Modulation of lipid nanoparticle-formulated plasmid DNA drives innate immune activation promoting adaptive immunity

Summary

Nucleic acid vaccines have grown in importance over the past several years, with the development of new approaches remaining a focus. We describe a lipid nanoparticle-formulated DNA (DNA-LNP) formulation which induces robust innate and adaptive immunity with similar serological potency to mRNA-LNPs and adjuvanted protein. Using an influenza hemagglutinin (HA)-encoding construct, we show that priming with our HA DNA-LNP demonstrated stimulator of interferon genes (STING)-dependent upregulation and activation of migratory dendritic cell (DC) subpopulations. HA DNA-LNP induced superior antigen-specific CD8⁺ T cell responses relative to mRNA-LNPs or adjuvanted protein, with memory responses persisting beyond one year. In rabbits immunized with HA DNA-LNP, we observed immune responses comparable or superior to mRNA-LNPs at the same dose. In an additional model, a SARS-CoV-2 spike-encoding DNA-LNP elicited protective efficacy comparable to spike mRNA-LNPs. Our study identifies a platform-specific priming mechanism for DNA-LNPs divergent from mRNA-LNPs or adjuvanted protein, suggesting avenues for this approach in prophylactic and therapeutic vaccine development.

Graphical abstract

Highlights

• •Higher N/P ratio DNA-LNP formulations improve biophysical profile and immunogenicity
• •DNA-LNP platform induces strong cGAS-STING-dependent innate immune inflammation
• •DNA-LNP platform induces superior T cell responses to mRNA-LNPs and adjuvanted protein
• •Spike DNA-LNP is protective in a lethal SARS-CoV-2 challenge model

Tursi et al. adapt lipid nanoparticle formulations to stabilize plasmid DNA incorporation resulting in an improved biophysical profile. These drive enhanced cellular and humoral immunity in models of influenza and SARS-CoV-2.

Introduction

Vaccine approaches that can elicit both arms of adaptive immunity are valuable tools for combating diverse pathogens as correlates of protection are frequently complex or unvalidated. This is exemplified in an emerging infectious disease scenario, such as the beginning of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. Nucleic acid vaccines can elicit both humoral and cellular immunity as well as have production and redosing advantages relative to other platforms.^1,2,3,4

Lipid nanoparticle-formulated nucleoside-modified mRNA vaccines (mRNA-LNPs) have demonstrated efficacy for the prophylaxis of COVID-19, with two initial product licenses for Moderna’s SpikeVax (mRNA-1273) and Pfizer-BioNTech’s Comirnaty (BNT162b2).^5,6 Immunization with mRNA-LNPs in animal models and humans is associated with robust immunogenicity.^3,7 Translation of mRNA has been reported to generate a strong, short burst of antigen production.⁸ The introduction of modified nucleosides has been shown to reduce innate reactogenicity and improve translation efficiency of in vitro-transcribed (IVT) RNA.^8,9 Recently, beyond mRNA delivery, the lipid nanoparticle (LNP) component has been described as an adjuvant for the resulting immune response.^10,11,12 Some LNP formulations have also been reported to adjuvant recombinant protein vaccines.¹¹

IVT mRNA reactions have distinct synthetic requirements including template DNA, RNA polymerase, and ribonucleotides to form mRNA transcripts as well as a process to add a 5′ cap structure and polyA tail.^13,14 After transcription, IVT mRNA requires purification to remove byproducts of the reaction, such as double-stranded RNA species.¹⁵ Additionally, mRNA thermostability leads to cold chain-dependent storage requirements. Recently, there have been advances for improving mRNA-LNP vaccine stability including lyophilization.¹⁶ Overall, additional gene-vectored vaccine tools remain important to evaluate.

DNA vaccines have demonstrated clinical safety and efficacy while also conferring product profile advantages including temperature stability and simple production.^{17,18,19,20,21,22} Recent advancements in DNA vaccine technology have led to clinical efficacy in human papillomavirus (HPV)-associated recurrent respiratory papillomatosis and liver cancer therapy.^23,24 Naked DNA is poorly immunogenic as has been previously reported, and advancements in device delivery have improved in vivo immunogenicity and clinical impact.¹ Physical transfection modalities for DNA vaccines include in vivo electroporation, gene gun, and jet delivery among others; a licensed DNA vaccine for SARS-CoV-2 (ZyCoV-D) utilizes jet delivery.^25,26,27 The direct encoding of gene adjuvant sequences (e.g., cytokines) to further enhance DNA vaccine-induced responses has increased potency in preclinical models and in the clinic.^23,28,29 Additional tools for understanding and improving in vivo DNA vaccine immunogenicity remain important considering the favorable manufacturing and stability aspects of plasmid DNA.

LNP formulations of DNA have been studied, and LNPs enabling in vivo transfection efficiency have been reported,^30,31,32 with some progress in this area.^33,34 However, lipid-based formulation of plasmid DNA has demonstrated poor in vivo immunogenicity in the clinic.³⁵ It has been hypothesized that DNA delivered in the cytoplasm is inefficiently transferred to the nucleus; early studies using naked DNA as vaccines demonstrate that plasmid delivered to the cytoplasm can be trafficked to the nucleus for transcription.^36,37,38 Critically, LNP-encapsulated DNA could be of value if it retains the positive product profile of DNA vaccines, and device-free formulation would provide additional options in combination with or separate from current delivery systems.

Here, we describe that tailored formulation of plasmid DNA into LNPs (DNA-LNPs) stabilizes particle formation promoting an improved immune profile with a different phenotype relative to mRNA-LNPs or adjuvanted protein. We demonstrate that the relationship between lipid components and DNA is a modulator of biophysical parameters and immunogenicity for DNA-LNPs. Our studies illustrate that these DNA-LNPs exhibit an innate priming phenotype dependent on double-stranded DNA (dsDNA) sensing via cyclic GMP-AMP synthase (cGAS), STING, and TANK-binding kinase 1 (TBK1). Further, DNA-LNP formulations exhibit dose-sparing humoral potency while additionally inducing superior CD8⁺ T cell responses. These findings extended to a rabbit model, with DNA-LNP eliciting responses comparable or superior to mRNA-LNP at the same dose. These data support DNA-LNPs as a next-generation genetic vaccine platform with robust immunogenicity.

Results

Relationship between lipid components and DNA impacts immunogenicity

The association between lipid amine groups to nucleic acid backbone phosphates (N/P ratio) in LNP-formulated nucleic acids can modulate particle size, affecting stability. Plasmid is double stranded (having many more phosphate groups) and normally includes significantly longer sequences compared to mRNA expression cassettes. We examined whether different N/P ratios impact biophysical parameters and immunogenicity for DNA-LNPs. N/P ratios for mRNA-LNPs have been well characterized, with a 6:1 N/P ratio utilized for licensed SARS-CoV-2 vaccines.³⁹ We formulated CA09 hemagglutinin (HA)-expressing plasmid DNA-LNPs (HA DNA-LNP) at a range of N/P ratios including 10.5, 5.3, and 2.6 (Figure 1A). Encapsulation efficiency was similar between 10.5 and 5.3 N/P ratio (91% and 89%, respectively) but lower for 2.6 N/P ratio, approaching 80%. Additionally, 2.6 N/P ratio formulation particles were larger in size with a lower (more anionic) zeta potential (Figures 1B–1D). Nanoparticles with a zeta potential between −10 and 10 mV are considered neutral and associated with less cell wall destruction-associated toxicity seen with cationic particles.⁴⁰ To examine how this relationship impacts immunogenicity, mice were immunized with HA DNA-LNPs at different N/P ratios or empty vector DNA-LNP (pVAX DNA-LNP) at a 10.5 N/P ratio. We used antigen-specific germinal center (GC) and splenic T cell responses as an immunogenic readout. Interestingly, there was a significant attenuation in the frequency of total GC B cells (Figures 1E and 1F) with formulations at 5.3 and 2.6. We also observed a trend toward a decrease in HA-specific GC B cells with the 2.6 formulation (Figure 1G). In the T cell compartment, we observed a similar activated T follicular helper (Tfh) cell response with formulations at all three N/P ratios (Figure 1H). Additionally, in line with our Tfh cell data, antigen-specific effector CD4⁺ and CD8⁺ T cell responses appear unaffected with the different lipid-to-DNA ratios examined. We did not observe a significant difference in interferon (IFN)γ-secreting cells by enzyme-linked immunospot (ELISpot) assay (Figure 1I) or cytokine-expressing CD8⁺ (Figures S1A–S1C) or CD4⁺ T cells (Figures S1D–S1F). Taken together, these data suggest that higher N/P ratios for HA DNA-LNPs can lead to improved immunogenicity exemplified by improved B cell responses within the GC, but both cytotoxic and helper T cell responses remain durable at low N/P ratios.

Figure 1.

Initial immune characterization of HA DNA-LNP formulations and immunogenicity

(A) Schematic of DNA-LNP N/P ratios and immunization regimen.

(B–D) Biophysical characterization of DNA-LNPs at different N/P ratios. (B) Particle size; (C) polydispersity index (PDI); (D) zeta potential.

(E) Representative fluorescence-activated cell sorting (FACS) plots of GC B cells.

(F) Bar plots quantifying frequency of GC B cells.

(G) Frequency of CA09 HA-specific GC B cells.

(H) Frequency of activated Tfh cells.

(I) IFNγ ELISpot of splenocytes.

(J) Representative TEM images of HA DNA-LNP (top) and HA mRNA-LNP (bottom). Scale bar 100 nm

(K and L) Fold change cytokine induction in DLNs at 4 h (K) and 24 h (L) after immunization quantified using Luminex.

(M and N) ELISpot assay measuring IFNα (M) and IFNγ (N) 20 h after stimulation of splenocytes ex vivo with DNA-LNP, plasmid DNA, or DNA-LNP in the presence of chemical inhibitors to the indicated DNA sensors.

(O) Schematic of relevant pathways implicated in DNA-LNP sensing.

Dots represent individual animals; n = 8–9 (E–H), n = 5 (I, K, and L), or n = 3–4 animals per group (L and M); data pooled or representative from two independent experiments (E–H, M, and N) or from one independent experiment (I–L). Plots show mean with SD (B–D and I) or geometric mean with geometric SD (F–H and K–N). Unpaired one-way ANOVA adjusted for multiple comparisons with Bonferroni corrections was used to compare groups (F–H, K, and L) or compared to DNA-LNP control (M and N). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

We performed biophysical characterization of DNA-LNPs and mRNA-LNPs using transmission electron microscopy (TEM). We utilized HA DNA-LNPs and a CA09 HA-expressing mRNA-LNP (HA mRNA-LNP) as sample LNPs. Particles looked phenotypically similar (Figure 1J), supporting additional biophysical characterization data (zeta potential, size, polydispersity index, and encapsulation efficiency) (Table S1). We next assessed antigen expression and biodistribution kinetics after intramuscular immunization using in vivo imaging. Mice were injected once with a firefly luciferase (Luc)-expressing DNA-LNP or mRNA-LNP and imaged longitudinally. After an initial expression signal, Luc DNA-LNPs rebound and signal persists in the muscle through 40 days after immunization (Figures S2A and S2B). In contrast, once the Luc mRNA-LNP signal wanes to baseline, no additional expression is observed, supporting described dynamics of mRNA-LNP expression in vivo.⁴¹ With Luc DNA-LNP, signal was observed only at the site of injection in the tibialis anterior (TA) muscle, whereas mice injected with Luc mRNA-LNP showed luciferase expression in the liver at 4 h that nearly resolved by 1 day (Figure S2B). These data support that DNA-LNPs and mRNA-LNPs have different kinetics of antigen expression.

