Biodegradable Polyester Nanoparticle Vaccines Deliver Self‐Amplifying mRNA in Mice at Low Doses

Delivery of self‐amplifying mRNA (SAM) has high potential for infectious disease vaccination due to its self‐adjuvanting and dose‐sparing properties. Yet a challenge is the susceptibility of SAM to degradation and the need for SAM to reach the cytosol fully intact to enable self‐amplification. Lipid nanoparticles are successfully deployed at incredible speed for mRNA vaccination, but aspects such as cold storage, manufacturing, efficiency of delivery, and the therapeutic window can benefit from further improvement. To investigate alternatives to lipid nanoparticles, a class of >200 biodegradable end‐capped lipophilic poly(beta‐amino ester)s (PBAEs) that enable efficient delivery of SAM in vitro and in vivo as assessed by measuring expression of SAM encoding reporter proteins is developed. The ability of these polymers to deliver SAM intramuscularly in mice is evaluated, and a polymer‐based formulation that yields up to 37‐fold higher intramuscular (IM) expression of SAM compared to injected naked SAM is identified. Using the same nanoparticle formulation to deliver a SAM encoding rabies virus glycoprotein, the vaccine elicits superior immunogenicity compared to naked SAM delivery, leading to seroconversion in mice at low RNA injection doses. These biodegradable nanomaterials may be useful in the development of next‐generation RNA vaccines for infectious diseases.


Introduction
Vaccines are one of the few strategic tools available for wide-spread utilization in combatting infectious disease epidemics and global pandemics. Yet, traditional vaccine development of live attenuated, inactivated, or subunit vaccines typically requires years of development and production time for wide-spread distribution. [1] Genetic vaccines using mRNA to encode pathogenic antigens are one of the most promising advancements in vaccine development strategies, as they allow for rapid development of functional vaccine candidates as soon as the sequence of the desired protein target is known. [2] This approach has been adopted with great efficiency by Moderna and BioN-Tech among others in response to the COVID-19 pandemic with great efficacy. [3,4] Self-amplifying mRNA (SAM) technology is an innovative vaccine platform for high expression of a target antigen. [5] In addition to the antigen gene of interest, SAM also encodes four alphavirus derived nonstructural proteins, which constitute the replication complex responsible for self-amplification of the original mRNA in the cytosol. [5,6] During the self-amplification process doublestranded RNA replication intermediates are formed and contribute to the induction of a robust Type I interferon response, that can be highly beneficial for eliciting a strong B cell response. [5] Due to its self-amplifying nature, SAM vaccines have the potential for high potency and dose-sparing, which could result in the production of a higher number of vaccine doses at an equivalent amount of mRNA when compared to conventional mRNA vaccines. [7,8] These features make SAM a platform optimal for vaccination and its utilization has been previously demonstrated to elicit protective immunity in different preclinical models, including mice and nonhuman primates against viral pathogens, including rabies, [9,10] influenza, [11] Ebola, [11] and HIV. [12] The biggest challenge associated with SAM is to achieve effective cytosolic delivery, as mRNA is highly susceptible to nuclease damage, which would eliminate its ability to self-amplify. [10,13] Delivery of SAM with gene delivery platforms including lipid nanoparticles, [10,14] polymeric dendrimers, [11] and cationic nanoemulsions [15] demonstrated varying degrees of success, but www.advancedsciencenews.com www.advtherap.com each strategy used to date has drawbacks. Importantly, requirements for cold storage, and in some cases, the nondegradability of a mixture of synthetic components pose a challenge. Slow degradability, such as with polymeric dendrimers based on polyamidoamine modified with alkyl-epoxides, may also create issues such as inducing inflammation in vivo. [11] Certain end-capped poly(beta-amino ester) (PBAE) terpolymers have recently been developed for systemic delivery of mRNA, resulting in high efficacy of delivery to lung endothelial cells when administered intravenously in mice. [16] Compared to PBAEs lacking alkyl-amine side chain monomers, these amphiphilic PBAEs enabled highly improved mRNA complexation, protection and delivery and further allowed co-complexation with PEG-lipids to reduce rapid reticuloendothelial system (RES) clearance of cationic nanoparticles. [17] The alkyl chains of these PBAEs also enhanced structural stability of the nanoparticles by the hydrophobic effect [16] in contrast to canonical PBAE polymers which rely more on electrostatic interactions with nucleic acids to drive nanoparticle nucleation. [18] These features of amphiphilic PBAE terpolymers mimic much of what makes lipid nanoparticles highly effective for nucleic acid delivery, while also incorporating the primary benefits of PBAE polymers, including increased avidity compared to ionizable lipids, rapid ester degradation catalyzed by the tertiary amines in the backbone of the polymer, [19,20] and structural end-cap monomer diversity enabling differential cell targeting. [21][22][23] Recent data have also highlighted that PBAE nanoparticles can be an order of magnitude more effective at endosomal escape to the cytosol compared to leading commercially available polymer and lipid-based transfection agents. [22] Due to the self-amplifying nature of SAM in the cytosol, modified nucleosides replacement cannot be used to generate viable SAM with improved stability [24] and requires intact delivery of full-length SAM molecules to the cytosol for activity; [2,5] these features coupled with the long transcript length of SAM makes SAM especially susceptible to degradation by extracellular nucleases compared to other nucleic acid cargos. Thus, SAM vaccines or therapeutics can directly benefit by encapsulation into nanoparticles, both to increase intracellular delivery efficiency and to offer protection from degradation.
