Phenotypic, molecular and functional characterisation of human in vitro-generated IL-17A+ CD8+ T-cells

IL-17A+ CD8+ T-cells, often referred to as Tc17 cells, have been identified at sites of inflammation in several immune-mediated inflammatory diseases including psoriasis and spondyloarthritis. Whilst much of our understanding of IL-17A+ CD8+ T-cells has been discerned from murine studies, human IL-17A+ CD8+ T-cells remain less-well characterised. We optimised an in vitro polarisation protocol to expand human IL-17A+ CD8+ T-cells from PBMC or bulk CD8+ T-cell populations for phenotypic and functional assessment. We show that T-cell activation in the presence of IL-1β and IL-23 significantly increased the frequencies of IL-17A+ CD8+ T-cells, which was not further enhanced by the addition of IL-6, IL-2 or anti-IFNγ mAb. In vitro-generated IL-17A+ CD8+ T-cells from healthy donors displayed a distinct type-17 profile compared with IL-17A- CD8+ T-cells, as defined by transcriptional signature (IL17A, IL17F, RORC, RORA, MAF, IL23R, CCR6, CXCR6); high surface expression of CCR6 and CD161; and polyfunctional production of IL-17A, IL-17F, IL-22, IFNγ, TNFα and GM-CSF. A significant proportion of in vitro-induced IL-17A+ CD8+ T-cells expressed TCRVα7.2 and bound MR1 tetramers, indicative of a MAIT CD8+ T-cell population. Using an IL-17A secretion assay, we demonstrate that the in vitro-generated IL-17A+ CD8+ T-cells were biologically functional and induced pro-inflammatory IL-6 and IL-8 production by synovial fibroblasts from patients with psoriatic arthritis. Collectively, we report an in vitro culture system to expand IL-17A+ CD8+ T-cells and further characterise their phenotype, transcriptional regulation and functional relevance to human health and disease.


Introduction
Interleukin-17A (IL-17A) was originally identified as an effector cytokine produced by T-helper (Th) CD4+ T-cells (1) and together with fellow family member IL-17F, has come to characterise a distinct lineage termed Th17 cells. Multiple studies have described the factors and conditions that drive the differentiation of murine and human Th17 cells as well as their involvement in both host-protective and pathological immune responses (2,3). Additionally, robust evidence implicates IL-17A (either independently or synergistically with other proinflammatory mediators (4,5)) to have a pivotal function in the pathogenesis of immunemediated inflammatory diseases such as psoriasis, psoriatic arthritis (PsA) and axial spondyloarthritis for which IL-17A is now a therapeutic target (6,7).
In vivo and in vitro studies have highlighted IL-17A production is not restricted to Th17 cells; instead, a collection of conventional, unconventional T-cells and innate-like cell types may serve as alternative sources of IL-17A and IL-17F either in health or disease-relevant tissues (reviewed in (8,9)). In particular, IL-17A+ CD8+ T-cells, often referred to as Tc17 cells, are known to share phenotypic and functional features of Th17 cells (reviewed in (10)). This includes expression of the lineage-committing transcription factors retinoic acid-receptor (RAR)-related orphan receptor (ROR)γt and musculoaponeurotic fibrosarcoma (c-MAF), surface expression of the typical type-17 markers CD161, CCR6 and IL-23R, and concomitant expression of Th17 (IL-17A, IL-17F, IL-21, IL-22 and granulocyte macrophage colony-stimulating factor (GM-CSF)) and cytotoxic CD8+ Tc1 (interferon (IFN)γ and tumour necrosis factor (TNF)α) related cytokines. The presence of these Tc17 cells has been described at different tissue sites in a variety of human infectious, autoimmune and inflammatory diseases, including tuberculosis (11), multiple sclerosis (12), inflammatory bowel disease (13) and psoriasis (14)(15)(16). We have also previously identified an enrichment of Tc17 cells in the joints of PsA patients that correlated with disease activity measures (17,18). Whilst phenotypic and transcriptomic profiles of Tc17 cells derived ex vivo from patients with inflammatory disease have been reported, functional studies remain limited due to the relative paucity of these cells (19,20). In vitro expansion of Tc17 cells could overcome this limitation and allow phenotypic and functional studies of this cell population.
Human in-vitro generated IL-17A+CD8+ T-cells 4 To date, several murine in vitro studies have demonstrated that Tc17 cell differentiation can be achieved by applying Th17 polarising protocols, largely inclusive of a combination of transforming growth factor (TGF)β, IL-6, IL-1β, IL-21, IL-23, anti-IL-4 and anti-IFNγ (reviewed in (10)). Analogous to Th17 cells, murine Tc17 cells were shown to be a heterogeneous population including non-pathogenic and pathogenic subtypes, the latter typically defined by enhanced cytotoxicity with dual IL-17 and IFNγ expression promoted by IL-23 (21). However, few studies have described factors driving human Tc17 cells. The small collection of studies thus far offers a consensus in that IL-1β, IL-6, IL-23 without or with TGFβ and anti-IFNγ can direct Tc17 cell responses (11,13,(22)(23)(24)(25)(26). In addition, evidence for involvement of cellular drivers comes from reports that activated monocytes from tumour sites rather than nontumour tissues more potently induced IL-17A+ CD8+ T-cells in vitro (27), and that pleural mesothelial cells from patients with tuberculosis infection significantly enhanced IL-17 production by patient blood-derived CD8+ T-cells, in a cell-cell contact-dependent manner (11).

