Synthesis and characterization of gardiquimod-polylactide nanoparticles (TLR7-NPs)
The TLR7-NPs were made by co-nanoprecipitation of TLR7-PLA polymer conjugates and poly(ethylene glycol)-b-poly(lactic-co-glycolic acid) (PEG-PLGA) (Fig. 1a). TLR7-PLA polymer conjugates were synthesized by using gardiquimod, a potent agonist for TLR7 receptor expressed in both mouse and human, to initiate the ring-opening polymerization (ROP) of lactide27, 28, 29. This method allowed for quantitative incorporation of gardiquimod into PLA polymers and resulted in TLR7-PLA conjugates with precisely controlled composition and molecular weights (Fig. S1 and S2). At a monomer/initiator (LA/gardiquimod) ratio of 25, gardiquimod loading was achieved as high as 14.8 wt % with nearly 100% incorporation efficiency. The resultant TLR7-PLA polymer conjugates were then mixed with PEG-PLGA in dimethylformamide (DMF) followed by nanoprecipitation into rapidly stirred water to self-assemble into TLR7-NPs with 77 nm hydrodynamic diameter and narrow size distributions (polydispersity index = 0.105) characterized by dynamic light scattering (DLS) (Fig. 1b).
In the design of TLR7-NPs, gardiquimod was conjugated to PLA polymer through ester linkages and could be released from NPs subjected to hydrolysis of ester bonds under physiological conditions. Released gardiquimod from TLR7-NPs (red) shared the identical UV-vis absorbance spectrum as the original gardiquimod compound (black), providing evidence of releasing unmodified TLR7 agonist without any residual chemical groups (Fig. 1c). To mimic the release profile of gardiquimod from TLR7-NPs in the body, both in circulation and in endosomes/lysosomes, we next conduct the kinetic studies of TLR7-NPs at different pH 5.0 (endosomes/lysosomes) and 7.4 (circulation). Consistent with expectations, gardiquimod was released from NPs in a sustained manner without burst release effects, potentially minimizing the undesired systemic toxicity during the circulation. The release rate of gardiquimod from NPs was accelerated at increased acidities, likely due to faster hydrolysis at lower pH, which is important for robust activation of the intracellular TLR7 receptors once NPs are internalized into cells (Fig. 1d).
In vivo retention, lymph node trafficking and cellular internalization of TLR7-NPs
To determine how TLR7-NPs influence the in vivo trafficking of gardiquimod, gardiquimod in the TLR7-PLA was labeled with AlexaFluor647 (AF647) and co-nanoprecipitated with PEG-PLGA to formulate into NPs (Fig. S3). For comparison, we made gardiquimod adsorbed on alum, termed as TLR7-Alum, as previously reported25 and labeled gardiquimod with AF647 (Fig. S4). These two AF647-labeled adjuvants combined with alum-adsorbed OVA antigen were subcutaneously injected into C57BL/6 mice tail base, and the fluorescence of AF647 at the injection site was tracked by whole-body fluorescence imaging. TLR7-Alum was rapidly cleared from the injection site only 1 day post injection (p.i.), whereas AF647 labeled gardiquimod in the TLR7-NPs persisted for more than 3 days (Fig. 2a). Draining lymph nodes were dissected at days 1 and 3 post injection to evaluate the accumulation of gardiquimod through measuring the fluorescence intensity of AF647. As shown in Fig. 2b, the fluorescence intensity in the draining lymph nodes was significantly enhanced with TLR7-NPs as early as 1 day post injection, and the increased fluorescence intensity was maintained for as long as 3 days post immunization. Compared to TLR7-Alum, TLR7-NPs led to a 20-fold increase at day 1 and a 10-fold increase at day 3 in fluorescence signals in draining lymph nodes after immunization. These results suggest that TLR7-NPs were preferentially targeted and retained in the draining lymph nodes.
