key: cord-0278324-1j5uiqzi
authors: Argyriou, A.; Wadsworth, M. H.; Lendvai, A.; Christensen, S. M.; Hensvold, A.; Gerstner, C.; Kravarik, K.; Winkler, A.; Malmstrom, V.; Chemin, K.
title: Single cell sequencing reveals expanded cytotoxic CD4+ T cells and two states of peripheral helper T cells in synovial fluid of ACPA+ RA patients
date: 2021-06-02
journal: nan
DOI: 10.1101/2021.05.28.21255902
sha: 0aecbf1683837ec00936e1a2d3dbe0ed360a3a62
doc_id: 278324
cord_uid: 1j5uiqzi

Rheumatoid arthritis is an autoimmune disease affecting the synovial joints where different subsets of CD4+ T cells are suspected to play a pathogenic role. So far, our understanding of the contribution of cytotoxic CD4+ T cells is incomplete, particularly in the context of the recently described peripheral helper T-cell subset (TPH). Here, using single cell sequencing and multi-parameter flow cytometry, we show that cytotoxic CD4+ T cells are enriched in synovial fluid of anti-citrullinated peptides antibody (ACPA)-positive RA patients. We identify two distinct TPH states differentially characterized by the expression of CXCL13 and PRDM1, respectively. Our data reveal that the adhesion G-Protein Coupled Receptor 56 (GPR56), a marker of circulating cytotoxic cells, delineates the synovial TPH CD4+ T-cell subset. At the site of inflammation, GPR56+CD4+ T cells expressed the tissue-resident memory markers LAG-3, CXCR6 and CD69. Further, TCR clonality analysis revealed that most expanded clones in SF are contained within the cytotoxic and the CXCL13+ TPH CD4+ T-cell populations. Finally, the detection of common TCRs between the two TPH and cytotoxic CD4+ T-cell clusters suggest a shared differentiation. Our study provides comprehensive immunoprofiling of the heterogenous T-cell subsets at the site of inflammation in ACPA+ RA and suggests GPR56 as a therapeutic target to modulate TPH cells and cytotoxic CD4+ T cell function

CD4 + T cells with cytotoxic functions (CD4 + CTL) have gained attention in recent years and accumulating evidence supports their importance in the defense against human viral infections such as CMV 1 , EBV 2 , dengue 3 , HIV 4,5 and SARS-CoV-2 6 . They share common functional characteristics with NK and CD8 + T cells, including the expression of cytolytic proteins PRF1 (perforin-1), GZMB (granzyme B) 7 and NKG7 (Natural Killer Cell Granule Protein 7) 5 . CD4 + CTL can originate from Th1 cells 8 but can also differentiate from a distinct precursor expressing the class I-restricted T cell-associated molecule (CRTAM) 9 . CD4 + CTL differentiation processes are still unclear but antigen dose 10 , IL-2 11 and type 1 IFN 12 signaling are suspected to contribute. 4-1BB (CD137) co-receptor triggering induces CD4 + CTL differentiation through the induction of the transcription factor Eomesodermin (Eomes) 13 and CD4 + CTL cytotoxic capacity is increased after IL-15 exposure 14 . Eomes cooperates with Runx3 to induce the transcription of Prf1 in mouse CD8 + T cells 15 . Together with Blimp-1, the transcription factor Hobit sustains the transcription of Gzmb 16 . Hobit and Blimp are also involved in the maintenance of tissue resident memory T cells (TRM) 17 . Both Eomes 18, 19 and Hobit 20 are expressed in circulating human cytotoxic CTL, including CD4 + CTL, suggesting that they convey similar functions in a human setting. GPR56 is an adhesion G-protein coupled receptor encoded by the ADGRG1 gene expressed on human circulating NK, CD8 + and CD4 + CTL 21 . Many of the molecular regulators of CD4 + CTL have been identified in mouse models or in circulating human blood. However, CD4 + CTL have not been investigated in great detail at the site of inflammation in human inflammatory disorders, especially in the context of TRM. In this environment, it is still unclear from which precursors they differentiate from.

