Suppression of Th1 priming by TLR2 agonists during cutaneous immunization is mediated by recruited CCR2⁺ monocytes.

Abstract

Effective subunit vaccines require the incorporation of adjuvants that stimulate cells of the innate immune system to generate protective adaptive immune responses. Pattern recognition receptor (PRR) agonists are a growing class of potential adjuvants that can shape the character of the immune response to subunit vaccines by directing the polarization of CD4 T cell differentiation to various functional subsets. In the present study, we applied a high-throughput in vitro screen to assess murine CD4 T cell polarization by a panel of PRR agonists. This identified lipopeptides with TLR2 agonist activity as exceptional Th1 polarizing adjuvants. In vivo, we demonstrated that intravenous administration of TLR2 agonists with antigen in mice replicated the findings from in vitro screening by promoting strong Th1 polarization. In contrast, TLR2 agonists inhibited priming of Th1 responses when administered cutaneously in mice. This route-specific suppression was associated with infiltrating CCR2⁺ cells in the skin draining lymph nodes, and was not uniquely dependent on any of the well-characterized subsets of dendritic cells known to reside in the skin. We further demonstrated that priming of CD4 T cells to generate Th1 effectors following immunization with the Mycobacterium bovis BCG strain, a lipoprotein-rich bacterium recognized by TLR2, was dependent on the immunization route, with significantly greater Th1 responses with intravenous compared to intradermal administration of BCG. A more complete understanding of route-dependent TLR2 responses may be critical for informed design of novel subunit vaccines, and for improvement of BCG and other vaccines based on live-attenuated organisms.

Introduction

Vaccine adjuvants, particularly TLR agonists, are promising tools to augment and influence the character of the immune response to vaccines. Adjuvants can induce protective immunity to co-administered antigens through various mechanisms including the polarization of specific CD4 T cell effector subsets (1–4). Promoting robust Th1 polarization is of specific interest to elicit protective responses against intracellular pathogens, including major causative agents of morbidity and mortality such as Mycobacterium tuberculosis (5). Despite the broad range of potential vaccine adjuvants, only a handful are currently approved and used routinely, including alum and the TLR4 agonist monophosphoryl lipid A (6, 7) . New approaches to screen and validate vaccine adjuvants are needed, but current in vitro protocols to efficiently screen adjuvants for key properties, such as relevant effects on CD4 T cell priming, are not standardized and do not accurately reflect the CD4 T cell priming environment in vivo (8, 9).

In the current study, we sought to develop an in vitro priming assay that accurately reflected the in vivo priming environment to facilitate the screening of multiple PRR agonists for utility as adjuvants. A novel in vitro co-culture system using TCR transgenic naïve CD4 T cells, primary splenocytes and splenic fibroblastic reticular cells (FRCs) was used to perform initial screening of adjuvant-induced CD4 T cell priming and polarization to different defined effector subsets. Analysis in vivo showed that the in vitro screening accurately predicted the results with i.v. immunization. However, we observed route-specific differences in the ability of TLR2 agonists to induce Th1 differentiation during naïve CD4 T cell priming when administered intradermally (i.d.) with foreign protein antigen. Detailed analysis of the cellular mechanisms underpinning this discrepancy between intravenous (i.v.) and cutaneous immunizations incorporating TLR2 agonists implicated monocytes infiltrating the skin draining lymph nodes as major mediators of suppression of Th1 priming. These findings are relevant to situations in which vaccines containing TLR2 agonists are administered by injection into the skin, as in the case of the Bacillus Calmette-Guérin (BCG) live-attenuated vaccine for tuberculosis (10, 11), and potentially also for subunit vaccines co-administered with adjuvants acting through TLR2 stimulation.

Materials and Methods

Mice

6–8 week old BALB/c and C57BL/6 females were purchased from Jackson Laboratories. T cell receptor transgenic DO11.10 mice on a BALB/c Rag2^−/− background were purchased from Taconic Biosciences. The TCR transgenic P25 and BatF3^−/− mice on C57BL/6 background were purchased from the Jackson Laboratory and maintained by breeding in our facility. Mice expressing diphtheria toxin receptor under direction of the human Langerin promoter (huLangerin-DTR) and mice expressing diphtheria toxin receptor under direction of the Mgl2 promoter (Mgl2-DTR) were maintained in our facility from founders generously provided by D. Kaplan (University of Minnesota, MN). Mice expressing diphtheria toxin receptor under direction of the CCR2 promoter (CCR2-DTR mice) were generously provided by E. Pamer (Memorial Sloan Kettering Cancer Center New York, NY). The P25 transgenic mice expressing GFP on a C57BL/6 background (P25^GFP mice) were produced by a single F1 cross of homozygous P25 Tg^+/+ mice with mice homozygous for green fluorescent protein under the direction of the human ubiquitin C promoter (UBI-GFP^+/+ mice; Jackson Laboratory). The OT-II TCR transgenic mice expressing green fluorescent protein (OT-II^GFP) were bred in our facility from crossing OT-II and UBI-GFP^+/+ founders originally obtained from the Jackson Laboratory. All mice were maintained in specific pathogen-free conditions. All procedures involving the use of animals were in compliance with protocols approved by the Einstein Institutional Animal Use and Biosafety Committees.

PRR agonists

All isolated or synthetic PRR agonists were obtained from Invivogen (San Diego, CA) and used at the following concentrations for in vitro experiments unless otherwise indicated: Pam3CSK4 (1 μg/ml), lipoteichoic acid (LTA) from S. aureus (1 μg/ml), zymosan from S. cerevisiae (1 μg/ml), lipoarabinomannan (LAM) derived from M. smegmatis (1 μg/ml), high molecular weight Poly (I:C) (10 μg/ml), Poly (A:U) (10 μg/ml), ultrapure lipopolysaccharide (LPS) derived from E. coli O111:B4 (1 μg/ml), monophosphoryl lipid-A (MPL-A) (1 μg/ml), ultrapure flagellin derived from S. typhimirium (0.1 μg/ml), imiquimod (2 μg/ml), loxoribane (100 μM), CL097 (1 μg/ml), CpG ODN 1585 (CpG-A) (0.05 μM), CpG ODN 1668 (CpG-B) (0.05 μM), CpG ODN 2395 (CpG-C) (0.05 μM), CpG ODN 2088 (Inhibitory ODN) (0.5 μM), muramyl dipeptide (MDP) (10 μg/ml), and C12-iE-DAP (10 μg/ml), Nigericin (0.05 μM), curdlan from A. faecalis (5 μg/ml), trehalose dimycolate (TDM) (2 μg/ml). Adenosine-5’-triphosphate (ATP) (100 μM), 2’(3’)-O-(4-Benzoylbenzoyl)-ATP (BzATP) (50 μM), Nicotinamide adenine dinucleotide (NAD) (50 μM), all-trans retinoic acid (aTRA) (0.5 μM), and prostaglandin E2 (PGE2, 2 μM) were from Sigma-Aldrich (St. Louis, MO). Macrophage activating lipopeptide-2 (MALP-2) (0.1 μg/ml), and gardiquimod (1 μg/ml) were purchased from Enzo Life Sciences (Farmingdale, NY). The synthetic glycolipid trehalose dibehenate (TDB) was purchased from Avanti Polar Lipids (Alabaster, AL), and monomycoyl glycerol (MMG) (2 μg/ml) was a generous gift from Dr. Branch Moody (Harvard Medical School, Boston, MA).

