Enhanced Efficacy of a Next-Generation EEEV Self-Replicating RNA Platform for Combination Cancer Immunotherapies

Abstract

Self-replicating RNA (srRNA) technology has received its first approval with Kostaive and EUA for GEMCOVAC COVID-19 vaccines. The effective doses for srRNA vaccines are far lower than for conventional mRNA, as srRNA drives greater protein expression and has attributes of viral vectors that can potentiate immunogenicity. Importantly, srRNA encodes for a viral replicase, frequently derived from non-structural proteins (nsP) from Venezuelan equine encephalitis virus (VEEV), that helps amplify the encoded transgene. nsPs also influence critical cellular processes, including protein translation and processing, inflammation and autophagy, that can impact downstream performance. Here we show that a next-generation srRNA vector derived from Eastern equine encephalitis virus (EEEV), identified from a wider screen of alphaviruses, that has high potency in two immunotherapeutic applications, RBI-1000 and RBI-2000. RBI-1000, a multigenic vaccine targeting acquired resistance mutations in breast cancer, was able to elicit antigen-specific immune responses to tumor-associated antigens and neoantigens. In addition, RBI-2000, a combination IL-12 and IL-1RA biotherapeutic, resulted in therapeutic levels of protein expression that bolstered anti-tumor immune responses in vivo. Both RBI-1000 and RBI-2000 successfully demonstrated efficacy in preclinical tumor models. These data show the importance of empirically evaluating new srRNA vectors for the development of fit-for-purpose therapies.

Graphical Abstract

Self-replicating RNA is a validated platform for vaccines and immunotherapy but limited to the use of VEEV as the base vector. Here, Miyake-Stoner and colleagues report an Eastern equine encephalitis virus srRNA platform for delivery of vaccine antigens and multigenic immunotherapeutics that resulted in strong immune and anti-tumor effects in preclinical studies.

Introduction

The overall goal of tumor immunotherapy is to initiate the tumor immunity cycle leading to activation of immunity, killing of tumor cells and priming of subsequent waves of anti-tumor adaptive immunity¹. This has been achieved therapeutically using two distinct strategies, antigen-specific approaches (such as cancer vaccines or antibody-based approaches, or cell-based therapies), or antigen-independent approaches (such as immune checkpoint inhibitors (CPI) and cytokines). Clinical successes in all modalities have led to approved drugs in all cases across numerous solid and liquid malignancies². It is likely that combinations of these strategies will extend the efficacy of future immunotherapy treatments³.

Since the SARS-CoV-2 pandemic, RNA is seen as a successful therapeutic platform for infectious disease vaccines^4–7. Within this class, both conventional mRNA and self-replicating RNA assets have received regulatory approvals^8,9. Both approaches can induce protective antibody responses, however srRNA can achieve this at a fraction of the dose of mRNA due to the higher protein (i.e. antigen) expression, compounded with attributes of viral vectors that can further adjuvant immune responses and potentiate vaccine performance¹⁰. This difference is due to the presence of a viral replicase in srRNA, made up of non-structural proteins (nsPs) frequently derived from a parental Venezuelan equine encephalitis (alpha)virus (VEEV), that enables the amplification of progeny RNA molecules for translation, including the transgene¹¹. Importantly, nsPs also impact host cellular pathways, including protein processing and innate immune pathways, that can affect downstream effector responses¹². Additional improvements to the canonical VEEV srRNA platform resulted in even further increases in clinical bioactivity and lower human doses. A recent clinical study showed that a rabies vaccine using an optimized VEEV vector resulted in seroprotection at doses as low as 0.1 micrograms in the majority of participants, demonstrating that improvements to the srRNA platform translate to enhanced clinical performance^13,14. In addition to vaccine-elicited antibody responses, induction of cell-mediated immunity, particularly the generation of cytotoxic CD8 T cells, is also critical for vaccine applications. In head-to-head preclinical studies, VEEV srRNA has outperformed conventional mRNA in generation of antigen-specific CD8 T cell responses to a pathogen, highlighting the benefits of the viral nature of srRNA at activating both arms of the immune system¹⁵.

In addition to infectious disease vaccines, RNA vaccines have also shown promise for tumor immunotherapy in recent clinical trials. Specifically, mRNA cancer vaccines have been evaluated using a range of tumor antigens including tumor-associated antigens (TAA)^16,17, as well as personalized and shared neoantigens (neoAg)^18–21. As mutations found in tumors are often heterogenous, cancer vaccines need to target multiple proteins and epitopes to prevent immune escape. For larger antigens, such as TAAs in an mRNA vaccine, each antigen is encoded on a separate mRNA molecule, likely to limit the size of the drug substance for ease of manufacturing^16,17. However, as the number of antigenic targets increases, the relative amount of each mRNA is diluted, leading to suboptimal dosing. mRNA vaccines encoding neoAg, encoding short multi-epitope cassettes, have proven more successful, particularly in pancreatic cancer^19,20. Similarly to mRNA, srRNA²² has also shown success in this area with a personalized neoAg vaccine priming antigen-specific CD8 T cells in advanced metastatic solid tumors at a substantially lower dose of RNA (30 μg was considered optimal) than most mRNA trials (routinely in the 100–1000 μg dosing range)²¹.

