Comparative phenotype of circulating versus tissue immune cells in human lung and blood compartments during health and disease

Abstract The lung is a dynamic mucosal surface constantly exposed to a variety of immunological challenges including harmless environmental antigens, pollutants, and potentially invasive microorganisms. Dysregulation of the immune system at this crucial site is associated with a range of chronic inflammatory conditions including asthma and Chronic Pulmonary Obstructive Disease (COPD). However, due to its relative inaccessibility, our fundamental understanding of the human lung immune compartment is limited. To address this, we performed flow cytometric immune phenotyping of human lung tissue and matched blood samples that were isolated from 115 donors undergoing lung tissue resection. We provide detailed characterization of the lung mononuclear phagocyte and T cell compartments, demonstrating clear phenotypic differences between lung tissue cells and those in peripheral circulation. Additionally, we show that CD103 expression demarcates pulmonary T cells that have undergone recent TCR and IL-7R signalling. Unexpectedly, we discovered that the immune landscape from asthmatic or COPD donors was broadly comparable to controls. Our data provide a much-needed expansion of our understanding of the pulmonary immune compartment in both health and disease.


Introduction
CD45 + hematopoietic cells comprise a significant portion of cellular networks within any given organ and provide signals to non-hematopoietic cells to determine tissue function and health. This includes resident innate immune cells surveying the environment for foreign antigen and contributing to tissue homeostasis [1][2][3], tissue-resident memory cells awaiting re-exposure to their cognate antigen [4,5], and non-resident cells trafficking through the local vasculature. In the context of immune responses, interaction of resident and non-resident CD45 + cells with local tissue cells and the extracellular matrix environment in which they are found is fundamental to the induction, regulation, and resolution of such responses [6,7]. Consequently, immune responses are highly contextual to tissue environments in which they occur [8].
In humans, sample access for studying tissue-specific immunity is challenging, with our understanding of fundamental human cellular immunity largely derived from peripheral blood cells. While this can be informative [9,10], studies that have drawn comparisons between immune populations isolated from specific tissues, such as the liver and brain, and those from peripheral blood find clear phenotypic differences between these sites [11,12]. Moreover, it is becoming apparent that there are clear phenotypic distinctions that can be drawn between tissue-resident cells and their non-resident counterparts [11,12]. Therefore, there is a need for a deeper understanding of immune cell populations and their phenotypes in different tissues.
The lung presents a unique and challenging tissue context for immune populations. As a mucosal barrier, it is constantly exposed to environmental stimuli, necessitating robust defences against pollutants, particulates, and pathogens. However, in contrast to other mucosal surfaces and to facilitate gaseous exchange, this barrier, comprised of a singlecell layer, must remain in constant close contact with the innervating vasculature [13]. As a result, pulmonary immune responses are highly regulated, with dysregulated inflammation considered a hallmark of a wide variety of pulmonary disorders including asthma and chronic pulmonary obstructive disease (COPD) [14,15]. Due to the challenges in obtaining airway samples, our understanding of the human pulmonary immune system is severely limited, in either health or disease.
Mononuclear phagocytes (MNPs)-including dendritic cells (DCs), macrophages, monocytes, and monocyte-derived cells-represent a highly heterogeneous and plastic population of cells with overlapping functions and surface marker expression [16,17]. These MNPs can be further subdivided into an array of subsets based on cell surface expression of defining markers [16,18]. MNPs are well represented in the 3 lung, occupying both the alveolar and interstitial spaces as well the underlying parenchyma [13]. DCs in particular play a vital role in initiating and regulating immune responses to a wide variety of pathogens, environmental antigens, and self-antigens [16]. In humans, our understanding of MNPs is derived primarily from analysis of blood and skin samples [17,19]. However, it is evident that there is substantial inter-tissue variation in the composition of the MNP compartment [20][21][22] meaning that these peripheral sites make for poor proxies of less accessible tissues. Whilst these cells have been implicated as major players in a variety of lung disease settings [23], our understanding of the composition of MNP populations in the human lung, its relationship to other tissues, and whether it is significantly modified in inflammatory disease settings is currently limited.