In vitro priming assay

Fibroblastic reticular cells (FRCs) were isolated from single cell suspensions of naive BALB/c splenocytes using a modification of previously published methods (12). First, the splenocytes were depleted of CD45⁺ cells using anti-CD45 beads (Miltenyi Biotec, Cologne, Germany) and magnetic separation. The resulting CD45 negative splenocytes were cultured for four weeks in complete medium (RPMI-1640 with 50 μM 2-mercaptoethanol, 10 mM HEPES (all from ThermoFisher) supplemented with 10% heat inactivated fetal calf serum (Atlanta Biologicals, Flowery Branch, GA). Remaining cells at this point were adherent with fibroblast morphology, and were designated FRCs. These were harvested by trypsinization and gentle scraping, irradiated (5000 Rad) using a Shepard Mark 1 Cesium-137 irradiator and frozen in FCS with 10% DMSO as single use aliquots. For each experiment, FRCs were thawed and plated in flat bottom 96-well polystyrene plates at 1.25×10⁴ per well in Iscove’s modified Dulbecco’s Medium (IMDM) supplemented with 1% sodium pyruvate, 1% Non-essential amino acids solution, 100 U/ml penicillin and 100 μg/ml streptomycin, and 0.1% β-mercaptoethanol (all from ThermoFisher) plus 10% v/v heat inactivated FCS and incubated at 37°C for 24–48 hours. Pooled single cell suspensions of BALB/c spleen and lymph nodes were added to the FRC monolayer at 7.3 × 10⁵ per well, followed by addition of EndoFit Ovalbumin (Invivogen) at 25 μg/ml and PRR agonists. Following incubation for 2 hours at 37°C, 5 × 10⁴ splenocytes from DO11.10 TCR transgenic mice were added to the culture (the overall cellular composition of the priming co-culture at this stage is illustrated in Supplemental Fig. 1). After a further 48 hr incubation at 37°C in a humidified 5% CO₂ incubator, two thirds of the medium was removed and replaced with fresh IMDM. This was repeated after another 48 hours, followed by incubation for an additional 48 hours (i.e., for a total in vitro priming period of 6 days) until restimulation and flow cytometry analysis.

In vitro restimulation and analysis of functional CD4 T cell subsets

For analysis of differentiation of DO11.10 T cells primed in vitro with antigen plus the various PRR ligands, cultures were restimulated and analyzed by intracellular staining to identify functional subsets (i.e., Th1, Th2, Th17, Tfh and iTreg). Briefly, approximately two thirds of the media were removed by aspiration from the 96 well plate wells containing primed DO11.10 T cells following the 6 day priming culture. This was replaced with 150 μl of fresh complete IMDM with 10% FCS containing 5 × 10⁵ freshly isolated BALB/c splenocytes, 5 μg/ml Brefeldin-A, 2 μM monensin sodium (Sigma) 10 μg/ml OVA323–339 peptide (Mimotopes), 1 μg/ml anti-mouse CD28 (clone 37.51, ThermoFisher) and 1 μg/ml anti-mouse CD49d (clone R1–2, eBioscience). The cells were then incubated at 37°C in a 5% CO₂ humidified incubator for 6 hours to allow antigen restimulation, and then collected, washed twice in PBS and stained with the violet-excitable viability dye (AqVID, Invitrogen) for 25 minutes on ice. The cells were then washed in PBS + 5% FCS, and then treated with ~10 μg/ml anti-mouse FcγII/III antibody (clone 2.4G2) for 5 minutes at room temperature to block Fc receptor binding. Surface staining was first carried out by addition of fluorochrome labelled mAbs against mouse CD4 (clone GK1.5 APC-H7) and DO11.10 TCR (clone KJ1–26 PE), and a cocktail of biotin-labeled antibodies against mouse CD45/B220, CD8α, and MHC Class II (anti-I-A^d/I-E^d (clone 2G9, ThermoFisher) for 30 minutes at 4°C, washed and then stained with streptavidin-Alexa Fluor 350 (Invitrogen) for an additional 15 minutes at 4°C. Cells were then washed twice in PBS + 5% FCS and fixed with 2% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) for 20 minutes at room temperature. Following this initial fixation step, the cells were subjected to further fixation using a proprietary FoxP3 Fixation Solution (eBioscience™ Foxp3 / Transcription Factor Fixation/Permeabilization Concentrate and Diluent, eBiosciences) for 30 minutes at room temperature. The cells were then washed 3 times in ICS permeabilization buffer (PBS with Ca²⁺ and Mg²⁺, 1 mM HEPES, 0.1% Saponin, and 0.05% NaN₃), and blocked with permeabilization buffer containing 1 μg/ml 2.4G2 and 5% normal mouse serum (Jackson Immunoresearch) for 30 minutes at room temperature. The cells were then labeled at 37°C for 1 hour with recombinant IL-21R Fc Chimera (R&D Systems, Minneapolis, MN) and fluorochrome conjugated monoclonal antibodies against mouse IL-2 (clone JES-5H4, PE-CF594, BD Bioscience), TNF-α (clone XT22, PE-Cy7, ThermoFisher), IFN-γ (clone XMG1.2, Alexa Fluor 700, ThermoFisher), IL-17A (clone eBio17B7, PerCP-Cy5.5, eBiosciences), FoxP3 (clone FJK-15s eFlour 450, eBioscience), and Alexa Fluor 647 labeled anti-mouse IL-4 produced from the B-cell hybridoma 11B11 (BEI Resources, Manassas, VA). Samples were then washed twice with permeabilization buffer followed by a single wash with PBS, and then resuspended in PBS + 5% FCS. All samples were subsequently analyzed using a BD LSRII flow cytometer with a 5-laser configuration (355 nm, 405 nm, 488 nm, 561 nm, and 640 nm) using FACSDiva software. Data were analyzed using a Boolean gating strategy to quantitate different cytokine-producing and FoxP3 expressing CD4 T cell subsets with FlowJo software (TreeStar, Ashland OR). For each sample from the in vitro DO11.10 T cell priming assay, expression of the six different cytokines (IFN-γ, IL-2, IL-4, IL-17A, IL-21, and TNF-α) and one transcription factor (FoxP3) that were interrogated by FACS were used to calculate frequencies and absolute numbers of Th1, Th2, Th17, Tfh, and iTreg cells that resulted from each priming condition tested. A novel metric was then calculated for each data set to weigh the magnitude of the Th1 response, penalizing it for the extent to which responses were of mixed phenotype, as opposed to pure Th1 cells only (the ‘D-score’; see Supplemental Fig. 2 for details), and so reflecting both the strength and purity of the Th1 polarization of DO11.10 T cells in the cultures. Three separate screens were performed on separate days, and the calculated D-scores from these replicate screens were normalized to standard deviations from the mean within each data set for each experimental condition to calculate D-scores, using Prism software.