Rheumatoid arthritis (RA), which affects around 1% of the global population, is an autoimmune disease characterized by articular bone erosion leading to physical disability, pain and decreased quality of life. Bone damage is particularly exacerbated in the subset of RA patients who present with antibodies against citrullinated proteins (ACPA) 22, 23 . Although the use of biologics has revolutionized the treatment of RA, 30-40% of patients fail to respond to treatment stressing the need for novel therapeutic targets 24 . The role of CD4 + T cells in RA pathogenesis is evidenced by the HLA-DRB1 association with ACPA+ RA 25 . Expanded patient-specific T-cell clones have been identified in the synovium and synovial fluid (SF) of RA patients using bulk TCRb repertoire analysis 26 . Over the years, several CD4 + T cell subsets, including Th1 and Th17 cells have been described in the synovial joints of RA patients 27 . Recently, PD-1 high HLA-Class II + peripheral helper T cells (TPH) were identified in the synovial fluid and tissue of seropositive RA patients 28, 29 where they are proposed to facilitate B-cell recruitment and activation through CXCL13 and IL-21 production; however, there is still no satisfying marker that defines the TPH subset. It is also unclear how this population is generated in the synovial joint. We have recently shown that CD4 + T cells with cytotoxic features are observed in SF of RA patients carrying the PTPN22 1858T risk allele variant 30 . This finding, as well as previous studies reporting expanded CD4 + CD28 null cytotoxic T cells in the blood of RA patients 31, 32 , suggests that CD4 + CTL might contribute to RA. However, a transcriptomic and clonality analysis of CD4 + CTL, in the context of previously described CD4 + T cell subsets, is still missing. The identification of T-cell effector functions is deemed important since it can lead to new therapeutic strategies.

Here, we performed single cell sequencing in combination with 5´ TCRab sequencing on peripheral blood (PB) and synovial fluid (SF) from ACPA+ RA patients to investigate CD4 + CTL. We demonstrate that clonal expansions are prominent amongst CD4 + CTL in SF of ACPA+ RA. We also describe two TPH CD4 + T-cell clusters differentially characterized by the expression of PRDM1 and CXCL13, suggestive of two different stages of TPH differentiation. Finally, we find that TPH cells express the receptor GPR56 and the TRM markers LAG-3, CXCR6 and CD69 implicating that these T cells are maintained in the ACPA+ RA synovial joint. Our data provide a comprehensive immunoprofiling map of pathogenic CD4 + T cells at the site of inflammation in ACPA+ RA.

CD4 + T cells expressing cytotoxic effector molecules are enriched in synovial fluid of ACPA+ RA patients.

SFMC from ACPA-(n=9) and ACPA+ (n=12) RA patients were screened for the expression of cytotoxic effector molecules and transcription factors in both CD4 + and CD8 + T cells by flow cytometry (Fig. 1, supplementary Fig. 1 ). ACPA+ SFMC presented with a significantly increased frequency of GZMB + PRF1 + (p=0.0072), Hobit + (p=0.0018), NKG7 high (p=0.0036) and GPR56 + (p=0.0007) CD4 + cells ( Fig. 1ab ). No significant difference was observed for the expression of GZMA and Eomes.

Amongst CD8 + T cells, Hobit expression was increased in ACPA+ SFMC (p=0.0170) whereas a similar trend was observed for the expression of GZMA, GZMB and NKG7 without reaching statistical significance (supplementary Fig. 2 ). No significant difference was observed for the expression of PRF1 and the frequency of GPR56 in All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint CD8 + T cells in ACPA+ versus ACPA-RA. The frequency of GZMB + PRF1 + CD4 + T cells in SF positively correlated with CCP (cyclic citrullinated peptides) antibody titers (r=0.5701, p=0.0070) (Fig. 1c) . In ACPA+ synovial fluid, GPR56 identified a distinct CD4 + T cell population with a frequency that also correlated with CCP antibody titers (r=0.7871, p<0.0001) (Fig. 1d) .

