Alphavirus replicase and regulatory RNA elements in host interactions and viral vector engineering Authors: Dan T. Boghici, Danni Yong, Silvia M. Vidal, GuanQun Liu Published: 2026 ABSTRACT ABSTRACTAlphavirus-derived vector systems have been used for heterologous gene expression for nearly four decades, providing foundational insights into RNA virus replication, host interactions, and genome engineering. The recent success of conventional mRNA vaccines has intensified interest in leveraging the alphavirus replicase and its associated conserved sequence elements (CSEs) to develop self-amplifying mRNA (sa-mRNA) platforms characterized by more sustained antigen expression and dose-sparing. In this minireview, we summarize key advances that define the coordinated functions of the nonstructural proteins and CSEs in viral RNA synthesis and host modulation. We then contextualize the developmental milestones of alphavirus replicons that underpin modern sa-mRNA technologies. We further discuss emerging strategies to engineer replicase functions and RNA architecture in different sa-mRNA applications, while outlining critical gaps in alphavirus biology that currently constrain rational sa-mRNA design. Renewed investigation of nonstructural proteins and CSEs will accelerate optimization of next-generation sa-mRNA platforms and reinvigorate fundamental studies of alphavirus replication, evolution, and host adaptation. INTRODUCTION The Alphavirus genus comprises ~32 enveloped, positive-sense RNA viruses with ~11.5–12 kb genomes that are 5′ capped and 3′ polyadenylated (1). These viruses are maintained in enzootic and epizootic transmission cycles between mosquitoes and diverse vertebrate hosts, including non-human primates, birds, and fish (2). Sustained urban transmission requires close human-mosquito cohabitation, sufficient viremia in infected individuals to permit human-to-mosquito transmission, and efficient viral replication in the mosquito midgut following a blood meal (3). Based on disease phenotype and geographic origin, alphaviruses are broadly classified into Old World and New World clades, which predominantly cause arthritogenic and encephalitic disease, respectively (4). Early foundational studies in alphavirus biology were largely based on the mildly pathogenic Old World viruses, Sindbis virus (SINV) and Semliki Forest virus (SFV). Following several recurrent epidemics in Southeast Asia, Africa, and South America since 2004, chikungunya virus (CHIKV) has emerged as the most clinically relevant alphavirus due to its substantial public health impact (5–8). Acute CHIKV infection typically presents as a febrile illness with rash, arthralgia, and myalgia, with peak viremia occurring 2–3 days post-infection and viral clearance within approximately one week. However, chronic and debilitating musculoskeletal symptoms can persist for months to years (9). Given the nonspecific clinical presentation and co-circulation with other arboviruses, CHIKV infections are likely underreported. In the absence of widespread prophylaxis, epidemic regions are estimated to experience recurrent outbreaks approximately every 6 years, with roughly 8% of the population remaining susceptible (10). A deeper understanding of the alphavirus infection cycle and host interactions is thus needed to define the determinants of viral emergence, disease severity, persistence, and evolving neurotropism (11). Beyond their clinical relevance, alphaviruses are notable for their remarkable genomic plasticity. Replacement of the structural gene cassette with heterologous sequences enables robust and sustained transgene expression in the absence of infectious particle production. These alphavirus replicon systems form the molecular basis of self-amplifying mRNA (sa-mRNA) vaccines and therapeutics, a platform now largely advanced using backbones derived from the New World virus, Venezuelan equine encephalitis virus (VEEV). However, encephalitic alphaviruses such as VEEV and eastern equine encephalitis virus (EEEV) are emerging threats to human health and have demonstrated potential for aerosol transmission, which can result in more severe disease than mosquito-borne infection (12). Consequently, the risks of laboratory-acquired infection and persisting concerns regarding the potential misuse of New World alphaviruses have led to their classification as biosafety level 3 pathogens. In this minireview, we summarize the functions of alphavirus nonstructural replicase proteins and conserved regulatory RNA elements that direct viral RNA synthesis and host interactions, and discuss how these components can be engineered for emerging sa-mRNA applications. We also review the historical development of alphavirus-based expression systems and highlight key knowledge gaps in alphavirus biology that currently limit the rational design of improved sa-mRNA backbones. ALPHAVIRUS GENOME ORGANIZATION Alphavirus genomes encode two open reading frames that express four nonstructural proteins (nsP1–4) and six structural proteins (C–E3–E2–6k/TF–E1) (13) (Fig. 1A). The 5′ open reading frame (~7 kb) encodes the viral nonstructural polyprotein, which is translated primarily as P123 and through readthrough of an opal stop codon (UGA) at approximately 10%, as P1234 (14). The viral protease nsP2 mediates proteolytic processing of the polyprotein to yield the mature nonstructural proteins nsP1–4, which collectively orchestrate viral RNA replication and transcription while also modulating host responses such as antiviral innate immunity and apoptosis in vertebrate cells (15). The intact P123 precursor is required for the synthesis of full-length negative-strand RNA, which is directed by the virally encoded RNA-dependent RNA polymerase (RdRp), nsP4. Subsequent cleavage events release nsP1–3 and enable transition to positive-strand genomic and subgenomic RNA synthesis (16–18) (Fig. 1B). Alphavirus genome organization, replication, and proviral host interactions. (A) The positive-sense alphavirus genomic RNA (+gRNA) contains 5′ and 3′ untranslated regions (UTRs), a 3′ poly(A) tail, and two polyprotein open reading frames (ORF1 and ORF2) separated by an intergenic region that overlaps with the 26S subgenomic promoter. Following cap-0-dependent translation of ORF1, the nonstructural polyproteins P123 and, via readthrough at the opal stop codon, P1234 are produced. Proteolytic cleavage by nsP2 at the nsP3|4 junction releases nsP4, which, together with P123, localizes to host membranes and initiates formation of the viral replication organelle known as the spherule. (B) At the spherule, the P123–nsP4 complex mediates synthesis of the antigenome (−gRNA). Subsequent processing of P123 into individual nsPs directs production of +gRNA and subgenomic RNA (sgRNA), the latter serving as the template for translation of structural polyproteins (C, E3, E2, 6K/TF, and E1). At the base of the spherule, nsP1 assembles into a dodecameric pore complex that caps and exports newly synthesized RNAs. nsP3 oligomerizes into a scaffold adjacent to the nsP1 pore, contributing to RNA sorting and structural integrity of the replication organelle while serving as a hub for proviral host interactions. Key host factors include stress granule proteins G3BP1/2 (Old World alphaviruses) and FXR1/2 (New World alphaviruses), as well as amphiphysin (AMPH), which promote spherule assembly. The LIM domain protein FHL1 supports replication of CHIKV and ONNV in muscle and joint tissues, while the host RNA helicase DHX9 contributes to viral RNA synthesis. Phosphorylation of nsP3 by IKKβ enhances negative-strand RNA synthesis and virion assembly. An internal 26S promoter on the antigenome directs synthesis of the subgenomic RNA, from which the structural proteins (capsid, E3, E2, 6K/TF, and E1) are translated as two polyproteins. The transframe (TF) protein is generated through −1 ribosomal frameshifting within the 6K coding region, producing an alternative structural polyprotein that terminates upstream of E1. Capsid undergoes rapid autocleavage and remains cytosolic, while E3 directs the remaining polyprotein to the secretory pathway, through which E1 