mRNA癌症疫苗的免疫调节框架

Fig. 1 | An mRNA cancer vaccine cycle. The components of a messenger RNA (mRNA) vaccine are illustrated, comprising mRNA and the four components of a lipid nanoparticle (LNP): ionizable lipid, cholesterol, helper lipid and polyethylene glycol lipid. The LNP encapsulates the mRNA to prevent its degradation in the bloodstream. Upon endocytosis of the mRNA cancer vaccine by an antigen-presenting cell (APC) (1), the LNP facilitates endosomal release of the mRNA into the cytoplasm (2), where it is translated to protein (3). This protein can be secreted to act as an exogenous source of antigen, or it can be processed intracellularly for peptide loading onto MHC molecules (4). The result is antigen presentation at the cell surface and activation of adaptive immune cells (5). The mRNA cancer vaccine itself is not immunologically ‘silent’ as the LNP and mRNA components can activate innate immune sensing pathways (6), such as Toll-like receptor 4 (TLR4), TLR7, TLR8, complement, inflammasomes and RIG-I. The extent of activation is governed by factors outlined in axis 1 (adjuvanticity vs immunopathology) of our immunological framework (see main text and Fig. 2), such as manufacturing, LNP composition and mRNA chemistry. Activation of these innate immune pathways results in the release of type I interferon (IFN) into the tumour microenvironment, which, in addition to other inflammatory functions, facilitates metabolic reprogramming of the APC to favour aerobic glycolysis and production of acetyl-Co-A (7). These metabolic changes are necessary for the induction of trained immune responses. TCR, T cell receptor.

Fig. 2 | An immunological axes framework for mRNA cancer vaccines. The four panels illustrate the four immunological axes that determine the efficacy of messenger RNA (mRNA) cancer vaccines: innate immune sensing, antigen presentation, adaptive immune memory and trained immunity. Within each axis, opposing immunological forces operate: adjuvanticity vs immunopathology, immunogenicity vs immune tolerance, immune memory vs immune exhaustion, and beneficial vs maladaptive trained immunity. The central box within each panel represents the programmable design features of mRNA cancer vaccines that can be modified to balance each axis. Axis 1: innate immune sensing through pattern recognition receptor (PRR) signalling, for example, on dendritic cells (DCs), influences the balance between immune activation (adjuvanticity) through antigen presentation to T cells and immunopathology, for example, through excess production of pro-inflammatory cytokines such as interferon-γ (IFNγ). This axis can be tuned through lipid nanoparticle (LNP) composition including the density of PEGylated lipids, mRNA chemistry (mainly nucleoside modification), vaccine manufacturing protocols and addition of vaccine adjuvants. Axis 2: antigen presentation to T cells regulates immunogenicity vs immune tolerance, which can be tuned by the route of administration, by provision of costimulatory ligands or by antibodies and LNPs that target the vaccine construct to professional antigen-presenting cells (APCs) such as DCs rather than non- professional APCs such as fibroblasts. Axis 3: the adaptive immune response can result in either memory or exhaustion, which can be modulated by tuning antigen immunodominance or epitope spreading through LNP composition and mRNA chemistry. Anti-carrier B cell responses, such as polyethylene glycol (PEG)-specific antibodies, can promote antigen clearance, which blunts memory responses. Axis 4: trained immunity, mediated by epigenetic and metabolic changes in innate immune cells, can either enhance or suppress beneficial (or suppressive) programmes of immune gene expression (such as CXCL13, which is required for productive B cell responses). These pathways can be modulated using heterologous prime–boost strategies or alternative combination therapies and depend on the vaccine dose and half-life, the route of administration and the tumour microenvironment. dsDNA, double-stranded DNA.

Table 1 | Programmable immunological axes in mRNA cancer vaccine design. Full table, title and abbreviations are included in the archived image.

Fig. 3 | Tuning innate immune sensing for balanced adjuvanticity. This figure illustrates the concept of an ‘optimal zone’ of innate immune activation that can be achieved through design considerations of the messenger RNA (mRNA) vaccine. Factors such as high ratios of cationic lipids in lipid nanoparticles (LNPs), unmodified RNA and contaminating double-stranded RNA (dsRNA) in the mRNA vaccine can result in excessive immune activation, causing toxicities such as cytokine release syndrome. By contrast, immune tolerance can be induced through vaccine designs that have a high content of polyethyleneglycol- modified (PEGylated) lipids or high levels of saturated helper lipids (such as 1,2-distearoyl-sn-glycero-3-phosphocholine) in the LNP or that target antigen to non-professional antigen-presenting cells (APCs). The ‘optimal zone’ of activation, which will differ depending on tumour, patient and immune context, can be achieved through the inclusion of costimulatory signals such as 4-1BBL or cytokines (IL-7 and IL-21), pseudouridine modification of the mRNA, or LNP design choices such as mannosylated lipids to target conventional type I dendritic cells (cDC1s) or ionizable lipids to promote Toll-like receptor 4 (TLR4) signalling.

Fig. 4 | Perioperative prime–boost mRNA vaccine model. A hypothetical example of a heterologous messenger RNA (mRNA) cancer vaccination strategy involving the neoadjuvant administration of a universal mRNA cancer vaccine followed by surgical resection of the tumour and adjuvant administration of an mRNA vaccine encoding a personalized neoantigen. Following surgical resection, the tumour could be immunophenotyped to guide design of the secondary, personalized neoantigen vaccine. For example, for immunologically ‘cold’ tumours, vaccine inclusion of mRNA encoding the cytokine IL-12 together with self-amplifying mRNA encoding the neoantigen could be used to increase costimulation and antigen persistence, respectively, and enhance antitumour immunity. By contrast, immunologically ‘hot’ tumours with chronic inflammatory signalling could benefit from lipid nanoparticle (LNP) modifications that reduce pattern recognition receptor signalling, such as a high concentration of polyethylene glycol-modified (PEGylated) lipids.