In response to pathogen infection or antigen exposure, naive T cells primed by antigen-presenting cells proliferate and differentiate into functional effector T cells. Following elimination of immunologic threat, major effector cells (about 90-95%) die during contraction phase, and only a small proportion survives and develops into long-lived memory T (T_M) cells which are capable of self-renewal and surviving in the absence of further antigen stimulation ( Figure 1 ) (1–3). CD8⁺ and CD4⁺ T_M cells can be typically subdivided into CD62L⁺CCR7⁺central memory T (T_CM) cells, CD62L^-CCR7^- effector-like memory T (T_EM) cells and CD69⁺CD103^+or- tissue-resident memory T (T_RM) cells based on their functions and migration patterns (4, 5). As T_M cells play critical roles in protective immunity and immunopathology, elucidating mechanisms underlying their generation and maintenance is essential for the design of future vaccines capable of eliciting T cell-based immunity (6).

The core theories of memory generation are primarily derived from studies on differentiation of CD8⁺ T_M cells (1, 7). The identification of CD127^hiKLRG1^lo memory precursor effector cells (MPECs) and CD127^loKLRG1^hi short-lived effector cells (SLECs) at the peak of primary response (day 7–8 post-infection) provides a guiding framework for a deeper understanding of the fate decision between effector and memory CD8⁺ T cells ( Figure 2 ) (8–11). TCF1^hi cells within CD127^hi MPEC pool, which exhibit stem-like properties and undergo less cytotoxic differentiation, substantially give rise to T_CM cells, while TCF1^loCD127^hi population contracts and becomes T_EM cells following infection elimination (3, 12–14). Preferential localization of TCF1^hi T_CM precursors in paracortex (T cell zone) of secondary lymphoid organs (SLOs), through CCR7-mediated chemotaxis toward CCL19/21, facilitates their encounter with IL-7, thus enhancing their transition into CD8⁺ T_CM cells ( Figure 2 ) (15–20). Meanwhile, T-bet-induced CXCR3 drives the migration of activated CD8⁺ T cells to peripheral region of SLOs and inflamed non-lymphoid tissues (NLTs), facilitating their interaction with inflammatory signals and thereby promoting their differentiation into CD127^loKLRG1^hi terminal effector cells (9, 19, 21). CD69⁺CD103⁺ CD8⁺ T_RM cells are suggested to derive from a MPEC-like (CD127^hi) population residing in NLTs, and Hobit and Blimp-1 potentially identify T_RM precursors in different NLTs (22–25). Moreover, early-activated TCF1^hi and TCF1^lo cells (day 2–4 post-infection), which already possess distinct memory potential, exhibit reversible plasticity between effector and memory fates in response to inflammatory stimulation or the withdrawal of such stimulation (12, 14, 26, 27). The differentiation fate of antigen-activated CD8⁺ T cells between effector and memory is collaboratively determined by intrinsic factors, extrinsic factors and positioning profiles that facilitate intrinsic-extrinsic interactions during the whole priming phase under acute infection (1, 3, 28–31).

Though there is no unifying framework, the current understanding of CD8⁺ T_M cell generation provides a valuable guidance for investigations into differentiation of CD4⁺ T_M cells. However, much less is known about generation of CD4⁺ T_M cells and their precursors to date (5, 32). Unlike CD8⁺ T cells, antigen-activated CD4⁺ T cells can differentiate into Th1, Th2, and Th17 effector cells, which are typically identified by secretion of signature cytokines, under different types of infections and exposures (33). Th1 cells, characterized by IFNγ secretion, are induced by the master transcription factor T-bet in response to intracellular viral and bacterial infections. IL-4-secreting Th2 cells are promoted by transcription factor Gata3 upon exposure to extracellular parasites/helminths and allergens. The differentiation of IL-17-secreting Th17 cells depends on expression of transcription factor RORγt in response to fungal and extracellular bacterial infections. Additionally, Th1/2/17 responses are all associated with Bcl6-dependent parallel differentiation of CXCR5⁺PD-1⁺ follicular helper T (Tfh) cells, which play pivotal roles in promoting B cell responses within germinal center (GC) (34). Th1/2/17 and Tfh cells both could survive contraction and become T_M cells after immunological threat is eliminated (5, 35–39). The heterogeneity of activated CD4⁺ T cells adds complexity to studying the generation of CD4⁺ T_M cells.