To further characterize cytotoxic CD4 + T cells in SF of ACPA+ RA, we performed 10X single cell sequencing in combination with 5´ TCRab sequencing on purified CD4 + T cells from paired SFMC and PBMC from seven ACPA+ RA patients (Fig. 2a, supplementary table 1). After quality control (supplementary Fig. 3a ), we obtained transcriptomic data from 66,360 CD4 + T cells in SFMC and 69,373 CD4 + T cells in PBMC. Unsupervised analysis of the transcriptomic data from all CD4 + T cells (n=135,733) from PB and SF using the Seurat package 33 generated a total of 16 clusters annotated based on known and high scoring genes ( Fig. 2b- 28, 36, 37 , two TPH cell clusters were subclassified as TPH CXCL13 + (cluster 3) and TPH PRDM1 + (cluster 12). Cluster 4 defines cytotoxic CD4 + T cells expressing cytotoxicity related genes including NKG7, GZMH and PRF1. This population was present in SF of GPR56 expression defines the subset of TPH cells (Fig. 3b -c) as illustrated by the increased expression of PD-1 (p<0.0001) and MHC-II (p<0.0001) in GPR56 + CD4 + T cells when compared to GPR56 -CD4 + T cells (Fig. 3d) . GPR56 was not detected on cytotoxic CD4 + T cells in SF ( Supplementary Fig. 6c -d) as opposed to ADGRG1 (Fig   3a, 6a-b) , highlighting the difference between transcriptomic and protein expression.

These data indicate that, while GPR56 is a marker for cytotoxic CD4 + T cells in PB, it also delineates the subsets of PD-1 high MHC-Class-II + TPH cells in ACPA+ RA SF.

Besides being an inhibitory receptor, PD-1 is also included in the signature associated to T-cell tissue residency 38 and implicated in follicular helper T cells maintenance in germinal centers 39 . We therefore assessed the expression of known tissue resident memory (TRM) T-cell markers: LAG3, ITGA1 (CD49a), CD69, CXCR6 but also CX3CR1 which is downregulated on TRM T cells 38 . In SF, we found that LAG-3 was mainly expressed on TPH cells (cluster 3 and 12) and proliferating T cells (clusters 13 and 14) (Fig. 4a) . A similar tendency was also observed for CXCR6. These genes were not expressed on cytotoxic CD4 + T cells, consistent with their recirculation in the blood.

Since we previously observed that GPR56 delineates the subset of TPH cells, we assessed the expression of tissue-resident memory markers in the context of GPR56 by flow cytometry (Supplementary Fig.7 ). LAG-3 (p<0.0001) and CXCR6 (p=0.0052) frequencies as well as CD69 (p=0.0002) expression was enriched on synovial GPR56 + CD4 + T cells (Fig. 4b-c ). In contrast, CD49 frequency was higher on GPR56-negative CD4 + T cells (p<0.0001). In general, LAG-3 frequency was also significantly increased on CD4 + T cells in ACPA+ SF as compared to ACPA-SF (p=0.0005) (Fig. 4d ) and correlated with CCP titers (r=0.8088, p<0.0001) (Fig. 4e) . In CD8 + T cells, CD49a was the only marker for which an increased frequency was observed in ACPA+ SF All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table 3 and 4). Further, we examined overlapping CDR3 sequences between SF CD4 + T-cell subsets that would be indicative of a possible shared differentiation pathway (Fig. 5c ). In SF, the CXCL13 + TPH cluster shared TCR sequences with the PRDM1 + TPH, Humanin + CD4 + and proliferating CD4 + T-cell clusters. Common CDR3s were also identified between cytotoxic CD4 + T cells and CXCL13 + TPH cells. Of note, in SF, Tregs shared only few CDR3 sequences with other CD4 + T-cell subsets, suggesting that they could be thymus-derived Tregs 40 and not induced at the site of inflammation from conventional CD4 + T cells. We then further All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint focused our TCR analysis on the clusters of interest (cytotoxic CD4 + T cells, PRDM1 + TPH, CXCL13 + TPH and proliferating CD4 + T cells 1 and 2) in PB and SF (Fig. 5d) . We observed that some CDR3 sequences were shared between cytotoxic CD4 + T cells in PB and SF. Similarly, CXCL13 + TPH cells had common CDR3s in PB and SF. Overall, these data show that, in ACPA+ synovial fluid, expanded clones are mainly found within the cytotoxic and the CXCL13 + TPH CD4 + T-cell subsets and that shared CDR3 sequences are identified between clusters in particular between the two TPH clusters and between CXCL13 + TPH and cytotoxic CD4 + T cells.

