Professional antigen presenting cells (APCs) including dendritic cells (DCs) are key regulators of innate and adaptive immune responses. During primary viral/bacterial respiratory infection, lung-resident DCs process and present the pathogen’s antigens and migrate to the mediastinal lymph node (MLN) to prime naïve T cells and stimulate their proliferation ( Figure 1 ). Migratory lung DCs within the MLN imprint T cell lung homing through site-specific surface molecular signatures (15, 16) and help influence pulmonary T_RM generation. In human and humanized mice, pulmonary CD1c⁺ and CD141⁺ DCs have both been shown to present viral antigens, however only CD1c⁺ DCs drive the expression of CD103 (a key marker of T_RM – see “Phenotypic Characterisation”) on both naïve and memory CD8⁺ T cells (17). Multiple chemokine receptors involved in lung trafficking are expressed by T_RM including C-X-C Motif Chemokine Receptor 3 (CXCR3), CXCR6 and CCR5 (11, 18–20). Although no specific combination of homing markers have been identified for pulmonary T_RM, CD4⁺ are likely recruited to the airway during Respiratory Syncytial Virus (RSV) infection in human via C-X-C motif chemokine 10 (CXCL10 - the ligand for CXCR3), as chemokine levels correlated with activated CD4⁺ T cell recruitment in bronchoalveolar lavage (BAL) (18).

Shortly after activation in the MLN, effector T cells migrate to the lungs and contribute towards pathogen clearance. The majority of pathogen-specific T cells then undergo apoptosis, however a minority differentiate into T_RM in response to environmental cues (21), with the number of T cells persisting in the lung following infection correlating with the efficiency of T_RM differentiation (22).

Effector T cells entering the lung express sphingosine-1-phosphate receptor (S1PR1), sensing increasing sphingosine-1-phosphate (S1P) gradients in blood and lymph, leading to tissue egress (10). S1PR1 expression is regulated by local cytokine-induced transcriptional downregulation and early activation marker CD69-mediated post-transcriptional antagonism (10). CD69 is a cell-surface receptor that is rapidly and transiently expressed on all recently activated T cells. Induction of the membrane‐bound type II C‐lectin receptor CD69 by antigen stimulation and inflammatory cytokine exposure leads to downregulation of S1PR1, which when combined with inflammation-induced chemotactic signalling, supports effector T cell retention and T_RM generation (10, 23). Transition of recruited effector T cells to T_RM in murine lung requires simultaneous tissue damage and T cell receptor (TCR) activation by pulmonary cognate antigen encounter (24–27). Overlapping TCR genes from human T_RM and non-T_RM indicate that environment, rather than epitope specificity, drives T_RM formation (19). Antigen-dependent cross-competition however does promote T_RM formation, with effector T cells recognising antigen presented by infected tissue cells preferentially entering the local T_RM pool (28). Although demonstrated in murine skin, it is possible the same rules also apply to the lung. Naïve T cells in LNs may also be epigenetically preconditioned during steady state conditions by migratory DCs to differentiate into T_RM upon exposure to cognate antigen (29). Dependent on DC-driven, transforming growth factor β (TGF-β), altering local or systemic TGF-β activity prior to vaccination may help promote T_RM formation (29).

Once established, T_RM remain lung-resident and contribute towards immunosurveillance and homeostasis (6). Maintained in a quiescent state, human transplant studies have demonstrated donor CD4⁺ and CD8⁺ T_RM to persist in the lungs for over 15 months, with single cell transcriptome analysis confirming de novo T_RM generation via the identification of a “mature T_RM” and an immature “T_RM-like” population that gradually acquire T_RM markers (CD69, CD103 and CD49a) over time (30, 31). Pulmonary CD8⁺ T_RM are however more short-lived than those found in other tissues such as the skin and intestine (32, 33). As microbes are constantly being inhaled, the limited longevity of pulmonary CD8⁺ T_RM may provide a mechanism for avoiding unnecessary inflammation and pathogenesis in this tissue (34).

In human, Notch signalling alongside low levels of T-bet and Eomesodermin (EOMES) are required for the development and maintenance of CD4⁺/CD8⁺ T_RM, with Notch regulating T_RM metabolic programs (11, 20). Human pulmonary CD8⁺ T_RM display elevated levels of the transcription factors Hobit (encoded by the gene ZNF683) and Runx3, that may be involved in T_RM generation and/or maintenance (30). Interestingly, despite showing elevated mRNA levels, Hobit protein expression was reported absent in human CD4⁺ T_RM, suggesting differences between CD4⁺/CD8⁺ T_RM formation/maintenance (11).