Here, we describe the development of a biodegradable cationic polyester for intramuscular delivery of SAM and demonstrate its efficiency in delivering SAM both in vitro to a myoblast mouse cell line and in vivo in mice following intramuscular administration. The resulting nanoparticles formed via a bulkmixing and dialysis process were demonstrated to be consistent in size as well as stable following lyophilization. This work builds upon research utilizing similar polymers for systemic nucleoside modified mRNA delivery, where prior generation polymers have shown effectiveness for delivery to the lungs and spleen. [16,22] When used to deliver a rabies antigen encoding SAM, the new nanoparticles enabled protective immunity by eliciting neutralizing serum antibodies at doses of only 200 nanograms of RNA per mouse. Overall, this polymeric nanoparticle platform using rapidly biodegradable cationic esters holds promise as an effective delivery vehicle for intramuscular delivery of large RNA molecules, such as SAM. The hydrophobic PBAEs used here benefit from rapid polymer backbone degradation that limits cytotoxicity, have reduced complexity in the number of components required in the nanoparticle formulation, and as self-assembly is driven by electrostatic interactions with the nucleic acid cargo, can be manufactured flexibly from either bulk batch processes or fluidic mixing. [16,25,26] In addition, PBAEs have high avidity to nucleic acids due to the repeating amine groups throughout the polymer, enhanced endosomal escape compared to leading commercially available polymer and lipid-based transfection agents, [22] and the ability to precisely tune chemical structure to facilitate cell-targeting. [21] The polymer and RNA self-assemble to small, low polydispersity nanoparticles with high mRNA encapsulation efficiency, and that can be further surface shielded in a modular fashion with a sheddable PEG-lipid in the same manner as commercially approved lipid nanoparticles.

Biophysical Properties/Characterization of PBAE Nanoparticles Encapsulating SAM
To assess properties of PBAE terpolymers (Figure 1) with SAM ( Figure 1A), we synthesized a library of PBAEs from small molecule monomers with structure denoted as described (Figure 1B) following the naming scheme for B:base monomers, S:side-chain monomers, Sc:alkyl-side chain monomers and E:end-cap monomers ( Figure 1B) and using hyphens to separate classes of monomers. [27] While PBAE terpolymers enable co-complexation with nucleic acids using exclusively aqueous buffers for mixing, co-formulation with lipids and a mixing strategy using acidified ethanol and dialysis yielded similar nanoparticle diameters ( Figure 1C) with improved RNA encapsulation as previously observed in related structures. [16,22,28] Analysis of the PEG-lipid stabilized PBAE terpolymer eGFP SAM nanoparticles by transmission electron microscopy (TEM) revealed dried nanoparticles of ≈100 nm in diameter with a spherical shape ( Figure 1D). Inclusion of the alkyl-amine side chain monomer [29] was crucial to enable co-complexation of PEG-lipid molecules for surface shielding of the nanoparticles. We selected use of DMG-PEG2k as a saturated, 14-carbon diacyl lipid for sheddable PE-Gylation, which has been previously demonstrated to be useful for progressive shedding of the PEG-lipids in vivo to enable cell uptake. [30,31] In our experiments using acidified ethanol and dialysis to encapsulate SAM, admixing DMG-PEG2k at a mass percentage between 0 and 20% resulted in smaller, less polydisperse nanoparticles, with a neutral zeta potential for DMG-PEG2k content ≥5% by mass ( Figure 1E). This approach was applicable to PBAE terpolymers containing 12-carbon alkyl-amine side chain monomers of varied structure including for the two lead nanoparticles later identified ( Figure 1F), 7-90,c12-63, 50% Sc12, and 5-3,c12-39, 30% Sc12. Following screening of a polymer library with diverse structures, both lead nanoparticle formulations had a convergence of biophysical properties when assessed at a 30:1 weight PBAE:weight SAM ratio with the addition of 10% DMG-PEG2k by mass. At these conditions both types of nanoparticles had high encapsulation efficiency (≥94%), a particle size of ≈115 nm, and neutral zeta potential.

