Opportunities and Challenges in the Delivery of mRNA-Based Vaccines

4.1.1. Cap Analog

A synthetic cap analogue can readily be added to the mRNA because the 5’ end cap is not encoded by the DNA template. Natural eukaryotic mRNA has a 7-methylguanosine (m7G) cap coupled to the mRNA during the transcription process via a 5’-5’-triphosphate bridge (ppp) [32]. The m7GpppN structure at the 5’ end of the mRNA cap serves several functions [33]. First, it protects the mRNA from rapid degradation by exonucleases. Second, it plays an indispensable role during translation because the eukaryotic initiation factor (eIF) 4E recognizes and binds to the cap of the mRNA. It further plays a role in preventing innate immune sensors from recognizing the mRNA [34]. The mRNA may contain one of three distinct caps, i.e., cap-0 [m7G(5’)pppN1pN2p], cap-1 [m7G(5’)pppN1mpNp] and cap-2 [m7G(5’)pppN1mpN2mp], respectively [11]. Capping of IVT mRNA can be performed using two different approaches: The first approach includes the addition of a second step with recombinant vaccinia virus-derived capping enzymes after the transcription, resulting in a cap identical to the most frequent endogenous eukaryotic cap structure i.e., 7-methylguanosine (m⁷G) cap [30]. Alternatively, a synthetic cap analogue may be added during the in vitro transcription reaction, hence capping and in vitro transcription is performed in a single step. This approach is referred to as co-transcriptional capping [35]. A major drawback of this approach is the competition between the cap analog and the GTP nucleotide required for in vitro transcription, which ultimately results in a fraction of uncapped and translationally inactive mRNA [36]. The fraction of uncapped mRNA containing 5’ ppp groups is more immune-stimulating, which can be rectified by treatment with phosphatase to remove the ppp group at the 5’ end of uncapped mRNA [37].

Three different classes of m7GpppG cap analogs are used [38]: (i) anti-reverse cap analogs (ARCAs) [39], (ii) 3’-O-Me-m7GpppG [40], and (iii) modified ARCAs [41]. Initial mRNA research was performed using mRNA containing the m7 cap analog (GpppG) [42], and it is currently the most commonly used mRNA cap in clinical trials. Unfortunately, a fraction of the m7GpppG cap analog used during in vitro transcription becomes incorporated in the opposite orientation and is therefore not recognized by the ribosomes, eventually resulting in lower translational activity. To avoid this, the so-called ARCA with only one 3’-OH group instead of two 3’-OH groups (ARCAs; m₂^7,3’−OGpppG) has been introduced to prevent incorporation in the opposite orientation [43]. ARCAs juxtaposing the traditional cap analogues have been shown to exhibit more than four times RNA transcription efficiency [44]. In addition, the duration and the levels of protein expression have been found to be enhanced in cells transfected with ARCA-capped IVT RNA [41]. Recently, new types of chemically-modified cap analogues have been introduced, e.g., of phosphorothioate, phosphorothiolate [35], imidiphosphate [45], locked nucleic acid [46], boranophosphate bonds [47] and other types of modifications, which provide the mRNA with resistance to decapping by the mRNA-decapping enzyme 2, eventually resulting in a longer half-life of the mRNA [48].

Conventionally, the synthetic 5’-Cap 0-capped RNA is transcribed by performing an in vitro transcription, where more than 80% of the GTP added to the reaction is substituted with a dinucleotide cap analog (i.e., m7G[5’]ppp[5’]G), resulting in initiation of transcription with the cap analog [49]. However, this approach has been shown to exhibit a number of efficiency-related drawbacks, which have been overcome by the introduction of ScriptCap™. ScriptCap™ involves the addition of enzymatically built cap-0 structures onto the RNA transcripts by employing a capping enzyme with 100% reaction efficiency [50]. Native eukaryotic mRNA may contain a cap-1 or cap-2 but never a cap-0 [11]. Therefore, IVT mRNA should contain a cap-1 or a cap-2 in order to be less immune-stimulating [51]. A hallmark of these cap structures is the methylation status of the 2’ position of the 5’ second last and third last nucleoside. Prior to the advent of the novel technology of CleanCap™ introduced by TriLink, synthetic mRNA with a cap-1 could only be prepared by enzymatic capping [52,53]. However, cap-1 or cap-2 can now be incorporated during co-transcriptional capping at a capping efficiency of approximately 94% [52]. Notably, the capping efficiency has been shown to be significantly higher than the efficiency achieved by traditional co-transcriptional capping with cap-0 or ARCA [54].