Given the preferential in vivo retention and enhanced LN targeting of TLR7-NPs, we next sought to evaluate how TLR7-NPs impact the cellular internalization of gardiquimod at a single-cell level using flow cytometry. Strikingly, there was a 5.5 to 27.8-fold increase in the cellular uptake of TLR7-NPs in the vast majority of antigen-presenting cells (APCs) in draining lymph nodes, including B cells, dendritic cells, macrophages, and Ly6Chi inflammatory monocytes, compared to that of TLR7-Alum as early as 1 day post injection. This enhanced cellular uptake of TLR7-NPs in various APCs remained at a high level as long as 3 days post injection, with a 5.4 to 22.9-fold increase compared to TLR7-Alum (Fig. 2c). Interestingly, the persistent adjuvant capture of APCs led to potent and persistent mobilization of DCs and cells of monocytic lineage in draining lymph nodes (Fig. 2d and Fig. S5).
TLR7-NPs potently boosts and prolongs both GC and extrafollicular B cell responses
The enhanced and sustained activation of innate immune cells by TLR7-NPs encouraged us to assess the subsequent effect on adaptive immune responses. Vaccine efficacy largely depends on germinal center (GC) activity, which selects for high-affinity antibody producing B cells and the formation of durable B cell memory. To test whether TLR7-NPs as adjuvants might alter GC dynamics, two groups of C57BL/6 mice received an equivalent dose of NP-OVA as the antigen along with TLR7 agonist two different forms: one group was immunized with alum-adsorbed NP-OVA and TLR7-Alum and the other group with alum-adsorbed NP-OVA and TLR7-NPs. We then did a time course study from day 4 to day 22 post injection. As shown in Fig. 3a, GC B cells were detected starting at day 7 in draining lymph nodes, and their numbers were significantly higher in the TLR7-NP-treated group compared to TLR7-Alum group. The difference in GC B cell number between the two groups was maintained to day 22, the latest time point we have tested (Fig. 3a). Notably, despite decreasing GC B cell numbers after day 7, GC B cell numbers at day 22 in the TLR7-NP group were still higher than the peak numbers in the TLR7-Alum group. GC selection relies on B-T cell interaction, in which T cells provide essential help to support GC B cells survival and selection30. Consistent with GC B cell kinetics, follicular T cells expressing CXCR5 and PD1 were also highly expanded in TLR7-NP group. Follicular T cells contain T follicular helper cells (Tfh) and T follicular regulatory cells (Tfr). Tfh are the actual helper cells that promote GC responses, whereas Tfr express regulatory T cell marker FOXP3 and function to restrain GC reaction. To characterize the composition, we first gated on total follicular T cells (BCL6+PD1+) and further gated Tfr subset based on FOXP3 expression (Fig. 3b). Strikingly, there were ~ 30–40% follicular T cells on day 7 post immunization from TLR7-Alum group expressing FoxP3 (defined as Tfr) (Fig. 3b). In TLR7-NP group, the percentage of Tfr dropped significantly to ~ 10%. All these data strongly suggest the nanoparticle adjuvant formulation is very potent in promoting GC response. Interestingly, TLR7-NPs could also promote an extrafollicular B cell response with a markedly increased number of early plasmablasts in TLR7-NP group compared to TLR7-Alum group (Fig. 3a). With these drastic changes in B cell characteristics, we found that TLR7-NPs were not only enhanced the magnitude of antibody secretion, but also altered the quality of antibody responses. All mice immunized with TLR7-NPs adjuvanted NP-OVA, in which contains immunodominant epitope NP and subdominant epitope OVA, generated high levels of antibodies against the subdominant OVA epitope. In contrast, 83.3% (5 out of 6) of mice immunized with TLR7-Alum adjuvanted NP-OVA couldn’t (Fig. 3c).
