bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
CD4+ follicular helper-like T cells are key players in anti-tumor immunity
Singh D1,2,7, Ganesan AP1,3,7, Panwar B1, Eschweiler S1, Hanley CJ2, Madrigal A1, Ramírez-Suástegui C1,
Wang A1, Clarke J1,2, Wood O2, Garrido-Martin EM4, Chee SJ2,5, Seumois G1, Belanger S1, Alzetani A5, Woo
E5, Friedmann PS4, Crotty S1, Thomas GJ2, Sanchez-Elsner T4, Ay F1,8, Ottensmeier CH2,8, Vijayanand
P1,4,5,6,8,9
1La Jolla Institute for Allergy and Immunology, La Jolla, California, USA.
2Cancer Sciences Unit, Faculty of Medicine, University of Southampton, Southampton, UK.
3 Division of Pediatric Hematology Oncology, Rady Children’s Hospital, University of California San Diego,
San Diego, California, USA.
4Clinical and Experimental Sciences, National Institute for Health Research Southampton Respiratory
Biomedical Research Unit, University of Southampton, Faculty of Medicine, Southampton, UK.
5Southampton University Hospitals NHS Foundation Trust, Southampton, UK.
6Department of Medicine, University of California San Diego, San Diego, California, USA.
7These authors contributed equally to this work
8Senior author
9Lead Contact
Correspondence should be addressed to P.V. ([email protected]), F.A. ([email protected]), C.H.O.
The authors have declared that no conflict of interest exists.
1 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
ABSTRACT
To determine the nature of CD4+ T cells that provide ‘help’ for generating robust anti-tumor CD8+ cytotoxic T cell (CTL) responses, we profiled the transcriptomes of patient-matched CD4+ and CD8+ T cells present in the tumor micro-environment (TME) and analyzed them jointly using integrated weighted gene correlation
+ network analysis. We found the follicular helper T cell (TFH) program in CD4 T cells was strongly associated
+ with proliferation and tissue-residency in CD8 CTLs. Single-cell analysis demonstrated the presence of TFH- like cells and features linked to cytotoxic function and their provision of CD8+ T cell ‘help’. Tumor-infiltrating
TFH-like cells expressed PD-1 and were enriched in tumors following checkpoint blockade, suggesting that they may respond to anti-PD-1 therapy. Adoptive transfer or induction of TFH cells in mouse models resulted
+ in augmented CD8 CTL responses and impairment of tumor growth, indicating an important role of TFH-like
CD4+ T cells in anti-tumor immunity.
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
INTRODUCTION
CD8+ cytotoxic T cells (CTLs) are vital components of anti-tumor immunity (1). A distinct subset of
+ CD8 CTLs, tissue-resident memory (TRM) T cells, have recently emerged as critical players in mediating robust anti-tumor immune responses (2-6). Cancer immunotherapies designed to potentiate CD8+ CTL responses have led to remarkable clinical success, albeit in a small proportion of patients (7, 8).
Therapeutic failure is due, at least in part, to an incomplete understanding of the signals and cell types in the tumor microenvironment (TME) and how they modulate effector and resident-memory CTL responses.
+ CD4 T helper cells (TH) are central orchestrators of an efficient immune response. During immunization or infections, they are necessary for robust primary CD8+ CTL effector responses and
+ memory transition (9-12). CD4 TH cells work through a multitude of mechanisms involving dendritic cell
(DC) licensing and activation to initiate CD8+ CTL responses (13, 14) and enhance CTL proliferation (15).
They mediate CD8+ CTL recruitment to cognate DC (16) or pathological tissue (17), and by regulation of
CD8+ T cell TRAIL expression, they promote CTL secondary expansion on restimulation following vaccine
+ or viral infections (18). However, within the TME, it is unknown what type of CD4 TH cells or CD4 helper-
+ derived signals are essential to generate robust CD8 CTL effector and TRM anti-tumor immune responses.
Single-cell sequencing studies in different tumor types, have revealed tumor cell (19) or stromal cell programs (20, 21) that regulate immune response (19) or metastases (21). The analysis of tumor-infiltrating immune cells has demonstrated ‘pre-exhausted’ and exhausted CD8+ T cell states and multiple CD4+ T cell subtypes (22-24). Previous reports on CD4+ T cells have focused on the evaluation of specific CD4+ T cell
+ + subsets such as regulatory T cells (Treg), CD4 TH1 and CD4 TH17 cells (22, 25-27). Whilst these studies provided valuable insights into tumor-infiltrating CD8+ T cell or CD4+ T cell subsets in isolation, the molecular characterization of the cross-talk between them has not been elucidated. In order to understand the interplay between CD4+ T cells and CD8+ CTLs within the TME and the resultant impact on anti-tumor immune responses, it is critical to undertake an integrated assessment of their transcriptional programs.
We have previously reported on the transcriptomic features of tumor-infiltrating CD8+ CTLs in a well- characterized cohort of patients with non-small cell lung cancer (NSCLC) (3). Here, we generated the transcriptional profiles of patient-matched, purified tumor-infiltrating CD4+ T cells from the same cohort of
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
patients to define the molecular interactions between CD4+ T cells and CD8+ CTLs in the TME. We performed integrated weighted gene correlation network analysis (iWGCNA) to fully characterize the molecular landscape of adaptive immune responses in the TME and the differences therein between tumors
+ with or without robust CD8 CTL effector and TRM responses. We found that a follicular helper T cell (TFH) program in CD4+ T cells was strongly associated with CTL proliferation and tissue-residence in the TME.
Single-cell transcriptomic analysis of tumor-infiltrating CD4+ T cells confirmed the presence of CXCL13- expressing TFH-like cells, which, despite expressing PDCD1, were enriched for transcripts linked to cell proliferation, cytotoxicity and provision of ‘help’ to CD8+ T cells, indicative of superior functionality. Using a
+ murine tumor model, we showed that adoptively transferred or induced TFH cells augmented CD8 CTL responses and impaired tumor growth in vivo. Thus, based on the molecular identity and functional
+ properties of tumor-infiltrating CD4 T cells, we show that the TFH subset is associated with robust anti-
+ tumor CD8 CTL and TRM responses.
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
RESULTS
Follicular program in CD4+ T cells is associated with CD8+ CTL proliferation, cytotoxicity and tissue residency in the TME
To characterize the molecular interplay between cells in the TME and to define the properties in
+ + CD4 T cells that are strongly associated with robust anti-tumor CD8 CTL effector and TRM responses across our cohort of patients with lung cancer, we developed a novel method, iWGCNA, for joint transcriptome analysis of co-expression patterns across different cell types (Figure 1A). Standard co- expression analysis (WGCNA (28)), is designed for studying one cell type at a time, and genes with similar expression patterns across a set of samples are grouped together into discrete clusters or modules. These modules can then be correlated with specific molecular traits to determine the group of genes whose expression is strongly associated with that trait(28). However, for a complex system, as found in the TME, where multiple cell types interact and co-regulate the trait of interest, it is important to analyze the cell transcriptomes in an integrated fashion. The goal of iWGCNA is to group transcript expression from different cell types in the TME of matched patients to form integrated gene modules that can reveal the molecular cross-talk between cell types and their relationship to specific traits.
Here, we performed iWGCNA by merging the transcriptomes from patient-matched CD4+ and CD8+
T cells present in the TME (n = 36) (3) and generated 29 gene network modules, each of which was composed of varying proportions of CD4+ T cell- and CD8+ T cell-transcripts (Figure 1A and B;
Supplementary Tables 1 and 2). To determine what properties in CD4+ T cells were associated with robust
CD8+ T cell responses in the TME, we correlated these gene modules with CD8+ T cell proliferation signature as a trait (Methods), as it represented a feature of robust anti-tumor T cell responses in the TME.
Module 7 (407 CD8+ T cell- and 178 CD4+ T cell-transcripts) exhibited the highest correlation with the CD8+
T cell proliferation signature (r = 0.915, adjusted P = 1.65E-13) (Figure 1B) and, as expected, nearly 25% of the CD8+ T cell-transcripts in Module 7 were cell cycle-related genes. Clustering analysis of the 407 CD8+
T cell-transcripts present in Module 7 identified a tightly correlated and co-expressed subset of transcripts (n
= 171), which, besides cell cycle genes, included several genes encoding products linked to effector and cytotoxic functions such as GZMB, CCL3, STAT1, FKBP1A, KIR2DL4 (Figure 1C and 1D; Supplementary
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Table 2). This highly co-expressed cluster of CD8+ T cell-transcripts also contained ITGAE (which encodes
+ for the α−subunit of the integrin molecule αEβ7), a well-established marker of lung CD8 TRM cells (29, 30), which we have recently shown to be a critical determinant of survival outcomes in lung cancer (Figure 1C and 1D; Supplementary Table 2) (3). Together, these findings indicated that cell proliferation, effector
+ functions and TRM features are highly correlated and interconnected processes in CD8 CTLs present in the
TME.
To assess the properties in CD4+ T cells that are associated with these functional features of CD8+
CTLs in the TME, we next analyzed the CD4+ T cell-transcripts (n = 178) present in Module 7. We observed a tightly correlated and co-expressed cluster of transcripts (n = 61), which included several genes linked to
+ follicular helper CD4 T cells (TFH), such as CXCL13, BATF, CD38, PDCD1 (31) (Figure 1C and 1D;
+ Supplementary Table 2). TFH cells are the principal CD4 T cell subpopulation that provides essential ‘help’ to B cells, promoting maturation of antibody affinity in germinal centers (GC) (32). Among the TFH-related transcripts in this cluster, BATF encodes for a transcription factor involved in TFH differentiation (33).
CXCL13 is a chemokine produced by human TFH cells, but not by their murine counterparts; CXCL13 has been shown to play an important role in the homing of B cells to follicles (34-36). The Module 7 cluster was also composed of CD4+ T cell-transcripts linked to cell proliferation (e.g., KI67, TOP2A, STMN1, CDK1) and
T cell activation, such as CD38 (37-39), TNFRSF9 (which encodes for 4-1BB (40, 41)), TNFRSF18 (which encodes for GITR (42)), and TNFRSF8 (which encodes for CD30 (43)) (Figure 1C; Supplementary Table
2). In an unbiased overlap analysis using hypergeometric test, we confirmed that among the 29 gene modules generated from the iWGCNA, module 7 exhibited the highest enrichment for both the TFH and cell cycle signature genes (Figure 1E). Consistent with these findings, GSEA also showed significant
+ enrichment of proliferation and TFH gene signatures in tumor-infiltrating CD4 T cells from tumors with high
TRM density relative to those with low TRM density (Figure 1F). These results suggest that the TFH program in
CD4+ T cells is tightly coupled with features of cell proliferation and activation in the TME. Taken together with results from the CD8+ T cell-transcripts in Module 7 (Figure 1C), our iWGCNA analysis indicates that a
+ + TFH-like transcriptional program in CD4 T cells is strongly associated with CD8 CTL proliferation, effector function and TRM features in the TME, all features of a robust anti-tumor immune response.
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Major transcriptional changes characterize tumor-infiltrating TFH-like cells
TFH cells have been reported within human cancers, where they were shown to support B cell responses, tertiary lymphoid structures and were associated with improved survival outcomes (44, 45).
Recent studies in murine breast cancer models showed the importance of IL21, presumed to be TFH- derived, in mediating response to checkpoint blockade by impacting B cell infiltration and antibody
+ production (46). Given that we found association of tumor-infiltrating TFH-like cells with CD8 CTL responses, we next sought to characterize these TFH cells in order to underpin their mechanistic role in anti- tumor immunity.
Conventional GC TFH cells are characterized by the expression of CXCR5, however in the context of tumor and inflammation, TFH cells lacking expression of CXCR5 have also been described (47, 48).
Therefore, to capture the entire spectrum of CD4+ T cells with a follicular program, we performed single-cell
RNA-seq of both CXCR5+ and CXCR5– CD4+ T cells in the TME (Figure 2A; Supplementary Table 1).
Unbiased single-cell transcriptomic analysis of 5317 tumor-infiltrating CD4+ T cells revealed 9 distinct clusters (Figure 2B). Given that GC TFH are the major producers of the B cell chemoattractant CXCL13, we utilized CXCL13 expression to mark CD4+ T cells with a follicular program. Cells expressing CXCL13 transcripts were highly enriched in cluster 3 (~70% of cells expressed CXCL13), which suggests that
CXCL13-expressing cells likely represent a distinct CD4+ T cell subset (Figure 2B; Supplementary Figure
1A and 1B; Supplementary Table 3). Single-cell analysis of the 2783 lung-infiltrating CD4+ T cells (N-TIL) showed very little CXCL13 expression (Supplementary Figure 1C). These findings were confirmed at the protein level by flow cytometric analysis of CXCL13 expression in tumor and lung tissue (Supplementary
Figure 1D).
Single-cell differential gene expression analysis of the CXCL13-expressing versus CXCL13-non- expressing CD4+ tumor-infiltrating lymphocytes (TILs) (Methods) revealed over 1000 differentially expressed transcripts (Supplementary Table 4). We found both higher expression and higher percentage of cells expressing TFH-related genes (MAF, SH2D1A, PDCD1, BTLA, CD200, BCL6(31)) in CXCL13- expressing cells than in CXCL13-non-expressing cells (Figure 2C; Supplementary Figure 1E). PDCD1 encodes for the inhibitory checkpoint molecule PD-1, which is known to be constitutively expressed in GC 7 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
+ + TFH cells and regulates GC localization and helper functions (49). Similarly, the majority of CXCL13 CD4
TILs expressed PD-1 when compared to CD4+ TILs not expressing CXCL13 (Figure 2D). GSEA confirmed significant enrichment of TFH signature genes in the CXCL13-expressing cells (Figure 2C). Together, these findings clearly established that the expression of CXCL13 delineated a TFH program, i.e., CXCL13- expressing cells represent TFH-like cells in TME. We found that CXCL13-expressing cells were present in equal proportions in CXCR5+ and the two CXCR5– subsets (CD25+CD127– and CD25–) (Figure 2B;
Supplementary Figure 1B), which indicated that using CXCR5 as a surface marker for TFH cells would have missed the majority of TFH cells. Consistent with iWGCNA results (Figure 1C-E), higher proportions of
CXCL13-expressing cells expressed cell cycle-related transcripts, and GSEA also demonstrated enrichment of cell cycle genes in the CXCL13-expressing cells (Figure 2E). Together these results indicate that despite expressing high levels of PDCD1 (Figure 2C and D), TFH-like cells actively proliferate in the tumor microenvironment presumably in response to tumor-associated antigens and may be important cellular targets of anti-PD-1 therapies.
In agreement with our results, analysis of nine published single-cell studies of CD4+ TIL transcriptomes (n = 25,149) from several cancer types showed that CXCL13-expressing TFH-like cells represented 9-36% of the CD4+ T cells in the TME (Figure 2F; Supplementary Table 5). Importantly, we
+ found a positive correlation between the proportion of CXCL13-expressing TFH-like CD4 T cells and proliferating CD8+ T cells in the TME (n = 63, Figure 2G). Across all cancer types, a greater proportion of
CXCL13-expressing TFH-like cells expressed PDCD1 transcripts relative to their CXCL13-non-expressing
+ counterparts (Figure 2H). Hence, we evaluated whether anti-PD-1 therapies targeted TFH-like CD4 T cells, which, if so, would lead to their enrichment in tumor samples post-treatment with anti-PD-1 agents. As
+ expected, we found a strong enrichment of TFH signature genes in CD4 TIL transcriptomes and an increase in the proportion of CXCL13-expressing TFH-like cells from post-treatment samples compared to pre- treatment samples (Figure 2I and J). Overall, these results indicate that TFH-like cells are an important
CD4+ T cell subset in the TME that is linked to robust CD8+ CTL responses, and they are likely to be responsive to anti-PD-1 therapy.
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
+ Tumor-infiltrating TFH-like CD4 T cells possess superior functional properties
To further probe the functional properties of these TFH-like cells, we performed pathway analysis of the transcripts differentially expressed in CXCL13-expressing cells relative to CXCL13-non-expressing cells
+ in our single-cell CD4 TIL transcriptomes. CXCL13-expressing TFH-like cells showed significant enrichment of pathways linked to co-stimulation (CD28 and ICOS-ICOSL signaling), which is important for TFH activation
(31) (Figure 3A; Supplementary Table 6). As expected of activated and co-stimulated T cells in the tumor, these TFH-like cells showed increased expression of transcripts encoding for cytokines (IFN-γ and IL21) and co-stimulation molecules (GITR and OX-40), which are known to play an important role in CD4+ T cell- mediated ‘help’ to CD8+ CTLs (50-56) (Figure 3B). Notably, the cytokine IL21 has been shown to support
CD8+ CTL survival and function (50-53). Another important finding from our analysis was that Tregs, were seen mainly within CXCL13-non-expressing CD4+ T cells, which also showed differential expression of
FOXP3 transcripts (Supplementary Figure 2A).
A surprising finding was enrichment of the cytotoxicity pathway in CXCL13-expressing cells (Figure
3A and 3C; Supplementary Table 6). We found higher expression and a higher percentage of cells expressing cytotoxicity-related transcripts (57) such as GZMB, GZMM, FKBP1A, RAB27A, CCL4 and ZEB2 in CXCL13-expressing than in CXCL13-non-expressing cells (Figure 3C). This finding was confirmed by
GSEA, which showed significant enrichment of cytotoxicity signature genes in the CXCL13-expressing cells
(Figure 3C; Supplementary Figure 2B). We verified the expression of granzyme B (a canonical
+ + cytotoxicity marker) in tumor-infiltrating CD4 TFH-like cells (CXCL13-expressing CD4 T cells) using three independent approaches: a) intracellular staining by flow cytometry, b) ImageStream imaging cytometry, and c) immunohistochemical (IHC) analyses of human lung tumor samples (Figure 3D).
Because anti-tumor functions were not previously ascribed to conventional TFH cells, we sought to investigate the precise nature of the cells that harbored these functions using our single-cell RNA-seq data where cells are annotated based on CXCR5 protein expression (Figure 2A). Since the CXCL13-expressing
+ – + – TFH-like cells were present in equal proportions in CXCR5 and the two CXCR5 subsets (CD25 CD127 and CD25–), we first asked whether the superior functional properties were attributes of all or were unique to one subset. Pairwise comparisons for differential expression analysis between the three subsets showed 9 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
that TFH signature genes were significantly enriched in all three TFH subsets (Supplementary Figure 2C), which suggested that all subsets had switched on a TFH molecular program. However, transcripts linked to superior anti-tumor properties, such as cytotoxicity (KLRB1, GZMB, CCL3, CCL4, FKBP1A, SOD1, ZEB2
+ – (57)) and provision of CD8 T cell ‘help’ (IFNG, GITR, OX40) were mainly enriched in both CXCR5 TFH subsets (Figure 3E; Supplementary Figure 2C; Supplementary Table 7). Notably, cell cycle-related
– transcripts were also expressed predominantly in this CXCR5 TFH subset, which together suggest that the functionally important TFH cells that proliferate in the TME are contained within this subset (Figure 3E;
Supplementary Figure 2C; Supplementary Table 7).
+ + TFH-like CD4 T cells colocalize with CD8 TRM cells in the TME
We undertook multi-parametric immunohistochemistry to gain insights into the organization of CD4+
TFH-like cells within the TME, and importantly, determine the spatial relationship between tumor cells,
+ tertiary lymphoid structures (TLS) and CD8 TRM cells, the density of which has been linked to good survival
+ outcomes (2-6). As expected, CXCL13-expressing TFH-like CD4 T cells were localized in TLS (48), but they
+ + were also present in the tumor core and its invasive margins (Figure 4A). TRM cells (CD8 CD103 cells) in the tumor core and invasive margins were seen in close proximity to CXCL13-expressing CD4+ T cells, which suggested potential for crosstalk and ‘help’ (Figure 4B). Therefore, we asked whether the density of
+ TFH-like CD4 T cells in the tumor core and invasive margins was positively associated with density of TRM cells in the tumor. We found a significant positive correlation (r = 0.72, P < 0.0001) between the density of
+ + + + TRM cells (CD8 CD103 cells) and TFH cells (CD4 CXCL13 cells) in both tumor core and invasive margins
(Figure 4C). Importantly, the proportion of CD4+ T cells that were CXCL13+ also positively correlated (r =
+ + 0.58, P < 0.0001) with the density of TRM cells (CD8 CD103 cells) (Figure 4C), a finding that further supports the association between a follicular program in CD4+ T cells and robust CD8+ T cell responses in tumors.
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
+ TFH cells augment CD8 CTL responses and impair tumor growth
In order to evaluate the functional role of antigen-specific TFH cells in anti-tumor immunity in vivo, we utilized the B16F10-OVA murine syngeneic tumor model in which the aggressive growth of melanoma tumors is unhindered by most interventions (58). OT-II TCR transgenic (specific for OVA 323-339) mice were first immunized with OVA in alum to generate TFH cells in vivo. We then adoptively transferred OT-II
+ + CD4 TFH cells or OT-II CD4 TEFF cells to tumor-bearing mice at 11 days post-tumor inoculation and assessed the fold increase in tumor volume over the next 48 hours (Figure 5A; Supplementary Figure 3).
+ Only the transfer of OT-II CD4 TFH cells resulted in significant reduction in tumor growth when compared to mice that received no adoptive transfer (Figure 5A). As an alternative strategy, we adoptively transferred
+ naïve OT-II CD4 cells and immunized tumor-bearing mice to induce antigen-specific TFH responses in vivo and assessed effects on tumor growth. Immunized mice demonstrated significant reduction in tumor volume relative to unimmunized controls (Figure 5B). The tumors in immunized mice had a higher proportion of TFH
neg cells (Figure 5C; Supplementary Fig. 4), and notably, the tumor-infiltrating TFH cells were CXCR5 similar to the TFH-like cells observed in human tumors. Enhanced TFH cell infiltration was also accompanied by an increased frequency of CD8+ T cells, higher proportions of which also expressed granzyme B and Ki-67, implying greater cytotoxic potential and cell proliferation, respectively (Figure 5C). Taken together, these
+ results reveal that induction of TFH response functionally bolsters anti-tumor CD8 CTL response and improves tumor control.
