ to 5.9 × 10⁷ CFU of Danish Strain 1331 (IAVI). Cryopreserved BCG was thawed and diluted in 2 ml of PBS containing 0.05% tyloxapol and 0.002% antifoam Y-30, as described (43). Peripheral blood was collected as follows: 1) 2BXy animals at 4 h, 1 d, 3 d, and then weekly; 2) LPS animals at 2/4 h, 1 d, 3 d, and then weekly; 3) BCG animals at 2 h, 6 h, 1 d, and 1, 2, 3, 4, and 7 wk (Fig. 1). The timing of sampling was based on published reports (43, 44) and prior in-house experience. Bronchoalveolar lavage (BAL) was collected in BCG-vaccinated animals only (baseline; 4 and 7 wk after each immunization). PBMCs were isolated from EDTA blood collections using Ficoll-Paque Plus gradient separation (Cytiva). Cell pellets from BAL (3 × 20-ml washes of room temperature PBS) were isolated by centrifugation at 311 × g, aspiration, resuspension in warm R10 (RPMI 1640 with 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 U/ml Benzonase nuclease [MilliporeSigma], and 10% heat-inactivated FBS), and filtering through a 70-μm cell strainer before counting and downstream immune assays.

Ethics approval for this project was obtained from the institutional Vaccine Research Center Animal Care and Use Committee. Macaques were housed and cared for in accordance with local, state, federal, and institutional policies in facilities accredited by the American Association for Accreditation of Laboratory Animal Care, under standards established in the Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals.

IFN-α was measured in plasma using a VeriKine cynomolgus/rhesus IFN-α serum ELISA kit (PBL Assay Science). TNF-α levels were measured in plasma using either a monkey TNF-α ELISA kit (MyBioSource) (−7 d to 35 d in Supplemental Fig. 1) or a rhesus macaque TNF-α Quantikine ELISA kit (R&D Systems) (Fig. 2, 42–57 d in Supplemental Fig. 1, with this change having been made due to the greater sensitivity of this kit). ELISA assays were performed according to the manufacturers’ instructions.

Multiplex cytokine measurements were performed using the MILLIPLEX MAP nonhuman primate (NHP) cytokine magnetic bead panel/premixed 23-plex Luminex kit (MilliporeSigma) according to the manufacturer’s instructions and a Bio-Plex MAGPIX multiplex reader (Bio-Rad). An asymmetric sigmoidal, 5PL, x is log(concentration) model curve with least squares fit was used to interpolate the cytokine concentration with GraphPad Prism software v9. Values above the limit of detection (LOD) were assigned a value equal to the upper LOD, whereas values below the lower LOD were assigned a value half of this limit when determining fold change between samples.

Cryopreserved PBMCs were thawed, washed, and then stained with a Live/Dead fixable blue dead cell stain kit for UV excitation (Thermo Fisher Scientific) for 20 min at room temperature. Cells were stained with fluorescently labeled Abs (all BD Biosciences unless otherwise stated) specific for cell surface markers CD3 Cy7 allophycocyanin (clone SP34-2), CD4 Cy5.5 PE (clone S3.5; Life Technologies), CD8 BUV805 (clone RPA-T8), CD20 Cy5 PE (clone 2H7), NKG2A PE Cy7 (clone Z199; Beckman Coulter), CD14 BV750 (clone M5E2), CD16 BV570 (clone 3G8; BioLegend), CD28 allophycocyanin (clone CD28.2), CD45RA BUV737 (clone 5H9), CD38 PE (OKT10; Caprico Biotechnologies), CD69 ECD (clone TP1.55.3; Beckman Coulter), HLA-DR BV480 (clone G46-6), CD64 BUV395 (clone 10.1), CD80 BV421 (clone L307.4), and CD169 BB515 (clone 7-239). In some experiments, cells were fixed and permeabilized using the transcription factor buffer set (BD Biosciences) according to the manufacturer’s instructions prior to staining with the anti-Ki67 BV650 (clone B56). Finally, cells were resuspended in 0.5% paraformaldehyde and staining was measured on a BD FACSymphony. Data were analyzed using FlowJo v10 software (Tree Star).

