mRNA therapeutics in cancer immunotherapy

One of the approaches to pulse DCs with mRNA encoding tumor antigens is the use of bulk mRNA extracted from autologous tumors. Whole tumor mRNA-loaded DC vaccines demonstrated safety in phase I and I/II clinical trials in patients with renal cell carcinoma (RCC; NCT00006431) [48], pediatric and adult brain cancer [49], pediatric neuroblastoma [50], melanoma (NCT01278940) [51], and androgen-resistant prostate cancer [52]. In a phase I/II study (NCT01278940), whole tumor mRNA-transfected DC vaccines were administered alone or together with interleukin (IL)-2 and induced T-cell responses against mRNA-transfected DCs ex vivo in 16 of the 31 patients with advanced melanoma [53]. Immune-responders exhibited significantly enhanced survival in comparison to non-immune-responders (14 vs. 6 months). As the mRNA is extracted from the individual patient’s tumor, feasibility depends on sample availability and quality. Vaccine production is therefore highly variable and sometimes impossible. Within the extracted mRNA, truly cancer-associated antigen encoding mRNAs represent a very small fraction. A low number of relevant epitopes needs to compete with the majority of irrelevant peptides for presentation on MHC molecules for T-cell priming. In vitro mRNA amplification attempting to increase the abundance of all mRNA species in bulk preparations including those representing potential tumor rejection antigens, may even promote under-representation or loss of immunogenic antigens or epitopes in the process [54].

Another approach is to resort to defined TAAs and generate synthetic mRNA for ex vivo DC transfection, as has been done with prostate-specific antigen (PSA) [55], carcinoembryonic antigen (CEA) [56, 57], gp100/tyrosinase [58‐60], human telomerase reverse transcriptase (hTERT) [61] and Wilms’ tumor 1 (WT1) antigen [27, 62]. TAA mRNA DC vaccines exhibited good safety profiles in phase I and I/II clinical trials without serious adverse effects, such as autoimmunity, and did not show dose-limiting toxicity. In a phase II clinical study (NCT00965224), patients with acute myeloid leukemia (AML) in remission were vaccinated with WT1 mRNA-electroporated DCs [27]. Prevention or delay of relapse was observed in 43% of patients and the five-year overall survival of the vaccinated patients compared favorably to historic controls.

In a randomized phase II clinical trial (NCT01446731), a DC vaccine with mRNA encoding PSA, prostatic acid phosphatase (PAP), survivin and hTERT failed to significantly increase the disease-specific survival of docetaxel-treated, castration-resistant prostate cancer patients (docetaxel: 21.9 months, docetaxel and DC vaccination: 25.1 months) [34]. TAA-specific T-cell responses were present in 9 of the 18 combination therapy-treated patients, of which three were de novo.

In 78 melanoma patients with resected regional lymph node metastases (NCT00940004, NCT01530698, NCT02285413, NCT00243529), vaccination with gp100- and tyrosinase-mRNA loaded DC vaccines as an adjuvant treatment showed a favorable safety profile and induced tumor-specific T-cell responses in 71% of the patients [63]. The overall survival of DC-vaccinated patients was doubled as compared to that of 209 matched controls (63.6 months vs. 31.0 months).

Generally speaking, T-cell responses induced by mRNA-loaded DC vaccines have been rather weak, which may contribute to the low clinical efficacy. DC differentiation, maturation and antigen loading directly impact DC homing and T cell co-stimulation (reviewed in Gu et al. [64]), and are targets for further improvement of this approach.

Cognate T-cell help is a critical factor that determines the extent of vaccine-induced T-cell responses. This requires the vaccine antigen to be presented on MHC class II molecules of the DC. mRNA transfected into DCs undergoes cytoplasmic translation, enters the endogenous processing compartment and is therefore presented on MHC class I rather than MHC class II molecules. Several strategies have been followed to enhance MHC class II antigen presentation by routing the antigen through lysosomal compartments, such as fusion of the vaccine antigen to lysosome-associated membrane proteins (LAMPs) [65, 66], or to the MHC class I cytoplasmic and transmembrane domain [67]. Fusion of a lysosomal targeting signal (LAMP-1) to CEA in a chimeric mRNA (CEA/LAMP-1) enhanced the expansion of CEA-specific CD4⁺ T-cell responses and led to a more potent CEA-specific CTL response in vitro [68]. CEA LAMP mRNA-transfected DC vaccines were evaluated in phase I studies in patients with newly diagnosed glioblastoma multiforme (NCT02529072, NCT00626483, NCT00639639) [32, 33, 69, 70], which suggested prolongation of overall survival as compared to non-transfected DC vaccines or historical controls [33]. Four ongoing phase II clinical trials are evaluating CEA/LAMP-1 mRNA engineered DC vaccines in glioblastoma patients in combination with tetanus toxoid preconditioning to enhance DC migration, granulocyte-macrophage colony-stimulating factor (GM-CSF), and other drugs, such as basiliximab (antagonistic anti-CD25 monoclonal antibody (mAb)), varilumab (agonistic anti-CD27 mAb) or temozolomide (NCT02366728, NCT03927222, NCT03688178, NCT02465268).