4.1.2. 5’ and 3’ Untranslated Regions (UTRs)

The importance of incorporation of 5’- and 3’-UTRs has been noted during in vitro post-transcriptional regulation of gene expression [55]. The numerous roles that UTRs play include (i) regulation of mRNA export from the nucleus, (ii) regulation of translation efficiency [56], (iii) orchestration of subcellular localization [57], and (iv) mRNA stability [58]. Introduction of α-globin 3’ end UTRs results in stabilization of mRNA, while the incorporation of beta-globin 5’ end and 3’ end UTRs leads to enhanced translational efficiency [59]. The optimal outcome is achieved by using two β-globin 3’-UTRs aligned in a head-to-tail configuration. α-globin and β-globin UTRs have been incorporated for tweaking the RNA for optimized in vitro transcription followed by mRNA electroporation of autologous T cells [60] and intranodal injection of naked antigen-encoding RNA [61]. Moreover, DCs transfected with antigen-encoding UTR-optimized mRNA have been used in a study involving immunization of cytomegalovirus-seropositive individuals and cancer patients [62]. In some situations, destabilizing the mRNA might be a viable approach to reduce the duration of protein synthesis. This may be accomplished by introducing adenylate-uridylate-rich elements in the 3’-UTRs of the mRNA, eventually resulting in faster mRNA degradation and shortening of the duration of protein expression [63].

4.1.3. Poly(A) Tail

The poly(A) tail plays a significant role in mRNA translation as well as for the enzymatic stability of mRNA. The poly(A) tail binds to several polyadenosyl binding proteins (PABPs) while working synergistically with 5’m7Gcap sequences to regulate translational efficiency [64]. Eukaryotic translation initiation factor eIF4E binds to the 5’m7G cap, which in turn complexes with eIF4G and eIF4A. PABP then interacts with the N-terminus of the eukaryotic translation initiation factor eIF4G, which forms an mRNP (messenger ribonucleoprotein) or a polysome complex [65]. The former depicts the mRNA-protein complex not yet involved in protein synthesis, while the latter is one that is already being translated. An adequately long poly(A) tail is required to circularize the mRNA via binding of PABPs to the poly(A) tail and the cap [55,66]. It has been observed that increasing the poly(A) tail length improves the efficiency of polysome generation and consequently influences the protein expression levels [67].

It has been shown that a gradual increase in the poly(A) tail length of IVT mRNA to 120 bases commensurately increases the protein expression level, while an increase in the number of bases beyond 120 does not further enhance protein expression [68]. Poly(A) tails can be added to mRNA by encoding the poly(A) tail in the DNA template, or by extension of the IVT RNA after transcription using recombinant poly(A) polymerase. However, polyadenylation with recombinant poly(A) polymerase results in variable poly(A) tail length, thereby yielding polyadenylated mRNA with varying lengths. Therefore, the preferred approach is the generation of poly(A) tails with well-defined length from the mRNAs transcribed from poly(A) tail-encoding DNA templates [69]. The physical interactions between the 5’ and 3’ ends of mRNA take place between the cap and the poly(A) tail [70]. The poly(A) tail also plays a role in preventing decapping and mRNA degradation because removal or shortening of the poly(A) tail to less than 12 residues results in degradation of the mRNA through cleavage of the 5’ cap structure and 5’ to 3’ exonucleolytic digestion or 3’ to 5’ degradation [71].