and E2 are glycosylated (19–21). E3 chaperones E2 as the p62 precursor, shielding the E1 fusion loop until furin-mediated cleavage generates mature, fusion-competent glycoprotein spikes (22, 23). In addition, 6K and TF promote efficient budding and virion assembly, and TF has been further implicated in antagonism of the host type I interferon (IFN) response (24–26). Mature virions consist of an enveloped nucleocapsid displaying 240 E1–E2 heterodimers arranged into 80 trimeric spikes that mediate receptor-dependent entry and pH-triggered membrane fusion (27–30). ALPHAVIRUS NSP1 AND NSP2 COORDINATE REPLICATION AND IMMUNE EVASION nsP1 is a multifunctional membrane-associated protein that anchors the viral replication machinery to host membranes. Its amphipathic helix inserts into anionic lipid domains at the plasma membrane to drive the formation of bulb-like evaginations known as spherules (31, 32) (Fig. 1B), with species-specific requirements for cholesterol-rich microdomains that more greatly affect CHIKV than SINV (33). At the spherule neck, nsP1 assembles into a dodecameric pore complex that exports newly synthesized viral RNAs to the cytosol (34). These membrane-bound compartments sequester double-stranded RNA (dsRNA) replication intermediates from innate immune sensors such as RIG-I and MDA5, which nevertheless contribute to IFN induction during alphavirus infection (35–37). However, the precise ligands responsible for activating these sensors remain incompletely defined. Recent studies further indicate that alphavirus replicases, particularly those of SFV and SINV, can generate immunostimulatory dsRNA when expressed in isolation, including species derived from host RNA templates. These RNA byproducts are potent activators of RIG-I and MDA5 in immunocompetent cells (38, 39). Thus, although replication organelles provide partial immune shielding, efficient replication requires active modulation of innate immune signaling pathways. For example, CHIKV nsP1 directly engages a cytosolic loop of STING, dampening cGAS-mediated IFN signaling while stabilizing nsP1 expression in infected cells (40) (Fig. 2). Antiviral host interactions and viral antagonism during alphavirus replication. Alphavirus infection activates RNA sensors RIG-I and MDA5, leading to type I interferon (IFN) production. Downstream IFN-stimulated genes (ISGs), such as IFIT1, restrict cap-0-dependent translation of viral RNAs. In parallel, cellular stress responses triggered during replication activate the PERK–eIF2α pathway, suppressing translation of genomic RNA (+gRNA). Mono-ADP-ribosylation of nsP2 by PARP10 inhibits its protease activity. To overcome these restrictions, alphavirus nsPs deploy multiple antagonistic mechanisms. The extreme 5′ terminus of +gRNA adopts a conserved stem-loop structure that functionally substitutes for cap-1 modification and limits IFIT1 binding. nsP2, a central antagonist of host responses, translocates to the nucleus to promote ubiquitin-dependent degradation of RNA polymerase II (Pol II) and nuclear export of STAT1, thereby inhibiting IFN signaling. In both Old and New World viruses, nsP2 elevates intracellular cAMP, activating protein kinase A (PKA) and promoting eEF2 phosphorylation to suppress host translation. nsP4 counteracts PERK-mediated translational arrest to sustain viral protein synthesis, while the hydrolase activity of the nsP3 macrodomain reverses PARP10-mediated ADP-ribosylation of nsP2. In addition, CHIKV nsP1 interacts with STING at the endoplasmic reticulum (ER) to suppress cGAS-mediated IFN induction and stabilize nsP1. CHIKV nsP2 impairs antigen presentation by downregulating peptide loading onto major histocompatibility complex class I (MHC-I) molecules. P, phosphorylation; U, ubiquitination. In addition to its role in spherule formation, nsP1 possesses methyltransferase and guanylyltransferase activities that, together with the nsP2 triphosphatase function, generate a cap-0 structure on positive-sense viral RNAs (41–44). Of note, a fraction of genomic RNAs remain uncapped. Enhanced capping efficiency, as observed for a SINV nsP1 D355A mutant, reduces viral infectivity and particle assembly through mechanisms that remain incompletely understood and appear independent of RNA synthesis (45–47). Recent work has further revealed that nsP1, despite lacking canonical Nudix hydrolase motifs or a phosphatase domain, exhibits de-capping activity toward viral RNAs mediated by reversible guanylyl transfer. CHIKV nsP1 can also remove 5′ caps from host mRNAs, suggesting a direct role in host translational shutoff (48, 49). In contrast to many positive-sense RNA viruses whose RNAs carry a cap-1 structure with 2′-O-methylation that mimics cellular mRNA, alphaviral RNAs lack this modification and are thus susceptible to sensing by RIG-I and restriction by specific IFN-stimulated genes (ISGs), particularly IFIT1 (50, 51). To counteract this, the extreme 5′ ends of genomic and subgenomic alphavirus RNAs adopt a conserved stem-loop structure that functionally substitutes for cap-1 modification by limiting IFIT1 binding and permitting efficient translation (52, 53) (Fig. 2). nsP2 plays central roles in polyprotein processing, RNA replication, and immune evasion. Its protease activity mediates cleavage of the nonstructural polyprotein precursors, releasing nsP4 and coordinating the transition from negative- to positive-strand RNA synthesis through sequential P1|23 and P2|3 cleavage events (54, 55). The amino-terminal region of nsP2 contains a superfamily 1 helicase domain whose RNA unwinding activity requires the presence of intact carboxyl-terminal protease and methyltransferase-like (MTL) domains. These domains bind strongly to viral RNA and direct helicase-domain binding to RNA through a flexible interdomain linker region (56, 57). Disruption of conserved helicase–RNA stacking interactions, exemplified by the F287A mutation, impairs subgenomic RNA synthesis (58, 59). The coordination of polyprotein processing by nsP2 also determines the accumulation and function of specific precursor intermediates, such as P23, which serves as a critical regulator of spherule formation and viral RNA synthesis (59, 60) (Fig. 1). Structural and biochemical studies on P23 reveal an extended RNA-binding interface spanning the nsP2 protease and the nsP3 alphavirus unique domain (AUD) and that the P2|3 cleavage site is occluded in cis, rendering its processing highly regulated (59). Old World nsP2 is a potent suppressor of type I IFN responses and a major determinant of cytopathogenicity (61, 62). Cleavage-defective mutants that fail to generate free nsP2 can produce near-wild-type levels of positive-strand RNA in IFN-deficient baby hamster kidney (BHK-21) cells but are strongly restricted in cells with intact type I IFN signaling, highlighting the dual replicative and immunomodulatory roles of nsP2 (63). Mechanistically, nsP2 traffics to the cell nucleus and promotes the ubiquitin-dependent proteasomal degradation of Rpb1, the catalytic subunit of RNA polymerase II, resulting in transcriptional shutoff (64) (Fig. 2). Encephalitic alphaviruses achieve host transcriptional shutoff through the capsid protein, which forms tetrameric complexes with nuclear import and export proteins to block nuclear pores and impede the trafficking of host transcriptional factors (65). As part of the host antiviral response, ADP-ribosylation of nsP2 by host poly(ADP-ribose) polymerase PARP10 suppresses its protease activity, thereby disrupting polyprotein processing and impairing viral replication (66–68). However, this restriction is counteracted by the N-terminal macrodomain of nsP3, which possesses ADP-ribose hydrolase activity (Fig. 2). Consistent with this, mutations that reduce macrodomain hydrolase function in CHIKV nsP3 impair replicase activity and virus viability in proportion to the degree of enzymatic loss, although it remains unclear whether these defects arise solely from impaired nsP2 function or reflect broader effects on host ADP-ribosylated targets involved in replication (66, 69). In addition, the MTL domain of CHIKV nsP2 is implicated in downregulating