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
DISCUSSION
Published transcriptional studies of tumor-infiltrating CD4+ T cells from patients with cancer have largely focused on the analysis of specific CD4+ T cell subtypes or CD4+ TILs in isolation, without integration with CD8+ CTL responses (25, 27, 59). We interrogated the transcriptomes of patient-matched CD4+ and
CD8+ TILs using a novel iWGCNA-based approach to identify molecular features of CD4+ TILs that were linked to robust CD8+ anti-tumor immune responses and thus better patient outcomes. This combined TIL transcriptomic dataset, generated from patient-matched samples, facilitated the capture of in vivo TIL interactions within the TME and functional mapping of CD4+ TIL responses with those of CD8+ TILs. Such cell-specific, context-dependent, cross-talk would have otherwise been challenging to decipher in patients.
+ Using our cohort of patients with a range of CD8 and TRM TIL densities, we discovered a link
+ + between a TFH program in CD4 TILs and features of robust CD8 T cell anti-tumor immune responses such
+ as proliferation, cytotoxicity and tissue residency. Furthermore, we found that TFH-like CD4 TILs possessed superior functional properties including proliferation, cytotoxic potential and provision of ‘help’ to CD8+
CTLs. Conventional GC TFH are known to provide B cell ‘help’ during viral infections by promoting GC development, B cell affinity maturation and class switch recombination (31). However, the association of
+ + tumor-infiltrating TFH cells with CD8 T cell ‘help’ and robust CD8 CTL responses within tumors has not been described before. We uncovered increased expression of transcripts encoding molecules that mediate
+ + + CD8 T cell ‘help’ (TNFRSF18, TNFRSF4, IFNG, IL21) in TFH-like CD4 TILs. CD4 helper-derived IL21 has a prominent role in CD8+ T cell ‘help’ by inducing the BATF-IRF4 axis to sustain CD8+ T cell maintenance and effector response (50, 51, 60). OX40 and GITR signaling on CD4+ T cells critically impacts CD8+ T cell priming, accumulation and expansion (55, 56). The role of interferon-γ produced by CD4+ T cells in helping
+ + CD8 CTLs and CD8 TRM cells has been well established (54, 61). We further showed the co-localization of
+ + + CD8 TRM cells with CD4 CXCL13 TFH cells in tumor invasive margins and tumor core, which lends support
+ to the notion that TFH cells may mediate CD8 T cell ‘help’.
Our single-cell RNA-seq data unraveled another novel finding, the expression of granzyme B by the
+ + TFH-like CD4 TILs. The existence of MHC class II-restricted CD4 CTLs has been demonstrated in viral infections where they may play a particularly important role in viral clearance in the face of virus strategies
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
+ to escape CD8 CTL responses (62). Recently, GC TFH cells with cytotoxic potential have been reported in recurrent tonsillitis (63). Hence, it is plausible to hypothesize that a CTL program may also be induced in tumor-infiltrating TFH cells within tumors that have downregulated their MHC I expression. Interestingly,
+ + TGF-β signaling has been shown to promote differentiation of both CD4 CTLs (64-66) and CD8 TRM cells
(30, 67), whilst IL2 depletion may induce CXCL13 production and TFH development (68). Strikingly, both these signaling cues are abundant within tumors that harbor Tregs, suggesting that such CD4+ T cell plasticity may have evolved as a mechanism to provide these TILs with survival fitness in such immunosuppressive IL2-deprived environments. We utilized a number of complementary methods and provided a spatially resolved analysis to confirm the presence and location of these GZMB- and CXCL13- expressing CD4+ TILs.
+ In further dissecting the molecular profile of the TFH-like CD4 TILs at single-cell resolution, we
+ – + revealed TFH features in both the CXCR5 and CXCR5 CD4 T cell subsets. Our results are consistent with recent studies both in breast cancer and rheumatoid arthritis, which demonstrated the presence of CXCL13- producing TFH cells that lacked CXCR5 expression (47, 48). An additional finding from our studies was that the superior functional properties such as cytotoxicity, provision of CD8+ ‘help’ and proliferation observed in
+ – TFH-like CD4 TILs, specifically resided in the CXCR5 subset.
Previous studies have reported the presence of TFH cells in cancers, where they were shown to impact B cell activation and antibody production (44-46). Our study presents novel insights into the
+ functional role of TFH cells in augmenting CD8 CTL responses within tumors. In addition to our human data, functional evidence is provided by our in vivo murine studies, in which induction of TFH response was associated with increased CD8+ CTL infiltration, proliferation and granzyme B expression within tumors, culminating in tumor control. A further important implication for TFH cells derives from their high PD-1 expression, rendering them targets of anti-PD-1 therapy. CD8+ CTL subsets are considered to be the primary cellular responders to anti-PD-1 therapy, however, our re-analyses of published single-cell datasets indeed showed significant enrichment of tumor-infiltrating TFH cells following checkpoint blockade with anti-
+ PD-1 agents, which suggested that TFH-like CD4 T cells may also be important targets of anti-PD-1 therapies.
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
+ In summary, our study has revealed that TFH-like CD4 T cells in the TME constitute a distinct functional subset that supports CD8+ CTL responses. Further studies will enable understanding of the mechanisms underlying generation and long-term maintenance of these cells and their functional
+ significance in preventing relapse. Our findings suggest that eliciting a TFH program in CD4 T cells may be an important component of immunotherapies and vaccination approaches aimed to generate robust and
+ durable CD8 CTL and TRM responses against neo-antigens or shared tumor antigens.
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
MATERIALS AND METHODS
Human subjects
Newly diagnosed, untreated patients with non-small cell lung cancer (Supplementary Table 1) referred to Southampton University Hospitals NHS Foundation Trust and Poole Hospital NHS Foundation trust, UK between 2014 and 2017 were prospectively recruited. Freshly resected tumor tissue and, where available, matched adjacent non-tumor tissue was obtained from patients with lung cancer following surgical resection.
Mice
All mice were of C57BL6/J background and bred at LJI or purchased from the Jackson Laboratory.
All mice inoculated with tumor were females and between 8-10 weeks of age at the beginning of experiments. Within each cage, age-matched mice were randomly allocated to control or experimental groups and the investigator was not blinded to the allocation during the experiment.
Tumor model
Tumor cell lines tested negatively for mycoplasma infection and Plasmocin (InvivoGen) was used as a routine addition to culture media to prevent mycoplasma contamination. Mice were inoculated with
5 1.5x10 B16F10-OVA cells subcutaneously into the right flank. For generation of TFH cells for adoptive transfer to tumor-bearing mice, OT-II mice were injected intra-peritoneally with 100µg OVA (Invivogen) in
+ alum (Invivogen) and spleens were harvested 1 week post-immunization for flow sorting OT-II CD4 TFH or
5 5 TEFF cells (Supplementary Figure 3). Identical numbers (6x10 - 9x10 cells) of TFH or TEFF cells were transferred adoptively into tumor-bearing mice by retro-orbital injection. For induction of TFH cells in tumor- bearing mice, immunization was performed by footpad and tailbase injection of 10µg OVA in alum. Tumor size was monitored every other day, and tumor harvested at indicated time points for analysis of tumor- infiltrating lymphocytes. Tumor volume was calculated as ½ x D x d2, where D is the major axis and d is the minor axis, as described previously(69).
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Flow cytometry of fresh samples
Patient samples were processed as described previously (3). For sorting of fresh CD4+ TILs for transcriptomic analysis, cells were first incubated with FcR block (Miltenyi Biotec), then stained with a mixture of the following fluorescence-conjugated antibodies (BD Biosciences or BioLegend): anti-human
CD45 (HI30), CD4 (RPA-T4), CD3 (SK7), CD8a (SK1), HLA-DR (L243), CD14 (MφP9), CD19 (HIB19) and
CD20 (2H7) for 30 min at 4°C. Live/dead discrimination was performed by DAPI staining. CD4+ T cells were sorted into ice-cold TRIzol LS reagent (Ambion) using a BD FACSAria™ (BD Biosciences).
Murine samples –Tumor tissue was dispersed in 2ml of PBS, followed by incubation of samples in a shaker at 37°C for 15 min with DNase I (Sigma) and Liberase DL (Roche). The suspension was then diluted with MACS buffer and passed through a 70-µm cell strainer to generate a single cell suspension. Cells were first incubated with FcR block (clone 2.4G2, BD Biosciences), then with a mixture of antibodies for 30 min at 4°C. The following fluorescence-conjugated antibodies (BD Biosciences or BioLegend) were used in different panels for surface staining: anti-mouse Ctla-4 (UC10-4B9), Cd3 (145-2C11), Cd4 (RM4-5), Cd8
(53-6.7), Pd-1 (29F1.A12), Cd19 (6D5), Gitr (DTA-1), Cd45 (30-F11), Icos (C398.4A), Cxcr5 (L138D7),
Tcrb, Samples were then sorted or fixed. Intracellular staining was performed using anti-mouse Ki67 (B56),
FoxP3 (FJK-16s), Bcl6 (K112-91), GzmB (QA16A02) and FoxP3 transcription factor kit as per manufacturer’s protocol (eBioscience). Cell viability was determined using fixable viability dye
(ThermoFisher). Samples were analyzed on a BD FACS Fortessa.
+ + For sorting for adoptive transfer of OT-II CD4 TEFF cells or OT-II CD4 TFH, splenocytes were first enriched for OT-II CD4+ T cells using EasySep™ mouse CD4+ T cell isolation kit (StemCell Technologies), incubated with FcR block and stained with the following fluorescence-conjugated antibodies: anti-mouse
Cd45 (30-F11), Cd3 (145-2C11), Cd4 (RM4-5), Cd8 (53-6.7), Pd-1 (29F1.A12), Cd19 (6D5), Gitr (DTA-1),
Icos (C398.4A), Cxcr5 (L138D7), Cd25 (PC61), Cd44 (IM7) and Cd62L (MEL-14). Cell viability was determined using fixable viability dye (ThermoFisher). Samples were sorted on a BD FACS Fusion (BD
Biosciences).
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
Flow cytometry of cryopreserved samples
For 10x single-cell transcriptomic analysis and phenotypic characterization, patient tumor and lung samples were first processed and cryopreserved in freezing media (50% complete RMPI (Fisherscientific),
40% human decomplemented AB serum, 10% DMSO (both Sigma). Cryopreserved samples were thawed, incubated with FcR block (Miltenyi Biotec), then stained with a combination of the following fluorescence- conjugated antibodies (BD Biosciences or BioLegend) for sorting: anti-human CD45 (HI30), CD3 (SK7),
CD8A (SK1), CXCR5 (RF8B2), CD25 (MA251), CD127 (eBioRDR5), CD19/20 (HIB19/2H7), CD56 (HCD56) and CD4 (OKT4). Live/dead discrimination was performed using propidium iodide (PI). 1500 TILs from each of the three subsets, CD4+CXCR5+, CD4+CXCR5–CD25– and CD4+CXCR5–CD25+CD127– from tumor of each patient and 4500 CD4+ T cells (N-TILs) from adjacent uninvolved lung of each patient were sorted into
50% ice cold PBS, 50% FBS (Sigma) using a BD Aria-III (BD Biosciences).
For intracellular staining for the chemokine CXCL13 and granzyme B, TILs and N-TILs were incubated in RPMI 1640 medium (Life Technologies) containing brefeldin A (5ug/ul) for 3.5 hrs. TILs and N-
TILs were stained using Zombie Aqua fixable viability kit (Biolegend), following which surface staining was performed with a mixture of fluorescence-conjugated antibodies (BD Biosciences or BioLegend): anti- human CD45 (HI30), CD3 (SK3), CD4 (RPA-T4), CD8 (SK1), CXCR5 (RF8B2), CD25 (2A3), CD127
(A019D5), PD-1 (EH12.2H7) for 30 min at 4°C. After fixation (BD Cytofix/Cytoperm) and permeabilization
(BD Perm/Wash buffer), intracellular staining was performed with fluorescence-conjugated antibodies, anti- human CXCL13 (53610, R&D Systems) and GZMB (REA226, Miltenyl Biotec), for 30 min at 4°C. For intracellular staining for T regulatory cells, the following fluorescence-conjugated antibodies and buffers were used: anti-human Foxp3 (PCH101, ThermoFisher), CXCL13 (53610, ThermoFisher) and Foxp3 staining buffer kit (eBioscience). Samples were analyzed on a BD LSRII or ImageStreamX MkII imaging flow cytometer (Amnis, Seattle).
ImageStream Analysis
Samples were processed as described above for flow cytometry. Images were acquired on a 2- camera ImageStreamX MkII imaging flow cytometer (Amnis, Seattle) at low speed with 40X objective and
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
INSPIRE software version 200.1.620.0. The cytometer passed all ASSIST performance checks prior to image acquisition. BB515 (Ch02, 480-560 nm), PE (Ch03, 560-595 nm) PE-Dazzle594 (Ch04, 595-642 nm), PerCP-Cy5.5 (Ch05, 648-745 nm) and PE-Cy7 (Ch06, 745-780 nm) were excited at 488nm (40 mW).
BV421 (Ch07, 435-505 nm) was excited at 405 nm (20 mW). APC (Ch11, 640-745 nm) and APC-Cy7
(Ch12, 745-780 nm) were excited at 642 nm (150 mW). The acquisition gate was set to include all single, in-focus, live, CD3+ events. Data was compensated and analyzed with IDEAS software version 6.2.64.0 using the default masks and feature set.
Histology and immunohistochemistry.
Deparaffinisation, rehydration, antigen retrieval and IHC staining was carried out using a Dako PT
Link Autostainer. Antigen retrieval was performed using the EnVision FLEX Target Retrieval Solution, High pH (Agilent) for all antibodies. The primary antibodies used for IHC includes anti-CD103 (EPR4166(2);
1:500; Abcam), anti-CXCL13 (polyclonal; 1:100; ThermoFisher Scientific), anti-CD8 (C8/144B; pre-diluted;
Agilent Dako), anti-CD4 (4B12; pre-diluted; Agilent Dako), anti-granzyme B (GrB-7; 1:50; Dako) and anti-
PanCK (AE1/AE3; pre-diluted; Agilent Dako). Primary antibodies were detected using EnVision FLEX HRP
(Agilent Dako) and either Rabbit or Mouse Link reagents (Agilent Dako) as appropriate. Chromogenic visualization was completed with either two washes for five minutes in DAB or one wash for thirty minutes in
AEC and counterstained with hematoxylin. To analyze multiple markers on single sections, multiplexed IHC staining was performed as described previously(70). 4 micron tissue sections were stained with anti-PanCK antibody, visualized using DAB chromogenic substrate and scanned using a ZEISS Axio Scan.Z1 with a
20x air immersion objective. Each immune marker was then visualized using AEC chromogenic substrate and scanned. Between each staining iteration, antigen retrieval was performed along with removal of the labile AEC staining and denaturation of the preceding antibodies.
For each tissue section, regions within the tumor core (1 per section) or at the invasive margin (2 per section) were identified by a pathologist (GJT). These regions were exported as ome.tiff files and processed using Fiji image analysis software(71) as follows. The PanCK alone image was used as a reference for registering each iteration of staining, using the linear stack alignment with SIFT plugin. Color deconvolution
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
for hematoxylin, DAB and AEC staining was performed using a customized vector matrix(72). 8bit deconvoluted images were then visually inspected to determine a pixel intensity threshold of positive staining for each marker and this value was subtracted from each image to remove non-specific staining.
This color deconvolution approach resulted in DAB positive regions also being identified as AEC positive, therefore the PanCK alone image was used to generate a 0/255 pixel intensity binary “DAB mask”, which was then subtracted from each AEC image. Cell simulation and analysis was then performed using Tissue
Studio image analysis software (Definiens). A machine learning classifier was trained to recognize epithelial and stromal regions using hematoxylin and PanCK staining. Cells were then identified by nucleus detection and cytoplasmic regions were simulated up to 5µm. CD4+CXCL13+ and CD8+CD103+ cells were then enumerated within the stromal regions of each image. This analysis was performed for 41 patients out of the total 45 patients in the cohort; due to insufficient sample, 4 patients were not analyzed.
Bulk RNA sequencing
Total RNA was purified using a miRNAeasy micro kit (Qiagen, USA) and quantified as described previously (73) (on average, ~8000 CD4+ T cells per sample were processed for RNA-seq analysis).
Purified total RNA was amplified following the smart-seq2 protocol(73, 74). cDNA was purified using
AMPure XP beads (1:1.1 ratio, Beckman Coulter). From this step, 1 ng of cDNA was used to prepare a standard Nextera XT sequencing library (Nextera XT DNA sample preparation kit and index kit, Illumina).
Samples were sequenced using HiSeq2500 (Illumina) to obtain 50-bp single-end reads (Supplementary
Table 1). Quality control steps were included to determine total RNA quality and quantity, optimal number of
PCR pre-amplification cycles, and cDNA fragment size (73). Samples that failed quality control were eliminated from further downstream steps.
10x Single-cell RNA sequencing
Samples were processed using 10x v2 chemistry as per manufacturer’s recommendations; 11 and
12 cycles were used for cDNA amplification and library preparation respectively (57). Barcoded RNA was collected and processed following manufacturer recommendations, as described previously. Libraries were
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
sequenced on a HiSeq2500 (Illumina) to obtain 100- and 32-bp paired-end reads using the following read length: read 1, 26 cycles; read 2, 98 cycles; and i7 index, 8 cycles (Supplementary Table 1).
Bulk-RNA-seq analysis and iWGCNA
Bulk RNA-seq data were mapped against the hg19 reference using TopHat (75) (v1.4.1: --library- type fr-unstranded --no-coverage-search) and read counts were calculated using htseq-count -m union -s no -t exon -i gene_id (part of the HTSeq framework, version 0.7.1)) (76). Cutadapt (v1.3) was used to remove adapters.
WGCNA was completed using a R package WGCNA (v1.61) from the TPM data matrix generated from HTSeq-based read counts (28). Expressed genes with TPM >1 in at least 25% of the samples, were used in both CD4+ and CD8+ TIL data. In the integrated WGCNA approach, highly correlated genes from combined transcriptomes of patient-matched CD4+ TILS and CD8+ TILS were identified and summarized with a modular eigengene (ME) profile (28). Gene modules were generated using blockwiseModules function (parameters: checkMissingData = TRUE, power = 6, TOMType = "unsigned", minModuleSize = 50, maxBlockSize = 25426, mergeCutHeight = 0.40) (Supplementary Table 2). Module 30, which represented the default ‘grey’ module generated by WGCNA for non-co-expressed genes, was excluded from further analysis. For each gene module, individual MEs were also calculated for CD4+ TIL-genes and CD8+ TIL- genes separately. As each module by definition is comprised of highly correlated genes, their combined expression may be usefully summarized by eigengene profiles, effectively the first principal component of a given module. A small number of eigengene profiles may therefore effectively ‘summarize’ the principle patterns within the cellular transcriptome with minimal loss of information. This dimensionality-reduction approach also facilitates correlation of ME with traits. Cell cycle signature was used to generate an eigengene vector from CD8+ TIL-genes, which was then used as a trait and correlated with MEs.
Significance of correlation between this trait and MEs was assessed using linear regression with Bonferroni adjustment to correct for multiple testing.
To visualize co-expression network, we used the function exportNetworkToCytoscape at weighted = true, threshold = 0.05. A soft thresholding power was chosen based on the criterion of approximate scale-
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
free topology. Networks were generated in Gephi (v0.92) (77, 78) using Fruchterman Reingold and
Noverlap functions (Supplementary Table 2). The size and color were scaled according to the Average
Degree as calculated in Gephi, while the edge width was scaled according to the WGCNA edge weight value.
The CD103 status of TILs was determined as previously described (3). GSEA determines whether an a priori defined ‘set’ of genes show statistically significant cumulative changes in gene expression between phenotypic subgroups using Kolmogorov-Smirnov statistic and was generated as previously described (3, 79). Genes used in the GSEA analysis are shown in Supplementary Table 8.
Single-cell RNA-seq analysis
Raw 10x data was processed as previously described, merging multiple sequencing runs using cellranger count function in cell ranger, then merging multiple cell types with cell ranger aggr (v2.0.2). The merged data was transferred to the R statistical environment for analysis using the package Seurat (v2.1)
(57, 80). Only cells expressing more than 200 genes and genes expressed in at least 3 cells were included in the analysis. The data was then log-normalized and scaled per cell and variable genes were detected.
Transcriptomic data from each cell was then further normalized by the number of UMI-detected and mitochondrial genes. A principal component analysis was then run on variable genes, and the first 6 principal components (PCs) were selected for TILs for further analyses based on the standard deviation of
PCs, as determined by an elbow plot in Seurat. Cells were clustered using the FindClusters function from
Seurat with default settings, resolution = 0.6. Clusters with less than 50 cells were excluded from analysis.
Seurat software was used to identify cluster-specific differentially expressed gene sets (cutoff used is q <
0.05) (Supplementary Table 3).
Differential expression between two groups was determined by converting the data to CPM and analyzing group-specific differences using MAST (q < 0.05, v1.2.1) (57, 81, 82) (Figure 2; Supplementary
Table 4). For differential expression between three groups (CXCL13-expressing single cells in CXCR5+ subset (n = 300), CXCR5-CD25- subset (n = 319), CXCR5-CD25+CD127- subset (n = 336)) (Figure 3E), pairwise comparisons were performed (Supplementary Table 7). A gene was considered significantly
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
different (unique to a group), only if the gene was commonly positively enriched in every comparison for a singular group (57, 79). A gene was considered shared between two groups if the gene was commonly positively enriched in the two groups compared to the third group. For this analysis in Figure 3E, only genes that overlapped with those differentially expressed in CXCL13-expressing versus CXCL13-non-expressing cells (Figure 2; Supplementary Table 4) were used.
The mean CPM and percentage of cells expressing a transcript expressing cells was calculated with custom R scripts. Further visualizations of exported normalized data were generated using the Seurat package and custom R scripts. Average expression across a cell cluster was calculated using the
AverageExpression function, and downsampling was achieved using the SubsetData function (both in
Seurat).
The biological relevance of differentially expressed genes identified by MAST analysis was further investigated using the Ingenuity Pathways Analysis platform (Supplementary Table 6) as reported previously (3).