Flow cytometry was performed on BAL samples collected from animals that received BCG vaccination. BAL samples were analyzed on the day of collection, as described (43). Briefly, ∼3 million BAL cells were either stained immediately ex vivo to characterize the overall leukocyte composition or stimulated with R10 alone or 20 μg/ml M. tuberculosis H37Rv whole-cell lysate (BEI Resources) for 6 h (4 h with protein transport inhibitor cocktail, eBioscience) before intracellular cytokine staining to detect Ag-specific CD4⁺ T cell responses (43).

2BXy-specific Abs were measured in plasma from animals that received 2BXy. 2BXy was coated onto amine-binding, maleic anhydride 96-well plates (Pierce) at 1 μg/well in 100 μl PBS/well and incubated, shaking, at room temperature overnight. The plate was then blocked with B3T block buffer (250 μl/well; 150 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, 3.3% FBS, 2% bovine albumin, 0.07% Tween 20, 0.02% thimerosal). All subsequent steps used total volumes of 100 μl/well, in B3T buffer, with incubations at room temperature for 1 h. Serial dilutions of plasma were added to the plate as the primary Ab. Binding was detected with anti-monkey HRP-conjugated Ab (clone SB108a, SouthernBiotech) at a 1:5000 dilution. Reactions were developed for 10 min using 100 μl/well SureBlue tetramethylbenzidine one-component microwell peroxidase substrate (KPL), and reactions were stopped using 100 μl/well 1 N sulfuric acid. Absorbance was measured at 450 nm on an Epoch microplate spectrophotometer (BioTek).

Principal component (PC) analysis (PCA) was performed using the R programming language and the prcomp, ggplot, and factoextra libraries. Initially, the Luminex and flow cytometry datasets were merged based on corresponding sample IDs. The combined dataset was subsequently scaled and centered to guarantee equal contribution from all variables. The prcomp function was employed to conduct PCA on the scaled data, obtaining the PCs that accounted for the maximum variation in the dataset. Immune measurement contributions to the PCA results were determined using the factoextra R library. The PCA results were visualized by plotting the scores of the first three PCs (PC1, PC2, and PC3) using ggplot2 and ggthemes.

Statistical comparisons of immunological parameters between groups were performed using GraphPad Prism v9.4.1. Primary immunogenicity outputs were compared across groups using a Kruskal–Wallis test. Nonparametric pairwise comparisons between groups were made using the post hoc Dunn’s test. Statistical significance was preset at an α level of 0.05.

We performed a series of studies in rhesus macaques to assess the impact of sequential administration of TLR agonists on innate immune activation. We also aimed to determine the optimal dosage, route, and sampling schedule for each of these agonists. The TLR agonists selected were as follows: 1) 2BXy, a small-molecule imidazoquinoline-based TLR7/8 agonist with a C2-butyl substituent and an N¹-xylylamine substituent (40, 41), selected due to the promising results of TLR7 agonists in HIV cure studies (31, 33); 2) LPS, a TLR4 agonist (39) that has been previously shown to reactivate virus in SIV-infected rhesus macaques (44); and 3) i.v. BCG, expected to be a highly potent immune activator, due to multiple components that engage TLRs (primarily TLR2 and TLR4 [45–47] but also TLR9 [48]) and in vivo replication. BCG is a live attenuated vaccine derived from the Gram-positive Mycobacterium bovis, typically delivered intradermally as a childhood vaccination against M. tuberculosis infection. Administration of BCG i.v. protects against M. tuberculosis infection in NHPs, an effect associated with increased CD4⁺ and CD8⁺ T cell immunity over conventional intradermal vaccination (43).