4.2.1. Naked RNA

The simplest administration strategy comprises intramuscular (i.m.) injection of naked mRNA, and proof of concept was originally demonstrated by in vivo reporter gene expression in mice [24]. Since then, the effectiveness of naked mRNA has been confirmed upon i.m. [86], subcutaneous (s.c.) [87] or intradermal (i.d.) [88] injections. I.d. and s.c. administration of mRNA has been shown to mediate healing of various skin diseases and to ameliorate wound healing by in situ expression of specific proteins in the skin [75,89]. Adopting this approach circumvents several obstacles otherwise associated with systemic administration of mRNA, e.g., clearance from the bloodstream via the liver, the kidneys, and the spleen [75]. Very efficient translation of the encoded protein has been shown for mRNA administered s.c. [90,91,92]. Interestingly, more efficient translation has at times been measured when compared to mRNA-loaded nanoparticle-based delivery systems [90,93]. Hence, it circumvents the need to employ carriers, eventually contributing to reduced cost and potential risk. Another advantage of administering mRNA-based vaccines via the s.c. route is that both cellular and humoral immune responses are induced [94] because the mRNA is expressed by both skin-resident DCs [95] and non-immune cells [96]. However, the outermost stratum corneum layer of the epidermis serves as a tight barrier to the absorption of topically administered drugs [97]. Heretofore, various approaches have been adopted to overcome this barrier, including physical (e.g., microporation [98], microneedles [99], and jet injection [100,101]), active (e.g., electroporation [102], iontophoresis [103], and sonophoresis [104]), and passive methods (e.g., nanoparticles [105] and liposomes [106]). Electroporation and sonoporation can transiently permeabilize the skin by means of electric pulses and low-frequency ultrasound, respectively, for efficient delivery of genes into the skin [107]. Modified mRNA-encoding vascular endothelial growth factor-A was formulated in citrate-buffered saline without the use of a delivery system and was used for i.d. vaccination of patients with type 2 diabetes [108]. However, in spite of pronounced, sustained, dose-dependent and cargo-specific vasodilation, blood flow increase, oxygen-metabolic upregulation, angiogenesis and neovessel formation in animal models, the vasodilatory and angiogenic activity did not translate into humans [108,109]. Microneedle-based delivery is also an efficient technique where micron-sized needle patch/arrays composed of water-soluble polymeric or sugar excipients are employed, into which the mRNA is incorporated [76]. The patches/arrays provide mechanical strength needed for the needle to permeate the stratum corneum and penetrate into the viable skin layers. Following injection, depending upon the type of microneedles, the patches/arrays degrade/dissolve and the encapsulated drug is released in the interstitial fluid of the skin [75]. Administration of mRNA by using dissolvable microneedles provides an important advantage, i.e., delivery in a solid dosage form circumvents the necessity of dealing with mRNA in a liquid dosage form [110], which therefore eliminates the menace emanating from RNase contamination, along with increasing mRNA stability, and shelf life [76]. Scavenger receptor-mediated endocytosis and micropinocytosis have been shown to be the active uptake mechanisms for naked mRNA in immature DCs [16]. However, naked mRNA displays a short plasma half-life, is prone to ribonuclease degradation, and faces difficulties in entering the cell. Therefore, delivery systems have been propounded to protect the mRNA and shield its negative charge.

4.2.2. Viral Vectors

Delivery of mRNA can be mediated by viral and non-viral vectors. Non-viral vectors can further be categorized into lipid-based delivery systems, polymer-based delivery systems, and lipid-polymer hybrid systems [111]. For viral RNA delivery, there has been a great deal of interest in the engineering of adeno-associated viruses to carry nucleic acid cargoes [112]. Genetically-modified viruses are usually employed for mRNA/gene delivery. The genes of these viruses are partially or fully substituted with model or therapeutic genes. A benefit of RNA viruses is that they are replicated and expressed locally in the cytoplasm. Positive strand RNA viruses are distinguished by a genomic sequence that can be translated directly into proteins of interest by host ribosomes. Notably, alphaviruses (e.g., Sindbis and Semliki Forest virus) [113], picornaviruses [114], and flavivirus [115] (e.g., Kunjin virus) have been employed for mRNA delivery. Various alphavirus vectors can be used to express high levels of exogenous protein in a wide spectrum of hosts [116]. Commonly used approaches include the direct substitution of structural genes with heterologous expression or placing the nonstructural genes downstream of the RNA sub-genomic promoter [117]. However, alphaviruses induce severe cytopathogenic effects, which restrict their application in gene therapy, although different strategies can be employed to surmount this challenge [118]. Some of these strategies include engineering of mutant vectors with mitigated cytotoxicity and temperature-inducibility, and self-inactivating vectors with point mutations in the nsP2 gene (especially at position 726, 259 and 650) [119]. Sendai virus (SeV) (murine parainfluenza virus type 1 or hemagglutinating virus of Japan) that belongs to the Paramyxoviridae family is worth mentioning owing to its popular application as a vector. It is favored for its high but transient gene expression levels, wide host cell specificity, low pathogenicity, and strong immunogenicity [120]. As a vaccine platform, the Venezuelan encephalitis virus is of particular interest [121]. These vaccines encompass the live-attenuated viral vaccine TC-83 and a formalin-inactivated variety of it is referred to as C-84, which boosts the efficacy and increases the duration of immunity upon administering the TC-83 vaccine via different administration routes, intranasal being the most widely used [122]. However, using viral vectors embodies crucial drawbacks associated with genome integration, and possible host rejection (immunogenicity and cytotoxicity) among others [123], hence provoking the need for non-viral vectors for mRNA delivery [10].