Meta-analysis of published single-cell RNA-seq studies
From each of the 9 published single-cell RNA-seq datasets (6, 19-24, 26, 27), we extracted all the cells from the cluster(s) that were annotated as tumor-infiltrating CD4+ or CD8+ clusters. For each cell type
(either CD4+ or CD8+), we then filtered out the cells with expression (>1 CPM for UMI data or >10 TPM for
Smart-seq2 data) of the other cell type’s representative transcript (i.e., filter out cells from the CD4+ clusters with CD8B expression and from the CD8+ clusters with CD4 expression) (Supplementary Table 5). We then integrated the remaining cells and their corresponding clusters from each study using the R package
Seurat v3.0 for each cell type (83). For each dataset, cells that expressed less than 200 genes were considered outliers and discarded. FindIntegrationAnchors function was used to find correspondences across the different study datasets with default parameters (dimensionality = 1:30). IntegrateData function was used to generate a Seurat Object with an integrated and batch-corrected expression matrix. For each cell type, the 2000 most variable genes were used for clustering. We used the standard workflow from
Seurat, scaling the integrated data, finding relevant components with PCA and visualizing the results with
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
UMAP. The number of relevant components was determined from an elbow plot. UMAP dimensionality reduction and clustering were applied with the following parameters: 2000 genes, 15 principal components, resolution of 0.2, min.dis 0.05 and spread 2. CXCL13-expressing or PDCD1-expressing cells were identified based on criteria defined in Supplementary Table 5. Patient tumor samples with total CD4+ cells or total
CD8+ cells <30 were excluded from further analysis to avoid sampling bias due to low cell numbers.
Cell cycle signature (Figure 2G) was derived by comparing the CD8+ cell cycle cluster with other
CD8+ clusters. The CD8+ cells were then ranked based on expression of the cell cycle signature. In this rank order, the threshold for cell cycle signature-positive cells was marked at both the lower (5.3%) and upper
(11.9%) bound of the percentage of cells that belong to the cell cycle cluster across different studies.
Quantification and statistical analysis
Hypergeometric test using phyper function and p.adjust in R was used to calculate adjusted significance values for gene enrichment tests (Figure 1E). Comparison between two groups was assessed with Mann-Whitney test (Figure 5A-C and Supplementary Figure 1D) or two-tailed paired Student’s t-test
(Figure 2D, 2H, 2J and Supplementary Figure 2A) using GraphPad Prism 7. Spearman correlation coefficient (r value) was calculated to assess significance of correlation between any two parameters of interest (Figure 2G, 4C).
Contact for reagents and resource sharing
Please contact Dr. Vijayanand ([email protected]) for reagents and resources generated in this study.
Data availability
RNA sequencing data reported in this paper has been deposited in NCBI GEO (GSE118604).
Study approval
The Southampton and South West Hampshire Research Ethics Board approved the study (Ref.
14/SC/0186), and written informed consent was obtained from all subjects (3). The Institutional Animal Care
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
and Use Committees (IACUC) of the La Jolla institute for Immunology (LJI) approved all animal studies
(Ref. AP00001126).
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
AUTHOR CONTRIBUTIONS
A.-P.G., P.S.F., T.S.-E., C.H.O., F.A., and P.V. conceived the work, designed and analyzed experiments; D.S. and A.-P.G. performed micro-scaled RNA-seq experiments under the supervision of G.S. and P.V.; D.S., A.-P.G., B.P., A.M., C.R.S. performed data analysis under the supervision of F.A., C.H.O. and P.V.; A.-P.G., S.E. and A.W. performed experiments using mouse models; C.J.H. performed immunohistochemistry and data analysis under the supervision of G.J.T.; J.C., O.W. and E.M.G.-M. helped perform the cell isolations; S.J.C., A.A. and E.W. assisted in patient recruitment, obtaining consent and sample collection; S.C., S.B., A.-P.G. and P.V. designed experiments for TFH induction in mice; A.-P.G.,
C.H.O., F.A. and P.V. wrote the manuscript. All authors read and approved the final text of the manuscript.
ACKNOWLEDGMENTS
We thank M. Chamberlain, K. Amer, B. Johnson for assistance with recruitment of study subjects and processing of samples; J. Greenbaum at the LJI bioinformatics core for help with processing and analysis of sequencing data; Y. Altman at the Sanford Burnham Prebys Flow Cytometry Core (NCI grant
P30 CA030199) and the James B. Pendleton Charitable Trust for support and access to Amnis Image
Stream analysis.
Supported by the Wessex Clinical Research Network and the National Institute of Health Research,
UK (sample collection), Cancer Research UK (digital pathology, accelerator award C11512/A20256 to
C.H.O., G.J.T., P.V.), Faculty of Medicine of the University of Southampton (P.V., T.S.-E, C.H.O.), National
Institutes of Health (K08 CA230164-01A1 to A.-P.G.) and the William K. Bowes Jr Foundation (P.V.).
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
REFERENCES
1. Ribas, A., and Wolchok, J.D. 2018. Cancer immunotherapy using checkpoint blockade. Science 359:1350-1355.
2. Djenidi, F., Adam, J., Goubar, A., Durgeau, A., Meurice, G., de Montpreville, V., Validire, P., Besse, B., and Mami-Chouaib, F. 2015. CD8+CD103+ tumor-infiltrating lymphocytes are tumor-specific tissue-resident memory T cells and a prognostic factor for survival in lung cancer patients. J Immunol 194:3475-3486.
3. Ganesan, A.P., Clarke, J., Wood, O., Garrido-Martin, E.M., Chee, S.J., Mellows, T., Samaniego- Castruita, D., Singh, D., Seumois, G., Alzetani, A., et al. 2017. Tissue-resident memory features are linked to the magnitude of cytotoxic T cell responses in human lung cancer. Nat Immunol 18:940- 950.
4. Wang, Z.Q., Milne, K., Derocher, H., Webb, J.R., Nelson, B.H., and Watson, P.H. 2016. CD103 and Intratumoral Immune Response in Breast Cancer. Clin Cancer Res.
5. Webb, J.R., Milne, K., Watson, P., Deleeuw, R.J., and Nelson, B.H. 2014. Tumor-infiltrating lymphocytes expressing the tissue resident memory marker CD103 are associated with increased survival in high-grade serous ovarian cancer. Clin Cancer Res 20:434-444.
6. Savas, P., Virassamy, B., Ye, C., Salim, A., Mintoff, C.P., Caramia, F., Salgado, R., Byrne, D.J., Teo, Z.L., Dushyanthen, S., et al. 2018. Single-cell profiling of breast cancer T cells reveals a tissue- resident memory subset associated with improved prognosis. Nat Med 24:986-993.
7. Sharma, P., and Allison, J.P. 2015. Immune checkpoint targeting in cancer therapy: toward combination strategies with curative potential. Cell 161:205-214.
8. Drake, C.G., Lipson, E.J., and Brahmer, J.R. 2014. Breathing new life into immunotherapy: review of melanoma, lung and kidney cancer. Nat Rev Clin Oncol 11:24-37.
9. Laidlaw, B.J., Craft, J.E., and Kaech, S.M. 2016. The multifaceted role of CD4(+) T cells in CD8(+) T cell memory. Nat Rev Immunol 16:102-111.
10. Janssen, E.M., Lemmens, E.E., Wolfe, T., Christen, U., von Herrath, M.G., and Schoenberger, S.P. 2003. CD4+ T cells are required for secondary expansion and memory in CD8+ T lymphocytes. Nature 421:852-856.
11. Sun, J.C., and Bevan, M.J. 2003. Defective CD8 T cell memory following acute infection without CD4 T cell help. Science 300:339-342.
12. Sun, J.C., Williams, M.A., and Bevan, M.J. 2004. CD4+ T cells are required for the maintenance, not programming, of memory CD8+ T cells after acute infection. Nat Immunol 5:927-933.
13. Schoenberger, S.P., Toes, R.E., van der Voort, E.I., Offringa, R., and Melief, C.J. 1998. T-cell help for cytotoxic T lymphocytes is mediated by CD40-CD40L interactions. Nature 393:480-483.
14. Bennett, S.R., Carbone, F.R., Karamalis, F., Flavell, R.A., Miller, J.F., and Heath, W.R. 1998. Help for cytotoxic-T-cell responses is mediated by CD40 signalling. Nature 393:478-480.
15. Feau, S., Arens, R., Togher, S., and Schoenberger, S.P. 2011. Autocrine IL-2 is required for secondary population expansion of CD8(+) memory T cells. Nat Immunol 12:908-913.
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
16. Castellino, F., Huang, A.Y., Altan-Bonnet, G., Stoll, S., Scheinecker, C., and Germain, R.N. 2006. Chemokines enhance immunity by guiding naive CD8+ T cells to sites of CD4+ T cell-dendritic cell interaction. Nature 440:890-895.
17. Nakanishi, Y., Lu, B., Gerard, C., and Iwasaki, A. 2009. CD8(+) T lymphocyte mobilization to virus- infected tissue requires CD4(+) T-cell help. Nature 462:510-513.
18. Janssen, E.M., Droin, N.M., Lemmens, E.E., Pinkoski, M.J., Bensinger, S.J., Ehst, B.D., Griffith, T.S., Green, D.R., and Schoenberger, S.P. 2005. CD4+ T-cell help controls CD8+ T-cell memory via TRAIL-mediated activation-induced cell death. Nature 434:88-93.
19. Jerby-Arnon, L., Shah, P., Cuoco, M.S., Rodman, C., Su, M.J., Melms, J.C., Leeson, R., Kanodia, A., Mei, S., Lin, J.R., et al. 2018. A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade. Cell 175:984-997 e924.
20. Lambrechts, D., Wauters, E., Boeckx, B., Aibar, S., Nittner, D., Burton, O., Bassez, A., Decaluwe, H., Pircher, A., Van den Eynde, K., et al. 2018. Phenotype molding of stromal cells in the lung tumor microenvironment. Nat Med 24:1277-1289.
21. Puram, S.V., Tirosh, I., Parikh, A.S., Patel, A.P., Yizhak, K., Gillespie, S., Rodman, C., Luo, C.L., Mroz, E.A., Emerick, K.S., et al. 2017. Single-Cell Transcriptomic Analysis of Primary and Metastatic Tumor Ecosystems in Head and Neck Cancer. Cell 171:1611-1624 e1624.
22. Guo, X., Zhang, Y., Zheng, L., Zheng, C., Song, J., Zhang, Q., Kang, B., Liu, Z., Jin, L., Xing, R., et al. 2018. Global characterization of T cells in non-small-cell lung cancer by single-cell sequencing. Nat Med 24:978-985.
23. Li, H., van der Leun, A.M., Yofe, I., Lubling, Y., Gelbard-Solodkin, D., van Akkooi, A.C.J., van den Braber, M., Rozeman, E.A., Haanen, J., Blank, C.U., et al. 2019. Dysfunctional CD8 T Cells Form a Proliferative, Dynamically Regulated Compartment within Human Melanoma. Cell 176:775-789 e718.
24. Sade-Feldman, M., Yizhak, K., Bjorgaard, S.L., Ray, J.P., de Boer, C.G., Jenkins, R.W., Lieb, D.J., Chen, J.H., Frederick, D.T., Barzily-Rokni, M., et al. 2018. Defining T Cell States Associated with Response to Checkpoint Immunotherapy in Melanoma. Cell 175:998-1013 e1020.
25. De Simone, M., Arrigoni, A., Rossetti, G., Gruarin, P., Ranzani, V., Politano, C., Bonnal, R.J.P., Provasi, E., Sarnicola, M.L., Panzeri, I., et al. 2016. Transcriptional Landscape of Human Tissue Lymphocytes Unveils Uniqueness of Tumor-Infiltrating T Regulatory Cells. Immunity 45:1135-1147.
26. Zhang, L., Yu, X., Zheng, L., Zhang, Y., Li, Y., Fang, Q., Gao, R., Kang, B., Zhang, Q., Huang, J.Y., et al. 2018. Lineage tracking reveals dynamic relationships of T cells in colorectal cancer. Nature 564:268-272.
27. Zheng, C., Zheng, L., Yoo, J.K., Guo, H., Zhang, Y., Guo, X., Kang, B., Hu, R., Huang, J.Y., Zhang, Q., et al. 2017. Landscape of Infiltrating T Cells in Liver Cancer Revealed by Single-Cell Sequencing. Cell 169:1342-1356 e1316.
28. Langfelder, P., and Horvath, S. 2008. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9:559.
29. Gebhardt, T., Palendira, U., Tscharke, D.C., and Bedoui, S. 2018. Tissue-resident memory T cells in tissue homeostasis, persistent infection, and cancer surveillance. Immunol Rev 283:54-76.
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
30. Mackay, L.K., Rahimpour, A., Ma, J.Z., Collins, N., Stock, A.T., Hafon, M.L., Vega-Ramos, J., Lauzurica, P., Mueller, S.N., Stefanovic, T., et al. 2013. The developmental pathway for CD103(+)CD8+ tissue-resident memory T cells of skin. Nat Immunol 14:1294-1301.
31. Crotty, S. 2019. T Follicular Helper Cell Biology: A Decade of Discovery and Diseases. Immunity 50:1132-1148.
32. Victora, G.D., and Nussenzweig, M.C. 2012. Germinal centers. Annu Rev Immunol 30:429-457.
33. Ise, W., Kohyama, M., Schraml, B.U., Zhang, T., Schwer, B., Basu, U., Alt, F.W., Tang, J., Oltz, E.M., Murphy, T.L., et al. 2011. The transcription factor BATF controls the global regulators of class- switch recombination in both B cells and T cells. Nat Immunol 12:536-543.
34. van de Pavert, S.A., Olivier, B.J., Goverse, G., Vondenhoff, M.F., Greuter, M., Beke, P., Kusser, K., Hopken, U.E., Lipp, M., Niederreither, K., et al. 2009. Chemokine CXCL13 is essential for lymph node initiation and is induced by retinoic acid and neuronal stimulation. Nat Immunol 10:1193-1199.
35. Kim, C.H., Lim, H.W., Kim, J.R., Rott, L., Hillsamer, P., and Butcher, E.C. 2004. Unique gene expression program of human germinal center T helper cells. Blood 104:1952-1960.
36. Luther, S.A., Lopez, T., Bai, W., Hanahan, D., and Cyster, J.G. 2000. BLC expression in pancreatic islets causes B cell recruitment and lymphotoxin-dependent lymphoid neogenesis. Immunity 12:471- 481.
37. Ma, C.S., and Deenick, E.K. 2017. The circulating life of a memory T-follicular helper cell. Clin Transl Immunology 6:e141.
38. Herati, R.S., Reuter, M.A., Dolfi, D.V., Mansfield, K.D., Aung, H., Badwan, O.Z., Kurupati, R.K., Kannan, S., Ertl, H., Schmader, K.E., et al. 2014. Circulating CXCR5+PD-1+ response predicts influenza vaccine antibody responses in young adults but not elderly adults. J Immunol 193:3528- 3537.
39. Bentebibel, S.E., Lopez, S., Obermoser, G., Schmitt, N., Mueller, C., Harrod, C., Flano, E., Mejias, A., Albrecht, R.A., Blankenship, D., et al. 2013. Induction of ICOS+CXCR3+CXCR5+ TH cells correlates with antibody responses to influenza vaccination. Sci Transl Med 5:176ra132.
40. Vinay, D.S., and Kwon, B.S. 1998. Role of 4-1BB in immune responses. Semin Immunol 10:481- 489.
41. Watts, T.H. 2005. TNF/TNFR family members in costimulation of T cell responses. Annu Rev Immunol 23:23-68.
42. Ma, J., Feng, D., Wei, Y., Tian, J., Tang, X., Rui, K., Lu, L., Xu, H., and Wang, S. 2016. Blockade of Glucocorticoid-Induced Tumor Necrosis Factor-Receptor-Related Protein Signaling Ameliorates Murine Collagen-Induced Arthritis by Modulating Follicular Helper T Cells. Am J Pathol 186:1559- 1567.
43. Guo, Y., Sun, X., Shibata, K., Yamada, H., Muta, H., Podack, E.R., and Yoshikai, Y. 2013. CD30 is required for activation of a unique subset of interleukin-17A-producing gammadelta T cells in innate immunity against Mycobacterium bovis Bacillus Calmette-Guerin infection. Infect Immun 81:3923- 3934.
44. Bindea, G., Mlecnik, B., Tosolini, M., Kirilovsky, A., Waldner, M., Obenauf, A.C., Angell, H., Fredriksen, T., Lafontaine, L., Berger, A., et al. 2013. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity 39:782-795. 28 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
45. Gu-Trantien, C., Loi, S., Garaud, S., Equeter, C., Libin, M., de Wind, A., Ravoet, M., Le Buanec, H., Sibille, C., Manfouo-Foutsop, G., et al. 2013. CD4(+) follicular helper T cell infiltration predicts breast cancer survival. J Clin Invest 123:2873-2892.
46. Hollern, D.P., Xu, N., Thennavan, A., Glodowski, C., Garcia-Recio, S., Mott, K.R., He, X., Garay, J.P., Carey-Ewend, K., Marron, D., et al. 2019. B Cells and T Follicular Helper Cells Mediate Response to Checkpoint Inhibitors in High Mutation Burden Mouse Models of Breast Cancer. Cell 179:1191-1206 e1121.
47. Rao, D.A., Gurish, M.F., Marshall, J.L., Slowikowski, K., Fonseka, C.Y., Liu, Y., Donlin, L.T., Henderson, L.A., Wei, K., Mizoguchi, F., et al. 2017. Pathologically expanded peripheral T helper cell subset drives B cells in rheumatoid arthritis. Nature 542:110-114.
48. Gu-Trantien, C., Migliori, E., Buisseret, L., de Wind, A., Brohee, S., Garaud, S., Noel, G., Dang Chi, V.L., Lodewyckx, J.N., Naveaux, C., et al. 2017. CXCL13-producing TFH cells link immune suppression and adaptive memory in human breast cancer. JCI Insight 2.
49. Shi, J., Hou, S., Fang, Q., Liu, X., Liu, X., and Qi, H. 2018. PD-1 Controls Follicular T Helper Cell Positioning and Function. Immunity 49:264-274 e264.
50. Elsaesser, H., Sauer, K., and Brooks, D.G. 2009. IL-21 is required to control chronic viral infection. Science 324:1569-1572.
51. Yi, J.S., Du, M., and Zajac, A.J. 2009. A vital role for interleukin-21 in the control of a chronic viral infection. Science 324:1572-1576.
52. Moroz, A., Eppolito, C., Li, Q., Tao, J., Clegg, C.H., and Shrikant, P.A. 2004. IL-21 enhances and sustains CD8+ T cell responses to achieve durable tumor immunity: comparative evaluation of IL-2, IL-15, and IL-21. J Immunol 173:900-909.
53. Shi, W., Dong, L., Sun, Q., Ding, H., Meng, J., and Dai, G. 2018. Follicular helper T cells promote the effector functions of CD8(+) T cells via the provision of IL-21, which is downregulated due to PD- 1/PD-L1-mediated suppression in colorectal cancer. Exp Cell Res 372:35-42.
54. Laidlaw, B.J., Zhang, N., Marshall, H.D., Staron, M.M., Guan, T., Hu, Y., Cauley, L.S., Craft, J., and Kaech, S.M. 2014. CD4+ T cell help guides formation of CD103+ lung-resident memory CD8+ T cells during influenza viral infection. Immunity 41:633-645.
55. Pascutti, M.F., Geerman, S., Slot, E., van Gisbergen, K.P., Boon, L., Arens, R., van Lier, R.A., Wolkers, M.C., and Nolte, M.A. 2015. Enhanced CD8 T cell responses through GITR-mediated costimulation resolve chronic viral infection. PLoS Pathog 11:e1004675.
56. Song, A., Song, J., Tang, X., and Croft, M. 2007. Cooperation between CD4 and CD8 T cells for anti-tumor activity is enhanced by OX40 signals. Eur J Immunol 37:1224-1232.
57. Patil, V.S., Madrigal, A., Schmiedel, B.J., Clarke, J., O'Rourke, P., de Silva, A.D., Harris, E., Peters, B., Seumois, G., Weiskopf, D., et al. 2018. Precursors of human CD4(+) cytotoxic T lymphocytes identified by single-cell transcriptome analysis. Sci Immunol 3.
58. Kleffel, S., Posch, C., Barthel, S.R., Mueller, H., Schlapbach, C., Guenova, E., Elco, C.P., Lee, N., Juneja, V.R., Zhan, Q., et al. 2015. Melanoma Cell-Intrinsic PD-1 Receptor Functions Promote Tumor Growth. Cell 162:1242-1256.
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
59. Tirosh, I., Izar, B., Prakadan, S.M., Wadsworth, M.H., 2nd, Treacy, D., Trombetta, J.J., Rotem, A., Rodman, C., Lian, C., Murphy, G., et al. 2016. Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. Science 352:189-196.
60. Xin, G., Schauder, D.M., Lainez, B., Weinstein, J.S., Dai, Z., Chen, Y., Esplugues, E., Wen, R., Wang, D., Parish, I.A., et al. 2015. A Critical Role of IL-21-Induced BATF in Sustaining CD8-T-Cell- Mediated Chronic Viral Control. Cell Rep 13:1118-1124.
61. Green, A.M., Difazio, R., and Flynn, J.L. 2013. IFN-gamma from CD4 T cells is essential for host survival and enhances CD8 T cell function during Mycobacterium tuberculosis infection. J Immunol 190:270-277.
62. Takeuchi, A., and Saito, T. 2017. CD4 CTL, a Cytotoxic Subset of CD4(+) T Cells, Their Differentiation and Function. Front Immunol 8:194.
63. Dan, J.M., Havenar-Daughton, C., Kendric, K., Al-Kolla, R., Kaushik, K., Rosales, S.L., Anderson, E.L., LaRock, C.N., Vijayanand, P., Seumois, G., et al. 2019. Recurrent group A Streptococcus tonsillitis is an immunosusceptibility disease involving antibody deficiency and aberrant TFH cells. Sci Transl Med 11.