CD4 T cell adoptive transfer and analysis of in vivo polarization

Single cell suspensions from spleens were collected from Rag^−/−DO11.10, OT-II^GFP, or P25^GFP age and sex matched donor mice. In the case of OT-II^GFP and P25^GFP donors, CD4 T cells were enriched by magnetic negative selection using a CD4 T cell isolation kit (Miltenyi) with the addition of biotinylated anti-CD44 (clone IM7, ThermoFisher) to deplete CD44^hi antigen experienced cells. Enriched CD4 T cells were washed with sterile PBS, counted and injected i.v. into sex, background strain and age-matched recipient mice via lateral tail vein at 2 × 10⁴ cells per recipient mouse. For analyses of transferred cells, single cell suspensions from recipient mouse spleens and inguinal lymph nodes were treated with Liberase and DNAse (Roche) in serum free media for 30 minutes at 37°C. Single cell suspensions were prepared by disrupting organs with a syringe plunger through a 70-μm cell strainer. To remove red blood cells, splenocytes were treated with RBC lysing buffer Hybridmax (Sigma) and washed with PBS. For intracellular cytokine staining, single cell suspensions were incubated for 5 hr with appropriate peptide antigen, monensin (2 mM; Sigma-Aldrich), brefeldin A (5 mg/ml; Sigma- Aldrich), anti-CD28, anti-ICOS, and anti-CD49d in DMEM medium supplemented with HEPES, penicillin-streptomycin (Life Technologies), 10% FBS, 2-ME, essential amino acids, and nonessential amino acids (Life Technologies) in 96-well round-bottom polypropylene plates. The final concentration of OT-II peptide (ISQAVHAAHAEINEAGR) and P25 peptide (FQDAYNAAGGHNAVF) was 5 mg/ml (synthetic peptides obtained from Sigma). After restimulation, cells were washed and resuspended in PBS and stained with viability dye (LIVE/DEAD Fixable Blue Dead Cell Stain; Molecular Probes) and subsequently stained with fluorochrome-conjugated Abs against surface markers in PBS containing FBS (2%) and sodium azide (0.05%; Sigma-Aldrich). In some experiments, cells were fixed using paraformaldehyde (2% in PBS; Electron Microscopy Sciences), permeabilized (Fixation & Permeabilization Buffer; eBioscience), and stained for intracellular cytokines. or transcription factors. Stained cells were then analyzed by flow cytometry using an LSRII bench top flow cytometry (BD Bioscience) using gating strategies illustrated in Supplemental Fig. 3.

Bone marrow chimeras

6–8 week old female C57BL/6 recipients were lethally irradiated with 1000 Rads using a Shepard Mark 1 Cesium-137 irradiator and reconstituted with 1 −2 × 10⁶ bone marrow cells i.v. from a single age and sex matched Mgl2-DTR⁺ donor. Recipient mice were maintained on antibiotic water for two weeks and rested for 3 additional weeks prior to use.

Immunizations

Mice were immunized with ovalbumin by intradermal injections at the base of the tail or i.v. injection into the lateral tail vein. Endotoxin free ovalbumin (EndoFit Ovalbumin, Invivogen) was used with a dose of 10 μg per injection in sterile PBS. Where noted, adjuvants were added to the immunogen, including Vaccigrade CpG-B (20 μg), Pam2CSK4 (10 μg), Pam3CSK4 (10 μg), high molecular weight Poly(I:C) (50ug) (all purchased from Invivogen), and MALP2 (1 μg, purchased from Enzo Life Sciences). For immunizations with live BCG, M. bovis BCG-Danish was acquired from the Statens Serum Institute (Copenhagen, Denmark) and cultured in Sauton medium (Difco Laboratories, BD Diagnostic Systems, Sparks, MD) with .05% Tween-80. Bacteria were grown from low passage number frozen stocks to mid-log phase and frozen in medium with 5% glycerol at −80°C in single use aliquots. For immunization, bacteria were thawed, washed and resuspended in PBS containing 0.05% Tween-80, and sonicated to obtain single-cell suspensions. Mice were vaccinated with 5 × 10⁶ colony forming units (CFU) i.d. at the base of the tail or i.v. via lateral tail vein.

Serum ELISA

Serum samples were collected by retro-orbital bleeds from immunized mice. ELISA was performed by coating flat-bottom 96-well EIA/RIA high binding polystyrene plates (Corning) with 1 μg/well ovalbumin in 0.2 M carbonate buffer, followed by blocking with 1% BSA in PBS. Serial dilutions of sera were added to wells were incubated overnight, followed by washing with PBS and subsequent addition of primary biotinylated rabbit anti-Mouse IgG subclass specific polyclonal antibodies (ThermoFisher). After further incubation for 2 hours at room temperature, the wells were washed with PBS, followed by addition of Streptavidin-HRP (ThermoFisher) for two hours, final washing with PBS and addition of One-Step Turbo TMB-ELISA substrate (ThermoFisher). Color development was allowed to proceed for up to one hour, and absorbance was read at 450 nm (A₄₅₀) using a Victor plate reader (Perkin Elmer, Waltham, MA). Background A₄₅₀ levels obtained for wells incubated with PBS only (i.e., no primary antibody) followed by Streptavidin-HRP were subtracted from experimental values, and non-linear regression was applied to the titration curves. The EC50 values (e.g., 50% maximum binding) were calculated from these curves and expressed as the reciprocal of the titer generating this level of binding. Levels of antibody binding in this assay for animals immunized with ovalbumin in PBS only in absence of any adjuvant generated values that were indistinguishable from naïve control sera.

Fluorescent microscopy of fibroblastic reticular cells

A monolayer of adherent FRCs was cultured on glass coverslips in DMEM supplemented with HEPES, penicillin-streptomycin, 10% FBS, 2-ME, essential amino acids, and nonessential amino acids. Cover slips were removed when cells reached 80% confluence and washed with PBS. Cells were fixed with 4% PFA, washed, and permeabilized with 0.1% Triton X-100. Cells were then blocked with 10% FBS in PBS. Cells were stained with 0.5 units Texas Red-X phalloidin (ThermoFisher) in PBS with 10% FBS for 1 hour in the dark and subsequently washed. The nuclear stain DAPI was added, and cells were mounted on glass slides with ProLong Gold antifade medium (ThermoFisher). Cells were visualized using an inverted fluorescence microscope with a 20X objective, and images were captured with an Axiocam MRm CCD camera using ZEN software (Zeiss USA, Thornwood, NY). Images were subjected to linear adjustment for overall brightness.

Statistical Analysis

GraphPad Prism 7 software was used for statistical analyses. Unless otherwise stated, the simple one-way Analysis of Variance (ANOVA) with Bonferroni post-test for multiple comparisons was used to generate adjusted P values for the indicated comparisons. Values for P less than or equal to 0.05 were considered significant.

Results

Establishment of in vitro co-culture model supporting CD4 T cell expansion and differentiation

Secondary lymphoid organs contain a network of fibroblastic reticular cells (FRCs), which provide a scaffold for efficient interaction between T cells and APCs and support for T cell survival and function (13, 14). We hypothesized that co-culture of splenocytes with FRCs would facilitate in vitro modeling of the in vivo CD4 T cell priming environment, and potentially enable high throughput screening of adjuvants to predict their impact on CD4 T cell priming during vaccination. As a first step toward developing such a predictive in vitro priming system, we established cultures of normal FRCs from BALB/c spleen using previously published methods (12). The cultivation of CD45 depleted splenocytes under the conditions used generated homogeneous populations of viable fibroblastic cells that appeared capable of long term survival and growth in a contact inhibited manner. These cells were uniformly positive for the established FRC surface marker podoplanin, and lacked detectable CD31 which distinguished them from other types of resident nonlymphoid cells in lymphatic tissues (13) (Fig. 1A).