DNA-LNPs induce cGAS-STING pathway-dependent inflammation

We next sought to characterize the immune response in depth after immunization with HA DNA-LNP formulated at a higher N/P ratio (10.5) in comparison with benchmark HA mRNA-LNP as well as AddaVax-adjuvanted recombinant protein, which is similar to the adjuvant MF59 in Food and Drug Administration-approved Fluad. First, we characterized innate immune sensing of DNA-LNPs in the draining lymph nodes (DLNs) through induction of local pro-inflammatory cytokine production at 4 and 24 h after immunization with 2 μg HA DNA-LNP, 2 μg HA mRNA-LNP, or 1 μg adjuvanted HA protein. Relative to a PBS-only negative control, immunization with both DNA-LNP and mRNA-LNP at 4 and 24 h led to robust upregulation of a number of different pro-inflammatory cytokines, notably IFNγ and interleukin (IL)-6 (Figures 1K and 1L). Levels trended downward over time, which ameliorated responses at 24 h relative to 4 h. mRNA-LNP led to increased IFNγ-induced protein 10 (IP-10) upregulation at 4 h, which was largely similar with DNA-LNP at 24 h. Together, these data suggest that DNA-LNPs, along with mRNA-LNPs, are potent drivers of early inflammatory responses.

To gain a mechanistic understanding of how DNA-LNPs drive inflammatory responses, we treated naive splenocytes ex vivo with HA DNA-LNP in combination with or without various chemical inhibitors of DNA sensing pathways and measured the production of IFNα and IFNγ. Treatment with DNA-LNP alone led to robust production of IFNα and IFNγ, while plasmid DNA alone did not, suggesting that LNP encapsulation of plasmid DNA is driving innate immune inflammation (Figures 1M and 1N). However, in the presence of inhibitors to the cGAS-STING-TBK1 pathway, which senses dsDNA, we observe little to no secretion of either IFNα or IFNγ. This is in contrast to CpG-sensing Toll-like receptor 9 (TLR9), which, when inhibited, does not impact IFNα or IFNγ secretion. Together, these data suggest that sensing of dsDNA via cGAS-STING-TBK1 (Figure 1O) is a driver of inflammatory responses to DNA-LNPs, while TLR9 sensing of CpG motifs is dispensable.

HA DNA-LNP induces robust early adaptive and innate immune activation

Considering the strong inflammatory responses we observed with HA DNA-LNP, we profiled early activation of both adaptive and innate immune populations after vaccine priming. Previous work has demonstrated that both mRNA-LNP vaccines (e.g., BNT162b2) and AddaVax-formulated protein elicit innate immune modulation in the DLN, particularly mediated through myeloid activation.^12,42 We observed robust upregulation of early activation marker CD69 on CD8⁺ T cells, CD4⁺ T cells, and natural killer (NK) cells 24 h after immunization from mice immunized with both DNA-LNP and to a lesser extent mRNA-LNP (Figures S3A–S3F). Among NK cells, DNA-LNP immunization also led to secretion of IFNγ in both DLNs as well as in the spleen (Figures S3G–S3H). We next profiled myeloid populations in depth 24 h after immunization (Figure 2A). There was a robust induction in total migratory dendritic cells (mDCs) in the DLN with DNA-LNP contrasting with a marked decline in these populations after mRNA-LNP or adjuvanted protein immunization (Figure 2B). CD11b⁺ mDCs followed a similar trend, with a decline in the mRNA-LNP and protein groups relative to DNA-LNP (Figure 2C). There was a stark increase in CD103⁺ mDCs with DNA-LNP immunization relative to both control and the two other platforms (Figure 2D), a subset similar to CD8α⁺ resident dendritic cells (rDCs) important for cross-presentation.⁴³ All platforms drove a decrease in the frequency of plasmacytoid dendritic cells (pDCs) (Figure 2E). Immunization with DNA-LNP led to a striking increase in the frequency of neutrophils relative to other platforms or the control naive baseline (Figure 2F), while mRNA-LNP immunization drove an increase in the frequency of monocytes in DLNs (Figure 2G). Using the expression of activation marker CD86, we interrogated the activation profile among myeloid populations. DNA-LNP immunization induced activation in all populations, particularly among total mDCs and both CD11b⁺ and CD103⁺ mDCs (Figures 2H and 2I). We observed lower overall CD86 expression on neutrophils, pDCs, and monocytes relative to mDCs; however, DNA-LNP elicited significantly more CD86 expression on these populations compared with the other vaccine platform. In the spleens of immunized animals, immunization with DNA-LNP, and to a lesser extent mRNA-LNP, led to downregulation in the frequency of total rDCs, CD11b⁺ rDCs, and CD8α⁺ rDCs (Figures S4A–S4C). However, the frequencies of mDCs in the spleen remained largely similar to naive, with a downregulation in total mDCs with protein immunization (Figure S4D), downregulation in CD11b⁺ mDCs with mRNA-LNP immunization (Figure S4E), and an upregulation in CD103⁺ mDCs with both mRNA-LNP and protein immunization (Figure S4F). This is in contrast to the DLN, with a striking increase in mDC populations after DNA-LNP immunization, indicating differential modulation of DC populations at proximal and distal sites. Similar to the DLN, we observed upregulation in neutrophils after DNA-LNP immunization (Figure S4G). However, in contrast to the DLN, we observed an upregulation in pDCs with mRNA-LNP immunization (Figure S4H). In terms of activation profile, DNA-LNP immunization induced an upregulation in CD86 expression among rDC and mDC populations, despite decreases in total frequency (Figure S4I). In contrast, mRNA-LNP immunization led to a slight increase in CD86 expression in neutrophils and, along with DNA-LNP immunization, in pDCs. Taken together, these data suggest that DNA-LNP induces robust innate immune activation at proximal and distal sites and upregulates the frequency of migratory dendritic cell (DC) subpopulations and neutrophils 24 h after immunization in the DLN.

Figure 2.

Early immune profiling of innate and adaptive populations reveals STING-dependent activation status after HA DNA-LNP immunization

(A) Schematic of immunization regimen.

(B–G) Frequency of total mDCs (B), CD11b⁺ mDCs (C), CD103⁺ mDCs (D), pDCs (E), neutrophils (F), and monocytes (G) in the iliac DLN.

(H) Representative histograms of CD86 expression on total mDC and mDC subpopulations.

(I) Quantification of CD86 expression.

(J) Schematic of immunization regimen.

(K–P) Frequency of total mDCs (K), CD11b⁺ mDCs (L), CD103⁺ mDCs (M), pDCs (N), neutrophils (O), and monocytes (P) in the iliac DLN.

(Q) Representative histograms of CD86 expression on total mDC and mDC subpopulations.

(R) Quantification of CD86 expression.

Dots represent individual animals; for (A–I), n = 9–10 animals per group; for (J–R), n = 3–6 animals per group. Data are representative of two independent experiments. Plots show geometric mean with geometric SD. Unpaired one-way ANOVA adjusted for multiple comparisons with Bonferroni corrections was used to compare groups. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

To determine whether innate immune activation was dependent on cGAS-STING signaling, wild-type or STING knockout (KO) mice were immunized with HA DNA-LNP and myeloid populations were interrogated in iliac DLN 24 h later (Figure 2J). There was striking modulation of myeloid populations within the DLN in STING KO mice, with decreases in mDC populations and neutrophils and increases in pDCs and monocytes (Figures 2K–2P). In terms of activation status, mDC populations and neutrophils had significantly decreased CD86 expression (Figures 2Q and 2R). In contrast to the changes observed with DNA-LNPs, STING KO mice immunized with HA mRNA-LNP exhibited no significant difference in the frequency of induced myeloid populations (Figures S5A–S5F) or their activation profile (Figure S5G). Thus, these data demonstrate that sensing of DNA-LNPs in vivo is dependent on the cGAS-STING pathway as illustrated with the modulation in frequency and decrease in activation status of myeloid populations.

Single-cell transcriptomics highlights platform-specific modulation of innate immune populations

To further elucidate the mechanisms of innate immune modulation after vaccination with DNA-LNP, we performed single-cell RNA sequencing (scRNA-seq) on sorted innate immune cells from the DLN 24 h after immunization as in Figure 2A. We identified 5 clusters each corresponding to a different innate immune subset—NK cells, neutrophils, monocytes, pDCs, and conventional DCs (cDCs)—and utilized uniform manifold approximation and projection (UMAP) for visualization (Figure 3A). Sample-wise distribution demonstrates platform-specific modulation of populations, particularly in neutrophil, monocyte, and pDC clusters (Figure 3B). To understand the activation signature after immunization with DNA-LNP, we examined the top differentially expressed genes within each cluster that were up- or downregulated in DNA-LNP relative to naive and the other vaccination platforms. Among NK cells, we observed downregulation of Ccl5 and upregulation of Runx3; Runx3 is important for NK cell activation and improved function within the tumor (Figure 3C).^44,45 Additionally, we observed Traf1 upregulation in NK cells after DNA-LNP immunization, a gene downstream of tumor necrosis factor (TNF) superfamily receptors induced by genes such as TNF-α.⁴⁶ In neutrophils, we observed upregulation of migration-associated genes including chemokine Ccl4 and atypical chemokine receptor Ccrl2 (Figure 3D).^47,48 Additionally, we observed downregulation of Cd244a in neutrophils from DNA-LNP immunized mice; CD244 expression is a functional identifier of myeloid-derived suppressor cells.⁴⁹ Among monocytes, we observed upregulation of numerous chemokine or chemokine receptor genes associated with immune cell migration, such as Ccl4, Ccl2, Ccr5, and Ccl12 (Figure 3E). In pDCs, we observed upregulation of pro-inflammatory chemokine Ccl5 and alarmin S100a8, implicated in chemotaxis and activation of innate immune subsets (Figure 3F). Among cDCs, we observed a strong activation signature with upregulation of Cd80 and Cd86, the latter of which mirrored the flow cytometry data (Figure 3G). We also observed upregulation of Ext1, a glucosyltransferase implicated in DC migration⁵⁰; these findings further reflect the robust upregulation in mDC populations observed by flow cytometry. We next examined genes that were differentially expressed among all active vaccination groups relative to the naive control. A pro-inflammatory signature consistent with IFN signaling among NK cells, neutrophils, monocytes, and pDCs in multiple gene families was observed (Figures S6A–S6D). Active vaccination also induced NK cells upregulation of pro-inflammatory cytokine genes including Ifng and Gzmb. Among cDCs, all vaccine groups led to an upregulation in Stat1 and Cd274 expression; CD274 (PD-L1) is a negative regulator of T cell activation through PD-1-PD-L1 interactions, acting as an immune checkpoint during priming (Figure S6E). Together, these data demonstrate that immunization with DNA-LNPs generates a pro-activation and pro-migration signature in numerous innate immune subsets, with all active vaccination modalities resulting in potent IFN-mediated inflammatory responses.