Antigen-independent tumor immunotherapies represent an important class of drugs for universal off-the-shelf cancer therapies or in the case of cold tumors with a poorly defined antigenic landscape. Antigen-independent strategies have also been explored with combinations of cytokine and proinflammatory molecules being delivered using a cocktail of mRNA molecules. The drawback of this admixed cocktail strategy, as described above for mRNA cancer vaccines, is that this leads to lower doses of each cytokine to compensate for the overall increase in mRNA/LNP requiring up to 4 dosing days per week in some mouse models^23,24. Longer srRNAs can encode multiple transgenes on a single molecule, and their self-replicating nature leads to a dose-sparing effect that limits side effects and prolongs antigen expression compared to conventional mRNA^21,25.

Interestingly to date, most late-stage srRNA assets are derived from the alphavirus VEEV, i.e. encode nsPs from VEEV to form the viral replicase^8,9,26. In addition to their role in srRNA amplification, nsPs can also affect host cell pathways and thus impact how the heterologous protein is processed and seen by the immune system¹². To determine the impact of nsPs on protein expression and immunogenicity, we built and screened a library of srRNA backbones from several alphaviruses for protein expression and in vivo bioactivity. We report in this paper that an srRNA backbone derived from Eastern equine encephalitis virus (EEEV) leads to higher protein expression and improved T cell priming when used as a vaccine. In this paper we describe two distinct therapeutic uses for this srRNA platform in oncology, namely a cancer vaccine, RBI-1000, and an expression vector for biotherapeutic proteins, RBI-2000.

RBI-1000 is designed to target common resistance antigens/mutations found in estrogen receptor (ER) positive, HER2 negative breast cancer. Anti-estrogen targeted therapies, particularly aromatase inhibitors, are frequently used in ER+ HER2− breast cancer. Resistance to these therapies is driven by mutations in the ESR1 gene with hotspot neoantigens frequently reported in 30–56% of patients with metastatic ER+ HER2− breast cancer after receiving aromatase inhibitors^27–32. ER+ HER2− breast cancer treated with standard of care therapies may also induce resistance by upregulating expression of compensatory bypass proteins. Amplifications in ERBB2/HER2 (~10%) and ERBB3/HER3 (~14%) as well as mutations in PIK3CA/PI3K (~40% frequency) have been reported in subsets of patients as alternative resistance mechanisms to anti-ER therapy^33–39. To avoid these bypass resistance mechanisms, outside of ESR1 mutations, RBI-1000 was designed to co-encode HER2 and HER3 TAAs, as well as a multi-epitope neoAg cassette encoding PIK3CA mutations. This approach is supported by some of our previous studies that have documented vaccines targeting individual resistance genes and neoepitopes have allowed for significant antigen-specific anti-tumor immunity^40–45. Notably, this current approach illustrates the versatility of the srRNA platform in its ability to co-encode multiple TAA and neoAg cassettes on the same RNA molecule. This is in contrast to multi-drug cocktails used in conventional mRNA cancer vaccine trials that can result in higher total administered doses, which may impact safety and reactogenicity, and complicate manufacturing due to the need of synthesizing separate drug substances^16,17.

RBI-2000, an srRNA expressing the cytokine IL-12 in combination with IL-1 receptor antagonist (IL-1RA), was designed to induce tumor-specific immunity when injected intratumorally. IL-12 has been evaluated in numerous clinical trials and is known to have potent anti-tumor effects due to induction of IFNγ, leading to priming of T cells and initiation of the tumor immunity cycle^46–49. IL-1RA functions to block the action of the inflammatory cytokines IL-1α and IL-1β and its recombinant form, Anakinra, is approved for use in rheumatoid arthritis as well as other inflammatory conditions^23,50,51. In addition, Anakinra and anti-IL-1 blocking antibodies have been reported to have anti-tumor effects by blocking IL-1-mediated angiogenesis, as well as differentiation and proliferation of suppressive immune cell types in the tumor microenvironment (TME)^52–55. We hypothesized that blocking the pro-tumor IL-1 pathway, in combination with IL-12-mediated T cell activation, would synergize to promote anti-tumor responses.

This report demonstrates two multigenic immunotherapies, RBI-1000 and RBI-2000, using a next-generation EEEV srRNA vector that outperforms the prototypical VEEV srRNA. Thus, expansion of the srRNA platform to include additional alphaviral properties can lead to superior preclinical efficacy that may translate to more effective immunotherapeutics.

Results

Generation of RBI-1000 – an srRNA cancer vaccine targeting acquired resistance in ER+ breast cancer

For targeting of acquired resistance to estrogen therapy that occurs in ER+ breast cancer, activating mutations in the ESR1 gene were selected^27–32. In addition, bypass pathways that can also render estrogen therapy ineffective, such as activating mutations in PIK3CA or elevated expression of HER2 and HER3, were also included to allow for a multi-pronged approach in suppressing resistance^33–39. To identify an optimal expression cassette for ESR1 targeting, the top 6 most frequent ESR1 neoAgs were selected and encoded in multi-epitope cassettes, which differed by epitope ordinality as well as the choice of the connecting linker sequences (Figure 1A and S1). NeoAg sequences were 30 amino acids long, with the mutated residue in the middle to increase the frequency of both MHC class I and class II epitopes upon cellular processing. Cassettes were encoded into the VEEV srRNA vector and screened for their ability to induce antigen-specific T cell responses in immunized mice. Of the 16 different cassettes screened, a lead was selected based on strongest and most balanced T cell responses to each of the 6 encoded ESR1 neoAgs (Figure 1B). As predicted, and previously reported, in vivo immunogenicity was influenced by both the order of the ESR1 neoAgs and the linker sequences within each cassette⁵⁶.