2BXy was administered four times to two groups of three rhesus macaques either via i.v. or i.m. routes (Fig. 1A). Dosing escalated from 5 (i.v.) or 25 (i.m.) to 125 µg/kg. LPS was initially administered i.v. to two groups of three animals four times receiving either 10 or 50 µg/kg (Supplemental Fig. 1A). However, plasma TNF-α responses, measured by ELISA as a marker of TLR4-induced inflammation, were below the assay LOD (Supplemental Fig. 1B), prompting a revised study design with a shorter interval blood collection at 2 h rather than 4 h after LPS administration (Fig. 1B). The same six animals received the four additional LPS doses every 2 wk. We hypothesized that earlier sampling would capture rapidly expressed and cleared cytokines such as TNF-α. Initially, this second phase of the study administered LPS at a dose of 100 µg/kg. Upon the second administration of 100 µg/kg, one of the first three animals to receive this dose experienced an adverse reaction, with seizures and an erratic heartrate (Supplemental Fig. 1C). No additional animals received the second dose of 100 µg/kg (the additional three animals received no injection), and the remaining third and fourth doses, administered to all six animals, were reduced to 50 µg/kg, which three of the animals had previously received four times with no adverse events. No further adverse events were observed. BCG was administered i.v. with ∼5 × 10⁷ CFU three times, 8 wk apart (Fig. 1C).

To assess immune activation and production of inflammatory cytokines, plasma IFN-α (a type I IFN expected to be produced upon TLR7 engagement) and TNF-α were measured longitudinally after 2BXy and LPS administration, respectively. Circulating IFN-α increased in all six 2BXy animals, with the highest levels of ∼100–50,000 pg/ml observed at 4 h after each 2BXy administration (Fig. 2A, 2B). IFN-α persisted in plasma for a longer duration after the first, lower doses, where it was measurable out to 7 d, compared with subsequent administrations where IFN-α levels dropped below the assay LOD by 3 d (Fig. 2A), possibly due to the repeated dosing. Peak IFN-α increased with 2BXy dose, with significantly higher plasma IFN-α concentrations following 125 µg/kg i.m. compared with the second 25 µg/kg i.m. administration (Fig. 2B). The route of administration (i.v. versus i.m.) did not affect IFN-α response rate or magnitude.

Although small-molecule chemical drugs are not typically immunogenic, the potential reduction observed in peak IFN-α cytokine levels following repeat administrations of 2BXy at identical doses (25 and 125 µg/kg i.m.) prompted us to consider the possibility of anti-drug Ab (ADA) development limiting 2BXy activity with subsequent exposures. IgG titers against 2BXy were measured by ELISA. Two animals had pre-existing low-level 2BXy-reactive titers, which increased following the third administration, and again after the fourth administration in one animal (Fig. 2C). One animal without baseline 2BXy reactivity developed measurable, albeit low, anti-2BXy Abs after repeat dosing. Humoral reactivity to 2BXy at the time of the third administration did not inversely correlate with plasma IFN-α concentration, either at 4 h or 1 d (Fig. 2D), suggesting that pre-existing (cross-reactive) anti-2BXy Abs do not impact 2BXy responsiveness.

As LPS TLR4 stimulation induces TNF-α production by myeloid cells (44, 49), we measured TNF-α in the plasma of animals that received LPS. As noted above, the dynamics of innate immune stimulation by LPS are such that short-interval blood sampling was required to measure cytokine responses in circulation. Two hours following LPS administration with this modified sampling schedule (fifth LPS dose in total), TNF-α spiked in peripheral blood of all six animals, with a geometric mean titer of 515 pg/ml (Fig. 2E, 2F; range of 143–786 pg/ml). TNF-α was cleared rapidly as described previously (44, 50), with concentrations below the LOD 4 h or 1 d after LPS administration (Fig. 2E, Supplemental Fig. 1B). Peak TNF-α responses measured following the eighth, and final, administration remained robust, albeit lower than the fifth dose with a geometric mean of 79 pg/ml (Fig. 2F; range of 16–148 pg/ml). This reduction may suggest a tolerization to the agent or may be due to the reduced 50 µg/kg dosage.