Figure 3.

Single-cell transcriptomics elucidates pro-activation and migration signature among innate immune subsets after priming with HA DNA-LNP

(A) UMAP plot of innate immune subsets.

(B) UMAP plot of clusters colored by sample.

(C–G) Differentially expressed genes upregulated after immunization with DNA-LNP relative to naive, mRNA-LNP, and protein in adjuvant immunization in NK cells (C), neutrophils (D), monocytes (E), pDCs (F), and cDCs (G). Data represent one independent experiment of 10 pooled mouse popliteal LNs per group.

HA DNA-LNP elicits robust T cell responses durable at low doses

Considering the activation status of innate immune subsets, particularly CD103⁺ mDCs, we next examined whether HA DNA-LNP induced effector T cell responses (Figure 4A). HA DNA-LNP demonstrated a >3-fold or 7-fold increase in IFNγ-secreting CD8⁺ T cell responses compared with mRNA-LNP and adjuvanted protein formulations, respectively (Figures 4B and 4C). We additionally observed significantly increased CD107a-expressing (Figure 4D) and TNF-α-secreting (Figure 4E) CD8⁺ T cell responses with HA DNA-LNP immunization compared to both other platforms. Differing from the CD8⁺ T cell response, HA DNA-LNP and mRNA-LNP elicited comparable IFNγ-secreting (Figure 4F) and TNF-α-secreting (Figure 4G) CD4⁺ T cell responses that were significantly increased over adjuvanted protein. However, all three immunization groups elicited similar IL-2-secreting CD4⁺ T cells (Figure 4H).

Figure 4.

HA DNA-LNP elicits potent antigen-specific CD8⁺ and CD4⁺ T cell responses

(A) Schematic of immunization regimen.

(B and C) Representative FACS plots (B) and frequency (C) of IFNγ⁺ effector CD8⁺ T cells.

(D) CD107a⁺ effector CD8⁺ T cells.

(E) TNF-α⁺ effector CD8⁺ T cells.

(F) IFNγ⁺ effector CD4⁺ T cells.

(G) TNF-α⁺ effector CD4⁺ T cells.

(H) IL-2⁺ effector CD4⁺ T cells.

(I–K) Frequency of effector CD8⁺ T cells expressing IFNγ (I), CD107a (J), or TNF-α (K) after dose de-escalation.

Dots represent individual animals; for (C–H), n = 10 animals per group; for (I–K), n = 5 animals per group. Data pooled from two (C–H) or one (I–K) independent experiment(s). Plots show geometric mean with geometric SD. Unpaired one-way ANOVA adjusted for multiple comparisons with Bonferroni corrections was used to compare groups. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Given the magnitude of CD8⁺ T cell responses elicited by HA DNA-LNP, we examined T cell induction at lower doses. Importantly, mice immunized with 0.2 μg of HA DNA-LNP exhibited similar IFNγ-secreting CD8⁺ T cell responses to 2 μg of nucleoside-modified mRNA-LNP, demonstrating a 10-fold dose-sparing effect (Figure 4I). Similar trends were observed with CD107a-expressing (Figure 4J) and TNF-α-secreting (Figure 4K) CD8⁺ T cell responses, where 0.2 μg of HA DNA-LNP and 2 μg of HA mRNA-LNP were functionally equivalent. HA DNA-LNP and HA mRNA-LNP elicited similar levels of IFNγ-secreting (Figure S7A), TNF-α-secreting (Figure S7B), and IL-2-secreting (Figure S7C) CD4⁺ T cells at varying doses. Taken together, these data show that HA DNA-LNP formulations elicit robust CD8⁺ T cell responses that are durable at low doses.

HA DNA-LNP elicits GC responses and functional serum antibodies

We next sought to examine the GC response as LNPs are known to generate potent GCs when delivered with nucleoside-modified mRNA as well as recombinant protein.^11,51 GC reactions after mRNA-LNP immunization has been correlated with serum antibody functionality.⁵¹ HA DNA-LNP elicited an attenuated frequency but similar absolute numbers of GC Tfh cells (Figures 5A–5C) and total GC B cells compared with HA mRNA-LNP (Figures 5D–5F). Immunization with HA DNA-LNP generated antigen-specific GC B cell responses, which were attenuated in frequency and absolute number relative to HA mRNA-LNPs (Figures 5G–5I).

Figure 5.

HA DNA-LNP induces robust GC and serum responses

Mice were immunized with HA DNA-LNP (2 μg), HA mRNA-LNP (2 μg), or adjuvanted HA protein (1 μg). GC responses were assessed in the DLNs 14 days post immunization and serum responses longitudinally.

(A) Representative FACS plots of activated Tfh cells.

(B and C) Bar plots show quantification of frequency (B) and numbers (C) of activated Tfh cells.

(D) Representative FACS plots of total GC B cells.

(E and F) Bar plots show quantification of frequency (E) and numbers (F) of total GC B cells.

(G) Representative FACS plots of CA09 HA-specific GC B cells.

(H and I) Bar plots show frequency (H) and numbers (I) of CA09 HA-specific GC B cells.

(J) Area under the curve (AUC) of total A/California/04/2009 HA-specific serum IgG ELISA data.

(K) Serum endpoint titers at week 8 to various H1N1 HAs.

(L) HAI titers at week 8 to A/California/07/2009 X-179A.

(M and N) AUC of serum binding antibodies to A/Guangdong-Maonan/SWL1536/2019 HA (M) and A/Victoria/4897/2022 HA (N).

(O and P) HAI titers to A/Netherlands/602/2009 (O) and A/New York City/PV63249/2022 (P).

Dots represent individual animals (B, C, E, F, H, I, K, and L); n = 9–10 animals per group; data pooled from two independent experiments. Plots show geometric mean with geometric SD. Unpaired one-way ANOVA adjusted for multiple comparisons with Bonferroni corrections was used to compare groups (A–I) or active immunization groups (K). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

We examined the induction of influenza HA-specific serum antibody responses after DNA-LNP immunization. Relative to HA mRNA-LNP and adjuvanted protein, HA DNA-LNPs elicited comparable serum immunoglobulin G (IgG) responses to matched A/California/04/2009 HA over time, with durable titers persisting through at least 24 weeks (Figure 5J). We studied the breadth of binding antibody responses to related pandemic-lineage H1N1 influenza viruses at 8 weeks after immunization. We obtained comparable HA-specific binding antibody titers to H1N1 A/Michigan/45/2015, A/Wisconsin/588/2019, A/Sydney/5/2021, and A/Victoria/4897/2022 (Figure 5K) with all 3 vaccine platforms. In terms of functionality, all three active vaccination groups elicited robust and equivalent hemagglutinin inhibition (HAI) titers to A/California/07/2009 X-179A (Figure 5L). To further probe the functional breadth of serum antibody responses as the immune response matured, we measured binding and functional antibodies at week 6, week 14, and week 24. All treatment groups elicited binding antibodies to A/Guangdong-Maonan/SWL1536/2019 (Figure 5M) and A/Victoria/4897/2022 (Figure 5N). Additionally, we observed strong HAI titers over time to near-matched virus A/Netherlands/602/2009 (Figure 5O) and recent clinical isolate A/New York City/PV63249/2022 (Figure 5P), although titers to the latter were attenuated relative to older isolates. Importantly, both binding antibody titers and HAI titers were stable across 24 weeks. These data demonstrate that, despite an attenuated HA-specific GC B cell response in the DNA-LNP group, serum antibody responses were similar in terms of titer, functionality, and breadth in all active vaccination groups.

HA DNA-LNP induces durable memory responses in mice and rabbits

We next sought to assess whether priming with HA DNA-LNP induced memory T and B cell responses (Figure 6A). HA DNA-LNP induced approximately 4-fold greater IFNγ-secreting cell responses by ELISpot relative to HA mRNA-LNP or adjuvanted protein (Figure 6B) beyond one year after immunization. Concordantly, we observed similarly robust effector memory CD8⁺ T cell responses secreting IFNγ, expressing CD107a, or secreting TNF-α among HA DNA-LNP immunized mice using flow cytometry (Figure 6C; Figures S8A and S8B).

Figure 6.

HA DNA-LNP induces potent memory responses in mice and rabbits

(A) Schematic of mouse immunization regimen.

(B) IFNγ-secreting cells in splenocytes by ELISpot.

(C) IFNγ-secreting effector CD8⁺ T cells by flow cytometry.

(D) CA09 HA-specific ASC responses in bone marrow by ELISpot.

(E) Representative FACS plot of CA09 HA-specific MBCs.

(F and G) Bar plots show frequency (F) and numbers (G) of CA09 HA-specific MBCs.

(H) Schematic of rabbit immunization regimen.

(I–K) IFNγ ELISpot on peripheral blood mononuclear cells (PBMCs) at day 42 (I), day 105 (J), and day 202 (K).

(L) AUC of total A/California/04/2009 HA-specific serum IgG ELISA data.

(M and N) HAI titers to A/Netherlands/602/2009 (M) and A/New York City/PV63249/2022 (N).

Dots represent individual animals (C, D, F, and G); n = 9–10 animals per group (B–D, F, and G), n = 5 animals per group (I–N); data pooled from two independent experiments. Plots show mean with SD (B and I–K) or geometric mean with geometric SD (C, D, F, G, and L–N). Unpaired one-way or two-way ANOVA adjusted for multiple comparisons with Bonferroni corrections was used to compare groups. ANOVA was performed at the final time point for (L–N). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Antibody-secreting plasma cells (ASCs) and memory B cells (MBCs) are key mediators of immune memory, derived from GC reactions. Bone marrow-resident plasma cells secrete antibody into serum, while MBCs have the capability re-seed GCs or rapidly differentiate into plasma cells during secondary responses.^52,53 Importantly, immunization with HA DNA-LNP induced comparable bone marrow ASC responses to HA mRNA-LNP and adjuvanted protein (Figure 6D), mirroring serum responses over time. Among antigen-specific MBCs, we observed comparable HA-specific MBC responses in terms of frequency and absolute number after immunization with HA DNA-LNP or HA mRNA-LNP, which was approximately 2-fold greater than what was observed with adjuvanted protein (Figures 6E–6G). Taken together, these data demonstrate that immunization with DNA-LNPs induces strong adaptive immune populations that persist beyond one year after immunization.

Translation to larger animal models is crucial for vaccine development. We assessed the immunogenicity of HA DNA-LNPs in New Zealand white rabbits relative to HA mRNA-LNP or a naked HA DNA control (Figure 6H). We initially examined HA-specific T cell responses in peripheral blood by ELISpot. At Day 42 (14-days after the boost), we observed strong IFNγ responses by ELISpot after immunization with HA DNA-LNP, approximately 2-fold stronger than HA mRNA-LNP (Figure 6I). Although responses wane over time, HA DNA-LNP maintains responses in peripheral blood through 202 days (Figures 6J and 6K). Humoral immune responses were robust with both HA DNA-LNP and mRNA-LNP; interestingly, CA09 HA-specific IgG responses measured by ELISA exhibited some waning with HA mRNA-LNP but not HA DNA-LNP through 202 days (Figure 6L). However, in both groups, we observed robust HAI titers to matched and heterologous virus (Figures 6M and 6N). These data show that HA DNA-LNP is potent in rabbits with superior or comparable immunogenicity to mRNA-LNPs at the same dose.