peptide loading onto MHC class I via an unknown mechanism, thereby promoting viral persistence in joint and muscle tissues (70) (Fig. 2). HOST INTERACTIONS MEDIATED BY NSP3 ARE HIGHLY DIVERGENT The precise role of nsP3 in alphavirus replication remains incompletely understood, particularly due to its modular architecture and extensive functional plasticity. In addition to the essential roles of the macrodomain described above (60, 66, 69), the C-terminal hypervariable domain (HVD) serves as a hub for host protein recruitment but exhibits striking sequence divergence across the genus (71–74). This divergence is a major determinant of species- and cell type-specific host interactions and contributes to differences in tropism and pathogenesis (75, 76). For example, a CHIKV chimera expressing nsP3 from its closest relative, o’nyong-nyong virus (ONNV), enabled productive infection of Anopheles mosquitoes, a natural vector of ONNV, demonstrating that nsP3 is a critical determinant of vector specificity (77). The HVDs of both Old World and New World nsP3 sequester distinct components of stress granules via conserved short linear motifs that bind G3BP1/2 in Old World viruses and FXR family proteins (FXR1/2 and FMR1) in New World viruses (78–80) (Fig. 1B). In CHIKV and Semliki Forest virus (SFV), duplicated FGDF motifs (e.g., 449/466L/ITFGDFD455/472) constitute the primary G3BP-binding sites, with an additional arginine-rich region (398PVAPPRRRR406) proposed to contribute to this interaction (80, 81). Genetic ablation of G3BP is lethal for CHIKV and strongly attenuates SFV, indicating that these interactions are not merely antagonistic to stress granule formation but are essential for efficient viral RNA replication (82, 83). Additional proviral HVD interactors include amphiphysin-1/2, which bind proline-rich SH3 motifs in the nsP3 HVD of SFV, SINV, and CHIKV (84), and the cytoskeletal adaptors CD2AP and SH3KBP1, recruited by HVDs of multiple Old World and New World alphaviruses to organize the replication complex (85). The nsP3 HVD of CHIKV also engages the RNA helicase DHX9, which enhances early nonstructural protein translation but is later selectively degraded, relieving inhibition of RNA synthesis and promoting the transition from translation to genome replication (86). nsP3 also engages tissue-specific host factors to shape tropism and pathogenesis. Interaction with four-and-a-half LIM domain protein 1 (FHL1), a regulator of myocyte differentiation, is critical for CHIKV and ONNV replication within muscle and joint tissues (87–89) (Fig. 1B). Notably, the West African genotype of CHIKV, which is associated primarily with enzootic transmission, exhibits reduced FHL1 binding compared with the more virulent Asian and East/Central/South African lineages, linking strain-specific nsP3-host interactions to pathogenicity (88). Conversely, FHL1 is not required for other arthritogenic viruses such as Mayaro virus (MAYV) and Ross River virus (RRV). Both viruses caused wild-type-like disease in Fhl1−/− C57BL/6 mice (89). More broadly, the extreme sequence variability of the HVD permits tolerance of mutations and large deletions, as exemplified by the successful experimental insertion of fluorescent reporter genes and by the 62-amino-acid C-terminal truncation that constitutes the principal attenuating lesion in the live-attenuated IXCHIQ vaccine strain of CHIKV (90–92). Additional motifs within the HVD, such as a conserved YXXM sequence, have been implicated in metabolic reprogramming and enhanced virulence of RRV (93). The central region of nsP3, termed the AUD, is structurally conserved and oligomerizes into a helical scaffold at the spherule neck juxtaposed to the nsP1 pore (Fig. 1B). This structure associates with nascent RNA and may direct genomic and subgenomic RNAs toward translation or virion assembly (94, 95). The AUD contains a cysteine-coordinated zinc-binding motif. Mutations within this motif, as well as in other conserved residues, impair or abolish viral replication. Whether these defects arise from disruption of helical scaffold assembly itself, however, remains unclear (96). The opal stop codon at the nsP3|4 junction regulates the P123/P1234 ratio and is essential for proper polyprotein processing and spherule stability in vertebrate cells. In addition, it contributes to evasion of antiviral RNA interference in mosquito vectors (76, 97–100). Replacement of the opal codon with amber or ochre stop codons imposes more stringent translational termination and reduces readthrough efficiency at 37°C. Conversely, substitution with sense codons promotes polyprotein processing and attenuates viral replication in vertebrate cells, whereas such modifications are generally better tolerated in mosquito cells (101). Circulating SFV and ONNV strains naturally encode arginine or cysteine in place of the opal stop codon, a feature proposed to favor controlled replication in mosquitoes and potentially limit their pandemic potential in humans (102). Finally, post-translational modification of nsP3 also contributes to its multifunctionality. In VEEV, IKKβ-mediated phosphorylation of nsP3 has been shown to promote negative-strand RNA synthesis and virion assembly (103, 104) (Fig. 1B). NSP4 STABILITY AND REPLICATION FIDELITY The viral RdRp nsP4 is inactive and highly unstable when expressed in the absence of P123, owing to a conserved N-terminal tyrosine that targets the protein for ubiquitin-dependent N-end rule degradation (105–108). Substitution of this residue with a primary methionine (Y1M) or with aromatic or histidine residues stabilizes nsP4 but attenuates replication. Except for its first ~100 amino acids, nsP4 is the most conserved component of the alphavirus replicase and is functionally interchangeable across the genus. Many replicase chimeras composed of the parental P123 precursor and heterologous nsP4, including that from the insect-specific Eilat virus, remain competent for viral RNA synthesis (109). In addition to its polymerase activity, nsP4 exhibits terminal adenylyltransferase activity, suggesting a role in poly(A) tail synthesis and repair (110). Mutations that target nsP4 fidelity modulate viral replication dynamics. A C483G substitution reduces polymerase fidelity and increases mutagenesis and template switching, thereby promoting the accumulation of truncated, defective interfering (DI) genomes (111, 112). Conversely, the C483Y substitution has been proposed to generate a high-fidelity variant of CHIKV nsP4 that maintains near-wild-type levels of genomic RNA synthesis but displays context-dependent attenuation in vivo. In susceptible neonatal mice infected subcutaneously, the C483Y mutant virus caused reduced disease, presumably reflecting stronger immune control of a more genetically constrained viral quasispecies (113). However, immunocompetent adult mice challenged in the footpad with the same mutant virus unexpectedly developed more severe arthritis and myalgia. These discordant outcomes have led subsequent studies to question whether C483Y reliably confers a stable high-fidelity phenotype (114, 115). HISTORY OF ALPHAVIRUS-DERIVED EXPRESSION SYSTEMS Conserved sequence elements (CSEs) refer to the conserved features of the viral genome that are essential for RNA replication, transcription, packaging, and assembly. Modern recombinant alphavirus replicons exploit these elements in the context of native viral genomes and replace the structural gene cassette with a heterologous gene of interest (GOI) expressed under control of the 26S subgenomic promoter. These systems retain the ability to form membrane-associated replication complexes and to generate (sub)genomic RNAs, but are incapable of producing infectious virus particles because they lack the essential structural proteins. The conceptual basis of alphavirus replicon technology emerged from early studies of DI genomes that sought to identify the minimal CSEs required for replication and packaging. When structural proteins are supplied in trans, DI RNAs or full-length replicons can be selectively encapsidated into single-cycle viral replicon particles (VRPs). Early recombinant