Tissue resident memory T cells inhabit the deep human conjunctiva

Mucosal linings of the body, including the conjunctiva, are enriched in tissue-resident memory T cells (TRMs) whose defining feature is their continual tissue protection that does not rely on migration to lymphoid organs to elicit immune responses. Hitherto, conjunctival TRMs have only been identified in the superficial epithelium. This work aims to develop a more complete understanding of the conjunctival immunological capacity by investigating the presence of TRMs within the deeper, more stable layers of the healthy human conjunctiva. Using immunofluorescence microscopy and antibodies against CD3, CD4, CD69 and HLA-DR on bulbar conjunctival biopsies obtained from 7 healthy adults (age range = 32–77 years; females = 4), we identified CD69+TRM subsets in all layers of the human conjunctiva: the superficial epithelium, the basal epithelium, the adenoid, and the fibrous layers. Interestingly, the adenoid layer showed significantly higher densities of both CD4 and CD8 TRMs when compared to the fibrous layer and conjunctival epithelia. Additionally, CD4 TRMs predominated significantly over CD8 TRMs in the adenoid layer. The abundance of deep conjunctival CD69+TRMs within the healthy human may suggest the presence of defence mechanisms capable of inducing long-term immunogenic memory. Understanding this spatial distribution of conjunctival CD69+TRMs is essential to improving mucosal vaccine design.

www.nature.com/scientificreports/ Recently, tissue-resident memory T cells (T RMs ) have been critical cells of interest in mucosal vaccine studies. These cells are a subtype of non-circulating memory T cells that remain localised in peripheral tissues after antigenic challenge to enhance long-term immunity and immunosurveillance upon re-exposure 11,12 . Their broad and long-lasting immune protection at mucosal surfaces, including their ability to secrete inflammatory and antiviral cytokines, and communicate with dendritic cells and macrophages [13][14][15][16][17] , may provide a key immune effector mechanism for mucosal vaccines. Furthermore, as contagious pathogens are most often encountered at peripheral mucosal surfaces, the continual presence of a mucosal T RM population is important for pathogen control and elimination at the site of entry 18 . In contrast, tissues without a local T RM population allow pathogens to multiply and circulate within the body whilst the host recruits circulating memory T cells to the site of infection 18,19 . Therefore, the induction of T RMs has become a new vaccine development strategy.
A common clinical protocol for conjunctival tissue collection is ocular surface impression cytology, in which an absorbent filter paper is pressed onto the surface of the conjunctiva to remove the most superficial layer of cells 20 . Cells present within this superficial layer are perpetually exposed to friction from the blinking eyelid, and thus are sloughed off into the tears to be replaced by basal epithelial cells 21,22 . In 2017, Bose et al. 23 used impression cytology to examine the resident populations of T cells in the healthy conjunctiva. Their flow cytometric analysis determined that the conjunctiva is protected by two T RM subsets, CD69 + CD103 − and CD69 + CD103 + cells. However, the continual sloughing of the superficial epithelium where these T RM subsets were identified, minimises their most crucial function: their long-term protection of the ocular surface. In contrast, deep conjunctival layers remain stable with their resident cell populations as they are not exposed to these frictional forces from blinking. Thus, immune cells populating the deep conjunctiva may provide the ocular surface with longer-lasting, specific immune protection. Additionally, T RMs in the substantia propria could indicate potential communication with antigen-presenting cells that are abundant within this layer 24 and may allow for a more robust elimination of pathogens. Nevertheless, the presence and relative proportions of T RM cell subsets in the deep conjunctiva are yet to be studied in healthy humans. Such information could provide insight into the immunological response profile of the normal conjunctiva and allow for the diagnosis of abnormal profiles from inflammatory ocular diseases.
In the present study, we investigated healthy human conjunctival biopsies with complete and preserved histomorphologies of the epithelium and substantia propria to determine the presence and distribution of T RM cell subsets in the deep conjunctival layers using immunofluorescent microscopy.