To interrogate a broad array of immunomodulatory cytokines potentially induced by TLR stimulation, plasma samples were further analyzed by a multiplex Luminex assay to quantify 23 cytokines/chemokines. The first, second, and fourth 2BXy administrations (5 or 25, 25, and 125 μg/kg, respectively, Fig. 3A; 4 h after 2BXy administration), and the first, third, and fourth LPS administrations (100, 50, and 50 μg/kg, second study phase, Fig. 3B; 2 h after LPS administration) were assessed. Both agents induced a large increase in plasma cytokines within the 2–4 h sampling interval. Cytokine responses to 2BXy increased in a dose-dependent manner. Plasma IL-6 increased from an 8-fold induction at the 5 µg/kg dose to 1650-fold at the 125 µg/kg dose, whereas IL-1RA increased from 60-fold to 480-fold. IFN-γ, which did not rise following the 5 µg/kg dose, increased by 15-fold upon 125 µg/kg 2BXy. TNF-α and G-CSF expression increased predominantly after the 125 µg/kg dose (although TNF-α levels may have been higher if measured at 2 h, as for LPS). IL-12, which is induced by TLR8 stimulation, increased from a 2-fold induction following 5 µg/kg to 7-fold at 125 µg/kg. TGF-α was induced in all animals, and IL-1β and IL-18 were induced in five of six animals following the second dose of 25 µg/kg, with minimal induction following the 125 µg/kg dose, suggesting either inhibition of these cytokines at this higher dose or different response kinetics. At 1 d postadministration, the cytokines in both groups largely returned to baseline levels. One notable exception was IL-1RA, which remained consistently elevated, particularly among 2BXy recipients. Cytokines with the most apparent induction by LPS included IL-6 (geometric mean 1300-fold increase), IL-1RA (350-fold), IL-10 (250-fold), G-CSF (140-fold) and, as expected, TNF-α (90-fold). These responses did not appear to change with LPS dose. LPS elicited greater overall plasma cytokine levels than did 2BXy in the early hours following administration, although we cannot exclude an effect of the shorter sampling interval. By 1 d, responses to 2BXy exceeded those of LPS. This was especially apparent for IL-1RA, but could also be seen with cytokines such as MCP-1, IL-6, and IFN-γ. The route (i.v. versus i.m.) of 2BXy administration did not impact cytokine responses.

Immune activation can also be measured directly on cells by their expression of activation markers. Adaptive and innate immune cell populations were phenotyped by flow cytometry 1 d and 1 wk following the administration of 2BXy (second and fourth doses) and LPS (first and third doses). Insufficient cell numbers limited analyses at <24 h. T cells (both CD4⁺ and CD8⁺) upregulated surface CD38, which is induced by inflammatory stimuli (51), and both T and NK cells upregulated CD69, a marker of recent activation (52) at 1 d after 2BXy infusion (Fig. 4A; cell subset identification in Supplemental Fig. 2). NK cell CD38 expression significantly increased at 1 d and 1 wk; however, high constitutive CD38 expression observed at baseline suggests that this change may not be biologically meaningful. Neither HLA-DR, a late T cell activation marker (53), nor Ki67, a marker of proliferation (54), expression was altered by 2BXy. Animals that received LPS displayed greater heterogeneity in cell surface activation marker expression (Fig. 4B). CD69 expression increased 1 d after LPS administration on memory CD4⁺ T cells (p = 0.08, p = 0.03 following the two infusions, respectively) and on five of six animals for CD8⁺ T cells and four of six animals for NK cells, although significance was not reached on these two cell types. CD80, a marker of B cell activation, was modestly increased on B cells 1 d following the fourth 2BXy administration (Supplemental Fig. 2D), with no differences following LPS administration.

Monocytes, key sensors of pathogen-associated molecular patterns, were similarly evaluated for expression of the following activation surface markers: CD64 (FcγRI, whose expression has been shown to increase on monocytes following TLR agonist stimulation [55]); CD80, a monocyte activation marker (56); and CD169, a marker of activated monocytes with augmented capacity to stimulate CD8⁺ T cells (57). CD80 and CD169 expression significantly increased 1 d following one of the two infusions analyzed for both agonists (Fig. 4C, 4D), and trended higher at the other infusion (2BXy: CD80 p = 0.12, CD169 p = 0.15; LPS: CD80 p = 0.26, CD169 p = 0.08). Ki67 was measured following the later administrations of each agonist. Its expression significantly increased 1 d after 2BXy (the fourth administration) and 1 wk after LPS (the third administration). The upregulation of CD169 was especially striking in animals that received 2BXy. CD169 expression was maintained at high levels 4 wk following the first administration and 2 wk following the third administration in three of six animals—the three animals that received 2BXy via the i.m. route. It is also possible that these animals had greater baseline CD169 expression.