4.2.3. Polymer-Based Vectors

Diethylaminoethyl (DEAE) dextran was the first polymer to be tested as a delivery reagent for IVT mRNA [124]. Later, it was shown that lipid-mediated mRNA transfection is 100 to 1000 times more efficient than DEAE-dextran [125]. This discovery stalled the progress of polymeric carriers and paved the way for lipid-based transfection reagents for nucleic acids, including mRNA. A comprehensive study compared the polymers poly-beta-amino-esters (PBAE) and polyethylenimine (PEI) with commercial transfection reagent Lipofectamine™ 2000 and 1,2-dioleoyl-3-trimethylammonium propane (DOTAP)/ 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) for functional, antigen-specific T-cell responses after mRNA delivery [126]. All carriers were complexed with mRNA encoding the HIV-1 antigen gag. Gag-specific, IFN-γ secreting T cells were measured in the spleen and lymph nodes of mice immunized with gag mRNA complexed with cationic lipids but not in mice immunized with naked and polymer-complexed mRNA. PEI and its derivatives are among the most commonly employed cationic polymers [127]. They are water-soluble, display a high density of positive charge associated with the amino groups, and are proven mRNA carriers for in vitro transfection [128]. However, PEI displays toxicity issues owing to the high molecular weight (>25 kDa), which may arise from the adsorption of anionic serum proteins onto the polyplex surface between cationic polymers and anionic serum plasma proteins. However, the resultant increase in size is only transient as the proteins adsorbed on the surface of polyplexes prevent particle–particle aggregation in the long run [129]. Various efforts have been made to mitigate these challenges. The first proof of concept for safe and efficacious mRNA vaccine transfection employing cationic polymer was obtained by intranasal administration of 2 kDa PEI conjugated to cyclodextrin. Cyclodextrin conjugated to PEI enabled delocalization of the charge density on the polyamine backbone, hence reducing cytotoxicity and at the same time maintaining protonatable groups, resulting in improved transfection [130]. Polymeric nanoparticles composed of biodegradable polymers, e.g., poly(lactic-co-glycolic acid) (PLGA), are well suited for incorporation of hydrophobic and positively-charged molecules. They provide good colloidal stability, low toxicity, and the possibility of sustained release. However, due to the anionic nature of PLGA at physiological pH [131], the mRNA encapsulation efficiency is very low. Polymer-based carriers exhibit considerable potential for gene therapy owing to the substantial transfection efficiency and tolerable toxicity [132]. A series of multifunctional block copolymers, i.e., dimethylaminoethyl methacrylate (DEAMA), poly(ethylene glycol) methacrylate, and DEAEMA-co-n-butyl methacrylate, demonstrated a transfection efficiency of 77% and 50% in RAW 264.7 macrophages and DC2.4 dendritic cells, respectively, thereby exhibiting potential as a carrier for mRNA-based intracellular vaccine delivery [133]. While different types of polymers and copolymers have been tested, the correlation between the structure of polymers and their biological response, e.g., transfection and toxicity, was found to be poor and thus, design of various polymer-based delivery systems relies on empirical, rather than rational approaches [134]. Despite the advantages mentioned above, polymer-based delivery systems are not as clinically advanced as lipid-based delivery systems owing to their polydispersity and challenges pertaining to metabolism of large molecular weight polymers [21].

