For influenza virus, CD8+ T cells initiate cytotoxic responses early in the infection by recognizing highly conserved epitopes of the viral nucleoprotein (NP) and matrix protein (M1). This leads to the direct lysis of virus-infected respiratory epithelial cells, significantly reducing viral load and the risk of severe disease (15, 16). These epitopes, due to their involvement in the assembly of the virus’s core structure, are less likely to undergo mutations, making CD8+ T cell responses have cross-strain protective potential. For example, cross-protection against different subtypes such as H1N1 and H5N1 can be mediated by these responses (17). An experiment using a single-cycle influenza vaccine (S-FLU) in female mice aged 6 to 8 weeks has confirmed that NP-specific CD8+ TRM cells deposited in the respiratory tract can recognize conserved epitopes from at least 12 different influenza strains. Their polyclonal nature ensures that mutations in a single epitope cannot completely escape immune surveillance (15, 18). At the same time, Th1 cells activate the antiviral state of alveolar macrophages by secreting cytokines such as IFN-γ. This not only directly inhibits the function of the viral polymerase but also upregulates the expression of MHC-I molecules on infected cells, enhancing the epitope presentation efficiency to CD8+ T cells (17, 19) ( Table 1 ).

For SARS-CoV-2, the pressure to escape cellular immunity primarily comes from the CTL response targeting the S2 subunit of the spike protein (S protein), a region that plays a critical role in viral membrane fusion and, therefore, is constrained in terms of mutations (20, 21). Whole-genome T cell epitope scanning shows that approximately 60% of CD8+ T cell responses target non-structural proteins, such as ORF1ab. These regions are conserved in more than 85% of the Omicron variant (22, 23). Structural studies indicate that HLA-A*02:01-restricted epitopes of the spike protein (such as S269-277) can induce potent perforin-granzyme release. Even in the presence of key mutations like K417N and E484K, these epitopes maintain over 70% TCR recognition efficiency (24, 25). Respiratory TRM cells localize to the nasopharyngeal-associated lymphoid tissue (NALT) through CXCR3-mediated homing. Their phenotypic characteristics (CD69+CD103+) are similar to those of EBV-specific TRM cells, but their function is somewhat weaker. This may explain the phenomenon of repeated upper respiratory tract infections (26, 27). It is noteworthy that SARS-CoV-2 employs three molecular strategies to escape CTL surveillance: first, by using the NSP7 protein to mimic human T cell epitopes (for example, NSP795–102 shares 67% similarity with human PPIA), thereby achieving immune camouflage; second, by mutating the furin cleavage site of the spike protein to interfere with antigen processing; and third, by utilizing the ORF8-mediated MHC-I degradation pathway to systematically downregulate antigen presentation (25, 28, 29). Hybrid immunity (post-infection mRNA vaccination) induces the broadest TRM response. These cells exhibit unique transcriptional characteristics—high expression of TCF7 to maintain stem cell-like properties, and upregulation of KLF2 to enhance tissue residency. As a result, their survival time in bronchoalveolar lavage fluid is extended to more than 12 months after infection (30, 31). Single-cell TCR sequencing has also revealed that CD8+ T cell clones targeting SARS-CoV-2 exhibit a “layered cross-reactivity” feature: the foundational clones recognize conserved epitopes from seasonal coronaviruses, while the dominant clones specifically target unique epitopes of SARS-CoV-2. This dual recognition architecture significantly enhances the breadth of defense against variants (32, 33).

The core challenge in the development of respiratory syncytial virus (RSV) vaccines lies in the immune pathology induced by a Th2-type immune bias. This bias is characterized by the excessive secretion of cytokines such as IL-4, IL-5, and IL-13 following vaccination, leading to pathological features such as eosinophil infiltration, airway hyperreactivity, and excessive mucus production (34, 35). This phenomenon was particularly evident in the 1960s formalin-inactivated RSV vaccine (FI-RSV) clinical trials, which manifested as exacerbated pulmonary inflammation and defective immune memory (36, 37). This pathological reaction may stem from the damage to viral surface proteins (e.g., PreF conformation) during vaccine preparation, leading to impaired innate immune recognition (e.g., TLR signaling) and aberrant Th2 polarization (38, 39). Notably, Th2 cytokines (e.g., IL-4, IL-5) not only promote eosinophil infiltration but also synergize with the IL-17 pathway—Th17 cells further amplify the inflammatory cascade through IL-17A/F secretion. This imbalance in T-cell subsets (Th2/Th17 dominance) has been confirmed in RSV infection models to exacerbate tissue damage (40–43). However, protective immunity requires coordinated T-cell responses: CD8+ T cells can directly clear virus-infected cells, while Th1-type responses (IFN-γ-dominated) and regulatory T cells (Tregs) can suppress excessive inflammation (44–46).