As ssRNA and dsRNA trigger TLR activation, pulsing of DCs with mRNA provides a maturation signal irrespective of the biological activity of the encoded antigen. Several strategies aim to further improve the T-cell priming capacity of tumor-antigen loaded DC vaccines. These include co-transfection with immunostimulatory ligands and receptors for enhanced DC maturation and T-cell co-stimulation, which can be introduced by mRNA. In preclinical studies, DC co-transfection of cancer antigens with OX40 ligand (OX40L), 4-1BB ligand (4-1BBL), glucocorticoid-induced tumor necrosis factor receptor (GITR) ligand (GITRL), CD40 ligand (CD40L), constitutively active (ca)CD40 or inducible T cell co-stimulatory ligand (ICOSL) mRNA improved DC co-stimulation and resulted in enhanced T-cell priming [71‐76]. One of the most clinically advanced DC vaccines is Rocapuldencel-T, an autologous tumor mRNA-transfected DC vaccine co-transfected with CD40L mRNA [28, 29]. In a phase III trial with 462 metastatic RCC (mRCC) patients (NCT01582672), Rocapuldencel-T failed to improve the overall survival of sunitinib-treated mRCC patients. Co-transfecting TAA mRNA-loaded DC vaccines with CD40L, CD70 and caTLR4 mRNA (TriMix) represents an alternative approach [77‐79]. Superior T-cell priming capacity was observed by T-cell stimulation with TriMix-transfected DCs compared to that induced by irrelevant mRNA-transfected immature DCs, or by DCs matured in culture supplemented with cytokines (IL-1β, IL-6, tumor necrosis factor (TNF)α, and prostaglandin E2 (PGE2)) [77]. In a single-arm phase II clinical study, TriMix-DCs, transfected with TriMix and four TAAs (melanoma-associated antigen (MAGE)-A3, MAGE-C2, tyrosinase, or gp100) fused to an MHC class II targeting signal, plus the mAb ipilimumab (blocking the immune checkpoint cytotoxic T-lymphocyte-associated protein 4 (CTLA-4)) were combined in 39 pretreated advanced melanoma patients (NCT01302496) [37]. The combination was well tolerated and after a median follow-up of 36 months, eight complete and seven partial responses were observed. T-cell responses were detected in 12 of 15 patients by IFNγ enzyme-linked immunospot (ELISpot) [38]. The study lacked a direct comparison of TriMix-DCs and ipilimumab against ipilimumab alone.

mRNA-encoded cytokines are another class of molecules used to enhance DC maturation and T-cell priming capacity. In the context of autologous DC vaccines loaded with whole tumor mRNA preparations or with synthetic TAA-encoding mRNA, co-transfection with cytokine-encoding mRNA, such as GM-CSF, IL-12 and IL-15, was explored [80‐84]. Co-transfection of GM-CSF mRNA into tumor mRNA-loaded DCs significantly enhanced the anti-tumor efficacy of DC vaccines in CT26 tumor-bearing mice, which was associated with increased cytotoxicity of bulk splenocytes against CT26 tumor cells in vitro [80]. In another study, co-delivery of IL-12 mRNA enhanced the ex vivo migratory and immunostimulatory capacity of healthy donor-derived human DCs transfected with melanoma antigen recognized by T cells (MART)-1 [81]. Proliferation, effector function and memory potential of MART-1-specific CD8⁺ T cells were significantly enhanced by co-culture with IL-12 and MART-1 mRNA-transfected DCs compared to MART-1 mRNA-transfected DCs alone.