64. Mucida, D., Husain, M.M., Muroi, S., van Wijk, F., Shinnakasu, R., Naoe, Y., Reis, B.S., Huang, Y., Lambolez, F., Docherty, M., et al. 2013. Transcriptional reprogramming of mature CD4(+) helper T cells generates distinct MHC class II-restricted cytotoxic T lymphocytes. Nat Immunol 14:281-289.
65. Reis, B.S., Rogoz, A., Costa-Pinto, F.A., Taniuchi, I., and Mucida, D. 2013. Mutual expression of the transcription factors Runx3 and ThPOK regulates intestinal CD4(+) T cell immunity. Nat Immunol 14:271-280.
66. Cheroutre, H., and Husain, M.M. 2013. CD4 CTL: living up to the challenge. Semin Immunol 25:273- 281.
67. Zhang, N., and Bevan, M.J. 2013. Transforming growth factor-beta signaling controls the formation and maintenance of gut-resident memory T cells by regulating migration and retention. Immunity 39:687-696.
68. Locci, M., Wu, J.E., Arumemi, F., Mikulski, Z., Dahlberg, C., Miller, A.T., and Crotty, S. 2016. Activin A programs the differentiation of human TFH cells. Nat Immunol 17:976-984.
69. Juneja, V.R., McGuire, K.A., Manguso, R.T., LaFleur, M.W., Collins, N., Haining, W.N., Freeman, G.J., and Sharpe, A.H. 2017. PD-L1 on tumor cells is sufficient for immune evasion in immunogenic tumors and inhibits CD8 T cell cytotoxicity. J Exp Med 214:895-904.
70. Remark, R., Merghoub, T., Grabe, N., Litjens, G., Damotte, D., Wolchok, J.D., Merad, M., and Gnjatic, S. 2016. In-depth tissue profiling using multiplexed immunohistochemical consecutive staining on single slide. Sci Immunol 1:aaf6925.
71. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. 2012. Fiji: an open-source platform for biological-image analysis. Nat Methods 9:676-682.
72. Ruifrok, A.C., and Johnston, D.A. 2001. Quantification of histochemical staining by color deconvolution. Anal Quant Cytol Histol 23:291-299.
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.01.08.898346; this version posted January 9, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
73. Seumois, G., Zapardiel-Gonzalo, J., White, B., Singh, D., Schulten, V., Dillon, M., Hinz, D., Broide, D.H., Sette, A., Peters, B., et al. 2016. Transcriptional Profiling of Th2 Cells Identifies Pathogenic Features Associated with Asthma. J Immunol 197:655-664.
74. Rosales, S.L., Liang, S., Engel, I., Schmiedel, B.J., Kronenberg, M., Vijayanand, P., and Seumois, G. 2018. A Sensitive and Integrated Approach to Profile Messenger RNA from Samples with Low Cell Numbers. Methods Mol Biol 1799:275-301.
75. Trapnell, C., Pachter, L., and Salzberg, S.L. 2009. TopHat: discovering splice junctions with RNA- Seq. Bioinformatics 25:1105-1111.
76. Anders, S., Pyl, P.T., and Huber, W. 2015. HTSeq--a Python framework to work with high- throughput sequencing data. Bioinformatics 31:166-169.
77. Ottensmeier, C.H., Perry, K.L., Harden, E.L., Stasakova, J., Jenei, V., Fleming, J., Wood, O., Woo, J., Woelk, C.H., Thomas, G.J., et al. 2016. Upregulated Glucose Metabolism Correlates Inversely with CD8+ T-cell Infiltration and Survival in Squamous Cell Carcinoma. Cancer Res 76:4136-4148.
78. Mellone, M., Hanley, C.J., Thirdborough, S., Mellows, T., Garcia, E., Woo, J., Tod, J., Frampton, S., Jenei, V., Moutasim, K.A., et al. 2016. Induction of fibroblast senescence generates a non-fibrogenic myofibroblast phenotype that differentially impacts on cancer prognosis. Aging (Albany NY) 9:114- 132.
79. Engel, I., Seumois, G., Chavez, L., Samaniego-Castruita, D., White, B., Chawla, A., Mock, D., Vijayanand, P., and Kronenberg, M. 2016. Innate-like functions of natural killer T cell subsets result from highly divergent gene programs. Nat Immunol 17:728-739.
80. Macosko, E.Z., Basu, A., Satija, R., Nemesh, J., Shekhar, K., Goldman, M., Tirosh, I., Bialas, A.R., Kamitaki, N., Martersteck, E.M., et al. 2015. Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets. Cell 161:1202-1214.
81. Soneson, C., and Robinson, M.D. 2018. Bias, robustness and scalability in single-cell differential expression analysis. Nat Methods 15:255-261.
82. Finak, G., McDavid, A., Yajima, M., Deng, J., Gersuk, V., Shalek, A.K., Slichter, C.K., Miller, H.W., McElrath, M.J., Prlic, M., et al. 2015. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol 16:278.
83. Stuart, T., Butler, A., Hoffman, P., Hafemeister, C., Papalexi, E., Mauck, W.M., 3rd, Hao, Y., Stoeckius, M., Smibert, P., and Satija, R. 2019. Comprehensive Integration of Single-Cell Data. Cell 177:1888-1902 e1821.
31
Color Key Color and Histogram and Color Key and Histogram 35000 35000 30000 30000
Figure 1 25000 25000 20000 20000 Count Count 15000 15000 10000 A C 10000 5000 5000 +
Module 7 CD8 TIL genes 0 0 0.5 0 0.5 − −0.5 0 0.5 Value + Value SCARNA17 − CD8 SH2B3 − CD8 PITPNC1 − CD8 XPO6 − CD8 LSR − CD8 CD8−SCARNA17 TRIM44 − CD8 CD8−SH2B3 FXYD2 − CD8 CD8−PITPNC1 C1orf21 − CD8 CD8−XPO6 CD4 TCF7 − CD8 CD8−LSR ICAM2 − CD8 CD8−TRIM44 LTB − CD8 CD8−FXYD2 C20orf3 − CD8 CD8−C1orf21 ZDHHC2 − CD8 CD8−TCF7 LYAR − CD8 CD8−ICAM2 CD RAB9A − CD8 CD8−LTB TP73 − CD8 CD8−C20orf3 KIF18B − CD8 CD8−ZDHHC2 MND1 − CD8 CD8−LYAR CDC20 − CD8 CD8−RAB9A UBE2L6 − CD8 CD8−TP73 TK1 − CD8 CD8−KIF18B CDKN3 − CD8 CD8−MND1 FKBP1A − CD8 CD8−CDC20 GAPDH − CD8 CD8−UBE2L6 PGAM1 − CD8 CD8−TK1 TUBB − CD8 CD8−CDKN3 HMGN2 − CD8 CD8−FKBP1A MCM4 − CD8 CD8−GAPDH ASF1B − CD8 CD8−PGAM1 TPI1 − CD8 CD8−TUBB CD38 − CD8 CD8−HMGN2 HIST1H2AH − CD8 CD8−MCM4 KIF15 − CD8 CD8−ASF1B MCM6 − CD8 CD8−TPI1 MCM2 − CD8 CD8−CD38 CDC45 − CD8 CD8−HIST1H2AH CCNE2 − CD8 CD8−KIF15 CKAP2L − CD8 CD8−MCM6 FEN1 − CD8 CD8−MCM2 CD8−CDC45 Cell cycle KIR2DL4 − CD8 TNS3 − CD8 CD8−CCNE2 ZWINT − CD8 CD8−CKAP2L ETV7 − CD8 CD8−FEN1 KPNA2 − CD8 CD8−KIR2DL4 HAVCR2 − CD8 CD8−TNS3 RRM2 − CD8 CD8−ZWINT STMN1 − CD8 CD8−ETV7 KIAA0101 − CD8 CD8−KPNA2 DLGAP5 − CD8 CD8−HAVCR2 BIRC5 − CD8 CD8−RRM2 AURKB − CD8 CD8−STMN1 + CDCA2 − CD8 CD8−KIAA0101 DTL − CD8 CD8−DLGAP5 PKMYT1 − CD8 CD8−BIRC5 TYMS − CD8 CD8−AURKB TOP2A − CD8 CD8−CDCA2 LOC100507600 − CD8 CD8−DTL TPX2 − CD8 CD8−PKMYT1 RACGAP1 − CD8 CD8−TYMS PLK1 − CD8 CD8−TOP2A CCNA2 − CD8 CD8−LOC100507600 NUSAP1 − CD8 CD8−TPX2 KIF23 − CD8 CD8−RACGAP1 FBXO5 − CD8 CD8−PLK1 BUB1 − CD8 CD8−CCNA2 GINS2 − CD8 CD8−NUSAP1 SLC25A5 − CD8 CD8−KIF23 NEIL3 − CD8 CD8−FBXO5 TOMM34 − CD8 CD8−BUB1 MKI67 CDK1 MELK − CD8 CD8−GINS2 HIST1H3G − CD8 CD8−SLC25A5 HIST1H2AJ − CD8 CD8−NEIL3 EPSTI1 − CD8 CD8−TOMM34 ANXA5 − CD8 CD8−MELK CASC5 − CD8 CD8−HIST1H3G UHRF1 − CD8 CD8−HIST1H2AJ CLSPN − CD8 CD8−EPSTI1 RANBP1 − CD8 CD8−ANXA5 HJURP − CD8 CD8−CASC5 MYBL2 − CD8 CD8−UHRF1 CD8−CLSPN HIST1H2AM − CD8 CD8−RANBP1 FABP5 − CD8 CD8−HJURP MLF1IP − CD8 CD8−MYBL2 GZMB − CD8 CD8−HIST1H2AM MAD2L1 − CD8 CD8−FABP5 PSMC3 − CD8 CD8−MLF1IP RAN − CD8 CD8−GZMB PRDX6 − CD8 CD8−MAD2L1 NCAPG − CD8 CD8−PSMC3 CENPF − CD8 CD8−RAN MKI67 − CD8 CD8−PRDX6 MTHFD2 − CD8 CD8−NCAPG UBE2T − CD8 CD8−CENPF HIST1H4C − CD8 CD8−MKI67 HIST1H1B − CD8 CD8−MTHFD2 ESCO2 − CD8 CD8−UBE2T CDCA7 − CD8 CD8−HIST1H4C VDR − CD8 STMN1 CDCA8 CD8−HIST1H1B GTSE1 − CD8 CD8−ESCO2 PIF1 − CD8 CD8−CDCA7 CDCA8 − CD8 CD8−VDR CCNB2 − CD8 CD8−GTSE1 TROAP − CD8 CD8−PIF1 LOC541471 − CD8 CD8−CDCA8 GEM − CD8 CD8−CCNB2 FANCI − CD8 CD8−TROAP CCNB1 − CD8 CD8−LOC541471 WDHD1 − CD8 CD8−GEM RFC2 − CD8 CD8−FANCI PDIA6 − CD8 CD8−CCNB1 ENO1 − CD8 CD8−WDHD1 CD82 − CD8 CD8−RFC2 CHMP4A − CD8 CD8−PDIA6 CSF1 − CD8 CD8−ENO1 PTPN7 − CD8 CD8−CD82 PRKAG1 − CD8 CD8−CHMP4A C3orf14 − CD8 CD8−CSF1 GBP4 − CD8 CD8−PTPN7 STAT1 − CD8 CD8−PRKAG1 GBP1 − CD8 CD8−C3orf14 PTMA − CD8 CD8−GBP4 RRM1 − CD8 CD8−STAT1 HPRT1 − CD8 CD8−GBP1 ARPC2 − CD8 CD8−PTMA PGK1 − CD8 CD8−RRM1 LDHA − CD8 CD8−HPRT1 TOP2A KIF11 − CD8 CD8−ARPC2 GALNT2 − CD8 CD8−PGK1 PKM2 − CD8 CD8−LDHA MCM5 − CD8 CD8−KIF11 GBP2 − CD8 CD8−GALNT2 ITGAE − CD8 CD8−PKM2 SMC2 − CD8 CD8−MCM5 COTL1 − CD8 CD8−GBP2 COX5A − CD8 CD8−ITGAE DUT − CD8 CD8−SMC2 TUBA1B − CD8 CD8−COTL1 ACOT7 − CD8 CD8−COX5A TALDO1 − CD8 CD8−DUT WARS − CD8 CD8−TUBA1B CHEK1 − CD8 CD8−ACOT7 TME NDFIP2 − CD8 CD8−TALDO1 WDR34 − CD8 CD8−WARS BST2 − CD8 CD8−CHEK1 PSME2 − CD8 CD8−NDFIP2 CALM3 − CD8 CD8−WDR34 TOX2 − CD8 CD8−BST2 SHMT2 − CD8 CD8−PSME2 TRAFD1 − CD8 CD8−CALM3 ID3 − CD8 CD8−TOX2 SARDH − CD8 CD8−SHMT2 + HAPLN3 − CD8 CD8−TRAFD1 IGFLR1 − CD8 CD8−ID3 PSMD8 − CD8 CD8−SARDH IFI35 − CD8 CD8−HAPLN3 GBP5 − CD8 CD8−IGFLR1 OBFC2B − CD8 CD8−PSMD8 ANKRD35 − CD8 CD8−IFI35 E2F2 − CD8 CD8−GBP5 PSMB9 − CD8 CD8−OBFC2B SNRPB − CD8 CD8−ANKRD35 CD8 FDPS − CD8 CD8−E2F2 SPC24 − CD8 CD8−PSMB9 MAD2L2 − CD8 CD8−SNRPB CDK1 − CD8 CD8−FDPS CKS1B − CD8 CD8−SPC24 SNAP47 − CD8 CD8−MAD2L2 CXCR6 − CD8 CD8−CDK1 P2RY1 − CD8 CD8−CKS1B CXorf69 − CD8 CD8−SNAP47 CCL3 − CD8 CD8−CXCR6 C16orf59 − CD8 CD8−P2RY1 PBK − CD8 CD8−CXorf69 EXO1 − CD8 CD8−CCL3 CCDC74A − CD8 CD8−C16orf59 PCNA − CD8 CD8−PBK KIF2C − CD8 CD8−EXO1 PARK7 − CD8 CD8−CCDC74A SLC27A2 − CD8 CD8−PCNA AKAP5 − CD8 CD8−KIF2C GLDC − CD8 CD8−PARK7 SIRPG − CD8 CD8−SLC27A2 TNFRSF9 − CD8 CD8−AKAP5 SCCPDH − CD8 CD8−GLDC LIMK1 − CD8 CD8−SIRPG TSPAN17 − CD8 CD8−TNFRSF9 NXT2 − CD8 CD8−SCCPDH CHST14 − CD8 CD8−LIMK1 Cytotoxicity- RASD1 − CD8 CD8−TSPAN17 MAOB − CD8 CD8−NXT2 ACAT2 − CD8 CD8−CHST14 KLRC2 − CD8 CD8−RASD1 IL21 − CD8 CD8−MAOB LOC729234 − CD8 CD8−ACAT2 IFNG − CD8 CD8−KLRC2 SLC50A1 − CD8 CD8−IL21 HIST1H2BM − CD8 CD8−LOC729234 CCDC121 − CD8 CD8−IFNG HIST1H4L − CD8 CD8−SLC50A1 TMEM97 − CD8 CD8−HIST1H2BM HIST1H3J − CD8 CD8−CCDC121 PRRG4 − CD8 CD8−HIST1H4L H1F0 − CD8 CD8−TMEM97 HIST1H3F − CD8 CD8−HIST1H3J HIST1H3H − CD8 CD8−PRRG4 HIST1H2AE − CD8 CD8−H1F0 PIK3AP1 − CD8 CD8−HIST1H3F SPRY2 − CD8 CD8−HIST1H3H CHAF1A − CD8 CD8−HIST1H2AE IRF7 − CD8 CD8−PIK3AP1 FAM3C − CD8 CD8−SPRY2 MYO1E − CD8 CD8−CHAF1A KCNK1 − CD8 CD8−IRF7 WDR65 − CD8 CD8−FAM3C UQCRC2 − CD8 CD8−MYO1E LOC254559 − CD8 CD8−KCNK1 LOC100506274 − CD8 CD8−WDR65 407 CD8−UQCRC2 related SLC7A5 − CD8 CD8−LOC254559 GALNT1 − CD8 CD8−LOC100506274 POC1A − CD8 CD8−SLC7A5 FKBP4 − CD8 CD8−GALNT1 H2AFV − CD8 CD8−POC1A FAM64A − CD8 CD8−FKBP4 GSTZ1 − CD8 CD8−H2AFV CCDC86 − CD8 CD8−FAM64A POLE2 − CD8 CD8−GSTZ1 PTMS − CD8 CD8−CCDC86 GPR19 − CD8 CD8−POLE2 AMZ1 − CD8 CD8−PTMS MAPK12 − CD8 CD8−GPR19 PMVK − CD8 CD8−AMZ1 CHAC2 − CD8 CD8−MAPK12 HIST1H2AL − CD8 CD8−PMVK E2F1 − CD8 CD8−CHAC2 NDC80 − CD8 CD8−HIST1H2AL HIST1H2BF − CD8 CD8−E2F1 SCUBE1 − CD8 CD8−NDC80 DPF3 − CD8 CD8−HIST1H2BF TMEM206 − CD8 CD8−SCUBE1 HMGB3 − CD8 CD8−DPF3 CDCA5 − CD8 CD8−TMEM206 CENPH − CD8 CD8−HMGB3 PAQR4 − CD8 CD8−CDCA5 NCAPH − CD8 CD8−CENPH WDR76 − CD8 CD8−PAQR4 TRAIP − CD8 CD8−NCAPH GZMB CCL3 PARP2 − CD8 CD8−WDR76 HIST1H3C − CD8 CD8−TRAIP STIL − CD8 CD8−PARP2 EME1 − CD8 CD8−HIST1H3C NUF2 − CD8 CD8−STIL CENPN − CD8 CD8−EME1 Tumor HADH − CD8 CD8−NUF2 GBP1P1 − CD8 CD8−CENPN YWHAE − CD8 CD8−HADH KIF4A − CD8 CD8−GBP1P1 APOBEC3B − CD8 CD8−YWHAE FOXM1 − CD8 CD8−KIF4A E2F7 − CD8 CD8−APOBEC3B PRDX3 − CD8 CD8−FOXM1 CDT1 − CD8 CD8−E2F7 DHFR − CD8 CD8−PRDX3 SHCBP1 − CD8 CD8−CDT1 RAD51 − CD8 CD8−DHFR CISD1 − CD8 CD8−SHCBP1 UBB − CD8 CD8−RAD51 CKS2 − CD8 CD8−CISD1 FAM111B − CD8 CD8−UBB DEPDC1 − CD8 CD8−CKS2 TMEM106C − CD8 CD8−FAM111B C11orf75 − CD8 CD8−DEPDC1 UBE2C − CD8 CD8−TMEM106C RAD51AP1 − CD8 CD8−C11orf75 CEP55 − CD8 CD8−UBE2C ORC6 − CD8 CD8−RAD51AP1 STAT1 FKBP1A HMGB2 − CD8 CD8−CEP55 DIAPH3 − CD8 CD8−ORC6 KIFC1 − CD8 CD8−HMGB2 HMMR − CD8 CD8−DIAPH3 CENPW − CD8 CD8−KIFC1 CENPM − CD8 CD8−HMMR samples from CEND1 − CD8 CD8−CENPW SPAG5 − CD8 CD8−CENPM ANLN − CD8 CD8−CEND1 E2F8 − CD8 CD8−SPAG5 PHGDH − CD8 CD8−ANLN MCM10 − CD8 CD8−E2F8 SKA3 − CD8 CD8−PHGDH NEK2 − CD8 CD8−MCM10 PPA1 − CD8 CD8−SKA3 CMC2 − CD8 CD8−NEK2 CDCA3 − CD8 CD8−PPA1 CENPA − CD8 CD8−CMC2 SAE1 − CD8 CD8−CDCA3 ENSA − CD8 CD8−CENPA DNAJB11 − CD8 CD8−SAE1 ECH1 − CD8 CD8−ENSA DONSON − CD8 CD8−DNAJB11 PDLIM7 − CD8 CD8−ECH1 ID2 − CD8 CD8−DONSON GOLIM4 − CD8 CD8−PDLIM7 CTNNAL1 − CD8 CD8−ID2 GMNN − CD8 CD8−GOLIM4 CD8−CTNNAL1 KIR2DL4 CACYBP − CD8 COMMD7 − CD8 CD8−GMNN TRIP13 − CD8 CD8−CACYBP CCZ1B − CD8 CD8−COMMD7 PTPRK − CD8 CD8−TRIP13 PPAP2A − CD8 