In partnership with MNPs, T cells function as major regulators of lung immune responses. CD4 + T cells are key sources of cytokines essential in feeding back to the myeloid compartment, coordinating adaptive immune responses, and supporting B-cell antibody responses [24]. CD8 + T cells are also capable of providing inflammatory cytokines and contribute to direct cytotoxic killing of infected cells [25], while CD4 + FOXP3 + regulatory T cells (Tregs) are key regulators of inflammation in a wide variety of settings, essential for tissue protection [26,27]. TCRγδ + T cells also reside in the lung, but they remain poorly characterized and their function in lung tissue physiology is unclear even in murine systems [28]. In recent years, clear phenotypic differences between resident T cells, non-resident T cells, and peripheral blood T cells have been reported in a variety of tissue contexts, the delineation of which has centred around the expression of the canonical T cell tissue-residency markers CD69 and CD103 [11,12,[29][30][31][32][33]. CD69, in particular, seems to demarcate longlived tissue-resident memory (TRM) cells [24], with one elegant study showing that the majority of donor-derived CD4 + and CD8 + T cells from liver allografts express this marker for years following transplantation [31]. The requirement for CD103 in tissue retention is less clear as, whilst it is expressed on the majority of tissue-resident CD8 + T cells in the adult human intestines [30], its expression on CD4 + T cells and CD8 + T cells in other tissue compartments is more variable [33][34][35]. Additional complexity is presented by the fact that CD69 and CD103 are not always co-expressed within a given tissue, often with the concomitant presence of discrete CD69 + CD103 − , CD69 − CD103 + , and CD69 + CD103 + T cell populations [34], suggesting that CD103 expression denotes T cell specialization correlated with-but independent from-tissue residence. Indeed, CD103 + CD69 + CD8 + T cells in the liver have a phenotype distinct from CD103 − CD69 + CD8 + T cells, expressing higher levels of chemokines, MHCII, and IL-2 [11].
At present, our understanding of composition and phenotype of the MNP and T cell compartments within the human lung is highly limited. This is particularly true in the context of chronic inflammatory disorders such as asthma and COPD. Few studies have performed detailed immune phenotyping on primary lung tissue and those that have are reliant on relatively small sample sizes [13]. More data is available for peripheral blood cells isolated from donors with such conditions, but it is unclear to what extent peripheral blood represents the pulmonary immune environment, meaning depth of understanding of the immune phenotype of lung-resident MNPs and T cells in these conditions is currently lacking.
To expand the scope of our understanding of the pulmonary immune environment, we performed flow cytometric phenotyping on cells isolated from resected lung tissue from 115 donors, with matched peripheral blood from 69 of these donors. We provide a detailed comparison of the immune phenotype of these two sites, identifying clear differences in the overall composition of immune populations and expression of phenotypic markers between lung tissue and blood sites. We demonstrate that pulmonary DCs express higher levels of activation markers compared to those in circulation and identify CD103 as a marker of recent receptor signalling in pulmonary CD4 + and CD8 + T cells. Finally, we demonstrate that, at resting state, the composition of the pulmonary MNP and T cell compartments is not significantly different in donors with asthma or COPD, relative to controls. Together, these data expand our overall understanding of the pulmonary immune network and its phenotypic distinction from peripheral blood, supporting a model in which immune dysregulation in the lung is a product of functional changes within cells rather than dramatic changes in the pulmonary immune cell landscape.

Results
To interrogate the immune phenotype of human lung, we collected tissue resection samples from donors undergoing thoracic surgery and performed flow cytometric analysis on enzymatically digested lung single-cell suspensions from 115 donors ( Table 1), 69 of whom we were also able to acquire PBMCs.
We observed cDC2s and pDCs to be the dominant DC populations in lung, existing at significantly greater frequency than cDC1s and moDCs (Fig. 1a, Supplementary  Fig. S2a). Although this pattern was consistent in the blood pDCs, were present at significantly greater frequency than cDC2s in this compartment (Fig. 1a, Supplementary  Fig. S2a). Consistent with previous literature [19], cDC1s represented a small fraction of peripheral blood DCs and accounted for less than 1% of total PBMCs (Fig. 1b). The C-type lectin receptor CD206 (mannose receptor) is a pattern recognition receptor (PRR) associated with signalling by type-2 cytokines, particularly IL-4 and IL-13 [38], and used as a moDC marker on CD1a+ cells [37]. We noted that CD206 expression on cDC2s was dramatically increased in the lung relative to the blood ( Fig. 1c and d), indicating expression is dependent on cells that are associated with the lung environment.
Analysis of monocytes showed that CD14 + CD16 − (classical) monocytes were the dominant subset (vs. CD14 − CD16 + and CD14 + CD16 int populations) in both lung and blood ( Fig.  1e and f). Of these smaller subsets, we observed a small but statistically significant difference between the frequency of CD14 + CD16 − and CD14 + CD16 Int monocytes present in both the blood and lung (Fig. 1e, Supplementary Fig. S2b). In the blood, CD14 − CD16 + monocytes were present at higher frequency whereas the reverse was true in the lung (Fig. 1f).

The lung environment promotes the expression of activation markers by myeloid cells
To assess whether pulmonary myeloid cells in the lung displayed a different level of activation compared to their counterparts in circulation, we measured expression of the canonical myeloid maturation markers CD40 and CD86 as well as MHCII (Fig. 2, Supplementary Fig. S3).
