Tuning DNA Dissociation from Spherical Nucleic Acids for Enhanced Immunostimulation

Abstract

The stability of the core can significantly impact the therapeutic effectiveness of liposome-based drugs. While the spherical nucleic acid (SNA) architecture has elevated liposomal stability to increase therapeutic efficacy, the chemistry used to anchor the DNA to the liposome core is an underexplored design parameter with potentially widespread biological impact. Herein, we explore the impact of SNA anchoring chemistry on immunotherapeutic function by systematically studying the importance of hydrophobic dodecane anchoring groups in attaching DNA strands to the liposome core. By deliberately modulating the size of the oligomer that defines the anchor, a library of structures has been established. These structures, combined with in vitro and in vivo immune stimulation analyses, elucidate the relationships between and importance of anchoring strength and dissociation of DNA from the SNA shell on its biological properties. Importantly, the most stable dodecane anchor, (C12)₉ is superior to the n=4-8 and 10 structures and quadruples immune stimulation compared to conventional cholesterol-anchored SNAs. When OVA1 peptide antigen is encapsulated by the (C12)₉ SNA and used as a therapeutic vaccine in an E.G7-OVA tumor model, 50% of the mice survived the initial tumor, and all of those survived tumor rechallenge. Importantly, the strong innate immune stimulation does not cause a cytokine storm compared to linear immunostimulatory DNA. Moreover, a (C12)₉ SNA that encapsulates a peptide targeting SARS-CoV-2 generates a robust T cell response; T cells raised from SNA treatment kill >40% of target cells pulsed with the same peptide and ca. 45% of target cells expressing the entire Spike protein. This work highlights the importance of using anchor chemistry to elevate SNA stability to achieve more potent and safer immunotherapeutics in the context of both cancer and infectious disease.

Graphical Abstract

Liposomes have been shown to be a versatile and effective platform for drug delivery due to their biocompatibility and ability to encapsulate and protect a variety of therapeutic cargo.^1-3 Densely modifying the surface of a liposome with nucleic acids to create a spherical nucleic acid⁴ (SNA) architecture alters the biodistribution⁵ and significantly improves the stability and cellular uptake of the cargo.⁶ SNAs not only offer a way to improve liposomal formulations and elevate the delivery of nucleic acid cargo, but they also have properties distinct from their linear nucleic acid components.^7-9 In contrast with linear nucleic acids, SNAs rapidly enter cells,⁷ resist nuclease degradation,⁸ and exhibit higher binding constants for their complements when compared to free strands of the same sequence.⁹ These SNA properties have led to the development of several nucleic acid-based nanomedicines that are under evaluation in multiple clinical trials.^10-12 Liposomal SNAs, in particular, are a promising platform to design vaccines by manipulating their structure at the nanoscale,^13-16 thereby enabling the investigation of structure-function relationships driving enhanced immune activity: an approach termed rational vaccinology.¹⁴ Through this approach, various features have been identified as being critical in elevating vaccine potency.^{17, 18} For example, the stability of these structures has been shown to impact their pharmacokinetic profile,⁵ and high throughput methods have identified anchor chemistry as a vital parameter in the design of SNA vaccines.¹⁷ Other efforts to improve liposomal SNA stability have included altering the composition of the lipid bilayer in such a way that it raises its melting transition¹⁸ and modifying the lipophilic anchor which tethers the oligonucleotide to the liposome.¹⁹ However, the ability to further extend the stability of these constructs and elevate potency beyond these thresholds, has been limited by the solubility of lipophilic anchors and inefficient packing of the anchoring oligonucleotide into the lipid bilayer.

Herein, we report a method to extend the stability of liposomal SNAs by oligomerizing hydrophobic units that anchor immunostimulatory oligonucleotides into the core and subsequently explore how these changes in stability impact immunotherapeutic function. Specifically, we evaluated dodecane oligomers (C12, n=4-10) in the 3’-terminus of immunostimulatory cytosine–guanine “CpG” motif oligonucleotides as anchors for immunostimulatory SNAs. These anchors are readily incorporated during solid phase oligonucleotide synthesis, circumventing the requirement for aqueous solubility during anchor coupling. Importantly, through the addition of discrete C12 units, we tuned both the stability and potency of immunostimulatory liposomal SNAs, finding that the most serum stable (C12)₉ SNAs potentiated T cell immunity in vivo with minimal inflammatory side effects commonly found with linear oligonucleotides. (C12)₉ SNAs utilized as a cancer vaccine treatment limit the growth of E.G7-OVA lymphoma tumors and resulted in the survival of 50% of the mice. This is a 6-fold increase in survival compared to conventional cholesterol-anchored SNAs. Notably, when surviving mice were re-challenged with E.G7-OVA cells 70 days post initial tumor inoculation, no tumor growth was observed suggesting immunological protective memory against tumor rechallenge. When harnessed as a prophylactic vaccine, we observe that (C12)₉ SNAs strongly prime antigen-specific T cell responses against SARS-CoV-2. These (C12)₉ SNAs incorporate a conserved peptide from the SARS-CoV-2 spike protein, which demonstrates the generalizability of the construct to raise potent cellular immunity. This work highlights the potential of the liposomal SNA platform in the development of T cell-stimulating vaccine therapies.

RESULTS AND DISCUSSION

Tuning SNA Stability with C12 Oligomers

We hypothesized that tuning the oligomerization of dodecane (C12) units as anchors would modulate the stability of resulting SNAs (Figure 1A). To test this hypothesis, class B CpG oligonucleotides²⁰ with varying quantities of C12 units (i.e., (C12)_n anchor where n indicates the number of C12 units ranging from 4 - 10) on their 3’ terminus were synthesized. Following purification, products were confirmed by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS, SI Table 1).

Figure 1. The anchoring chemistry used to tether oligonucleotides to a DOPC liposome impacts stability of the SNA shell.

A) Schematic of an immunostimulatory SNA displaying the chemical structures of the hydrophobic anchors used in these studies. B) FRET % signal remaining after incubation of FRET SNAs with varied anchor groups in a solution containing 60 M excess DOPC liposomes. C) Half-lives of mono exponential fits of (B). D) Change in FRET % over time after incubation of FRET SNAs with varied anchoring groups with 10% FBS. E) Half-lives of mono exponential fits of (D). Individual points represent mean values of an independent experiment (n=3). Data in (B, D) were fit to a one-phase exponential decay, half-lives (C, E) are presented with significance from ANOVA with Dunnett-corrected comparisons. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns = not significant. Black dotted line in B, D represents 0.1% SDS baseline. Data in C, E is presented as mean ± s.e.m.