CD8−CCZ1B Tumor CD8−PTPRK ETV1 − CD8 36 patients CD8−PPAP2A TNFSF4 − CD8 LINC00158 − CD8 CD8−ETV1 INPP5F − CD8 CD8−TNFSF4 TIL high CD8−LINC00158 ZBED2 − CD8 TIL low CD8−INPP5F PDCD1 − CD8 PHEX − CD8 CD8−ZBED2 IFITM10 − CD8 CD8−PDCD1 CAMK1 − CD8 CD8−PHEX LAYN − CD8 CD8−IFITM10 NAB1 − CD8 CD8−CAMK1 PDLIM4 − CD8 CD8−LAYN PET112 − CD8 CD8−NAB1 WIPF3 − CD8 CD8−PDLIM4 Key Color Color Key AFAP1L2 − CD8 CD8−PET112 TWF2 − CD8 CD8−WIPF3 SEMA4A − CD8 CD8−AFAP1L2 Histogram and and Histogram LOC100506668 − CD8 CD8−TWF2 MCM7 − CD8 CD8−SEMA4A ZNF367 − CD8 CD8−LOC100506668 DDX49 − CD8 CD8−MCM7 SMTN − CD8 CD8−ZNF367 REEP2 − CD8 CD8−DDX49 TTYH3 − CD8 CD8−SMTN ORC1 − CD8 CD8−REEP2 KIF20A − CD8 CD8−TTYH3 KCNK5 − CD8 CD8−ORC1 MYO5B − CD8 CD8−KIF20A SLC4A2 − CD8 CD8−KCNK5 KRT86 − CD8 CD8−MYO5B with NSCLC TIGIT − CD8 CD8−SLC4A2 UBE2F − CD8 CD8−KRT86 6000 TSHZ2 − CD8 CD8−TIGIT VOPP1 − CD8 CD8−UBE2F NOTCH1 − CD8 CD8−TSHZ2 6000 GPR25 − CD8 CD8−VOPP1 PKIA − CD8 CD8−NOTCH1 CALR − CD8 CD8−GPR25 CD160 − CD8 CD8−PKIA MXD3 − CD8 CD8−CALR CD8−CD160 SPIN4 − CD8 CD8−MXD3 LRRC61 − CD8 CD8−SPIN4 CLIP3 − CD8 CD8−LRRC61 MAP1LC3A − CD8 CD8−CLIP3 FASN − CD8 CD8−MAP1LC3A OAZ1 − CD8 CD8−FASN ATP5B − CD8 CD8−OAZ1 HNRNPF − CD8 CD8−ATP5B CCT3 − CD8 CD8−HNRNPF FAM110A − CD8 CD8−CCT3 IDH2 − CD8 CD8−FAM110A MDH2 − CD8 CD8−IDH2 CPNE7 − CD8 CD8−MDH2 5000 TNFSF10 − CD8 CD8−CPNE7 TREX1 − CD8 RM 5000 CD8−TNFSF10 NKG7 − CD8 CD8−TREX1 IL32 − CD8 T CD8−NKG7 PSMB10 − CD8 CD8−IL32 FIBP − CD8 CD8−PSMB10 ELOF1 − CD8 CD8−FIBP COPE − CD8 CD8−ELOF1 LAG3 − CD8 CD8−COPE PFN1 − CD8 CD8−LAG3 PSME1 − CD8 CD8−PFN1 LSM2 − CD8 CD8−PSME1 ASNA1 − CD8 CD8−LSM2 CCL5 − CD8 CD8−ASNA1 PPIB − CD8 CD8−CCL5 C18orf56 − CD8 CD8−PPIB MVD − CD8 CD8−C18orf56 C9orf16 − CD8 CD8−MVD SDHC − CD8 CD8−C9orf16 KIAA1671 − CD8 CD8−SDHC CCRL2 − CD8 CD8−KIAA1671 4000 PSMB7 − CD8
CD8−CCRL2 4000 Range of TIL status across patients GSTM4 − CD8 CD8−PSMB7 GPR34 − CD8 CD8−GSTM4 CSNK2B − CD8 CD8−GPR34 ECHS1 − CD8 CD8−CSNK2B UBL7 − CD8 CD8−ECHS1 EIF3I − CD8 CD8−UBL7 RALY − CD8 CD8−EIF3I FBXO6 − CD8 CD8−RALY YBX1 − CD8 CD8−FBXO6 MCM3 − CD8 CD8−YBX1 ITGAE FABP5 GTF3C6 − CD8 CD8−MCM3 POLD1 − CD8 CD8−GTF3C6
H2AFX − CD8 CD8−POLD1 Count PMF1 − CD8 CD8−H2AFX Count BARD1 − CD8 CD8−PMF1 PHF19 − CD8 CD8−BARD1 TFDP1 − CD8 CD8−PHF19 RDM1 − CD8 CD8−TFDP1 SFXN2 − CD8 CD8−RDM1 DAPK2 − CD8 CD8−SFXN2 3000 3000 DCPS − CD8 CD8−DAPK2 FDFT1 − CD8 CD8−DCPS SUMO2 − CD8 CD8−FDFT1 CD8−SUMO2 ID2 ID3 LTB TK1 DTL IL32 IL21 LSR PBK UBB DUT VDR RAN STIL PIF1 TPI1 − − IRF7 MVD FIBP PPIB IDH2 PKIA IFI35 GEM − LYAR IFNG EIF3I E2F8 E2F7 E2F1 E2F2 TP73 − PPA1 H1F0 PLK1 − RALY UBL7 ETV1 BST2 TPX2 ETV7 TCF7 LAYN TOX2 LAG3 − − − ID3 ID2 CCL5 PFN1 TIGIT DPF3 NXT2 CCL3 CSF1 CD82 TNS3 FEN1 CD38 YBX1 SAE1 SKA3 − FASN OAZ1 BUB1 CCT3 CDT1 NUF2 RFC2 LSM2 NAB1 NEK2 CKS2 − − PMF1 CALR ECH1 ANLN AMZ1 EXO1 CDK1 FDPS GBP5 GBP2 KIF11 LDHA PGK1 GBP1 GBP4 KIF23 KIF15 TUBB XPO6 − − − − − LTB NKG7 TWF2 PHEX ENSA EME1 PKM2 ENO1 MELK − DCPS CLIP3 MXD3 ORC1 ORC6 DHFR KIF4A PTMS GLDC PCNA SMC2 PTMA NEIL3 TYMS − − TK1 RDM1 SDHC COPE MDH2 SMTN CMC2 KIFC1 HADH PMVK KIF2C RRM1 RRM2 MND1 − − − − − STAT1 DTL GZMB SPIN4 LIMK1 PDIA6 MKI67 MAOB FANCI LSR IL21 IL32 MCM3 MCM7 MCM5 MCM2 MCM6 MCM4 CISD1 TRAIP BIRC5 − WARS − − − − − − − PBK ITGAE GINS2 − − − DUT UBB ATP5B WIPF3 GMNN ICAM2 − ACAT2 KRT86 − − − − − − FABP5 − − RAN VDR TPI1 PIF1 STIL FDFT1 HMMR − PHF19 CD160 SIRPG P2RY1 − − − − − − − − − − − − − − − − IRF7 PARP2 PARK7 ZWINT TFDP1 ELOF1 TSHZ2 TTYH3 CEP55 FKBP4 SPC24 ASF1B SH2B3 − − COTL1 SPAG5 − − − − − − − − − FIBP MVD SFXN2 H2AFX TREX1 UBE2F DDX49 ZBED2 RAD51 POLE2 GSTZ1 H2AFV KLRC2 AKAP5 PTPN7 UBE2T FXYD2 SPRY2 DAPK2 TOP2A PAQR4 HPRT1 − − − − − − − − − − − − − − − − − − − ACOT7 GEM IFI35 PKIA IDH2 PPIB POLD1 FBXO6 GPR34 CCRL2 ASNA1 GPR25 REEP2 CCZ1B NDC80 GPR19 CKS1B GTSE1 MYBL2 ANXA5 FBXO5 KPNA2 CDC45 CDC20 RAB9A − VOPP1 − − − − − − − − FAM3C CENPA COX5A BARD1 ECHS1 CPNE7 KCNK5 PTPRK UBE2C PRDX3 KCNK1 RASD1 CALM3 CHEK1 ARPC2 PRDX6 HJURP CLSPN UHRF1 CASC5 KIF20A KIF18B LYAR CHAC2 − − − − − − − − − − − − − − IFNG TP73 E2F2 E2F1 E2F7 E2F8 EIF3I FOXM1 PDCD1 CDCA3 CEND1 CDCA5 CXCR6 SHMT2 CCNB1 CCNB2 CDCA8 CDCA7 CENPF CCNA2 CDCA2 STMN1 CCNE2 CDKN3 PSMB7 PSME1 TRIP13 POC1A PSMB9 PSME2 ESCO2 AURKB − TROAP MYO5B MYO1E − − − − − − − − − − − − − PPA1 PLK1 H1F0 GSTM4 CAMK1 PRRG4 SNRPB PSMD8 PSMC3 EPSTI1 − − RALY − − − − − TCF7 ETV7 TPX2 BST2 ETV1 UBL7 TOX2 LAYN − INPP5F MCM10 CENPN WDR76 NCAPH CENPH WDR65 IGFLR1 SARDH WDR34 MLF1IP PGAM1 − LAG3 − − − − − − − − − − − − − CD38 FEN1 TNS3 CD82 CSF1 CCL3 NXT2 DPF3 TIGIT PFN1 CCL5 SKA3 SAE1 YBX1 SUMO2 HMGB2 HMGB3 NDFIP2 NCAPG GAPDH TRIM44 C9orf16 C3orf14 C20orf3 C1orf21 − FASN OAZ1 − − BUB1 RFC2 NUF2 CDT1 CCT3 CKS2 NEK2 NAB1 LSM2 PHGDH CENPM HMGN2 ZNF367 PET112 PDLIM4 PDLIM7 DIAPH3 − − − − − − XPO6 TUBB KIF15 KIF23 GBP4 GBP1 PGK1 LDHA KIF11 GBP2 GBP5 FDPS CDK1 EXO1 AMZ1 ANLN ECH1 CALR PMF1 − WDHD1 − YWHAE CXorf69 − − − − − − − − − − − − MELK ENO1 PKM2 EME1 ENSA PHEX TWF2 NKG7 SLC4A2 IFITM10 CENPW SLC7A5 UBE2L6 − − − − − − − − − − − − − TYMS NEIL3 PTMA SMC2 PCNA GLDC PTMS KIF4A DHFR ORC6 ORC1 MXD3 CLIP3 DCPS − − PPAP2A TALDO1 TNFSF4 GOLIM4 − − − − − − − − − − − − − − − − − FAM64A − MND1 RRM2 RRM1 KIF2C PMVK HADH KIFC1 CMC2 SMTN MDH2 COPE SDHC RDM1 − − − − GTF3C6 LRRC61 CHST14 SNAP47 CKAP2L FKBP1A − − − − − − STAT1 − − − − − − − − − − − − − − − − − − − GZMB MKI67 PDIA6 LIMK1 SPIN4 − TRAFD1 GBP1P1 GALNT1 MAD2L2 HAPLN3 TUBA1B GALNT2 MAD2L1 − MAOB FANCI − − − − − − − − − − − − − − − − − − − − − MCM4 MCM6 MCM2 MCM5 MCM7 MCM3 BIRC5 TRAIP CISD1 − WARS − PSMB10 MAPK12 CHAF1A DLGAP5 − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − GINS2 ITGAE HAVCR2 − − − CSNK2B SHCBP1 SCUBE1 CCDC86 OBFC2B RANBP1 NUSAP1 PKMYT1 − − − PRKAG1 ATP5B − − − − − − − ICAM2 GMNN WIPF3 − ACAT2 DEPDC1 TMEM97 MTHFD2 ZDHHC2 SEMA4A PIK3AP1 KIR2DL4 KRT86 − − − − − − FABP5 NOTCH1 − − CACYBP − − − − − − − − − − − − HMMR FDFT1 − SIRPG CD160 PHF19 P2RY1 C18orf56 C11orf75 C16orf59 − − − − − − − − − − − − − − TOMM34 PARK7 PARP2 − CD8 − − CD8 − − − − − − − − ZWINT SH2B3 ASF1B SPC24 FKBP4 CEP55 TTYH3 TSHZ2 ELOF1 TFDP1 − − COTL1 HNRNPF SPAG5 − CHMP4A − − − − − − − − − − − − − − − − AFAP1L2 FXYD2 UBE2T PTPN7 AKAP5 KLRC2 H2AFV GSTZ1 POLE2 RAD51 ZBED2 DDX49 UBE2F TREX1 H2AFX SFXN2 SPRY2 TOP2A DAPK2 PAQR4 UQCRC2 SCCPDH PITPNC1 HPRT1 − − − − − − − − − − − − − − − − − − − ACOT7 − − − RAB9A CDC20 CDC45 KPNA2 FBXO5 ANXA5 MYBL2 GTSE1 CKS1B GPR19 NDC80 CCZ1B REEP2 GPR25 ASNA1 CCRL2 GPR34 FBXO6 POLD1 VOPP1 SLC50A1 SLC27A2 SLC25A5 − − − − − − − − FAM3C CENPA − − − − − CD8 COX5A CASC5 UHRF1 CLSPN HJURP PRDX6 ARPC2 CHEK1 CALM3 RASD1 KCNK1 PRDX3 UBE2C PTPRK KCNK5 CPNE7 ECHS1 BARD1 KIF18B KIF20A TSPAN17 TNFSF10 CHAC2 − − − − − − − − − − − − − − − − − − FOXM1 FAM110A FAM111B CDKN3 CCNE2 STMN1 CDCA2 CCNA2 CENPF CDCA7 CDCA8 CCNB2 CCNB1 SHMT2 CXCR6 CDCA5 CEND1 CDCA3 PDCD1 ESCO2 PSME2 PSMB9 POC1A TRIP13 PSME1 PSMB7 AURKB TROAP DONSON DNAJB11 MYO1E MYO5B − − − − − − − − − − − − − − − − − CD8 − − − EPSTI1 PSMC3 PSMD8 SNRPB PRRG4 CAMK1 GSTM4 − COMMD7 − − − − − − CD8 − − − − − − − − PGAM1 MLF1IP WDR34 SARDH IGFLR1 WDR65 CENPH NCAPH WDR76 CENPN MCM10 INPP5F − TNFRSF9 KIAA1671 KIAA0101 − − − − − − − − − − − CD8 − − CD8 − CD8 TRIM44 GAPDH NCAPG NDFIP2 HMGB3 HMGB2 SUMO2 C1orf21 C20orf3 C3orf14 C9orf16 CTNNAL1 CCDC121 − − − − − − − CD8 − − − − HMGN2 CENPM PHGDH DIAPH3 PDLIM7 PDLIM4 PET112 ZNF367 − − TMEM206 HIST1H3J ANKRD35 − − − − − CD8 − CD8 − − − − − WDHD1 − CXorf69 YWHAE − − − HIST1H4L CCDC74A − − − − − − CD8 CD8 − − CD8 − CD8 CD8 UBE2L6 SLC7A5 CENPW IFITM10 SLC4A2 − − − HIST1H3F − − − − − − − − − − − − TALDO1 PPAP2A RACGAP1 − CD8 − GOLIM4 TNFSF4 HIST1H1B − − − − − − − − − − − − − − − − FAM64A − − − − CD8 CD8 − − − − CHST14 LRRC61 GTF3C6 FKBP1A CKAP2L SNAP47 − HIST1H3C HIST1H3H HIST1H4C − − − − − − − − − − − − − − − − − − − − CD8 CD8 CD8 CD8 CD8 TRAFD1 − − − MAD2L1 GALNT2 TUBA1B HAPLN3 MAD2L2 GALNT1 GBP1P1 − − − HIST1H3G − − − − − − − − − − − − − − − − − − − − − − DLGAP5 CHAF1A MAPK12 PSMB10 − − − − − − − − − − − − − − − − − CD8 − − − − − − − − − − HAVCR2 − − CD8 − − CD8 − CD8 CD8 CD8 CD8 CD8 PKMYT1 NUSAP1 RANBP1 OBFC2B CCDC86 SCUBE1 SHCBP1 CSNK2B PRKAG1 − − − − − − − CD8 − ZDHHC2 MTHFD2 TMEM97 DEPDC1 CD8 CD8 KIR2DL4 PIK3AP1 SEMA4A NOTCH1 CACYBP CD8 − − − − − − − − − − − − LINC00158 CD8 CD8 − CD8 CD8 CD8 CD8 C16orf59 C11orf75 C18orf56 − − CD8 CD8 TOMM34 CD8 − − − − CD8 − − CD8 − RAD51AP1 − − − − HNRNPF CD8 CD8 CD8 − CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CHMP4A CD8 CD8 CD8 − − − − − − − − CD8 AFAP1L2 CD8 CD8 CD8 CD8 CD8 CD8 PITPNC1 SCCPDH UQCRC2 CD8 CD8 CD8 CD8 − − − − MAP1LC3A − − CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 SLC25A5 SLC27A2 SLC50A1 HIST1H2AJ SCARNA17 CD8 − − − − − TSPAN17 TNFSF10 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 − − − TMEM106C − − HIST1H2AL FAM111B FAM110A DONSON DNAJB11 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 − CD8 − − − APOBEC3B HIST1H2BF − − − LOC254559 LOC729234 LOC541471 COMMD7 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 − CD8 CD8 − CD8 CD8 CD8 − − − − − − − − HIST1H2AE CD8 TNFRSF9 CD8 KIAA0101 KIAA1671 CD8 CD8 CD8 CD8 − CD8 CD8 − CD8 − − CD8 − HIST1H2AH CD8 CCDC121 CTNNAL1 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − − − − CD8 − − − − − ANKRD35 HIST1H3J TMEM206 CD8 CD8 − − CD8 − CD8 − − − − HIST1H2BM HIST1H2AM − CD8 − CCDC74A HIST1H4L CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 − CD8 − − CD8 CD8 − HIST1H3F CD8 CD8 CD8 CD8 − CD8 CD8 RACGAP1 − − CD8 CD8 CD8 HIST1H1B CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 − − − CD8 CD8 HIST1H4C HIST1H3H HIST1H3C CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − − CD8 HIST1H3G CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 − − CD8 − CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 − CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 − − − CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 LINC00158 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 − − CD8 CD8 CD8 RAD51AP1 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − − CD8 CD8 CD8 CD8 MAP1LC3A CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 CD8 SCARNA17 HIST1H2AJ CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 − TMEM106C HIST1H2AL CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 HIST1H2BF APOBEC3B CD8 CD8 LOC541471 LOC729234 LOC254559 CD8 CD8 CD8 CD8 − CD8 CD8 − CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 HIST1H2AE CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 HIST1H2AH CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 HIST1H2AM HIST1H2BM CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 − − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 LOC100506668 LOC100506274 LOC100507600 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 LOC100507600 LOC100506274 LOC100506668 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 − − − CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 CD8 2000 2000 1000 1000 Module 7 CD4+ TIL genes 0 0 0.5 0 0.5 − −0.5 0 0.5 Value Value
GPSM3 − CD4
CERS2 − CD4 CD4−GPSM3
LY9 − CD4 CD4−CERS2
DEGS1 − CD4 CD4−LY9
SLC46A3 − CD4 CD4−DEGS1
SLC2A1 − CD4 CD4−SLC46A3
EMP3 − CD4 CD4−SLC2A1
AKTIP − CD4 CD4−EMP3
NUDCD3 − CD4 CD4−AKTIP
DHRS3 − CD4 CD4−NUDCD3
RBM23 − CD4 CD4−DHRS3
PWP2 − CD4 CD4−RBM23
PSAT1 − CD4 CD4−PWP2
GDPD5 − CD4 CD4−PSAT1
SPATS2L − CD4 CD4−GDPD5
LOC541471 − CD4 CD4−SPATS2L IL12RB2 − CD4 CD4−LOC541471 Cell cycle STMN1 − CD4 CD4−IL12RB2
BATF − CD4 CD4−STMN1
TNIP3 − CD4 CD4−BATF
CD38 − CD4 CD4−TNIP3
GBP2 − CD4 CD4−CD38
C9orf16 − CD4 CD4−GBP2
TNFRSF18 − CD4 CD4−C9orf16