Cytotoxic CD4 + CD28 null T cells are expanded in the blood of patients with RA 31 but their presence and clonality at the site of inflammation in ACPA+ RA was still not deeply characterized. Leveraging single cell technology, we identified the presence of cytotoxic CD4 + T cells in ACPA+ RA SF that are clonally expanded in synovial fluid and, for some patients, in their corresponding peripheral blood. We further define the TPH CD4+ T-cell subset at the single cell level showing that it is composed of two independent clusters with different CXCL13 and PRDM1 (Blimp-1) expression levels.

So far, the TPH CD4 + T-cell subset was only defined by high expression of PD-1 and MHC-Class-II and more accurate markers were still lacking. We show that GPR56 clearly delineates the TPH subset and associates with the expression of tissue resident memory markers LAG-3, CXCR6 and CD69. Together, these data provide a refined characterization of cytotoxic and TPH CD4 + T cells, two pathogenic CD4 + T-cell subsets in ACPA+ RA (Fig. 6 ).

In synovial tissue of ACPA+ RA patients, TPH CD4 + cells produce CXCL13 28,36,41 , localize with B cells and have the ability to induce in vitro plasma cell differentiation All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint through IL-21 production 28 . TPH cells are defined by the high expression of PD-1 and MHC-Class II, the cytokines CXCL13, IL-21 and IL-10, and the transcription factors Blimp-1, TOX, MAF and SOX4 28, 36, 37 . TPH CD4 + T-cell differentiation is induced by TGFb in IL-2 neutralizing conditions 37 . Single cell sequencing identified two distinct clusters of TPH cells: PRDM1 + TPH cells and CXCL13+ TPH cells. We postulate that these two clusters correspond to two states of differentiation: PRDM1 + TPH CD4 + T cells could be the precursors of the more differentiated cytokine producing CXCL13 + TPH CD4 + T cells. Indeed, Blimp-1 encoded by PRDM1 negatively regulates the expression of CXCL13 37 . In line with this hypothesis, common TCR sequences observed between PRDM1 + TPH cells and CXCL13 + TPH cells clusters suggest of a common origin between these two subsets.

The impact of CD4 + CTL in the synovial joint is less well characterized. Expansions of CD4 + CTL are frequently observed in chronic viral infections and CD4 + CTL are capable of directly killing EBV-infected B cells in a MHC-class II dependent manner 2 .

Whether CD4 + CTL are protective or pathogenic in viral infections is still a matter of debate and their role may vary depending on the type of infections 5 . Recently, single cell sequencing data has also identified infiltrating CD4 + CTL in liver cancer 42 and in bladder cancer where they provided anti-tumor killing in the context of MHC-Class II 43 . Interestingly, a CD4 + CTL signature in bladder tumors predicted clinical response to anti-PD-L1 43 . Furthermore, expansions of cytotoxic CD28 null CD4 + T cells have been repeatedly reported in the blood of patients with rheumatic diseases such as RA 31 , myositis 44 and vasculitis 45 as well as in cardiovascular diseases 46, 47 . PRF1 + CD4 + T cells were also identified by immunofluorescence in synovial tissue of RA patients 48 .