The addition of hydrophobic units to the terminus of an oligonucleotide allowed facile synthesis of SNAs through insertion^{6, 21, 22} unto ~50 nm 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes (SI Figure 1). A significant size increase was observed by dynamic light scattering (DLS) following the addition of oligonucleotides, confirming the presence of the DNA shell. We next determined the loading of oligonucleotides for the far ends of the hydrophobic range tested (n=4 and n=9) through gel electrophoresis (SI Figure 2). Analysis of the gel electrophoresis revealed that 75 (C12)₄ DNAs compared to 400 (C12)₉ DNAs could be loaded onto the liposome before DNA dissociation was observed on the gel. This corresponds to densities of 1.6 pmol cm⁻² and 7.4 pmol cm⁻² for the (C12)₄ and (C12)₉ SNAs, respectively, and highlights how increasing anchor hydrophobicity strongly influences the adsorption of DNA onto the liposome surface. This (C12)₉ SNA density exceeds what has been previously reported with other anchoring groups (i.e., cholesterol or diacyl lipid). Since DNA loading greatly influences SNA-like properties,^{19, 23} we kept this parameter constant at 75 DNA:liposome for all subsequent studies.

Next, we assessed the stability of the SNAs as a function of the anchoring group through Förster resonance energy transfer (FRET). We prepared FRET SNAs containing rhodamine dye lipids and Cy5-labeled DNA for a range of selected (C12)_n anchors that cover a representative spectrum of hydrophobicity, as well as previously reported cholesterol and diacyl lipid anchors (SI Table 1). The FRET SNAs were first used to probe the DNA shell stability in the presence of 60 M excess of DOPC liposomes (Figure 1B). FRET signal (λ_exc=550 nm, λ_em=680 nm) is indicative of an intact DNA shell since the excited rhodamine core can transfer energy to the Cy5 DNA on the surface. Increasing the hydrophobicity of the anchor strengthened the affinity of the CpG DNA for the liposome and resulted in a prolonged half-life for tested SNA groups (Lipid SNA 24 ± 12 min, (C12)₆ SNA 36 ± 6 min, (C12)₉ SNA 53 ± 5 min, and (C12)₁₀ SNA 53 ± 7 min, Figure 1C). SNAs with more hydrophobic anchors also displayed greater FRET signal at the end of the study (8 h), indicating that more DNA is retained on the surface of the liposome when stronger anchoring groups are used (SI Figure 3). Less hydrophobic anchoring groups (cholesterol and (C12)₄) exhibited a quick exponential decay in FRET signal (apparent SNA half-lives of 10 ± 1 min and 9 ± 1 min, respectively, Figure 1C), indicating that these anchors are more dynamic and lead to more rapid diffusion of the DNA shell. The higher hydrophobicity exhibited a slower exponential decay in FRET signal, highlighting the benefit in reduced DNA diffusion from the liposome core that is obtainable with C12 oligomers. The addition of more than nine C12 units did not significantly enhance SNA half-life, indicating a limit to enhancing SNA stability through C12 oligomerization. Importantly, by oligomerizing hydrophobic units, the lifetime of the DNA shell on the SNA was increased 5-fold compared to the commonly employed cholesterol anchor. The addition of 0.1% sodium dodecyl sulfate (SDS) was used to disrupt liposomal structures and reduced FRET signal to ~4% of its initial value, proving that the FRET signal observed is due to intact SNA structures.

Similar trends were observed when FRET SNAs were incubated with 10% fetal bovine serum (FBS) to assess their stability in a physiologically-relevant environment (Figure 1D). In this experiment, a more rapid reduction in FRET signal was observed due to the interactions of the SNAs with serum proteins. (C12)₉ SNAs exhibited the greatest serum half-life (26 ± 3 min) compared to all other groups (Figure 1E). While (C12)₁₀ SNAs displayed an equivalent affinity for the liposome core in Figure 1C, the reduced serum stability compared to (C12)₉ SNAs can be attributed to an increased affinity to serum albumin.^24-26 Overall, the (C12)₉ anchor imparts improved serum stability to the SNA and is thus positioned to be more effective for therapeutic applications

Dendritic Cell Activation Is Determined by SNA Stability

To gain an understanding of how improvements in stability affect the SNA’s biological properties, we studied the interactions of SNAs with murine bone marrow-derived dendritic cells (BMDCs) since immunostimulatory SNAs are known to interact with DCs to generate robust T cell responses in vivo.^{14, 16, 27} As cellular uptake of SNAs depends on the density of oligonucleotides on the surface,²³ we hypothesized the enhanced stability of the C12 oligomers would also enhance SNA uptake and activation of BMDCs via the Toll-like receptor 9 (TLR9) pathway. To ensure that the (C12)₉ anchor did not alter cellular entry, we used confocal microscopy (Figure 2A, representative gating strategy is provided as Figure S4) to confirm uptake into BMDCs for both cholesterol and (C12)₉ SNAs. Flow cytometry validated that the SNA structure with the greatest serum stability ((C12)₉ SNAs) exhibited the most rapid cellular uptake (Figure 2B). BMDCs treated with (C12)₉ anchored SNAs for 30 min showed a ca. 9-fold increase in Cy5 median fluorescent intensity (MFI) compared to its linear (C12)₉ DNA counterpart and a 3.6-fold MFI enhancement over cholesterol-anchored SNAs. The observed decrease in cellular uptake with the addition of a tenth C12 unit compared to (C12)₉ SNAs can be rationalized by the decrease in serum stability for (C12)₁₀ anchored SNAs. Overall, the uptake results are consistent with prior observations that enhancing the stability of liposomal SNAs enhances their cellular uptake^{18, 19} and shows that oligomerization of C12 units provides a tunable path towards programming stability and resultant SNA-like properties.

Figure 2. Dendritic cell uptake and activation is determined by SNA anchor chemistry.