LINC00152 − CD4 CD4−TNFRSF18 NME1 − CD4 CD4−LINC00152 CXCL13 − CD4 CD4−NME1 PDCD1 − CD4 CD4−CXCL13 UCP2 − CD4 CD4−PDCD1 CD7 − CD4 CD4−UCP2 MKI67 CDK1 TYMP − CD4 ~~~~~~~~ ~~~~~~~~ CD4−CD7 TK1 − CD4 CD4−TYMP PKM2 − CD4 CD4−TK1 ENO1 − CD4 CD4−PKM2 RRM2 − CD4 CD4−ENO1 TPI1 − CD4 CD4−RRM2 GAPDH − CD4 CD4−TPI1 TRIM69 − CD4 CD4−GAPDH IL1R2 − CD4 RNA-Seq of CD4−TRIM69 ACTG2 − CD4 CD4−IL1R2 SOD1 − CD4 CD4−ACTG2 NCAPG − CD4 CD4−SOD1 ZWINT − CD4 STMN1 CDCA8 CD4−NCAPG ~~~~~~~~ H2AFV − CD4 ~~~~ CD4−ZWINT MCM4 − CD4 ~~~~ CD4−H2AFV TNFRSF9 − CD4 CD4−MCM4 GINS2 − CD4 CD4−TNFRSF9 FAM96A − CD4 CD4−GINS2 KIF2C − CD4 CD4−FAM96A LDHA − CD4 CD4−KIF2C MYBL2 − CD4 CD4−LDHA LMNB1 − CD4 purified CD4−MYBL2 SLC1A4 − CD4 CD4−LMNB1 TNFRSF8 − CD4 CD4−SLC1A4 CDCA5 − CD4 CD4−TNFRSF8 CKAP2L − CD4 TOP2A CD4−CDCA5 MELK − CD4 CD4−CKAP2L ~~~~~~~~ UHRF1 − CD4 ~~~~~~~~ CD4−MELK CEP55 − CD4 CD4−UHRF1 SMC4 − CD4 + CD4−CEP55 ARL6IP1 − CD4 CD4−SMC4 TOP2A − CD4 CD4−ARL6IP1 MTHFD1 − CD4 + CD4−TOP2A TPX2 − CD4 + CD4−MTHFD1 CD4 TIL MKI67 − CD4 CD4−TPX2 MLF1IP − CD4 CD4−MKI67 FAM111B − CD4 CD4−MLF1IP CD4 TIL DLGAP5 − CD4 CD4−FAM111B CD8 TIL FKBP1A − CD4 CD4−DLGAP5 BIRC5 − CD4 CD4−FKBP1A ~~~~~~~~ CDCA8 − CD4 ~~~~ CD4−BIRC5 KIAA0101 − CD4 ~~~~ CD4−CDCA8 CDK1 − CD4 CD4−KIAA0101 UBE2C − CD4 CD4−CDK1 DTL − CD4 CD4−UBE2C GLMN − CD4 CD4−DTL CDKN2C − CD4 CD4−GLMN ARL15 − CD4 CD4−CDKN2C FACS GBP4 − CD4 CD4−ARL15 EED − CD4 transcriptome PARP9 − CD4 CD4−GBP4 transcriptome SYT11 − CD4 CD4−EED FH CD4−PARP9 T -related ~~~~~~~~ ~~~~~~~~ STAT1 − CD4 GBP1 − CD4 CD4−SYT11
ASB2 − CD4 CD4−STAT1
UBE2L6 − CD4 CD4−GBP1
PSMB9 − CD4 CD4−ASB2
GBP5 − CD4 CD4−UBE2L6
SPCS1 − CD4 CD4−PSMB9
HIST1H2AH − CD4 CD4−GBP5
GBP1P1 − CD4 CD4−SPCS1 sorting CCZ1B − CD4 CD4−HIST1H2AH FAM166B − CD4 CD4−GBP1P1 CD4−CCZ1B
178 HIST1H2BJ − CD4 ~~~~ ~~~~~~~~ NEK2 − CD4 CD4−FAM166B CXCL13 BATF ~~~~ CAMK1 − CD4 CD4−HIST1H2BJ BTN2A2 − CD4 CD4−NEK2
DEPDC1 − CD4 CD4−CAMK1
CCNB1 − CD4 CD4−BTN2A2
ALDOC − CD4 CD4−DEPDC1
SLBP − CD4 CD4−CCNB1
RAC3 − CD4 CD4−ALDOC
IDH2 − CD4 CD4−SLBP
SLC27A2 − CD4 CD4−RAC3
TYMS − CD4 CD4−IDH2
TPM3 − CD4 CD4−SLC27A2
CDC45 − CD4 CD4−TYMS CALM3 − CD4 CD4−TPM3 PDCD1 CD38 LEMD1 − CD4 CD4−CDC45 ~~~~~~~~ BUB1 − CD4 CD4−CALM3 ~~~~~~~~ KIF15 − CD4 CD4−LEMD1 MCM10 − CD4 CD4−BUB1
AURKA − CD4 CD4−KIF15
DHFR − CD4 CD4−MCM10
ASAH2B − CD4 CD4−AURKA
MYO5C − CD4 CD4−DHFR
SLC3A2 − CD4 CD4−ASAH2B
HNRPLL − CD4 CD4−MYO5C
H2AFY − CD4 CD4−SLC3A2
SP140 − CD4 CD4−HNRPLL TRPC4AP − CD4 CD4−H2AFY IL12RB2 ~~~~~~~~ RBPJ − CD4 CD4−SP140 ~~~~~~~~ HMGB1 − CD4 CD4−TRPC4AP SNORD70 − CD4 CD4−RBPJ
C11orf82 − CD4 CD4−HMGB1
RAB27A − CD4 CD4−SNORD70
LYST − CD4 CD4−C11orf82
NCAPH − CD4 CD4−RAB27A RFC5 − CD4 CD4−LYST ZNF593 − CD4 CD4−NCAPH TMSB10 − CD4 CD4−RFC5 COX8A − CD4 CD4−ZNF593 SUB1 − CD4 CD4−TMSB10 CKS1B − CD4 CD4−COX8A EPSTI1 − CD4 CD4−SUB1 TUBB − CD4 CD4−CKS1B RANBP1 − CD4 CD4−EPSTI1 LMCD1 − CD4 CD4−TUBB LOC100507582 − CD4 CD4−RANBP1 TIGIT − CD4 CD4−LMCD1 SIRPG − CD4 CD4−LOC100507582 MAP1LC3A − CD4 CD4−TIGIT WARS − CD4 CD4−SIRPG ZBED2 − CD4 CD4−MAP1LC3A SLC44A3 − CD4 CD4−WARS RARRES3 − CD4 CD4−ZBED2 IDO1 − CD4 CD4−SLC44A3 COPS2 − CD4 Activation CD4−RARRES3 SLC35A3 − CD4 Integrated WGCNA CD4−IDO1 HELLS − CD4 CD4−COPS2 NEBL − CD4 CD4−SLC35A3 KIAA1524 − CD4 CD4−HELLS RFC3 − CD4 CD4−NEBL TBK1 − CD4 CD4−KIAA1524 RAD17 − CD4 CD4−RFC3 NETO2 − CD4 CD4−TBK1 SNORD74 − CD4 CD4−RAD17 CHCHD4 − CD4 CD4−NETO2 GPSM2 − CD4 CD4−SNORD74 LOC284581 − CD4 CD4−CHCHD4 TNFRSF18 (GITR) CDT1 − CD4 CD4−GPSM2 AK4 − CD4 CD4−LOC284581 TUBA1B − CD4 CD4−CDT1 JAKMIP1 − CD4 CD4−AK4 SKA2 − CD4 CD4−TUBA1B ABI1 − CD4 CD4−JAKMIP1 COX4NB − CD4 CD4−SKA2 SAMD9L − CD4 CD4−ABI1 CHST11 − CD4 CD4−COX4NB VDAC2 − CD4 CD4−SAMD9L SEPT2 − CD4 CD4−CHST11 ATAD5 − CD4 CD4−VDAC2 KIF20B − CD4 CD4−SEPT2 TNFRSF9 (4-1BB) FAM193A − CD4 CD4−ATAD5 MRC2 − CD4 CD4−KIF20B RDH10 − CD4 CD4−FAM193A PRR19 − CD4 CD4−MRC2 CTLA4 − CD4 CD4−RDH10
CD4−PRR19 LY9 TK1 DTL AK4 CD7 EED TPI1 ABI1 − IDH2 IDO1
LYST CD4−CTLA4 − BATF − − TBK1 TPX2 TIGIT RBPJ CD38 SKA2 SLBP ASB2 − BUB1 CDT1 RFC3 RFC5 NEBL SUB1 NEK2 RAC3 − TUBB KIF15 TPM3 GBP5 GBP1 GBP4 CDK1 LDHA IL1R2 UCP2 GBP2 − MELK SOD1 ENO1 PKM2 EMP3 − DHFR TYMS SMC4 TYMP NME1 MRC2 KIF2C RRM2 TNIP3 PWP2 − STAT1 GLMN MKI67 AKTIP − MCM4 − BIRC5 WARS ATAD5 GINS2 SP140 − PSAT1 SYT11 ARL15 − − CTLA4 SIRPG − − − − − − PARP9 ZWINT SEPT2 HELLS CEP55 − − − − − − − − PRR19 RAD17 ZBED2 H2AFY H2AFV TOP2A − − − − − − − − − − − VDAC2 RDH10 CKS1B CDC45 CCZ1B SPCS1 MYBL2 NETO2 ACTG2 − − − − − COX8A LEMD1 CALM3 UBE2C UHRF1 LMNB1 RBM23 CERS2 KIF20B − − − − − LMCD1 CCNB1 CDCA8 CDCA5 PDCD1 STMN1 DHRS3 COPS2 PSMB9 DEGS1 AURKA − − − − − EPSTI1 ALDOC CAMK1 GDPD5 − MYO5C − − − GPSM2 NCAPH MCM10 MLF1IP GPSM3 − − − HMGB1 NCAPG TRIM69 GAPDH C9orf16 − − − ZNF593 − − − − − SLC3A2 UBE2L6 SLC1A4 SLC2A1 − − − − − BTN2A2 CXCL13 − − − − − FAM96A − CHST11 RAB27A FKBP1A CKAP2L − − − − − − − − − TUBA1B TMSB10 GBP1P1 IL12RB2 − − − − − − − − − HNRPLL ASAH2B DLGAP5 − − − − − − − − − − − SAMD9L RANBP1 ARL6IP1 − − − − COX4NB − DEPDC1 MTHFD1 − − − − − CDKN2C C11orf82 − − − − − JAKMIP1 SPATS2L CHCHD4 NUDCD3 − CD4 SLC35A3 SLC44A3 SLC27A2 SLC46A3 − − − − − − FAM193A FAM166B FAM111B − − CD4 − − − CD4 − CD4 − − − TNFRSF8 TNFRSF9 KIAA1524 KIAA0101 − − − − − CD4 − TRPC4AP − − CD4 − − − CD4 SNORD74 RARRES3 SNORD70 − − − − CD4 CD4 − − − − CD4 CD4 − − − CD4 TNFRSF18 LINC00152 CD4 CD4 − − − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 MAP1LC3A CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 HIST1H2BJ − − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 LOC284581 LOC541471 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 HIST1H2AH CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 LOC100507582 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 LY9 CD4 CD4 − TK1 DTL AK4 CD4 CD4 CD7 EED CD4 CD4 TPI1 ABI1 CD4 − IDH2 IDO1 LYST − BATF − − TPX2 TBK1 CD38 RBPJ TIGIT ASB2 SLBP SKA2 − BUB1 RFC5 RFC3 CDT1 NEK2 SUB1 NEBL RAC3 − GBP2 UCP2 IL1R2 LDHA CDK1 GBP4 GBP1 GBP5 TPM3 KIF15 TUBB − EMP3 PKM2 ENO1 SOD1 MELK − NME1 TYMP SMC4 TYMS DHFR PWP2 TNIP3 RRM2 KIF2C MRC2 − STAT1 GLMN AKTIP MKI67 − MCM4 − BIRC5 WARS ATAD5 GINS2 SP140 − PSAT1 ARL15 SYT11 − − SIRPG CTLA4 − − − − − − PARP9 ZWINT CEP55 HELLS SEPT2 − − − − − − − − H2AFV H2AFY ZBED2 RAD17 PRR19 TOP2A − − − − − − − − − − − VDAC2 MYBL2 SPCS1 CCZ1B CDC45 CKS1B RDH10 NETO2 ACTG2 − − − − − COX8A CERS2 RBM23 LMNB1 UHRF1 UBE2C CALM3 LEMD1 KIF20B − − − − − DHRS3 STMN1 PDCD1 CDCA5 CDCA8 CCNB1 LMCD1 DEGS1 PSMB9 COPS2 AURKA − − − − − GDPD5 CAMK1 ALDOC EPSTI1 − MYO5C − − − GPSM3 MLF1IP MCM10 NCAPH GPSM2 − − − GAPDH TRIM69 NCAPG HMGB1 C9orf16 − − − ZNF593 − − − − − SLC2A1 SLC1A4 UBE2L6 SLC3A2 − CD4 − − − − CXCL13 BTN2A2 − − − − − FAM96A − CHST11 CKAP2L FKBP1A RAB27A − − − − − − − − − IL12RB2 GBP1P1 TMSB10 TUBA1B − − − − − − − − − DLGAP5 ASAH2B HNRPLL − − − − − − − − − − − RANBP1 SAMD9L ARL6IP1 − − − − COX4NB − MTHFD1 DEPDC1 − − − − − CDKN2C C11orf82 − − − − − JAKMIP1 SPATS2L NUDCD3 CHCHD4 − CD4 SLC46A3 SLC27A2 SLC44A3 SLC35A3 − − − − − − FAM111B FAM166B FAM193A − CD4 − − − − CD4 − − − CD4 − TNFRSF9 TNFRSF8 KIAA0101 KIAA1524 − − − CD4 − − − TRPC4AP − − CD4 − − CD4 − SNORD70 RARRES3 SNORD74 − − − − CD4 CD4 − − − − CD4 CD4 − − − CD4 TNFRSF18 LINC00152 − − CD4 CD4 − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 MAP1LC3A CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 HIST1H2BJ CD4 CD4 CD4 CD4 CD4 − − − CD4 CD4 CD4 CD4 CD4 LOC541471 LOC284581 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 HIST1H2AH CD4 CD4 CD4 CD4 CD4 CD4 CD4 − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 LOC100507582 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 CD4 − CD4 CD4 CD4 CD4 CD4 TNFRSF8 (CD30) Gene modules SpearmanCD4 correlation 1 2 3 4 5 n ~~~~ -0.5 0.5 ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ ~~~~ D
TIGITTNFRSF9 Correlation of modules with CD8+ TIL proliferation ITGAE
B
14 TOP2A CXCL13 )
10 Module 7 log
- Adjusted P = 1.65E-13 CD38 MKI67 Significance of correlation with value( 7 CD8+ TIL proliferation P Significance threshold (P = 0.001) CD38 3 CCNA2
Adjusted STMN1 CDK1 0 LAG3 6 TOP2A CD4+ TIL MKI67 HAVCR2 CD8+ TIL CDK1 3 STMN1
Modulesize(x1000) 0 GZMB
0.4 1 6 11 16 21 26 0.5
0.6 BATF
0.7
0.8 +
0.9 CD4 TIL genes 80 degrees
1.0 1.1 CD8+ TIL genes 20 degrees E Edge weight 25 0.1 0.05
Module 7 TIL (%)
+ Adjusted P = 3.08E-25
12.5 F
TFH signature genesinCD4
FH 0.8 NES = 2.2 T q < 0.0001 0 1 6 11 16 21 26 0.4 1 6 11 Module16 ID 21 26 0.0 25 -0.2
TIL (%) Module 7 + Adjusted P = 1.2E-29 Cell cycle signature
12.5 0.6 NES = 2.0 q < 0.0001 0.3
0 0.0 Cellcyclegenes inCD4 11 66 11 16 2121 26 -0.2 RES Module ID 32 TRM high TRM low
Figure 1. iWGCNA of transcriptomes from patient-matched CD4+ and CD8+ T cells present within lung tumors. (A) Schematic representation of study method, iWGCNA of the CD4+ and CD8+ TIL transcriptomes and generation of integrated modules; dotted black box indicates previously published CD8+
TIL transcriptomic data (3). (B) Barplots (bottom panel) show module size (number of genes, left margin) and composition of CD4+ T cell- and CD8+ T cell-transcripts (key above plot) for each module; number below bar represents the corresponding module ID. Significance of correlation (top panel) of proliferation- related eigengene to CD8+ T cell-transcripts within each module represented by symbols (Spearman correlation, left margin); red line denotes significance threshold of Bonferroni adjusted P value = 0.001; red symbol denotes module with significant correlation. Blue box highlights Module 7. (C) Hierarchical clustering analysis showing Spearman correlation co-expression matrix of the CD8+ T cell- (above) and CD4+-T cell transcripts (below) in Module 7 (bottom key); black frame within matrix delineates gene clusters; left margin, number of genes in module; right margin, key genes enriched in the clusters. (D) Module 7 genes visualized in Gephi, the nodes are colored according to cell of origin and sized according to the number of edges
(connections), and the edge thickness is proportional to the strength of co-expression. (E) Enrichment of TFH signature genes (above) or cell cycle genes (below) in CD4+ TILs within each module; horizontal axis represents module ID; vertical axis represents percentage of genes (symbols); red symbols denote
Bonferroni adjusted P value < 0.001 (hypergeometric test). Blue box highlights Module 7. (F) GSEA of TFH
+ high low (top) or cell cycle signature (bottom) in the transcriptome of CD4 TILs from TRM (n = 11) versus TRM (n
= 11) tumors, presented as running enrichment score (RES) for the gene set, from most over-represented genes at left to most under-represented at right; values above the plot represent the normalized enrichment score (NES) and false discovery rate (FDR)-corrected significance value; Kolmogorov-Smirnov test.