Results
A total of 18 healthy cataract surgery patients with no underlying systemic conditions were enrolled, and all of them met the inclusion criteria. The analysis was completed on seven conjunctival biopsies from 7 healthy cataract patients (age range = 32-77 years; females = 4) (see supplementary Table S1). Excluded cases, 11 in total, had tissues demonstrating inflammation or severe folding determined by histological analysis due to their collection using an unoptimized protocol at the time. The surgical excisions preserved the entire histological anatomy of the conjunctival tissue and comprised the entire superficial and basal epithelia and the underlying loose layers of the substantia propria; except for one conjunctival sample where the fibrous layer was not surgically collected.  CD4, CD69 and HLA-DR was optimised for immunofluorescence microscopy and used on healthy conjunctival sections to enumerate T RM cell subsets. As most CD3 + T cells mature to become either CD4 + or CD8 + singlepositive lymphocytes 25 , this study classified CD4 T RMs as CD3 + CD4 + CD69 + HLA-DR − and CD8 T RMs as CD3 + CD4 − CD69 + HLA-DR − ( Fig. 2A). We show that both CD4 T RMs and CD8 T RMs are present in all layers of the bulbar conjunctiva, including the deep conjunctiva (see supplementary Tables S2-S5). Specifically, the average number of CD4 T RM cells per mm 2 was significantly higher in the adenoid layer (292.9 ± 92.2) compared to the superficial epithelium (17.9 ± 3.8; p = 0.0034), basal epithelium (61.9 ± 28.0; p = 0.0225) and the fibrous layer (28.6 ± 7.6; p = 0.0143). When we focused on the average number of CD8 T RMs per mm 2 , we saw no significant difference in the cell count across the conjunctival layers. However, there was a trend for CD8 T RMs to congre- www.nature.com/scientificreports/ gate within the basal epithelium (65.1 ± 33.7) with lower cell counts in the superficial epithelium (28.3 ± 13.2), adenoid layer (24.1 ± 10.7) and fibrous layer (9.2 ± 6.0). Within the conjunctival layers, CD4 T RMs predominated over CD8 T RMs in the adenoid layer of the substantia propria (292.9 ± 92.2 CD4 T RMs /mm 2 ; 24.1 ± 10.7 CD8 T RMs / mm 2 ; p = 0.0024). CD4 T RM and CD8 T RM cell counts per mm 2 across the superficial epithelium, basal epithelium and fibrous layer showed no significant difference in abundance (Fig. 2B).
CD69 + T RM subsets constitute a large proportion of conjunctival CD3 + T cells. CD69 + T RMs of the superficial epithelium made up 86% of the total conjunctival CD3 + T cell population with CD8 T RMs constituting 53% of CD3 + T cells while CD4 T RMs making up the remaining 33% (Fig. 2C). In the basal epithelium, both CD4 T RMs and CD8 T RMs formed 14% of the CD3 + T cell population each. We show that within the substantia propria, the CD4 T RM population formed the predominate percentage (51% in the adenoid and 29% in the fibrous) of the CD3 + T cell population compared to the CD8 T RMs (4% in the adenoid and 9% in the fibrous). Finally, we demonstrate, by means of a Pearson's Chi-squared test, that there is a significant relation between the tissue layer and the type of CD3 + T RM cell subset residing within the layer (p < 0.0001) (Fig. 2C).