Figure 1. Design and validation of an in vitro CD4 T cell polarization assay.

A) Surface expression of podoplanin and CD31 on irradiated splenic FRCs from BALB/c mice as measured by flow cytometry (left panel). Microscopic image of isolated FRCs in culture stained with phalloidin and DAPI to determine cell morphology (right panel, scale bar 100 μm). B) Naive Rag^−/− DO11.10 CD4 T cells were cultured in the presence of vehicle control, murine IL-7 (20 ng/ml), 5% FRC conditioned growth medium or co-cultured with monolayer of irradiated FRCs. CD44^lo Rag^−/− DO11.10 T cells were quantitated after five days by flow cytometry. Mean ± SEM for at least 5 replicates is shown, and data are representative of 3 independent experiments. ****P ≤ 0.0001 compared to control (one-way ANOVA). C) Expansion of Rag^−/− DO11.10 T cells was quantitated by flow cytometry after no treatment or activation with Curdlan or Poly (I:C) in the absence or presence of irradiated FRCs. Data plotted as mean ± SEM for 3 replicates, and is representative of at least 4 independent experiments. **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 for comparisons of plus adjuvant and plus adjuvant/FRC (one way ANOVA with Bonferroni post-test). D) Effect of adjuvant dose titration on Rag^−/− DO11.10 CD4 T cells expansion. Absolute cell numbers were determined by cell counting beads and flow cytometry. Expansion is represented on a scale of 0–100% of maximum expansion measured for each individual adjuvant. Data are representative of at least 4 independent experiments.

It is known that FRCs produce IL-7 and potentially other factors that improve survival and support proliferation of naive CD4 T cells(13). We thus assessed the ability of our cultured FRCs to carry out these functions using naïve ovalbumin-specific CD4 T cells from the spleens of DO11.10 TCR transgenic mice. A mixture of normal BALB/c splenocytes and DO11.10 splenocytes (40:1 ratio) was cultured in complete medium without antigen (OVA) for five days, and then recovered and analyzed for residual DO11.10 cell numbers by FACS. Cultures in medium alone showed recovery of only ~20% of the input cells, which was increased approximately 3-fold by addition of recombinant IL-7 and more than 6-fold by co-culture with splenic FRCs (Fig. 1B). Conditioned medium from FRCs did not replicate the effect of the co-culture, suggesting that factors other than secreted cytokines may also play a role in the pro-survival effect. To assess the effect of FRCs in this co-culture model on antigen specific stimulation and expansion, we added OVA to the co-culture system, with or without FRCs or inflammatory PRR ligands as adjuvants (Fig. 1C). The expansion of OVA-specific DO11.10 T cells to OVA stimulation was minor with FRCs alone in the absence of PRR ligands. In contrast, addition of PRR ligands (LPS 100 ng/ml or Curdlan 10 μg/ml) provided a strong adjuvant effect with respect to dose dependent expansion to OVA stimulation, which was moderately enhanced by inclusion of FRCs in the culture. Expansion of DO11.10 T cells in the co-culture system containing FRCs and antigen (25 μg/ml OVA) showed a dose dependent response to PRR agonists with peak effects at between 10 to 100 ng/ml for most PRR agonists tested, including CpG-B, LPS and Pam3CSK4 which activate through TLR9, TLR4, and TLR1/2 respectively (Fig. 1D). This result indicated that co-culture of CD4 T cells with irradiated FRCs not only supported improved naïve CD4 T cell survival in vitro, but also improved adjuvant-associated CD4 T cell expansion.

Differentiation of naïve CD4 T cells to Th1 phenotype in the in vitro priming system

Since the in vitro co-culture model incorporating FRCs showed enhanced support of survival and expansion of naïve T cells in response to antigen plus adjuvants, we next assessed the utility of this system to support differentiation to well defined functional CD4 T cell subsets. Our initial focus for this was Th1 differentiation, which can be reliably induced in vitro using TCR stimulation in the presence of IL-12p70 and anti-IL-4 (15). Thus, a mixed culture of BALB/c and DO11.10 splenocytes (40:1) and FRCs were cultured with antigen (OVA 100 μg/ml) plus anti-IL-4 antibody and IL-12p70 to stimulate Th1 polarization. Transgenic DO11.10 T cells were assessed by flow cytometry for production of signature effector cytokines indicating polarization state using intracellular staining after 6 days of culture (Figs. 2A). A majority of polarized DO11.10 T cells became IFN-γ producing Th1 cells, while relatively few polarized to IL-21 producing follicular helper T cells (Tfh) and FoxP3 expressing regulatory T cells (Tregs) and essentially none differentiated to IL-4 producing Th2 or IL-17A producing Th17 cells (Fig. 2B). Despite the greater yield of OVA-specific T cells recovered from cultures containing FRCs, the distribution of DO11.10 T cells into the different polarized subsets was not significantly different in cultures with or without FRCs (Fig. 2C), These results indicated that the mixed co-culture model for in vitro priming could faithfully replicate strong polarization of naïve DO11.10 T cells to a Th1 phenotype.

Figure 2. In vitro priming assay for Th1 responses.

A) Mixed priming culture demonstrates accurate CD4 T cell priming with soluble cytokine or adjuvant induced priming. Rag^−/− DO11.10 cells were cultured in the presence of plate immobilized anti-CD3 and anti-CD28, soluble murine IL-12p70 and anti-murine IL-4. After seven days, cells were stained for intracellular cytokines and transcription factors and analyzed by flow cytometry. Representative dot plots are shown. B) Quantitation of polarized Rag^−/− DO11.10 CD4 T cells under Th1 polarizing conditions assessed by intracellular staining of transcription factors and analyzed by flow cytometry. Analysis of 12 replicates is shown. C) Distribution of DO11.10 T cells into various polarized subsets in mixed priming cultures stimulated with OVA and either containing FRCs (filled bars) or without FRCs (open bars). D) Strength and purity of Th1 priming by individual adjuvants in the in vitro priming system was determined by flow cytometry and analysis of effector T cell subsets. Results are plotted as Z-scores representing the number of standard deviations below or above the mean for the entire panel of 30 adjuvants. Results shown means ± SEM for data from three separate analyses. *P ≤ 0.05, **P ≤ 0.01, *** P ≤ 0.001, ****P ≤ 0.0001 compared to no adjuvant condition (cross hatched bar), as determined by one-way ANOVA and Bonferroni post-test.

Given the strong influence of PRR agonists on the antigen-induced expansion of naïve CD4 T cells in this system, we next sought to establish the Th1 priming potential of a panel of various PRR agonists when used as adjuvants in the mixed priming culture (Fig. 2D). Using a flow cytometry analysis of DO11.10 T cells recovered from in vitro priming cultures, we identified multiple PRR agonists with significant Th1 biasing adjuvant activity, as measured by combining both the magnitude and polarization of the response. Out of the panel of 30 potential adjuvants tested, the TLR2 agonists MALP2 and Pam3CSK4 established the most robust Th1 priming and expansion. Furthermore, among the 12 adjuvants with positive D-score values, one third of these were known TLR2 agonists (Pam3CSK4, MALP-2, LTA and Zymosan).