alphavirus expression systems were developed using SINV DI genomes. In its first published use, a ~2.2 kb DI genome was cloned into a plasmid under the control of an SP6 bacteriophage promoter. This cDNA was transcribed and delivered into chicken embryonic fibroblasts together with a replication-incompetent SINV helper virus (116). Systematic deletions in the DI cDNA subsequently defined a minimal set of essential CSEs, including a 162-nt region at the 5′ terminus comprising the 5′ untranslated region (CSE1) and a 51-nt stem-loop-rich region within nsP1 (CSE2), as well as a 19-nt AU-rich region from the 3′ end of the genome (CSE4). Together with the 26S subgenomic promoter (CSE3), these elements were shown to be collectively required for efficient RNA amplification and production of infectious particles (117) (Fig. 3). Conserved sequence elements and regulatory RNA structures in alphavirus genomes. Viral RNA replication is governed by four conserved sequence elements (CSEs) distributed along the genome. At the 5′ end, CSE1 forms a stem-loop structure that functions as the promoter for the synthesis of full-length genomic RNA from the negative-strand template in coordination with CSE2. Immediately downstream, the 51-nt CSE2 within the nsP1 coding region supports efficient positive- and negative-strand RNA synthesis. The subgenomic promoter (CSE3), spanning from the C-terminus of nsP4 to the intergenic region (−98 to + 14 in SINV), directs transcription of subgenomic RNA. At the 3′ end, CSE4, a conserved 19-nt AU-rich sequence (AUUUUGUUUUUAAUAUUU) located immediately upstream of the poly(A) tail, acts as the promoter for antigenome synthesis, likely through long-range RNA interactions with CSE2. Several additional RNA elements contribute to viral replication and gene expression. Packaging signal loops (PS; arrowheads), located within nsP1 or nsP2 coding regions depending on the virus, are recognized by the capsid protein to ensure selective incorporation of full-length genomic RNA into virions. The nsP3|4 junction frequently contains an opal stop codon (UGA) embedded within a stem-loop structure that regulates readthrough and production of the P1234 polyprotein. In SINV and SFV, a translation-enhancing downstream loop (DLP) within the N-terminal 34 amino acids of capsid promotes efficient subgenomic RNA translation. A −1 ribosomal frameshifting element within the structural region drives expression of the transframe protein (TF). Repeated sequence elements (RSEs) within the 3′ UTR vary in number among alphaviruses and influence RNA stability, replication efficiency, and host adaptation. The substitution of native viral sequences from a functional DI genome with a heterologous reporter gene (chloramphenicol acetyltransferase, CAT) demonstrated that these elements were sufficient to drive foreign gene expression, establishing the core design principles of alphavirus replicon vectors (118). Breakthroughs in cloning the entire SINV genome enabled the construction of the first full-length alphavirus replicon system, which achieved robust CAT expression under control of the viral 26S subgenomic promoter across a wide range of vertebrate and invertebrate cells (119, 120). However, early systems were limited by low transfection efficiency and recombination between replicon and helper RNAs (121), which were later alleviated by electroporation-based transfection and by bipartite helper systems in which capsid and glycoprotein genes were expressed from separate RNAs to minimize recombination (122–124). Further refinements stemmed from the observation that the N-terminal 30–34 amino acids of the SINV and SFV capsid proteins function as translational enhancers for subgenomic RNA expression (124–126) (Fig. 3). Incorporation of this enhancer upstream of the SFV glycoprotein cassette, linked through a foot-and-mouth disease virus ribosomal skipping 2A peptide, enabled native-like expression levels of the separated structural proteins in helper systems (124). Similarly, in-frame N-terminal fusion of full-length capsid to a GOI enhanced fluorescent reporter expression in SINV and CHIKV backbones (127). Importantly, the intrinsic C-terminal autoproteolytic activity of capsid ensures efficient release of the downstream protein, thereby preserving proper folding and intracellular localization of the GOI product. Subsequent studies using VRPs elucidated the molecular basis of alphavirus RNA packaging. It was found that capsid selectively recognizes conserved stem-loop structures containing GGG trinucleotide motifs, which are distributed within the nsP1 (e.g., SINV, VEEV) or nsP2 (e.g., CHIKV, SFV) coding regions rather than being confined to a single, discrete packaging signal (128–130). Importantly, capsid can indiscriminately recognize GGG-containing stem-loops for packaging. Chimeric alphaviral RNAs encoding VEEV nsPs can be encapsidated by CHIKV capsid proteins to produce viable, albeit attenuated, viruses (131). Such chimeric viruses represent promising candidates for the development of live-attenuated vaccines. Given the malleability of the alphavirus genome and the naturally low virulence of SINV in humans, dual-subgenomic promoter viruses had been proposed as candidate live-attenuated vaccines. The duplication of the 26S promoter within full-length viral genomes enabled simultaneous expression of structural proteins and a GOI from two subgenomic RNA molecules (132). A shift to New World dual-promoter systems emerged upon the observation that the more pathogenic VEEV is naturally lymphotropic in C57BL/6 mice and Syrian golden hamsters via efficient infection of dermal dendritic cells that traffic to draining lymph nodes (133, 134). The construction of avirulent derivatives for dual-promoter vaccine design leveraged targeted glycoprotein point mutations that markedly reduced viremia and neurotropism while restricting replication in lymphoid tissues (135, 136). While these systems proved immunogenic and protective in animal models, they were ultimately constrained by polymerase template switching, accumulation of truncated genomes due to structural limits on packaging oversized genomic RNA, and poor nsP4 proofreading during replication of a large subgenomic payload (136, 137). Although dual-promoter alphavirus vectors have fallen out of favor, attenuated VEEV-based platforms have maintained considerable interest as vaccine vectors. A VEE-VRP system protected BALB/c mice against influenza challenge while limiting recombination through bipartite defective-helper RNAs (123). Sequential immunization experiments demonstrated limited vector immunity, even following prior exposure to the homologous replicon backbone. Subsequent use of lymphotropic SINV bipartite helper RNAs enabled VEE-VRP production outside biosafety level 3 containment (138, 139). Both wild-type and chimeric VEE-VRPs produced higher transgene expression and cytotoxic T cell responses than reciprocal SIN-VRPs, highlighting the strength of New World alphavirus expression systems (139). Furthermore, unlike SINV and SFV systems, high-level transgene expression from VEEV replicons does not require a subgenomic capsid enhancer. Despite these advances, scalable manufacturing of alphavirus VRPs remains a practical limitation, even with the use of stable packaging cell lines (140). FROM CONVENTIONAL MRNA VACCINES TO SELF-AMPLIFYING RNA PLATFORMS The COVID-19 pandemic catalyzed the first large-scale use of conventional mRNA vaccines against SARS-CoV-2. These vaccines consist of capped, in vitro-transcribed mRNAs encoding viral antigens that are delivered using lipid nanoparticle (LNP) formulations (141). Although highly effective at inducing protective immunity, conventional mRNA vaccines are limited by the intrinsic instability and short intracellular half-life of mRNA, which constrain the duration of antigen expression. This transient antigen production has been proposed as one factor contributing to the limited durability of immune memory and the need for periodic booster immunizations (142). A logical extension has thus been the development of sa-mRNA platforms that encode both the antigen