SARS-CoV-2 spike-expressing DNA-LNP elicits serum antibody titers and is protective in challenge

To extend our findings to an additional antigen of interest and study the protective efficacy elicited from DNA-LNPs, we utilized a SARS-CoV-2 model. Mice were immunized with DNA- and mRNA-LNP vaccines expressing full-length SARS-CoV-2 spike^54,55,56 (Figure 7A). Spike DNA-LNP induced robust antigen-specific total IgG responses similar to spike mRNA-LNP against matched spike receptor binding domain (RBD) (Figure 7B); RBD binding is a target of neutralizing antibody responses.^57,58 Additionally, spike DNA-LNP elicited antigen-specific total IgG responses that were competitive to spike mRNA-LNP against near wild-type or against variant-of-concern spike trimers, including D614G (Figure 7C), Delta (Figure 7D), and BA.2 (Figure 7E). These data show that spike DNA-LNP can elicit comparable antibody titers to spike mRNA-LNP against matched and distant SARS-CoV-2 spike antigens. We next assessed neutralization titers against wild-type SARS-CoV-2 pseudovirus. Spike DNA-LNP induced a competitive neutralization titer compared with spike mRNA-LNP against wild-type SARS-CoV-2 pseudovirus (Figure 7F) at the equivalent dose.

Figure 7.

SARS-CoV-2 spike DNA-LNP elicits serum antibody responses and is protective in lethal challenge

(A) Schematic of immunization regimen and challenge.

(B–E) Serum endpoint titers at week 8 after immunization to wild-type spike RBD (B), D614G full-length spike (C), B.1.617.2 (Delta) full-length spike (D), and BA.2 full-length spike (E).

(F) Serum neutralization infective dose (ID)₅₀ against wild-type SARS-CoV-2 pseudovirus.

(G and H) Weight loss (G) and survival (H) of mice challenged with 1 × 10⁵ plaque-forming unit (PFU) mouse-adapted SARS-CoV-2 with an 80% weight loss cutoff.

(I) Clinical score representing clinical signs of morbidity 4 days after challenge.

Dots represent individual animals; n = 5–8 animals per group; data combined from two independent studies (A–F). Plots show geometric mean with geometric SD. Non-parametric Mann-Whitney U test (B–F) or log rank (Mantel-Cox) test (H) was used to compare groups. ∗∗p < 0.01.

To investigate the protective efficacy of spike DNA-LNP, mice were challenged with mouse-adapted SARS-CoV-2 (Figure 7A). We observed no weight loss with complete survival in mice immunized with either the spike DNA-LNP or mRNA-LNP, in contrast to empty vector control mice, which exhibited significant weight loss resulting in overall 80% mortality (Figures 7G and 7H). Empty vector-immunized mice also exhibited significant morbidity, where both spike DNA-LNP and mRNA-LNP immunized mice demonstrated no signs of outward morbidity (Figure 7I). Taken together, these data demonstrate that a single immunization with spike DNA-LNP elicits protective immunity comparable to spike mRNA-LNP.

Discussion

In this study, we compare the immunological profile of plasmid DNA-LNP vaccines relative to mRNA-LNPs and adjuvanted protein. The LNP component of mRNA-LNPs has been reported to be highly pro-inflammatory and serves as an adjuvant for nucleoside-modified mRNA as well as protein vaccines.¹¹ This is due in part to the transient upregulation of numerous pro-inflammatory cytokines, such as IL-6, which can potentiate adaptive immune engagement.^12,59 The adjuvant effect of LNPs has been linked to the ionizable component.¹¹ LNP-formulated plasmid DNA, however, has previously not shown similar levels of in vivo immunogenicity, despite robust transfection efficiency.⁶⁰ While we were preparing our studies, there have been two recent reports of immunogenic formulations of spike-expressing plasmid DNA in ionizable LNPs^33,34; these studies utilized a prime-boost approach to follow protective efficacy in a SARS-CoV-2 model of viral challenge. These studies did not explore the mechanisms of immunity conferred by these DNA formulations.

We sought to alter aspects of the platform to understand its contribution to immune responses. An increased N/P ratio has recently been associated with increased in vitro gene expression for RNA.⁶¹ Our study implicates the lipid-to-DNA ratio as a modulator of particle size and its immunogenicity in vivo. We examined three different lipid-to-DNA ratios for our formulations; this is not exhaustive in characterizing how immunogenicity may change with different N/P ratios beyond the range tested, specific changes to the molar amounts of lipid components within the LNP formulation, or other described ionizable lipids. Biophysical characterization and immunogenicity comparisons of this formulation relative to other DNA-LNP formulations are warranted. Additionally, we use one vector (pVAX1) in our studies, supporting further characterization of how immunogenicity may be modulated by the composition, length, or CpG content of alternate vectors; however, our numerous DNA-LNPs feature glycoproteins that vary in those attributes, supporting that the platform can accommodate variability.

We observed platform-specific modulation of innate immune populations in the DLN and spleen; immunization with DNA-LNPs drove activation of numerous innate immune subsets, particularly with mDCs. CD103⁺ mDCs are functionally similar to CD8α⁺ rDCs important for CD8⁺ T cell priming,⁴³ which support the robust antigen-specific CD8⁺ T cell responses observed with DNA-LNPs. Additionally, we observed a stark increase in the frequency of neutrophils after DNA-LNP immunization; neutrophils have numerous pattern recognition receptors and contribute to activating adaptive immune responses.⁶² Future studies to elucidate the specific role of individual innate immune populations in priming adaptive immune subsets appear warranted. Additionally, the role for nucleic acid sensing has been documented for mRNA-LNP vaccines.^11,12 We observed STING-dependent changes in innate immune populations, with KO mice demonstrating a loss in specific innate immune populations and activation in the DLN relative to wild type. Additional studies to further characterize DNA-LNP-induced signaling through TLR9, STING, and other DNA sensors could be of value. The role of each innate immune subset in the migration and activation of mDCs remains unclear. Additionally, the degree to which inflammation from innate immune cells impacts adaptive immunity and how modulation of this inflammation could be utilized to further improve immune responses warrant further study.

In the context of cytotoxic T cells, nucleic acid vaccines, both DNA as well as mRNA, can generate CD8⁺ T cell responses superior to that of protein in adjuvant.⁶³ However, it has been recently described that the incorporation of N1-methylpseudouridines in IVT mRNA transcripts, such as those in the licensed SARS-CoV-2 mRNA vaccines, can cause ribosomal frameshifting and induces off-target T cell responses to +1 frameshifted protein products in humans.^64,65 We observed superior antigen-specific CD8⁺ T cell responses with DNA-LNP immunization; studying the role of CD8⁺ T cell responses induced by DNA-LNP vaccines in the context of other infectious disease antigens as well as for anti-tumor immunity appears warranted. Specific applications that would benefit from the robust cytotoxic T lymphocyte (CTL)-associated profile of DNA-LNPs could include cancer immunotherapy or neoantigen vaccines. Vaccine applications in high-risk populations such as the elderly, where immune responses are typically attenuated, could also benefit from the overall potency of DNA-LNPs. Immune responses were maintained beyond one year after immunization, indicating that cells are entering the memory phase. Additionally, the performance of the DNA-LNP platform was comparable or superior to that of mRNA-LNPs in rabbits, supporting further scaling up of this technology in larger animals to enable possible translational studies.

We observed that, despite HA DNA-LNP eliciting attenuated GC effector cell frequencies compared with mRNA-LNP, humoral responses were comparable in antibody titer, breadth, and functionality compared with benchmark platforms. This observation represents a decoupling of the GC response with functional antibodies, as these have been demonstrated to be correlated with nucleoside-modified mRNA-LNPs.⁵¹ This could be due to plasmid DNA having longer protein production kinetics as compared with mRNA,^66,67 allowing for longer antigen exposure. We observed extended protein production kinetics with Luc DNA-LNP using in vivo imaging (IVIS), but further characterization of antigen expression kinetics as well as cell type and tissue-specific distribution of expression is warranted.

LNP formulation of plasmid DNA as studied demonstrated induction of robust antigen-specific humoral immunity while maintaining a strong CTL bias. These findings are relevant for DNA vaccine platform development providing an additional delivery method that can further expand potential product applications. Manipulating innate immune sensing utilizing cGAS-STING-TBK1 pathway inhibitors can potentially serve to modulate or even silence immunogenicity, facilitating development of DNA-LNPs for therapeutic applications. The identification of improved formulations for DNA delivery represent important tools for the generation of immunity against infectious agents and targeting pathogenic cells.

Limitations of the study

Although our studies utilized varied N/P ratios, additional studies to characterize other N/P ratios and phosphate groups on plasmid DNA available for lipid binding remain important. Our studies featured immunological analyses in mice and rabbits; utilizing larger animal models to understand dose escalation kinetics and tolerability will be crucial for further translational studies. We characterized the immune response in primary and secondary lymphoid organs; further study to characterize responses at other immune sites or at mucosal surfaces, such as the lungs, is warranted. Selection of adjuvants for comparisons to recombinant protein impacts the resulting immune response; comparing DNA-LNPs to other protein in adjuvant formulations remains of interest. In the DNA space, gene-encoded adjuvants have enabled potency advancement as well as tailoring of responses. We did not explore the potential benefit of gene adjuvants in the DNA-LNP platform. Additionally, DNA vaccines have traditionally been delivered using physical delivery methods; we did not explore how DNA-LNPs could complement existing delivery modalities.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, David B. Weiner (dweiner@wistar.org).

Materials availability

All unique/stable reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability

scRNA-seq data were deposited in the NCBI Gene Expression Omnibus under GEO: GSE271745. Further inquiries can be directed to the corresponding authors. This paper does not report original code. Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

Acknowledgments

We would like to thank the Animal Facility staff, the Flow Cytometry Facility, and the Genomics Facility at the Wistar Institute. We would like to thank Peter Hewins and Kellie Jurado from the University of Pennsylvania for assistance with the SARS-CoV-2 challenge. D.B.W. is supported by NIH P01AI165066, NIH/NIAID Collaborative Influenza Vaccine Innovation Centers (CIVIC) contract 75N93019C00051, and INOVIO Pharmaceuticals SRA 21-05. Additional funding to D.B.W provided by the W.W. Smith Charitable Trust Distinguished Professorship in Cancer Research and The Jill and Mark Fishman Foundation. N.P. is supported by NIAID R01AI146101 and R01AI153064. Work in the Krammer laboratory was supported by NIH/NIAID CIVIC contract 75N93019C00051. We thank the Mount Sinai Pathogen Surveillance Program (co-directed by Drs. van Bakel, Sordillo, and Simon) for providing rapid access to clinical influenza virus isolates. Graphical schematics were made in BioRender.

Author contributions

N.J.T., N.P., and D.B.W. conceptualized the study and designed experiments. N.J.T., S.T., N.B., T.K., E.P., N.O., K.L., A.S., C.L., M.I.T., E.N.G., M.V., N.L., J.L., M.S.K., K.B., C.H., V.M., and I.M. performed experiments. S.O.-L., H.M., T.R.F.S., L.M.H., A.P., A.K., J.S.B., D.A., and F.K. provided crucial reagents or resources. N.J.T., N.P., and D.B.W. analyzed the data. All authors contributed to writing and revision of the manuscript.