In terms of cellular immunity durability and cross-protection, tissue-resident memory T cells (TRM) and central memory T cells (TCM) form a dual barrier against RSV reinfection. TRM cells (such as CD103+ CD69+ CD8+ T cells) reside in the respiratory mucosa, and their presence has been directly linked to reduced viral load in experimental RSV infections (47, 48). Their expansion depends on type I interferon signaling mediated by the MAVS and MyD88/TRIF pathways (49–51). In MAVS-deficient mice, TRM numbers can be restored to 80% of wild-type levels following IFN-α treatment (52). Single-cell RNA sequencing reveals that TCM cells initiate a transcriptional reprogramming process to differentiate into TRM cells within 2 days following intranasal LAIV booster immunization. This process involves the recruitment of chemokines dependent on CXCR3 and the upregulation of CD8+ TRM characteristic genes, such as Itgae and Cxcr6 (53). Circulating memory T cells maintain systemic immune surveillance through the TCM (CD62L+ CCR7+) and effector memory T cell (TEM) subsets. Under stimulation with RSV F protein peptides, PBMCs from uninfected individuals can generate a stable memory T cell response that lasts for more than 10 years (54). Long-lived TRM and circulating memory T cells (TCM) provide rapid recall responses upon re-exposure to the pathogen.

Epitope conservancy is key to achieving cross-protection. The preF conformation epitopes of the RSV F protein (such as antigenic sites Ø and V) are conserved by more than 95% between the A and B subtypes (55). In contrast, the mutation rate of T cell epitopes (such as M187–195 of the F protein) is only one-third of that of B cell epitopes (56). This makes it an ideal target for broad-spectrum vaccines. Multi-epitope vaccine design, by integrating HLA supertype epitopes of the F protein (such as the DRB104:01-restricted CD4+ epitope F254–268 and the HLA-A02:01-restricted CD8+ epitope F85-93), can simultaneously stimulate a balanced Th1/Th2 antibody response and CTL activity in mouse models (56, 57). It is worth noting that the multi-epitope display virus-like particles (VLP) based on the Round Leaf Bat Hepatitis Core Antigen (RBHcAg) can induce cross-neutralizing antibodies against RSV A2, B18537, and the clinical isolate hRSV/C-Tan/BJ 202301. The potency of these antibodies is comparable to that of approved vaccines (57, 58).

In contrast, traditional inactivated vaccines induce an immune response using inactivated virus particles. These vaccines inactivate the virus through physical or chemical methods, rendering it unable to cause infection (72). The vaccine contains inactivated whole viruses or viral components, and after injection, the viral antigens are directly presented, triggering an immune response. The antigens of inactivated vaccines enter cells primarily via the exogenous pathway, where they are taken up and processed, and then presented to CD4+ T cells through MHC class II molecules. This typically leads to the activation of Th2 cells. Th2 cells secrete cytokines that promote humoral immunity (such as IL-4 and IL-5), stimulating B cells to produce antibodies, particularly IgE antibodies, thereby enhancing the humoral immune response (73–75). However, the cellular immune response induced by inactivated vaccines is weaker because the antigens are mainly presented via MHC class II, which is insufficient to activate CD8+ T cells (76). Because inactivated vaccines typically induce a weaker cellular immune response, the quantity and functionality of the generated memory T cells may be insufficient. Compared to the strong immune responses induced by mRNA vaccines, inactivated vaccines may have lower durability and functionality of memory T cells (77). In addition, the persistence of antigens from inactivated vaccines in the body is relatively short, which may affect the formation and maintenance of memory T cells. The development and production cycles for traditional inactivated vaccines are longer, and the costs are higher. However, their safety profile is excellent, as the viruses in inactivated vaccines have lost their infectivity, ensuring that no infection is triggered after vaccination. Therefore, inactivated vaccines are highly safe and suitable for various populations, especially those with weakened immune systems (73, 78–80). The immune response induced by inactivated vaccines is primarily focused on humoral immunity, making them effective at inducing antibody responses, providing short-term protection against the virus. However, they are less effective at inducing cellular immunity, particularly in terms of activating CD8+ T cells (81).