PBAE Nanoparticle Library Synthesis and In Vitro Screening
To enable rapid screening of a diverse library of lipophilic 4component PBAEs (Table S1, Supporting Information) for SAM delivery, we adapted a semi-high-throughput strategy we previously employed for screening canonical three-component PBAE polymers for plasmid DNA delivery. [32] This entailed synthesizing acrylate-terminated polymers in mid-size batches in parallel in glass vials and performing the final end-cap monomer reaction in parallel in 384-well plates. [32] For this approach, we synthesized sets of PBAEs to be acrylate-terminated (Table S2, Supporting Information), characterized the polymers (Table S3, Supporting Information), and then performed end-capping reactions with a set of previously identified high efficacy end-cap monomers ( Figure  S1, Supporting Information). [32] With this strategy we combinatorially synthesized a new library of 196 end-capped PBAEs with varying base monomer and side-chain hydrophobic monomer (horizontal on heatmap), and end-cap monomers (vertical on heatmap) (Figure 2A). Each polymer was then complexed with SAM at weight:weight ratios of 30 and 60 and assessed for delivery efficiency in differentiated C2C12 murine myoblasts in 384-well plates. Differentiated C2C12 cells were utilized as an in vitro model system to better recapitulate the muscular microenvironment, given the eventual intramuscular delivery route The approach used for initial transfection screening of this PBAE library in C2C12 focused on differential polymer structures (without inclusion of DMG-PEG2k) and aqueous mixing of SAM and PBAE to best enable high-throughput nanoparticle self-assembly in parallel in multi-well plates and assessment of a greater number of polymer structures. In our method validation ( Figure S2, Supporting Information), inclusion of DMG-PEG2k or use of the acidified ethanol dialysis encapsulation method did not improve transfection in vitro and similar trends in transfection efficacy were observed regardless of whether the nanoparticles were screened with or without the PEG-lipid. This is likely due to intracellular delivery and endosomal escape to the cytosol both being driven by the PBAE component, which makes up ≈87%-98% of the nanoparticle formulations evaluated by mass. With this approach to screening PBAE structures, we identified multiple PBAE structures effective for SAM delivery to differentiated C2C12 cells in vitro at extremely low doses of only 1 ng/50 μL media in each well of a 384-well plate equivalent to ≈1 ng per 20 000 cells/well as assessed by nuclei counting. Alternative endcap monomers previously effective for DNA delivery by PBAEs (including E6 and E7) were also included in an initial trial library of polymers but were not selected for our expanded 384-well screening library ( Figure S3, Supporting Information). At these doses, in addition to being efficacious, the PBAE nanoparticles tested were also observed as noncytotoxic in vitro, with negligible effects on C2C12 viability ( Figure S3, Supporting Information).
With the structural variation included in the 384-well screening library, we identified multiple polymer characteristics that improved delivery efficacy. Specifically, increasing alkyl-amine Sc side chain mole fraction (see Figure 1B, 7-90,c12-X, N% Sc12 series and 7-4,c12-X, N% Sc12 series) improved delivery efficacy in vitro even when alkyl-amine side chain mole fraction was increased up to ≥80%. Increasing alkyl-amine side chain length greater than 12 carbons was also beneficial for improving delivery efficacy in vitro. Finally, among the 12 end-cap monomers evaluated, the most effective end-caps across multiple polymers were E63, E58, and E39, which all possess three ionizable amines but differ by their hydrophobicity and ratio of their primary, secondary, and tertiary amines following reaction to the acrylate terminated polymers.
To demonstrate increased efficiency of expression with SAM compared to mRNA, PBAE nanoparticles encapsulating eGFP SAM and 5mou-modified eGFP mRNA were prepared in parallel and used to transfect differentiated C2C12 myoblasts at doses across multiple orders of magnitude ( Figure 2B,C). Under these conditions, SAM yielded the same transfection efficacy at a 180fold lower dose compared to 5-methoxyuridine-modified mRNA in vitro. Recapitulating this two-orders of magnitude level of efficiency improvement in vivo has the potential to enable dosesparing that could dramatically reduce the supply constraints of mRNA vaccines for global vaccination, as encountered for SARS-CoV-2.
Top nanoparticle formulations identified by screening the 384well library were evaluated with dose titration for transfection of differentiated C2C12 cells. Among these nanoparticles, lead poly-mers achieved >70% transfection of cells at a dose of 5 ng/well and >40% transfection at a dose of only 185 pg/well ( Figure 2D). To better understand the influence of the alkyl-amine side-chain monomer on in vitro transfection, we synthesized two series of polymers varying either the length of the alkyl-amine sidechain monomer or the mole fraction (increasing the total polymers in the library evaluated to >200) ( Figure S4, Supporting Information). These experiments clearly demonstrated that both increased length of the alkyl-amine side-chain monomer and increased alkyl-amine side-chain mole fraction improved SAM transfection in vitro.