Declaration of interests

D.B.W. has received grant funding; participates in industry collaborations; has received speaking honoraria; and has received fees for consulting, including serving on scientific review committees. Remunerations received by D.B.W. include direct payments and equity/options. D.B.W. also discloses the following associations with commercial partners: Geneos (consultant/advisory board), AstraZeneca (advisory board, speaker), INOVIO (board of directors, consultant), Sanofi (advisory board), BBI (advisory board), Pfizer (advisory Board), and Advaccine (consultant). N.P. is named on patents describing the use of nucleoside-modified mRNA in LNPs as a vaccine platform. He has disclosed those interests fully to the University of Pennsylvania, and he has in place an approved plan for managing any potential conflicts arising from the licensing of these patents. N.P. served on the mRNA strategic advisory board of Sanofi Pasteur in 2022 and the advisory board of Pfizer in 2023 and 2024. N.P. is a member of the Scientific Advisory Board of AldexChem and BioNet-Asia. The Icahn School of Medicine at Mount Sinai has filed patent applications relating to SARS-CoV-2 serological assays, Newcastle disease virus (NDV)-based SARS-CoV-2 vaccines influenza virus vaccines, and influenza virus therapeutics, which list F.K. as a co-inventor. Mount Sinai has spun out a company, Kantaro, to market serological tests for SARS-CoV-2 and another company, CastleVax, to develop SARS-CoV-2 vaccines. F.K. is a co-founder and scientific advisory board member of CastleVax. F.K. has consulted for Merck, Curevac, GSK, Seqirus, and Pfizer and is currently consulting for 3rd Rock Ventures, Gritstone, and Avimex. The Krammer laboratory is collaborating with Dynavax on influenza vaccine development. N.J.T., D.B.W., and N.P. have filed a patent application related to aspects of this work.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
B220, Clone RA3-6B2, AF700	BioLegend	Cat# 103231, RRID:AB_493716
B220, Clone RA3-6B2, BUV805	BD Biosciences	Cat# 748867, RRID:AB_2873270
CD103, Clone 2E7, PE-Cy7	BioLegend	Cat# 121425, RRID:AB_2563690
CD107a, Clone 1D4B, FITC	BioLegend	Cat# 121606, RRID:AB_572007
CD11b, Clone M1/70, APC-eFluor780	Thermo Fisher Scientific	Cat# 47-0112-82, RRID:AB_1603193
CD11b, Clone M1/70, BV605	BioLegend	Cat# 101257, RRID:AB_2565431
CD11c, Clone N418, PerCP-Cy5.5	BioLegend	Cat# 117328, RRID:AB_2129641
CD16/32, Clone S17011E	BioLegend	Cat# 156604, RRID:AB_2783138
CD19, Clone 1D3, BUV395	BD Biosciences	Cat# 563557, RRID:AB_2722495
CD19, Clone 1D3, BV650	BioLegend	Cat# 152427, RRID:AB_3675002
CD38, Clone 90, BUV805	BD Biosciences	Cat# 741955, RRID:AB_2871263
CD3e, Clone 145-2C11, BUV395	BD Biosciences	Cat# 563565, RRID:AB_2738278)
CD4, Clone GK1.5, BV421	BioLegend	Cat# 100437, RRID:AB_10900241
CD4, Clone GK1.5, PE	BioLegend	Cat# 100408, RRID:AB_312693
CD4, Clone RM4-5, APC-eFluor780	Thermo Fisher Scientific	Cat# 47-0042-80, RRID:AB_1272219
CD4, Clone RM4-5, BUV496	BD Biosciences	Cat# 741050, RRID:AB_2870665
CD44, Clone 1M7, BV605	BioLegend	Cat# 103047, RRID:AB_2562451
CD45, Clone 30-F11, AF700	BioLegend	Cat# 103127, RRID:AB_493714)
CD45, Clone 30-F11, FITC	BioLegend	Cat# 103108, RRID:AB_312973
CD62L, Clone MEL-14, BUV805	BD Biosciences	Cat# 569201, RRID:AB_3668954
CD69, Clone H1.2F3, BV421	BioLegend	Cat# 104527, RRID:AB_10900250
CD8, Clone 53–6.7, APC-eFluor780	Thermo Fisher Scientific	Cat# 47-0081-82, RRID:AB_1272185
CD86, Clone GL-1, BV421	BioLegend	Cat# 105032, RRID:AB_2650895
CD8α, Clone 53–6.7, APC-Cy7	BioLegend	Cat# 100713, RRID:AB_312752)
CD8α, Clone 53–6.7, BUV563	BD Biosciences	Cat# 748535, RRID:AB_2872946
CXCR5, Clone SPRCL5, Biotin	Thermo Fisher Scientific	Cat# 13-7185-82, RRID:AB_2572800
Donkey anti-rabbit IgG, HRP-conjugated	Santa Cruz Biotechnology	Cat# sc-2313, RRID:AB_641181
F4/80, Clone BM8, APC-eFluor780	Thermo Fisher Scientific	Cat# 47-4801-82, RRID:AB_2735036
Fas, Clone Jo2, BV510	BD Biosciences	Cat# 563646, RRID:AB_2738345
GL7, Clone GL7, PE	BioLegend	Cat# 144608, RRID:AB_2562926
Goat anti-mouse IgG H + L, HRP-conjugated	Bethyl	Cat# A90-216P, RRID:AB_67184
Goat anti-mouse IgG H + L, HRP-conjugated	Rockland	Cat# 610–1302, RRID:AB_219656
Goat anti-mouse IgG, Biotin	SouthernBiotech	Cat# 1030-08, RRID:AB_2794296
I-A/I-E, Clone M5/114.15.2, AF647	BD Biosciences	Cat# 562367, RRID:AB_11152078
IFNγ, Clone XMG1.2, APC	BioLegend	Cat# 505809, RRID:AB_315403
IgD, Clone 11-26c.2a, PE-Cy7	BioLegend	Cat# 405720, RRID:AB_2561876
IgM, Clone II/41, PerCP-eFluor710	Thermo Fisher Scientific	Cat# 46-5790-82, RRID:AB_1834435
IL-2, Clone JES6-5H4, PE-Cy7	BioLegend	Cat# 503832, RRID:AB_2561750
Ly-6C, Clone HK1.4, BV785	BioLegend	Cat# 128041, RRID:AB_2565852
Ly-6G, Clone 1A8, BUV395	BD Biosciences	Cat# 563978, RRID:AB_2716852
NKp46, Clone 29A1.4, BV650	BioLegend	Cat# 137635, RRID:AB_2734200
PD-1, Clone 29F.1A12, PE-Cy7	BioLegend	Cat# 135216, RRID:AB_10689635
PDCA-1, Clone 927, PE	BioLegend	Cat# 127010, RRID:AB_1953285
Streptavidin, BV421	BioLegend	Cat# 405225
TCRβ, Clone H57-597, FITC	BioLegend	Cat# 109205, RRID:AB_313428
TCRβ, Clone H57-597, PE	BioLegend	Cat# 109207, RRID:AB_313430
TNFα, Clone MP6-XT22, PE	BioLegend	Cat# 506306, RRID:AB_315427
Bacterial and virus strains
Influenza H1N1 A/Netherlands/602/2009	Florian Krammer	N/A
Influenza H1N1 A/California/07/2009 X-179A	International Reagent Resource	Cat# FR-246
Influenza H1N1 A/New York City/PV63249/2022	Florian Krammer	N/A
SARS-CoV-2-MA30	Stanley Perlman	N/A
Chemicals, peptides, and recombinant proteins
18:0 PC (DSPC)	Avanti Polar Lipids	Cat# 850365
A/California/04/2009 Peptide Pools	Genscript	N/A
ACK Lysis Buffer	Quality Biologicals	Cat# 118-156-101
Addavax	Invivogen	Cat# vac-adx-10
Amlexanox (TBK1 inhibitor)	Invivogen	Cat# inh-amx
APC Lightning Link Kit	Abcam	Cat# ab201807
Britelite plus Reporter Gene Assay System	Revvity	Cat# 6066769
Cell Stimulation Cocktail	eBioscience	Cat# 00-4970-03
Cholesterol	Avanti Polar Lipids	Cat# 700000
D-luciferin	Regis Technologies Inc.	Cat# 103404-75-7
DMG-PEG-2000	Avanti Polar Lipids	Cat# 880151
FITC Lightning Link Kit	Abcam	Cat# ab102884
Fixable Viability Dye eFluor780	eBioscience	Cat# 65-0865-14
H-151 (STING inhibitor)	Invivogen	Cat# inh-h151
Influenza H1N1 A/California/04/2009 HA	Sino Biologicals	Cat# 11055-V08H
Influenza H1N1 A/California/04/2009 HA, B cell probe	Sino Biologicals	Cat# 11055-V08B1
Influenza H1N1 A/California/04/2009 HA HA0 (full-length)	Sino Biologicals	Cat# 11055-VNAB
Influenza H1N1 A/Guangdong-Maonan/SWL1536/2019 HA	Florian Krammer	N/A
Influenza H1N1 A/Michigan/45/2015 HA	Immune Technology	Cat# IT-003-00105ΔTMp
Influenza H1N1 A/Sydney/5/2021 HA	Immune Technology	Cat# IT-003-00119 ΔTMp
Influenza H1N1 A/Victoria/4897/2022 HA	Florian Krammer	N/A
Influenza H1N1 A/Victoria/4897/2022 HA	Immune Technology	Cat# IT-003-00120 ΔTMp
Influenza H1N1 A/Wisconsin/588/2019 HA	Immune Technology	Cat# IT-003-00117 ΔTMp
ODN2088 (TLR9 inhibitory CpG)	Invivogen	Cat# tlrl-2088
Protein Transport Inhibitor Cocktail	eBioscience	Cat# 00-4980-03
Receptor destroying enzyme (RDE)	Hardy Diagnostics	Cat# 370013
RU.521 (cGAS inhibitor)	Invivogen	Cat# inh-ru521-2
SARS-CoV-2 B.1.1.529 Spike	Sino Biologicals	Cat# 40589-V08H26
SARS-CoV-2 B.1.617.2 Spike	Sino Biologicals	Cat# 40589-V08H10
SARS-CoV-2 BA.2 Spike	Sino Biologicals	Cat# 40589-V08H28
SARS-CoV-2 D614G Spike	Sino Biologicals	Cat# 40589-V08H8
SARS-CoV-2 Wild-type Spike RBD	Sino Biologicals	Cat# 40592-V08H
SM-102	Echelon Bioscience	Cat#N-1102
Uranyl Formate Hydrate	Electron Microscopy Sciences	Cat# 22541
Whole turkey blood	Lampire	Cat# 7209403
Zombie Aqua Fixable Viability Kit	BioLegend	Cat# 423101
Critical commercial assays
Britelite plus Luminescence reporter gene assay system	Perkin Elmer	Cat# 6066769
Chromium Next GEM Single Cell 3ʹ Kit v3.1, 4 rxns	10x Genomics	Cat# PN-1000269
Luminex Mouse Cytokine/Chemokine Magnetic Bead Panel	Millipore Sigma	Cat# MCYTOMAG-70K-24C
Mouse IFNα ELISpot	Mabtech	Cat# 3326-2A
Mouse IFNγ ELISpot	Mabtech	Cat# 3321-4APT-2
Quant-iT PicoGreen	Thermo Fisher Scientific	Cat# P7589
Quant-iT RiboGreen	Thermo Fisher Scientific	Cat# R11490
Rabbit IFNγ ELISpot	Mabtech	Cat# 3110-4HPW-10
Deposited data
Single cell RNAseq	This manuscript	GEO Accession# GSE271745
Experimental models: Cell lines
huACE2-CHO cells	Creative Biolabs	Cat# VCeL-Wyb019
293T cells	ATCC	Cat# CRL-3216
Experimental models: Organisms/strains
BALB/c-Sting1em3Vnce/J (STING KO) mouse	The Jackson Laboratory	Strain# 036638
BALB/cAnNCrl mouse	Charles River Laboratories	Strain# 028
BALB/cJ mouse	The Jackson Laboratory	Strain# 000651
New Zealand White Rabbit	Charles River Laboratories	Strain# 052
Recombinant DNA
Firefly luciferase DNA plasmid	Genscript	N/A
Influenza H1N1 A/California/04/2009 Hemagglutinin DNA plasmid	Genscript	N/A
SARS-CoV-2 spike DNA plasmid	Genscript	N/A
Software and algorithms
Bio-Plex Manager Software 6.1	Bio-Rad	N/A
CTL Immunospot Analyzer	Immunospot	N/A
FACS DIVA	BD Biosciences	N/A
FlowJo V10.10.0	FlowJo LLC	N/A
GraphPad Prism 10	GraphPad	N/A
Living Image 4.7.4	Revvity	N/A
Mabtech IRIS	Mabtech	N/A
Seurat	Hao et al.68	https://satijalab.org/seurat/
SingleR	Aran et al.69	N/A
Other
CA09 HA mRNA-LNP	Pardi et al.70	N/A
Spike mRNA-LNP	Synthesized as in Laczkó et al.71	N/A
Cu300 EM Grids	Fisher Scientific	Cat# CF300-Cu
EasiGlow Glow Discharging System	Ted Pella Inc.	Cat# 91000S