Discussion
This study reports the in-situ observation of CD4 and CD8 T RMs in the superficial epithelium, the basal epithelium, the adenoid and the fibrous layers of the healthy human bulbar conjunctiva. To our knowledge, this is the first spatial distribution analysis of T RMs done with complete human conjunctival biopsies. To perform this, we designed and optimised an immunofluorescence panel with antibodies staining against the CD3, CD4, CD69 and HLA-DR surface markers. Using this approach, we demonstrate that human CD4 T RMs and CD8 T RMs are present throughout the entire conjunctival tissue. This confirmed the presence of these T RMs in the superficial epithelium by Bose et al. 23 , where impression cytology was used, as well as identified their presence for the first time in all three deep conjunctival layers: the basal epithelium, the adenoid layer and fibrous layer. As cells of the superficial conjunctival epithelium are continuously sloughed off and shed due to frictional forces from blinking 21,22 , determining the spatial distribution of T RMs in the stable layers of the deep conjunctiva is essential for providing the ocular surface with long-term local immune memory and defence. CD69 + T cells identified using our optimised panel reflect tissue-resident memory T cells that were easily distinguished from circulating memory T cells by their positive CD69 expression 26 . CD69 is an activation marker as well as an inhibitor of sphingosine 1-phosphate receptor-1 that functions to prevent cell egress from tissues 27,28 . HLA-DR is a prominent T-cell activation marker that defines activated T cells as well as myeloid cells including dendritic cells 29,30 . Thus, CD69 detection in the absence of other activation markers such as HLA-DR suggests that CD3 + T cells identified using the optimised immunofluorescence panel are CD4 or CD8 T RMs . Mean values of CD4 T RM cell numbers per mm 2 indicated that the majority of these cells congregated abundantly within the adenoid layer of the substantia propria just below the basement membrane. T RMs have a distinct profile dependent on specific cytokines including IL (Interleukin)-15, IL-17, IL-33 and transforming growth factor beta (TGF-β) that may be required for their differentiation and continual residency within the tissue 18,31 . Such cytokines are produced by epithelial cells and immune cells, viz lymphocytes and macrophages 16,[32][33][34] . The greater abundance of immune cells in the substantia propria compared to the epithelium 24 , and the close association of the substantia propria's adenoid layer to the basal epithelial cells may explain the larger distribution and establishment of T RMs within the adenoid layer. The presence of CD8 T RMs below the superficial epithelium is important as they may be capable of providing adaptive immune responses, especially antiviral cytokines such as interferon-gamma (IFN-γ) 13 , that prevent the establishment of infection in the epithelium and spread to deeper layers of this tissue.
Within the mucosal epithelia of the body, including the conjunctival epithelium, CD8 + T cells are found more abundantly than CD4 + T cells; the opposite is true in the subepithelial lamina propria 35 . Similarly, CD4 T RMs in the human skin accumulate within the dermis as opposed to CD8 T RMs that are more abundant within the epidermis 24,36 . However, our results demonstrate that conjunctival CD8 T RMs do reside in the substantia propria, and are only significantly outnumbered by CD4 T RMs in one layer, the adenoid layer. As the downregulation of the transcription factors Eomesodermin and T-bet is a requirement for CD8 T RM formation 37 , functional analysis on the expression levels of such transcription factors by conjunctival CD8 T RMs may determine if they form a distinct subset of CD8 T RMs that could explain their similar distributions across all conjunctival layers and also their residence within the unique microenvironment of the conjunctival lamina propria-the substantia propria. CD8 T RMs have been demonstrated to move slowly in the epidermis of the skin, allowing them to target cells of interest 38 . This calls into question whether conjunctival CD8 T RMs are also capable of clearing viral pathogens in infected cells. In other mucosal tissues, T RMs have two antiviral functions: the first is the direct cytotoxicity of infected cells, limited in scope by their relative immobility, and the second is the secretion of antiviral cytokines which diffuse more widely 15 . In general, CD4 + T cells secrete antiviral cytokines and act synergistically with CD8 + T cells via cognate interactions with local dendritic cells 39 . They also support and maintain the CD8 + T cell responses through the production of cytokines such as IL-21 and IFN-γ 40,41 . Therefore, our identification of CD8 and CD4 T RMs in the deep layers of the conjunctiva is vital for understanding the immune response of this tissue. Our findings that T RMs populate the deep conjunctiva denotes that they may be induced in these layers following antigenic challenges administrated to the conjunctival tissue to establish immunogenic memory against specific pathogenic antigens. Thus, T RMs induced in the deep conjunctiva would be able to provide long-term immunity against specific pathogens and, in the case of T RMs of the substantia propria, they may be able to communicate with plasma cells and mononuclear phagocytes such as macrophages that are present within this layer 24 to elicit a rapid, yet vigorous immunogenic response before the pathogen can enter the lymphatics or blood vasculature.