For all DC subsets captured by our panel, we observed significant increases in the frequency of expression of both CD40 and CD86 in lung DCs when compared to blood, which was particularly pronounced in cDC2s ( Fig. 2a and b). In cells positive for these markers, we detected significant increases in the median fluorescence intensity (MFI) of CD40 on cDC1s and cDC2s, and of CD86 on cDC2s and pDCs, in the lung relative to the blood (Fig. 2b). Conversely, when assessing the expression of MHCII we found that DCs isolated from the lung tissue expressed significantly lower levels than those isolated from the blood ( Supplementary Fig. S3b). Strikingly, when we compared CD206 + cDC2s with those that did not express CD206 we observed a significant increase in the frequency of CD40 and CD86 expression by CD206 + cells ( Fig. 2c and d). This increase in CD40/CD86 positivity was noticeable in the blood (Fig. 2d) but was substantially greater in the lung, indicating that CD206 expression may demarcate a mature population of cDC2s.
Changes in expression of activation markers were less pronounced in monocytes ( Supplementary Fig. 3a). However, we observed statistically significant increases in the frequency of CD40 expressing CD14 + CD16 − and CD14 + CD16 int monocytes, as well as significant increases in the MFI of CD40 in all monocyte populations ( Supplementary Fig. S3a). In contrast to DCs, we detected no change in the frequency of monocytes expressing CD86 in the lung relative to the blood, although there was a small but statistically significant increase in the MFI of CD86 on CD14 + CD16 int monocytes in the lung relative to the blood. Additionally, CD14 + CD16 − monocytes expressed higher levels of MHCII in the lung tissue relative to peripheral blood, whilst CD14 -CD16 + monocytes had reduced MHCII expression ( Supplementary  Fig. S3c).

Differential expression of residency markers is associated with distinct T-cell phenotypes
In tandem with and coordinated by DCs, T cells represent major regulators of immune responses. Given that DCs presented a distinct profile in the lung compared to the blood (Figs. 1 and 2), we undertook phenotyping of the T cell compartment ( Supplementary Fig. S4) in these sites to identify whether comparable distinction in T cell phenotype could be observed between locations. CD4 + FOXP3 − and CD8 + T cells were the dominant T cell populations in the lung (Fig.  3a), present at comparable frequencies ( Supplementary Fig.  S5a). As expected, CD4 + FOXP3 + and γδ T cells were found with substantially less frequency than CD4 + FOXP3or CD8 + T cells, but at comparable levels to one another ( Fig.  3a, Supplementary Fig. S5a). This pattern was similar in the blood (Fig. 3a) although CD4 + FOXP3 − cells were present at a statistically greater frequency than CD8 + cells at this site ( Fig.  3b, Supplementary Fig. S5a). CD69 and CD103 have been used to delineate tissue residency for T cells in a number of sites [34], with CD69 expressing CD4 + T cells within tissues thought to represent a distinct subset phenotypically distinct from circulating T cells [29]. We observed increased CD69 and CD103 expression in the lung relative to the blood for CD4 + FOXP3 − , CD4 + FOXP3 + , and CD8 + T cells, although CD69 − CD103 − cells still accounted for >50% of all T cells present ( Fig. 3c and d, Supplementary Fig. S5b). Notably, whilst CD69 or CD103 expression was low in the blood, a clear population of CD69 − CD103 + cells was observed in this site, whilst CD69 + CD103 + and CD69 + CD103 − cells were virtually undetectable (Fig. 3d). This observation was consistent for CD4 + FOXP3 − , CD4 + FOXP3 + , and CD8 + T cells (Fig. 3d).
Differential expression of CD45RA and CCR7 yielded four T cell phenotypes: naïve T cells (CD45RA + CCR7 + ), central memory T cells (TCM, CD45RA − CCR7 + ), effector memory T cells (TEM, CD45RA − CCR7 − ), and effectors re-expressing CD45RA T cells (TEMRA, CD45RA + CCR7 − ) [34] (Fig. 3e). We observed significant differences in the frequencies of different CD4 + FOXP3 − and CD8 + T cell phenotypes based on expression of CD69 or CD103 both in the blood and the lung (Fig. 3f, Supplementary Fig. S5c). Notably CD103 expression, on its own or in conjunction with CD69, was associated with an increase in the frequency of CD4 + FOXP3 − and CD8 + T cells with a TEM phenotype, and a concomitant decrease in those with a naïve phenotype, in both sites (Fig. 3f).   CD103 defines a TCR and IL-7R signalling experienced population of effector memory T cells in the lung Given that differential expression of CD69 and CD103 was associated with differing phenotypes in both CD4 + FOXP3 − and CD8 + T cells, of which TEMs were dominant in the lung, we assessed whether the expression of these markers was associated with other readouts of phenotypic activation in CD4 + FOXP3 − and CD8 + TEMs. To do this we measured expression of the TCR co-stimulatory receptor, CD28, and the IL-7R, CD127. Since downregulation of these receptors is associated with experience of TCR and IL-7R signalling respectively, cells expressing these receptors are likely to be unstimulated, while decreased expression can be associated with recent activation [12,34].