To assess systemic immune activation following BCG vaccination, plasma cytokine concentrations were assessed longitudinally by Luminex postvaccination (Fig. 5A). Although plasma cytokine levels remained largely unchanged at 2 h after BCG administration relative to baseline, increases were observed by 6 h and many cytokines remained elevated at 24 h postvaccination. At 6 h, MCP-1 (∼22-fold), G-CSF (∼25-fold), IL-6 (∼76-fold), and IL-1RA (∼42-fold) concentrations were particularly elevated. At 1 wk postvaccination, plasma IL-12 (∼8-fold), IL-18 (∼20-fold), IFN-γ (∼10-fold), and IL-1RA (∼15-fold) remained above baseline. Some cytokine responses increased with BCG boosting. For example, plasma IFN-γ remained unchanged at 6 h following the first vaccination, then increased ∼9- and ∼35-fold following the second and third vaccinations, respectively. Similarly, at 24 h, IFN-γ increased by 37-fold at the first vaccination and then 136-fold at the third vaccination. This contrasted with the 2BXy and LPS administration, where plasma IFN-γ levels remained unchanged. When comparing overall trends in all plasma cytokine levels, the magnitude of cytokine elicitation increased with each subsequent immunization, with increasing fold changes at the 6- and 24-h time points (Fig. 5B).

Cellular activation elicited by i.v. BCG immunization was assessed as described above. HLA-DR and Ki67 significantly increased 1 wk after vaccinations on NK cells and memory CD4⁺ and CD8⁺ T cells (CD28-negative and/or CD45RA-negative; Fig. 6A, 6B). HLA-DR⁺ cell frequency was greatest after the third BCG vaccination for all three of these populations, suggesting a recall response to BCG vaccination. CD8⁺ T cell activation was also evidenced by elevated CD38 at 1 wk after the first and third administrations and CD69 at 1 d following the third vaccination, whereas CD4⁺ T cell CD69 expression increased 1 wk following the first vaccination (Supplemental Fig. 3A, 3B). B cell activation decreased 1 wk following the initial BCG administration, evidenced by diminished CD80 surface expression, and similar trends were observed following subsequent vaccinations (Supplemental Fig. 2D). Monocytes, which express TLR2 and TLR4, were also activated in response to BCG, with marked induction of Ki67 1 d after the third BCG administration. Similarly, the latter immunizations stimulated CD80 expression (Fig. 6C). Monocyte activation was more modest and delayed following the first BCG administration (7–14 d). Neither CD64 nor CD169 expression by monocytes was altered by BCG (Supplemental Fig. 3C).

In contrast to 2BXy or LPS, BCG is expected to elicit strong adaptive immune responses that may in turn regulate inflammatory and innate activation. To determine whether adaptive immune responses continue to develop during repeated i.v. BCG exposures and in the context of marked innate activity and pre-existing adaptive responses, we assessed longitudinal T cell responses. As previously shown (43), i.v. BCG immunization increased the number of total cells in the BAL, consistent with cell recruitment to the airways (Fig. 7A). The magnitude of this increase was greatest following the first two immunizations, with elevated cell numbers sustained following the second vaccination. Vaccination also altered the BAL cellular composition: prior to BCG administration, ∼60% of BAL cells were macrophages, whereas T cells predominated after BCG administration, making up 70, 82, and 84 after each successive BCG administration (Fig. 7B). CD4⁺ T cells were especially enriched, including Vγ9⁺ and mucosal-associated invariant T cells following the first vaccination. M. tuberculosis–specific T cells in the BAL were quantified by intracellular cytokine staining following in vitro stimulation with M. tuberculosis Ags. BAL CD4⁺ T cells were largely M. tuberculosis–specific, comprising an average of 34% of memory CD4⁺ T cells or 3.7 × 10⁶ cells 4 wk after the second immunization (Fig. 7C). The quality of the cytokine response largely remained the same after the repeat vaccinations; however the proportion of peak effector triple-positive IFN-γ, IL-2, and TNF-α cells decreased following the third vaccination, whereas more differentiated single IFN-γ⁺ or dual IFN-γ⁺TNF-α⁺ cells increased (Fig. 7D).