Mucosal-associated invariant T (MAIT) cells are an innate-like T-cell subset defined by high
expression of CD161 and their semi-invariant αβ T-cell receptor (TCR) restricted to Vα7.2-Jα33/Jα12/Jα20 (28). Expression of TCRVα7.2 restricts MAIT cells to the non-polymorphic MHC class Ib molecule MHC-related protein 1 (MR1), which presents microbial-derived metabolites of riboflavin (vitamin B2) biosynthesis including 5-(2-oxopropylideneamino)-6ribitylaminouracil (5-OP-RU) (29). These unconventional T-cells are abundant in human tissues such as the gut and liver as well as in peripheral blood where they are predominantly CD8+ (enriched for CD8αα). Owing to their innate-like capacity, MAIT cell activation is elicited through either a TCR-dependent, TCR-independent, or synergistic manner, with the local cytokine milieu purportedly enhancing their effector function (30,31). This innate-like functionality and particularly, a type-17 program is imparted by expression of the transcription factor promyelocytic leukemia zinc finger (PLZF) in MAIT cells as well as in other unconventional T-cells (32,33). Subsequently, MAIT cells can express several type-17 associated markers including IL-23R, CCR6, RORγt and are potent producers of IL-17A and IL-17F that, together with IFNγ, TNFα and granzymes, rapidly orchestrate protective antimicrobial responses (reviewed in (34)). Moreover, and akin to classical Tc17 cells, MAIT cells (notably of a type-17 phenotype) have also been implicated in several inflammatory diseases including psoriasis and spondyloarthritis (35)(36)(37).
Human in-vitro generated IL-17A+CD8+ T-cells 5 The accumulating evidence for the presence of Tc17 cells in human inflammatory diseases together with the relative paucity in knowledge regarding their induction and function provides a strong rationale for detailed characterisation of these cells. Here, we describe a protocol to polarise and increase the frequency of human IL-17A+ and IL-17F+ CD8+ cells from either human PBMC or purified CD8+ T-cells, and we report their phenotypic and transcriptional profile. Furthermore, we provide evidence that human in vitro-induced IL-17A+ CD8+ T-cells are bioactive and elicit disease-relevant pro-inflammatory responses from PsA patientderived synovial fibroblasts, suggestive of a functional role of these cells in PsA joint inflammation. penicillin/streptomycin, 2% L-glutamine and 1mg/mL amphotericin B (all ThermoFisher Scientific). Tissue explants were incubated at 37°C in an atmosphere of 5% CO2;

Samples and cell isolation
supplemented DMEM medium was replenished every 3 days. Synovial fibroblasts that had migrated out of the synovial tissue were collected and maintained in T175 flasks for cell line Human in-vitro generated IL-17A+CD8+ T-cells 6 generation. Fibroblasts were passaged with1X trypsin (Sigma-Aldrich) once they reached 80% confluency and were used in cultures when between passages 3-7.