Intramuscular (IM) Delivery of PEGylated SAM PBAE Nanoparticles
To identify nanoparticles effective for intramuscular SAM delivery in vivo, we used a SAM construct encoding luciferase (FLuc) to rapidly assess overall SAM delivery efficiency when injected intramuscularly with live animal imaging. We first chose a single PBAE nanoparticle formulation and assessed the duration of expression when injected intramuscularly in BALB/c mice (Figure S5A, Supporting Information), observing that peak expression occurred at ≈10 d post-injection and expression at that dose of SAM persisted through ≈40 d post-injection. From this experiment, we selected 10 d as the time-point for live animal imaging studies for nanoparticle efficacy assessment in vivo. We further assessed optimal w/w ratio for a single PBAE nanoparticle formulation, evaluating 15, 30, and 60 w/w ratios with DMG-PEG2k included at a 10% mass fraction (Figure 3A), selecting a 30 w/w ratio as optimal under these conditions. In contrast to in vitro assessment ( Figure S2, Supporting Information), where inclusion of DMG-PEG2k moderately reduced transfection, PEGylation of the nanoparticle surface with DMG-PEG2k was shown to be required for efficacy following intramuscular injection of these polymeric nanoparticles ( Figure 3B), presumably by improving their extracellular transport properties. Particles not formulated with DMG-PEG2k yielded lower transfection intramuscularly than naked SAM injected under matched conditions and particles formulated with 10% mass DMG-PEG2k yielding the highest expression. This result was in some ways surprising, as cationic nanoemulsions, [12,15] cationic liposomes, [9] and cationic polyethyleneimine polyplex nanoparticles [33] have previously been demonstrated to be effective for intramuscular administration of nucleic acids despite their positively charged zeta potentials. Yet, the observation that cationic PBAE nanoparticles (even with alkyl side chains for increased hydrophobicity) yielded no expression with intramuscular administration is consistent with prior observations of PBAE nanoparticles suppressing expression of plasmid DNA when injected intramuscularly [33] unlike when DNA or RNA containing PBAE nanoparticles are injected by other local routes such as intratumorally [34][35][36] or intraocularly, [37] where there is high expression. This is most likely due to the extracellular environment of the muscle, perhaps by unPEGylated PBAE nanoparticles with a positive zeta potential sticking to the high concentration of extracellular matrix proteins in the intramuscular space and failing to reach cell plasma membranes to initiate endocytosis prior to the PBAE backbone beginning hydrolytic degradation. [38] Using polymers with different modes of degradation (like self-immolative oligo(alphaamino ester)s) may be a possible pathway to improve efficacy without the use of PEG, as it was shown that that such chargealtering releasable transporters (CART) have demonstrated successful intramuscular mRNA delivery at injection doses of 7.5 μg mRNA. [39] In any case, mixing in 10% by mass DMG-PEG2k into the PBAE/SAM nanoparticles during formulation is an easy and modular way to functionalize them to be PEG-shielded for in vivo applications at low doses and the safety of using DMG-PEG2k has been demonstrated in humans via its inclusion in the Moderna COVID-19 vaccine. [4] Intramuscular dosing of naked mRNA in mice can present a challenge in reproducibility and utility for assessing delivery efficacy, as murine quadriceps are quite small and naked mRNAs injected in buffer are capable of transfecting muscle cells primarily because of the high hydrostatic pressure achieved when injecting a relatively large volume, without intrinsic ability for the mRNA to safely reach the cytosol itself. [6,[39][40][41][42] Naked nucleic acid expression following intramuscular injection in mice does not reproduce in human patients when typical intramuscular injection volumes in mice (20-50 μL) do not scale with either body surface area (228-fold higher in adult humans, equivalent of 5-11 mL) or body mass and routine intramuscular injection volumes in humans are limited to 500 μL ( Figure S5, Supporting Information). The two primary mRNA-based vaccines brought to market for SARS-CoV-2 by BioNTech/Pfizer and Moderna use doses of 30 and 100 μg in injection volumes of 300 and 500 μL, respectively. [4,43] Scaling down from these doses and injection volumes to approximate a clinically relevant intramuscular injection in mice is not feasible due to the accuracy and precision constraints of injecting 5 μL intramuscularly. We evaluated using 7-90,c12-63, 50% Sc12 nanoparticles compared to naked SAM injected intramuscularly at a variety of doses between 7.5 and 0.05 μg and injection volumes between 50 and 5 μL ( Figure S5, Supporting Information). Based on these results we selected a dose of 0.2 μg in 20 μL injection volume (0.01 μg μL −1 ) for all following experiments ( Figure S5, Supporting Information).