To further investigate the unique and shared features among responses to the three immune agonists (2BXy, BCG, and LPS) in a multivariate approach, we explored the combined plasma cytokine and cellular activation phenotyping data by PCA. For harmonization, all data were derived from 1 d following TLR agonist administration. The three groups formed discrete clusters defined by three PCs (Fig. 8A), suggesting distinct response profiles to each agonist. PC1 primarily separated LPS from 2BXy and BCG. Analysis of the variable contributions to this PC revealed separation primarily driven by plasma IL-1RA, which was elevated following 2BXy and BCG administration compared with LPS (Fig. 8B, Supplemental Table I). PC1 variance was also driven by monocyte CD64 expression, which increased following BCG vaccination, and CD69 expression by NK cells, which was induced most by 2BXy. PC2 separated the three different agonists on a continuum from BCG to 2BXy to LPS, stemming from increases in plasma IFN-γ and IL-6 by BCG and soluble CD40L (sCD40L) by LPS (Fig. 8B, Supplemental Table I). PC3, which best separated 2BXy from the LPS and BCG profiles, derived most of its variance from sCD40L and IL-1RA (both of which also contributed to PC1 and PC2) as well as monocyte CD169 expression, which was most strongly upregulated by 2BXy (Fig. 8B).

This study investigated immune activation generated by three TLR agonists administered separately in rhesus macaques, including 2BXy, a TLR7/8 agonist; LPS, a TLR4 agonist, and i.v. BCG vaccination, a replicating pathogen that engages TLR2/4/9. All agonists were safe and immunostimulatory and induced distinct immune activation profiles. LPS elicited transient immune activation, dominated by plasma TNF-α, IL-6, IL-10, IL-1RA, MCP-1, and monocyte activation. An adverse event was observed at the higher LPS dose, indicating that close attention to LPS dosing and administration is warranted in future studies. 2BXy, the TLR7/8 agonist, induced activation of T and NK cells in addition to monocytes, as well as plasma type I IFNs. Dose escalation of 2BXy increased immune activation, whereas the route of administration (i.v. versus i.m.) did not alter responses. Three repeated doses of i.v. BCG at 8-wk intervals were safe, well tolerated, and induced a pronounced increase in the proliferation marker Ki67 on T and NK cells. Immunotherapy strategies aiming to increase activation or abundance of these cell populations may benefit from incorporating these or related agents.

The dynamics of immune activation differed among the TLR agonists. LPS effects were largely transient, with plasma TNF-α rising rapidly within 2 h and resolving by 4 h, and, compared with the other TLR agonists, cytokine expression was minimal at 1 d. This is consistent with previous reports of early TNF-α expression in both LPS-stimulated human whole blood (mRNA 2–4 h) (50) or LPS-administered rhesus macaques (peak at 2–6 h) (44). 2BXy induced marginally longer immune activation than did LPS, with persistent elevation of plasma IL-1RA at 1 d, consistent with type I IFN induction by TLR7 stimulation (58). The longer half-life of 2BXy compared with other TLR agonists may contribute to the prolonged response (59). BCG generated the most protracted inflammatory state, with substantial cytokine induction observed at 1 d and even 1 wk following administration. The persistence of BCG bacterium at least 1 mo after vaccination in the lymph nodes, lung, and spleen (43) may provide sustained stimulation. The affinity of each of these TLR agonists for their ligands may also vary, which could influence their immunostimulatory properties. For example, 2BXy is a high-affinity TLR7 agonist (60) whereas LPS requires other proteins for its interaction with TLR4 (61). The use of different TLR agonists with altered affinity may achieve different results from those reported in the current study.