Flow cytometric analysis
For intracellular cytokine staining, PBMC or CD8+ T-cells were stimulated, either ex vivo or following 3-day culture, with PMA (50 ng/mL) and ionomycin (750 ng/mL, both Sigma-Aldrich)  subsets were either stored in TRIzol ® Reagent (ThermoFisher) at -80°C for later qPCR array analysis or used directly for functional assessments. For the latter, sorted T-cells were either added to fibroblasts as described below or cultured for 20 hours at 37°C (5% CO2) in complete culture medium for supernatant generation.

PsA synovial fibroblast co-cultures
PsA synovial fibroblasts were seeded (1x10 4 per well) in a flat-bottomed 96-well plate in supplemented DMEM medium and incubated for 24 hours at 37°C (5% CO2). Following Human in-vitro generated IL-17A+CD8+ T-cells 8 supernatant removal, fibroblasts were cultured in supplemented DMEM in the absence or presence of 20% (v/v) cell culture supernatants from FACS-sorted CD8+ T-cell populations for a further 24 hours, after which supernatants were collected and analysed for IL-6 and IL-8 production. Alternatively, fibroblasts were co-cultured for 24 hours with CSA-FACS sorted IL-17A+ or IL-17A-CD8+ T-cells at a 1:2.5 fibroblast to T-cell ratio, in the absence or presence of anti-IL-17A mAb (secukinumab, Novartis) and/or anti-TNFα blocking antibodies (adalimumab, Abbott Laboratories) or isotype control mAb (all human IgG1 and added at 5μg/mL).

Cytokine detection
The presence of IL-17A in T-cell culture supernatants and of IL-6 and IL-8 in fibroblast culture supernatants was quantified by ELISA according to manufacturer instructions (deluxe or standard kits, respectively, all BioLegend). Plates were read at 450 nm using a Spark 10M IL-17A and IL-17F were measured on separate assay plates due to the cross-reactivity of the magnetic beads with IL-17AF.

RNA extraction and cDNA synthesis
Total RNA was extracted from healthy donor CSA-FACS sorted IL-17A+ CD8+ and IL-17A-CD8+ T-cell subsets (4,175 -43,856 cells or 1x10 6 cells, respectively) using the TRIzol ® Reagent phenol-chloroform extraction method in combination with Phasemaker tubes (ThermoFisher). To increase total RNA precipitation, during the isopropanol step, samples were kept at -80°C for 24 hours and 1 µl GlycoBlue TM Coprecipitant (ThermoFisher) was added. RNA yield and integrity (mean RIN = 8.9) were assessed by Bioanalyzer (Waterloo Genomics Centre, KCL) and RNA was stored short-term at -80°C. Amount of RNA input for complementary DNA (cDNA) transcription was standardised within donor paired samples by adjusting RNA concentrations to the total extracted from the IL-17A+ CD8+ T-cell subset (range 30.2ng -213.0ng). First-strand cDNA synthesis was performed as a 20 µl reaction using the LunaScript ® RT SuperMix Kit (NEB, following the manufacturer's protocol) and cDNA was stored short-term at -20°C before transcriptional assessment.