In vivo effects of TLR2 agonists on induction of Th1 responses

To assess the accuracy of the in vitro co-culture system for predicting CD4 T cell polarization in vivo in response to priming with antigen plus adjuvants, we used an in vivo CD4 T cell priming model with transfer of naïve transgenic CD4 T cells. Following adoptive transfer of DO11.10 T cells into naïve BALB/c mice, we immunized by i.d. route with OVA alone or in combination with MALP2, Pam3CSK4, or CpG-B (Fig. 3A). While OVA-specific CD4 T cells expanded in response to i.d. administration of antigen alone or combined with each of these adjuvants, we were surprised to find that DO11.10 T cells failed to generate Th1 cells in response to TLR2 agonists Pam3CSK4 or MALP2 given that these were the most strongly Th1 polarizing adjuvants in the in vitro screening. Given the well-known inherent bias of BALB/c mice toward Th2 versus Th1 responses(16), we carried out the same experiment using adoptive transfer of OT-II^GFP T cells into C57BL/6 recipients, which generally show robust Th1 priming. As in the case of the DO11.10 model, the OT-II^GFP cells also expanded in response to i.d. administered antigen and adjuvants, but showed little polarization to Th1 effector type based on T-bet expression when immunized with ovalbumin plus TLR2 agonists (Fig. 3B). Conversely, the TLR9 agonist CpG-B was an effective Th1 priming adjuvant in both of these in vivo models following i.d. immunization, while TLR2 agonists were effective at inducing Th2 (Gata-3+) and TfH (CXCR5+ Bcl6+) responses but failed to stimulate differentiation to Th1 cells.

Figure 3. TLR2 ligands demonstrate route-specific CD4 T cell polarization.

A) DO11.10 splenocytes (2 × 10⁴) were transferred into each of five age and sex matched BALB/c female mice. One day later, recipients were immunized i.d. with ovalbumin alone or in combination with adjuvant (MALP2, Pam3CSK4 or CpG-B as indicated). Skin draining inguinal lymph nodes were harvested 7 days post immunization, and DO11.10 T cells were stained and analyzed for cell surface markers and transcription factors to identify effector subsets (Th1, Th2 and Tfh). Data shown as mean ± SEM. * P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001compared to ovalbumin alone (two-way ANOVA).

B) Similar analysis as shown in A) except using transfer of 2 × 10⁴ CD44^lo OT-II^GFP cells into age and sex matched C57BL/6 recipients.

C) Rag^−/− DO11.10 T cells were adoptively transferred into groups of four sex and age matched BALB/c mice. Recipients were vaccinated with ovalbumin and different amounts of Pam3CSK4 as indicated. Spleens were harvested after seven days and the numbers of DO11.10 cells were determined by flow cytometry. Data are representative of two independent experiments.

D) Groups of four mice were immunized with ovalbumin and Pam3CSK4 i.d or i.v. at day 0 and again at day 28. After an additional 28 days, sera were collected and titers of ovalbumin specific IgG1, IgG2a and IgG2b were measured by sandwich ELISA and plotted as reciprocal of serm dilutions required to give 50% of maximal binding. **P ≤ 0.01, ***P≤ 0.001 for comparisons of serum titers in i.d. versus i.v. immunized mice (Mann-Whitney test). Data are representative of two independent experiments.

Given the known effects of the route of antigen administration on biasing the differentiation of CD4 T cells in vivo(17), we next considered that the apparent discrepancy between the strong Th1 responses observed with TLR2 agonists in vitro and in our initial in vivo priming studies may have been related to the i.d. route that was used for immunization of the mice. We assessed this possibility directly by transferring DO11.10 T cells into naïve recipient BALB/c mice, and immunizing with OVA and a range of doses of Pam3CSK4 either i.v. or i.d. (Fig. 3C). After 7 days, splenocytes were harvested and DO11.10 T cells were analyzed by flow cytometry. We observed that i.v. administration of Pam3CSK4 resulted in marked expansion of DO11.10 cells in the spleen in a dose dependent manner, with the majority of TCR transgenic T cells polarizing to the Th1 effector state as predicted by our in vitro data. In striking contrast, i.d. administration showed little to no Th1 responses to OVA administered with Pam3CSK4 throughout the entire dose range.

Specific IgG isotypes are well established to be produced in response to immunoglobulin class switching induced by particular CD4 T cell effector subsets (18). In general, class-switching to the IgG2a and IgG2b isotypes is dependent on the Th1 cytokine IFN-γ, whereas switching to IgG1 is dependent on the Th2 cytokine IL-4. To further assess the effector polarization of endogenous OVA specific CD4 T cells in vivo, we examined the isotype of serum antibodies induced by priming and subsequent boosting of BALB/c mice with OVA plus Pam3CSK4 administered by i.v. or i.d. routes (Fig. 3D). Sera were collected from the immunized mice 28 days after boosting, and assayed for anti-ovalbumin IgG isotypes by ELISA. We observed that i.v. immunized animals produced high titers of IgG1, IgG2a and IgG2b, compared to mice that were immunized i.d. which only produced IgG1 in reliably detectable levels. Thus, while i.d. immunization with antigen plus a TLR2 agonist resulted in detectable antibody responses, these were relatively weak and showed no evidence for Th1 mediated effects compared to the robust Th1 helper activity resulting from i.v. immunization.

Overall, our results demonstrated that administration of antigen with TLR2 agonists i.v. resulted in robust Th1 priming in vivo, while administration of the same combination i.d. resulted in the absence of expected Th1 responses. In contrast, this apparent block to Th1 priming with cutaneously administered TLR2 agonists was not observed with the TLR9 agonist CpG-B, which could drive significant Th1 responses when co-administered with antigen by the i.d. route.

Dominant suppression of Th1responses by TLR2 agonists in the skin

To explore the mechanism for the absence of Th1 priming with intradermally administered TLR2 ligands, we carried out experiments to examine whether TLR2 ligands could actively suppress Th1 priming against cutaneously delivered antigens. We used two adjuvants that signal through TLRs other than TLR2 for these experiments. These were CpG-B, which signals through TLR9, and Poly(I:C) which is a ligand for TLR3. It is noteworthy that all TLRs except for TLR3 utilize the adaptor protein MyD88 to initiate the intracellular signaling cascade that results in pro-inflammatory responses and adjuvant effects. On the other hand, TLR3 mediated pro-inflammatory responses are not dependent on MyD88 and instead engage signaling only through the adaptor Trif (16, 19). Using the OT-II^GFP adoptive transfer model, we found that both of these adjuvants stimulated strong Th1 priming when administered i.d. with OVA (Figs. 4 A and B) based on intracellular staining for the transcription factor T-bet and IFN-γ. However, when these were combined with Pam3CSK4, there was a marked suppression of Th1 priming. This inhibition of Th1 priming was observed with both Pam3CSK4, which is a ligand for TLR2/1 heterodimers, or Pam2CSK4 which signals through TLR2/6 heterodimers (Figs. 4 A and B) (20). The inhibition of Th1 priming was not associated with skewing of the OT-II cells to either Th2 or Th17 effector subsets, based on analysis by intracellular transcription factors (Gata-3 or RORγt) or signature cytokines (IL-4 or IL-17A). Thus, signaling via TLR2 in the skin or the draining lymph nodes of this tissue actively suppressed the Th1 priming adjuvant effects stimulated by both Trif and MyD88 dependent signaling.