Heterogeneity in effector function and phenotype is evident within T_RM populations, particularly within CD4⁺ T_RM (19). Transcriptome profiling of human lung CD69⁺ T_RM has revealed the differential expression of 31 core genes associated with migration, adhesion and regulatory molecules when compared to CD69^- subsets (19). This transcriptional profile is conserved across CD4⁺/CD8⁺ CD69⁺ lineages as well as tissues (19). Pulmonary T_RM exhibit high transcript levels for genes encoding for several chemokine receptors, pro-inflammatory cytokines and cytotoxic mediators, enabling them to be recruited and retained within the lung and undergo rapid, polyfunctional responses (11, 20). T_RM respond rapidly with effector functions, however, expression of regulatory genes (e.g. cytotoxic T-lymphocyte-associated protein 4 [CTLA4] and B-and T-lymphocyte attenuator 4 [BTLA4]) in CD8⁺ T_RM may present a safety mechanism to minimise aberrant activation and associated inflammation/tissue damage (20).

Human antigen-experienced lungs are enriched with B cells containing a resident memory phenotype (35). As human and non-human primate (NHP) B_RM data are limited, most findings are derived from mouse studies. During primary respiratory infection, naïve B cells, primed by either free antigen or antigen delivered by subcapsular sinus (SCS) macrophages (36), interact with cognate CD4⁺ T cells at the T-B border within the MLN (9, 37). Following initial proliferation at the outer follicles, B cells may differentiate into extrafollicular short-lived plasma cells, early (germinal centre (GC)-independent) memory cells or proliferate to form the GC ( Figure 1 ). Following somatic hypermutation, B cells can exit as long-lived plasma cells, migrating to the bone marrow where they secrete antibodies for decades, or memory B cells (9, 37). Having migrated to the lungs to participate in pathogen clearance, most of the responding B cells undergo apoptosis, leaving a few resting memory cells in the respiratory tract and lymphoid organs where they wait for the same antigen.

Murine parabiosis studies have demonstrated B_RM generation requires local antigen encounter and is dependent on early CD40-interactions with T cells (7). Once established, B_RM remain lung resident due to expression of CD69 (7). Here they undergo metabolic reprogramming, switching from anabolic to catabolic pathways to reduce their requirement for high levels of cytokines for their maintenance (37). In mice, B_RM are quiescent and long-lived, maintained from precursors within persisting GCs in areas known as inducible bronchus-associated lymphoid tissue (iBALT) (14), however B_RM have also been detected in the absence of iBALT (39). Established one week after influenza infection, murine pulmonary B_RM have been demonstrated to be phenotypically and functionally distinct from their systemic counterparts (7).

Few studies have investigated gene regulation in pulmonary B_RM, particularly in humans. Although the possibility of a “master transcription factor” for B_RM generation has been suggested, no unique transcription factor has been identified so far (9). Increased expression of the transcription factors Bach2, KLF2, ZBTB32, ABF1 and STAT5 are associated with B_RM formation in mice, however their exact roles are yet to be understood (9, 40). The transcriptional regulation of pulmonary B_RM differentiation is likely to be unique – understanding these transcription factors may help identify methods for modulating their formation (41).

Due to their similarities to human, NHPs provide an invaluable tool for investigating host response to respiratory infection and vaccination. Although heterogenous within the lung, human and NHP T_RM are phenotypically distinct from T_CM and T_EM and are primarily identified by the high expression of the C-type lectin receptor CD69, and integrins CD103 and CD49a (30). The transmembrane CD69 is a key marker of pulmonary T_RM, distinguishing memory T cells in tissue from those in circulation (19), however murine evidence suggests its expression is not essential for the establishment and maintenance of T_RM in the lung (25, 42, 43). Although considered as an early activation marker for TCR signalling, T_RM CD69 expression is not associated with markers of recent activation and appears to be a function of previous antigen exposure (19).

Preferentially expressed on CD8⁺ T_RM compared to CD4⁺, CD103 promotes adherence to E-cadherin, an adhesion molecule expressed by epithelial cells (22, 30, 44). CD103 expression is driven by membrane-bound TGF-β (mediated by IL-10) on APCs (CD1c⁺ DCs and monocytes) (17, 45) and is thought to contribute towards initial recruitment and persistence of CD8⁺ T_RM to aide surveillance rather than long-term maintenance (42). CD49a, expressed by both CD4⁺ and CD8⁺ T_RM, is an integrin specific to collagen IV that facilitates locomotion for surveillance and is essential for T_RM survival by limiting apoptosis following ligand engagement (42). Other recognised surface markers of pulmonary T_RM are outlined in Table 1 - understanding the full function of these markers, whether they represent different subsets/maturation states and whether they are pathogen-dependent remains to be determined.