A) Confocal microscopy of BMDCs following 30 min treatment with Cy5-labeled (red) linear CpG or CpG on an SNA. Nucleus stained with DAPI (blue). Scale bar: 5 μm. B) Uptake of Cy5-labeled SNAs or a linear (C12)₉ DNA control into BMDCs after incubation for 30 min. C) CD80 expression by CD11c⁺ BMDCs after incubation with different constructs all containing 1 μM by CpG. D) CD86 expression by CD11c⁺ BMDCs after incubation with different constructs all containing 1 μM by CpG. For all panels, error bars represent mean +/− s.e.m. Statistical comparisons to (C12)₉ SNAs were performed using ANOVA with Dunnett-corrected comparisons. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

We next investigated how the (C12)_n anchors impact the activation of BMDCs. To this end, BMDCs were treated for 20 h and subsequently evaluated for the expression of co-stimulatory markers. Specifically, co-stimulatory markers CD80 (Figure 2C) and CD86 (Figure 2D) were measured since these markers play an important role in T cell activation.^{28, 29} The most stable (C12)₉ SNA exhibited the greatest enhancements in both CD80 and CD86 expression. For CD80 expression, (C12)₉ SNA treatment resulted in an 11.7-fold increase compared to untreated BMDCs and a 4.7-fold increase compared to cholesterol SNAs. Importantly, this data validates that the benefit observed with (C12)₉ SNAs is a result of the radial arrangement of CpG oligonucleotides on the SNA surface, as linear (C12)₉ CpG failed to potently activate co-stimulatory markers. These trends were consistent when analyzing the upregulation of CD86 expression. (C12)₉ SNA treatment resulted in an 8.9-fold enhancement in CD86 expression compared to untreated BMDCs and a 2.2-fold enhancement over cholesterol SNAs. The linear (C12)₉ CpG oligonucleotides did not significantly activate CD86 expression. From these results, we conclude that enhancements in DC immunostimulation by the SNA are due to increased cellular uptake and intracellular TLR9 activation which results from the enhanced half-life and stability of the more hydrophobic C12 anchors.

(C12)₉ SNAs Elicit Powerful CD8⁺ T Cell-Mediated Antitumor Immunity

We next evaluated how enhancing SNA stability using the nine C12 oligomers impacted immune responses in vivo. To evaluate antigen-specific immune responses from SNA vaccination, we employed the model OVA1 antigen (OVA_257-263 from the ovalbumin protein). The antigen was either encapsulated into liposomal SNA structures with varying anchors (i.e., cholesterol OVA1 SNA or (C12)₉ OVA1 SNAs) or administered free in solution as a simple mixture of adjuvant and antigen (i.e., linear CpG + OVA1). C57BL/6J mice were vaccinated three times biweekly (Figure 3A, 6 nmol by CpG and OVA1) through subcutaneous (SQ) injections in the abdomen to promote drainage to the inguinal lymph nodes.^{13, 14} On day 35, the mice were sacrificed, and splenocytes were isolated to analyze antigen-specific T cell immune responses raised from the treatments.

Figure 3. (C12)₉ SNAs elicit potent antigen-specific CD8 T Cell immunity and inhibit tumor growth.

A) Injection timeline for immunizations for B-E. B) Percentage of raised OVA1-specific CD8⁺ splenic T cells as detected by H-2K^b-Ig/OVA dimer staining after SNA vaccination. C) Percentage of polyfunctional (CD107a⁺IFN-γ⁺) splenic CD8 T Cells upon restimulation with OVA1 peptide. B and C: n=7, from two independent experiments. D) IFN-γ spot-forming cells (SFCs) after stimulation of splenocytes with OVA1 peptide (n=4). E) Representative counted ELISpot images. F) Dosing schedule for treatment of E.G7-OVA lymphoma tumors (G-I). G) E.G7-OVA tumor growth with different treatments (n=10-13). H) Kaplan-Meier survival curves of E.G7-OVA lymphoma with statistical significance assessed via the Mantel-Cox log-rank test. I) Surviving mice are resistant to rechallenge with E.G7-OVA lymphoma cells. All graphs highlight mean +/− s.e.m. Statistical significance in B, C, D, G, and I was determined by one-way ANOVA with Tukey corrected comparisons. *p<0.05, **p<0.01, ***p<0.001 ****, p<0.0001.

Consistent with the in vitro observations where the (C12)₉ SNAs promoted the strongest DC activation, (C12)₉ OVA1 SNA vaccination also led to the most robust OVA-specific CD8⁺ T cell immune responses in vivo. There was a greater frequency of OVA1-specific CD8⁺ splenic T cells raised from (C12)₉ OVA1 SNA vaccination (Figure 3B). In particular, (C12)₉ OVA1 SNA vaccination yielded a 2-fold enrichment in OVA1-specific T cells above cholesterol OVA1 SNA treated mice and a 2.6-fold enrichment above the mixture of linear CpG and OVA1. The frequency of CD8⁺ T cells expressing the effector cytokine IFN-γ and degranulation marker CD107a was also quantified in response to restimulation with the OVA1 peptide (Figure 3C). The (C12)₉ OVA1 SNA immunization led to the strongest response with 22% of CD8⁺ T cells producing both IFN-γ and CD107a, a 3.9-fold increase compared to cholesterol OVA1 SNAs, and a 3.2-fold increase above the mixture of linear CpG and OVA1. This indicates elevated levels of polyfunctional T cells, which are considered potent against chronic infections and tumors.¹³ We next confirmed the antigen-specific secretion of IFN-γ through an Enzyme-Linked Immunosorbent Spot (ELISpot) assay (Figure 3D,E). (C12)₉ OVA1 SNA vaccination resulted in 2.3-fold more spot-forming cells (SFCs) than cholesterol OVA1 SNAs and a 1.9-fold higher number of SFCs than the linear CpG and OVA1 mixture. Together, these results demonstrate how elevated stability of the (C12)₉ SNAs corresponds to an increased antigen-specific CD8⁺ T cell response in vivo.