Figure 4. TLR2 ligands suppress Th1 responses during skin immunization independent of skin specific resident dendritic cells.

A) Purified CD44^lo OT-II^GFP T cells (2 × 10⁴) were transferred into groups of four age and sex matched C57BL/6 female mice. One day later, in plus CpG-B alone or in combination with Pam2CSK4 or Pam3CSK4 as indicated. OT-II^GFP cells were harvested from draining inguinal lymph nodes seven days post immunization and stained for Th1 (T-bet), Th2 (Gata3), and Th17 (RORɣt) polarization specific transcription factors and analyzed by flow cytometry. Data show mean ± SEM for triplicate samples, and are representative of at least three independent experiments.

B) Similar to analysis in A) except recipient mice wer immunized i.d. with poly(I:C) alone or in combination with Pam2CSK4 or Pam3CSK4 as indicated.

C) Purified CD44^lo OT-II^GFP T cells (2 × 10⁴ ) were adoptively transferred into groups of four sex and age matched huLangerin-DTR^+/− mice or control huLangerin-DTR negative littermates, and treated with 200 ng diphtheria toxin i.p. Recipient mice were immunized i.d. 24 hours later with ovalbumin plus a mixture of Pam3CSK4 and CpG-B. Draining inguinal lymph nodes were harvested seven days post immunization and analyzed as above.

D) Purified CD44^lo OT-II GFP T cells (2 × 10⁴) were adoptively transferred into groups of four age and sex matched BatF3^−/− littermates. Recipient mice were immunized i.d. 24 hours later with ovalbumin plus a mixture of Pam3CSK4 and Poly(I:C), or with ovalbumin combined with Poly(I:C) alone. Draining inguinal lymph nodes were harvested 7 days post immunization and analyzed as above. Data are representative of at least 2 independent experiments. (D) 2×10⁴ purified CD44^lo OT-II GFP cells were adoptively transferred into groups of four age and sex matched Mgl2-DTR bone marrow chimeras or wild type C57BL/6 recipient mice. Recipients were treated with 200 ng diphtheria toxin i.p. 24 hours before immunization with ovalbumin plus the mixture of Pam3CSK4 and Poly(I:C), or ovalbumin with Poly(I:C) alone. Draining inguinal lymph nodes were harvested seven days post immunization and analyzed as above. Data show mean ± SEM for triplicate samples. *P ≤. 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, n.s., not significant (two-way ANOVA).

Analysis of resident skin dendritic cells in suppression of Th1 priming

A potential explanation for the ability of ligands for TLR2 to suppress Th1 priming in a tissue specific manner, while ligands of other MyD88 dependent TLRs such as TLR9 stimulate such responses, would be the selective expression of TLR2 on a subset of cells in the skin that mediates the suppression when activated. The skin is populated by a heterogeneous group of tissue specific DCs. Langerhans cells populate the epidermis while two subsets populate the dermis distinguished by expression of the C-type lectin Langerin (21). The Langerin⁺ DCs in the dermis are closely related to CD8α⁺ DCs found in secondary lymphoid organs, while Langerin negative DCs are more closely related to CD11b⁺ DCs. To determine the involvement of skin resident DCs we used a variety of genetic models to examine the suppressive phenotype in each subset’s absence.

First, to investigate the involvement of epidermal Langerhans cells in the mechanism of Th1 priming suppression, we transferred OT-II^GFP cells into naïve human langerin-DTR⁺ and wild type control litter mates. Previous detailed characterization of the human langerin-DTR⁺ mice, which express a diphtheria toxin receptor under transcriptional control of the human langerin promoter, shows that treatment with diphtheria toxin (DT) enables specific ablation of epidermal Langerhans cells (Supplemental Fig. 4) but not langerin⁺ dermal DCs (22). We treated each group with 200 ng of DT to ablate Langerhans cells, and then immunized groups i.d. with ovalbumin combined with a mixture of CpG-B and Pam3CSK4 (Fig. 4C). Th1 responses were suppressed in both the presence and absence of Langerhans cells, indicating that Langerhans cells were dispensable for TLR2-mediated Th1 suppression in the skin.

We next carried out experiments to examine the potential role of langerin⁺ dermal DCs in suppression of Th1 priming by TLR2 agonists in the skin. For this, we used BatF3^−/− mice, which lack a critical transcription factor for the development of CD8α⁺ DCs as well as langerin⁺ dermal DCs and thus constitutively lack these populations (23). We immunized BatF3^−/− mice i.d. with ovalbumin and Poly(I:C) as a control or in combination with Pam3CSK4 (Fig. 4D). BatF3^−/− mice that received only OVA and Poly(I:C) demonstrated strong Th1 responses, indicating that there is no severe defect in Th1 priming in Batf3^−/−animals. However, the addition of Pam3CSK4 suppressed Th1 responses to a similar extent as in WT mice. These data suggested that langerin⁺ dermal DCs did not play a major, non-redundant role in the TLR2 mediated suppression of Th1 responses in the skin.

Lastly, to assess the role of the langerin negative DC subset we made use of the Mgl2-DTR mice in which DT treatment ablates both langerin negative dermal DCs and Langerhans cells (24). To avoid ablation of Langerhans cells, which are regenerated locally in the skin and not from bone marrow precursors in adult mice, we constructed Mgl2-DTR bone marrow chimeras. After successful engraftment of Mgl2-DTR bone marrow stem cells into WT mice, we treated the chimeras and wild type control mice with DT followed by adoptive transferred of OT-II^GFP cells. Twenty-four hours later, we immunized i.d. with OVA combined with the mixture of Poly(I:C) and Pam3CSK4. A control group of wild type mice was also immunized with OVA plus Poly(I:C) only, as a positive control for Th1 priming (Fig. 4E). This showed inhibition of Th1 priming in the Mgl2-DTR chimeras that was comparable to the similarly treated wild type mice, indicating that ablation of langerin negative DCs did not restore Th1 priming. These data, taken together, indicated that none of the known cutaneous DC subsets was individually necessary to suppress Th1 priming in response to cutaneous administration of TLR2 ligands.

Association of suppression of Th1 priming with rapid monocyte infiltration of draining lymph nodes

To better understand the spatiotemporal context of the priming environment of the skin draining lymph node (sdLN) in response to Pam3CSK4, we immunized naïve BALB/c mice i.d. with OVA alone or in combination with CpG-B or Pam3CSK4 and analyzed the myeloid cells arriving in the sdLN at 2 hours post immunization (Figs. 5 A and B). Based on flow cytometry using staining for CD11b and Ly6C, antigen alone or in combination with CpG-B showed very modest increases in neutrophil recruitment and no detectable monocyte recruitment at this early time point. Conversely, immunization with OVA plus Pam3CSK4 induced a striking infiltration of the sdLN by both neutrophils and monocytes. We then analyzed the numbers of resident monocytes and neutrophils in the sdLNs after i.d. immunization with OVA plus CpG-B or Pam3CSK4 at 24, 72, and 96 hours (Fig. 5C). Interestingly, numbers of monocytes in the sdLN were similar between CpG-B and Pam3CSK4 treated mice at 24 hours, but substantially increased in CpG-B treated mice through 96 hours, while monocyte numbers in Pam3CSK4 treated mice begin to decrease after 72 hours. Conversely, Pam3CSK4 treated mice retained large numbers of neutrophils at 24 hours in the sdLN. but numbers declined markedly by 72 hours. In contrast, CpG-B treated mice showed a measurable increase in neutrophils only at 96 hours.