Although no specific marker of B_RM residency has been described, pulmonary B_RM are phenotypically distinct from their systemic memory and non-memory counterparts (7) – see Table 2 . As well as lacking CD62-L, murine pulmonary B_RM express markers associated with T_RM, such as CD69, CXCR3 and CD44, which retain B_RM within the lung (7, 12, 35, 39). CD69 has also been found on human pulmonary B_RM (35). Whether other markers found in mice are also expressed on human and NHP pulmonary B_RM requires further investigation.

Functional studies have revealed B_RM established early after murine influenza infection are positive for immunoglobulin M (IgM⁺) which are later followed by isotype-switched B_RM (7). Following murine pneumococcal infection, the majority of isotype-switched B_RM are IgG⁺, with a small fraction IgA⁺ (35). The majority of B_RM found in healthy human lung are also isotype-switched (35).

Growing evidence indicates that virus-specific T cells resident along the respiratory tract are highly effective at providing potent and rapid protection against inhaled pathogens. In human influenza infection, CD8⁺ T_RM have been shown to recognise the internal, conserved proteins of the virus whereas CD4⁺ T_RM recognise both internal and external proteins, with both cell types contributing towards heterosubtypic protection (33). CD8⁺ T_RM have been shown to be cross-reactive against three influenza strains (51), with single cell sequencing revealing diverse TCR profiles “capable of recognising newly emerging viral escape variants” (22).

Influenza-specific CD8⁺ T_RM have a low activation requirement, requiring only cognate antigen in the absence of helper cell-derived signals (52). Once stimulated, they are highly proliferative, producing polyfunctional progeny (producing ≥2 cytokines – IFN-γ, TNF, Granzyme B and IL-2) with effector function superior even to their parent population (22, 44). Polyfunctional T_RM offer enhanced protection by producing higher levels of cytokines whilst simultaneously driving effector responses (53) - activated CD8⁺ T_RM exert their cytotoxic function to kill infected cells (10) whilst CD4⁺ T_RM interact with B cells in iBALT to generate new neutralising antibodies (14, 54). A newly identified, long-lived CD4⁺ T resident helper (T_RH) population with functional and phenotypical similarities to lymphoid T follicular helper cells (T_FH) has also been described following murine influenza infection. Residing within iBALT, T_RH are tightly localised with B_RM to support local antibody production following reinfection (55).

In an experimental human RSV infection model, the abundance of RSV-specific, pulmonary CD8⁺ T_RM before infection was associated with reduced symptoms and viral load, implying that CD8⁺ T_RM can confer protection against severe respiratory viral disease when humoral immunity is overcome (8). RSV-specific CD8⁺ T_RM displayed phenotypic changes representative of advanced differentiation, with downregulation of both co-stimulatory and cytotoxicity markers, suggesting cells can respond rapidly to reinfection, but function is restricted to minimise excessive tissue damage (8).

RSV infection in African Green Monkeys (AGM) also induced virus-specific airway CD8⁺ T_RM capable of reducing viral titres, however failed to induce robust CD4⁺ T_RM and humoral responses (21). Previously protective RSV-candidate vaccines in AGM induced a strong T cell response, whilst those eliciting a strong neutralisation antibody response without detectable T cell response were not as effective (56). Similar to influenza, CD8⁺ T_RM recognise internal proteins of RSV, whilst CD4⁺ T_RM recognise external proteins (18). RSV-induced immunopathology relates to a dysregulated T cell response – RSV-specific memory CD8⁺ T cells in blood display little evidence of multiple cytokine production unlike those seen against influenza (8, 18). CD8⁺ T_RM however appear to be more polyfunctional, generating IFN-γ, IL-2 and TNF (21), however fail to undergo proliferation when activated and express reduced cytotoxicity markers compared to peripheral memory cells (8).

Activated CD4⁺ and CD8⁺ T_RM have been identified in the lungs of patients infected with Mycobacterium tuberculosis (Mtb), where they help limit intracellular macrophage Mtb replication (46). T_RM were polyfunctional, expressing IFN-γ, TNF ± IL-2, and exhibited a highly cytotoxic profile, with CD4⁺ T_RM appearing more polyfunctional than CD8⁺ T_RM (46). CD49d is upregulated on airway CD4⁺ T_RM and optimises the localisation of human Mtb -specific recall responses (47). Mtb infection in macaques drives a cellular T helper 1 (T_H1) and humoral response, without protective efficacy, however repeated pulmonary Bacillus Calmette-Guérin (BCG) delivery was shown to induce polyfunctional, T_H17 CD4⁺ T_RM, leading to airway IgA secretions in BAL (48), presumably through the generation of B_RM. Interstitial CD4⁺ depletion with simian immunodeficiency virus (SIV) following Mtb infection identified CD4⁺ T_RM (57), proliferating CD8⁺ memory T cells (T_CM, T_EM and likely T_RM) and B cells within iBALT (58) as critical for suppressing latent Mtb reactivation.