Non-formulated or “naked” mRNA administered intradermally or into lymph nodes (intranodally) has been shown to prime T-cell responses in mice [85‐88] and humans [45, 89, 90]. The rationale for intranodal injection is to deliver the vaccine directly into the area of T-cell priming for uptake by resident DCs. The major uptake mechanism of naked mRNA by immature DCs in lymph nodes [91] and in the dermis [92] is macropinocytosis. The feasibility and safety of naked mRNA vaccines was first demonstrated in a phase I/II trial in 15 metastatic melanoma patients using intradermally injected autologous tumor mRNA and GM-CSF [93]. In five of 13 evaluable patients, enhanced antigen-specific CD4⁺ and CD8⁺ T-cell responses in ex vivo expanded PBMC were detected by intracellular IFNγ staining upon re-stimulation with tumor mRNA-transfected PBMC. Vaccine-induced humoral responses were analyzed by incubating allogenic melanoma cell lines with plasma samples taken before or during the therapy followed by flow cytometry analysis to detect cell-bound IgG antibodies. Upon vaccination, an increase in the amount of tumor-cell specific IgG antibodies was observed in the plasma of four out of the 13 patients. Objective responses were not apparent in this trial. In a phase I/II clinical study with 30 stage IV RCC patients, intradermal administration of naked mRNA encoding mucin 1 (MUC1), CEA, human epidermal growth factor receptor-2 (HER-2)/neu, telomerase, survivin and MAGE-A1 combined with GM-CSF was well tolerated and associated with stable disease in 15 patients and one partial response [90]. Antigen-specific T-cell responses were observed by IFNγ ELISpot or chromium-release cytotoxicity assays in 12 out of 17 patients.

In 2012, a phase I study of 29 advanced melanoma patients was initiated, which proved that intranodal injection of mRNA is both feasible and safe (NCT01684241) (unpublished results). Vaccination induced either de novo or expanded existing T-cell responses against the vaccine-encoded antigens New York esophageal squamous cell carcinoma-1 (NY-ESO-1) and tyrosinase in six out of 13 evaluable patients. New lesions occurred in only eight of the 29 patients during the course of the study. In another phase I study of 20 melanoma patients (NCT03394937), intranodal injection of tyrosinase, gp100, MAGE-A3, MAGE-C2, and melanoma antigen preferentially expressed in tumors (PRAME)-encoding mRNA together with TriMix mRNA was well tolerated. As observed by IFNy ELISpot and intracellular cytokine staining, T-cell responses against at least one TAA were induced in 35% of patients [45].

Mutated neoantigens are promising vaccine targets as they are abundant in many tumor types and highly tumor-specific (reviewed in Vormehr et al. [94]). Neoantigen-specific T cells are likely not subject to central tolerance and have a higher likelihood to induce high-affinity T-cell responses, while TAAs come with the challenge of breaking tolerance. However, as each single tumor has its unique profile of mutations, the full potential of mutated neoantigens as an ideal source for vaccine antigens can only be leveraged in the context of individualized vaccines, with each and every patient being administered their personal set of mutated epitopes. As mRNA vaccines can be produced fast and cost-effectively, they are particularly suited for the purpose of “on demand” vaccine design and manufacturing. Intranodal administration of a neoantigen-specific mRNA vaccine, comprising 20 patient-specific mutations, was investigated in a phase I clinical trial in stage III and IV melanoma patients (NCT02035956) [89]. Neoantigen-specific mRNA vaccines induced CD4⁺ and CD8⁺ T-cell responses against multiple vaccine neoepitopes in 13 evaluable melanoma patients. Objective responses were observed and the cumulative rate of recurrences was significantly reduced by vaccination, resulting in a sustained progression-free survival. The vaccine was well tolerated.

Nanoparticulate formulations protect mRNA from extracellular RNases and improve uptake by APCs in vivo. Formulations typically contain polymers, such as protamine, or cationic and ionizable lipids with or without polyethylene glycol (PEG) derivatives to enable complexing with mRNA via electrostatic interactions and condensing of the RNA molecules (reviewed in Zeng et al. [95]).