Motivated by the robust antigen-specific CD8⁺ T cell immune responses generated by (C12)₉ OVA1 SNA vaccination; we evaluated the potency of these SNAs against E.G7-OVA lymphoma tumors which express the entire ovalbumin protein.³⁰ Mice were SQ inoculated with 5 × 10⁵ lymphoma cells in the right hind flank and treated weekly with equimolar doses (6 nmol by CpG and OVA1) of either (C12)₉ OVA1 SNA, cholesterol OVA1 SNA, or a mixture of linear CpG and OVA1 (Figure 3F). Both (C12)₉ OVA1 SNAs and cholesterol OVA1 SNAs limited tumor growth at early stages, resulting in 21.9-fold and 4.6-fold reductions in tumor volumes compared to saline-treated mice on day 17, respectively (Figure 3G, spider plots of individual tumor growth curves are provided in SI Figure 5). However, (C12)₉ OVA1 SNA treatment proved superior to cholesterol OVA1 SNA treatment in the long-term after treatments were no longer administered, likely due to the enhanced antitumor activity of the OVA1-specific CD8⁺ T cells raised from (C12)₉ OVA1 SNA vaccination. Overall, (C12)₉ OVA1 SNA treatment resulted in a significant extension of median survival (53.5 d) compared to cholesterol OVA1 SNA (28.5 d), linear CpG and OVA1 (22.5 d), and the saline control (22.5 d) (Figure 3H). Notably, 6 of the 12 mice treated with (C12)₉ OVA1 SNAs were kept tumor free at the end of the study (70 d). Moreover, when the surviving mice were rechallenged on day 70 with 5 × 10⁵ E.G7-OVA tumor cells, no tumor growth was observed compared to naïve untreated mice (Figure 3I). This highlights the ability of (C12)₉ OVA1 SNAs to mount a more effective antitumor T cell memory response, leading to complete tumor elimination in these animals.

(C12)₉ SNA-mediated CpG Delivery Reduces Risk of Acute Cytokine Release

To gain greater insight into the biological responses arising from (C12)₉ SNA administration, we sought to characterize the proinflammatory serum cytokine profile after administration of adjuvant compared to the linear CpG form and cholesterol SNAs. Prior work has shown that linear CpG administration can lead to rapid cytokine release, which upon repeated administration, can attenuate the generation of an effective immune response.³¹ Additionally, the excessive activation of the immune system can lead to severe, possibly fatal systemic inflammation.^32-35 We hypothesized that the SNA architecture would provide a targeted delivery of CpG, which would reduce the risk of acute cytokine release. C57BL6/J mice were treated with 6 nmol of CpG by either linear CpG, cholesterol SNAs, or (C12)₉ SNAs, and cytokine release in response to the stimulation was measured.

Analysis of systemic cytokine secretion shortly after treatment (1 h) revealed a spike in several pro-inflammatory cytokines in mice treated with linear CpG but not in mice treated with either cholesterol SNAs or (C12)₉ SNAs. Compared to (C12)₉ SNA treated mice, linear CpG induced a 23-fold increase in IL-2 concentration (Figure 4A), a 96-fold increase in IL-6 concentration (Figure 4B), a 14-fold increase in IL-10 concentration (Figure 4C), a 42-fold increase in IL-4 concentration (Figure 4D), a 7-fold increase in the concentration of chemokine CXCL1 (Figure 4E), and a 64-fold increase in TNF-α concentration (Figure 4F). IL-6 and TNF-α are particularly noteworthy due to their association with toxic cytokine storms in disease.³⁶ Mice treated with either cholesterol or (C12)₉ SNAs did not show significant serum increase in these pro-inflammatory cytokines at this time point compared to untreated mice. Similar trends were observed with additional cytokines (i.e., IFN-γ, IL-12p70, IL-1β, IL-5) (SI Figure 6) with linear adjuvant inducing a substantial spike at 1 h in serum cytokine secretion. This highlights the enhanced safety of SNA-mediated therapy and demonstrates the (C12)₉ SNA’s ability to effectively deliver a potent immunostimulatory payload with reduced systemic release of pro-inflammatory cytokines that can cause fatal cytokine storms.

Figure 4. SNA adjuvants reduce the risk of acute cytokine release.

Pro-inflammatory serum cytokine concentrations after subcutaneous administration of 6 nmol CpG oligonucleotide (n=4). Cytokines include A) IL-2, B) IL-6, C) IL-10, D) IL-4, E) CXCL1, and F) TNF-a. All significant differences in cytokine concentrations are presented via ANOVA with Tukey-corrected comparisons. Data is shown as mean with error bars representing s.e.m. *p<0.05, **p<0.01, ***p<0.001, **** p<0.0001.

(C12)₉ SNAs Potentiate Antigen-Specific T Cell Immunity Towards A Conserved SARS-CoV-2 Epitope

Cytokine storms are associated with severe cases of COVID-19,^{37, 38} which necessitates the development of potent therapies which do not cause off-target immune responses. Since (C12)₉ SNAs promote potent antigen-specific CD8⁺ T cell immunity with minimal off-target effects, we sought to employ this construct to generate potent cellular immunity towards the coronavirus, SARS-CoV-2. We identified a human CD8⁺ epitope conserved from the original SARS-CoV virus³⁹ that has affinity for the human leukocyte antigen allele A*201 (HLA-A*201) (SI Table 2). Sequence alignments of different SARS-CoV-2 variants (Alpha (B.1.1.7)⁴⁰; Beta (B.1.351)⁴¹; Delta (B.1.617.2); Epsilon (B.1.427)⁴²; Kappa (B.1.617.1)⁴¹; Omicron (B1.1.529)⁴³) reveal that this epitope has remained highly conserved and is thus a judicious target for therapeutic development. Due to the modularity of the SNA structure, encapsulation of this peptide, termed herein as CoV, within the liposomal SNA to generate CoV SNAs was readily achieved using a protocol identical to the generation of (C12)₉ OVA1 SNAs, exchanging the OVA1 peptide for the CoV peptide.