TLR7-NPs adjuvanted vaccine elicits a robust antigen-specific CD8 + T cell response
TLR7-NPs have efficiently induced persistent mobilization and activation of conventional type 1 dendritic cells (cDC1), which are vital for cross-priming CD8+ T cells. This, we sought to determine if TLR7-NPs adjuvant could also enhance antigen-specific CD8+ T cell responses compared to other immunization strategies, which generally poorly stimulate CD8+ T cells. As shown in Fig. 3d, there was a marked enhancement in the number of effector memory (CD44hiCD62Llow) OVA-specific CD8+ T cells in draining lymph nodes 7- and 14-days post immunization of TLR7-NPs adjuvanted vaccines (Fig. 3d). The proportion of CD69hiCD8+ T cells rose significantly at 4 days post immunization and persisted until 14 days post immunization, indicating the early and persistent activation of OVA-specific CD8+ T cells in response to TLR7-NPs adjuvanted vaccination (Fig. 3d). Further, flow cytometry analysis showed TLR7-NPs adjuvanted vaccine efficiently induced Granzyme B (Gzmb) producing CD8+ T cells in draining lymph nodes at 7 days post immunization. In contrast, the TLR7-Alum adjuvanted vaccine failed to induce Gzmb producing CD8+ T cells, demonstrating TLR7-NPs augment not only the quantity but also the quality of cellular immune responses (Fig. 3d).
As CD8+ T cells-mediated killing of intracellular pathogen-infected cells is critical for controlling viral replication in the lung, we next analyzed the CD8+ T cell responses in the lungs after 14 days post immunization. As shown in Fig. 3e, a large number of functional CD8+ T cells expressing Gzmb were observed in the lungs after receiving TLR7-NPs adjuvanted NP-OVA vaccine, whereas Gzmb producing CD8+ T cells were barely detected in the lungs of mice receiving TLR7-Alum adjuvanted vaccines (Fig. 3e). Altogether, TLR7-NPs were superior to TLR-Alum in inducing potent CD8+ T-cell responses.
TLR7-NPs adjuvanted HA immunization induces cross-reactive and stem specific antibodies
Development of a broadly cross-reactive universal influenza vaccine for clinical use has been challenging and thus far, unsuccessful. The primary reason is immunodominance, in which antibody and memory responses are focused on dominant epitopes, which leads to selection pressure and frequent viral escape mutations. The potent effect of TLR7-NP on promoting GC responses suggests their potential to tackle this issue. More importantly, TLR7-NPs adjuvant can drastically increase the ratio of Tfh to Tfr, which might increase the diversity of B cell clones in the GC by providing additional T help. As such, we hypothesized that TLR7-NPs adjuvant might help overcome immunodominance and increase the breadth of antibody response against influenza HA. First, we performed the immunofluorescence staining of the draining lymph nodes from mice who received the immunization of HA plus TLR7-NPs and confirmed more GCs were observed at day 10 post immunization (Fig. 4a). Next, to test whether TLR7-NPs can increase the breadth of the response, we immunized mice with H3 HA from influenza strain A/Hong Kong/1/1968 (HK68), a representative strain from the group 2 influenza viruses, and did serology for antibodies against HAs from heterologous strains at week 2 post primary immunization (Fig. 4b). The result showed that TLR7-NPs not only increased antibody titers against the immunized H3 HA from HK68, but also enhanced antibody responses against HA from another H3N2 strain, A/Hong Kong/4801/2014 (HK14). More importantly, 80% of mice immunized with TLR7-NPs adjuvanted H3 HA were also able to make antibodies against HA from H7N9 (strain SH13), an influenza virus variant from a different subtype. In contrast, these heterosubtypic antibodies could not be detected in the mice immunized with HA adjuvanted by TLR7-Alum. To test whether TLR7-NPs could induce HA-specific antibodies against group 1 viruses as well, we used H1 HA from PR8 in a similar immunization experiment setting. Again, we found that mice immunized with TLR7-NPs adjuvanted PR8 HA generated significantly higher levels of antibodies against HA of cross-subtype or even cross-group (Fig. 4c). The head region of HAs from different influenza strains is highly variable and mutable, while the stem region is more conserved. We hypothesized that antibodies against stem might explain such cross-reactivity. To test if TLR7-NPs could skew the immunodominance from mutable HA head region to conserved stem region, we did ELISA with a headless HA protein derived from PR8 HA. We found that TLR7-NPs adjuvanted PR8 HA immunization did induce a much greater level of antibodies binding to the stem region (Fig. 4c). Next, we immunized mice with H1 HA from another strain (NC99) and tested for antibodies against headless and intact HA from PR8. With the change of immunogen to NC99 H1 HA, TLR-NPs still could enhance the antibody response towards stem region as well as intact HA from PR8 (Fig. 4d), confirming that the stem region is indeed conserved and targeted by cross-reactive antibodies. Hence, using three different HAs as the immunogen, we demonstrated the potential of TLR7-NPs as an adjuvant to boost cross-reactive antibody responses.