and a viral replicase, allowing RNA amplification and prolonged antigen expression. Alphavirus replicon systems, given their extensive studies over decades, provided the conceptual and molecular foundation for a dose-sparing approach in vaccine prophylaxis (13, 143–148). Among available backbones, VEEV replicons have emerged as the dominant platform, owing to their pronounced lymphotropism, relatively low vector immunity, strong innate immune activation, and robust heterologous gene expression compared with Old World alphaviruses such as SINV or SFV (149). Consistent with this trend, all licensed sa-mRNA vaccines (KOSTAIVE and GEMCOVAC) and publicly described alphavirus-derived sa-mRNA candidates currently in FDA clinical trials for SARS-CoV-2, influenza, or rabies employ VEEV replicase backbones (150). However, despite this progress and extensive prior work on alphavirus expression systems, key aspects of nonstructural protein function and CSEs in the context of sa-mRNA remain incompletely understood. HOST CYTOTOXIC AND STRESS RESPONSES ELICITED BY ALPHAVIRUS REPLICONS Alphavirus replicons are often reported to achieve very high intracellular RNA copy numbers, although quantitative estimates vary and are frequently poorly substantiated (151–153). Regardless of amplification magnitude, replicon activity induces substantial cellular stress even in the absence of structural genes, making control of cytopathic effects (CPE) essential for sustained antigen expression, efficient antigen presentation, and reduced vaccine reactogenicity. Most contemporary VEEV-based sa-mRNA platforms are derived from the TC-83 live-attenuated vaccine strain, generated by serial passage of the Trinidad Donkey (TRD) strain in guinea pig heart cells (154). In packaging-deficient TC-83 replicons, attenuation is largely attributed to a G-to-A substitution at position 3 of the 5′ UTR, which alters the local stem-loop secondary structure and generates a longer, more exposed 5′ overhang, thereby enhancing IFIT1-mediated restriction of translation initiation (52, 155, 156) (Fig. 2). Consistent with this, TC-83 displays virulence comparable to wild-type TRD VEEV in type I IFN receptor (IFNAR)-deficient mice, while inducing reduced CPE in IFN-primed mouse fibroblasts (157, 158). Thus, attenuation of TC-83 replicons is context-dependent and driven largely by intact type I IFN signaling. In permissive cell lines such as BHK-21, VEEV and western equine encephalitis virus (WEEV) replicons are intrinsically less cytopathic than SINV replicons, and the TC-83 5′ UTR mutation further reduces CPE (159). Comparative studies across SINV, VEEV, and WEEV dual-promoter replicons further revealed that non-cytopathic variants can be selected through adaptive mutation, often converging on substitutions within the C-terminal MTL domain of nsP2 (160–162). A conserved proline residue at position 726 (P726) in SINV and SFV and at position 718 in the West African genotype of CHIKV appeared to be a key determinant of cytopathogenicity. Subsequent studies demonstrated that enzymatic activities of the nsP2 MTL domain are not strictly required for CPE induction, suggesting instead a dominant role for nsP2-mediated inhibition of host transcription (61, 62). Non-cytopathic nsP2 variants often exhibit disrupted GTPase, helicase, or protease activities and can be maintained in permissive cell lines, although compensatory mutations in other nsPs are frequently required for productive replication in human cells, where isolated nsP2 mutations may otherwise be lethal (160, 162, 163). In SINV replicons, P726 is additionally implicated in host translational shutoff by interfering with PKR-dependent phosphorylation of eIF2α, and substitution of this residue (e.g., P726G) alleviates translational repression (164). Conversely, CHIKV nsP4 appears to suppress PERK-mediated eIF2α phosphorylation early in infection, thereby relieving translational inhibition and promoting viral protein synthesis (165) (Fig. 2). Interestingly, non-cytopathic nsP2 mutations are often associated with hyperactive polyprotein processing, which reduces antigenome synthesis and thereby limits the intracellular pool of nsP2 capable of antagonizing host transcription and translation (160). Despite reduced translation of the nonstructural polyprotein upon eIF2α phosphorylation, Old World alphaviruses retain efficient transgene expression due to a stable RNA hairpin, termed the downstream loop (DLP), located within the capsid translational enhancer (Fig. 3). This element permits robust translation of subgenomic RNA in the absence of eIF2α and eIF4G and can support initiation from non-AUG start codons (166, 167). Consequently, non-cytopathic Old World replicons can sustain sufficient antigen expression for effective processing and presentation. However, the translational advantage conferred by the DLP may depend on the context of active viral replication, as subgenomic RNA delivered by replicon transfection has been observed to preferentially engage canonical translation pathways (168). Given the divergent host shutoff strategies employed by Old World and New World alphaviruses, targeting nsP2 nuclear translocation has emerged as a means to partially restore innate immune signaling and reduce cytopathogenicity for Old World viruses, although these effects can be highly virus- and strain-dependent (169). For example, mutation of the nsP2 nuclear localization signal (NLS) in SFV enhanced type I IFN responses and reduced cytotoxicity when combined with an additional P718T substitution (163, 170). Mutations at conserved proline residues (P718 in CHIKV or P726 in SINV) or within the NLS (e.g., KR649-650AA) improved nuclear import of phosphorylated STAT1 and restored functional IFNAR signaling (171–173) (Fig. 4). It is noted that putative NLS mutations at analogous positions do not necessarily confer a non-cytopathic phenotype across different alphavirus species, or even among strains of the same virus (162, 174). Later work on developing non-cytopathic Old World replicons identified a weakly conserved peptide motif (called “VLoop”) within the nsP2 MTL domain, whose mutations reduced nsP2 nuclear localization and subsequent host transcriptional arrest (Fig. 4). These mutations conferred additional attenuation by impairing polyprotein processing in CHIKV, but not in SINV (37, 175). In parallel, a PKR-independent mechanism of translational shutoff has been proposed, in which the nsP2 helicase domain of CHIKV and VEEV elevates intracellular cAMP levels, thereby activating protein kinase A and downstream phosphorylation of EEF2 (176) (Fig. 2). Engineering nonstructural proteins and RNA elements for optimized sa-mRNA backbones. Schematic of an alphavirus-derived sa-mRNA highlighting viral nonstructural proteins (nsP1–4) and conserved sequence elements (CSEs) that can be rationally modified to tune replication, innate immune sensing, cytotoxicity, and transgene expression. The 5′ UTR is a key regulatory hub controlling replicase specificity and the balance between IFIT1 restriction and type I IFN induction. Current VEEV TC-83-based sa-mRNA production uses CleanCap AU (cap-1) to enhance translation and mitigate IFIT1 restriction of incoming RNA, despite increased IFIT1 susceptibility associated with the G3A mutation. The 5′ UTR sequence also influences template selection by heterologous replicases, an emerging strategy in trans-amplifying RNA systems. Mutations that destabilize the 51-nt CSE2 within nsP1 can reduce RNA replication and alleviate cytotoxicity. Likewise, nsP1 variants that enhance capping efficiency or weaken membrane anchoring can modulate replication output and cellular stress. Targeting residues in Old World nsP2 associated with cytopathogenicity can relieve host shutoff and attenuate cytopathic effects. Mutations in the nsP3 macrodomain can shift the balance between genomic replication and subgenomic RNA transcription, potentially enhancing transgene expression, while modifications in the nsP3 hypervariable domain (HVD) can alter host factor recruitment to reduce cytopathogenicity or enable cell- and tissue-specific replication. Variants