To generate immunity towards this human T cell epitope in vivo, we vaccinated immunocompetent humanized B6.Cg-Immp2l^{Tg(HLA-A/H2-D)2Enge}/J (AAD) mice with either a simple mixture of linear CpG DNA and CoV peptide, cholesterol CoV SNAs, or (C12)₉ CoV SNAs. These mice express a hybrid MHC I complex that contains the alpha-1 and alpha-2 domains of the human HLA-A*201 complex and an alpha-3 domain from the murine H-2D^d allele.⁴⁴ Therefore, an innate immune response can be propagated by antigen-presenting cells through activation of murine TLR9 while presenting human epitopes. The same dose (6 nmol) and treatment schedule as Figure 3A was followed, with spleens harvested on day 35 for ex vivo analysis. Potent, antigen-specific immune responses were only observed through (C12)₉ CoV SNA vaccination. A shift in the splenic CD8⁺ T cell phenotype towards an effector memory (CD62L⁻CD44⁺) state was 4.7-fold and 2.9-fold higher compared to cholesterol CoV SNA and the mixture of linear CpG and CoV peptide, respectively (Figure 5A). We also assessed the polyfunctionality of T cells raised from vaccination. The T cells raised from (C12)₉ CoV SNA vaccination and restimulated with CoV antigen produced doubly positive IFN-γ and CD107a T cells at 11.5- and 6.3-fold greater frequencies than the cholesterol CoV SNA and mixture of linear CpG and CoV peptide, respectively (Figure 5B). ELISpot measurements corroborated this result (Figure 5C and D), with only (C12)₉ CoV SNA treatment capable of raising splenocytes that produce a significant and large number of SFCs (ca. 358). We explored the cytolytic capability of these T cells to target human T2 cells pulsed with CoV peptide (Figure 5E). T2 cells are deficient in the transporter associated with antigen processing (TAP) protein and present exogenous peptide fragments on HLA-A*0201^{45, 46} similar to the cells infected by SARS-CoV-2 in HLA-A*0201⁺ individuals. At all tested ratios of T cells to T2 target cells, (C12)₉ CoV SNA vaccination produced cytolytic T cells capable of inducing apoptosis in the T2 target cells, with a 100:1 ratio able to lyse nearly half of the T2 population. Even at ratios of 12.5:1, 25:1, and 50:1, where negligible killing was observed from T cells raised from mice immunized with the mixture or cholesterol CoV SNA, T cells from (C12)₉ CoV SNA-immunized mice killed ca. 16%, 28%, and 39% of the T2 cell population, respectively. The cytolytic capability of anti-CoV CD8⁺ T cells from was further explored in vitro using recombinant CHO-K1 cells that stably overexpress human SARS-CoV-2 spike protein on the surface (Figure 5F). Detection of caspase-3⁺ CHO-K1 cells revealed a significant increase for (C12)₉ CoV SNA-raised human T cells, which induced apoptosis in ca. 45% of the CHO-K1 Spike cell population. These results show that the (C12)₉ SNA structure enhances the cellular immunity of a desired antigen, regardless of disease class, and results in effective antiviral T cell immunity.

Figure 5. (C12)₉ SNAs enable generation of an antiviral CD8⁺ response towards a conserved region of SARS-CoV-2.

A) Effector memory subset (CD62L⁻CD44⁺) of splenic CD8⁺ T cells from AAD mice after vaccination. B) Intracellular staining of splenic CD8⁺ T cells for IFN-γ production and CD107a expression after restimulation with CoV antigen. A and B (n=6 from two independent experiments). C) Number of IFN-γ expressing cells per 2×10⁵ splenocytes via ELISpot after incubation with CoV antigen is increased with (C12)₉ CoV SNA vaccination (n=4). D) Representative ELISpot images of the data presented in C. E) Antigen-specific killing of T2 target cells presenting the CoV antigen by CD8⁺ T cells raised from in vivo vaccination. Percentage of T2 cells double-positive for Annexin V (early apoptosis) and 7-aminoactinamycin D (7AAD, necrotic) are shown (n=3). F) Percentage of caspase-3 detected in target CHO/K1-Spike cells following co-culture with human T cells that were primed with respective treatments (25 μM). Statistical significance among treatment groups was assessed by ANOVA with Tukey-corrected comparisons. Data shown as mean ± s.e.m. *p<0.05, **p<0.01, ***p<0.001, **** p<0.0001.

CONCLUSIONS

This study defines a method for tailoring and enhancing liposomal SNA stability through the use of hydrophobic C12 DNA anchoring units. In general, the dissociation rate of the oligonucleotides from the SNA shell decreases as a function of oligomer length with optimum stability and immune stimulation at n=9 for the C12 units. The nine C12 moieties provide an ideal anchor hydrophobicity for strong intercalation into the liposome core and maximize oligonucleotide packing without affecting the phospholipid bilayer of the liposome. It is worth noting that increasing C12 moieties beyond nine results in increased and detrimental construct affinity for serum albumin, as opposed to greater liposome intercalation. Importantly, this approach to increasing anchor hydrophobicity though iterative C12 units to stabilize LSNAs is general and independent of the oligonucleotide sequence, as it is easily controllable with a conventional synthesizer. Therefore, one can use it to define the optimum stability profiles across immunostimulatory and immunoinhibitory applications. The programmable nature of this approach is advantageous over other nanoparticle systems, which require alteration of the chemical composition to achieve similar modifications in stability. Moreover, critically, the use of C12 optimized constructs avoids acute and nonspecific cytokine release, likely due to increased intracellular uptake prior to target binding compared to linear DNA. Remarkably, regardless of cancer or infectious disease model, the constructs designed via the C12 approach raise robust T cell responses, and therefore provide a path toward developing potent therapeutic vaccines.

EXPERIMENTAL SECTION

Materials, Animals, and Instrumentation:

Oligonucleotides were synthesized as described below utilizing phosphoramidites and synthesis reagents purchased from Glen Research (SI Table 1). Peptides were synthesized and purified (>95%) by Northwestern’s Peptide Synthesis Core. Nanopure water was used from a MilliQ system equipped with a 0.22 μm filter All other reagents were purchased from commercially available sources as described below. E.G7-OVA and T2 cells were purchased from ATCC. CHO-K1/Spike cells were purchased from GenScript. Female C57BL/6 (10-12 weeks old, JAX stock #000664) and B6.Cg-Immp2l^{Tg(HLA-A/H2-D)2Enge}/J (AAD) mice (8-9 weeks old, JAX stock #004191) were purchased from Jackson Laboratory and used in accordance with national and local guidelines and regulations. Animal protocols were approved by the institutional animal care and use committee (IUCAC) at Northwestern. Mass spectra were collected on a Bruker MALDI Rapiflex Tissue Typer in linear negative mode. UV-Vis measurements were performed on Agilent Cary 60 spectrophotometer baselined using water in a 1 cm² path length cuvette. Cell counts were performed utilizing a ViCELL Blue Viability Analyzer. Cytometry data was collected on a FACS Symphony A3 Cell Analyzer (BD). ELISpot SFCs were counted on a CTL ImmunoSpot Analyzer.