Experimental models and study participant details

Mouse studies

All mouse studies were conducted under protocols approved by the Wistar Institute and the University of Pennsylvania Institutional Animal Care and Use Committees (IACUC). Six-to eight-week-old female BALB/cJ mice and BALB/c-Sting1em3Vnce/J (STING KO, Strain# 036638) were housed in the Wistar Institute Animal Facility (purchased from The Jackson Laboratory) or University of Pennsylvania Animal Facility (purchased from Charles River Laboratories). Animals were housed under specific pathogen free conditions.

Rabbit study

The work was conducted under an IACUC approved protocol in compliance with the Animal Welfare Act, PHS Policy, AALACi guidelines, USDA. Eight-month-old female White New Zealand rabbits were purchased from Charles River Laboratories. Rabbits were divided into three groups (n = 5) according to average group weight.

Cell lines

HEK 293T (obtained from ATCC) and huACE2-CHO (obtained from Creative Biolabs) cells were cultured in DMEM +10% fetal bovine serum (FBS) + 1% Penicillin/Streptomycin (10 U/mL penicillin and 10 μg/mL streptomycin) (D10). Cells were authenticated by the manufacturer, tested to be mycoplasma negative, and maintained at 37°C in 5% CO₂ conditions.

Method details

Preparation of DNA-LNPs and nucleoside-modified mRNA-LNPs

Codon optimized DNA plasmids were obtained from Genscript with the antigen in a pVAX1 vector under the control of an IgE leader sequence to facilitate secretion. The construct encoding the full-length SARS-CoV-2 spike glycoprotein (with the D614G mutation, which arose early during the COVID-19 pandemic) has been described previously.^54,55,56 Full length CA09 HA was cloned into a pVAX1 vector behind an IgE leader sequence and obtained from Genscript. mRNA vaccines encoding the full-length wild-type SARS-CoV-2 spike glycoprotein were synthesized as described in Laczkó et al.⁷¹ CA09 HA mRNA-LNPs were described previously.⁷⁰ To create LNP-formulated vaccines, plasmid DNAs and mRNAs were formulated using microfluidic mixing of organic phase containing lipids and aqueous phase containing plasmid DNA or mRNA. In brief, lipids were dissolved in ethanol at a molar ratio of 50:38.5:10:1.5 (SM102:Cholesterol:Distearoylphosphatidylcholine(DSPC):DMG-Polyethelene glycol (DMG-PEG)). Lipids utilized include SM102 (Echelon Bioscience, Cat#N-1102), cholesterol (Avanti polar lipids, Cat#700000), DMG-PEG 2000 (Avanti polar lipids, Cat#880151), and 18:0 PC (DSPC; Avanti polar lipids, Cat#850365).The plasmid DNA or mRNA were diluted in citrate buffer (citrate buffer 50 mM, pH-4) at a concentration of 129 μg/mL and mixed with lipid containing ethanol at a volumetric ratio of 1:3 (ethanol: citrate buffer) using microfluidic mixing device (NanoAssembler Ignite, Precision Nanosystems). Formulations were synthesized using an N/P ratio of ∼10.5 (total lipid to nucleic acid ratio 40:1 w/w) unless otherwise specified. N/P ratios above 6 for mRNA-LNPs do not exhibit particle size differences.⁷² The LNPs were then dialyzed against 1x PBS buffer in a 12–14 kDa dialysis membrane (Fisher Scientific Cat#08700158) for 2 h. The LNP formulations were characterized for their size, polydispersity index and zeta potential using Zetasizer Pro ZS (Malvern Panalytical). The encapsulation efficiency of DNA and mRNA in LNPs were analyzed using Quant-iT RiboGreen assay and Quant-iT PicoGreen assay (ThermoFisher Scientific).

Immunizations

In all studies, mice were primed once in the tibialis anterior muscle in a total volume of 50 μL diluted in PBS. Mice were immunized with indicated DNA-LNPs or mRNA-LNPs at doses indicated in the legends/text. For protein vaccination, 1 μg of recombinant full-length CA09 HA (Sino Biologicals Cat# 11055-VNAB) was formulated 1:1 v/v with Addavax Adjuvant (Cat# vax-adx-10, Invivogen). Mice were bled by submandibular bleed for antibody titers. Rabbits were dosed 50 μg of naked HA DNA, HA DNA-LNP, or HA mRNA-LNP intramuscularly to the left quadricep on days 0 and 28.

Tissue processing

Mice were euthanized using CO₂ at indicated timepoints. Popliteal and iliac draining lymph nodes were harvested into ice-cold RPMI 1640 + 10% FBS +1% penicillin/streptomycin (R10). Lymph nodes were mechanically dissociated over a 40 μm strainer, followed by washing with R10. Lymph nodes were counted using a ViCell Blu Cell Viability Analyzer (Beckman Coulter) and subjected to downstream staining for flow cytometry. Spleens were harvested similarly and mechanically dissociated using a Stomacher 80 (Seward). Splenocytes were filtered using a 40 μm strainer before Ammonium-Chloride-Potassium (ACK) lysis for 5 min at room temperature (RT). Cells were quenched by dilution with PBS and resuspended in fresh R10 before a second 40 μm filtration. Splenocytes were then counted using a ViCell Blu and subjected to downstream assays.

Mouse IFNγ ELISpot

To measure secreted IFNγ from splenocytes, mouse IFNγ-ELISpot plates (Mabtech) were used according to manufacturer’s protocol. Briefly, plates were washed in sterile PBS four times before blocking in complete R10 for 30 min at RT. Plates were seeded with 1x10⁵ cells in duplicate in the presence of 5 μg/mL of overlapping CA09 influenza virus HA protein peptide pools. Peptides were 15 amino acids in length with a 9 amino acid overlap. DMSO or Cell Stimulation Cocktail (eBioscience) were used as negative and positive controls respectively in lieu of peptides. Plates were incubated for 20 h at 37°C in 5% CO₂ before being developed according to manufacturer’s protocol. Plates were counted using a Mabtech IRIS (Mabtech) and values from DMSO treated wells were subtracted.

T and NK cell activation flow cytometry

Lymph node and spleen suspensions were resuspended in Fixable Viability Dye eF780 (eBioscience) in FACS buffer for 10 min at RT. Cells were next washed before resuspension with a surface stain cocktail for 30 min at RT containing the following antibodies: BUV563 anti-mouse CD8α (Clone 53–6.7, BD), BV421 anti-mouse CD69 (Clone H1.2F3, Biolegend), BV650 anti-mouse NKp46 (Clone 29A1.4, Biolegend), and PE anti-mouse CD4 (Clone GK1.5, Biolegend). Cells were washed in FACS and then fixed/permeabilized using BD CytoFast/CytoFix (BD) according to manufacturer’s protocol for 20 min at 4°C. Cells were washed in 1x Perm/Wash before stained with a cocktail of intracellular antibodies for 30 min at 4°C: BUV395 anti-mouse CD3e (Clone 145-2C11, BD) and APC anti-mouse IFNγ (Clone XMG1.2, Biolegend). Cells were resuspended in FACS buffer and acquired as above. Complete gating strategy shown in the Figure S9. IFNγ⁺ NK and CD69⁺ NK, CD4⁺ T, CD8⁺ T cells were defined as marker+ shown as a percent of parent population.

Myeloid population flow cytometry

Lymph node and spleen suspensions were resuspended in Fixable Viability Dye eF780 (eBioscience) and Fc block (anti-CD16/32, Clone S17011E, Biolegend) in FACS buffer for 10 min at RT. Cells were next washed before resuspension with a surface stain cocktail for 30 min at RT containing the following antibodies: BUV395 anti-mouse Ly-6G (Clone 1A8, BD), BUV563 anti-mouse CD8α (Clone 53–6.7, BD), BUV805 anti-mouse B220 (Clone RA3-6B2, BD), BV421 anti-mouse CD86 (Clone GL-1, Biolegend), BV605 anti-mouse CD11b (Clone M1/70, Biolegend), BV650 anti-mouse CD19 (Clone 1D3, Biolegend), BV785 anti-mouse Ly-6C (Clone HK1.4, Biolegend), FITC anti-mouse TCRβ (Clone H57-597, Biolegend), PerCP-Cy5.5 anti-mouse CD11c (Clone N418, Biolegend), PE anti-mouse PDCA-1 (Clone 927, Biolegend), PE-Cy7 anti-mouse CD103 (Clone 2E7, Biolegend), AF647 anti-mouse I-A/I-E (Clone M5/114.15.2, BD), and AF700 anti-mouse CD45 (Clone 30-F11, Biolegend). Cells were washed and resuspended in FACS buffer and acquired as above. Complete gating strategy shown in the Figure S10, with markers defined as in.^12,73,74 Populations are shown as a frequency of CD45⁺ cells.