Figure 1: Generation of a next generation srRNA cancer vaccine, RBI-1000.

(A) Schematic presentation of a bead-on-a-string design of a multi-epitope neoAg cassette using short amino acid linkers. (B) In vivo immunogenicity, as measured by ex vivo neoAg-specific T cell responses, was assessed to different srRNA vaccine designs in immunized mice. srRNA vaccines encoding ESR1 neoAgs in a bead on a string cassette using 5 different linkers, labeled AAY, EAAAK, RVRR, GGGS, and GPGPG were evaluated for a single neoAg ordinality (Ordinality 1). The GPGPG linker was also evaluated in an additional 11 ordinalities. Splenic neoAg-specific T cell responses shown as Spot Forming Units (SFU) in an IFNγ ELISpot assay as geometric means, with each symbol representing individual animals. The red box indicates the lead construct based on these data. (C) Schematic of RBI-1000 tetragenic srRNA design encoding ESR1 and PIK3CA neoAg cassettes, along with HER2 and HER3 tumor-associated antigens. (D) In vitro screening of 4 different tetragenic srRNA designs for relative protein expression. (E) In vivo immunogenicity evaluation of tetragenic srRNA designs, along with monogenic and bigenic comparators. Antigen-specific T cell responses to ESR1 neoAgs (blue), HER2 (grey), and HER3 (pink) were measured following two immunizations. PI3K T cell responses were not measured due to haplotype incompatibility in this mouse strain. Graph shows geometric mean and standard deviation of antigen-specific T cell responses as SFU in an ELISpot assay, where each symbol represents an individual animal. (F) Schematic design of srRNA vaccines in VEEV or EEEV-based alphavirus vector backbones depicting the nsP that form the viral replicase and 26S subgenomic promoter that drives expression of the encoded antigens. (G) In vitro relative protein expression from different srRNA vectors. (H) In vivo immunogenicity evaluation of different srRNA vaccines utilizing different alphavirus-based vector backbones, as measured by antigen-specific T cell responses to ESR1 neoAg (blue), HER2 (grey), and HER3 (blue). PI3K T cell responses were not measured due to haplotype incompatibility in this mouse strain. Graph shows geometric mean and standard deviation of antigen-specific T cell responses as SFU in an ELISpot assay, where each symbol represents an individual animal. Statistical testing shows Mann-Whitney U-test. Additional statistical testing for panels (E) and (H) is shown in Supplemental Tables S1 and S2.

Similarly, four separate tetragenic designs where the ESR1 neoAgs cassette was combined with bypass resistance targets, HER2, HER3 and PIK3CA neoAgs, were built and evaluated for protein expression in vitro. Each tetragenic construct differed by ordinality of the 4 encoded transgenes (Fig. S2). Expression of all four targets, ESR1, PI3K, HER2 and HER3 were confirmed from each srRNA construct, albeit with differing levels (Figure 1C, 1D, and S3). In vivo, all four tetragenic constructs primed T cell responses to ESR1 neoAgs, HER2 and HER3. However, Tetragenic 1 showed the most balanced immunogenicity across all antigens, that were non-inferior when compared to monogenic srRNA controls or admixed bigenic srRNA constructs (Figure 1E and Table S1). PI3K mutations did not yield positive ELISPOT responses in this strain of mice, likely due to the MHC haplotype, thus were not considered in lead construct determination.

We have previously reported alternate alphaviruses can be used to design srRNA vectors distinct from the most frequently used VEEV^57,58. Importantly, the nature, i.e. parental alphavirus origin, of the srRNA backbone can influence both protein expression and immunogenicity, with some vectors demonstrating enhanced induction of cell-mediated immunity, critical for anti-tumor immunity^12,57,59,60. However, the mechanism underlying these differences is poorly characterized, requiring empirical screening of vectors for the desired properties. Thus, to identify an optimal srRNA vector, we evaluated several proprietary srRNA backbones for induction of antigen-specific T cell responses to the encoded antigens. Although antigen expression and in vivo immunogenicity was confirmed from all srRNA vectors tested (Figures S4 and S5), an srRNA vector derived from EEEV trended higher for relative protein expression in vitro, when compared to a VEEV-derived construct, specifically for ESR1 (Figure S4). In agreement with this, although all srRNA backbones were able to induce antigen-specific T cell responses, ESR-1-specific T cell responses were significantly higher with the EEEV construct to ESR1 in vivo (Figure 1H, S5, and Table S2). Importantly, T cell responses to HER2 and HER3 were observed and comparable to VEEV (Figure 1H and Table S2). As HER2 and HER3 are also validated antibody targets, we confirmed that EEEV can also induce antibody responses to these two tumor-associated antigen targets (Figure S6). Thus, for this set of antigens and cassette design, EEEV was determined to have the optimal balance of immunogenicity and significantly higher anti-ESR1 T cell responses relative to VEEV and the other srRNA vectors.

Following these series of screens, EEEV Tetragenic 1 was designated as RBI-1000 for subsequent studies.