Drawing on insights from the theory of CD8⁺ T_M cell generation, this paper presents current advances in identifying precursors of Th1, Th2, Th17, and Tfh memory cells, as well as elucidating the mechanisms underlying fate decision of CD4⁺ T_M cells. We also discuss the challenges and prospects for identifying common precursor of specific CD4⁺ T_M cells under various types of infections and exposures.

In comparison to Th1 cells, memory generation of Th2 cells has been much less studied. Th2 memory cells, including T_CM cells in lymph nodes, T_EM cells, T_RM cells in lungs, and Tfh-like memory cells all can be generated after antigen clearance under helminth/parasite infection (63, 87, 99, 100). However, Th2 and Tfh-like cells in LNs exhibit great plasticity under parasite infection, and a proportion of activated CD4⁺ T cells appear to co-express Tfh marker Bcl6/CXCR5 and Th2 marker Gata3/IL-4 (101, 102). Meanwhile, entry of activated CD4⁺T cells into lungs, which promotes terminal IL-4⁺IL-13⁺ Th2 effector cell differentiation, depends on downregulation of the hallmark Tfh transcription factor Bcl6, indicating a distinction between them (99, 101). The lack of clear delineation between Th2 and Tfh cells complicates the investigation into their memory generation under Th2 conditions.

Without considering the existence of Tfh cells, antigen-activated CD44^hiCD62L⁺ CD4⁺T cells, which preferentially accumulate in LNs rather than lungs, are suggested to enrich T_CM precursors (103, 104). CD44^hiCD62L⁺ cells, with high co-expression of CCR7, maintain a less differentiated state as indicated by lower expression of Gata3 and IL-4 (103–105). It is consistent with the observation that CCR7-mediated retention of activated CD4⁺ T cells in T cell zone inhibits Th2 effector differentiation (87, 106). Meanwhile, CXCR5-mediated migration towards peripheral region of LNs, in a CXCL13-dependent manner, is required for differentiation of both IL-4-producing Th2-like cells and Tfh cells (87, 106). It is indicated that positioning into periphery and center of SLOs also contributes to differentiation of Th2 effector cells and T_CM cells, respectively, during Th2 responses. On the other hand, retention of Th2 effector cells in NLTs facilitates their encounter with local extrinsic signals and thus promotes Th2-T_RM cell generation. Local IL-7 signaling is essential for maintenance of Th2-T_RM cells in lungs (105). IL-2/CD25 signaling has been shown to promote Th2-T_RM generation through enhancing optimal effector differentiation and residency of Th2 cells in lungs (99). CD69, which is highly expressed on Th2-T_RM cells, might be required for their generation through promoting lung residency of Th2 effector cells (99). Although the underlying mechanisms are not well-defined, these investigations indicate similarities in spatial requirements for memory generation between Th2 cells and Th1/CD8⁺ T cells. In addition, ablation of ACC1, which is required for Th1 memory cell formation, also enhances generation of Th2 memory cells via regulating fatty acid oxidation under helminth infection (65). It suggests that Th2 memory cell generation depends on some mechanisms shared with Th1 cells.

Despite the similarities mentioned above, it should not be simply presumed that factors involved in formation of CD8⁺ T_M and Th1 memory cells play the same role in Th2 memory cell generation. CD127, the key marker for identifying CD8⁺ T_M cell precursors, is barely expressed on Th2-like effector cells in LNs during priming and thus cannot mark Th2 memory precursors, although it has been shown to be essential for homeostasis of Th2 memory cells (104, 105). Ly6C and PSGL1, which are used to identify memory precursors from Th1 effector cells, are barely expressed on ovalbumin-specific Th2-like effector cells at the peak of Th2 responses (104). CXCR5, a key marker to distinguish between Th1 effector cells and Tfh/T_CM cells, is expressed on both Tfh and Th2 cells in lymph nodes during priming (87, 106). RNA-seq data also demonstrate that TCF1, Thpok and Id3, which are essential for T_CM and Th1-T_EM generation under viral infection, are comparably expressed on CD62L^- Th2 cells and T_CM precursor-enriched CD62L⁺ Th2 cells (104). Overall, independent investigations into generation of Th2 memory cells are definitely required.