TLR7-NPs adjuvanted HA immunization induces broad protection against heterosubtypic influenza viruses
Given the potent cross-reactive antibody responses induced by TLR7-NP assisted vaccination, we reasoned that this adjuvant might help induce cross-protection against different influenza viruses. To evaluate this possibility, we immunized two groups of C57BL/6 mice with an equivalent dose of alum-adsorbed NC99 HA as the antigen along with TLR7 agonist in TLR7-Alum or TLR7-NPs, at week 0 and week 3. Four weeks post boost; mice were challenged with a lethal dose of distinct PR8 H1N1 virus (Fig. 4e). In line with the observation of enhanced cross-reactivity with TLR7-NP adjuvantation (Fig. 4d), mice receiving the TLR7-NPs adjuvanted NC99 HA lost significantly less body weight upon challenge compared to mice receiving TLR7-Alum adjuvanted vaccine (Fig. 4f). Most of the mice that received TLR7-Alum adjuvanted NC99 HA failed to develop cross-protection against PR8 H1N1 viral challenge as only ~ 40 % of mice from this group survived the challenge (Fig. 4g). In contrast, more than 92% of mice immunized with TLR7-NP adjuvanted HA fully recovered after the infection (Fig. 4g). Furthermore, we performed a histologic analysis of lung sections from mice survivors 14 days post PR8 H1N1 viral infection. As illustrated in Fig. 4h, severe pathologic peribronchiolar inflammation associated with diffuse damage of structures were readily identified within the lung tissues of unvaccinated mice and mice who received TLR7-Alum adjuvanted vaccines. In contrast, TLR7-NPs adjuvanted HA immunization fully protected the mice from the infection induced pulmonary damage, maintaining similar architecture of lung tissues compared to uninfected mice (Fig. 4h). Combined, these results strongly demonstrate that a TLR7-NPs adjuvanted HA vaccine not induced cross-reactive antibodies but also offered strong cross protection against heterologous influenza infection.
TLR7-NPs adjuvanted spike immunization induces cross-reactive antibodies against the SARS-CoV-2 variants
The emergence of several SARS-CoV-2 variants with multiple amino acid changes in the Spike (S) protein and especially within the RBD has implications for the future control of the COVID-19 pandemic. Several of these mutations directly affect ACE2 receptor binding affinity, impacting infectivity, viral load, or transmissibility. Variants of concern include the Alpha variant United Kingdom (U.K.) 501Y.V1 lineage B.1.1.7, the Beta variant South Africa (S.A.) 501Y.V2 lineage B.1.351, the Gamma variant Brazil (B.R.) 501Y.V3 lineage P.1, and the Delta variant India B.1.617.2. To test if a TLR7-NPs adjuvant could increase the breadth of the antibody response against these variants, we immunized C57BL/6 mice with the full-length spike protein from SARS-CoV-2 virus with TLR7-Alum or TLR7-NPs as an adjuvant (Fig. 5a) and characterized the antibody responses against the wildtype RBD, spike and S1 protein, as well as RBD and S1 protein from several variants of concern (Fig. 5b and 5c). After two immunizations, all mice immunized with TLR7-NP adjuvanted spike protein (Wuhan reference strain) developed high titers of antibodies against RBD with 501Y mutation and RBD of P.1 and B.1.617 variant, as well as S1 protein from B.1.1.7 and B.1.351 variants. In contrast, the mice immunized with TLR7-Alum adjuvanted spike had significantly lower antibody responses against immunized spike protein. Only 50–60% of them developed detectable antibody responses against these variants. Similarly, for the Indian Delta variant (B.1.617.2), a highly contagious SARS-CoV-2 virus strain now circulating globally, TLR7-NPs were able to induce high titers of antibodies against its RBD domain in 89% of vaccinated mice, whereas only 50% of mice immunized with TLR7-Alum adjuvanted spike generated 1–2 log fold lower antibody titers (Fig. 5c).