RBI-1000 shows effective anti-tumor responses in breast cancer mouse model

To determine whether the immune response primed by the EEEV backbone vector, RBI-1000, is of sufficient quality and magnitude to control tumor growth, we used the MM3MG cells expressing oncogenic driver human HER2-Δ16⁴¹. MM3MG-HER2Δ16 tumor-implanted mice were immunized with two administrations of 0.1, 1, and 10 μg of RBI-1000 or a control LNP-formulated srRNA to assess the impact of dosing and potentially control for innate inflammation associated with the drug product. Significant tumor growth control and survival was observed down to 0.1 μg dose level when compared to control mice (Figure 2A–B). Bodyweight loss was dose-dependent but animals never lost more than 15% of bodyweight and recovered quickly post-dosing (Figure S7). Mechanistically, we also observed that RBI-1000 primed both polyfunctional CD8 and CD4 T cell responses against HER2 at all dose levels tested (Figure 2C and S8). This T cell response was also confirmed using IFNγ ELISpot following restimulation of splenocytes with HER2, RBI-1000 primed statistically significant levels of cytokine-producing T cells over control srRNA (Figure 2D).

Figure 2: RBI-1000 controls tumor growth and increases probability of survival in HER2-expressing tumor model.

(A) Study schematic: animals (n=10 per group) were implanted with MM3MG-HER2Δ16 tumor cells on Day 0 then immunized with RBI-1000 or an control srRNA on Day 3 and 17 and monitored for tumor growth. (B). RBI-1000 promoted tumor growth control at all doses tested. (left) Tumor volume growth shown as mean and standard deviation for each treatment cohort. Statistical testing was performed using 2-way ANOVA with all RBI-1000 groups compared to control srRNA. (right) Probability of survival graph shows Kaplan-Meier log rank test with all RBI-1000 groups compared to control srRNA. (C) RBI-1000 promoted HER2-specific T cell responses tumor-bearing mice. Graphs show the mean frequency of cytokine producing CD8 (left) or CD4 (right) of HER2-specifc T cells following ex vivo stimulation of splenocytes in an ICS assay on day 27 (n=5 per group). (D) shows IFNγ ELISpot following HER2 stimulation of splenocytes. Graph shows mean and symbols represent individual mice. Statistical analysis was performed using a Mann-Whitney U-test (black) and Kruskal-Wallis test (blue).

Preclinical efficacy, following doses as low as 0.1 μg, shows a clear advantage of srRNA versus conventional mRNA vaccines that required 100-fold higher dose levels in published models^61,62.

Overall, these data provide strong preclinical proof-of-concept that alternate srRNA vectors encoding multigenic inserts can prime antigen-specific immunity against both neoantigen and tumor associated-antigen targets.

Design of RBI-2000 – a multigenic srRNA encoding anti-tumor immunotherapeutics

In addition to the utility of srRNA for generation of RBI-1000, a cancer vaccine, we also evaluated if this platform can be utilized for delivery of anti-tumor biotherapeutic proteins. The inflammatory cytokine IL-12 is a heterodimer of two proteins, IL-12p40 and IL-12p35⁶³, and has been shown to drive anti-tumor immune responses via production and activation of CD8 T and NK cells in preclinical and clinical studies^{46–49,64–66}. Previous iterations of srRNA technology using both VEEV and Semliki Forest virus have reported preclinical IL-12 expression^67,68. IL-1RA has also shown a role in anti-tumor responses by inhibiting tumor angiogenesis, proliferation, and metastasis^52–55. Thus, we generated an srRNA that combined IL-12p35, IL-12p40 and IL-1RA for a multi-prong immunotherapy approach. Similarly to the candidate selection process used for RBI-1000, screens to identify an optimal insert cassette configuration and a vector backbone were performed for RBI-2000 (Figure 3A). From these studies, EEEV showed superior protein expression in vitro and in vivo (Figure S9). Compared to VEEV, EEEV expressed significantly higher levels of the IL-12p70 heterodimer, the bioactive version of IL-12, and IL-1RA in vitro (Figure 3B–C) in either configuration. In vivo, EEEV outperformed VEEV for expression of both biotherapeutic proteins (Figure 3D&E).

Figure 3: Design of a multigenic srRNA, RBI-2000, to deliver anti-tumor immune modulators.

(A) Schematic of four srRNA designs based on two different alphaviral backbones, VEEV and EEEV, and 2 different transgene configurations. (B,C) In vitro relative protein expression of IL-12p70 (B) and IL-1RA (C) from srRNA candidates by ELISA. (D, E) In vivo protein expression of IL-12p70 (D) and IL-1RA (E) by ELISA in serum of mice, 3 days following administration of srRNA corresponding to VEEV Config. 2 and EEEV Config. 1, or saline as a control. Graph shows geometric mean and standard deviation. Each symbol represents an individual animal. Statistical testing was performed using a Mann-Whitney U test. ND, not detectable as the levels of cytokine is below the limit of detection of the assay. (F) In vivo induction of serum IFNγ by ELISA in mice that were administered EEEV Config. 1 srRNA at the indicated doses. Graph shows geometric mean and standard deviation. Each symbol represents an individual animal.

To assess if therapeutic levels of proteins were being expressed in vivo, we measured a pharmacodynamic marker of IL-12, namely induction of IFNγ, using a murine version of RBI-2000. A clear dose response was observed with significant pharmacodynamic effects of RBI-2000 seen as low as 10 ng in vivo (Figure 3F). These data demonstrate that functional IL-12 is being produced by RBI-2000 at levels that achieve a pharmacodynamic effect in vivo.