This study is not without its limitations. Notably, the small number of tissues and limited age range call for further studies to be conducted with a larger number of healthy subjects with a focus on the senior age group www.nature.com/scientificreports/ (80 + years old). Additionally, T RMs in this study were identified based on the consideration that CD69 is acting as a marker of cellular tissue residency rather than an activation marker when a cell is HLA-DR − : HLA-DR is considered a late activation marker 42 while CD69 is an early T cell activation marker, expressed as early as 2-3 h following stimulation 43 . In view of these kinetics, it is possible but unlikely to have included early activated T cells and thus overestimated the number of T RMs counted in the layers of the conjunctiva. It was not possible to incorporate into our IF panel additional activation markers such as CD25, CD71 or Ki67, whose expression levels start to peak around 24-48 h after antigenic challenge 44 compared to HLA-DR, which is increased even later in the T-cell activation process with its upregulation at 48-72 h after antigenic challenge 42,45 . However, we included the HLA-DR marker rather than earlier activation markers to avoid including CD4 + myeloid cells in the T RM cell count. Nevertheless, as all samples were biopsied from healthy participants, inflammatory and activation markers associated with inflammation were not expected to be prominent. Finally, although the substantia propria of tarsal conjunctivae demonstrates a larger abundance of immune CD8 + T cells than the bulbar conjunctiva 6 , this tissue is more difficult to collect from patients. Further work may reveal a more abundant presence of CD8 T RM subsets in the tarsal conjunctivae; hence, strengthening the incentive to expand research on human ocular mucosal vaccines to induce them. Nevertheless, this work shows that the bulbar conjunctiva alone may prove to be an effective site for T RM induction by conjunctively applied vaccines.
In the present study, we use immunofluorescence microscopy on conjunctival biopsies that preserve their entire histomorphology to determine the distribution of T RMs in the superficial epithelium and the deeper conjunctival layers. We demonstrate that the entire human conjunctiva is protected by an abundant number of both CD4 and CD8 T RMs in all conjunctival layers, with the majority of T RMs congregating in the adenoid layer just below the basement membrane. Additionally, we show a trend for CD4 T RMs to predominate over CD8 T RMs in the substantia propria. As increased presence of some T RM subsets may contribute to local ocular inflammation such as in dry eye disease and experimental autoimmune uveoretinitis 23,46 and were shown to maintain vitiligo disease in a murine model 47 , identifying abnormal levels of conjunctival T RMs based on insight from the normal biological milieu of the tissue will enable their selective targeting and deletion during therapeutic interventions. In relation to mucosal vaccine design, the use of eye drops to target the induction of antigen-specific T RMS within the deep conjunctiva via eye drops may boost the efficacy of prior systemic vaccine regimens. Such cell induction into the mucosa may be possible without the generation of inflammatory responses by the selective recruitment of endogenous antigen-specific T cells using chemokines after parenteral vaccine priming 48 . T RMs ' continual conjunctival surveillance and long-lasting immunogenic memory, along with their ability to interact with neighbouring immune cells to secrete cytokines may allow for effective immune mediation against local infections. Additionally, as mucosal sites of the body are functionally connected 7 , immunological activity against pathogens may not only occur in CALT, but also in NALT and BALT; allowing for wide protection across mucosal tissues where pathogen entry is prevalent.

Materials and methods
Ethics approval. This study was in strict compliance with the tenets of the Declaration of Helsinki and Australian Regulations concerning the use of human tissues for biomedical research. Experimental protocols within this study were approved by the Western Sydney Local Health District (WSLHD) Human Research Ethics Committee (HREC 2020/ETH02012).
Inclusion and exclusion criteria. Healthy cataract surgery patients, both men and women at least 18 years old and willing to provide informed consent were enrolled in this study. Patients were excluded from the study if they had any other ophthalmic disorder, additional systemic diseases or if they were pregnant women, Aboriginal or Torres Strait Islanders or from a vulnerable, intellectually, or mentally challenged group.