Few investigations have been performed into memory generation during Th17 responses, while the heterogeneity of T_M cells and the existence of Tfh memory cells have remained largely unexamined. Th17 memory cells are shown to derive from IL-17⁺ or RORγT⁺ Th17-like effectors; however, how effector cells contribute to the ultimate Th17 memory cell pool remains ill-defined (107, 108). IL-7 and IL-15 have been shown to promote the maintenance of Th17 memory cells at inflammatory site and draining lymphoid tissues (109). IL-23/IL-23R signaling directly drives effector to memory conversion of Th17 cells via upregulation of CD127 and IL-15 receptor during contraction phase, while IL-2 prominently impairs IL-23-induced Th17 memory cell generation (110). It suggests that surviving signals are also required for Th17 memory generation. CD30, which promotes Th1 memory generation, plays critical role in generation of Th17 memory cells, and thus can serve as a prospective marker for Th17 memory precursors (64). Moreover, retention of Th17 cells in the lung and skin to interact with IL-1α and local IL-23 is required for CD69⁺ Th17-T_RM generation, although factors that mediate their residency remain unclear (111, 112). Though the information are fragmented, the differentiation of Th17 memory cells exhibits both similarities and differences with CD8⁺ T_M cells and Th1/2 memory cells.

Despite the above advances in characterizing precursors and elucidating fate decision mechanisms of CD4⁺ T_M cells, current knowledge remains insufficient. A deeper understanding of generation of CD4⁺ T_M cells has been hindered by several technological and biological challenges. The first challenge is to trace antigen-specific CD4⁺ T cells in vivo. TCR-transgenic T cell adoptive transfer system and peptide:MHC tetramer technology, which are vital tools for studying antigen-specific CD8⁺ T cells, both exhibit limitations in studying CD4⁺ T cells (5, 32). Moreover, the number and expansion capability of antigen-specific naïve CD4⁺ T cells is much lower than CD8⁺ T cells, and CD4⁺ T_M cells are suggested to be less stable over time (32, 42, 113, 114). Low number of antigen-specific CD4⁺ T_M cells makes them harder to be detected in vivo.

In addition to technical limitations, the functional and phenotypic heterogeneity of CD4⁺ T cells at effector and memory phase poses significant challenges to establishing a unifying framework for CD4⁺ T_M cell generation. As mentioned above, Th1/2/17 and Tfh populations both can give rise to T_M cells after immunological threat is eliminated (5, 35–39). Plenty of T cell intrinsic and extrinsic factors, which contribute to generation and maintenance of CD4⁺ T_M cells, are also involved in polarization of naïve T cells towards Th1/2/17 and Tfh lineages (5, 33, 39, 58, 115). For instance, TCF1 and Thpok, which are shown to be required for generation of T_CM cells, also promotes Tfh cell differentiation during Th1 responses (41, 52, 55, 58). Bcl6, which plays pivotal role in Tfh differentiation, is also required for T_CM and Th1-T_RM generation (58, 59, 94). Moreover, lack of a unifying approach to faithfully identify Th1/Th2/Th17 and Tfh populations complicates the study of CD4⁺ T_M cell differentiation. Th1, Th2 and Th17 cells are generally identified by secretion of IFN-γ, IL-4 and IL-17 respectively; however, Bcl6⁺ follicular helper T (Tfh) cells are also able to produce these hallmark cytokines (34). Besides, the two commonly used approaches, based on expression of Ly6C/PSGL1/FR4 and CXCR6/CXCR5/CCR7, both cannot unambiguously identify the diverse CD4⁺ T cell subsets during effector and memory phases of Th1 responses (32, 37, 39, 45). Expression of CXCR5 and production of IL-4 are also not able to discriminate between Th2 and Tfh cells under parasite infection (101, 102). The developmental relationships between Th/Tfh polarization and memory generation remain incompletely elucidated, largely due to shared mechanistic pathways and phenotypic overlaps between diverse CD4⁺ T cell subsets.