TLR7-NPs adjuvanted spike protein subunit vaccine elicits B cell differentiation and antibody response in 3D human tonsil organoid cultures
To further investigate the translational potential of TLR7-NPs as an adjuvant for both mouses and human analyses, we used our recently developed human tonsil organoid system31 to assess the effects of TLR7-NPs when combined with the full-length SARS-CoV-2 spike protein. Organoid cultures were prepared and stimulated with spike protein with or without TLR7-NPs. After 14 days in culture, we assessed the phenotype changes of B cells in the culture and antibody specificities against SARS-CoV-2 spike in the supernatants. As shown in Fig. 6a and Fig. 6b, we observed a significant increase in Pre-GC B cells, GC-B cells and plasmablast differentiation in a subset of donors upon stimulation of Spike with TLR7-NPs compared to Spike protein only controls. Spike-specific IgM and IgA antibodies increased in 5 out of 6 donors when stimulating with Spike + TLR7-NPs compared to Spike alone (Fig. 6c). In addition, we also tested the adjuvanticity of TLR7-NPs in combination with a subunit vaccine currently under clinical testing, comprised of the SARS-CoV-2 spike protein RBD displayed on a I53-50 protein nanoparticle scaffold (termed as RBD-NP)32 in this organoid system. We performed a time-course single-cell RNA sequencing (scRNA-seq) study using the BD Rhapsody platform for sorted B cells from unadjuvanted RBD-NP cultures and RBD-NP + TLR7-NPs stimulated tonsil organoid cultures. We first examined B cells states within our single-cell transcriptomic atlas and identified 8 distinct B cell populations by uniform manifold approximation and projection (UMAP) based on their gene expression profile (Fig. 6d). As shown in Fig. 6e, both the frequencies and the numbers of GC B and plasmablasts (PB) notably increased as early as 4 days’ post stimulation with the RBD-NP plus TLR7-NPs compared to RBD-NP alone. This is well in line with our previous observation that TLR7-NPs induced the early promotion of both extrafollicular responses and GC responses in the mouse (Fig. 3). More importantly, IgM PB, IgA PB, and IgG PB were all significantly increased in RBD-NP plus TLR7-NPs stimulated cultures, indicating the broad effects of TLR7-NPs on the antibody isotype switching of B cells in human tonsil organoids (Fig. 6f). To further understand TLR7-NP-mediated signaling pathways on B cell differentiation, we then compared gene expression within B cells in both RBD-NP alone stimulated- and RBD-NP + TLR7-NPs stimulated cultures. Gene Ontology (GO) analysis showed that genes associated with responses to viruses and type 1 interferon are remarkedly upregulated on 4 -day-old tonsil cultures stimulated with RBD-NP plus TLR7-NPs compared to RBD-NP alone (Fig. 6g).
As possible adjuvant mediated toxicity is an important consideration, we next evaluated this possibility by subcutaneously injecting the TLR7-NP adjuvanted vaccine into C57BL/6 mice, using TLR7-Alum as a control. TLR7-Alum rapidly diffused into circulation within 3 hours after a single injection, leading to acute inflammation, characterized by multiple elevated serum inflammatory cytokines including MCP-1, MIP1B, IFNA, IL-6, IP-10, and GROA (Fig. S6a, left). Among them, GROA and IL-6 maintained at a high level in the serum as long as 24 hours post immunization (Fig. S6a, right). We further evaluated the long-term toxicity by clinical chemistry at day1, day4, and day7 post immunization. The result showed the significantly increased blood urea nitrogen (BUN) enzyme level in TLR7-Alum immunized group at day1, indicating the acute kidney toxicities, presumably caused by fast renal clearance of small molecular TLR7 agonist in this formulation (Fig. S6b). In contrast, TLR7-NPs adjuvanted vaccines showed negligible systemic inflammation and renal toxicity.