Figure 5. Intradermal administration of Pam3CSK4 induces rapid infiltration of myeloid cells that are necessary for Th1 suppression.

A) Groups of three mice were immunized i.d. with ovalbumin in combination with CpG-B or Pam3CSK4. Skin draining inguinal lymph nodes were harvested two hours post immunization, stained for cell surface markers. After doublet and viability discrimination, monocytes were identified as CD11b⁺ Ly6C^hi and neutrophils are identified as CD11b⁺ Ly6C^intermediate. Representative dot plots are shown with proportion of live single cells present in draining lymph node.

B) Proportions of monocytes and neutrophils of total cells are shown.

C) Groups of three BALB/c mice were immunized i.d. with CpG-B or Pam3CSK4 and ovalbumin. Monocyte and neutrophil populations in skin draining lymph nodes were analyzed by flow cytometry at 24, 48 and 96 hours post immunization. Total numbers of monocytes and neutrophils in the inguinal lymph node are shown.

D) Groups of at least 3 CCR2-DTR⁺ and C57BL/6 females received 2×10⁴ OT-II GFP cells and were treated with 200 ng diphtheria toxin i.p. 16 hours prior to immunization with ovalbumin in combination with Pam3CSK4 and CpG. After seven days, skin draining inguinal lymph nodes were harvested, restimulated, stained and analyzed by flow cytometry.

E) Repeat of experiment in panel D except that an additional group of C57BL/6 mice was included that received antigen with CpG only, and analysis was restricted to intracellular IFNγ and Tbet. Results in panels A through C are representative of at least two similar experiments. All data are plotted as mean ± SEM for three replicate samples. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 for pairwise comparisons (two-way ANOVA with Bonferroni post-test).

Overall, these analyses indicated a more rapid influx of myeloid cells in the first few hours after administration with a Th1-suppressing adjuvant like Pam3CSK4, compared to an adjuvant like CpG-B that gave a delayed but more sustained influx of these cells. This suggested that the early (i.e., 2 hours) accumulation of large numbers of myeloid cells from the blood could be required for the suppression of Th1 priming. Given the large influx of monocytes and neutrophils in the sdLN shortly after immunization with Pam3CSK4, we hypothesized that ablation of these cells prior to i.d. immunization with antigen plus a mixture of Pam3CSK4 and CpG-B could restore Th1 responses. To test this, we used CCR2-DTR mice, in which monocytes and other CCR2 expressing cells are rapidly and transiently depleted following a single injection of DT (Supplemental Fig. 4). Following adoptive transfer of OT-II^GFP cells into naive CCR2-DTR⁺ and wild type C57BL/6 mice, both groups were treated with DT, and sixteen hours later all animals were immunized i.d. with ovalbumin combined with the mixture of CpG-B and Pam3CSK4. Analysis of intracellular cytokine and transcription factor expression in OT-II cells recovered from the draining lymph nodes showed that ablation of CCR2⁺ cells prior to the immunization restored Th1 priming to levels that were significantly higher than those in wild type mice immunized in the same way (Fig. 5D). A second experiment using this approach confirmed the significantly increased Th1 priming in CCR2-DTR+ mice receiving antigen CpG plus Pam3CSK4, which reached levels at least as great or greater than were observed in wild type mice immunized with antigen plus CpG alone (Fig. 5E). These results implicate a CCR2⁺ cell in the mechanism of suppression of Th1 priming, and overall suggest that the monocytes or neutrophils arriving in the sdLN within 2 hours after exposure to TLR2 ligands may play a central role.

Potential effects of Th1 suppression by TLR2 agonists on cutaneous immunization with BCG

TLR2 is critical in the recognition and immune response to many bacteria, and particularly in the case of those which have an abundance and diversity of lipoproteins in their membranes. Among these, mycobacteria are prominent examples, as both Mycobacterium tuberculosis and the attenuated M. bovis BCG strain harbor close to 100 genes that encode putative lipoproteins (25) . Some of these, such as the 19 kDa lipoprotein, are known to be highly expressed during infection and contribute to pathogenesis and immune evasion (26). We hypothesized that live attenuated bacterial vaccines, such as BCG, when administered i.d. would be likely to engage the Th1 suppressing properties we discovered using pure synthetic TLR2 agonists such as Pam3CSK4. Thus, it would be expected that BCG vaccination would elicit greater Th1 responses when administered i.v. compared to the clinically relevant i.d. route. To demonstrate this, we transferred CD44^lo P25^GFP T cells, specific for the P25 peptide of the mycobacterial antigen Ag58b, into naïve recipient mice. These were immunized with 5 × 10⁶ CFU of live BCG-Danish i.v. or i.d., and analyzed 10 days later for P25^GFP polarization in the spleen and sdLNs (Figs. 6 A and B). We observed significant increases in the proportion of Th1 responses to BCG in the spleen when administered i.v. compared to i.d. Conversely, we found that a significantly higher proportion of P25 cells were Th2 polarized in the sdLN and spleen when the vaccine was administered by the standard i.d. route.

Figure 6. Intravenous immunization with BCG augments Th1 polarization compared to the intradermal route. Age matched naïve C57BL/6 females received 2 × 10⁴ P25^GFP cells i.v. After 24 hours, mice were immunized with 5 × 10⁶ CFU BCG-Danish i.d. or i.v. Spleens and skin draining inguinal lymph nodes were harvested 10 days post immunization. Cells were subsequently stained for cell surface markers and transcriptions factors, and donor P25^GFP cells were analyzed by flow cytometry for expression of T-bet, Gata3, and RORγT.

A) Representative plots of flow cytometry analysis of endogenous and donor P25^GFP CD4⁺ T cells resident in the spleen after i.d. (left) and i.v. (right) immunization with BCG. Percentages of T-bet⁺ donor P25^GFP cells as a proportion of total CD4 T cells are shown. B) Quantitation of transcription factor positive P25^GFP cells plotted as percentage of GFP⁺ cells positive for T-bet, Gata-3 or RORγt. Data show mean ± SEM for three replicate samples. * P ≤. 05, ***P ≤. 001 for comparisons between i.d. and i.v. (two-way ANOVA).

Discussion

Vaccine development relies on the ability to design formulations that elicit strong and effective immune responses to one or more protective epitopes. A comprehensive understanding of the innate recognition of potential vaccine adjuvants and the resultant CD4 T cell responses is critical to inform successful formulation strategies. However, widely accepted methods to accurately predict CD4 T cell polarization in response to adjuvants in a high throughput manner do not currently exist. In the present study we describe a novel in vitro high-throughput screening system to accurately predict the CD4 T cell priming potential of a wide variety of vaccine adjuvants. Accurate replication of the CD4 T cell priming environment is critical to the design of an in vitro screening system for adjuvant induced polarization. Fibroblastic reticular cells are increasingly appreciated as important mediators of CD4 T cell survival in secondary lymphoid organs through multiple mechanisms (27, 28). We have demonstrated that co-culture of splenocytes on a monolayer of irradiated splenic FRCs significantly improved survival of naive CD4 T cells ex vivo. Moreover, we have also demonstrated that CD4 T cell priming in this system is predictable and clonal expansion is improved by co-culture with FRCs in response to some adjuvants.