We observed significant differences in expression of both CD28 and CD127 on CD4 + FOXP3and CD8 + TEMs between blood and lung, with the lung generally demonstrating lower expression of these markers ( Fig.  4a and b), suggesting increased T cell activation within the lung tissue. Interestingly, CD103 expressing (either CD69 − CD103 + or CD69 + CD103 + ) CD4 + FOXP3 − and CD8 + TEM cells showed the lowest levels of CD28 and CD127 expression relative to the blood, and compared to other lung TEMs not expressing CD103 (Fig. 4b, Supplementary  Fig. S6). These data indicate that, whilst CD103 expressing TEMs formed a smaller fraction than CD103 -TEMs, they represented the portion of TEMs with TCR and/or IL-7R signalling experience.
Chronic pulmonary inflammatory disorders are not associated with significant changes in cellular immune composition of the lung tissue As a major immune barrier site, immunological dysregulation within the lung is associated with a variety of pulmonary disorders including asthma, COPD, and pulmonary fibrosis [14,15,40]. Within our cohort, two of the most prevalent conditions observed were asthma and COPD, accounting for 14.78% (n = 17) and 33.04% (n = 38) of subjects, respectively (Fig. 5a). Donors not suffering asthma or COPD, and who had not previously been excluded due to a history of other inflammatory disease, were treated as controls. Both men and women were represented in each group, although all groups had a greater proportion of women, with this being particularly pronounced in the asthma group (Fig. 5a). The  (e and f) Boxplots representing the frequency of (e) NK cells and (f) CD64 + cells in the lung tissue given as the frequency of CD45 + cells. Boxplots indicate the median and interquartile range, whiskers indicate the maximum/minimum value no greater/lesser than 1.5× the quartile limit. Points beyond the whiskers indicate statistical outliers. P values were calculated by unpaired Wilcoxon two-sample test and adjusted for multiple comparisons using the Holm method. *P < 0.05. distribution of ages between groups was not significantly different, as determined by Kolmogorov-Smirnov test (control vs. asthma P = 0.9906, control vs. COPD P = 0.6789, asthma vs. COPD P = 0.4049) (Fig. 5b). Comparable numbers of cells were isolated from donor lung tissue regardless of disease group or smoking status ( Supplementary Fig. S7a). The majority of donors in each group were within the clinically typical blood neutrophil (1.8-7.5 × 10 9 cells/l) and blood eosinophil (0.0-0.4 × 10 9 cells/l) ranges (Supplementary Fig.  7a) and statistical comparison did not identify significant differences in the frequency of immune cell populations between donors that exceeded the typical range and those within it for both neutrophils and eosinophils.
To determine whether asthma or COPD had a significant effect on the frequency of blood and lung immune cells we compared a range of immune cell types between groups. We did not observe a statistical difference in the frequency any of the DC, monocyte, or T cells subsets assessed previously, nor in the total number of B cells (SSC-A low CD45 + CD3 − CD19 + ), Neutrophils (SSC-A high CD45 + CD64 − CD66b + CD11b + CD16 + ), or NKT (SSC-A low CD45 + CD19 − CD3 + CD56 + ) cells when comparing asthmatic and COPD donors to controls in both peripheral blood and lung tissue (Fig. 5d). However, we identified a significant reduction in the frequency of NK cells (SSC-A low CD45 + CD19 − CD3 − CD56 + ) in the lung tissue of COPD donors relative to control and asthmatic donors (Fig.  5e). Additionally, there was a small but significant reduction in the frequency of CD64 + cells (SSC-A int CD45 + CD66b − CD64 + ) in the lung tissue of asthmatic donors relative to control and COPD donors (Fig. 5f). However, despite these differences, when overall composition was assessed by principle components analysis, we did not observe donor samples clustering by group, and multivariate statistical comparison by permutational ANOVA indicated that group did not have a significant effect on overall cellular immune composition (P = 0.715) (Fig. 5c).
We next assessed whether we could detect phenotypic differences in the MNP and T cell compartments between groups. However, consistent with cell subset frequencies, we did not observe any significant differences in expression of CD40, CD86, or MHCII on MNP subsets between groups ( Supplementary Fig. S7c-f). Similarly, the expression of residency markers and of CD28 or CD127 on T cell subsets was not statistically different between groups (Supplementary Fig.  S8). Collectively, these data indicate that asthma and COPD were not associated with substantial changes to the gross composition of the lung immune cell compartment.