4.2.4. Lipid-Based Vectors

Vectors based on lipids or lipid-like compounds (lipidoids) represent the most commonly used non-viral gene carriers [21]. Various synthetic and naturally-derived lipids have been employed to form liposomes (complexes of liposomes and nucleic acids are referred to as lipoplexes) or lipid nanoparticles (LNPs), both of which have been reported to efficiently deliver mRNA-based vaccines (Table 2). LNPs are often formulated by using cationic lipids displaying tertiary or quaternary amines to encapsulate the polyanionic mRNA. Cationic lipids spontaneously encapsulate negatively-charged mRNA, mediated by a combination of attractive electrostatic interactions with RNA and hydrophobic interactions, and thus have been used alone or in combination for lipofection of mRNA. The first reported use of LNPs as delivery system for mRNA came in 2015, with the delivery system consisting of ionizable cationic lipid/phosphatidylcholine/cholesterol/PEG-lipid in the ratio of (50:10:38.5:1.5 mol/mol) [86]. Examples of cationic lipids include e.g., 1,2-di-O-octadecenyl-3-trimethylammonium propane (DOTMA) [135], DOTAP [136], and zwitterionic DOPE [137,138]. They are structurally denoted by a cationic headgroup, a hydrophobic tail group, and a linking group in between [139]. However, cationic lipids have been observed to exhibit pro-inflammatory reactions and undesirable side effects [140]. Therefore, neutral lipids are also incorporated into cationic liposomes to decrease toxicity and attain high transfection levels in vivo [106]. The mechanism of LNP-mediated delivery of mRNA is not fully understood, but LNPs are suggested to be internalized by endocytosis and are attached electrostatically and fused with the cell membrane via inverted non-bilayer lipid phases [21].

Liposomes are closed membrane structures, which are formed by self-assembly when phospholipids are dispersed in aqueous systems [141]. They consist of at least one phospholipid bilayer, which mimics the cell membrane structure enclosing an aqueous core [8]. DOTAP/DOPE at a 1:1 molar ratio has been reported as an effective transfection agent for mRNA encoding the HIV-1 antigen Gag, which successfully induced an antigen-specific immune response in vivo in mice [126]. Additionally, 3β-[N-(N’,N’-dimethylaminoethane) carbamoyl](DC)-Cholesterol)/DOPE-based liposomes in a [1:2] ratio achieved high encapsulation efficiency of enhanced green fluorescent protein (eGFP) mRNA, along with high eGFP expression in vitro [106]. Furthermore, an additional multi-component LNP displayed a tumor-suppressant effect when loaded with herpes simplex virus I (HSV I) thymidine kinase encoding mRNA. The LNPs were composed of DOTAP/Cholesterol [1:1] liposomes along with 1,2-distearoyl-phosphatidylethanolamine (DSPE)-polyethylene glycol (PEG) and DSPE-PEG-anisamide (AA) [142]. The principle behind their effectiveness may be summarized as a combination of its electrostatic interactions attributable to opposite charges and hydrophobic interactions with mRNA. Additionally, the endosomal escape capabilities and self-assembling properties resulting in uniform layers enclosing polymeric cores also contribute to the wide application of cationic lipids [143]. However, in vivo studies are more challenging due to the fast elimination of cationic lipids by the mononuclear phagocytic system [144]. Cationic lipids consisting of only one quaternary ammonium headgroup pose safety issues such as toxicity and immunogenicity in vitro [145] and in vivo [146]. For instance, cationic liposomes when administered via the intravenous route may induce hepatotoxicity [147] and can trigger a strong IFN-γ response in mice resulting in inflammation [148,149]. Furthermore, positively-charged lipids, e.g., DOTAP and DOTMA, can be neutralized by anionic serum proteins, leading to toxicity and reduced efficacy [150]. Moreover, challenges like unrestricted protein binding, colloidal instability, and drug leakage may arise [151].