Furthermore, the plasticity of CD4⁺ T cell subsets at population level also increases complexity in studying CD4⁺ T memory generation (116). During Th1 responses, CXCR5⁺ T_M cells give rise to secondary Th1 and Tfh effector cells upon rechallenge, indicating heterogeneity or plasticity of them (44). Recent data demonstrate that CXCR5⁺ memory cells contain a T_CM-like (CCR7⁺ or Ly6C^loPSGL1⁺) population, which primarily gives rise to secondary Th1 effectors and generates few Tfh cells (41, 45, 117). However, remaining Tfh memory cells (Ly6C^loPSGL1^loFR4⁺ or CXCR5⁺CCR7^lo), which efficiently generate Tfh effector cells, equally give rise to secondary Th1 effectors upon challenge (41, 45). It might be because Tfh-like memory cells indeed possess plasticity, or they still represent a heterogeneous population. In addition, Gata3⁺ Th2 and Bcl6⁺ Tfh cells exhibit great plasticity, as each can give rise to both cell types following helminth challenge (101, 102). On the other hand, a specific CD62L⁺CXCR5⁺Bcl6⁺ Tfh cell population has recently been shown to efficiently generate Tfh effectors during secondary response, indicating lineage commitment of it (86). Besides, Th1-T_EM cells (Ly6C^hi and/or CXCR5^lo) almost exclusively generate secondary Th1 effector cells upon rechallenge (38, 42, 44, 45, 49). Some other fate mapping and single-cell studies also indicate that CD4⁺ T_M cells exhibit minimal plasticity and might be lineage-committed upon recall (38, 39). A possible explanation is that the plasticity of CD4⁺ T_M cells at population level is associated with additional heterogeneity (116). Nonetheless, the plasticity at the population level currently hinders a deeper understanding of CD4⁺ T_M cell generation.

As discussed above, CD4⁺ T_M cells are suggested to derive from antigen-activated T cell progenitors, and the memory fate is primarily dictated during priming phase of primary response (39, 62). However, the lack of a guiding model significantly hinders further investigations into the mechanisms underlying generation of CD4⁺ T_M cells. Identification and characterization of potential common precursors of specific CD4⁺ T_M cells under Th1, Th2 and Th17 conditions would be beneficial for further investigating the programs of fate decision between effector and memory CD4⁺ T cells.

Unfortunately, the current theory of CD8⁺ T_M cell generation provides only limited guidance for identifying precursors of CD4⁺ T_M cells. As a reliable marker for CD8⁺ T_M precursors (8, 9), role of CD127 in generation of CD4⁺ T_M cells has attracted considerable attentions. Multiple studies demonstrate that CD127 mediated IL-7 signaling is required for maintenance and homeostasis of CD4⁺ T_M cells; however, CD127 seems to be downregulated at effector phase and thus cannot identify memory precursors at the peak of Th1, Th2 and Th17 responses (48, 52, 69, 84, 86, 104, 108, 110, 118). Nonetheless, revealing programs that regulate re-expression of CD127 in activated CD4⁺ T cells during contraction phase might be a breakthrough for identifying common memory precursors, as IL-7/CD127signaling has been shown to be essential for survival of activated CD4⁺ T cells during this phase (52, 84, 110, 119). On the other hand, the dual roles of TCF1 in CD4⁺ T_M cell differentiation and Tfh polarization make it an unreliable marker for memory precursors during Th1/Tfh response under viral infections (39, 52–57). Besides, RNA-seq data indicates that TCF1 is not involved in formation of T_CM precursor-like CD62L⁺ T cells during Th2 response (104). The exact role of TCF1 in memory generation and Th/Tfh lineage polarization still needs to be defined. Collectively, CD127 and TCF1 are not reliable markers for discriminating CD4⁺ T memory precursors from terminally differentiated effectors.

Programs of CD4⁺ T_M cell differentiation also exhibit great differences under Th1, Th2 or Th17 conditions (5, 43, 94). Commonly used markers during Th1 responses, such as Ly6C, PSGL1 and CXCR5, seem unable to distinguish between T_CM precursors, Th2 effector cells, and Tfh cells under Th2 conditions (101, 102, 104). Besides, whether T_CM-associated factors during Th1 response, such as TCF1, Thpok and OCA-B, also participate in regulating memory formation under Th2 and Th17 conditions remains ill-defined (32, 58, 104). The mechanisms of T_CM and Tfh memory cell differentiation under various types of infections and exposures are far from clear. Moreover, the generation of CD4⁺ T_EM and T_RM cells are closely related to Th1, Th2, Th17 and even Tfh effector cell differentiation (5, 43, 94). For instance, the Th1 hallmark transcription factor T-bet is indispensable for terminal effector differentiation and the generation of T_EM and T_RM cells through multiple mechanisms during Th1 responses, whereas it is clearly not involved in effector and memory differentiation under Th2 and Th17 conditions (81, 93, 120). Whether Gata3 and RORγt contributes to T_EM and T_RM generation during Th2 and Th17 responses remains unclear. It seems that the more we delve into the differences between subsets of CD4⁺ T cells, the harder it becomes to develop a general model of their memory generation.