RBI-2000 inhibits tumor growth in CPI-sensitive tumor model

To assess the anti-tumor efficacy of RBI-2000, we utilized a MC38 colorectal model, known to be sensitive to PD-1 CPI therapy⁶⁹ (Figure 4A). As a monotherapy, RBI-2000 showed a dose-dependent inhibition of tumor growth, with significant anti-tumor effects shown at both the 1 and 10 μg doses (Figure 4B). Survival was also significantly improved at all doses of RBI-2000 including the lowest dose tested of 0.1μg dose (Figure 4C).

Figure 4: RBI-2000 demonstrates anti-tumor effect as both a monotherapy and in combination with CPI.

(A) Study design of MC38 model and RBI-2000 treatment. (B, C) Measurement of tumor growth volume (B) and probability of survival (C) following treatment with RBI-2000 as a monotherapy, a control RNA control (10 μg dose), and saline following intratumoral injection. (D) Individual tumor volume measurements for cohorts treated with RBI-2000 with an isotype control antibody or RBI-2000 in combination with anti-PD-1 antibody. Each line represents one animal. (E, F) Tumor volume (E) and probability of survival (F) following treatment with RBI-2000 and anti-PD-1 antibody. Additional treatment groups included a control srRNA (10 μg dose) with anti-PD-1 antibody and saline only mice. (G) Tumor volume following rechallenge of complete responders in the study. Tumor volume graphs show mean and standard deviation. Statistical testing was performed using 2-way ANOVA comparing each RBI-2000 dose level to the control srRNA group. Survival graphs show statistical testing by Kaplan-Meier log rank test comparing each RBI-2000 dose level to the control srRNA group. Primary tumor challenge experiments used n=10 animals per group. Animal numbers are indicated in (G) for rechallenge group.

When combined with an anti-PD-1 antibody, a dose-dependent improvement in tumor growth inhibition was observed with significance achieved at the 10 μg cohort (Figure 4D–F). Survival was significantly extended when anti-PD-1 was co-administered with RBI-2000, compared to RBI-2000 monotherapy, with 8 out of 10 complete responders. Furthermore, we found that these anti-tumor responses were durable, as rechallenge of complete responders resulted in an amnestic response that provided 100% protection in all groups, regardless of the primary treatment (Figure 4G). Bodyweight change shows a mild change of less than 10% in animals treated with 10 μg of either RBI-2000 or the control srRNA immediately post dosing but these animals recover quickly. No bodyweight loss was observed in lower dose cohorts (Figure S10)

RBI-2000 demonstrates anti-tumor effect in CPI-refractory tumor model

To further test the therapeutic effect of RBI-2000, we next employed a B16-F10 model of melanoma, which is considered “immunologically cold” due to its suppressive tumor microenvironment (TME), low immune cell infiltration and insensitivity to CPI (Figure 5A)^70,71. In this model, RBI-2000 was able to exhibit an anti-tumor effect at all doses tested with significant tumor growth inhibition at the 10 μg dose (Figure 5B). Improvements to survival were also observed in a dose dependent manner (Figure 5C). The addition of anti-PD-1 was tested in combination with RBI-2000 at the 10 μg dose, but no additional improvement was measured in either tumor size or survival. Bodyweight loss was mild with less than 5% loss in the 10 μg dose groups, no loss was observed at the lower doses (Figure 5D).

Figure 5: RBI-2000 demonstrates anti-tumor effect in CPI refractory tumor model.

(A) Study design of B16F10 model and RBI-2000 treatment. (B, C) Tumor volume (B) and probability of survival (C) following treatment with RBI-2000 as both a monotherapy and in combination with anti-PD-1 antibody. Control treatment groups included treatment with a control srRNA (10 μg dose), with either anti-PD-1 antibody or an isotype control antibody, and saline only with either anti-PD-1 antibody or an isotype control antibody. (D) Bodyweight change over time. Tumor volume graph shows mean and standard deviation. Statistical testing was performed using 2-way ANOVA. Survival graph shows statistical testing by Kaplan-Meier log rank test. All statistical comparisons were between RBI-2000 treated groups and the appropriate control srRNA groups. Bodyweight change shows mean with SEM. Tumor challenge experiments used n=10 mice per group.

These experiments clearly demonstrate an anti-tumor effect of RBI-2000 in mouse preclinical tumor models, the relative contribution of IL-12 and IL-1RA to this effect is still unknown. Both molecules have demonstrated anti-tumor effects in a range of preclinical tumor molecules (in some examples in combination with checkpoint inhibitors), but their synergistic effects remain to be determined^{23,24,67,68,72–74}.

Overall, these data show that RBI-2000 can inhibit tumor growth of multiple models representing distinct indications and TME types, either as a monotherapy or in combination with CPI.

Discussion

srRNA technology is a validated platform for infectious disease vaccine applications, and in clinical evaluation for both vaccines and expression of biotherapeutics in oncology^{8,9,13,21,75–77}. However, almost all current approved and late-stage assets are based on the same parental alphavirus, VEEV. Here, we show that an srRNA vector based on an EEEV can outperform the prototypical VEEV vector in eliciting antigen-specific immunity, as well as encoded protein expression. Moreover, we found that these properties allowed for significant anti-tumor efficacy for RBI-1000, a cancer vaccine application, and RBI-2000, a combination immunotherapeutic.