Our data shows that cytotoxic CD4 + T cells are enriched in SF of ACPA+ RA patients and correlates with CCP titers showing that cytotoxic CD4 + T cells are mainly All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. implicated in the subset of autoantibody-positive RA. Purified perforin-1 49 as well as supernatants from IL-15 stimulated synovial CD8 + T cells 50 induces histone citrullination in neutrophils suggesting that cytotoxic effector molecules and citrullination are tightly linked in RA. In our dataset, cytotoxic CD8 + T cells also tended to be more cytotoxic in ACPA+ RA and could contribute to a perforin-dependent citrullination mechanism. The presence of CD4 + CTL only in SF of ACPA+ RA suggest that they have a pathogenic function, but further studies are warranted to understand the functional consequences of their interactions with antigen presenting cells in the synovial joint.

The differentiation processes that lead to the differentiation of cytotoxic CD4 + T cells are not completely elucidated. IL-2, IL-15 and type I IFNs, which have been implicated in the differentiation of cytotoxic CD4 + T cells, are also expressed in the synovial joint possibly contributing to their development. We show that the transcription factor Hobit, which has been shown to regulate Gzmb expression in mouse 16 , was expressed in GZMB/PRF-1-expressing CD4 + T cells in both PB and SF suggesting that Hobit also drive CD4 + CTL differentiation in ACPA+ RA. We have previously shown that PRF-1-expressing Eomes + CD4 + T cells are increased in SF of PTPN22 risk carriers 30 . In dengue viral infections, an increased frequency of CD4 + CTL was observed in HLA-DRB1*0401+ individuals 51 . Hence, RA-associated genetic polymorphisms might contribute to increase the differentiation of CD4 + CTL through mechanisms that remain to be investigated.

Expanded TCR clones have been identified in synovial tissues and fluid from RA patients using bulk TCRβ sequencing 26 . In blood, expanded CD4 + CD8 null cells present with a bias in their TCRVβ usage 52, 53 . In our dataset, the most expanded T-cell clones All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint identified in SF of ACPA+ RA presented with cytotoxic or CXCL13 + TPH signatures.

The differentiation towards the cytotoxic CD4 + T cell lineage has been proposed to be linked to repeated antigen stimulation as illustrated in the case of chronic viral infections such as CMV 10 . Expanded cytotoxic CD4 + T cell clones in SF might therefore be the final stage of differentiation of autoreactive CD4 + T cells which have been exposed to several rounds of antigen stimulation. Several studies have shown that CMV reactivity is found in the subset of CD4 + CD8 null cell subset but so far, their autoreactive properties have not been investigated in the context of citrulline immunity.

In our data set, some TCR sequences were shared between TPH cells and cytotoxic CD4 + T cells and we can speculate that TPH cells, after repeated antigenic stimulation led to cytotoxic CD4 + T cell expanded clones.

GPR56 is an adhesion G-protein coupled receptor encoded by the ADGRG1 gene implicated in numerous migration/adhesion processes including neuron migration 54 and tumor growth inhibition 55 . In NK cells, GPR56 is an inhibitory receptor binding to the tetraspanin CD81 56 . We confirmed a previous report showing that GPR56 is also a marker of circulating cytotoxic CD4 + cells 21 . At the site of inflammation, however, we discovered that GPR56 was highly expressed on TPH cells in ACPA+ RA patients and could be used as a new marker to identify TPH CD4 + T cells. In SF CD4 + T cells, GPR56 expression correlated with the expression of PD-1, LAG-3, CXCR6 and CD69 expressed on tissue-resident memory T cells 38 . Similarly, in multiple sclerosis, CD8 + T cells from active lesions exhibit a tissue resident memory phenotype (CD69, CD103, PD-1, CD49a) associated with an intermediate expression of GPR56 without granzyme B expression 57 . In mouse, Blimp-1 and Hobit mediates the transcriptional program associated with T-cell tissue residency 17 . In ACPA+ SF, Blimp-1 was expressed on TPH CD4 + cells presenting with tissue resident memory markers whereas Hobit was only All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint expressed on cytotoxic CD4 + T cells suggesting that, Blimp-1 could drive tissue residency of TPH CD4 + T cells in the synovial joint. In human NK cells, Hobit triggers ADGRG1 expression 56 but, in SF, Hobit was not expressed in ADGRG1+ CD4 + T cells suggesting that a different transcription factor could regulate GPR56. GPR56 might be implicated in the migration and/or maintenance of TPH CD4 + T cells in the synovial joints where its ligand still remains to be determined.