Using this optimized dose and injection volume of SAM with intramuscular injections, we evaluated nine different unique PBAE nanostructures prepared at a 30 w/w ratio with SAM and 10% by mass DMG-PEG2k ( Figure 3C). Among these polymers, six polymers led to significantly higher luciferase expression than naked SAM, with lead performing formulations of 7-90,c12-63, 50% Sc12 and 5-3,c12-39, 30% Sc12 yielding an increase of 23and 37-fold higher luminescence over naked SAM ( Figure 3C,D). The structures of these lead polymers were considerably different from one another and demonstrate multiple structural possibilities for high levels of intramuscular delivery efficacy. The 7-90,c12-63 PBAE used a morpholino based ionizable side chain and bisphenol A (BPA) based diacrylate monomer, while the 5-3,c12-39 PBAE used an amino-alcohol based ionizable side chain and a pentanediol based diacrylate. In the context of vaccine development, the higher performing 5-3,c12-39 PBAE may also have a better-tolerability profile due to the avoidance of BPA. [44] One of the primary roles associated with nanoparticle-based encapsulation of SAM and other mRNAs for intramuscular administration is to improve the potency and robustness of delivery over the injection of naked nucleic acids. For the 10 injections shown in representative IVIS images of two lead formulations and naked SAM, PBAE nanoparticle-based encapsulation yielded strongly detectable expression of luciferase 7-9 times out of 10 compared to naked SAM, where only 1/10 injections yielded a strong luminescent signal ( Figure 3E). Even at these low injection doses and volumes, the results show that the biodegradable polymeric nanoparticles can facilitate in vivo intracellular delivery in muscle. It is critical that the PBAE nanoparticles were able to induce significant SAM expression in muscle at modest doses because these small scales have the potential to be translatable to human patients in a manner that larger volume hydrodynamic injection cannot. The hydrodynamic effect by which naked nucleic acids injected intramuscularly in mice mediate effective cy-tosolic delivery has been demonstrated to have scaling challenges to larger animals like non-human primates. [45,46] Many studies have assessed correlation between transfections in vitro and in vivo for screening different nanoparticle formulations, often revealing low correlation between the optimal materials identified for in vitro and in vivo based delivery. Notably, for lipid nanoparticles with delivery assessed using a barcoded, pooled library approach for delivery to macrophage cell lines in vitro and via intravenous administration in vivo, there was a linear correlation with R2 < 0.1. [47] In the context of intramuscular delivery, there have been strategies to improve in vitro models of muscle by using ECM mimicking substrates to influence the mechano-transduction of muscle cells and recapitulate endocytic pathways observed in vivo. [38] While the approach of creating more complex but applicable in vitro culture conditions for identifying nanoparticles effective for in vivo utilization is attractive, implementation can be difficult. For example, the low stiffness (10 kDa) hydrogel system developed by Bhosle et al. presents challenges for high-throughput screening due to its limited applicability to a multiwell plate format. [38] In the current study, PBAE structures were identified by in vitro multi-well plate screening utilizing differentiated C2C12 myoblasts and further evaluated via in vivo intramuscular delivery. Expression following intramuscular administration was weakly positively correlated with in vitro transfection efficacy (R2 = 0.313, Figure 3F). Differences between the in vitro and in vivo experiments, such as biodistribution, the effect of PEGylation on cellular uptake, and the nature of a 2D versus 3D environment may all contribute to lessening the strength of the correlation. Nonetheless, formulations observed to have low transfection efficacy during the screening in vitro also had low transfection in vivo. In vitro screening of the PBAE system showed utility in that all the nanocarriers that achieved greater than 50% transfection of C2C12 myoblasts in vitro also had the capacity to successfully transfect in vivo following incorporation of DMG-PEG2k and IM injection.