Adenoviral vaccines deliver target antigen genes into host cells using non-pathogenic viral vectors, efficiently activating multilayered cellular immune responses. The core mechanism lies in the ability of the viral vector to rapidly recognize receptors on the host cell surface, such as the coxsackievirus and adenovirus receptor (CAR) (82). The vector then enters the cell via endocytosis or membrane fusion, followed by antigen protein expression in the cytoplasm (83, 84). These antigens are degraded into peptides by the proteasome via the endogenous pathway, then bind to major histocompatibility complex class I (MHC-I) molecules and are presented on the cell surface, directly activating CD8⁺ T cells to differentiate into cytotoxic T lymphocytes (CTLs) (85). CTLs eliminate virus-infected cells directly by releasing perforin and granzymes, while also establishing a long-lasting pool of memory T cells to respond to future infections (86, 87). In addition, antigen-presenting cells (such as dendritic cells) transfer antigens to MHC class II molecules through the classical exogenous pathway, activating CD4⁺ T helper cells (Th1 subtype). These cells secrete cytokines such as interferon-gamma (IFN-γ), which further enhance the cytotoxic function of CD8⁺ T cells and promote macrophage activation (88, 89). The direct cytotoxicity of CD8⁺ CTLs and the immunoregulatory function of CD4⁺ Th1 cells work synergistically, not only enhancing antigen delivery efficiency (90, 91), but also inducing long-lasting immune memory by mimicking the natural infection pathway, thereby conferring (92).

Novel nanoparticle vaccines efficiently activate cellular immune responses through their unique delivery systems and multifunctional designs. Nanoparticles (50–250 nm) enhance the uptake efficiency by antigen-presenting cells (APCs) through size-dependent effects and surface charge modulation, such as positive charge modifications (93, 94). Passive targeting relies on the enhanced permeability and retention (EPR) effect to accumulate in inflammatory or lymphoid tissues (94, 95). Active targeting involves surface modification with antibodies or peptide ligands (such as DC-SIGN ligands) to precisely recognize APC surface receptors (96). After the particles enter the cell, pH-sensitive materials [such as poly (β-amino esters)] disassemble in the acidic environment of the endosome, releasing the antigen into the cytoplasm and promoting cross-presentation by MHC-I molecules, which directly activates CD8⁺ T cells to differentiate into cytotoxic T lymphocytes (CTLs) (97, 98). If the antigen enters the lysosome, it activates CD4⁺ Th1 cells via MHC-II molecules, which secrete IFN-γ to enhance CTL function. Sustained-release designs (such as PLGA degradation control) can extend antigen exposure for up to 28 days, continuously stimulating the generation of memory T cells. Nanocarriers can co-deliver antigens and adjuvants (such as TLR3/7/9 agonists), enhancing synergistic effects through spatiotemporal synchronized delivery. Furthermore, nanoparticles surface-modified with mannose target APC surface C-type lectin receptors, improving drug delivery specificity, bioavailability, and therapeutic efficacy, while reducing off-target effects and systemic toxicity (99). Multifunctional nanocarriers achieve microenvironmental regulation through the co-delivery of immunomodulatory factors (100, 101). For example, PLGA nanoparticles loaded with OVA antigen and rapamycin (an mTOR inhibitor) can induce the differentiation of regulatory T cells (Tregs) and suppress Th17-mediated inflammatory responses. Th17-associated pro-inflammatory factors (IL-17, IL-1β, IL-12) are significantly reduced in the PLGA-Rapa treatment group, with some literature reporting a decrease in IL-17a of over 50% (102, 103). The ratio of anti-inflammatory factors (TGF-β1, IL-10) increases, with the secretion of TGF-β1 in the Rapa&P-50k group approximately doubling (104). In the OVA inflammation model, PLGA nanoparticles (such as IL10-AMNP) increase the proportion of Tregs and decrease the proportion of Th17 cells. The exact proportions vary depending on the model, but the trend remains consistent (105, 106). At the same time, it maintains antiviral CTL activity, achieving a balance between therapy and protection (107). This ‘immune switch’ design provides a precise intervention strategy for chronic infections or autoimmune diseases ( Table 2 ).