Using this in vitro screening assay we identified TLR2 agonists, including Pam3CSK4 and other elated synthetic lipoprotein mimics, as exceptionally strong Th1 promoting adjuvants. In vivo, after administration of Pam3CSK4 i.v. we observed similarly robust Th1 priming as seen in the in vitro screen. However, when we administered TLR2 ligands through the more clinically relevant i.d. route for vaccinations, we failed to induce significant Th1 priming. This route specific loss of Th1 priming was interesting as several groups have demonstrated that TLR2 ligands induce a wide range of CD4 T cell responses in various contexts (29–32). Differential responses have been observed in independent studies using different routes. However, this has not been adequately studied directly and the mechanisms involved are not well understood. Surprisingly, in a direct comparison of different immunization routes, i.d. administration of TLR2 agonists suppressed Th1 responses to different TLR ligands that potently activate either MyD88 or Trif dependent signaling pathways, suggesting a cell extrinsic suppressive mechanism. This result led us to hypothesize that a skin resident DC was necessary to mediate suppression of Th1 responses in light of previously studies establishing skin DCs as negative regulators of a range of immune responses (33, 34). Our results showed that suppression of Th1 priming by TLR2 agonists was intact in the absence of each of the known skin DC subsets, suggesting that no one subset is necessary for Th1 suppression. However, we were not able analyze the compensatory ability of other DC subsets. Interestingly, subsets of DCs requiring the transcription factor BatF3 for development, including Langerin⁺ dermal DCs, have been shown to be important for priming Th1 responses, although in our experiments Batf3^−/− mice showed no observable defect in Th1 priming (35). This may be due compensatory mechanisms of other skin DC subsets or the presence of a small number of Langerin⁺ DCs in the sdLNs despite their preferential absence in the skin which has been described specifically in BatF3^−/− mice on the C57BL/6 background (36).

Further characterization of the CD4 T cell priming environment in sdLNs showed TLR2 agonists differentially recruited myeloid subsets compared to TLR9 agonists, notably inducing a very early infiltration of monocytes and neutrophils. Transient ablation of CCR2⁺ cells using CCR2-DTR mice and DT treatment prior to immunization restored Th1 responses, suggesting that this early recruitment of monocytes and initial priming events are important for the suppression of Th1 responses. It is important to note, however, that several skin DC subsets also express CCR2 and are ablated in this model. Notably, monocyte infiltration to the draining lymph node in response to Pam3CSK4 began to decrease after 48 hours, while monocyte numbers increased in CpG-B treated mice through at least 96 hours. Macrophages and DCs, which can be generated from inflammatory monocytes, have previously been shown to carry out suppressive functions in response to TLR2 activation (37–40). It is likely that early infiltrating monocytes encounter soluble TLR2 ligands immediately after immunization in the sdLNs, while monocytes infiltrating at later time points do not see these ligands at such high concentrations. Monocytes are unlikely to be sufficient alone to mediate Th1 suppression in vivo and likely require other tissue-resident cell types for their immediate recruitment to the sdLNs. Further research on the molecular mechanisms of Th1 suppression are required to fully explain the role played by early infiltrating monocytes.

TLR2 heterodimers recognize bacterial lipoproteins and are widely expressed on myelomonocytic cells including DCs but are also found on T cells, hepatocytes, and keratinocytes as well as a wide range of other cell types (41–44). TLR2 dimerizes with TLR1 or TLR6 on the cell surface to recognize triacylated and diacylated lipoproteins, respectively (45). TLR2 may also recognize various other components of the bacterial cell wall and endogenous damage-associated molecular patterns. However it is controversial whether or not non-lipoprotein ligands of TLR2 are efficiently recognized at physiological concentrations (46, 47). Similar to other TLRs except for TLR3, TLR2 signals primarily through MyD88 to activate NF-ϰB and downstream transcription targets (20). However, it has become apparent that immune responses to TLR2 ligands are highly context dependent. Macrophages and some dendritic cells exhibit an immunosuppressive phenotype in response to TLR2 ligation, while TLR2 signaling directly on CD4 T cells leads to strong pro-inflammatory responses (19, 48–50). CD4 T cell priming in response to TLR2 signaling has been demonstrated to promote Th1, Th2, Th17, and regulatory T cells in independent studies using a wide variety of in vitro and in vivo models, and the factors influencing these various outcomes remain to be fully understood (51–56).

TLR2 mediated suppression of Th1 responses may have significant implications for bacterial live-attenuated vaccines and bacterial vectors. Mycobacteria are lipoprotein rich and demonstrate potent TLR2 stimulating activity (57). TLR2 responses to Mycobacteria are critical for pathogenicity, particularly in host macrophages (10, 58). Th1 supported cell-mediated responses are thought to be critical for the effective control of tuberculosis (5). Interestingly, BCG is clinically administered in the skin and can elicit detectable Th1 responses, which is thought to be critical to its protective capacity (5, 59). It was therefore surprising when our data suggested that TLR2 recognition of bacterial lipoproteins in the skin could potently suppress Th1 responses. We hypothesized that Th1 priming in response to i.d. administered BCG was not optimal, and in support of this we observed a significant increase in Th1 priming in response to i.v. immunization over i.d. immunization in mice. Although a number of factors could contribute to this result, including differences in antigen levels and persistence in the lymphoid tissues following i.v. versus intradermal injection of BCG, our findings are suggestive that an active suppression mechanism in the skin could be relevant to limiting the Th1 response following cutaneous immunization with BCG. Furthermore, this result agrees with previous research on BCG immunization in non-human primates demonstrating that Th1 responses to BCG may be further improved by i.v. administration as opposed to i.d,, and that i.v. immunization with BCG provides greater protection from aerosol M.tb infection (60–62). Since i.v. immunization is not clinically feasible due to significant safety concerns, identifying the mechanisms of improved protection from i.v. versus cutaneously administered BCG may be a useful way to gain insight into modifications of BCG that could improve protection from i.d. administration. Our findings suggest that neutralization of TLR2 signaling with antibodies or small molecule inhibitors in the skin during immunization may be a safer and more practical alternative to i.v. immunization for improving the protective efficacy of BCG. In this regard, in vivo protection studies in mice and non-human primates are likely to be informative.

In conclusion, TLR2 ligands promote strong Th1 responses when administered with antigen i.v., but suppress Th1 priming in a route dependent manner when administered into the skin. TLR2 mediated Th1 suppression appears not to be mediated by a single unique subset of skin resident DCs, but CCR2⁺ myeloid cells may be necessary. Overall, these findings highlight the potential for modulation of TLR2 signaling as a target for the improvement of i.d. administered vaccines that contain TLR2 agonists such as BCG and other whole bacterial cell vaccines.

Abbreviations used in this article:

TLR

toll-like receptor

PRR

pattern recognition receptor

MyD88

myeloid differentiation primary response 88

TCR

T cell antigen receptor

DTR

diphtheria toxin receptor

Rag-2

recombinase activating gene-2

GFP

green fluorescent protein

FRC

fibroblastic reticular cells