In summary, we identify clonally expanded cytotoxic CD4 + T cells and two clusters of TPH CD4 + T cells in ACPA+ SF. We show that GPR56 is a marker that should be used to identify TPH CD4 + cells. Future studies aiming at understanding the function of GPR56 in TPH CD4 + and cytotoxic CD4 + T cells will contribute to evaluate the feasibility of targeting this receptor in ACPA+ RA. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint metrics, the libraries were sequenced on Illumina NextSeq500/550 high output 75-cycle v2.5 kits with the following read structure: read1: 26, read2: 58, index 1: 8. Libraries were sequenced to obtain a read depth greater than 20,000 reads/cell for the geneexpression (GEX) libraries and greater than 5,000 reads/cell for the V(D)J-enriched T Cell libraries.

Following sequencing, the pooled GEX libraries were demultiplexed using bc2fastq (v2.17) with modified parameters (--minimum-trimmed-read-length=10 --mask-shortadapter-reads=10 --ignore-missing-positions --ignore-missing-controls) to generate FASTQ files for each patient. FASTQ files were then uploaded to Terra (www.app.terr.bio) where the raw sequencing data were mapped and quantified using STAR within the 10X Genomics Inc software package CellRanger on Cumulus Aligned matrices were first filtered to remove low-quality barcodes, keeping only those with greater than 200 UMIs, less than 4,000 UMIs, and less than 20% mitochondrial reads. Cell doublets were then computationally removed using DoubletFinder 58 with a calculated doublet rate of 14% and default parameters.

Using the Seurat 59 (v4.0.0) package, filtered gene-by-cell matrices were merged and then processed using a standard unsupervised workflow (i.e., normalization, scaling, dimensionality reduction, batch correction, cell clustering, and differential gene expression analysis). First, the merged matrix was normalized and log-transformed All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint (scaling factor = 10,000). The top 3000 highly variable genes were identified and used for scaling the data. During scaling, two sources of unwanted variation, cumulative UMI capture and percent mitochondrial reads, were regressed out. Dimensionality reduction was then performed with principal component analysis (PCA) over the top 3000 variable genes and, using the JackStraw function in Seurat, we selected 53 statistically significant principal components (PCs, p < 0.01). Using these significant PCs, we corrected for compartment and patient-specific batch effects with the harmony 60 package, and then proceeded with constructing the nearest neighbor graph and Uniform Manifold Approximation and Projection (UMAP) plots. Cells were clustered using Louvain clustering and the package clustree 61 was used to generate a clustering tree and identify which resolution achieved stability. Differential gene expression analysis was then performed using the FindAllMarkers Seurat function with test.use set to 'MAST' 62 . We also used Nebulsosa 63 to visualize the joint density approximations of known T-cell phenotypic markers. We attempted to deconvolute the multiplexed PBMC and SFMC cell populations using the cell hashing antibodies, however, due to sub-optimal tagging we were unable to separate the cell populations.

To address this issue, we generated a compartment-specific reference using PBMC and SFMC samples from two more volunteers and then reference-mapped our mixed samples to assign compartment membership 59 . Only barcodes with a predicted id score greater than 0.5 were kept and used for downstream analyses. Finally, non-T cells were filtered out by removing barcodes with zero gene expression for CD3 genes. Taking both the results of the differential expression analysis and joint density approximations of known phenotypic markers, we identified 16 unique T cell subsets: 1. Activated CD4 + T Cells (CD38 + HLA-DRA + ); 2. CD4+ T Cells (IL7R + ); 3. CXCL13 + Tph (CXCL13 + CD3D+); 4. Cytotoxic CD4 + T Cells (GNLY + GMZB + PRF1 + GZMA + All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint NKG7 + ); 5. Effector CD4+ T Cells (CCR6 + KLRB1 + CD27 + ); 6. EGR1 + CD4 + T Cells (EGR1 + ); 7. Humanin + CD4 + T Cells (MTRNR2L12 + MTRNR2L8 + ); 8. IL3RA + CD4 + T Cells (IL3RA + ); 9. LAMP3 + CD4 + T Cells (LAMP3 + ); 10. Naïve and Central Memory (CM) T Cells (IL7R + , CCR7 + , SELL + ); 11. Mitochondrial High CD4 + T cells (MT-CO1 + MT-ATP8 + MT-ND4 + ); 12. PRDM1 + Tph (PRDM1 + ADGRG1 + ); 13.