Nevertheless, we cannot exclude the possibility that there exists a common mechanism for generation of multiple CD4⁺ T_M populations under different conditions. TCR-dependent IL-2/CD25 signaling at early priming stage contributes to Th1/2/17 effector cell differentiation, whereas its absence favors the formation of Tfh and T_CM cells during both Th1 and Th2 responses (57, 66, 95, 99, 121). IL-2/CD25 signaling during priming stage is also required for Th1-T_RM and Th2-T_RM generation through promoting effector differentiation and NLT residency (95, 96, 99). Although it has been argued that IL2/CD25 signaling promotes survival and persistence of CD4⁺ T cells at later stage (69, 70), it indeed contributes to determining the fate choice between Th1/2/17 effector cells and Tfh/T_CM cells at the early stage during priming ( Figure 4 ). Some other factors, including ACC1 and CD30, have also been shown to participate in regulating the transition of Th1/2/17 effector cells into lineages-committed Th1/2/17 memory cells under multiple conditions (64, 65). It is indicated that activated CD4⁺ T cells undergo some common processes that regulate their transition into memory cells, although decisive factors that dictate the fate decision between effector and memory under diverse conditions have not been revealed.

On the other hand, positioning of activated CD4⁺ T cells, which facilitates their interaction with specific extrinsic signals, plays a critical role in effector and memory differentiation of Th1, Th2, Th17 and Tfh cells ( Figure 4 ). Localization in periphery and center of SLOs to interact with inflammatory and survival signals controls effector and T_CM differentiation, respectively, during both Th1 and Th2 responses (20, 37, 41, 42, 52, 65, 84, 103). Retention of activated T cells in NLTs to interact with local inflammatory and survival signals is suggested to be indispensable for T_RM generation during Th1, Th2 and Th17 responses, though factors regulating their residency in various NLTs remain unclear (93–96, 98, 99, 105, 111, 112, 122). Localization in B cell follicle (and GC area) and insufficient interaction with B cells both are suggested to be required for Tfh memory generation under viral infections and antigen exposures (84, 86). Consistent spatial requirements for differentiation of specific CD4⁺ T_M subsets also provide valuable perspectives for identifying common memory precursors under various types of infections and exposures.

CD4⁺ T_M cells play critical roles in protective immunity and immunopathology. Revealing underlying mechanisms of their generation and maintenance is crucial for developing therapeutic approaches targeting CD4⁺ T_M cells in human diseases. The intriguing MPEC/SLEC model has led to remarkable advances in the understanding of generation of CD8⁺ T_CM, T_EM and T_RM cells. However, there is a lack of guiding model for further investigations into generation of CD4⁺ T_M cells, due to the diversity and plasticity of multiple CD4⁺ T subsets. Differentiation of specific CD4⁺ T_M subsets under Th1, Th2 and Th17 conditions depends on both common and distinct underlying mechanisms. The shared features discussed above might provide a path to a putative common model on generation of CD4⁺T_M subsets under all types of infections and exposures. Clearer delineations of Tfh memory, T_CM, T_EM and T_RM cell populations at memory stage, as well as Tfh and Th1/2/17 populations at priming stage, are definitely required. Future works are also needed to reveal undergo programs governing memory generation under specific conditions and to clarify the consistency of these programs across different types of infections and exposures. Considering that transcriptional networks and positioning in proper niches to receive extrinsic stimulation both contributes to effector and memory differentiation, integrating spatial transcriptomics and single-cell RNA-seq technology will be an informative approach for further investigations (123). In addition, the advancing MHC-II multimer technology will be a valuable tool for studying endogenous antigen-specific CD4⁺ T cells (124). Identifying potential common and distinct precursors for each CD4⁺ T_M cell subset would be highly beneficial for elucidating underlying mechanisms for their generation.

On the other hand, recent findings have suggested that CD8⁺ cytotoxic T (Tc) cells also differentiate into multiple subsets, in a manner similar to CD4⁺ T cells (125). Conventional IFNγ-producing Tc1 cells, IL-4-producing Tc2 cells, and IL-17-producing Tc17 cells, which are polarized by specific cytokine microenvironments, are all capable of differentiating into long-lived T_M cells (125–127). However, whether each CD8⁺ T_M cell subset derives from distinct or common precursor populations and whether they exhibit plasticity remains largely unclear. Elucidating the mechanisms underlying the generation of multiple CD4⁺ T_M cell subsets may provide valuable insights into the differentiation of memory CD8⁺ Tc1, Tc2, and Tc17 cells.