Alphaviruses are classified as Old World or New World, the latter includes both VEEV and EEEV. Previous studies have shown that srRNA vectors from New World alphaviruses are often advantaged for heterologous protein expression compared to some Old World vectors like Sindbis and Chikungunya^58,59, corroborated in these studies (Figures S3, S6). However, higher protein expression does not always translate to enhanced immunogenicity, as a CHIKV srRNA vaccine for SARS-CoV-2 was able to prime immune responses as well as a VEEV (TC-83 strain) srRNA, despite showing less durable protein expression⁶⁰. Similarly, we also observed CHIKV srRNA vectors performing as well as New World, VEEV and EEEV vectors (Figure S4). Within New World alphaviruses, while VEEV-based vectors are an established vaccine platform for generation of antibody responses and T cell responses, EEEV-based RBI-1000 led to superior T cell responses against the targeted neoAg and TAAs than the VEEV srRNA. Importantly, this increase in immunogenicity from an EEEV-based srRNA vaccine may be restricted to cellular immune responses and antigen-dependent, as a VEEV-based vaccine still outperforms other backbones for generation of humoral immune responses for a viral antigen⁵⁷. Despite growing knowledge of the properties of different alphaviral backbones for use as srRNA vectors, there do not appear to be predictive rules for how each vector will perform prior to experimental testing. Each vector will influence cellular factors such as protein expression, transport, processing and presentation of antigens leading to distinct immunological outcomes. Empirical testing of a vector library for each transgene in the context of the exact therapeutic indication may be still required to identify the optimal vector. Improving our knowledge of how each vector interacts with host cellular processes will allow us to make more precise predictions in the future.

With this body of work, we have shown that mining the alphavirus family has led to the identification of alternate srRNA vectors that can outperform the existing state-of-the-art srRNA technology. We have shown two applications of an EEEV-based platform in oncology: (1) RBI-1000, a multigenic cancer vaccine for ER+ breast cancer, and (2) RBI-2000, an IL-12 and IL-1RA combination immunotherapeutic. Both assets led to robust anti-tumor responses in preclinical models, with multiple-fold lower doses than mRNA and first generation srRNA approaches^61,62,78, warranting further clinical development of these two immunotherapies.

Methods

In vivo mouse studies

srRNA construct selection studies were performed at Bioqual Inc. (IACUC 20–135P). All procedures were conducted in compliance with all the laws, regulations, and guidelines of the National Institutes of Health (NIH) and with the approval of CRL Animal Care and Use Committee. CRL (formerly Explora Biolabs) is an AAALAC accredited facility.

For RBI-1000, 6–8 week old female BALB/c mice were immunized by intramuscular (IM) injection on Day 0, or Day 0 and 21 for prime-boost studies with the indicated dose of LNP-formulated srRNA. Spleens were collected 14 days following immunization for serum antibody and/or immune cell analysis. For RBI-2000, 6–8 week old female BALB/c mice were immunized by IM injection on Day 0. Serum was collected to assess protein expression.

For the MM3MG-HER2Δ16 efficacy study (Duke University IACUC A080-20-04), 6–8 week old female BALB/c mice were implanted in the mammary fat pad with 2×10⁵ MM3MG-HER2Δ16 tumor cells on Day 0 (n=12), then immunized with RBI-1000 or control LNP-formulated srRNA on days 3 and 17 by IM injection. Tumor growth was monitored by caliper measurements up to 2000mm³, and survival of mice (following protocol humane endpoints) was assessed. Assessment of systemic responses by peptide stimulation and flow cytometry was performed on day 27 post-implantation (n=5).

MC38 and B16F10 efficacy studies were carried out at Crown Bioscience under IACUC approved protocol CBSD-ACUP-001. Care and use of animals was conducted in accordance with the regulations of the association for assessment and accreditation of laboratory animal care (AAALAC). Tumor cells were implanted subcutaneously in female, 6–9 week old C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME) (MC38: 1×10⁶ cells per mouse; B16-F10: 2×10⁵ cells per mouse). Tumors were grown until 30–70mm³ in volume and then mice were randomized (n=10 per group). RBI-2000 was injected intratumorally at the indicated doses within 24 hours of randomization. Anti-PD-1 antibody (clone RMP1–14) or rat IgG2a isotype control (BioXCell, Lebanon, NH) was injected at 10mg/kg i.p. BIW 3 weeks (MC38) or on days 3–6 and days 8–12 (B16-F10). Tumors and body weight were measured twice weekly with a maximum tumor volume of 2000mm³.

srRNA construct generation

RBI-1000

The base srRNA vectors (ref US11510975B1) were linearized by restriction enzyme digest. The ESR1, PIK3CA, ERBB2, and ERBB3 gene variants were codon optimized/refactored for human expression in silico and synthesized de novo (GeneArt). The synthetic products were amplified using primers which added either 5’ and 3’ adaptor sequences to the ends of the genes, or primers which added sequences of homology to neighboring gene inserts. Vector digestion products and the PCR products were combined by Gibson Assembly^® to result in the final vectors.