The immunomodulatory properties of the TLR agonists did not diminish with repeated administrations. Rather, immune activation induced by BCG vaccination increased with repeated doses. This may be due to development of BCG-specific adaptive immune responses, such as increased IFN-γ expressed by M. tuberculosis–specific CD4⁺ T cells. Trained immunity could also contribute to augmented inflammatory responses, whereby innate immune cell (e.g., monocyte) responsiveness to a pathogen increases upon subsequent encounter (62), consistent with increased monocyte activation following repeat BCG administration observed in the current study. We are unable to distinguish between indirect effects from the anamnestic T cell recall response versus direct TLR stimulation of myeloid cells as the source for increased innate immune activation with repeated BCG administration. Repeat administration of the two other TLR agonists, in contrast, achieved comparable immune stimulation throughout the study. Although modest ADA titers were detected against 2BXy, largely in animals with pre-existing cross-reactive immune responses, the magnitude of these responses was low and did not correlate with immune activation or diminished 2BXy responsiveness over time. Larger studies with a greater number of animals and 2BXy administrations are needed to explore the potential link between TLR agonist innate immune stimulation and possible 2BXy sensitization and induction of small-molecule ADAs, whereas longer follow-up will clarify the impact of 2BXy ADAs on 2BXy clearance and activity retention in vivo. Prescreening animals for 2BXy cross-reactive Abs to reduce the likelihood of 2BXy ADA development may be useful. Although previous work reported ADA responses to LPS (63), LPS-reactive Abs were not assessed in the current study.

Separate clustering of animals within each TLR agonist group by PCA indicated upregulation of distinct immune activation pathways. Major drivers of this separation included the following: 1) more robust plasma IL-1RA responses generated by 2BXy and BCG compared with LPS, 2) elevated plasma IFN-γ and IL-6 by BCG versus greater sCD40L by LPS, 3) BCG stimulating IFN-γ production, and 4) more pronounced monocyte activation (CD169⁺) by 2BXy. Less sustained levels of IL-1RA, an anti-inflammatory molecule released in response to type I IFN or IL-1 production (58), by LPS likely reflects the transient nature of LPS innate stimulation relative to 2BXy and BCG. Induction of a Th1 adaptive T cell response may explain the pronounced IFN-γ cytokine levels observed following i.v. BCG vaccination. Increased monocyte CD169 by 2BXy is not unexpected, as monocyte activation is induced by type I IFN responses induced by TLR7 engagement (64, 65). Of note, monocytes expressing CD169 have enhanced CD8⁺ T cell activation capacity (57), which has previously been observed following 2BXy administration (60, 66). One major limitation to the analysis conducted in the current study is the different response kinetics to each TLR agonist, as discussed above. In addition, our analysis was primarily limited to peripheral blood sampling. Most of our innate immune measurements were performed on plasma and PBMCs, with the exception of BAL-infiltrating lymphocytes and Ag-specific T cells following BCG vaccination. Future studies interrogating secondary lymphoid tissue and mucosal surfaces will be important for understanding the broader effects of blood-based TLR stimulation across the immune system and at sites where most pathogen exposures occur.