Quantitative real-time polymerase chain reaction (RT-qPCR)
Gene expression profiles of in vitro-generated IL-17A+ CD8+ and IL-17A-CD8+ T-cells were assessed by bespoke TaqMan ® qPCR array (all 96 primer assays including 3 endogenous controls and 93 targets, were selected from ThermoFisher, details listed in Supplementary Table 2). Array cards and sample cDNA templates were prepared according to manufacturer's instructions using the TaqMan ® Fast Advanced PCR Master Mix (ThermoFisher). RT-qPCR was performed using a QuantStudio TM 7 Flex System (Applied Biosystems) with the following amplification conditions: hold at 50°C for 2 min then hold at 92°C for 10 min followed by 40 cycles of 95°C for 1 sec and 60°C for 20 sec. Housekeeping

Data analysis and statistical testing
Graphs were constructed and statistical tests performed using GraphPad Prism v9.1. Sample sizes with n<8 or when the population assumed a non-normal distribution were tested nonparametrically using a Wilcoxon signed-rank matched pairs test. Data sets with n>8 were tested for normality using the D'Agostino and Pearson omnibus normality test then tested for significance using the appropriate parametric or non-parametric test as stated in the figure legends. For transcriptional data analysis, undetectable genes (ZBTB32 and CD5L) were filtered from initial principal-component analysis and heatmap generation; both were computed in R using prcomp and pheatmap functions. Genes that were identified to have very low/negligible expression in both IL-17A+ CD8+ and IL-17A-CD8+ T-cells with reference to the geometric mean of CD4 normalised expression, were excluded from statistical testing and relative fold change analysis (IL17B, IL17C, IL17D, IL25, IL17RC and CCR9). Differential statistical analysis was performed on normalised expression values using a parametric paired Student's t-test with Holm-Šídák multiple comparisons test with adjusted p-values (p<0.05) reported.

Human peripheral blood IL-17+ CD8+ T-cells are detected at low frequencies ex vivo and can be expanded in vitro in the presence of IL-1β and IL-23
We first sought to determine the ex vivo frequencies of IL-17A+ and IL-17F+ CD8+ T-cells in human peripheral blood. Freshly isolated or cryopreserved healthy donor PBMC were   (Figure 5B, C). This Human in-vitro generated IL-17A+CD8+ T-cells 15 revealed that both in vitro-induced IL-17A+ CD8+ T-cell subsets predominantly comprised polyfunctional cells that express multiple pro-inflammatory cytokines (triple-positive for IL-17A+IFNγ+TNFα+, orange pie segment, or quadruple-positive for IL-17A+IFNγ+TNFα+GM-CSF+, red pie segment). We only observed a statistically significant difference in the proportion of single-positive IL-17A+ cells which was higher in IL-17A+ Vα7.2-compared with IL-17A+ Vα7.2+ CD8+ T-cells (purple pie segment, p=0.0104) (Figure 5B, C). Additionally, in line with our molecular profile, we found that a high proportion of in vitro-generated IL-17A+ Vα7.2-as well as Vα7.2+ CD8+ T-cells harboured intracellular protein stores of granzyme B (n=2, median 94.6% versus 97.9%, respectively) and to a lesser extent granzyme A (n=2, median 49.1% and 69.4%, respectively) suggestive that both cell types also shared cytotoxic potential (data not shown).
Using a Luminex® assay, we then measured the production of the cytokines IL-17AA/AF, IL-