Software and Statistical Analysis:

Extinction coefficients to calculate DNA concentrations were generated by the OligoAnalyzer^™ Tool on IDT’s website (https://www.idtdna.com/pages/tools/oligoanalyzer). MALDI peaks identified in SI Table 1 are from the m/z value at the center of the largest identified peak identified after baseline subtraction. Statistical analysis was performed as indicated in the figure captions using GraphPad Prism version 9.2.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com.

Oligonucleotide Synthesis:

Oligonucleotides were synthesized on an ABI 394 DNA synthesizer at 1 or 10 μmole scale utilizing standard phosphoramidite chemistry on a controlled-pore glass Unylinker support (Chemgenes) with a phosphorothioate backbone. Cholesterol-terminated and thiol-terminated DNAs were synthesized on 3’-cholesteryl-TEG and 3’ thiol modifier C3 S-S supports, respectively. Following synthesis, DNA was cleaved and deprotected in an aqueous solution of 30 wt% ammonia and 40 wt% methylamine 1:1 (v/v) (Sigma-Aldrich) at 55° C for 50 min. DNAs containing a C12 spacer (12-(4,4'-Dimethoxytrityloxy)dodecyl-1-[(2-cyanoethyl)-(N,N-diisopropyl)]-phosphoramidite) were synthesized without a 5’ dimethoxytrityl (DMT) group and required lengthened (16 h) deprotection time at 55 °C to remove the Unylinker support. Cy5-labeled DNA was deprotected under mild conditions (30 wt % ammonia for 20 h at room temperature). After deprotection, ammonia was evaporated under a stream of nitrogen gas and lyophilized. C12-containing strands were purified by preparative-scale denaturing (8 M urea) polyacrylamide (15%, 19:1 acrylamide:bis, BioRad) gel electrophoresis (PAGE). 0.5 μmole of DNA in 0.5 mL water was mixed 1:1 with loading dye (8M urea with bromophenol blue) and loaded onto the gel which was run for 30 min at 175 V then ramped to 350 V for an additional 3 hours. Bands were visualized with a 260 nm UV lamp against a thin-layer chromatography plate and the highest retention band was isolated, crushed, washed 3x with water, lyophilized, and washed 3-4 times with water in a 3 kDa Amicron ultra-centrifugal filter (Millipore-Sigma) prior to identification of the desired product by MALDI-ToF spectroscopy (Bruker RapiFlex Tissue Typer). For MALDI-ToF, 0.5 μL sample was spotted with 0.5 μL of dihydroxyacetophenone matrix. Other DNAs which contained a 5’ DMT were purified by reverse phase high-performance liquid chromatography with a gradient of 0.1M triethylammonium acetate (aq) to acetonitrile on a C18 or C4 (for Cy5 and cholesterolated DNAs) resin. Fractions with the greatest column retention were isolated, lyophilized, and the DMT group was removed in 20% acetic acid (aq) for 1 h, washed 3 times with ethyl acetate, and lyophilized again prior to resuspension in water and product identification by MALDI-ToF. Lipid-anchored DNAs were synthesized by reacting 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE, Avanti) with a crosslinker, succinimidyl 4-(p-maleimidophenyl)butyrate (SMBP Thermofisher) followed by reaction with a 3’thiolated DNA (CpG-SH) following previously established protocols.¹⁷

Liposome Synthesis and Peptide Encapsulation:

1 mL of DOPC in chloroform (25 mg mL⁻¹, Avanti Cat# 850375C) was added to a glass scintillation vial, evaporated under a stream of nitrogen, then lyophilized and stored under argon. 1 mL of DPBS was used to rehydrate this thin film followed by 10 freeze thaw cycles using liquid nitrogen and a sonicated water bath (37°C). For peptide encapsulated liposomes, thin films were instead hydrated with 1 or 2 mL of a 1 mg mL⁻¹ solution of the desired peptide in DPBS. OVA1 required the addition of a small amount (50-100 μL in 10 mL) of hydrochloric acid (Sigma) to solubilize the peptide. Immediately following the freeze-thaw cycles, the lipid solution was diluted to 10 mL and subjected to sequential pressurized extrusions through polycarbonate membranes (3x 200 nm, 1x 100 nm, 1x 80 nm, 3x 50 nm) to generate a population of 50 nm liposomes whose sizes were confirmed by DLS. Excess peptide was removed from peptide-encapsulated liposomes through dialysis in a 3.5 kDa Slide-A-Lyzer cassette (Thermofisher Cat# 66370) against 3.5 L of DPBS overnight. Liposome concentrations were determined through a fluorescent phosphatidylcholine assay (Sigma Cat# MAK049-1KT) using the approximation that each 50 nm liposome contained 18,140 lipids. Peptide concentrations were determined by a Pierce quantitative fluorometric assay (Thermofisher Cat# 23290). The encapsulation efficiency of peptide was calculated to be as high as 30%. The average number of peptides per liposome in a sample was determined by dividing the measured peptide concentration by the liposome concentration. Since loading varied in each sample, samples were combined such that the weighted average of peptides per liposome was 75 prior to the addition of DNA.

Liposomal SNA Synthesis:

SNAs were assembled by addition of DNA (typically 60 uM) with a 3’ hydrophobic anchor to preformed liposomes (typically 0.8 μM) in DPBS followed by brief sonication and shaking overnight at 37°C. No further purification was performed since DNA intercalation is quantitative at this ratio (SI Fig S2).