RBI-2000

The base srRNA vectors (US11873507B2) were linearized by restriction enzyme digest. The IL-12A, IL-12B, IL-1RN genes were codon optimized/refactored for human expression in silico and along with the EMCV IRES were synthesized de novo (IDT). The synthetic products were amplified using primers which added either 5’ and 3’ adaptor sequences to the ends of the genes, or primers which added P2A sequences and/or sequences of homology to neighboring gene inserts. The digestion product and PCR products were combined by Gibson Assembly^® to result in the final vectors.

The control vector encoding red firefly luciferase was similarly constructed.

RNA generation and formulation

RNA was prepared by in vitro transcription from a SapI or BspQI-linearized plasmid template with bacteriophage T7 RNA polymerase (HiScribe^™ T7 High Yield RNA Synthesis Kit, NEB) followed by enzymatic addition of a 5’ cap-1 (Vaccinia Capping System, mRNA Cap 2′-O-Methyltransferase, NEB). RNA was purified using LiCl precipitation, followed by an ethanol wash and resuspended in water. RNA concentration was determined by absorbance at 260 nm (Nanodrop, Thermo Fisher Scientific).

srRNA was formulated in lipid nanoparticles using a microfluidics mixer (NanoAssemblr^™ Ignite^™, Cytiva). Lipids were suspended in ethanol. RNA was suspended in 50 mM citrate pH 4 at a concentration of 103 μg/mL and was mixed at a flow rate ratio of 3:1 (aqueous:organic) with a total flow rate of 12 mL/min. Resulting particles were analyzed for particle size and polydispersity using dynamic light scattering. Concentration and encapsulation efficiency were determined by RiboGreen^® assay.

srRNA potency

Retained potency for the LNP formulations were measured by diluting and electroporating (EP) unformulated srRNA into BHK-21 cells using a Lonza 4D-Nucleofector^® (Lonza Group AG, Basel, Switzerland) at calibration doses in duplicate to generate a reference curve. Test samples were prepared by extracting srRNA from LNPs and the srRNA was electroporated at several test doses in duplicate. After overnight incubation, the cells were fixed and immunostained with a phycoerythrin (PE)-conjugated (Novus Biologicals, Centennial, CO; catalog #703-0010) anti-dsRNA monoclonal antibody J2 (Jena Biosciences, Jena, Germany; catalog # RNT-SCI-10010200) that specifically detects the dsRNA replication intermediate of the srRNA. Cells were analyzed by flow cytometry using a 488 nm laser with 585 nm filter to detect PE-positive cells.

The frequency of PE-positive cells in the reference samples was used to generate a dose-response curve for quantification of potency based on RNA replication. The frequency of PE-positive cells for the test samples was used to interpolate an RNA dose (ng), and the percentage potency was calculated by dividing the theoretical dose by the measured value.

RBI-1000 protein expression

RNA was transfected by electroporation into BHK-21 cells using a Lonza 4D-Nucleofector^® (Lonza Group AG, Basel, Switzerland). ESR1: At 15–22 hours following transformation, the cells were collected and lysed in RIPA buffer. Lysate protein concentration was normalized, then probed in an immunoblot with an anti-ERα rabbit antibody (A300–497A, Bethyl) and imaged using an AF800 conjugated anti-rabbit goat antibody (A32735, Thermo). The fluorescence signal from the cell samples transfected by a monogenic srRNA expressing ESR1 was used to normalize expression levels to evaluate relative ESR1 expression from the panel of multigenic srRNAs. PI3K: At 15–22 hours following transfection, the cells were collected and lysed in RIPA buffer. Lysate protein concentration was normalized, then probed in an immunoblot with an anti-PI3KCA rabbit antibody (PA587398, Thermo) and imaged using an AF800 conjugated anti-rabbit goat antibody (A32735, Thermo). The fluorescence signal from the cell samples transfected by a monogenic srRNA expressing PI3K was used to normalize expression levels to evaluate relative PI3K expression from the panel of multigenic srRNAs. HER2: At 15–22 hours following transfection, the cells were fixed and permeabilized (eBioscience^™ Foxp3 / Transcription Factor Staining Buffer Set, Invitrogen) and stained using an AF488-conjugated anti-HER2 mouse monoclonal antibody (24D2, Biolegend). The mean fluorescence intensity (MFI) of AF488 was used as the readout of HER2 expression. The MFI of cells transfected by a monogenic srRNA expressing HER2 was used to normalize expression levels to evaluate relative HER2 expression from the panel of multigenic replicons. HER3: At 15–22 hours following transfection, the cells were fixed and permeabilized (eBioscience^™ Foxp3 / Transcription Factor Staining Buffer Set, Invitrogen) and stained using an APC-conjugated anti-HER3 mouse monoclonal antibody (IB4C3, Biolegend). The mean fluorescence intensity (MFI) of APC was used as the readout of HER3 expression. The MFI of cells transfected by a monogenic srRNA expressing HER3 was used to normalize expression levels to evaluate relative HER3 expression from the panel of multigenic srRNAs.

RBI-2000 protein expression and bioassays

Human IL-12p70 and IL-1RA expression were detected by electroporation of BHK-21 cells with RBI-2000 srRNAs. Supernatants were harvested at approximately 24 and 48 hours after transfection and assayed by Human IL-12p70 ELISA from RnD Systems (cat# DY1270) and Human IL-1RA/IL-1F3 DuoSet ELISA from RnD Systems (cat# DY280).