Germinal center flow cytometry

Lymph node suspensions were resuspended in Fixable Viability Dye eF780 (eBioscience) in 0.2% bovine serum albumin (BSA) in PBS (FACS) for 10 min at RT. Cells were next washed before resuspension with biotin anti-mouse CXCR5 (Clone SPRCL5, eBioscience) for 30 min at RT in FACS. Cells were subsequently washed and resuspended with a surface stain cocktail for 30 min at RT containing a subset of the following antibodies or probes: BUV395 anti-mouse CD19 (Clone 1D3, BD), BUV496 anti-mouse CD4 (Clone RM4-5, BD), PE anti-mouse CD4 (Clone GK1.5, Biolegend), BUV805 anti-mouse CD38 (Clone 90, BD), BV421 streptavidin (Biolegend), BV510 anti-mouse Fas (Clone Jo2, BD), BV605 anti-mouse CD44 (Clone IM7, Biolegend), PE-Cy7 anti-mouse PD-1 (Clone 29F.1A12, Biolegend), FITC-conjugated CA09 HA (in house), or APC-conjugated CA09 HA (in house). Cells were washed and resuspended in FACS buffer before acquiring on a FACSymphony A3 or A5 SE analyzer (BD). FCS files were exported and analyzed using FlowJo (Treestar). Complete gating strategy shown in the Figure S11 Activated Tfh were defined as PD-1^hiCXCR5⁺ among CD44⁺ CD4⁺ T cells (shown as percent of CD44⁺).⁷⁵ GC B cells were defined as CD38⁻Fas⁺ among CD19⁺ B cells (shown as percent of CD19⁺).⁷⁶ Antigen-specific GC B cells were defined as HA++ among GC B cells (shown as a percent of CD38⁻Fas⁺).

T cell intracellular cytokine staining

Splenocytes were seeded in U bottom plates and incubated in the presence of CA09 HA overlapping peptide pools (as in ELISpot). Samples were co-incubated in the presence of Protein Transport Inhibitor Cocktail (eBioscience) and FITC anti-mouse CD107a (Clone 1D4B, Biolegend). DMSO and Cell Stimulation Cocktail were used as negative and positive controls respectively. After peptide stimulation, cells were first washed in PBS before incubation with Zombie Aqua viability dye (Biolegend) for 10 min at RT. Cells were washed in FACS and resuspended in a surface stain cocktail for 30 min at RT containing a subset of the following antibodies: BUV496 anti-mouse CD4 (Clone RM4-5, BD), BUV805 anti-mouse CD62L (Clone MEL-14, BD), BV421 anti-mouse CD4 (Clone GK1.5, Biolegend), BV605 anti-mouse CD44 (Clone IM7, Biolegend), and APC-Cy7 anti-mouse CD8a (Clone 53–6.7, Biolegend). Cells were washed in FACS and then fixed/permeabilized using BD CytoFast/CytoFix (BD) according to manufacturer’s protocol for 20 min at 4°C. Cells were washed in 1x Perm/Wash before stained with a cocktail of intracellular antibodies for 30 min at 4°C: BUV395 anti-mouse CD3e (Clone 145-2C11, BD), PE anti-mouse TNFα (Clone MP6-XT22, Biolegend), PE-Cy7 anti-mouse IL-2 (Clone JES6-5H4, Biolegend), and APC anti-mouse IFNγ (Clone XMG1.2, Biolegend). Cells were resuspended in FluoroFix buffer (Biolegend) and stored at 4°C until acquisition as above. Complete gating strategy shown in the Figure S11. Cytokine/marker expressing CD4⁺ and CD8⁺ T cell populations were defined as cytokine/marker+ shown as a percent of CD44⁺ CD62L⁻ effector cells.^77,78

Memory B cell flow cytometry

Spleen suspensions were resuspended in Zombie Aqua viability dye (Biolegend) in PBS for 10 min at RT. Cells were then washed in FACS buffer and resuspended in a surface stain cocktail for 30 min at RT containing the following antibodies: BUV395 anti-mouse CD19 (Clone 1D3, BD), BUV805 anti-mouse CD38 (Clone 90, BD), FITC-conjugated CA09 HA (in house), PerCP-eF710 anti-mouse IgM (Clone II/41, Invitrogen), PE anti-mouse GL7 (Clone GL7, Biolegend), PE-Cy7 anti-mouse IgD (Clone 11-26c.2a, Biolegend), APC-conjugated CA09 HA (in house), AF700 anti-mouse B220 (Clone RA3-6B2, BD), APC-eF780 anti-mouse CD4 (Clone RM4-5, Invitrogen), APC-eF780 anti-mouse CD8 (Clone 53–6.7, Invitrogen), APC-eF780 anti-mouse CD11b (Clone M1/70, Invitrogen), and APC-eF780 anti-mouse F4/80 (Clone BM8, Invitrogen). Cells were then washed and resuspended in FACS buffer before acquisition as above. Dump represents CD4, CD8, CD11b, and F4/80. Memory B cells are shown as a frequency of IgD⁻IgM^- B cells, with the complete gating strategy shown in Figure S12.

Antigen-specific B cell probes

B cell probes were generated via direct conjugation of fluorochromes to recombinant influenza virus CA09 HA (Cat# 11055-V08B1, Sino Biologicals). Lightning Link Kits for FITC (ab102884) or APC (ab201807) were used according to manufacturer’s protocol. Probes were stored at 4°C until use.

Transmission electron microscopy

A total of 3 μL of purified LNPs (∼0.2 mg/mL) was adsorbed onto carbon-coated Cu400 EM grids with a EasiGlow Glow Discharging System (Ted Pella) instrument for 30–45 s. The grids were blotted with Whatman filter paper and rinsed and blotted 3 times with 6 μL nuclease free water. The grids were then stained with 3 μL of 2% uranyl formate, blotted, and stained again with 3 μL of the stain for 90 s. Image collection was performed on a FEI Tecnai T12 microscope at 100kV and ∼1.6 μm defocus, equipped with Oneview Gatan camera at 110,000× camera magnification.

Rabbit sample processing

Blood samples (500 μL) for sera isolation were collected via ear vein on days 0, 28, 42, 105 and 202, and were centrifuged at 1500 g for 15 min at 4°C. Isolated sera were aliquoted and immediately frozen at −20°C until further use. Peripheral blood was drawn from the rabbit ear vein and transferred to EDTA blood collection tubes. Blood was diluted 1:1 with PBS (HyClone) in a new 15 mL conical tube. Diluted blood was slowly layered over 3.5 mL Ficoll-Paque Plus in a 15 mL SepMate tube (Stemcell Technologies). Tubes were centrifuged at 1200 g for 10 min at RT. PBMCs were washed followed by residual red blood cells lysis. Cells were counted and resuspended at 10⁶ cells/mL concentration in R10 medium (RPMI-1640 with 10% FBS, 1% penicillin/streptomycin, and 0.1% 2-mercapthoethanol) for use in the IFNγ ELISpot assay.

Rabbit PBMC IFNγ ELISpot

To quantify IFNγ cellular responses, a Rabbit IFNγ ELISpot Plus kit (Mabtech, Cat# 3110-4HPW-10) was used. Briefly, pre-coated ELISpot plates were blocked for 30 min with R10 medium. Peptide pools spanning the globular head domain (pools 1 and 2), and stem domain (pool 3), as well as positive control (PMA/Ionomycin) and negative control (DMSO) were added to the plates, followed by addition of PBMCs (1x10⁵ cells/well) at a 1:1 ratio stimuli to cells. Plates were incubated in a 37°C humidified incubator with 5% CO₂ for 18 h. Cells were removed and plates were washed five times with 200 μL/well of PBS. Detection antibody (MT318-biotin) was diluted to 0.1 μg/mL in PBS containing 0.5% FBS. 100 μL/well was added and plates were incubated for 2 h at RT. Plates were washed and streptavidin-HRP was diluted (1:1000) in PBS-0.5% FBS and 100 μL added to each well. Plates were incubated for 1 h at RT, washed and 100 μL/well of TMB (Mabtech) substrate solution was added. The reaction was stopped after 10 min by washing plates with DI water and left to dry overnight. Spots were scanned and counted the following day using a CTL ImmunoSpot analyzer.

Ex vivo DNA sensing inhibition

To interrogate relevant DNA sensing pathways, ELISpot plates were seeded with 2.5x10⁵ splenocytes and incubated in the presence of plasmid DNA (1 μg/mL), DNA-LNP (1 μg/mL), or co-incubated with one of the following inhibitors in the presence of DNA-LNP: cGAS inhibitor 40uM (RU.521; Cat# inh-ru521-2, Invivogen), STING inhibitor 10 μM (H-151; Cat# inh-h151, Invivogen), TBK1 inhibitor 100 μM (Amlexanox; Cat# inh-amx, Invivogen), TLR9 inhibitor 1 μg/mL (ODN2088; Cat# tlrl-2088, Invivogen). Cells and stimuli were incubated for 20 h and either mouse IFNγ pre-coated ELISpot plates (Mabtech) or mouse IFNα (Mabtech 3326-2A) developed according to manufacturer’s protocol.

ELISA

For HA binding ELISAs in Figure 5J-K, 96-well half area plates (Corning, Ref# 3690) were coated in PBS overnight at 4°C with 1 μg/mL of one of the following recombinant proteins where specified: H1N1 A/California/04/2009 HA protein (Cat# 11055-V08H, Sino Biologicals), A/Michigan/45/2015 HA (Cat# IT-003-00105ΔTMp, Immune Technology), A/Wisconsin/588/2019 HA (Cat# IT-003-00117ΔTMp, Immune Technology), A/Sydney/5/2021 HA (Cat# IT-003-00119ΔTMp, Immune Technology), or A/Victoria/4897/2022 HA (Cat# IT-003-00120ΔTMp). The following day, plates were washed 4 times with PBS containing 0.05% Tween 20 (PBS-T) before being blocked with 1x PBS containing 5% non-fat dry milk (LabScientific) and 0.2% Tween 20 (Fisher) for 1 h at RT. After washing, plates were subsequently incubated with mouse sera serially diluted in 1% newborn calf serum and 0.2% Tween 20 in 1x DPBS for 2 h at RT before washing and incubation with 1:10000 horseradish peroxidase (HRP)-conjugated Goat anti-mouse IgG H + L (Cat# A90-216P, Bethyl) for 1 h at RT. Following this, plates were washed a final time and developed with 1-Step Ultra 3,3′, 5,5′ tetramethylbenzidine dihydrochloride (TMB)-ELISA Substrate Solution (Thermo, ref. 34029) for 5 min at RT before being stopped with 2N H2SO4. Plates were read with the BioTEK Synergy 2 plate reader and absorbance values measured at 450nm were subtracted with 570nm background optical density (OD) values. Endpoint titers were calculated against naive mouse serum. The endpoint titer was defined as the highest dilution where the OD value was greater than cutoff determined using the following formula: Average (Naive Mice) + (4∗ SD (Naive Mice)).

For HA binding ELISAs using rabbit serum in Figure 6L, the assay was performed as above with the exception of using HRP-conjugated donkey anti-rabbit IgG (Cat# sc-2313, Santa Cruz Biotechnology) as a secondary detection reagent.

For SARS-CoV-2 spike binding ELISAs in Figures 7B–7E, assays were performed as above, except for coating with the following proteins where indicated: SARS-CoV-2 wild type spike RBD (Cat# 40592-V08H, Sino Biologicals), SARS-CoV-2 D614G spike (Cat# 40589-V08H8, Sino Biologicals), SARS-CoV-2 B.1.617.2 spike (Cat# 40589-V08H10, Sino Biologicals), SARS-CoV-2 B.1.1.529 spike (Cat# 40589-V08H26), or SARS-CoV-2 BA.2 spike (Cat# 40589-V08H28, Sino Biologicals).