FRET Stability Measurements:

To measure the stability of SNAs as a function of anchor group, 20 μL of either blank liposomes (1.2 nmol) or normal FBS was added to 180 μL of FRET SNA (20 pmol), quickly mixed, and then fluorescence intensities: I_A (λ_exc =550 nm, λ_em = 680 nm) and I_D (λ_exc =550 nm, λ_em = 610 nm) were measured every 4-5 min under 37 °C using a BioTek Cytation 5 microplate reader. The decay in FRET signal was calculated according to Equation 1 at each timepoint and fitted to a single-phase exponential decay in Prism 9.2.

$${FRET}_{\%}=\frac{I_A}{I_D+I_A}\times 100$$ #(1)

Dendritic Cell Uptake, Activation, and Imaging:

Bone marrow cells were isolated from both the femur and tibia of C57BL6/J mice by rinsing the bone marrow with 4-5 mL of Roswell Park Memorial Institute Media containing 10% (v/v) heat-inactivated FBS and 1% (v/v) penicillin-streptomycin (RPMI (+/+)). Cells were pelleted and red blood cells lysed with 2 mL of ACK lysing buffer (Gibco, Cat#A10492). Cells were washed with DPBS and 5x10⁶ cells were plated and differentiated to a dendritic cell phenotype through incubation with granulocyte-macrophage colony-stimulating factor (GM-CSF, Biolegend Cat# 576306, 40 ng mL⁻¹) in 10 mL RPMI (+/+) in a petri dish in a 37°C incubator supplied with 5% CO₂. 5 mL of RPMI (+/+) was added 4 and 6 d after initial plating. On day 7, cells were collected and pelleted, resuspended in RPMI (+/+) at 1.67x10⁶ cells mL⁻¹, and 3x10⁵ cells were transferred to individual flow tubes. For activation, cells were treated with 200 pmol (1 μM) of CpG DNA for 20 h, or the dose indicated in the figure where dose was varied. For uptake measurements, cells were treated with 20 pmol (100 nM) of Cy5-labeled DNA of CpG for 30 minutes. Following treatment, cells were washed with 800 μL PBS and diluted in an antibody staining cocktail containing 50 μL of DPBS with 0.5 μL of the following antibodies for activation: CD11c (PacBlue, Biolegend Cat# 117322), CD80 (BD Cat# 560526, PerCPCy5.5), CD86 (PE, BD Cat# 105008), 0.5 μL fixable blue live/dead stain (Thermofisher Cat# L23105). CD80 and CD86 antibodies were omitted from the uptake staining. Cells were briefly vortexed and incubated for 30 min at 4°C, then washed with 600 μL DPBS, and resuspended in 100 μL fixation buffer and stored at 4°C (Biolegend Cat# 420801). Flow cytometry was run the following day. For confocal microscopy, 20 μL of fixed cells were placed onto a microscope slide and sealed using a micro cover glass using previously established methods.¹⁴ Images were acquired using a Zeiss LSM800 confocal microscope (Zeiss).

Splenocyte Isolation Following Vaccination:

One week following three fortnightly vaccinations, mice were sacrificed, and spleens isolated and stored temporarily in ~3 mL of RPMI (+/+). Splenocytes were obtained by mashing individual spleens (or 2-3 pooled spleens for naïve controls) with a sterile syringe plunger and rinsing the cells through a 0.7 μm polycarbonate filter with ~30 mL DPBS. The cells were then pelleted, and red blood cells were lysed with 2 mL ACK lysing buffer (Gibco Cat# LSA1049201) for 4 minutes. Cells were diluted to 30 mL in DPBS, counted, pelleted, and resuspended in RPMI (+/+) at 1x10⁸ cells(mL)⁻¹ and used in the following assays where splenocytes are indicated.

OVA1-MHCI Dimer Staining:

OVA1 peptide was dissolved in DPBS at 0.5 mg(mL)⁻¹ and incubated with H-2K^b:Ig PE dimer (BD Cat# 552944) at a 160:1 molar ratio overnight following manufacturer’s instructions. 1x10⁶ splenocytes from vaccinated mice were then incubated for 1 hour at 4°C with 50 μL DPBS containing 0.25 μg of the prepared OVA1-MHCI dimer as well as 0.5 μL each of CD3 (PerCpCy5.5, BD Cat# 551163), CD8a (APC, BD Cat# 553035), and fixable blue live/dead stain (Thermofisher Cat# L23105). Cells were then washed with 600 μL DPBS and resuspended in 100 μL fixation buffer. After overnight storage at 4°C, flow cytometry was used to measure the frequency of splenic CD8⁺ T cells positive for the OVA1-H2k^b dimer.

Intracellular IFN-γ Staining:

4x10⁶ splenocytes from individually vaccinated mice were transferred to flow tubes and restimulated for 4 hours at 37°C in a 5% CO₂ incubator with 500 μL RPMI (+/+) solution containing monensin (2 μM, Biolegend Cat# 420701), brefeldin A (5μg mL⁻¹, Biolegend Cat# 420601), CD107a surface antibody (0.5μL, FITC, BD Cat# 553793) and the peptide antigen (10μg mL⁻¹). The no restimulation control contained splenocytes from a (C12)₉ vaccinated mouse with only RPMI (+/+) media. Following incubation, cells were washed with 500 μL DPBS and resuspended in 50 μL DPBS containing Trustain FcX Blocker (0.5 μL, Biolegend Cat# 5566704) prior to the addition of 50 μL DPBS containing 0.5 μL each of the following surface antibodies: CD4 (APC, BD Cat# 100412), CD8a (PE, BD Cat# 553033), and the fixable blue live/dead cell stain. The splenocytes were briefly vortexed then incubated at 4°C for 15 min. After, splenocytes were washed with 600 μL DPBS, and resuspended in 100 μL fixation and permeabilization solution (BD Cat# 51-2090KZ), vortexed, and incubated for 20 min at 4°C. 300 μL of 1X permeabilization wash buffer (BD Cat# 51-2091KZ) was added to the cells, and cells were resuspended in in 100 μL permeabilization wash buffer containing 0.5 μL IFN-γ (PE-Cy7, BD Cat# 557649) or an isotype control. Splenocytes were stored at 4°C overnight and analyzed using flow cytometry.

IFN-γ ELISpot Assay:

Enzyme-linked immunospot (ELISpot) assay for murine IFN-γ detection was performed according to the manufacturer’s protocol (BD Cat# 551083). In brief, each well of the provided 96-well plate was coated with IFN-γ capture antibody in PBS for ~16 hours of incubation at 4°C. The antibody solution was removed, and the plate was washed once with 200 μL of RPMI (+/+) and blocked with an additional 200 μL RPMI (+/+) at room temperature. After 2 hours, media was removed and replaced with 2x10⁵ splenocytes in 100 μL RPMI (+/+). An additional 100 μL of media containing either peptide antigen (10 μg/mL), or CD28 (Biolegend Cat# 102116) and CD3 (Biolegend Cat# 100340) antibodies (2 μg/mL each for a positive control), or media only (negative control) was immediately added, and samples were incubated for 48 h at 37°C in a 5% CO₂ incubator. After this incubation, the plate was washed, and the detection antibodies and substrate were added according to manufacturer’s protocol. The plate was dried overnight, and spots were counted on a CTL ImmunoSpot Analyzer.