Source of conjunctival samples. Human bulbar conjunctival biopsies (n = 7) of approximately one mm 2 were obtained immediately following surgical excisions performed under local sub-tenon anaesthetic procedures on healthy patients during their routine cataract surgery at Westmead Hospital. Surgical excisions for the purposes of specimen collection were performed at the beginning of the cataract surgeries. All patients provided written consent prior to sample collection. Tissue samples were spread using fine forceps onto an absorbent cellulose filter paper that oriented the specimen so that the superficial surface of the tissue faced upwards. The tissue was allowed to adhere to the paper for 30 s, then was placed immediately to be floating in 5 mL of 4% paraformaldehyde (Electron Microscopy Sciences, PA, USA) in 50 mL collection tubes for 4 h at room temperature in the dark. Tissues were stored in 70% ethanol prior to paraffin-embedding.
Histological processing. Fixed tissues were processed in an Excelsior ES tissue processor (ThermoFisher Scientific, Runcorn, England, UK) for 4 h. After paraffin embedding in a cross-sectional orientation, 4 μm paraffin sections were cut and collected onto SuperFrost glass slides (ThermoFisher Scientific, MA, USA). Slides were baked for 2 h at 60 °C, deparaffinised in xylene, and rehydrated with decreasing concentrations of ethanol. Some of the serially sectioned slides were stained with haematoxylin and eosin to determine tissue orientation and structure; others were stained with PAS-Mayer's haematoxylin to determine the distribution of goblet cells and to confirm the identity of the conjunctival tissue. The rest of the slides were stained with fluorochrome-labelled antibodies for immunofluorescence microscopy. Tissue samples excluded from immunofluorescent staining had folded regions, heavily rolled areas, inflamed epithelia defined by polymorph infiltration, missing goblet cells or large areas filled with erythrocytes from ruptured conjunctival capillaries. For all immunofluorescent staining, the tissues were stained in parallel with the following negative controls: (1) The omission of primary antibodies; (2) The omission of primary and secondary antibodies, leaving only DAPI nuclear staining; (3) The omission of autofluorescence quenching to rule out the possibility of overquenching and reducing signals from fluorophore-labelled antibodies.
Immunofluorescence microscopy. Widefield imaging of the entire conjunctival tissue samples was performed on the Olympus VS120 Slide Scanner using the VS-ASW imaging software (version 2.9) as a base system. The ORCA-FLASH 4.0 VS: Scientific CMOS sensor was used to capture all images under the × 20 objective lens (UPLSAPO 20X; NA = 0.75). Channels scanned were DAPI (10 ms exposure), FITC (100 ms exposure), TRITC (500 ms exposure), Cy5 (500 ms exposure) and Cy7 (1000 ms exposure). Channels from the sectioned images were converted to 16-bit TIFF format, then merged and analysed using FIJI (ImageJ). The maximum and minimum brightness and contrast were adjusted in all channels to create the final composite image. Conjunctival layers were separated by creating masks that allow for T RM cell counting in each layer using the ImageJ cell-counter plugin. For each donor, three sections were analysed for cell counts per mm 2 and the results were averaged. Statistical analysis. GraphPad Prism 9.1.1 (GraphPad Prism, San Diego, CA, USA) software was used for all statistical analyses and graph plotting. Square root transformation was used to improve the distribution of the data. The normality assumption was met, and the assumption of equal variance was checked by looking at the data's residual plot. Thus, significance in the number of T RMs between the different layers of the conjunctiva and in the number of CD4 T RMs to CD8 T RMs was determined by a two-way analysis of variance (ANOVA) test followed by a Tukey HSD multiple comparison test. Additionally, a chi-squared test was performed to determine if there is an overall association between tissue layer and cell type presence. Results are given as the mean ± standard error of the mean (S.E.M.). A value of P < 0.05 was considered significant.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.