IFN-γ (interferon-gamma) is a key cytokine secreted by immune cells such as CD8⁺ T cells and NK cells, and its secretion level is an important indicator for evaluating vaccine-induced cellular immune responses. IFN-γ plays a central role in anti-infection and anti-tumor immunity by activating macrophages, enhancing antigen presentation, promoting Th1-type immune responses, and directly inhibiting viral replication or tumor growth (108–110). In vaccine immunization, high levels of IFN-γ secretion are typically associated with stronger immune protection, especially against pathogens that require cellular immunity for clearance, such as viruses or intracellular bacteria (111, 112). mThe IFN-γ secretion by CD8⁺ T cells induced by the RNA vaccine (BNT162b2) was significantly higher than that induced by the inactivated vaccine (BBIBP-CorV) (2.02 vs. 0.24 IU/mL), indicating that mRNA vaccines induce a stronger multifunctional CD8⁺ T cell response. mRNA vaccines are significantly superior to inactivated vaccines in activating Th1-type immune responses. Another study showed that CD8⁺ T cells induced by mRNA vaccines have a broader epitope coverage and are associated with reduced viral load in the upper respiratory tract (113, 114). This difference primarily arises from the distinct antigen designs and immune activation mechanisms of the two vaccines. mRNA vaccines (such as BNT162b2) encode a single target protein (such as the SARS-CoV-2 spike protein), efficiently activate dendritic cells, and directly present antigens through the MHC I pathway, thereby more strongly stimulating CD8⁺ T cell differentiation into effector cells and the secretion of IFN-γ (70, 112, 115). In contrast, although inactivated vaccines can induce CD8⁺ T cell responses targeting multiple viral proteins (such as S, N, and M proteins), their antigen presentation efficiency is lower and relies on cross-presentation pathways, leading to relatively weaker IFN-γ secretion by CD8⁺ T cells (70, 115, 116). Another comparative study showed that the IFN-γ secretion by CD8⁺ T cells induced by the mRNA vaccine is twice that of the inactivated vaccine (70). Moreover, the response is more focused on the S protein epitopes (115). In addition, the lipid nanoparticle (LNP) delivery system of mRNA vaccines may directly enhance T cell activation, further promoting IFN-γ production (117, 118). These differences suggest that mRNA vaccines have an advantage in stimulating Th1-type immunity and CD8⁺ T cell-mediated cytotoxicity.

Viral vector vaccines (such as AZD1222) maintain moderate CD8⁺ T cell activity 28 days after vaccination, while the response from inactivated vaccines rapidly declines. The advantage of viral vector vaccines (like AZD1222) in inducing memory T cell durability compared to inactivated vaccines is primarily due to their unique antigen presentation methods and immune activation mechanisms. Viral vector vaccines mimic the natural infection process, continuously expressing the target antigen [for example, adenoviral vectors can infect fibroblasts over the long term) (119)]. This results in the phenomenon of ‘memory inflation,’ where effector memory CD8⁺ T cells (TEM) continue to expand and maintain high numbers in peripheral tissues (120). This persistent antigen stimulation is achieved through two mechanisms: first, viral vectors (such as adenoviruses or cytomegalovirus vectors) can persist in host cells for extended periods and express the antigen at low levels (121, 122). Second, it preferentially targets antigen-presenting cells such as dendritic cells, promoting cross-presentation and activating CD8⁺ T cells (123). In contrast, the antigen of inactivated vaccines has a short-lived effect (78). It cannot form persistent stimulation, leading to a rapid decline in CD8⁺ T cell responses. Viral vector vaccines also enhance durability by inducing tissue-resident memory T cells (TRM), which remain distributed in various tissues months after vaccination (121, 124). In contrast, inactivated vaccines primarily induce circulating memory T cells (10). In addition, viral vector vaccines can more effectively activate cytokine pathways such as IL-15, promoting the early generation of memory precursor cells (such as Tscm) (125). These stem cell-like memory T cells have self-renewal capabilities and can maintain the memory pool for a long time (126). Phenotypically, viral vector-induced CD8⁺ T cells exhibit high expression of cytotoxic-related genes (127). They also exhibit higher functional affinity (128). The T cell subset distribution induced by inactivated vaccines is more limited (78). It is worth noting that adenovirus vector vaccines can also drive the expansion of innate-like CD8⁺ T cells by inducing host IL-15 production (122). This mechanism has not been reported in inactivated vaccines. Taken together, viral vector vaccines achieve more durable memory T cell maintenance compared to inactivated vaccines through multiple mechanisms, including persistent antigen exposure, tissue-resident memory formation, cytokine signaling activation, and the induction of stem cell-like memory subsets.