One of the mRNA formulations in advanced clinical development is based on protamine, a polycationic peptide that complexes negatively charged mRNA for intradermal delivery [42‐44, 96‐100]. In mice, protamine-formulated mRNA is taken up by leukocytes and non-leukocytes in the skin, including APCs that migrate to the draining lymph node, express and present the encoded protein and induce T-cell responses [96]. The safety and feasibility of a protamine-formulated mRNA vaccine encoding MART-1, tyrosinase, gp100, MAGE-A1, MAGE-A3 and survivin was demonstrated in 21 metastatic melanoma patients (NCT00204607) [99]. The mRNA vaccine was injected intradermally with GM-CSF either alone or in conjunction with keyhole limpet hemocyanin as adjuvants [101]. Antigen-specific T cell responses in ex vivo expanded PBMC were detected by intracellular IFNγ staining after re-stimulation with irradiated, autologous PBMC transfected with TAA mRNA. Vaccination was associated with an increase in the number of antigen-specific CD4⁺ and CD8⁺ T cells in two of four immunologically evaluable patients. One complete response occurred in seven patients with measurable disease. In addition, a vaccine encoding six protamine-formulated prostate cancer-specific antigens (PSA, prostate-specific membrane antigen (PSMA), prostate stem cell antigen (PSCA), 6-transmembrane epithelial antigen of prostate (STEAP), PAP and MUC1) was investigated in a phase I/II trial in patients with asymptomatic or minimally symptomatic metastatic castrate-resistant prostate cancer (NCT01817738) [42‐44]. The vaccine was well tolerated. In the phase I part, vaccine-induced antigen-specific T-cell responses were shown in 26 of 33 patients by ex vivo IFNγ ELISpot assay, intracellular cytokine and tetramer flow cytometry [44]. In the randomized, placebo-controlled phase II part of the study (n = 197) [43], the vaccine failed to improve overall survival and progression-free survival. A vaccine derived from the same protamine-formulated mRNA platform encoding five tumor-associated antigens (NY-ESO-1, MAGE-C1, MAGE-C2, survivin and trophoblast glycoprotein (TPBG)) was further evaluated in a phase I/IIa dose-escalation trial in seven patients with locally advanced and 39 patients with metastatic non-small-cell lung cancer (NSCLC) and controlled disease after first-line therapy (NCT00923312) [100]. The vaccine was well-tolerated and induced T-cell responses against one or more antigens in 63% of evaluable patients. The median overall survival, however, was not improved by vaccine monotherapy compared to that typically seen with maintenance chemotherapy. An ongoing phase I/II trial is currently evaluating this vaccine, with the addition of the antigen MUC1 in combination with either the antagonistic anti-programmed death-ligand 1 (PD-L1) antibody durvalumab or durvalumab plus the antagonistic anti-CTLA-4 antibody tremelimumab in patients with metastatic NSCLC (NCT03164772).

A major limitation of local administration is that only the injected or the draining lymph nodes are accessed and participate in antigen-specific T-cell priming. Other formulations are required for intravenous administration to deliver mRNA to secondary lymphoid organs body-wide, including the spleen as the largest agglomeration of immune cells. Formulation technologies developed for intravenous administration not only protect RNA integrity, but also promote its targeted expression in lymph node-resident DCs.

mRNA lipoplexes (RNA-LPX) are based on cationic liposomes as broadly used for transfection reagents in vitro [102]. Specific delivery of nanoparticles to APCs and improved uptake for presentation of epitopes derived from the encoded antigens can be achieved by decorating such particles with targeting moieties, such as antibodies, single chain variable fragments (scFv) fragments and small molecules binding DEC-205, mannose receptors and C-type lectin domain family 9 (CLEC9A) (reviewerd in Phuha et al. [103]; [104, 105]). Alternatively, RNA-LPX can be modified to have a negative net charge when prepared by molar excess of negatively charged mRNA, or by shielding positive surface charges with PEG. mRNA-LPX with a negative net charge are subject to specific uptake by APCs in vivo without the need for targeting moieties (reviewed in Guan et al. [106]).

Intravenously administered negatively charged to near-neutral RNA-LPX vaccines have been shown in mice to target APCs resident in lymphoid organs, including the spleen, lymph nodes and bone marrow, while inducing strong local (cell-specific activation) and systemic (proinflammatory cytokine abundance in the circulation) type I IFN-dominated immune modulation [40]. In several mouse tumor models, RNA-LPX vaccines induce strong effector and memory T-cell responses directed against the encoded TSAs or TAAs, and mediate tumor growth control or rejection, and improve survival [40, 107‐109]. In a phase I clinical trial (NCT02410733), 119 melanoma patients were vaccinated with an RNA-LPX vaccine encoding the four melanoma TAAs MAGE-A3, transmembrane phosphatase with tensin homology (TPTE), NY-ESO-1 and tyrosinase [40, 41]. In the subset of antagonistic anti-programmed cell death protein-1 (PD-1) antibody-experienced patients with evaluable disease at baseline (n = 25), vaccination alone mediated one metabolic complete response, three partial responses and seven cases of stable disease. In combination with anti-PD-1 antibodies (pembrolizumab or nivolumab, n = 17), six partial responses and two cases of stable disease were observed. RNA-LPX vaccine-induced CD4⁺ and CD8⁺ T-cell responses were observed in the majority of patients reaching up to low double-digit percentages of total circulating CD8⁺ T cells. A randomized phase II study will investigate this melanoma TAA-specific RNA-LPX vaccine in combination with the antagonistic anti-PD-1 antibody cemiplimab (NCT04526899). The same platform encoding other tumor type specific TAAs is being evaluated in various phase I and I/II clinical trials in patients with triple negative breast cancer (TNBC; NCT02316457), ovarian cancer (NCT04163094) and prostate cancer (NCT04382898).