For HA binding ELISAs in Figure 5M–5N, recombinant influenza virus HA proteins were expressed in the baculovirus expression system as previously described.⁷⁹ Immulon 4HBXultra-high-binding 96-well plates (Thermo Fisher) were coated with 100 μL of 2 μg/mL recombinant HA diluted in phosphate-buffered saline (PBS; pH 7.4; Gibco) and stored overnight at 4°C. The next day, the plates were blocked with 220 μL of blocking solution consisting of PBS supplemented with 0.1% Tween 20 (PBS-T), 3% milk powder (American Bio) for 1 h at RT. Serum samples were 3-fold serially diluted starting at a concentration of 1:100 in PBS-T with 1% milk. Diluted serum samples (100 μL) were added to the plates and incubated at RT for 2 h. The plates were washed three times with PBS-T, and 100 μL of HRP-conjugated goat anti-mouse IgG (Rockland, cat no: 610–1302; 1:5,000) was added to each well. After 1 h, plates were washed four times with PBS-T. The plates were developed with SigmaFast o-phenylenediamine dihydrochloride (OPD; Sigma) for 10 min, and the reaction was stopped with 3 M HCl (Thermo Fisher). The plates were read at 490 nm with a microplate reader (BioTek). The data were analyzed in GraphPad Prism 10, and the area under the curve (AUC) were determined. The cutoff value was defined as the average of the values of blank wells plus 6 times the standard deviation of the blank wells.

Bone marrow ELISpot

BM was flushed from femurs and tibia from each mouse using a 23G X 3/4″ needle and syringe into FACS buffer and filtered through 63 μm Nitex mesh. Red blood cells were lysed in ACK lysis buffer for 5 min on ice. The resulting cells were counted using a Nexcelom Cellometer Auto 2000 Cell Viability Counter Profiler (Nexcelom Bioscience LLC). MultiScreenHTS IP Filter Plates (0.45 μm, Millipore Sigma), was coated with 1 μg/mL H1N1 A/California/04/2009 HA protein (Cat# 11055-V08H, Sino Biologicals) in sodium carbonate/sodium bicarbonate buffer (pH 9.6) (35 mM NaHCO₃ and 15 mM Na₂CO₃) for 1 h at 37°C. Plates were then washed with 200 μL PBS/well three times and blocked at 37°C in complete RPMI +10% FBS for 30 min. BM cells were plated in eight halving dilutions beginning at 1 million cells per well and incubated overnight in complete RPMI +10% FBS. Plates were then washed with wash buffer (1x PBS +0.1% Tween 20) five times and incubated with biotinylated polyclonal goat anti-mouse IgG detection antibody (SouthernBiotech) in PBS +2% BSA at RT for 1 h. Plates were once again washed five times, and streptavidin-alkaline phosphatase (1:20,000 dilution in PBS +2% BSA) was added prior to incubation at RT for 30 min. Plates were then washed five times with wash buffer, and 50 μL/well BCIP/NBT single solution (SigmaAldrich) was added for 5 min or until spots developed, at which time the reaction was quenched with 100 μL 1 M sodium phosphate monobasic solution per well. After plates were rinsed with dH2O and dried overnight, they were scanned using Mabtech IRIS and counted.

IVIS – Female BALB/c mice were injected with 3 μg of Firefly luciferase (Luc) DNA-LNP or mRNA-LNP or given PBS. During imaging, mice were put under 3% isoflurane-induced anesthesia and intraperitoneally injected with 100 μL of 30 mg/mL D-luciferin sodium salt (Regis Technologies Inc, 103404-75-7). Mice were then imaged using an IVIS Spectrum machine (PerkinElmer) belly up for whole body chemiluminescence imaging. 5–30 images were taken with 0.2 min delay and automatically determined exposure time, until the signal reached the peak intensity. The collected data was analyzed using Living Image 4.7.4 software.

Luminex

Draining popliteal and iliac lymph nodes were collected at 4- and 24-h post-administration and snap-frozen on dry ice and stored at −80°C until use. Frozen tissues were disrupted in 750 μL lysis buffer (M-PER mammalian protein extraction reagent, Thermo Fisher with protease inhibitor tablets, cOmplete) using TissueLyser II (Qiagen). Tissues were disrupted using 5 mm steel beads (QIAGEN) under the following conditions: 2 × 30 Hz, 30 s per cycle. Homogenized tissues were incubated on ice for 30 min with inversions every 10 min. Tissue lysates were cleared by centrifugation for 1 min at 16,000 g. Lysates were transferred to new tubes and assayed for the induction of 24 selected cytokines (GM-CSF, MCP-1, MIP2, IP-10, RANTES, Eotaxin, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12p70, IL-13, IL-17A, IFN-γ, TNF-α, IL-1α, MIP-1β, KC, LIF, LIX, VEGF) using the Luminex Mouse Cytokine/Chemokine Magnetic Bead Panel (Cat# MCYTOMAG-70K-24C, Millipore Sigma) according to the manufacturer’s instructions. Samples were thawed on ice, cleared by centrifugation (10,000 g, 10 min, 4°C), diluted 1:4 in assay buffer, and a volume of 25 μL transferred to assay. Beads were read using the FLEXMAP 3D instrument and Luminex xPONENT 4.2; Bio-Plex Manager Software 6.1. The total protein concentration of the supernatants was determined by bicinchoninic acid (BCA) Protein Assay Kit (Cat #23225, Thermo Scientific) using BSA as a standard. The levels of cytokines/chemokines in each sample were normalized by the total protein level of the supernatants.

Hemagglutinin inhibition (HAI) assay

For HAI in Figure 5L, sera were first treated with receptor destroying enzyme (RDE, Hardy Diagnostics) at a 1:3 ratio overnight at 37°C. RDE-treated sera were then heat inactivated at 56°C for 45 min, followed by preadsorption with 10% turkey red blood cells (RBCs, Lampire Biologicals) for 1 h at 4°C with gentle agitation. Samples were applied to V-bottom 96-well plates at starting dilution neat and 2-fold serial dilution in 0.85% saline. Samples were then incubated with 4 hemagglutinating units of virus (A/California/07/2009 X-179A) and 0.5% turkey RBCs in saline at RT for 30 min. HAI titers were scored using drip method.

For HAI in Figure 5O-P and Figure 6M–6N, the assay was performed as previously described.⁸⁰ Briefly, receptor-destroying enzyme (RDE)-treated serum samples were 2-fold serially diluted starting at a dilution of 1:10 to 1:1280. Twenty-five-microliters of diluted serum samples were then incubated with 4 HA units of virus in 25 μL PBS. After 30 min of incubation at RT, 50 μL of 0.5% Turkey red blood cells were added to the virus-serum mixture and incubated at 4°C for 30 min after which the HAI activity was read.

Pseudovirus neutralization assay

Pseudotyped wild-type SARS-CoV-2 was produced in HEK293T cells via transient transfection using GeneJammer (Agilent) at a 1:1 ratio of IgE-SARS-CoV-2 spike plasmid (Genscript) and pNL4-3.Luc.R-E plasmid (NID AIDS reagent). After 48 h, transfection supernatant was harvested and enriched with FBS to 12% final volume and stored at −80°C until use. For the SARS-CoV-2 pseudovirus neutralization assay, huACE2-CHO cells (Cat# VCeL-Wyb019; Creative Biolabs) were seeded into 96-well plates in D10 overnight at 37°C. The next day, mouse serum was heat inactivated, serially diluted, and incubated with SARS-CoV-2 pseudovirus for 90 min at RT, and subsequently incubated with plated huCHOAce2 cells for 72 h. Cells were then lysed using the Britelite plus Luminescence reporter gene assay system (Cat# 6066769, PerkinElmer) and relative luciferase units (RLU) were read using a BioTek synergy plate reader. Pseudovirus neutralization titers (ID₅₀) were reported as the reciprocal serum titer at which RLUs were 50% reduced compared to RLUs in virus control well after background subtraction.

Mouse-adapted SARS-CoV-2 challenge

Challenge was performed under appropriate animal use protocols in the animal BSL3 facility at the University of Pennsylvania. BALB/c mice were infected approximately 9-week post immunization with 1 × 10⁵ PFU mouse-adapted SARS-CoV-2 intranasally under ketamine/xylazine sedation. Animals were then weighed and monitored daily for clinical signs of disease. Animals were monitored through 10 days post challenge. Weight loss cutoff for survival reported in Figure 7 is 80% of Day 0 body weight.

Cell sorting and single cell transcriptomics

Popliteal lymph nodes were processed as described above. Cell suspensions were lysed with ACK lysis buffer for 5 min at RT before quenching by dilution with PBS and resuspended in fresh R10 before a second 40 μm filtration. Cells were subsequently resuspended in Fixable Viability Dye eF780 (eBioscience) in FACS buffer for 10 min at RT. Cells were next washed before resuspension with a surface stain cocktail for 30 min at RT containing the following antibodies: BUV395 anti-mouse CD19 (Clone 1D3, BD), FITC anti-mouse CD45 (Clone 30-F11, Biolegend), and PE anti-mouse TCRβ (Clone H57-597, Biolegend). Live CD45⁺ CD19⁻ TCRβ⁻ cells were then sorted using a FACSymphony S6 sorter into 1% BSA before downstream processing. Cells were barcoded and processed using the 10x Genomics Chromium Single Cell Kits (10x Genomics) according to manufacturer’s protocol and sequenced using a NextSeq 500 (Illumina) as in.⁸¹ CellRanger suite was used for pre-processing of the scRNA-seq data (pipeline v7.0.0, https://support.10xgenomics.com) with refdata-gex-mm10-2020-A transcriptome as a reference to map reads on the mouse genome (mm10) using STAR.⁸² Cells with over 5% mitochondrial content and cells with less than 200 genes with reads were filtered out as they are likely cells with low quality and/or cells that are likely dying. The remaining 13534 cells were used for downstream analysis. Batch effect was not observed and hence not corrected for. Seurat v4⁶⁸ was used for cell clustering, marker identification, and visualization. Cells that were Tcrg+ or exhibited a T cell signature were discarded. The R package SingleR⁶⁹ was used to determine initial cell types of the clusters using the ImmGen dataset as a reference for cell-specific gene signatures and then verified using known cell-type markers unique to clusters. Differential expression between samples in specific clusters was performed using Wilcoxon Rank-Sum Test.

Quantification and statistical analysis

All statistical tests were performed using GraphPad Prism 10. Graphs and error bars represent means ± SD or geometric means ± geometric SD. Area under the curve (AUC) analysis was calculated using a baseline of y = 0.05 in Figure 5J. Non-parametric Mann-Whitney U test, unpaired one-way or two-way ANOVA with Bonferroni corrections were used to compare groups. Log rank test used for survival curve. The number of samples in each graph is notated in the figure legend. Individual mice may be excluded in Figure 5J–5P in the event serum volume was limiting at individual timepoints due to the number of assays performed. In all datasets, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Published: March 21, 2025