Effector Memory Staining:

3x10⁵ splenocytes in flow tubes were washed with 600 μL DPBS prior to suspension in 50 μL DPBS containing 0.5 μL of the following antibodies: CD8a(APC, BD Cat# 553035), CD4 (FITC, BD Cat# 553729), CD44 (APC, Biolegend Cat# 103012), CD62L(PE-Cy7, Cat# 560516), and fixable blue live/dead stain. Splenocytes were vortexed and incubated at 4°C for 15 min, then washed with 600 μL DPBS, and resuspended in 100 μL fixation buffer. Fixed splenocytes were stored at 4°C overnight prior to flow cytometry analysis.

Serum Cytokine Profiling:

C57BL/6J mice were injected subcutaneously with 6 nmol CpG DNA or SNA. At the indicated timepoint following treatment, ~50 μL of blood was collected via retroorbital blood draw using a heparinized pipette tip and transferred to a collection tube. After 30-60 min incubation at room temperature the coagulate was centrifuged at 2000xg for 10 min and 30 μL of the serum supernatant was transferred for storage at −20°C prior to MSD analysis. Multiplex cytokine analysis was performed using an MSD multi-spot assay system with the proinflammatory panel 1 mouse kit (MSD Cat# K15048D) according to the manufacturer’s instructions. Briefly, 50 μL diluted serum samples and calibrators were added to the capture antibody precoated plate and incubated for 2 hr with shaking. The plate was then washed, and detection antibody solution was added, followed by another incubation for 2 hours with shaking. After the final wash, read buffer T was added to the plate and read immediately in a MESO Quickplex SQ120.

E.G7-OVA Tumor Study:

C57BL/6 mice were inoculated with 5x10⁵ E.G7-OVA cells in the right hind flank. Three weekly treatments began on day 4 following inoculation. Mice were treated (6 nmol CpG+peptide with either a simple mixture or SNA formulation) subcutaneously in the abdomen. Starting 6 days after tumor inoculation, blinded tumor measurements were made every 2-3 days by measuring the length and width of tumors. Tumor volume was approximated with these measurements using Equation 2. Mice were sacrificed after tumors reached 2000 mm³. On day 70, any surviving mice from the (C12)₉ SNA treatment group and an equivalent number of mice which never received treatment (naïve) were SQ reinoculated with 5x10⁵ E.G7-OVA cells and tumor growth was monitored via blind measurements.

$$\text{Tumor volume}=\frac{length\times {width}^2}{2}$$ #(2)

T2 Cell Killing:

T2 target human lymphoma cells which had been passaged 2-3 times with fresh Iscove’s modified dulbecco’s essential media containing 20% FBS and 1% penicillin-streptomycin (IMDM +/+) were washed with DPBS and resuspended in DPBS with 10 μM efluor450 cell proliferation dye (Thermofisher Cat# 65-0842-85). Cells were pelleted, the DPBS aspirated, and the T2 cells were incubated with the CoV peptide (10 μg/mL in RPMI +/+) for 2 h at 37°C in a 5% CO₂ incubator. During the incubation period, splenocytes from each immunized mouse were pooled and resuspended in 1x MOJO-sort buffer (Biolegend Cat# 480017) and CD8a⁺ T cells were isolated by positive selection with magnetic beads according to manufacturer protocol (StemCell Technologies, 18953). The purified T cells were resuspended in RPMI (+/+) and counted and resuspended at 5x10⁵ cells(mL)⁻¹. Either 100, 50, 25, or 12.5 μL of T cells were than plated on a round bottom 96-well plate and topped up to 100 μL with RPMI (+/+). T2 cells were counted following peptide incubation, then washed with RPMI, (+/+), and 5x10³ cells in 100 μL media were added to each well of the 96 well plate to generate the desired T cell:target ratio. Cells were co-cultured for 20 h at 37°C in a 5% CO₂ incubator, then transferred to flow tubes. The plate was washed with an additional 200 μL PBS to remove residual cells. Samples were washed with an additional 600 μL PBS in the flow tubes, pelleted, then the supernatant was aspirated, and the cells were suspended in 100 μL Annexin V binding buffer (Biolegend, 422201) containing 0.5 μL each of 7-AAD (Fisher Cat# 50169259) and Annexin V (Biolegend, 640906) for 15 min prior to running flow cytometry. Cytometry data was acquired within 2 h of completion of staining.

CHO-K1/Spike Cell Killing:

Human PBMCs were thawed from storage and cells were added to 10 mL of warmed RPMI (+/+) similar to previously published protocols.¹³ Briefly, the solution was centrifuged at 300 x g for 10 min to wash the cells with fresh media and subsequently count them. 1 x 10⁶ cells in 450 μL volume were added to each well of a 24 well plate, and cells were left in a 37 °C in a 5% CO₂ incubator to recover while samples were prepared. Samples were diluted in RPMI (+/+) media to 2´ the final concentration (final = 25 μM) and 450 μL of each was added to the respective wells. Following 48 h incubation, CD8⁺ T cells were isolated as described above using a Human CD8 Positive Selection Kit II (Stem Cell Technologies, 17853) and co-incubated with target CHO-K1/Spike cells (GenScript, M00803) at a 300:1 ratio for 2 h. Following treatment, the cells were washed with PBS and the entire volume was transferred to flow tubes. The tubes were spun at 1200 rpm for 5 min, after which the supernatant was aspirated, and the samples were stained in 100 μL PBS containing surface antibodies (0.5 μL per sample each of: CD45 (clone HI30) -BUV661, CD3 (clone SP34-2) -PE-Cy7) at 4 °C for 15 min. Cells were then washed with 600 μL PBS, centrifuged at 1200 rpm for 5 min, aspirated, and resuspended in 100 μL of Cytofix Fixation and Permeabilization solution (BD, 554722) for 20 min at 4 °C. Cells were then washed with 600 μL of Perm/Wash Buffer (BD, 554723), centrifuged at 1200 rpm for 5 min, aspirated, and resuspended in 100 μL of Perm/Wash Buffer with an additional 20 μL of Caspase-3 antibody (BD, BDB550914) and immediately analyzed using flow cytometry.