Innate lymphoid cells (ILCs) have recently been recognized as major players in the regulation of mucosal barrier responses, largely through work in murine models [41,42]. Little is known as to how ILCs are affected by asthma or COPD. Although previous studies have associated COPD with an increased frequency of ILC1s in the lungs and peripheral blood of donors relative to non-smoking controls [43,44], earlier data showed no difference in ILC1s and instead indicated that group 3 ILCs, which do not express natural cytotoxicity receptor (NCR -ILC3s) are increased in the lungs of COPD donors [45]. To increase our fundamental understanding of ILCs in pulmonary disorders, we assessed the frequency of ILCs (Lin -CD56 − CD127 + cells) in a subset of our cohort (96 donors) by flow cytometry (Supplementary Fig.  S9a). We found ILC1s to be the most common ILC subset present in the human lung, accounting for 42% (mean, SEM = ±2.1%) of total ILCs (Supplementary Fig. S9b). However, in keeping with the other cell types we assessed (Fig. 5), we observed no statistical difference in the total number of pulmonary ILCs between controls and asthmatic or COPD donors ( Supplementary Fig. S9c), nor did we identify a statistical difference in the frequency of any ILC subset between these groups (Supplementary Fig. S9d). In contrast to the lack of impact of asthma or COPD on lung cellularity, when we subdivided our donors based on their smoking status we observed a significant increase in the frequency of NCR + ILC3s in control and COPD donors who self-identified as current smokers when compared to never smoker controls ( Supplementary Fig. S9e). We also observed a non-significant trend for a reduction in the frequency of ILC2s in current smoker COPD donors ( Supplementary Fig. S9e), consistent with previous results in peripheral blood [44]. Contrary to previous reports [43,44] we did not detect a robust difference in the frequency of ILC1s in COPD donors and smoking controls compared to non-smoking controls in our cohort ( Supplementary Fig. S9d).
Given the effect of being a current smoker on NCR + ILC3s in COPD donors, we asked whether smoking had any effect on the frequency of other cell types in the lung. As we only identified one current smoker asthma donor and no never smoker COPD donors in our cohort, these groups were not included in our analysis. PCA analysis did not identify clustering based on smoking status (Supplementary Fig. S10a). However, we observed significant reductions in the frequency of cDC1s present in the lungs of current smokers in both control and COPD donors compared to non-or ex-smokers in each group, respectively. Additionally, in COPD donors we observed reduced frequencies of CD14 − CD16 + monocytes, CD14 − CD16 int monocytes, and CD4 + FOXP3 − T cells in the lungs, but not the blood, of current smokers compared to ex-smokers. These data indicate that the effects of smoking on lung immune cell populations are limited to certain specific subsets and are not readily identifiable by profiling of PBMCs.

Discussion
Our understanding of the immune compartment in the human lung remains limited in scope due to the relative inaccessibility of tissue samples in healthy donors and many disease states. Although studies on cellular responses in donors during immune activation tend to be restricted to peripheral blood, the extent to which the blood reflects the immune phenotype of the lung is also poorly understood. Thus, it is challenging to parse which observations in the periphery can be extrapolated to the lung itself and which cannot. To directly address this important and understudied area, we performed flow cytometric immune phenotyping on single-cell suspensions from lung tissue resections and matched peripheral blood. We identified significant differences in the expression of maturation and residency markers in major myeloid and T cell subsets between the peripheral blood and lung tissue compartments. We observed a strong upregulation of the activation markers CD40 and CD86 on lung DCs compared to those in circulation and identified CD206 as a marker of maturation in lung cDC2s. Within the T cell compartment, we linked CD103 expression in lung CD4 and CD8 TEMs with experience of IL-7R and TCR signalling. Finally, we demonstrated that the general composition of the pulmonary T cell, monocyte, DC, and ILC compartments of asthmatic and COPD donors are comparable to controls.
MNPs are a highly heterogeneous population with overlapping functions and expression of cell surface markers, making accurate delineation of populations by conventional flow cytometry challenging [17,46], further complicated by tissue-specific expression of markers in different myeloid subpopulations [16,17,36]. To gain an overview of the MNP compartment in the human lung, we applied a conservative gating strategy utilizing accepted markers of monocyte and DC subpopulations [21,22] to distinguish four major DC subsets and three monocyte subsets. Consistent with recent observations using single-cell sequencing [47], we identified cDC2s as the primary DC population in the lung, closely followed by pDCs, although both cDC1s and moDCs were also represented in this compartment. We detected a strong increase in the expression of CD40 and CD86 on all DC subsets in the lung compared to those in circulation, suggesting that the lung environment facilitates DC maturation either as a consequence or in anticipation of exposure to environmental antigens.