of nsP4 that alter polymerase stability or fidelity modulate RNA amplification, mutation rates, and DI RNA formation, thereby affecting cytotoxicity, transgene integrity, and expression durability. Replacement of the opal stop codon at the nsP3|4 junction with a sense codon can attenuate replication in vertebrate cells of certain alphaviruses. Incorporation of the capsid-derived translational enhancer sequence into the transgene can increase expression in Old World backbones. Finally, regulatory modules within the 3′ UTR, including riboswitches, tissue-specific microRNA target sites, and optimized poly(A) tail length, can be used to fine-tune expression kinetics and tissue tropism. By contrast, replicons derived from New World alphaviruses such as VEEV generally require fewer genetic modifications to achieve acceptable cytopathic profiles, making them more practical backbones for sa-mRNA applications. Nevertheless, nsP2-mediated translational shutoff is conserved among New World alphaviruses, and the C-terminal region of nsP2 remains a key barrier to further attenuation and optimization (177). In vitro evolution of VEEV TC-83 sa-mRNA under exogenous type I IFN pressure selected mutations in the nsP3 macrodomain that reduce its hydrolase activity and slow replication kinetics (178) (Fig. 4). Although the mechanisms are not fully resolved, these variants exhibit reduced innate immune activation and cytotoxicity with enhanced transgene expression (178), indicating that optimal sa-mRNA performance depends on a finely tuned balance among replicase activity, innate immune sensing, and viral antagonism rather than on maximal RNA amplification alone. CONSERVED SEQUENCE ELEMENTS TUNE SA-MRNA REPLICATION AND EXPRESSION Alphavirus CSEs also provide powerful levers for tuning replicon performance. In addition to enhancing IFIT1-mediated restriction, the G3A mutation in the 5′ UTR of VEEV TC-83 alters competition between the genomic and subgenomic promoters for viral replicase components, thereby reducing subgenomic RNA synthesis (155). The conserved 51-nt element (CSE2) within nsP1, present across alphaviruses, is essential for positive-strand RNA synthesis and functions in concert with the 5′ UTR (179, 180). Consistent with this, and drawing on DI genome architecture, insertion of a 228-nt sequence encompassing the genomic 5′ UTR and CSE2 downstream of the VEEV subgenomic promoter markedly enhances transgene expression across multiple mammalian cell types (181). To prevent fusion of the GOI to the nsP1-derived peptide encoded by CSE2, the design further incorporates self-processing elements, such as a 2A peptide, ubiquitin, or 2A in combination with an NLS-deficient, non-cytopathic capsid sequence, positioned upstream of the transgene (181). These findings delineate alternative engineering strategies for New World alphavirus-based sa-mRNA and VRP platforms. Substantial sequence divergence within the nsP3 HVD and the 3′ UTR reflects natural mechanisms for tuning host interactions and RNA stability (Fig. 4). Viruses in the VEE serogroup harbor relatively short 3′ UTRs (100–200 nt), whereas WEE-complex and SFV-complex viruses contain longer regions of ~200–300 nt and ~300–600 nt, respectively (182, 183). Repeat sequence elements (RSEs) within the 3′ UTR enhance RNA stability in vertebrate cells by recruiting host RNA-binding proteins such as Human antigen R (HuR), which suppresses poly(A) shortening and exonucleolytic decay while also promoting efficient replication in mosquito vectors (184–187) (Fig. 3). From an evolutionary standpoint, recombination events that lengthen the 3′ UTR have been proposed to enhance adaptation of alphaviruses to mosquito vectors during transitions from aquatic hosts and, more recently, to stabilize transmission bottlenecks of epidemic CHIKV strains in Aedes mosquitoes (188, 189). The presence of additional RSEs that exert neutral effects in SINV but detrimental effects in CHIKV on viral replication in mammalian cells warrants further investigation into the balance required for optimal fitness across vertebrate and invertebrate hosts (189, 190). The length and composition of the poly(A) tail further regulate antigenome synthesis. For example, antigenome production in SINV is reduced when the poly(A) tail is shorter than 25 residues and is completely abolished when it falls below 11 residues (191). Consistent with a role for 3′ end architecture in host adaptation, VEEV TC-83 chimeras bearing heterologous 3′ UTRs from epizootic and enzootic subtypes exhibited differential restriction by ISGs, such as IFIT2 (192). ENGINEERING RNA CHEMISTRY AND ARCHITECTURE TO OPTIMIZE SA-MRNA PERFORMANCE Chemical and structural optimization of the RNA template has become a central strategy in sa-mRNA engineering for balancing robust replicase activity with controlled innate immune activation (144). Systematic screening has identified nucleotide modifications compatible with alphavirus RNA replication, including 5-hydroxymethylcytidine (hm5C), 5-methylcytidine (m5C), and 5-methyluridine (m5U), which attenuate innate sensing while preserving RdRp function and efficient RNA amplification (193). The architecture of the poly(A) tail has also been explored as a determinant of RNA stability and translation efficiency. Conventional non-replicating mRNA vaccines are constructed with 60–120 nt poly(A) tails, similar in length to those of SINV and SFV; however, alphavirus negative-strand RNA synthesis is not further improved by increasing tail length from 25 to 34 adenosines (191, 194, 195). Thus, extension of sa-mRNA poly(A) tails beyond the minimally sufficient 25 nucleotides may not be a strong determinant of payload expression. Current sa-mRNA manufacturing further employs cap-1 analogs (e.g., CleanCap AU) to enhance translation and dampen innate sensing during the initial launch phase (144, 150). However, following replication, newly synthesized viral RNAs acquire the natural cap-0 structure and 5′ triphosphate intermediates generated by nsP1 and nsP2, rendering cap-1 modification transient and largely restricted to the input RNA population. Trans-amplifying RNA (taRNA) platforms are bipartite systems that reduce manufacturing barriers associated with encapsulating long sa-mRNA transcripts. In this approach, the replicase is encoded within a non-replicating conventional mRNA, whereas the trans-replicon (TR) adopts a DI-like architecture containing all essential alphaviral CSEs required for self-amplification of the GOI (196). Recent simplification of this design, achieved by removing CSE3 and eliminating all CSE1/2-localized AUG start codons upstream of the GOI start site, further enhanced RNA replication and transgene expression (197). Notably, taRNA systems can exploit the partial cross-compatibility of alphavirus replicase-template pairs, with SFV replicase supporting highly efficient amplification of heterologous SINV-derived TRs in human cells (197, 198). In vitro evolution of such hybrid systems in the presence of competing TRs from different alphavirus species selected variants with enhanced replication kinetics and transgene expression, such as SFV-derived mutants bearing a short 5′ extension (AUAAAA(A)), previously linked to efficient negative-strand initiation (197, 199). These findings highlight the 5′ UTR as a critical regulatory hub that warrants continued investigation (Fig. 4), both for its role in governing replicase-template specificity and for its involvement in balancing IFIT1-mediated restriction and type I IFN induction during alphavirus infection (53, 198). BIODISTRIBUTION AND BIOSAFETY CONSIDERATIONS FOR SA-MRNA PLATFORMS LNP-mRNA vaccines are typically administered intramuscularly, where immune activation arises from antigen expression in muscle and antigen-presenting cells or from LNP drainage to lymph nodes (200). Although LNP composition and size can be tuned to restrict biodistribution, low-level accumulation in off-target tissues, including the liver and spleen, is common (201–203). Given the self-replicative capacity of alphavirus backbones, careful evaluation of in vivo biodistribution remains essential, as