Moreover, several clinical trials are testing this intravenous RNA-LPX technology in conjunction with TSAs, which are expected to be less compromised by central immune tolerance. One example is a human papilloma virus (HPV)16 E6 and E7 mRNA-encoding RNA-LPX vaccine, which is being evaluated in a randomized phase II trial in combination with pembrolizumab in patients with HPV16⁺ and PD-L1⁺ head and neck squamous cell carcinoma (HNSCC) (NCT04534205). Another example is the so-called individualized neoantigen-specific immunotherapy (iNeST) program, which treats each patient with their ‘on-demand’ manufactured individualized set of cancer mutations. This truly personalized treatment approach is being tested in various tumor types and treatment lines in the phase I and II studies NCT03289962, NCT03815058, and NCT04486378.

The same RNA-LPX vaccine platform was utilized to expand chimeric antigen receptor (CAR)-engineered T cells against claudin 6 (CLDN6), a carcinoembryonic cell surface antigen, in mice [110]. Vaccination with CLDN6-encoding RNA-LPX resulted in the expression of CLDN6 on splenic DCs and macrophages in vivo and increased the proportion of proliferating CLDN6-CAR-T cells after adoptive transfer. Subtherapeutic doses of CLDN6-CAR-T cells were expanded by the CLDN6 RNA-LPX vaccine, resulting in regression of subcutaneous CLDN6⁺ Lewis lung tumors. Safety and efficacy of this concept are currently being evaluated in a phase I/II clinical study in patients with solid cancers (NCT04503278).

Lipid nanoparticles (LNPs) are mRNA formulations based on a mixture of cholesterol, ionizable and helper lipids, and PEG-derivatives [106]. Intramuscularly administered RNA-LNPs have recently gained attention, as this is the format of the first two SARS-CoV-2 vaccines to be approved, based on their demonstration of compelling efficacy in COVID-19 protection [22, 111‐117], which was in line with previous preclinical and clinical data showing induction of strong antigen-specific immune responses against other infectious diseases [118‐120]. Both BNT162b2 and mRNA-1273 contain N¹-methylpseudouridine-modified mRNA with blunted innate immune modulatory activity to promote potent T_H1, T follicular helper and germinal center B cell responses for the induction of neutralizing antibodies as part of the multi-effector immune response [115, 121, 122].

In the cancer therapeutic field, intramuscularly administered, LNP-formulated nucleoside-modified mRNA vaccines encoding 20 patient-specific mutated neoepitopes are currently being evaluated in combination with pembrolizumab in patients with solid cancer (NCT03313778, phase I) and melanoma (NCT03897881, randomized phase II). Data from a recent phase I clinical trial in four gastrointestinal cancer patients (NCT03480152) suggest that vaccination with neoantigen-specific RNA-LNP is safe and induces mutation-specific T-cell responses [123] with no objective clinical responses observed yet in the limited number of patients tested.

Besides the clinically explored mRNA formulations discussed above, a plethora of different in vivo DC-targeting mRNA nanoparticles are under development. A recent study analyzed over 1000 mRNA lipid formulations for transfection efficacy. A subset of these were tested for in vivo protein production, APC maturation in local lymph nodes, induction of T and B cell responses in naïve, and anti-tumor efficacy of Trp2, OVA and E7-encoding mRNA lipid formulations in B16, B16-OVA and TC-1 tumor-bearing mice [124]. The formulations performing best shared a common structure involving an unsaturated lipid tail, a dihydroimidazole linker and a cyclic amine head group and mediate APC maturation by signaling through the STING pathway. Insights such as these will promote rational approaches for formulation development.