Immunization with PBAE Nanoparticle Delivered SAM Improves Rabies Antibody Titers in Mice
Using a SAM construct encoding rabies virus glycoprotein, we assessed the potential of PBAE nanoparticles to deliver SAM to elicit neutralizing antibodies relative to injections of naked SAM. [10,13] Using a vaccination schedule of homologous prime/boost separated by three weeks, serum was collected two weeks following boost administration (day 35) (Figure 4A). Neutralizing antibody titers against live rabies virus were then assessed via rapid fluorescent focus inhibition technique (RFFIT) assay. [48] The RFFIT assay directly assesses the presence of rabies virus neutralizing antibodies (RVnAbs) that can neutralize the rabies virus and prevent infection of healthy cells, providing better correlation for protection than anti-rabies virus glycoprotein binding antibody (bAb) titers measured by ELISA. [48] Comparison between naked SAM and both PBAE NPs tested (7-90,c12-63, 50% Sc12 and 5-3,c12-39, 30% Sc12) RVnAbs levels using the Mann-Whitney test demonstrated statistically significant antibody titer produced after initial immunization and booster injection. An antibody titer above the 0.5 IU mL −1 threshold (indicated by the dotted line) is considered protective by the assay, [48] demonstrating at the low mRNA doses a seroconversion rate of 6/10 animals for both PBAE NP formulations and 2/10 for naked SAM ( Figure 4B). Thus, both biodegradable polymeric nanoparticle formulations improved the seroconversion of IM injected SAM compared to hydrodynamic injection of naked SAM. Delivery of SAM IM in various nanoparticle formulations has been previously reported in the literature. [49] Researchers have found that similar levels of titers can be achieved following IM dosing of 150 ng SAM in higher volume 50 μL injections (on day 0 and day 28) using 1,2-dioleoyl-3-trimethylammoniumpropane (DOTAP) or dimethyldioctadecylammonium bromide (DDA) liposomes. [50] As polymeric PBAE NPs reached a neutralizing antibody level with low volume injections (10 μL), these polymeric NPs may be able to satisfactorily deliver SAM without relying on high hydrostatic pressure injection to facilitate intracellular delivery. Future research is needed with additional doses, volumes, and controls to fully elucidate the potential advantages of PBAE NPs for SAM delivery. These findings merit further investigation as higher doses and additional tuning of intracellular delivery parameters towards antigen-presenting cells could further boost titers. It is also noteworthy that as PBAE nanoparticles are capable of being lyophilized and stored in nonfrozen conditions, [51] future investigation may prove that they can be beneficial from a supply chain perspective and/or in the development of alternative routes of administration, including via microneedles, which has been demonstrated using PBAEs and plasmid DNA. [52]

Conclusion
In this study, we developed biodegradable end-capped lipophilic poly(beta-amino ester) terpolymers to enable the delivery of SAM constructs both in vitro and in vivo for intramuscular vaccination. By screening a library of >200 PBAE terpolymers with varied base polymer structure, differential side chain hydrophobicity and varied end-cap monomers, we identified optimal structural properties to enable highly efficient SAM delivery to myoblasts at sub-nanogram doses per well. In particular, small changes to end-group structure and % of lipophilic side chain could make a profound difference to rates of transfection. Inclusion of alkyl side chains enabled admixing with DMG-PEG2k to yield nanoparticles with high encapsulation efficiency and neutral zeta potential for effective intramuscular administration. Among lead PBAE terpolymers further evaluated by intramuscular injection in mice, optimal PBAE formulations enabled up to 37-fold higher intramuscular expression of SAM compared to injected naked SAM constructs. In vitro screening with C2C12 myoblasts was found to be helpful in identifying polymeric nanocarriers with the capacity for successful intramuscular transfection. Delivery of SAM encoding rabies virus glycoprotein via two different PBAE nanostructures at low doses of mRNA led to protective seroconversion among most animals vaccinated and demonstrated higher humoral immunity compared to injection of naked SAM. This study reveals that biodegradable polymers, as a class of nanomaterials, can be promising delivery vehicles for next-generation mRNA-based vaccines. Although the field is dominated by lipid-based materials for nonviral mRNA delivery, biodegradable polymers have the potential benefits of a broader therapeutic window, ease in manufacturability, possibility for nonfrozen supply chain, and efficiency of delivery without high hydrostatic pressure required. This motivates future work in further optimizing dosing, excipients, scaleup, and storage to better realize the potential of this class of nanomaterials.

Experimental Section
Materials: Monomers were purchased from vendors listed in Table S1 (Supporting Information) and used without further purification. Acrylate monomers were stored with desiccant at 4°C, while amine monomers were stored with desiccant at room temperature. mRNA for eGFP (5methoxyuridine, 5mou) was purchased from TriLink Biotechnologies (L-7201). All solvents were purchased from MilliporeSigma.