Proliferating CD4 + T Cells 1 (STMN1 + ); 14. Proliferating CD4 + T Cells 2 (KI67 + ); 15 .Tregs (TIGIT + FOXP3 + IL2RA + ); 16. TTN + CD4 + T cells (TTN + ).

Following sequencing, the V(D)J pooled libraries were demultiplexed using bc2fastq (v2.17.1.14) to generate FASTQ files for each patient, uploaded to terra, and mapped and quantified using the 10X Genomics Inc software package CellRanger on Cumulus (https://cumulus.readthedocs.io/en/latest/cellranger.html, snapshot 15, Cellranger 5.0.1, default parameters) to generate consensus annotation files for each patient. Using the scRepertoire 64 package, patient-specific consensus annotation files were consolidated into a list of TCR sequencing results and then integrated with the Seurat object using the combineExpression function. Clonotypes were called based on the CDR3 gene and clonotype expansion were assigned based on the absolute frequency.

Finally, the interconnectivity between specific cell types, both within and between tissue compartments, were visualized using chord diagrams. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. White dots indicate GPR56 -CD4 + T cells and black dots indicate GPR56 + CD4 + T cells.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint oversaw the analyses and wrote the paper. All authors read, edited and approved the paper.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 2, 2021. ; https://doi.org/10.1101/2021.05.28.21255902 doi: medRxiv preprint 

Acquisition of direct antiviral effector functions by CMVspecific CD4+ T lymphocytes with cellular maturation

Direct killing of Epstein-Barr virus (EBV)-infected B cells by CD4 T cells directed against the EBV lytic protein BHRF1

Dengue virus-specific human T cell clones. Serotype crossreactive proliferation, interferon gamma production, and cytotoxic activity

Characterization of CD4(+) CTLs ex vivo

Cytotoxic CD4 T Cells-Friend or Foe during Viral Infection? Front Immunol 8

Imbalance of Regulatory and Cytotoxic SARS-CoV-2-Reactive CD4(+) T Cells in COVID-19

Expanding roles for CD4(+) T cells in immunity to viruses

CD134 plus CD137 dual costimulation induces Eomesodermin in CD4 T cells to program cytotoxic Th1 differentiation

CRTAM determines the CD4+ cytotoxic T lymphocyte lineage

Infection with cytomegalovirus but not herpes simplex virus induces the accumulation of late-differentiated CD4+ and CD8+ T-cells in humans

IL-2 and antigen dose differentially regulate perforin-and FasL-mediated cytolytic activity in antigen specific CD4+ T cells

Cytokine-dependent induction of CD4+ T cells with cytotoxic potential during influenza virus infection

Systemic 4-1BB activation induces a novel T cell phenotype driven by high expression of Eomesodermin

Transcriptional reprogramming of mature CD4(+) helper T cells generates distinct MHC class II-restricted cytotoxic T lymphocytes

Runx3 and T-box proteins cooperate to establish the transcriptional program of effector CTLs

Blimp-1 induces and Hobit maintains the cytotoxic mediator granzyme B in CD8 T cells

Hobit and Blimp1 instruct a universal transcriptional program of tissue residency in lymphocytes

Characterization of T-bet and Eomes in Peripheral Human Immune Cells. Front Immunol

Eomesodermin-expressing T-helper cells are essential for chronic neuroinflammation

The Transcription Factor Hobit Identifies Human Cytotoxic CD4(+) T Cells

Specific expression of GPR56 by human cytotoxic lymphocytes

The prognostic value of anti-cyclic citrullinated peptide antibody in patients with recent-onset rheumatoid arthritis