T_H17 CD4⁺ T_RM are also critical in protecting against murine nasal Bordetella pertussis (Bp) colonization (59). Although both capable of protecting against Bp lung infection, whole cell Bp vaccine, unlike the acellular vaccine, induced nasal IL-17-producing CD103⁺ CD4⁺ T_RM (similar to natural Bp infection) that recruited neutrophils to enhance bacterial clearance.

Lung T_RM also display “innate-like” behaviour, amplifying inflammation following noncognate bacterial infection. APC-derived IL-12/IL-18 activated virus-specific CD8⁺ T_RM within the lung parenchyma, leading to the rapid synthesis of IFN-γ. This “bystander activation” boosted neutrophil recruitment to improve bacterial clearance. Despite being performed in mice, the authors demonstrated in vitro that human CD8⁺ T_RM similarly synthesise IFN-γ in response to IL-12/IL-18 (60).

Alongside long-lived antibody-secreting plasma cells, B_RM contribute towards the protective humoral immune response to pulmonary reinfection (12). The presence of B_RM is a common feature of antigen‐experienced lungs and is important for acquired immunity (7). B cells in the airways secrete antibodies that act both locally and at mucosal surfaces. These antibodies, predominantly IgM and IgA, bind to glandular epithelial and mucosal surfaces to promote pathogen clearance (61). B cells activated in respiratory lymphoid tissue also differentiate into IgA‐secreting plasma cells that predominantly act in the airway. Current knowledge of B cell homing and class switching in the airway remains limited.

Murine parabiosis/adoptive transfer/depletion studies have demonstrated the protective role played by B_RM in response to both viral (7, 12, 54) and bacterial lung infection (35). B_RM provide rapid antibody-secreting cells (ASC), producing a range of class switched neutralising antibodies (7, 12). Lung B_RM produce greater numbers of ASC than splenic memory cells following exposure to drifted virus, indicative of heterosubtypic protection (14). Cross-neutralising antibodies to conserved, internal influenza proteins provide heterosubtypic protection (14, 54). Although IgA is more effective than IgG at preventing upper respiratory infection, in combination they achieve maximal neutralising activity against influenza in mice (12). Following murine pneumococcal infection, B_RM contribute towards bacterial clearance by rapidly secreting cross-reactive antibodies, even when reactivated by a serotype-mismatched strain (35). In macaques, iBALT persistence is associated with reduced Mtb reactivation due to enhanced B-cell and humoral immunity (58). B_RM are also potent APC, binding and endocytosing antigen via their BCR to increase peptide/MHC II presentation and further enhance CD4⁺/B cell responses (9, 13).

Optical endomicroscopy imaging (88), recently used for the detection of human alveolar neutrophils in situ (89), may provide a valuable tool for assessing pulmonary-resident memory lymphocytes and quantifying immunological memory following vaccination. Fluorescently tagged ligands or antibodies, capable of binding to specific T_RM/B_RM surface markers, can be delivered to the airways via a bronchoscope to enable visualisation ( Figure 3 ). The information gained can be combined with systemic data to evaluate vaccine efficacy and expected degree of protection against respiratory pathogens. In situ optical imaging may also be used to screen lungs for transplantation, as the presence of T_RM in donor tissue is associated with reduced adverse clinical events in the recipient (30).

The COVID-19 pandemic has highlighted how immune responses in the airways differ from those in the circulation and that it is tissues, not blood, where immune cells function (90). Assessing tissue-based immunity following infection and vaccination is therefore essential. Ex vivo lung perfusion (EVLP), using human lungs deemed non-suitable for transplantation, provides an ideal model for assessing tissue immunity and optimising in situ optical imaging. As well as studying populations in situ, EVLP offers the ability to isolate large numbers of human T_RM/B_RM, far higher than those obtained from a typical BAL, for in-depth analysis (including phenotype, function, and antigen-specificity). Intraperfusate delivery of a fluorescently tagged CD45 antibody can also differentiate circulating (labelled) from tissue-resident (non-labelled) cells. This technique has recently revealed how human lung T_RM colocalise with lung-resident macrophages, preferentially around the airways, where they receive costimulatory signals to augment effector cytokine production and degranulation (91).