Alternatively, new gene delivery vectors containing ionizable lipids [152] and lipid-like materials termed lipidoids [153] have been introduced to overcome challenges posed by conventional cationic lipids while retaining their advantageous transfection properties. Ionizable lipids for mRNA transfection are positively-charged at low pH (which aids in mRNA complexation when it is carried out in acidic buffer) but are neutral at physiological pH (for reduced toxicity post-injection) [154]. Unlike conventional cationic lipids, these lipidoids display a series of secondary and tertiary amines allowing for more efficient interactions with mRNA without remarkably increasing the overall charge of the delivery system [155]. Encapsulation of mRNA in nanoparticles serves to physically protect nucleic acids from degradation and depending on the specific chemistry can aid in cellular uptake and endosomal escape [156]. The combination of the ionizable lipid C12-200, cholesterol, DOPE and C14-PEG2000 at a 3.5:4.65:1.6:0.25 molar ratio, respectively, encapsulating erythropoietin mRNA (EPO-mRNA) displayed high efficacy in vivo when injected into mice, measured as the cellular expression of EPO [157]. The emphasis on the nanoparticle platform for mRNA delivery is in part due to the application of established DNA and siRNA delivery systems.

4.2.5. Lipid-Polymer Hybrid Nanoparticles

Lipid-polymer hybrid nanoparticles (LPNs) have been demonstrated previously to exhibit effective functional delivery of siRNA in vitro [158] and therapeutic delivery of siRNA in vitro and in vivo [159]. This hybrid delivery system has also shown promising results for delivery of mRNA, with the mRNA being encapsulated in a hybrid nanoparticle composed of the lipid-like material N1, N3, N5-tris(2-aminoethyl)benzene-1,3,5-tricarboxamide (TT) in TT3:DOPE:Cholesterol:DMG-PEG2000 (1,2-dimyristoyl-sn-glycerol, methoxypolyethylene glycol) with a polymeric PLGA core [160]. In addition, optimized LPNs consisting of the degradable polymer PBAE, formulated with PEG-Lipid C14-2000, showed successful delivery of mRNA to the lungs [161]. This, along with reported co-delivery of siRNA and mRNA with lipidoid polymer hybrid nanoparticles [162], shows that LPNs are an emerging nucleic acid delivery system. The hybrid formulation is thermodynamically favorable, with respect to hydrophobic, van der Waal, and electrostatic interactions [80]. Several lipids and polymers have been investigated to formulate stable nucleic acid lipid particles using this delivery system. Most common polymers employed are PLGA, polycaprolactone, polylactic acid, or their combinations, whereas the lipids used include DOTAP, 1,2-dilauroyl-sn-glycero-3-phosphocholine, 1,2-distearoyl-sn-glycero-3-phosphocholine, lecithin, DSPE, and PEG, among others. Structurally, based on small-angle X-ray scattering and cryogenic transmission electron microscopy, these nanoparticles are suggested to entail a polymeric matrix core with lamellar lipid structures with the nucleic acid localized in the core and in the corona [163].

4.2.6. Peptide-Based Vectors

Peptide-based systems for mRNA delivery are gaining momentum due to the versatility peptides can offer. Peptide-based delivery systems, both alone and in combination with other materials such as polymers, have been reported in the literature. In a study concerning ovarian cancer therapy, the commercially available cell-penetrating peptide (CPP) PepFect14 was complexed with eGFP mRNA via attractive electrostatic interactions [164]. This nanoparticulate formulation was more efficient in transfecting eGFP mRNA into cells associated with ovarian cancer than commercially available lipofectamine MessengerMAX. Similarly, the CPP RALA has been used to effectively deliver both eGFP and OVA mRNA and has been demonstrated to outperform the cationic lipid DOTAP and the fusogenic lipid DOPE [165]. However, current limitations include targeted cell delivery [164] and short circulation half-life due to the low stability in serum-containing medium [166].

Recently, a novel polymer-peptide hybrid mRNA delivery nanoplatform was introduced [167] combining both polymer (PLA) based micelles and a cationic fusogenic peptide (RALA) to achieve appropriate degradability, mRNA stability, and endosomolytic properties for translation. It was reported to protect eGFP as well as FLuc mRNA against serum nuclease degradation and achieve DC transfection. Indeed, peptide-based vectors and hybrids are promising and interesting additions to the various existing non-viral carriers for the delivery of mRNA.