In vitro-induced IL-17A+ CD8+ T-cells are functional, with capacity to induce proinflammatory cytokine production by PsA synovial fibroblasts
As our final step, we sought to determine the functional contribution of in vitro-induced IL-17A+ CD8+ T-cells, by investigating their ability to promote clinically relevant proinflammatory cytokine production in an in vitro model of joint inflammation. For this, cell culture supernatants were collected from in vitro-induced and then CSA-FACS-sorted IL-17A+ or IL-17A-CD8+ T-cells. Supernatants were added (20% v/v) to synovial tissue fibroblasts from patients with PsA. After 24 hours, fibroblast culture supernatants were collected for analysis Human in-vitro generated IL-17A+CD8+ T-cells 16 of the pro-inflammatory cytokines IL-6 and IL-8. Addition of IL-17A+ CD8+ T-cell culture supernatant led to a significant increase in IL-6 and IL-8 production by PsA fibroblasts as compared with fibroblasts cultured in medium alone, whilst no significant increase in either IL-6 or IL-8 secretion was observed when PsA fibroblasts were cultured in the presence of IL-17-CD8+ T-cell supernatants (Figure 6A).
We next aimed to determine whether there were differences in these pro-inflammatory responses when fibroblasts were cultured in the presence of secretory products from in vitro- (median 63,700pg/mL versus 29,500pg/mL and 1,500pg/mL versus 280pg/mL, respectively), this difference did not reach statistical significance (Figure 6B).
IL-17A is known to act synergistically with TNFα to promote pro-inflammatory cytokine production by stromal cells (4,5). Given our observation that in vitro-induced IL-17A+ CD8+ T-cell subsets actively secreted IL-17A and TNFα (Figure 5), we investigated the effect of single and dual blockade of these cytokines in co-cultures of PsA fibroblasts and FACS-sorted in vitro-induced IL-17A+ CD8+T-cells (1:2.5 cell ratio). First, we identified that as with T-cell supernatants, fibroblasts co-cultured with in vitro-induced IL-17A+ CD8+ T-cells produced significantly higher levels of IL-6 and IL-8 compared with fibroblast monocultures (p=0.0001 and p=0.0153, respectively) ( Figure 6C). Addition of an isotype control mAb to the co-cultures did not affect IL-6 or IL-8 production. IL-6 and IL-8 production was lower in the presence of either secukinumab (anti-IL-17A) or adalimumab (anti-TNFα) and was significantly reduced upon combined blockade of IL-17A and TNFα (p=0.0153). Collectively, these data demonstrate that in vitro-induced IL-17A+ CD8+ T-cells and their secretory products are biologically active, with capacity to significantly increase pro-inflammatory IL-6 and IL-8 production by synovial fibroblasts from patients with PsA.