We also identified differential expression of CD206 on cDC2s between the lung and the blood, with the vast majority of lung cDC2s expressing CD206 whilst the majority of blood DCs did not express this marker. Interestingly, CD206 + cDC2s in the lung expressed significantly higher levels of CD40 and CD86 than their CD206 − counterparts (Fig. 2). A previous report, that elegantly distinguished between pulmonary monocytes and DCs within the lung epithelium/stroma and those in the pulmonary vasculature, demonstrated that CD206 expression on pulmonary monocytes and DCs is restricted to extravascular cells [48]. This suggests that, in our dataset, CD206 − lung cDC2s may represent cDC2s within the pulmonary vasculature and supports CD206 as a potential marker of tissue residency in cDC2s. CD206, also known as the mannose receptor, recognizes a broad array of organisms spanning fungi, bacteria, and protists [49], with its specific expression on tissue-resident DCs implying a necessity for this receptor in the recognition of and response to invading organisms. Early investigation into expression of CD206 using in vitro derived DCs correlated LPS-induced maturation with down-regulation of CD206 [50]. We demonstrate that, relative to immature circulating cDC2s, CD206 is linked to increased frequency of CD40 and CD86 expression by cDC2s in the lung tissue. However, we cannot rule out that following stimulation CD206 expression is lost on DCs as they egress the lung. Irrespective, our data support clear phenotypic differences between circulating and tissue-resident DCs and elevate basic understanding of the lung myeloid compartment in this regard.
In terms of T cells, we identified pulmonary CD103 + CD4 + FOXP3 − and CD103 + CD8 + TEMs as being those with recent experience of IL-7R and TCR signalling, as determined by substantial downregulation of CD28 and CD127 expression within these T cell subpopulations (Fig.  4). This suggests that expression of CD103 is linked to recent T cell activation within the lung. Consistent with our results, a recent report demonstrated that pulmonary CD103 + CD4 + cells have a distinct phenotype from CD103 − CD4 + T cells, expressing a different pattern of chemokine receptors and exhibiting a greater capacity for IFNγ and TNFα production following stimulation [51]. Similarly, a previous analysis of CD103 expression on CD4 + TRM cells present in female reproductive tissue found that CD103 + cells had a greater capacity to produce IL-17A than CD103 -CD4 + cells following ex vivo PMA/ionomycin stimulation. This observation did not extend to canonically non-Th17-associated cytokines such as IFNγ, suggesting that CD103 denotes a Th17 signature at this barrier site and a specialized need for regulation of the local bacterial microbiota [32]. In agreement with this observation, CD103 + CD69 + liver CD8 + T cells have been shown to produce greater levels of IL-2 than their CD103counterparts whilst expressing comparable levels of IFNγ [11]. Similarly, CD69 + CD8 + T cells in the brain can be divided into CD103 + and CD103 − populations with distinct phenotypes [12]. In an elegant study of long-lived T cell populations in the liver following transplantation, it was observed that, whilst the majority of donor-derived CD4 + T cells express CD69, they display minimal CD103 expression [31], suggesting that CD103 is not required for long-term tissue retention or residency status of CD4 + T cells. However, a study of T cells isolated from the bronchioalveolar lavage of HLA mismatched lung transplant recipients demonstrated that 9 months post-transplantation the majority of donor T cells remaining in the airways were CD69 + , with a substantial proportion of those cells also expressing CD103 [52]. These populations existed at significantly greater frequency than at 1-month post-transplantation, indicating a correlation between airway retention and CD103 expression over time, although the presence of a significant proportion of CD69 + CD103 − CD4 + T cells would again suggest that CD103 is not required for tissue retention [52]. Interestingly, we observed comparable levels of CD28 and CD127 expression on pulmonary CD69 -CD103 + and CD69 + CD103 + CD4 + FOXP3 − and CD8 + TEM cells (Fig. 4), perhaps implying that these lung CD69 − cells represent non-resident, recently recruited, TEMs following TCR stimulation. Alternatively, these CD69 − CD103 + CD4 + FOXP3 − T cells may represent resident cells down-regulating their residency program to facilitate egress from the tissue environment. Further, we observed small CD103 + CD4 + FOXP3 − and CD103 + CD8 + T cell populations in peripheral blood, the majority of which had a TEM phenotype (Fig 3). However, unlike pulmonary T cells, these cells expressed high levels of CD28, implying a lack of recent TCR stimulation (Fig 4). A previous RNAseq analysis found that CD103 + blood CD8 + T cells were transcriptionally more similar to CD103 − blood CD8 + T cells than CD103 + pulmonary CD8 + T cells, suggesting that these cells do not represent CD8 + TRM cells which have recently egressed from the lung [53]. However, whether CD103 + blood T cells represent a population in-transit to various tissue environments or a distinct circulating population remains unclear. Collectively, these data support a model in which, independent of CD69 expression, CD103 expression on lung TEM cells marks a subpopulation of active tissue specialists with recent receptor signalling experience and functional adaptation to the tissue environment in which they are located.