RNA Synthesis: SAM is an alphavirus-derived mRNA, which comprises genes for nonstructural proteins (NSPs) and a gene of interest (GOI) whose expression was enabled via a subgenomic promoter (Figure 1A). Three SAM constructs were prepared coding for eGFP, firefly luciferase and a dual firefly luciferase-2A-rabies antigen SAM separated by a 2A ribosomal skip site. RNAs were transcribed in vitro from template DNA constructs using T7 polymerase and purified as previously described, [53] and RNA integrity was validated by agarose gel electrophoresis. Ability of the in vitro transcribed RNAs to self-amplify and express the target antigens was measured in BHK cells as previously described. [54] Polymer Synthesis: PBAEs were synthesized as previously described and shown in Figure S1 (Supporting Information) using a twostep Michael addition reaction with a combination of backbone, side chain, and end-cap monomers. Bioreducible monomer BR6 (2,2disulfanediylbis(ethane-2,1-diyl) diacrylate) was synthesized according to Kozielski et al. [55] PBAE polymers were synthesized at the molar ratios of monomers specified in Table S2 (Supporting Information). The first Michael addition reaction between the backbone and side chain monomers occurred at 90°C for 48 h with stirring producing acrylateterminated base polymers. The second Michael addition reaction occurred in anhydrous tetrahydrofuran (THF) at room temperature for 1 h with stirring resulting in end-capped PBAEs. These polymers were then precipitated into anhydrous diethyl ether with centrifugations at 3200 rcf and washing twice with anhydrous diethyl ether to purify the polymer. The PBAEs were then dried under vacuum for 48 h to eliminate residual diethyl ether, dissolved in anhydrous dimethyl sulfoxide (DMSO) at 100 mg mL −1 , and stored in aliquots at −20°C with desiccant to limit freeze/thaw cycles. PBAE nomenclature follows the numbering of diacrylate, hydrophilic amine, hydrophobic amine with percentage, and amine end-cap monomers shown in Figure S1 (Supporting Information).
Polymer Characterization: Prior to end-capping reactions, acrylate terminated polymers after the first Michael addition reaction were sampled and precipitated twice in anhydrous diethyl ether to yield a neat polymer that was then dissolved in a small amount of anhydrous DMSO-d6. The sampled acrylate-terminated polymers were dried under vacuum for 2 h then dissolved in additional DMSO for NMR spectrum analysis of acrylate peaks via Bruker 500 MHz NMR. Similar analysis was done with polymer samples post-end capping to confirm complete reaction by elimination of acrylate peaks between 5.5 and 6.5 ppm. Gel permeation chromatography (GPC) (Waters, Milford, MA) was also used to characterize the MN, MW, and polydispersity index (PDI) relative to linear polystyrene standards for the sampled acrylate-terminated polymers as previously described. [56] Encapsulation Efficiency Assay: Loading efficiency of SAM loaded into the nanoparticles was analyzed using the commercial Invitrogen Ribogreen RNA analysis kit (ThermoFisher) as described previously. [28] Nanoparticles were complexed using two different PBAE formulations (7-90,c12-63, 50% Sc12 and 5-3,c12-39, 30% Sc12) to encapsulate SAM with 10m% DMG-PEG2k and underwent dialysis. Nanoparticles were then diluted to ≈1 ng μL −1 SAM in PBS pH 7.4 buffer. Standards using the SAM molecules were between 0.125 and 2 ng μL −1 . Prepared nanoparticles were then mixed with either PBS buffer or 10 mg mL −1 heparin in TE buffer with the later disrupting the polymer binding allowing for the release of SAM. The nanoparticles were incubated in the buffer for 15 min at 37°C and then diluted Ribogreen reagent was added and incubated for 3 min at 37°C. Fluorescence of Ribogreen was measured using a plate reader (Biotek Synergy) at 500 nm/525 nm excitation/emission according to the supplied protocol to determine encapsulation efficiency.
Nanoparticle Preparation: Nanoparticles were prepared without incorporation of lipid-PEG for high-throughput screening of the polymers' ability to facilitate intracellular delivery in vitro or by adding DMG-PEG2k as an extra component to neutralize surface charge followed by dialysis as previously described. [16] For transfections in 96-well plates, nanoparticles were formed by dissolving synthesized PBAE polymers in DMSO and eGFP SAM separately in 25 × 10 −3 m NaAc pH 5.0 buffer and combining them at a 1:1 volume ratio. The mixture was incubated at room temperature for 10 min to allow for self-assembly into nanoparticles. For transfections in 384well plates, nanoparticles were formed by resuspending synthesized PBAE polymers in 25 × 10 −3 m NaAc pH 5.0 buffer in parallel using a ViaFlo 384 (Integra Biosciences). Resuspended PBAE polymer was then mixed in parallel with SAM to yield a final nucleic acid concentration of 0.03 μg μL −1 in a 384 polypropylene nanoparticle source plate.
Nanoparticle Characterization: For nanoparticle characterization via dynamic light scattering (DLS), SAM and PBAE polymer were prepared using DMG-PEG2k and dialysis. [16] The Z-average hydrodynamic diameters and zeta potential of the nanoparticle formulations in 25 × 10 −3 m NaAc and in sixfold dilution using isotonic 150 × 10 −3 m PBS pH 7.4 buffer at 25°C, respectively, were measured by DLS using a Zetasizer Pro (Malvern Instruments, Malvern, UK) with a 173°detection angle. For TEM, nanoparticles were prepared at 30 w/w ratio with 10 m% DMG-PEG2k using dialysis against PBS for 75 min. Twenty microliters of nanoparticles were used to coat a corona plasma-treated carbon film 400 square mesh TEM grid for 60 min. Grids were then briefly washed in ultrapure water to eliminate excess dried salt crystals and dried under vacuum before acquiring images using a Philips CM120 (Philips Research, Briarcliffs Manor, NY).