The TLR agonists administered in this study increased activation of a broad array of cell types, including T and NK cells, which are of particular interest for the control of viral diseases such as HIV-1. TLR agonists may directly activate these cells via receptors expressed on their surface; for example, T and NK cells express TLR7 and TLR4, albeit at low levels (67–70). For example, the elevated IL-2 and sCD40L seen in the current study may be due to direct engagement of TLR on CD4⁺ T cells. However, activation signals produced by other cell types that express high levels of TLRs, such as monocytes and dendritic cells, are likely indirect, primary mediators of lymphocyte activation. Dendritic cell activation was not assessed in the current study and will be an important component of future work investigating TLR agonist immunomodulation. HIV-1 cure “kick and kill” strategies often seek to activate CD4⁺ T cells, a major reservoir of latent HIV, to achieve HIV-1 latency reversal and promote infected cell clearance. However, TLR agonists have shown mixed results in vivo to date in their ability to reactivate HIV-1 and increase antiviral immunity. TLR2 (71) and TLR3 agonists (72) stimulate HIV-1 expression in humanized mouse models, but reactivation of SIV/simian HIV in rhesus macaques with the TLR4 agonist LPS and TLR7 agonists GS-986 and GS-9620 has not been consistent (34, 35, 44). BCG-induced T cell proliferation may represent a strategy to achieve HIV-1 reactivation in CD4⁺ T cells among people living with HIV-1; however, this also poses a potential risk of expanding infected cell reservoirs. The broad activation achieved by these TLR agonists may also be a drawback for studies where activation of limited cell types is desired. The detailed immunomodulatory characterization presented in the current study will aid in selection of TLR agonists with beneficial attributes.

The ability to augment pathogen-specific immunity via TLR adjuvanting effects is an active area of investigation and has shown promise against HIV-1 in animal models. CpG-B (TLR9), R848 (TLR7/8), and poly(I:C) (TLR3) can enhance T cell responses when used to adjuvant HIV vaccines in mice (72–74), as does poly(I:C) in macaques (75). Ad26/MVA therapeutic vaccination combined with TLR7 agonist GS-986 increased T cell immune responses in SIV-infected monkeys and delayed viral rebound upon ART interruption (76). Similarly, combining GS-986 or a related TLR7 agonist, GS-9620, with a broadly neutralizing HIV-1 mAb in ART-suppressed simian HIV–infected macaques also delayed or even prevented viral rebound (31–33). However, the mechanisms by which TLR7 stimulation supports virus control in these therapeutic contexts are unclear. In contrast, prophylactic mucosal administration of CpG (TLR9) or imiquimod (TLR7) failed to protect against mucosal SIV challenge in macaques, instead enhancing plasma viremia (77). Further study of 2BXy, LPS, and BCG in the setting of infection or vaccination is required to assess their ability to modulate adaptive immune responses.

The TLR agonists used in this study may elicit distinct activation profiles in rhesus macaque models compared with humans. For example, rhesus macaques are less sensitive to the pyrogenic activity of LPS than are humans (78). The absence of B cell activation following LPS administration seen in the current study further supports decreased LPS sensitivity in macaques. Suppression of inflammatory cytokines in rhesus macaques by theta defensins (79) may contribute to some degree of endotoxin resistance (80). Macaque neutrophil progenitor populations are also differently activated by LPS than are their human counterparts (81). Although NHPs are considered a robust model of tuberculosis infection (82), reduced sensitivity to immune activation by tuberculosis Ags in the context of mycobacterial protein-based tuberculosis skin testing has been described (83). In response to TLR7/8 stimulation, some B cell proliferation and cell surface marker expression patterns differ between rhesus and human samples in vitro (84), as do IFN-α subtypes elicited (85). Thus, although rhesus macaques are an extremely valuable preclinical model system sharing extensive genetic and immunologic similarities with humans, differences in immune agonist response profiles between humans and macaques are not unexpected.

In sum, we provide detailed characterization and comparison of the immunostimulatory properties of three distinct TLR agonists in NHPs, including performance following repeated administrations and at varying doses and routes. Due to the importance of rhesus macaques as a preclinical model for assessing prophylactic and therapeutic efficacy strategies against many infectious diseases, we believe these data will be useful in guiding future clinical and preclinical studies, including informing the selection of agonists as well as appropriate conditions for their application.

The authors have no financial conflicts of interest.

We thank the Vaccine Research Center Nonhuman Primate Immunogenicity Core for superb technical support, especially Amy T. Noe, Evan Lamb, Dillon R. Flebbe, Shing-Fen Kao, and Nadesh N. Nji. We also thank the veterinary staff and animal support staff at the Vaccine Research Center and Bioqual, Inc., especially Dr. Ruth Woodward, Dr. Diana Scorpio, and John N. Graves.