Human IL-12p70 and IL-1RA bioactivities were quantified by electroporation of BHK-21 cells with RBI-2000 srRNAs. Supernatants were harvested at approximately 24 and 48 hours after transfection and assayed with the GloMax^® Bioassay from Promega for IL-12 (cat# JA2601). IL-1RA was assayed using IL-1β Reporter HEK 293 Cells (Invivogen, cat# hkb-il1bv2) pre-incubated with 4 ng/mL IL-1β (1 ng/mL final) and supernatants from transfected BHK-21 cells as per the manufacturer’s protocol.

For in vivo RBI-2000 studies, ELISA kits were used to assess the serum concentration of human IL-1RA (Abcam), human IL-12p70 (R&D systems) and mouse IFNγ (R&D systems) at various time points and according to the manufacturer’s instructions. Absorbance was read at 450nm on a GlowMax^® Discover (Promega).

Flow cytometry

For RBI-1000 flow cytometry, cells were isolated from mouse spleens and mechanically dissociated with a 40-mm cell strainer (Greiner Bio-One). Red blood cells were lysed with RBC lysing buffer (Sigma-Aldrich). Cells were plated in 96-well plates (Corning; 1,000,000 cells/well) with brefeldin A (BioLegend), monensin (BioLegend), and CD28 (BioLegend; at 0.1mg/well) in the presence or absence of a pool of HER2 and HER3 peptides (1mg/mL; JPT) for 5 hours at 37°C and 5% CO₂. Irrelevant HIV-gag peptide mix (1mg/mL; JPT) was used as a negative control and PMA (50 ng/mL) and Ionomycin (1mg/mL; Sigma-Aldrich) were used as positive controls. Cells were stained with Fixable viability dye (Invitrogen) and additional fluorochrome-conjugated antibodies. For intracellular staining, a FoxP3 Fix/Perm Kit was used according to the manufacturer’s instructions (eBioscience). Antibodies used include: CD45, IL-2, CD4, CD8b, IFNγ, CD44, TNFα, and CD11b (all BioLegend). Data were collected using an LSR II flow cytometer (BD Bioscience) and analyzed with FlowJo software (TreeStar).

ELISpot

For RBI-1000 studies, mouse spleens were mechanically dissociated with a 40-μm cell strainer (Greiner Bio-One) and red blood cells were lysed with RBC lysing buffer (Sigma-Aldrich). Cells were then resuspended in AIM V media (Gibco) supplemented with 10% FBS (Invitrogen) at a final concentration of 5×10⁶ cells/ml. 500,000 cells were plated per ELISpot well and stimulated overnight with various ESR1 peptides (4 μg/mL), a pool of the ESR1 peptides (5 μg/ml for each peptide), a HER2 library (1 μg/ml for each peptide), a HER3 overlapping library (1 μg/ml for each peptide)) or the recombinant HER3 protein (4 μg/ml) as illustrated in the Table 1. ELISpot plates were developed using the ELISpot PLUS kit (Mabtech) as per the manufacturer’s instructions and shipped to an external CRO (Cellular Technology Limited) for spot counting.

Table 1:

Peptide sequences used for ex vivo restimulation

Peptide	Sequence
ESR1 K303R	LWPSPLMIKRSKRNSLALSLTADQM
ESR1 E380Q	VDLTLHDQVHLLQCAWLEILMIGLV
ESR1 Y537N	SMKCKNVVPLNDLLLEMLDAHRL
ESR1 Y537S	SMKCKNVVPLSDLLLEMLDAHRL
ESR1 Y537C	SMKCKNVVPLCDLLLEMLDAHRL
ESR1 D538G	SMKCKNVVPLYGLLLEMLDAHRL
ESR1 WT	SMKCKNVVPLYDLLLEMLDAHRL
HER2	PepMix HER2 ECD library (JPT)
HER3	Overlapping peptide library - Whole sequence (Biosynth)
HER3	Recombinant HER3 ECD (Sino Biological)

Anti-IgG ELISA

For RBI-1000 studies, 96-well microplates (Thermo Fisher) were coated overnight at 4°C with 100uL of either recombinant HER2 or HER3 proteins (Sino Biologicals) diluted in 1XPBS (Gibco). Plates were then blocked with 100μL of PBS plus 1% BSA (Biotium Inc) for 1 hour at room temperature and washed 3 times with 300μL of PBS plus 0.05% Tween (Sigma-Aldrich). Mouse anti-human HER2 or HER3 standards (R&D systems) and serum samples were appropriately diluted in PBS plus 1% BSA and added to the plates for 2 hours at room temperature. Plates were washed 3 times with PBS plus 0.05% Tween and incubated for 2 hours with 50μL of anti-mouse IgG-HRP. Plates were then washed 3 times with PBS plus 0.05% Tween and incubated for 1 hour at room temperature 50μL of HRP-conjugated anti-mouse IgG. Plates were again washed 3 times with PBS plus 0.05% Tween and incubated with 50μL of 1-Step^™ TMB Ultra (ThermoFisher) for 15 min at room temperature in the dark. Absorbance was read at 450nm on a GlowMax^® Discover (Promega).

Statistical analysis

Analysis was performed using Prism 10 (GraphPad software) and the specific tests are indicated in each figure legend.

Data availability statement

All preclinical source data have been provided with this paper or as supplemental information. Additional experimental details will be available upon request from the corresponding authors.