Both asthma and COPD are inflammatory disorders of the airways associated with immune dysfunction [14,15]. Allergic asthma, the primary asthma endotype, is driven by an inappropriate and excessive eosinophilic Th2 response to harmless environmental stimuli (Th2-high asthma) [54]. Those donors with non-allergic/non-eosinophilic asthma (Th2-low asthma) present with greater neutrophilia and a resistance to steroid treatment [55]. Similarly, as an inflammatory condition COPD is triggered by external factors, commonly viral and bacterial infection, which drive CD8 + T cell and neutrophilic inflammation [55]. However, there is currently little understanding as to what extent these conditions reshape the lung immune compartment. We identified 17 donors with asthma and 38 with COPD within our cohort through a combination of self-reporting and clinical presentation. Somewhat unexpectedly, we did not observe major differences in the number of cells per gram of tissue when assessing a wide range of cell types and subsets in asthmatic and COPD donors relative to controls, although we noted a small reduction in NK cells in COPD donors and reduced frequency of CD64 + cells in asthmatics (Fig. 5). Further, we could not detect differences in the expression of the maturation markers CD40 and CD86 in myeloid population, nor differences in the expression of residency (CD69 or CD103) or phenotypic markers (CD45RA, CCR7, CD28, CD127) on T cell subsets (Supplementary Figs. 7 and 8).
A previous study observed statistical increases in the frequency of CD4 + and CD8 + T cells in the lungs of COPD donors as a proportion of the CD45 + compartment, as well as an increase in moDCs as a percentage of their CD11b + CD11c + HLA-DR + DC fraction [56]. Notably, the authors identified accumulation of T and B lymphocytes in regions of airways of COPD donors where the secretory immunoglobulin A (SIgA) had been lost [56], suggesting that major changes in the immune landscape of the lung occur in highly localized fashion which may not be reflected in bulk analysis by flow cytometry. It has also been observed that cDC2s isolated from COPD donors have a distinct transcriptional profile, an increased capacity to induced naïve CD4 + T cell differentiation into a T follicular helper phenotype, and localize to distinct tissue regions relative to non-COPD controls [46]. An analysis of blood and bronchiolar lavage reported a reduced frequency of γδ T cells in COPD donors, although this observation was only evident when comparing COPD donors with smokers not presenting with COPD, as COPD donors showed statistically comparable levels of γδ T cells to non-smoker controls [57]. Notably, however, these studies used relatively small numbers of donors and/or focused on small, localized differences. The data presented here takes a broader view, capturing the overall lung phenotype in over 100 donors, which may explain differences in our observations.
A role for ILCs in COPD, and more generally in response to cigarette smoke and environmental pollutants, is emerging [58]. However, there is little consensus on how the ILC compartment is affected by COPD. The evidence that is available has indicated that either ILC1s or NCR -ILC3s are increased in the lungs of donors with COPD [43,45]. However, these studies have relied on relatively small sample sizes, especially with regard to the non-COPD controls. In the 96 donors within our cohort profiled for ILC phenotype, we could not observe statistically significant differences in the frequencies of ILC1s, NCR − ILC3s, or ILC2s in COPD donors relative to non-COPD controls. Indeed, differences only became apparent when donors were separated based on smoking status, where we observed a trend to a reduction in ILC2 frequency in COPD donors relative to non-smoking controls and an increase in the frequency of NCR + ILC3s in ex and current smokers, especially those with COPD. These inconsistencies between studies are likely a result of high inter-person variability, small sample sizes, and differences in flow cytometric gating strategies used to identify ILC subsets. Further compounding the difficulty in linking changes in the ILC compartment with COPD is the recent observation that NCR + ILC3s accumulate in the lungs of non-small cell carcinoma donors [59]. Therefore, as the majority of primary lung tissue-including that used in this study-is only accessible from resected tissue proximal to tumour sites, disentangling whether NCR + ILC3 recruitment to the lung, or indeed any changes in ILC subset frequency, is a consequence of smoking, COPD, or the proximity to a tumour site is challenging.
Our data from the lungs of asthmatic donors are consistent with previous analyses of MNPs in the blood [60] and sputum [61] which reported either no difference or only weakly significant minor changes in the frequencies of various MNP subsets at steady state, although more robust differences could be observed following allergic challenge [61,62]. It has been reported that DCs expressing BDCA1 can be found specifically within the airway epithelium of asthmatics with a Th2-high phenotype, but not in those with a Th2-low phenotype [63]. The highly localized nature of this observation, and its specificity to Th2-high asthma, implies that remodelling of the immune compartment at a steady state in asthmatics is likely localized to specific tissue structures and is dependent on the endotype of a given donor. Consistent with our data, a recent single-cell analysis of human lung biopsies from asthmatics did not identify major changes in the overall composition of the immune compartment in terms of frequency of major immune cell lineages relative to non-asthmatic controls [64]. However, the authors identified significant changes to the cell-cell signalling landscape of the asthmatic lung at a transcriptional level, supporting the notion that immunological changes occurring within donors of chronic inflammatory conditions do occur, but likely require more extensive analysis methods than we have been able to apply in the current study, including more detailed characterization of condition-related changes in pulmonary immune cells at a transcriptional level.