Discussion
We report herein that human IL-17A+ and/or IL-17F+ CD8+ T-cells can be expanded in vitro upon anti-CD3 and anti-CD28 stimulation in the presence of IL-1β and IL-23. These cells are characterised by typical type-17 phenotype, cytokine and transcriptional profiles.
Furthermore, we demonstrate that in vitro polarisation induces both conventional IL-17A+CD8+ T-cells as well as unconventional IL-17A+ CD8+ MAIT cells, but that these cell types have a shared phenotype and are functionally active, with the capacity to drive biologically relevant pro-inflammatory cytokine production from PsA synovial tissue-derived stromal cells.
Our study confirms previous reports that detected low ex vivo frequencies of IL-17A+ CD8+ T-cells in healthy human peripheral blood with a rare presence of IL-17F+ CD8+ T-cells (40,(43)(44)(45)(46) and adds weight to the few studies that reported on the expansion of human IL-17expressing CD8+ T-cells (22,23). We also demonstrate that equivalent IL-17A+ CD8+ T cell induction can be achieved using either cryopreserved or freshly isolated PBMC, indicating that our protocol can be used with bio-banked samples. Kondo (18,40,49). Additionally, some of the more novel markers associated with human Th17 and/or Tc17 cells were found more abundantly expressed in our IL-17A+ CD8+ T-cells including CTSL, HOPX, MCAM and GPR65 (encoding cathepsin L, homeodomain-only protein homeobox, melanoma cell adhesion molecule and G protein-coupled receptor 65, respectively) (50-54); as was CTLA4 (encoding cytotoxic T-lymphocyte associated protein 4) which has been described as a regulator of murine Tc17 cell differentiation and stability (55).
We were unable to conclude on the expression of IL17B, IL17C, IL17D or IL17E transcript in IL-17A+ CD8+ T-cells as these were filtered from downstream statistical assessments due to normalised expression lower than that of CD4. We also observed a degree of molecular similarity in both sorted subsets with similar mRNA expression of surface receptors (IL6R, IL18R1, IL21R), transcription factors (TBX21, EOMES, STAT3, IRF4, BCL11B) and effector molecules (IFNG, TNF). We confirmed by flow cytometry that in vitro-induced IL-17A+ CD8+ T-cells predominantly co-expressed the hallmark type-17 surface markers CCR6 and CD161.
Human in-vitro generated IL-17A+CD8+ T-cells 19 A small proportion of IL-17A-CD8+ T-cells also expressed CCR6 and CD161 (also at lower transcript levels than IL-17A+ counterparts), which is consistent with literature highlighting that these surface markers are not exclusively restricted to human IL-17 expressing T-cells (38,56). This finding reiterates the need for identification of additional lineage specific surface markers to better facilitate identification of Tc17 cells and type-17 cells more broadly. Taken together, these analyses validate the cytokine secretion assay approach for specifically purifying type-17 cells that secrete bioactive IL-17A.
The leading transcriptional parameter defining segregation of IL-17A+ from IL-17A-CD8+ Tcells was identified as PLZF, which is commonly associated with directing type-17 effector function in unconventional rather than conventional T-cells. With further immunoprofiling, our analysis indeed revealed that in vitro polarisation did not only induce IL-17A and IL-17F expression in conventional CD8+ T cells, but also in Vα7.2+/MR1-tetramer pos MAIT cells. This finding is in line with reports evidencing that IL-17A and IL-17F production by MAIT cells requires cooperative TCR signalling with specific cytokine signals, such as IL-1β and IL-23 (37,49), IL-7 (35), or IL-12 with IL-18 (31,57). Interestingly, the largest fraction of IL-17A/F expressing cells was contained within the MAIT population. One explanation could be that there was preferential expansion of IL-17A/F-producing MAIT cells upon in vitro culture, however our preliminary data did not reveal substantive differences in proliferative capacity between IL-17A+ CD8+ MAIT and conventional T-cells.
It was interesting to note that our PBMC polarising conditions led to a relative greater fold increase in IL-17F than IL-17A (7-fold versus 3-fold). Additionally, both conventional and MAIT IL-17A+ CD8+ T cells comprised IL-17A or IL-17F single and double positive cells. We have previously demonstrated that CD28 costimulation differentially regulates IL-17F versus IL-17A expression in CD4+ T-cells (5). Furthermore, a previous study showed a 10-fold higher production of IL-17F compared with IL-17A upon anti-CD3/CD28 stimulation of CD8+ T-cells (23), and IL-17F was found to be the dominant isoform when MAIT cells were stimulated in the presence of IL-12 and IL-18 (57). Collectively, these findings suggest differential regulation of IL-17A and IL-17F production in human T-cells. Mechanistically, they showed that co-transfer of Th17 cells with Tc17 cells mediated robust and long-lived anti-tumour immunity, consistent with previous publications which showed that Th17 cells can augment the activation of CD8+ T-cells (59,60). Hinrichs et al. also reported Human in-vitro generated IL-17A+CD8+ T-cells 21 that type 17-polarised CD8+ T-cells mediated enhanced anti-tumour immunity and demonstrated greater persistence than non-polarised CD8+ T-cells (61). An in-depth study by Gartlan et al., using fate mapping reporter mice also showed that mouse Tc17 cells differentiate during GVHD culminating in a highly plastic, hyperinflammatory, poorly cytolytic effector population, which they termed "inflammatory iTc17" (62). Targeted depletion of these inflammatory iTc17 cells resulted in protection from lethal GVHD.
These in vivo data together with our in vitro data strongly suggests that Tc17 cells are biologically relevant contributors to inflammation in diseases where an enrichment of these cells is found. In addition, our in vitro induction approach has important potential to improve mechanistic understanding of how Tc17 cells contribute to pathogenic, as well as homeostatic, immune responses, which could offer novel translational insights into therapeutic targeting of these cells.

Data Availability Statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Author Contributions
LT, BK and KS contributed to study conception and supervision. LT, US and EG designed experiments that were equally performed, the data acquired, analysed and interpreted by US and EG. LD and SL aided with CSA-FACS sorting experiments and AC kindly provided some synovial tissue fibroblast lines for functional assessments. BK coordinated PsA patient recruitment and sample collection during clinic at the Rheumatology Department, Guy's hospital. US wrote an initial draft of the manuscript, which was re-worked and finalized by EG with editing by LT. All authors except AC, who sadly passed away during completion of the project, were involved in critically revising the article and approved the final submitted version.

Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.