The immunopathogenesis of seropositive rheumatoid arthritis: from triggering to targeting

A genome-wide association study suggests contrasting associations in ACPA-positive versus ACPA-negative rheumatoid arthritis

In Rheumatoid Arthritis, Synovitis at Different Inflammatory Sites Is Dominated by Shared but Patient-Specific T Cell Clones

Effector Functions of CD4+ T Cells at the Site of Local Autoimmune Inflammation-Lessons From Rheumatoid Arthritis

Pathologically expanded peripheral T helper cell subset drives B cells in rheumatoid arthritis

Defining inflammatory cell states in rheumatoid arthritis joint synovial tissues by integrating single-cell transcriptomics and mass cytometry

EOMES-positive CD4(+) T cells are increased in PTPN22 (1858T) risk allele carriers

CD4+ CD7-CD28-T cells are expanded in rheumatoid arthritis and are characterized by autoreactivity

CD28nullCD4+ T cells--characterization of an effector memory T-cell population in patients with rheumatoid arthritis

Spatial reconstruction of single-cell gene expression data

Single-cell gene expression reveals a landscape of regulatory T cell phenotypes shaped by the TCR

Single-cell transcriptomics of human T cells reveals tissue and activation signatures in health and disease

A distinct human CD4+ T cell subset that secretes CXCL13 in rheumatoid synovium

Human Sox4 facilitates the development of CXCL13-producing helper T cells in inflammatory environments

Human Tissue-Resident Memory T Cells Are Defined by Core Transcriptional and Functional Signatures in Lymphoid and Mucosal Sites

PD-1 Controls Follicular T Helper Cell Positioning and Function

Natural versus adaptive regulatory T cells

Mature antigen-experienced T helper cells synthesize and secrete the B cell chemoattractant CXCL13 in the inflammatory environment of the rheumatoid joint

Landscape of Infiltrating T Cells in Liver Cancer Revealed by Single-Cell Sequencing

Intratumoral CD4(+) T Cells Mediate Anti-tumor Cytotoxicity in Human Bladder Cancer

T cell infiltrates in the muscles of patients with dermatomyositis and polymyositis are dominated by CD28null T cells

Costimulatory molecules in Wegener's granulomatosis (WG): lack of expression of CD28 and preferential up-regulation of its ligands B7-1 (CD80) and B7-2 (CD86) on T cells

TRAIL-expressing T cells induce apoptosis of vascular smooth muscle cells in the atherosclerotic plaque

The life (and death) of CD4+ CD28(null) T cells in inflammatory diseases

Perforin and granzyme A expression identifying cytolytic lymphocytes in rheumatoid arthritis

Immune-mediated pore-forming pathways induce cellular hypercitrullination and generate citrullinated autoantigens in rheumatoid arthritis

Synovial fluid CD69(+)CD8(+) T cells with tissue-resident phenotype mediate perforin-dependent citrullination in rheumatoid arthritis

Dengue virus infection elicits highly polarized CX3CR1+ cytotoxic CD4+ T cells associated with protective immunity

The repertoire of CD4+ CD28-T cells in rheumatoid arthritis

Clonally expanded CD4+CD28null T cells in rheumatoid arthritis use distinct combinations of T cell receptor BV and BJ elements

G protein-coupled receptor-dependent development of human frontal cortex

GPR56 inhibits melanoma growth by internalizing and degrading its ligand TG2

The Adhesion G Protein-Coupled Receptor GPR56/ADGRG1 Is an Inhibitory Receptor on Human NK Cells

Tissue-resident memory T cells invade the brain parenchyma in multiple sclerosis white matter lesions

Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors

Fast, sensitive and accurate integration of single-cell data with Harmony

Clustering trees: a visualization for evaluating clusterings at multiple resolutions

MAST: Model-based Analysis of Single Cell Transcriptomics

Nebulosa recovers single cell gene expression signals by kernel density estimation

scRepertoire: An R-based toolkit for single-cell immune receptor analysis

LAMP3+ CD4+ TCells 10. Naive and CM CD4+ T Cells