Collectively, our data suggest that immunological differences that arise during COPD and asthma are not associated with dramatic changes in the overall lung immune cell compartment in those suffering each of these conditions. Instead, they indicate that it is functional changes within key immune cells in these donors that are likely responsible for differing responses to the same stimuli, which may include altered localization of specific cell types, differing metabolic capacities or access to metabolic substrates, different signalling events, or increased exposure to microbial or damage signals.
The data presented here enrich our overall understanding of major immune compartments within the human lung. They demonstrate major phenotypic differences between immune populations in the lung and blood and help establish markers of maturation and activation in the MNP and T cell compartments. Finally, these data suggest that in donors with asthma and COPD, at resting state, the gross composition of the pulmonary immune landscape is comparable to control donors. This indicates that immune dysregulation during these conditions is not a product of changes in the frequency of major immune cell subsets, but rather is likely due to differential functional capacity of these cells upon a sensitization/challenge event or changes in tissue localization. This highlights a need for a shift in phenotypic studies to integrate a focus on cell function and localization, rather than simply surface marker expression.

Donor samples
Lung samples were obtained via the Manchester Allergy, Respiratory and Thoracic Surgery (ManARTs) biobank from donors undergoing resection surgery for suspected lung cancer at Wythenshawe hospital (donors: n = 115, median age = 71 year; age range = 28-89 year; 68 female). Written consent was obtained from all donors and the study was approved under the National Research Ethics Service Committee; North West-Haydock ethics reference 20/NW/0302. The research use of the samples was in accord with the terms of the informed consent obtained from the donors. Lung tissue taken forward for processing was a minimum of 40 mm from the tumour site as determined by experienced pathologists and varied from 5 to 25 g in weight. In most cases, samples were processed on the same day as the surgery. Where this was not possible samples were stored at 4 o C overnight prior to processing. Fresh tissue was deposited in containers prior to transport to prevent drying out. Peripheral blood was collected into EDTA coated tubes and kept a room temperature. Samples were transported to the University of Manchester at ambient temperature.
Asthmatics were self-identified at the point of providing consent (donors: n = 17, median age = 72 year, age range = 55-83 year, 14 female). COPD donors were identified through a combination of self-identification, presentation of clinical symptoms including emphysema on chest X-ray, poor lung function measure by spirometry (FEV1/FVC < 70%), and a history of smoking (donors: n = 38, median age = 69 year, age range = 45-89 year, 20 female). Control donors were identified as those donors that did not present with asthma or COPD, and did not have a history of other inflammatory conditions, e.g. rheumatoid arthritis (donors: n = 60, median age = 71, age range = 28-86, 34 female).
Donors were stratified on smoking status based on selfreported smoking habits. Donors were defined as never smokers if they reported never having smoked, ex-smokers if they reported a history of smoking but reported not having smoked for at least 1 year prior to the date of surgery, or current smokers if they reported having smoked within 1 year prior to the date of surgery.

Tissue processing
To isolate PBMCs blood was diluted 1:2 in PBS then layered over Ficoll-Paque and centrifuged for 40 min at 400 × g at room temperature. PBMCs were collected with a Pasteur pipette, washed in Lung buffer (PBS + 10% FCS + 2 mM EDTA), counted, and resuspended in freezing medium (10% DMSO in FBS) at ~10-20 million cells/ml. PBMCs were frozen to −150 o C using a Mr Frosty Freezing Container (ThermoFisher Scientific).
To create single-cell suspensions from lung samples, resected lung tissue was weighed, diced in a sterile petri dish using a sterile razor blade, and transferred into a tube containing digestion buffer (Liberase TL 125 μg/ml, DNase 150 U/ml, in Hanks balanced salt solution H9269). The samples were then incubated at 37 o C for 40 min in a shaking incubator. Following incubation the samples were passed through a 100 μm filter, 10% FBS Hanks media was added, and then the samples were passed through a 70 μm filter. The samples were then centrifuged (400 × g, 5 min), the supernatant removed and RBC lysis performed with distilled water. Samples were then washed in 10% FBS Hanks media and layered over Ficoll-Paque, and centrifuged for 40 min at 400 × g at room temperature. Mononuclear cells were collected with a Pasteur pipette, washed in lung buffer, counted, and resuspended in freezing medium (10% DMSO in FBS) at ~10-20 million cells/ ml. All tissue processing occurred under sterile conditions in a class II safety cabinet. Isolated mononuclear cells were frozen to −150 o C using a Mr Frosty Freezing Container and stored at −150 o C until use (ThermoFisher Scientific).

Supplementary Data
Supplementary data is available at Clinical and Experimental Immunology online.