even minimal dissemination to sensitive tissues such as the central nervous system (CNS) could be detrimental (204). To date, sa-mRNA has not been detected in the brains of vaccinated mice, possibly reflecting limited penetration of the blood-brain barrier (BBB) by the large LNP-formulated RNA cargo, although studies addressing this issue remain limited (205, 206). Nevertheless, alphavirus biology suggests additional complexity. Infectious virus-like vesicles (VLV) can be generated in the absence of capsid proteins. Here, genomic, subgenomic, or DI RNA may be nonspecifically packaged when in proximity to heterologous viral glycoproteins that drive vesicular budding from the cell surface by a poorly defined mechanism (207, 208). Such VLVs displaying foreign glycoproteins can be propagated in IFN-deficient cells and have been explored as vaccine candidates (209, 210). Although VLVs appear non-pathogenic in immunocompetent animals, their potential emergence following sa-mRNA administration and its impact on biodistribution remain poorly understood. In addition to vesicle-mediated dissemination, direct cell-to-cell RNA transmission may further alter biodistribution. Old World alphaviruses (SFV, SINV, CHIKV), and to a lesser extent VEEV, remodel the actin cytoskeleton through nsP1 to induce filopodia-like extensions that facilitate transfer of infectious entities independently of nucleocapsid assembly (211–214). Mutations that disrupt nsP1 palmitoylation or membrane-associated hydrophobic residues reduce filopodia formation and thus represent useful engineering targets for future sa-mRNA design (33, 215, 216). Consistent with these dissemination mechanisms, multiple, non-exclusive routes of alphavirus CNS invasion have been proposed, including neuronal transport, inflammation-induced BBB disruption, and vesicular transcytosis across endothelial cells. Glycoprotein determinants, particularly within E2, modulate neurotropism, raising theoretical biosafety considerations for sa-mRNA encoding foreign glycoproteins with CNS-targeting potential. Although such risks remain speculative, they warrant continued mechanistic investigation (217–222). Genetic recombination between sa-mRNA and circulating alphaviruses is also theoretically possible, although superinfection exclusion is expected to strongly limit the generation of infectious chimeras (223, 224). In Old World alphaviruses, exclusion is mediated in part by premature nsP2-driven polyprotein processing, whereas in New World viruses, it involves nsP3 HVD-dependent sequestration of host factors required for RNA synthesis (225–227). To improve controllability and safety, drug-inducible sa-mRNAs have been developed (228). Riboswitches inserted into the 5′ and 3′ UTRs of the subgenomic RNA region of VEEV TC-83 confer ligand-dependent regulation in DNA-launched and VRP-packaged replicons (229), although this strategy has yet to be applied to LNP-delivered sa-mRNAs. More broadly, spatial control of RNA distribution remains a central requirement for safe and effective gene therapy. A panel of alphavirus sa-mRNAs encapsulated in the SM-102 LNP formulation (used in the Moderna COVID-19 vaccine) exhibited tissue-specific accumulation patterns in murine organs following intramuscular or intravenous administration (230). Accordingly, incorporation of 3′ UTR microRNA target sites that restrict productive replication in undesired tissues represents a practical strategy to improve targeting precision and reduce off-target effects (231, 232) (Fig. 4). Although VEEV remains the most widely used sa-mRNA backbone, alternative alphavirus replicons are desirable to mitigate anti-backbone immunity and to support repeated or heterologous boosting (233, 234). Old World alphaviruses, with their intrinsic myotropism, warrant renewed consideration, although additional engineering will be required to achieve expression levels comparable to VEEV-based systems (127, 230, 235). Exploration of diverse backbones may also reveal conserved T-cell epitopes that could be harnessed to elicit dual immunity against both alphaviruses and the encoded pathogen, thereby broadening the protective scope of sa-mRNA vaccine platforms (236). CANCER AND GENE THERAPY APPLICATIONS OF SA-MRNA Beyond their application in prophylaxis against infectious diseases, which has been reviewed extensively elsewhere (150), sa-mRNA platforms have been actively explored for cancer immunotherapy and protein replacement applications (Fig. 5). Intramuscular delivery of a single dose of sa-mRNA expressing a chimeric human papillomavirus (HPV) and herpes simplex virus antigen generated a robust cytotoxic T cell response that reduced HPV-associated tumor burden and established a durable memory T cell population capable of preventing subsequent tumor development (237). In mouse-derived dendritic cells, however, HPV antigen expression from sa-mRNA was lower than that observed with modified non-replicating mRNA containing N1-methylpseudouridine, suggesting that (i) more cytopathic or IFN-suppressive alphavirus backbones may be advantageous in certain cancer treatment regimens, and (ii) management of double-stranded RNA intermediates remains a critical determinant of sa-mRNA vaccine performance. While the incorporation of modified bases in self-replicating systems can facilitate evasion of Toll-like receptor sensing, lower and more sustained antigen expression kinetics may, in some contexts, be preferable to the metabolic stress associated with excessive or prolonged immunogen production. Emerging applications for next-generation sa-mRNA systems. The inherent plasticity of the alphavirus genome enables deployment of sa-mRNA platforms across diverse biomedical applications. Expanded use of sa-mRNA for infectious disease prophylaxis should incorporate a broader repertoire of alphavirus backbones to mitigate anti-backbone immunity and enable repeated or heterologous boosting. These alternative backbones may also support the development of trans-amplifying RNA systems capable of accommodating larger payloads, thereby overcoming current limitations in long RNA production and encapsulation. A deeper understanding of antigen processing and presentation in vivo will clarify how sa-mRNA platforms differ from conventional vaccines in shaping immunogenicity and protective efficacy. Oncolytic applications of sa-mRNA involve direct intratumoral administration, where wild-type Old World backbones may be advantageous due to their intrinsic cytotoxicity. Transient expression of cytokines such as interleukin-12 (IL-12) or tumor-associated antigens using arthritogenic backbones can remodel the tumor microenvironment to promote pro-inflammatory responses and overcome immune tolerance or T cell exhaustion. For gene replacement therapies, targeted delivery of sa-mRNA to specific tissues offers a strategy for sustained therapeutic protein expression. Off-target effects can be minimized through engineering of the 3′ UTR, while replication-associated innate immune activation may be mitigated by co-expression of IFN antagonists. Non-cytopathic nsP2 variants may further enable safer expression in immune-privileged sites such as the eye. Beyond therapeutic applications, alphavirus-based sa-mRNA systems provide powerful experimental platforms for probing fundamental aspects of alphavirus biology, including the functions of nsP2 and nsP3, regulation of replicase activity in the context of innate immune responses, and the role of nsP4 fidelity in viral diversity and precision control of transgene expression. Conventional mRNA platforms have also been used to deliver cytokine adjuvants, such as interleukin-12 (IL-12), to expand and prime CD8+ T cell pools against melanoma and related malignancies (238, 239). An EEEV-derived sa-mRNA expressing IL-12 and IL-1 receptor antagonist induced superior antitumor responses compared to a VEEV-based backbone (240). Additional cytokines, including IL-15 and IL-18, have been evaluated for their capacity to promote natural killer (NK) cell activation and Th1-skewed immunity in preclinical models, alongside granulocyte-macrophage colony-stimulating factor (GM-CSF), which supports the differentiation and activation of monocytes into pro-inflammatory macrophages (241). Furthermore, replicon particle systems derived from VEEV, SFV, and SINV have entered clinical trials for adenocarcinomas, prostate cancer, and checkpoint inhibitor-based immunotherapy, generally demonstrating acceptable tolerability with adverse events limited primarily to fever and mild injection-site reactions (242). Despite these advances, anti-vector immunity remains a concern for these platforms, and efficient delivery and targeting of LNP-encapsulated sa-mRNA to tumor-resident dendritic cells continue to be key translational barriers. In this context, filamentous liposomal delivery systems have shown promising delivery efficiency for SFV plasmid DNA vectors encoding tumor suppressor and immune checkpoint targets such as p53 and PD-L1 (243). The development of RNA therapeutics for protein replacement has similarly provided a framework for evaluating appropriate sa-mRNA dosing in immunocompetent but anatomically privileged tissues. In the context of retinal disease, one study showed a minimal threshold for detectable transgene expression in ocular cell lines, while higher doses were paradoxically associated with reduced exogenous protein production, suggesting a delicate balance between GOI expression and innate immune activation (244). Future studies should more systematically evaluate the contribution of regulatory RNA elements, such as the Old World virus DLP, to enhancing payload expression relative to innate immune restriction in physiologically relevant cell systems. Incorporation of innate immune antagonists or decoy proteins, such as vaccinia virus B19 or nodavirus B2, which suppress sensing of alphaviral dsRNA, may help mitigate dose-dependent inflammatory responses and improve the safety profile of sa-mRNA therapeutics (244, 245). Encouragingly, a recent study showed that intramuscular delivery of sa-mRNA encoding the Nppa gene supported sustained atrial natriuretic peptide hormone production for up to four weeks in mice. The therapeutic benefit of Nppa sa-mRNA was further validated in a swine model (246), supporting its potential as a less invasive alternative to direct cardiac injection approaches commonly used in regenerative therapies. CONCLUDING REMARKS AND FUTURE DIRECTIONS Decades of work on alphavirus biology have established the framework for contemporary sa-mRNA platforms, revealing how conserved coding and non-coding elements define replicative competence, host interactions, and tissue tropism. Insights into entry pathways, membrane remodeling, innate immune evasion, and RNA synthesis indicate that alphavirus infection is more complex than a simple receptor-centric process and that distinctions between Old World and New World viruses are less absolute than once appreciated. Dissecting how nonstructural proteins and RNA regulatory elements diverge across species and strains will thus be critical for predicting alphavirus emergence and for rationally engineering sa-mRNA backbones that improve antigen expression while minimizing off-target effects. Several key questions now define the next phase of sa-mRNA development. First, the fidelity of the error-prone nsP4 RdRp in replicating long heterologous payloads, and its impact on antigen integrity and quasispecies diversity, remain incompletely defined. Chronic CHIKV disease is characterized by macrophage reservoirs of viral RNA that are thought to sustain joint inflammation through persistent CD4+ T cell recruitment (247). Although the mechanisms underlying alphavirus persistence are not fully understood, prolonged sa-mRNA activity may similarly generate aberrant DI-like RNA species or mutant payloads capable of redirecting host immune responses toward non-protective outcomes in prophylactic settings. Accordingly, next-generation sa-mRNA platforms should prioritize the development of high-fidelity nsP4 variants or inducible regulatory systems that provide tighter control over gene expression during extended replication cycles. Second, the relative immunogenicity of the nonstructural replicase proteins versus the encoded antigen, and the extent to which anti-backbone immunity may limit homologous or heterologous boosting, require systematic evaluation. Emerging evidence demonstrating anti-replicase immunity emphasizes the need to expand the repertoire of candidate backbones beyond the commonly used VEEV, SINV, and SFV systems (234). At present, our knowledge of T cell epitopes targeting nonstructural proteins remains limited, although genomic surveillance of circulating CHIKV strains proposes nsP3 as a prominent driver of viral evolution alongside E1 and E2 (248). Exploratory efforts to develop broadly protective alphavirus vaccines have highlighted conserved, immunostimulatory epitopes within nsP2, nsP3, and nsP4 (249). In this context, the flexibility in permitting diverse chimeric P123–nsP4 configurations or taRNA hybrid systems provides a practical strategy to mitigate anti-vector immunity (109, 198). Third, defining the kinetics and magnitude of replicase activity, innate immune activation, and the counteracting viral antagonism mediated by nsPs and CSEs during sa-mRNA amplification will be critical for optimizing existing backbones and extending the platform to alphaviruses not naturally adapted to humans. A clearer understanding of how and when sa-mRNA-driven antigen expression is turned off will also inform broader questions surrounding the unexpectedly high rates of chronic disease observed following infection with arthritogenic alphaviruses. The observation that many CHIKV replicase molecules do not directly participate in spherule formation or RNA synthesis suggests that the nonstructural proteins may perform additional regulatory or immunomodulatory functions (250). From a platform engineering perspective, the VEE complex represents a largely underexplored group of alphaviruses that exhibit antigen expression dynamics similar to those of VEEV itself (230). Likewise, next-generation sa-mRNA platforms should more systematically evaluate backbones derived from the SFV complex and related Old World alphaviruses. Future studies will need to reconcile the propensity of arthritogenic alphaviruses to establish chronic infection with the paradoxically strong innate immune restriction often encountered when these viruses are deployed as replicon systems. If appropriately harnessed, Old World alphaviruses may provide favorable platforms for sustained antigen expression in therapeutic settings such as cancer immunotherapy. Conversely, nsP2 variants that only partially inhibit type I IFN signaling may support prolonged, low-level immunogen expression, thereby promoting more durable adaptive immune memory. Given recent findings that ectopically expressed alphavirus replicases can differentially convert host RNA templates into immunostimulatory RNAs (38), it will also be important to determine whether similar processes occur during prophylactic and therapeutic sa-mRNA use. Such template promiscuity may contribute to backbone-specific differences in self-adjuvanticity, with direct implications for immune priming and durability in prophylactic applications. Finally, more comprehensive investigation of the nsP3 host interactome will be required to better define its roles in innate immune activation, tissue tropism, and RNA persistence. Analysis of RSEs within the 3′ UTR may further clarify mechanisms regulating sa-mRNA stability and turnover in vivo. In parallel, the expanding application of alphavirus-derived sa-mRNAs in veterinary vaccinology (251, 252), where dose sparing, rapid reprogrammability, and accelerated regulatory pathways are advantageous, provides a valuable proving ground for next-generation platform designs while underscoring the importance of understanding host range and cell- and tissue-specific tropism. Together, integration of alphavirus molecular virology with RNA engineering and systems-level host profiling promises to guide the rational evolution of sa-mRNA technologies toward greater efficacy, precision, and clinical versatility.