There are significant differences in cellular immune responses to vaccination across different age groups. Adults generally exhibit the strongest T-cell responses, while infants and the elderly show lower reactivity due to the unique states of their immune systems. In infants, the immune system is still under development, with a higher proportion of naive T cells that are functionally immature, and limited capacity of antigen-presenting cells. This results in a significant reduction in the proliferation and effector function of CD4+ and CD8+ T cells after vaccination compared to adults. In contrast, elderly individuals experience immunosenescence, characterized by a significant reduction in the naive T cell pool. Immunosenescence weakens T cell activation and sensitivity to antigens, reduces the diversity of memory T cells, and chronic inflammation (inflammaging) further suppresses vaccine-induced cellular immune responses, thereby significantly affecting vaccine efficacy in older adults (173–175). Additionally, studies have found that elderly individuals exhibit significantly lower CD8+ T cell reactivity after receiving the pertussis vaccine compared to younger individuals, and the functionality of effector T cells is also impaired (175, 176). A study on age and immune response (with CMV infection as an immune-related phenotype) showed that the CMV positivity rate was 42.86% in young individuals with an average age of 25.3 years (aged 20–31 years at enrollment) and 55.56% in elderly individuals with an average age of 76.96 years (aged 60–96 years at enrollment). According to the IMM-AGE score, the immune aging index of a 90-year-old was approximately 3.2 times higher than that of a 66–67-year-old. The final results indicated that immune senescence begins in mid-adulthood (ages 40–60) (177). A study by Wu et al. showed that the antibody titers (GMT values) induced by the mRNA vaccine were significantly higher in younger individuals compared to the elderly. Specifically, 28 days after vaccination, the GMT for younger individuals was 302.9, while the GMT for the elderly was 173.2 (178). These studies suggest that the immune system status at different age stages significantly affects the cellular immune response induced by vaccines. Both infants and the elderly face greater immune response deficiencies, highlighting the need to specifically optimize vaccination strategies for these populations.

Immunocompromised and chronic disease patients often exhibit weakened cellular immune responses following vaccination, which poses significant challenges to vaccine efficacy and clinical application. In immunocompromised patients, such as those who have undergone organ transplants, the long-term use of immunosuppressive drugs significantly reduces their T-cell-mediated immune response, leading to a weaker vaccine-induced cytotoxic T lymphocyte (CTL) response (179, 180). Studies have shown that kidney transplant recipients exhibit CD4+ and CD8+ T cell responses after receiving the mRNA COVID-19 vaccine that are only 30%-50% of those seen in healthy individuals. Chronic kidney disease patients also face similar issues, with research indicating a reduction in memory T cell function and lower protection after vaccination compared to healthy controls (181). In chronic disease patients, such as those with diabetes, vaccine-induced cellular immune responses are also significantly affected due to chronic inflammation and immune system dysfunction (182). These characteristics lead to a higher infection risk for these patient groups when facing emerging viruses like SARS-CoV-2. Even after completing vaccination, additional preventive strategies, such as booster doses and passive immunotherapy, should be considered. Yang indicated that the third or fourth dose of the SARS-CoV-2 vaccine significantly improved the immunogenicity rate in dialysis patients, and this beneficial effect was not altered by vaccine type (same or different immunogenic vaccine), dialysis method (HD or PD), or prior low response after two doses of the vaccine (183).