Cell Culture: C2C12 murine myoblast cells were purchased from ATCC (Manassas, VA, CRL-1772) and expanded in DMEM supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin. For differentiation to myotube-like cells, C2C12 cells were plated at a density of 31250 cells cm −2 in tissue culture plates in DMEM supplemented with 2% horse serum, 1% insulin, selenium and transferrin (ITS, 41400045, ThermoFisher). For 96-well plate transfection efficacy experiments, cells were plated on CytoOne 96-well tissue culture plates (USA Scientific, Ocala, FL) 4 d prior to transfection with 12 000 cells/well in 100 μL complete differentiation media and media was changed on days 2 and 4. For noted 384-well plate transfection experiments, C2C12 cells were plated at 2500 cells/well in 50 μL complete differentiation media in 384well tissue culture plates (Santa Cruz, sc-206081) 2 d prior to transfection and media was replaced on the day of transfection. Cells were confirmed periodically to be mycoplasma negative via the MycoAlert test (Lonza).
In Vitro Transfection: For 96-well transfections, 20 μL of SAM nanoparticle dilution with nucleic acid concentration between 1.03 and 250 pg μL −1 was added to each well of cells in 100 μL of complete media for a 2 h incubation before changing media. Cell viability 24 h post-transfection was assessed using the MTS Celltiter 96 Aqueous One (Promega, Madison, WI) cell proliferation assay. For 384-well transfections, 5 μL of nanoparticle dilution was added to each well containing cells in 50 μL of complete media and left to incubate for 2 h before a media change. Percent transfection efficiency was assessed after 48 h by staining nuclei with Hoechst stain and imaged for eGFP expression using a Cellomics Arrayscan VTI (Thermo Fisher Scientific, Laguna Hills, CA), an automated fluorescencebased high-content screening imaging system.
In Vivo Experiments: Animal work was performed in compliance with an approved protocol by the Johns Hopkins University Animal Care and Use Committee (ACUC; approval number MO21M384 ). Female BALB/c mice, 6-8 weeks old were purchased from The Jackson Laboratory and maintained in accordance with the JHH animal care facility. For in vivo transfection analysis, nanoparticles were made with luciferase (FLuc) SAM and PBAE polymers and injected intramuscularly in mice at 0.2 μg dose for bioluminescent luciferase expression assessment at specified time points. For assessment of luciferase expression, mice were injected intraperitoneally (i.p.) with 100 μL of 150 mg kg −1 d-luciferin (potassium salt solution in 1× PBS; Cayman Chemical Company, Ann Arbor, MI). After 7 min, mice were anesthetized using isoflurane and imaged using an In Vivo Imaging System (IVIS Spectrum; PerkinElmer, Shelton, CT) to measure bioluminescence. For the rabies vaccination studies, 6-8 weeks old female BALB/c mice were each injected intramuscularly with nanoparticles carrying SAM encoding both rabies virus glycoprotein antigen and the luciferase reporter protein (0.1 μg in 10 μL in opposite quadriceps for 0.2 μg total dose) on day 0 followed by a booster on day 21. Serum was then collected from the mice at day 35, 14 d after the booster vaccination.
Neutralizing Antibody Titer Assay: Serum samples were analyzed for rabies virus neutralizing antibody (RVNA) titer using a rapid fluorescent foci inhibition test (RFFIT) at the Kansas State University Rabies Laboratory. [48] Serum was first diluted fivefold and then serially fivefold before incubating with live rabies virus. Cultured cells were then combined with the serum dilutions with virus to test for protection resulting from RVNA presence via a titer value calculated from the percent of infected cells. KSU Rabies Laboratory and the World Health Organization (WHO) report a titer of 0.5 international units per millimeter (IU mL −1 ) as a protective response resulting from rabies vaccination and this level is indicated on data plots with a dotted line.
Data Analysis and Statistics: Cellomics HCS Studio (Thermo Fisher) was used for image acquisition-based in vitro transfection analysis as previously described. [32] Polymer structures were characterized in ChemDraw (Perkin Elmer, Boston, MA) and Marvin (ChemAxon, Cambridge, MA) to determine logP and logD values. Prism 8 (GraphPad, La Jolla, CA) was used for all statistical analyses and curve plotting. Unless otherwise specified, statistical tests were performed with a global alpha value of 0.05. Unless otherwise stated, absence of statistical significance markings where a test was stated to have been performed signified no statistical significance. Statistical significance was denoted as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.