33 ● ● ● ● ●●●● ●●●● ●● ●●●●●● ● ●●●●● ● ● ●●●●●●● ●● ●●●●● ● ● ●●● ● ●●●●●● ● ●● ●●●●●●●●●●● ● ●●●●●●● ●●●●●● ● ● ●●●●● ● ●●●● ● ●● ●● ● ●● ●● ●● ●●●●●● ● ●●● ●● ● ● ● ● ●●●●●●●●●●●●●● ●●● ● ● ●● ● ● ●●● ● ● ●●●●●●● ●●●●● ●●●●●●●●● ● ●● ●● ● ●●● ●● ●● ●●●●● ● ●●● ●● ●●●● ● ●● ● ●●●●●●● ●●●●● ● ●● ●●●●●●●●● ●●●●●● ● ●●● ●●● ●● ● ●● ● ●● ● ● ●●●● ●● ●●●● ● ● ● ●● ●● ●● ● ● ● ●● ● ●●● ●●●● ● ● ●●●● ●● ●● ● ● ●●●●● ● ●●●●●● ●● ●●●●●●● ●● ● ● ●●●●●● ● ●● ● ● ● ●● ● ● ●● ●● ●● ●●●● ●● ●●● ●● ●●● ●● ● ● ● ● ●● ●● ● ● ● ●●●● ●● ●● ● ●●●●●●●●● ● ●●●●●● ● ●●●● ● ●●● ●●●●●● ●●● ● ●● ●●● ●● ●● ●●●●● ●● ● ●●●●● ● ●● ●● ●●●●● ●●●●●●●●●● ●● ● ●● ●● ●● ● ● ● ●● ●●●●●●● ●● ●● ●●●● ● ●●●● ● ●●● ●● ●●●● ●● ●●●●● ●● ● ●● ●● ● ●● ● ● ●●●●●●● ●● ● ● ●●● ●●● ● ●●●●●● ● ●●●● ●● ●●●●●● ●● ●●●●●●●●●●●● ● ●●●● ●● ●●● ●●●●● ● ●● ● ●●● ● ●● ● ● ●●●●● ●●● ●● ●●●● ●●● ●●●●●● ● ●● ● ●●● ●●●● ●●●●●●● ●● ●●●●●●●● ● ● ●●● ●●●●●●● ● ●●● ●● ●●●●● ● ● ●● ●●●●●● ●● ● ● ●● ●●● ●●●● ●●● ●●●●●●●●●●● ●● ● ●●● ●●●●● ● ●●●●● ●● ●● ●● ● ●● 20 ●● ● ●● ●●● ●●● ● ●● ● ●●●● ● ●●●● ●●● ●● ●●●● ● ● ●● ● ●● ●●●●●●●● ● ● ● ● ●● ● ● ●● ● ●●● ●● ● ● ● ●●●● ●●●●●●● ● ● ● ●● ●●●●●●● ●●●●●●● ●●●●●●● ●●●●●●● ●●●●●● ● ●●●●●● ● ●● ●●●● ● ●●●●● ● ● ●●● ● ●● ●●● ●●●●●●●●●● ● ● ● ● ●● ●● ● ●●● ● ●●● ●●●● ●● ●● ●●●●● ● ● ●● ●● ● ●●●● ● ● ●●● ● ●●●●●●●● ● ● ● ● ●●● ●●● ●●●● ●●●● ●● ●●●●●● ● ●● ●● ●●●● ● ●●●●● ●●●●● ●●● ● ●●●● ●●● ●● ● ●●●●●● ● ●●●● ●● ●●● ●●●●●●● ● ●●●● ●●● ●●●● ●● ●●●●●● ● ● ●●● ● ● ●●●●●●●● ● ●●●●●●●● ●●● ●● ●●●●●●●●● ●●● ●● ●● ●● ●● ●● ●● ●● ● ● ● ●● ●● ●●● ● ● ●●●●●● ● ● ●●● ●● ●●● ● ●● ●● ● ● ● ●● ● ●●●● ●● ●●●●●●●● ●● ● ●● ● ●● ● ● ●● ● ●●● ● ●● ●●●●●●●● ●●●●●●●● ● ● ● ●●●●●● ● ● ● ●●● ●●● ● ● ●● ●●●●● ●●● ●●●●●● ●●●● ● ●● ● ● ● ●● ●●●●●●● ● ● ● ●● ●● ● ●● ● ● ●● ● ●● ● Figure 2 ● ● ●●●● ●● ● ●● ●●●●●●●●●● ● ●● ●● ●● ● ●● ●● ● ● ●●● ● ●● ●●● ●● ● ●●● ●● ●● ●●● ●●●●●● ●● ●●● ● ●● ●●●●●●●● ●●● ● ● ● ●●●●●●●●●●● ● ●●●● ●●● ●●● ● ● ● ●● ●●● ● ● ●● ● ●● ●●●●● ● ● ● ●●● ●●●●● ●●●●●● ● ●● ●● ● ●●● ●●●●● ● ●●● ●● ●●●●● ●●●●● ●● ● ●●●●●● ● ●●● ●●● ● ● ● ●● ● ●●● ● ●●● ● ● ● ●● ●● ●●● ●● ● ● ●● ● ●●● ● ●● ●●●● ●● ●●● ●●●●● ● ● ● ●●●●● ●●●●●● ●● ● ● ● ●●● ● ● ●● ●●● ● ●● ●●● ●●●●●●●●● ●● ●●●● ●● ●●● ●● ● ●● ●●●● ● ●●● ●●● ●●● ● ● ● ●● ●● ● ●● ●●● ● ● ●●● ●● ● ● ● ● ●● ● ●●●●●● ● ● ● ●●●●● ●●● B ●● ●● ●●●●● ●●●● ●●● ● ●● ●● ● ●● ● ●● ●● ●● ●●● ●●●●●● ●●● A ●● ●● ●●●● ●●●● ●●●●●●●● ●●● ● ● ● ● ● ●●●●● ●●●●●●●● ●● ●●●●●●●● ●● ●● ●●● ●● ● ●● ● ● ●● ●●●● ● ● ●●●● ●●●●●● ● ●●● ● ● ●● ●● ● ● ●●●●● ●● ● ●●● ●●● ●● ●●● ●● ● ●●● ●●●● ●●● ●● ● ●●●● ●● ●●●● ●●●● ●● ● ● ●● ● ● ● ●●●●● ●●● ●●●●●●●● ● ● ● ● ●● ● ●●● ● ● ●●●● ● ● ● ● ●● ●● ●● ●●●●● ● ● ● ● ●●● ●●● ●●●● ● ● ● ● ● ●● ● ● ●● ●●●●●●● ● ●● ●● ●● ● ● ● ●●●● ●●●●● ●● ● ●● ● expr.value ●● ●●● ●● ●●●●● ●●● ●● ●● ● ● ● ●●●●●●●●● ●●●● ●● ●●● ●●● ●● ●●● ● ● ● ●●● ●● ● ●●●●● ●●●●● ● 12 ● ● ●●●●●● ● ● ● ●●● ●●● ● ● ●●● ● ●● ●● 0 5 0 ● ●●●●● ● ● ● ●● ●● ●● ●● ●●● ● ●●●● ●●●●●●●●●● ●● ●● 12 ● ●● CXCL13‾ ●● ●● ●● ● ● ●●●● ● ●● ● ● ● ● ● ●● ● ● ● ● ●● ● ●●● ● ● ● ● ●●●● ●●● ● ●●●●●●● ● ●● ● ●● ●● ●● ● ●● ● ●● ● ●● ●● ● ● ● ● ● ●●● ●● ●●●●●● ● ● ●●●●● ● ● ● ●●●● ● ● ●●● ● ● ●● ● ● ●●● ●●●● ●●● ● ● ● ● ●●● ● ●●●●●● ● ● ● ● ●●● ●●● ● ● ● ● ● ●● ● ●● ●●● ●●●●●●●●● ● ● ●●● ● ●● ● ●●●● ● ● ●●● ● ● ● ● ● ● ●●● ●● ●●●●●●●●●●●●● ●●●●●● ● 1 ● ● ●● ●●● ● ●●●●● ●●●● ● ● ● ●● ● ● ● ● ● ●● ●●● ●●● ●●●● ● ● ● ●● ● ● ● ● ● ●● ● ●●● ● ● ● ● ● ● ●● ● ●● ● ● ●● ●●●●●● ● ●● 6 ● ●●● ●●●● ●●● ●●●● ● ●●● ●● ● ● ●●● ● ● ●● ●● ● ● ● ● 9 ●●● ● ●● ● ● ● ● ● ●●●●●● ● ●● ●● ●●●●●●● ● ●● ●●●● ● ● ●● ● ●● ●● ●●● ● ●● ●●● ● ● ● ●● ●●● ●● ●● ● ●●●●● ● ●●● ● ● ● ● ●● ● ● ● ●●● ● ●●●●● ●● ●●●● ●●● ●● ●● ●● ● ● ● ● ●●● ●●● ●●● ●● ● ● ●●●●●●● ●●●● ●● ●● ● ● ● ●● ●●● ●● ●● ● ●●●●● ●● ● ● ● ● ●● ●●●● ●●● ● ● ● ●●●●●● ● ●● ● ● ●●● ● ● ● ●● ●●●●●● ●●● + ●● ●● ● ● ●●●●● ● ●●●●●● ●● ● ●● ●● ● ●●●●● ●●●● ● ●● ●● ●● ●●●●● ●●●●●●●●●● ●●●●●● ●● ● ●●●● ● ● ●●●●● ●●● ●●●● ● ● ●●●●● ● ●● ●● ●● ● ●●● ●●●● ● ●● ●●● ●●●●●● ●●●●●●● ●●● ● ● ● ● ●● ● ● ● ● ●● ●●● ●●●●●● ●●● ●●● ●●● ● ●●● ●●●● ● ● CXCL13 ●● ●●●●●●●● ●●●●●●● ●● ● ● ●●● ● ● ●●●●● ● ●●●● ● ●●●●● ● ● ● ● ●● ●● ● ●●● ● ● ● ●●●●● ●●●● ● ●●● ●●● ● ● ● ●●● ● ● ● ● ● ● ●● ●● ● ●●●● ● ● ● ● ●●● ●●● ● ●●●●● ● ●●●●● ● ● ● ● ●● ● ● ●●● ● ●● ●● ● ●●● ●●●● ● ●●● ● ● ●● ● ●●● ● ● ● ● ●●● ● 2 ● ● ●●●● ●● ●● ● ● ●● ●●● ●●●● ●●●●●● ● ●●●● ●●● ● ● ● ●● ● ●● ● ● ● ●● ●●●● ● ● ●● ●●●● ● ●●● ● ●●● ●●●● ● ● ●● ●●●●● ●●●● ●●● ● 7 ● ● ●●● ● ●● ● ● ●● ● ●●●●● ● ●● ● ● ● ●● ● ● ● ● ●● ● ● ●●●●● ● ●● ● ●● ●● ● ● ● ●● ● ● ●●● ●● ● ● ●● ● ● ● ● ● ●● ●●●●●● ● ●●●●●●●●●●●●●●● ● ● ● ● ●●●● ●● ●● ● ●● ● ● ●●●●● ●● ● ●● ●●●● ●● ●●● ● ●●●●● ●●●● ● ●●● ● ●● ●●● ● ● ●●● ● ● ● ●●● ●● ● ●● ●●●●●●● ●●●●● ● ● ●●●●●● ● ●●●●●● ●●● ●●● ●●●●● ●● ● ●● ●● ●●●● 2020 6 ●● ● ●●●● ● ● ● ● ●●● ● ●●●● ● ●● ●●●●●●● ●● ●● ●● ●● ●● ● ●● ●●●● ● 20 ● ●●● ●● ●● ●● ●●● ●● ●●● ● ●●●●● ●●● ●● ●●● ●● ● ●● ●●●● ● ●●● ● ● ●●● ● ●●● ●● ● ●● ● ●●● ●●●● ●●●●● ●●● ●●●● ● ● ● ●●● ● ● ●●●● ●●●● ● ● ● ● ●● ●● ●● ●●● ●●● ●●●● ● ●●● ●● ● ● ●● ● ●● ●● ● ● ● 6 ●●● ● ● ●●●● ● ● ● ● ●●●●● ● ●● ●● ●●●● ● ● ● ●●●●● ●● ●● ●●● ● ●●● ● ● ● ●●●●● ● ● ● ●● ●●●●● ●●●●● ● ● ●●●●● ●●●●●● ● ● ●● ●● ● ●●●● ● ● ● ●● ●●● ● ● ● ● ●●● ● ● ●● ●●●●● ●● ● ●●● ● ●●●● ● ● ●●●●●●● ●●●●●2 ●●●●●●● ●●● ●●●● ● ● ●●● ●●●●●●● ● ● ●●●●● ● ● ●●● ● ● ●● ●●●●●● ● ●●●● ● ●● ●● ●●●● ● ● ●●●●●● ●●● ●● ●●●●● ●●● ●● ● ● ●●●●●● ● ● ●● ● ●● ● ● ●●●● ● ● ●●● ● ●●●● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● BB515 ●● ● ● ● ● ● ●●● ●●●● ● ●● ● ●●● ●● ● ● ●●● ●●●● ●● ●●● ●●●●●● ●●● ●●●●●●●●●● ● ●●● ●●● ● ● ●●● ●●●●●●● ● ●●●●●● ● ● ● ●● ● ●● ● ● ● ● ● ●● ●● ●●● ●●● ●● ● ●●● 3 8 ●●● ● ● ●● ●●● ●● ● ● ● ●● ●●●● ● ●● ●●●● ● ● ● ●●●●●●●●●● ● ●●●● ●●●● ● ●●●●●●●● ●●●● ●● ●● ●●●●●●●● ●●●●●●● ●● ● ● ● ● ● ● ● ●● ● ●●●●● ● ●●● ●●● ●●● ●●●●●●●● ● ●●●●● ● ●●●●●● ● ● ● ● ● ● ● ● ●● ●● ● ●●● ●●●● ● ● ● ●●● ●●●● ● ● ●●●● ● ● ● ● ● ● ●● ● ● ●●● ●● ● ● ●● ● ● ●● ● ● ● - ●● ● ● ● ●●● ● ● ● ● ●● ●●● ●● ●● ● ●● ●●● ● ● ● ●●● ●●● ● ● ●● ● ● ●● ●● ●● ●● ● ●● ● ● ●●● ●●● ●●●●●● ● ● ● ● ●● ● ●● ●● ● ● ●● ●●● ● ● ●● ●●●●●● ●●●●●●●●●●● ● ● ●●●●● ●●● ●●●● ●●●● ●●● ●●● ● ●●●●● ●●● ●● ●● ● ● ● ●●●● ●●● ● ●● ●● ●●●● ●●● ● ● ●●● ●●●●●●●●●●● ● ●● ● ●● ● ● ● ● ●● ●●●●●●● ●●●●● ● ●●●● ● ●●● ●●● ●●●●●●● ● ●●● ● ●●● ● ●●●●●●●● ●●● ●● ●●● ● ●● ●● ●●● ●● ●●● ● ●●● ● ● ●● ●●●●●● ● ●● ●●●●●●●●● ● ● ●● ●●● ●●● ● ●● ● ● ● ● ●●● ●● ●● ●● ●●●● ●●●●●●● ●●●●● ●●●●●● ●● ●● ●● ●● ●● ●●● ● ● ●● ● ●● ●● ● ●● ●●●●● ●● ●● ● ●● ● ●● ●● ●●●●●● ● ● ●●● ●● ●●● ● ●●●● ● ●● ● ●●●● ● ● ● ● ●●●●●●●●●● ●●● ●●●●●●●● ●●● ● ●●●● ●●●● ●● ●●●● ● ●● ●● ●●●●●●●● ● ●● ●●●●●●● ●● ●●●●● ● ●● ●●● ●●●● ● ●● ●●●●● ● ● ●● ● ● ●● tSNE_2 ● ● ● ● ●●●●● ●●● ● ● ●●● ● ●●●●●● ● ● ● ● ●● ●●●● ● ● ● 3 ●●● ●●●●● ●● ● ●● ●● ●● ● ● ● ●● ●●●● ● ●●●●● ●●●● ●●● ●●● ●● ●●●● ● ● ● ●●●●●●●● ● ●● ● ● ● ●● ● ●●●● ●● ●● ● ●●● ●● ●●● ●●● ●●● ●●● ● ● ●● ● ● ●● ●●●●●● ●●●●●●●● ●● ● ● ●●● ●● ● ● ●● ●● ● ● ●●●●●●● ●●● ● ● ●●●● ● ● ● ● ● ● ●● ●● ●●● ●● ●● ● ● ●● ●● ● ● ● ● ● ● ● ●●●●● ● ●●●●●●●●●● ● ●●●●●●●●●● ● ●●● ● ● ● ●● ● ● ●●● ● ● ●●●●● ●● ●●●● ●●● ●●●●● ●● ●● ● ● ● ● ●● ●●● ●● ●●●● 4 ● ●●●●●● ● ●●●●●●● ●● ●●● ● ●●●● ●●● ●● ● ●● ●● ●● ● ●●● ●●●●●●● ● ●●●●● ● ●●●● ● ● ● ● ● ●●●●● ●●●●● ● ●●●●● ●●●● ● ● ● ●●●● ●●●● ●● ●● ●●● ● ● ●●●●● ●●●●● ● ●● ● ● ●● ● ● 0●● ● ●●●● ● ● ● ● ● ●● ● ● ● ●●● ● ●● ● ●●● ● ●● ● ●● ●●5●●●●●● ●●●●●●● ●● ●●●●●● ● ●●● ● ●●● ● ●● ● ●● ●●●●● ●● ● ● ● ● ● ●●●●● ●●●●●●●● ●●●● ● ●● ● ●● ● ● ● ●● ●● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ●●● ●●●● ● ●● ● ● ● ● ● ● ●●●●●● ● ● ● ●●●●●●●●●●● ●●●●●● ● ●●●● ● ● ● ● ● ●● ● ●● ● ●● ● ● ●● ● ●●●● ● ● ● ● ●● ●●●●● ●●●● ●●●●●●● ● ●● ●● ● ● ● ●●● ● ● ●●● ● ●● ● ●● ● ● ●●● ● ●●●● ●● ● ●●●●●● ● ●●●●●● ● ●●●●●●●● ●●● ●●●●●● ● ●● ● ● ●●●● ●●● ●●●●●●● ●● ● ● ●● ●●● ● ●●●● ●●●●● ●●●● ● ●●●●●● ●● ● ● ●● ●●●● ● ●●●●● ●●●● ● ● ●● ●● ●●●●●●●● ● ●●● ●● ●● ● ● ●●●●●●●●●●●●●● ●●● ●●●●●● ● ●● ● ●● ● ● ●●●●●●●● ●●● ● ●●● ● ● ●● ●● ● ●● ● ● ●●● ● ●●●●●●●●●●●●● ● ●●●●●● ●● ● ● ●●● ● ●●● ● ● ●● ●● ● ●●● ● ● ●●●● ●● ●●●●●●●●● ●●●●●●●● ●● ● ●●●●● ● ●●● ●●●● ● ● ● ● ●● ● ●●●●● ● ● ● ● ●●● ● ●●● ●● ●● ● ●●● ●●●●● ● ● ● ●●● ● ● ● ●● ●● ● ●●●●●●● ●● ● ● ● ● ●●● ● ● ● ●● ●●●●● ●●● ● ●● ● ●● ● ● ●● ●●●●●● ●● ● ● ● ●● ●● ● ●●●● ●●● ● ● ●●●● ●● ● ● ● ●● ●● ●● ● ● ● ● ● ●● ●●●●●●●●●● ●●●● ●●● ● ● ●●● ●● ●●●●●● ●● ● ● ● ●● ●●●●● ● ● ●●● ●●● ●● ● ● ● ● ● ● ● ● ●●●●●●● ● ●●●●●● ●●●●●● ● ● ● ●● ●● ● ●●● ● ●● ●●●● ●● ● ●● ● ● ● ● ●●●●●● ● ● ●●●●● ● ●●● ● ●●●●●●●● ●● ●● ●● ● ●● ● ● 0 ● ●●●●●●● ●● ● ● ● ●●●●●● ●●●● ●●●●●● ●●● ●●●●● ●●●●● ● ● ●●●● ●● ● ● ● ● ● ● ● ●●●●●●● ●●●●●●●●●●●●●●●● ●●● ● ●●●●● ● ● ●●●●● ●●● ● ● ●● ● ●●● 0 ●● ●● ● ●●●●● ● ● ● ●●● ● ● ●● ● ●● ●● ●●●●● ● ●● ● ● ●●● ●● ● ● ● ● ● ● ● ●● ●● ●●●●● ● ●● ● ●●● ●● ●● ●●● ●●● ●●● ● ● ● ● ● ●● ● ● ●● ●● ● ●●● ● ● ●● ●●●●●●●● ● ●● ●● ● ● ●●●● ●● ●●● ●●● ●● ● ● ● ●●●●● ●●● ● ● ●● ● ●● ● ● ●● ●●●●●● ● ●●● ● ● ● ●●●●●●●●●●●●●● ● ● ● ●● ●●●●●●● ●●●● ● ● ● ● ● ●● ●● ●●●●●●●●●●●●●●●●●●● ●●● ● ● ● ● ●● ● ●●●●● ●● ● ●● ● ● ● ● ●●● ● ● ●● ●● ● ●●●● ●● ●●●●● ● ● ● ● ● ●●● ● ●● ● ● ● ●●● ● ● ● ●●●●●●●●●● ● ● ● ●● ●●● ● ● ●●● ● ●●●● ● ●● ● ● ●●●●● ● ● ●● ● ●●● ● ●●●● ●●●●● ● ● ●●●●●●●● ● ● ●● ● ● ●● ● ● ● ● ●●● ● ● ●●● ● ● ●●● ●●●●●●●●● ●●● ●●●●● ●● ●●● ●● ● ●● ● ● ● ●●●●●● ● ●●●● ●● ●● ● ● ● ● ●●● ● ●●●● ● ● ●●●● ●●●● ●●●● ●● ● ● ●●● ● ●●●● ● ●●●●● ●● ●● ● ●●●● ● ●● ● ● ● ● ●● ●●●●●● ● ● ●●●● ● ●● ● ●● ●● ●● ●● ● ● ● ● ● ●●●●● ● ●● ●● ● ●●● ●● ●● ●● ●● ●●●●●● ● ● ●●●● ● ●●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ●● CXCR5 ● ● ● ●●● ● ● ●●●●● ●● ● ● ●● ● ● ● ●● ●● ● ● ● ● ●● ●●● ● ●● ● ●● ●● ●● ● ● ●● ● ● ●●● ● ● ● ●●●● ● ● ● ● ● ● ● ●● ●● ● ●● ● ●● ●● ●●● ●●● ● ● ●●● ● ●● ● ● ●● ● ● ● ●●● ●● ● ●● ● ●● ● ● ●● ● ● ●●●● ●● ● ●● ● ● ●● ● ● ● ●● ●● ● ●●●●●●● ●● ●● ●● ● ●●● ●●● ● ● ●● ● ● ● ●●●●●● ●●● ●● ●●● ●● ● ●● ●●● ●●● ● ● ●●● ● ●●●●●●● ● ● ● ● ● ●● ●● ●● ● ●● ● ●●● ● ●● ● ● ●●● ● ● ●●● ● ●●●●●●●●● ● ●●● ●●● ●● ● ● ● ● ●● ● ● ● ●●● ●● ●● ●●●●●●● ● ●● ●● ● ● ●● ● ● ● ● ● ● ●● ● ●●● ●● ● ●● ●● ● ●● ●●●●●●● ●● ●●●●● ●● ●● ●● ●●● ●● ●●●● ●● ● ●● ●●●●●●● ●● ● ●●● ● ●● ●● ●●●● ● ●●● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●●● ●● ●●●● ● ● ● ● ●●●● ●●● ●●●● ●●●●●●●● ●●● ●● ●●●● ● ●● ● ● ●●●● ●●●● ● ●● ●● ●● ●●●●● ●● ● ● ● ●● ●●●● ●● ● ●● ●● ●● ●●● ● ● ● ● ●● ● ● ● ● ● ●●●●● ●● ●●● ● ●● ● ●●●●●●●●●●● ● ● ● ● ●●●●●● ●● ●● ● ● ●● ● ●●●● ● ●●●●●● ●● ●● ●● ● ●● ● ●● ●● ●● ●●● ●● ● ● ● ● ● ●●●●●●●●●● ● ● ● ●●●●● ● ●●● ● ●● ● ● ● ●● ●●●● ● ●● ● ● ● ● ●● ●●● ● ●●● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ●● ●● ● ●● ●● ●● ●● ●●●●● ●● ● ● ● ●● ●● ● ● ● ●● ● ●●● ●● ● ●●●● ●● ● ● ● ● ●●● ●● ● ● ●●● ●●● ●●●● ● ● ● ● ● ●● ● ● ●● ●●● ● ● ●● ● ● ● ●●● ● ● ● ●● ●●●●●●●●●●● ●●●● expr.value 18% ● ●●● ● ● ● ● ●●●●●●●●●● ●● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ●● ● ● ● ●●●● 18% ● ● ●● ●● ● ●● ●● ●● ● ●●● ● ● ●●●● ●● ● ● ● ●●● ● ●● ● ● ●● ●● ● ●● ● 2 ●●● ● ●●●● ●● ● ●●● ●● 2 ●● ● ● ●● ● ● ● ● ● ●● ● ● ● ●● ● ● ●●● ●●●●● ●● ●●●● ●● ● ●● ●● ● ●●●● ● ●● ● ● ● ● ● ●● ● ● ●● ● ●●● ● ● ●● ●● ● ●●● ●●●●●●●● ●●● ● ●●●● ● ● ● ● ● ● 12 ●● ●● ●●●● ●●●●●●●● ●● ● ● ● ● ● ● ● ●●●● ●● ● ● ●●● ● ● ● ●● ● ●●●● ●●● ● ● ● ●● ●● ●●●●● ● ● ● ● ● ●●● ●●●●●●●● ● ●●●●● ●●● ● ● ● ●●● ●●●●●●●●●● ●● ● ● ●● ●●●● ● ● ● ●●●●●● ●●●● ●●● ● ● ●● ● ●●● ●● ● ●●●● ● ●●●● ● ● ● ● ●●● ● ● ● ● ● ●●● ● ● ● ●● ● ●●●●● ● ● ●● ● ●●● ●●●●●● ● ●● ● ●● ● ●● ● ●● ● ●●●● ● ●●● ●● ●● ● ●●●●● ● ● ●● ● ● ●●●● ●● ●●● ●● ● ● ●● ●●●● ●●●● ●● ● ●● ●●● ● ● ●●● ●● ●● ●●● ●●−20● ● ●● ●● ●●●● ●●●●●●●●●●●● ●●● ● ●● ● ● ● ● ●●● ● ●●● ● ● ●●● ●● ● ● ●●● ● ●●● ●●●●●●●●● ● ● ●● ● ● ● ● ● ●●●●●●●● ● ●●● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ●●● ● ●●● ●● ● ● ● ●●● ●● ● ● ● ● ● ●●● ●●●● ●●● ● ●● ●● ●●● ●● ●●● ● 6 ● ●●● ● ● ● ●●● ●● ●● ● ● ● ●● ●● ● ●●● ●● ●●●● ● ●● ●●●● ●● ● ●● ●●●●●●●●●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● 0 ● ● ●● ●● ● ● ● ●● ●●● ●● ● ●● ● ● ● ●●●●● ●● ●● ● ● ● ●● ● ● ● ● ●● ● ●●● ●●●● 0 ● ● ●● ●●●●●● ●● ●● ●● ● ● ● ●● ● ●●●●●● ● ● 9 0 ● ● ●●● ● ●● ●● ● ●● ●● ● ● ●● ● ● ●● ●● ●● ●● ● ●●● ● ● ● ● ● ● ●●● ● ●● ● ● ● ●●●●● ● ● ●● ● ●● ●● ● ● ● ● ●●● ●●●●●●●● ● ● ●● ●● ●● ● ●● ●● ●● ●● ● ●● ● ● ● ●● ●●●●●●● ● ●●● ● ● ● ● ● ●● ● ● ● ●● ●●● ●● ●●●● ●●● ● ●● ● ● ● ● ●● ●● ●●● ● ● ●● ●●● ● ●● ● ●● ● ●● ● ● ● ● ● ●●●●●●●●●●●● ● ● ●● ● ●●● ● ●●● ● ●● ● ● ●● ● ●●●●●● ● ● ● ●● ●● ● ●● ● ●●● ●●● ● ● ●● ● ●●●●●● ●● ● ●● ● ●●● ● ● ●●●●●●●● ● ●● ● ● ● ●● ●● ● ● ●●●● ● ●● ● ●●●●●●● ● ●●●●●●●●●● ● ●● ● ● ● ● ● ● ●●●●●●● ●●● ●● ●●●●● ● ●●● ●●● ●● ●● ● ●●●● ●●●● ●● ● ●● ●●● ● ● ● ●●● ● ●●●● ● ●● ●●●●●●●●● ● ●● ●●●●● ● ● ● ●● ●●● ● ●● ● ●● ● ●● ● ●●●●● ● ● ●●●● ●● ●●● ● ● ●● ● ● ●● ●● ● ●● ●● ● ●●●●●●●● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ●●● ●●●● ● ● ● ●● ● ●●● ●●●●●●● ●●● ●● 6 ●●●●● ●●●● ●●●●●●●4● ●●● ● ● ● ●● ● ●● ●● ●● ● ● ●●● ●●●●● ●● ● ● ●● ● ●● ● ● ● ●●● ●● ●●●●●●●●●●● ●● ● ●●●● ●●●●● ● ● ●●● ●●●●● ●●●●●● ●● ● ● ● ●● ● ● ●● ●● ●● ●●● ● ●● ●● ● ●● ● ● ● ● ● ●●●● ● ●● ● ● ● ● ●● ●● ● ● ● ●● ● CD4-BV510 ●●●● ● ● ●●●●●●● ● ● ● ● ●●●● ● ● ●●● ●●● ●●●● ● ●●●●●●● ●●● ● ●● ●●● ● ●●● ● ● ● ● ● ● ●● ●● ● ●●●● ●● ●●●● ● ● ●● ●●●●●●●● ●●●●● ● ●●● ● ● ● ●● ● ●●●● ●● ● ● ●● ●●● ●● ● ●● ●● ● ●● ●● ●●● ● ●● ● ● ●● ● ● ● ● ● ● ● ●●● ●● ●●●●●●●●● ● ●●●●●●● ● ● ● ●●●●●● ● ● ● ●●●● ● ●● ●● ● ●●● ●● ● ● ●●● ● ●● ● ● ● ● ●●● ● ●● ●●●●●● ● ● ● ● ● ●●● ● ● ●●● ●● ● ● ● ●● tSNE ● ●● ●●●● ●● ● ● ● ●●●● ● ●●●● ● ● ● ● tSNE ● ●●●●● ●●●● ● ● ● ●●● ● ● ●●●●●● ●● ●● ● ● ●● ●● ●●●● ● ●●● ●● ● ●●●● ●● ●●●● ●●●● ●● ● ●● ● ●● ●●●● ● ● ●● ● ● ●●●●●●●●● ●● ●●●●● ●●● ● ●●●●●●● ● ● ● ● ● ●● ● ●● ●●● ● ● ●● ●● ● ●● ● ●●●● ● ●●● ● ●● ●● ●●●●●●●● ● ● ●●●●● ● ● ● ●●●●●●● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● tSNE_2 ●●●● ●●● ● ●● ● ●●●● ●●● ● ● 3 ● ●● ●●●●● ● ● ● ● ●● ●●● ● ● ●●●●●●●● ● ● ● ●● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ●●●●● ●●● ●● ●●● ●●● ● ●●●●●●●●● ● ● ● ●● ● ● ●● ●●●● ● ●●● ●●●● ● ● ● ●● ●●●●● ● ●● ● ● ●● ●●● ● ●● ● ● ● ● ● ● ●●● ●●● ● ● ● ●●●●●●● ● ●● ● ●● ● ●● ● ●● ● ●● ● ●●● ●● ●● ●●● ●● ●●●● ● ●● ●● ● ●● ●● ●● ●● ● ●●●● ● ● ●●●●●●●● ● ●●●● ● ●● ● ●● ●●●●●● ● ● ●● ●●● ● ● ● ●● ●●●●● ● ● ●●●●● ● ●●●●● ● ● ● ● ● ● ● ●●●● ●●● ●● ● ●● ●● ● ●●●●●● ● ● ●●●●●● ● ●●●● ● ●● ● ● ●●● ●● ● ● ● ● ●● ●●● ●●● ●●●● ●● ●●● ●●● ●● ● ● ●●●● ●●● ●● ●● ●● ●●● ● ● ●●●● ● ● ●●● ● ●● ●●● ● ●●●●●● ●●●● ● ● ● ● ●●●●●● ●●●● ● ● ● ● ● ●● ● ●● ●●● ● ●● ●●●●● ●●● ●●● ●●● ●●● ● ●● ● ●● ● ● ● ● ● ● ●● ● ●●●● ●● ● ● ●●●●● ●●● ●●● ●●●● ●● ●●● ●● ●● ● ●●● ●●●● ● ●● ● ● ● ●●●●●● ● ● ● ●●●●●● ● ●●● ●●●●●●● ● ●●●●●● ●●●●● ● ●● ●● ● ●●● ●●● ● ●● ● ●● ●●● ●● ● ● ●●●●●● ● ●● ●● ● ●●● ●●●●● ● 0 ●●●● ● ● ●●●●● ●●●●● ● ● ● ●●●●●●● ● ●●●● ● ● ● ●● ●●●●● ●●●● ●● ●●●●●● ● ● ●● ●●●●● ● ● ●●●●● ● ●● ● ● ● ●●●● ● ● ●●●● ● ● ●● ● ● ● ●●●●●● ● ● ●●● ●●●●●●●●●●●● ● ● ● ● ●● ● ● ● ●●● ●● ● ● ● ●● ●● ●●●● ●●●●●●● ● ● ● ● ●●● ●●●●●●● ●●● ● ● ● ● ● ●● ● ● ● ● ●●● ● ● ● ● ● ● ●●●●●●● ● ● ●● ● ● ●●●● ●●●●● ● ● ● ●●● ● ● ●●●●●●● ●●● ●● ● ● ●●● ●●●●● ● ●●● ●●● ● ● ● ●● ● ● ●●● ●●●● ●● ●● ● ●● ●●●●●● ● ●●● ●●●●●●● ●● ●● ●● ●●●●●●● ● ●● ●●●●● ●● ●● ● ●● ●● ● ● ●● ● ●●● ●●●●● ● ●●●●● ● ●●●● ●●●● ● ●●●●●●●● ● ●●●●●●●●● ● ● ● ●● ● ● ●● ●● ● ●●●● ● ●●● ●●●●● ●●●● ● ●●●● ● ● ●● ●●● ● ● ● ● ● ● ●● ●●● ● ● ●● ●●● ● ● ●●●●●● ● ●●●●● ●● ●●●● ● ●●● ●● ● ● ●●●●● ● ● ●●● ●● ● ● ● ● ●●●● ●●● ● ● ● ●●● ● ● ●●●●● ● ●● ●● ● ●●●● ● ●●●● ● ●● ● ●● ●● ●● ● ●● ● ● ●●●● ●●●●● ● ●● ●● ● ●● ●●●●● ● ●● ● ● ● ● ●●● ● ●● ●● ●● ● ●● ● ● ●●●● ●●●●● ● ●●●●●●●●●●● ●● ●● ● ●● ●●● ●●●● ● ●● ●●● ● ● ●●● ● ● ●●● ● ●● ● ●●● ●●●● ●●●● ● ● ●●●● ●● ● ●● ●●● ●●●● ● ●● ●●● ● ● ●●● ● ●●●● ● ● ●●● ●●● ● ●●● ● ●●● ●●● ● ● ● ● ●●●●●● ●●● ● ● ●● ● ● ● ● ●●●● ● ● ●●●● ●● ● ●●●● ● ●●● ● ● ●● ●●● ● ●● ● ●●●●●● ●● ● ●●● ● ● ● ● ●●●● ●● ●●●●●● ●● ● ●● ● ●● ● ● ●●●●●●●●● ● ●●●●●●●● ● ●●● ● ● ● ● ●● ●● ● ●● ●●●● ●● ● ●●●●● ●●● ● ● ●●● ●●●●●●● ● ●●●●●●●●● ●●●●● ● ● ● ● ●● ●● ● ●● ●●● ●● ● ●●●●●●● ● ● ●● ●●●● ●●● ●● ●● ● ●● ● ● ● ●●●●●●●●● ●●● ● ●●●●●●● ●●●● ●● ●●●● ●●● ●●● ● ●●● ● ●●● ●● ●●● ●● ● ● ● ●●●●●●●● ●● ●●●●●● ●●●●●●●●●●● ●● ●● ●● ●● ● ● ● ● ● ● ●● ● ●● ● ● ● ●● ● ● ● ●●●●● ●● ● ●●●●●●●●●●● ● ●●●●● ● ● ● ● ● ●●● ●●●● ● ●● ● ● ●●●● ●●● ●● ●●●●●●●●●● ● ●●●●● ●● ● ●● ●● ●●●●● ● ●●● ● ● ●●● ● ●●● ●●● ●●● ●● ●●●●●●●●● ● ● ●●●● ●● ●● ● ●●●●●● ●●● ● ●●● ● ●●● ●●● ● ●● ●●●●●● ●● ●●●● ●● ● ●●● ●●●●● 7● ● ● ● ●●● ●●● ● ●● ●●● ● ● ●● ● ●●●●●●●● ●●● ●●●● ●●● ●●●● ● ●●●●● ● ●● ●●●● ●●● ●● ● ●●● ● ●● ● ● ●●●●●●●● ● ● ●● ● ● ●● ● ● ● ● ●● ● ●●● ● ●● ● ● ● ● ●●● ● ● ● ● ●●●● ● ● ● ●●● ●●●● ● ● ●● ● ●● ●●●● ●● ● ● ●● ● ●●● ● ● ● ●● ●● ●● ●●● ● ● ●● ●●●● ●● ● ● ● ● ●●●●●●●● ●●● ●● ●●● ●●● ●● ●● ●●●●● ● ●● ●●● ●●●● ●● ●●● ● ● ●●● ● ● ● ● ● ● ● ●●●●●● ● ●● ● ● ● ●●● ●●● ●●●● ●●● ●● ●●●●● ● ●● ●● ●●●●●● ● ●● ● ●● ● ●●● ● ● ●● ●●● ● ●● ●● ●●●●●● ●●● ●● ●●● ●● ● ● ● ● ●●● ●●●●●●●●●● ● ●●● ●●●● ●● ● ●●● ●●●● ●● ● ●● ●● ●● ● ●●●●● ●●●●●●● ●● ● ● ●● ●● ● ●●●● ● ●● ● ● ●●●●● ●●●● ● ●●● ● ●●● ●● ● ● ● ● ● ● ●●●●●● ●●● ● ● ● ● ● ● ● ●● ● ●●●●●● ● ● ●●● ●● ● ● ●●● ●● ●●●●● ● ● ● ●●● ● ●●● ●●●●● ● ● ●● ● ●● ●● ● ● ●● ●●●● ● ●● ● ●●●●●● ●●● ●● ●● ●●● ●● ● ●● ● ●● ● ● ●●● ●●● ●●●●●● ● ●● ● ● ● ●●● ●●●● ● ● ●●● ● ●●●●●●●● ● ● ●● ● ● ● ● ●●● ● ● ● ●●● ●●●●●●●● ● ● ● ● ● ●●● ●● ●● ● ●●● ●●●●●●● ● ●● ● ●● ● ● ●●●● ●● ● ● ●●●●● ● ● ● ● ● ●● 3●●●● ● ●● ●●● ● ●● ●●●● ●● ● ●●● ●● ●●● ● ● ●● ●●● ●●● ● ● ● ● ● ● ● ● ● ●●●●●● ● ● ● ● ● ● ● ● ●● ● ● ●●●●●●● ● ●● ●● ●●● ●● ●● ● ●● ● ●● ● ●●● ●●● ● ● ● ● ●● ●● ●●● ● ● ● ●● ●● ● ● ● PE ● 1 ● ●● ● ●● ● ● ● ● ●●● ●● ● ● ● ●● ● ● ● ● ● ●●● ●● ● ● ● ● ● ● ● ● ●●●● ●●●● ● ● ● ●● ●●● ●● ● ●●● ●● ●●● ●●●●● ●●●● ●●●●●● ●●● ●● ●●● ●●● ●● ● ●●● ● ● ● ●●● ● ● ● ● ●●●● ●●● ●● ●●●● ●● ● ●●●●●● ●● ●●●● ●● ●● ●●● ● ● - ●● ●● ● ● ● ● ● ● ● -20 ● ●● ●● ● ● ●● ● -20 ●●● ● ● ● ● ● ●●●● ● ● ●● ●●● ● ● ●● ●● −20 ●●●●●●● ● ● ● ● ● ●●●● ●● ● ● ●●●●●●● ● ● ●●●● ● ●● ●● ● ● ●●● ● ●● ●●●●●●●●●●●● ●●● ● ●● ●● ●● ●●●● ● ●●● ●● ●●● ●●● ●●●●● ● ● ●●●●●●● ● ●● ● ●●●●● ●●●●●●●●● ● ●● ● ●●●●●● ●●●● ●●●● ●●●● ●● ●● ●● ● ● ●● ● ● ● ● ● ● ● ●●● ● ●●●●● ● ● ● ● ● ●●●●● ●●● ● ●●● ●● ● ● ●● ● ●●●●●● ●●● ●●●●●● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ●● ● ● ●● ●● ●● ●● ● ●●●●● ●● ● ●●● ● ● ●●●●●● ●● ●●● ●● ● ● ● ●● ● ● ●●● ● ● ● ● ● ●●●● ● ● ● ●●●●●●● ●● ●● ●●●● ● ● ● ●●●● ● ● ● ● ●● ● ● ●●●●● ●●● ● ●● ●●● ● ●●●●● ● ●●●● ●● ● ● ● ●●● ● ●●● ● ● ●●● ●●●●●●●●●●● ●● ●●● ●● ● ● ● ● ●●● ● ● ●●● ● ●●●● ●● ●●●● ●● ●● ● ●●●● ● ●●● ●● ●●● ●●●● ● ●●●● ● ●●● ● ●●●● ● ● ●●● ●●● ●● ● ● ● ● ●●● ●● ●●● ● ●●● ●●● ●●● ● ● ●●●●● ● ●● ●●● ●●●●●●● ●● ● ● ● ● ●●●●●● ● ● ● ● ● ●●● ●● ●● ●●●●●●● ●●● ●● ●●●●● ● ● ● ●● ●●● ●●●●●●●● ●● ●●●● ● ●●●● ●● ● ●● ●●●● ● ●●●●● ● ●● ● ● ●●●●●●●● ● ●●● ●● ●● ● ● ● ●● ● ● ● ●●● ● ●● ● ●●●●●●● ●●●●●● ● ●●●● ●● ● ●●●●●●●● ●●●●●● ● ● ● ●●●●● ● ●● ●●● ●● ●● ● ● ●● ● ● ●●● ● ● ● ● ●●●● ● ● ●●● ● ●● ● ●●●●●● ●● ●● ●●●●●●● ●●●●●● ● ● ● ●●●●● ● ● ●●● ● ●●● ● −40 ● ●●●● ● ●● ●●● ●●●● ● ● ●●●●●● ● ● ●● ● ●● ● ●●●●● ● ●●● ●●● ●●● ● ● ● ●● ● ●● ● ● ●●●● ●● + CD25 ●●●●●● ●●●●●● ● ●● ● ●●●● ●● ●● ●● ● ● ●●●●●●●●●●●●●●●●●● ● ●●●●●● ●●● ●●●●●●●●●●●●●●● ●● ● ●●●●●●●●●●● ● ●●●● ●●●●●●●●●●●●●●●●● CXCR5 ● ● ●● ●●●●● ● ●●●● ●●●●●●●● ● ●● ●● ●●● ●● ●● ●●●● ●●●●●●●●● ●● ● ●● ●● ● ● ●● ●●● ● ●●● ● ● ● ●● ● ●● ● ●● 8 ●● ● ● CXCR5‾CD25‾
−20 −10 0 10 20 30 CD127-APC -20 0 −40 −20 20 0 20 -20 tSNE_10 20 CXCR5‾CD25+ tSNE 1 tSNE_1 tSNE 1 CD127‾ C D E
TFH–related genes PDCD1 Cell cycle genes 60 15 15 15 8 10 5 4 STMN1
MAF log2(CPM+1) Expression 0 0
(normalized) 0 40 cxcl13nonexpr cxcl13expr 10 MAFSAPMAF ‾ + Mean MAD2L1 MAF expression SH2D1A CXCL13mean.geneCXCL13 PDCD1SAP 0 2 4 6 8 SMC2STMN1STMN1 (SAP)SAP BTLAPDCD1 MAD2L1MAD2L1 PDCD1PDCD1 1.8 18.9 NUSAP1 20 CD200BTLA 5 % of expressing of % cells SMC2SMC2 % of expressing of % cells BTLABTLA TOP2A BCL6CD200 APC NUSAP1NUSAP1 CD200CD200 - BCL6 TOP2A 0 BCL6BCL6 0 TOP2A ‾ + 47.2 32 ‾ + CXCL13
CXCL13 CXCL13 PD-1-BV605 CXCL13 CXCL13
100 ****
TFH signature Cell cycle signature
1.0 NES = 3.3 0.8 NES = 2.3 cells(%) 50
q < 0.0001 + q < 0.0001 1 0.5 - 0.4 RES RES PD 0.0 0 0.0 -0.2 ‾ + -0.2 + CXCL13+ CXCL13‾ CXCL13CXCL13 CXCL13 CXCL13‾ F G 30 r = 0.47 Lunglambrechts cancer (Lambrechts et al.) P < 0.0001 10x Breastsavas cancer (Savas et al.) Melanomali (Li et al.) MARS Melanomaarnon (Arnon et al.) T cells) T Lungguo cancer (Guo et al.) + 15 CD8 seq
- Headpuram and neck cancer (Puram et al.) of of Proliferatingcells
Melanomasade (Sade-Feldman et al.) (%
SMART Colorectalzhang cancer (Zhang et al.)
Liverzheng cancer (Zheng et al.) 0 0 20 40 0 40 80 CXCL13-expressing cells CXCL13-expressing cells (% of CD4+ T cells) (% of CD4+ T cells) H I J *** * * 100 TFH signature 35
0.5 NES = 1.9 q < 0.0001 0.25 ) RES
0.0 cells T -0.1 + 17.5
expressingcells 50 - Post-aPD1 Pre-aPD1 expressingcells CD4 - (% of of (% CXCL13 % of PDCD1 of %
0 0 ‾ + -aPD1 -aPD1 CXCL13 CXCL13 Pre Post 34
Figure 2. Single-cell transcriptomics reveal molecular profile of CXCL13-expressing tumor- infiltrating CD4+ T cells. (A) Sorting strategy for single-cell RNA-seq assays. Live, singlet gated, CD14–
CD19–CD20–CD8–CD56–CD45+CD3+CD4+ lymphocytes from 6 lung tumors were sorted as CXCR5+ (green),
CXCR5–CD25+CD127– (blue) and CXCR5–CD25– (coral) subsets. (B) Seurat clustering of 5317 single cell
TIL transcriptomes identifying 9 clusters (left); each symbol represents a cell; circle delineates CXCL13 cluster. tSNE visualization of cells in B (right); each symbol represents a cell; brown color indicates CXCL13 expression in counts per million (CPM). Pie chart represents the percentage of CXCL13-expressing cells (n
= 955) among all TILs (far right, above) and relative proportions of each of the sorted subsets that express
CXCL13 (far right, below). (C) Percentage (left margin) of CXCL13-non-expressing or CXCL13-expressing cells that express the indicated TFH-related genes. Below, GSEA of TFH signature in the transcriptome of
CXCL13-expressing versus CXCL13-non-expressing cells, presented as in Figure 1F. (D) Violin plots
(above) of expression of PDCD1 in CXCL13-non-expressing or CXCL13-expressing cells; shape represents the distribution of expression among cells and color represents expression (log2(CPM+1)). Flow-cytometric analysis (middle) shows the expression of PD-1 and CXCL13 in live, singlet-gated, CD45+CD3+CD4+ TILs
(n = 6) from patients with NSCLC. Plot (below) shows percentage of PD-1+ cells in CD4+CXCL13- or
CD4+CXCL13+ TILs; **** P < 0.0001 (two-tailed paired Student’s t-test). (E) Percentage (left margin) of
CXCL13-non-expressing or CXCL13-expressing cells that express the indicated cell cycles genes. Below,
GSEA of cell cycle signature in the transcriptome of CXCL13-expressing versus CXCL13-non-expressing cells, presented as in Figure 1F. (F) Percentage of CXCL13-expressing cells (horizontal axis) in tumor- infiltrating CD4+ T cells derived from integrated analysis of 9 single-cell RNA-seq datasets (right, key). (G)
Correlation of frequencies of patient-matched CXCL13-expressing CD4+ TILs with cell cycle genes- expressing CD8+ TILs in the assessed datasets (n = 63) (key as in F); each dot represents a donor; r value indicates the Spearman correlation coefficient. (H) Percentage (left margin) of CXCL13-non-expressing or
CXCL13-expressing CD4+ TILs that express PDCD1 transcripts in the assessed datasets (key as in F); ***
+ P < 0.0005 (two-tailed paired Student’s t-test). (i) GSEA of TFH signature in the transcriptome of CD4 TILs from pre- versus post-therapy tumor samples obtained from patients treated with aPD-1 mAB, presented as in Figure 1F. (J) Percentage (left margin) of CXCL13-expressing cells in CD4+ TILs from matched pre- or
35 post-therapy tumor samples obtained from patients (n = 5) treated with aPD-1 mAB; * P < 0.05 (two-tailed paired Student’s t-test).
36
Figure 3 A B 0.2424 4 CD4+ help-related genes IFNG IL21 Adjusted Adjusted 50 1515 1515 TNFRSF18 10 10 5 5 P TNFRSF4 log2(CPM+1) value ( value 0.1212 2 0 0 0log2(CPM+1) 0 cxcl13nonexpr cxcl13expr cxcl13nonexpr cxcl13expr IFNG TNFRSF4 TNFRSF18 (normalized) Genes(%) -
log) (OX40) (GITR) mean.gene 15 25 IL21 1515 0 2 4 6 15 mean.gene TNFRSF18TNFRSF18 0 2 4 6 10 10 0 0 (GITR)TNFRSF18 TNFRSF4 5 5 TNFRSF4 Expression log2(CPM+1)
TNFRSF4 log2(CPM+1) cells expressing of % cells TNFRSF4 0 0 0 H (OX40) 0 IFNG cxcl13nonexpr cxcl13expr+ cxcl13nonexpr cxcl13expr+ signaling junctions ‾ ‾ IFNGIFNG IL21 CXCL13 CXCL13 CXCL13mean.gene CXCL13mean.gene Cdc42 Signaling-iCOSL Integrin signaling IL21 junction signaling IL21IL21 0 2 4 6 0 2 4 6 mediated apoptosis adherens 0 - -based motility by rho iCOS + Mean expression -mediated phagocytosis ‾ CTL CD28 signaling in T Mitochondrial dysfunction adherens Upregulated 0 2 4 6 CXCL13 CXCL13 receptor γ Adjusted P value Epithelial Fc Regulation of actin Remodeling of epithelial
C
Cytotoxicity-related genes FKBP1A CCL4 Cytotoxicity signature 50 1515 2020 0.6 NES = 2.1 10 15 10 q = 0.0005 5 5 0.3 RES log2(CPM+1) FKBP1A 0log2(CPM+1) 0 0 0 cxcl13nonexpr cxcl13expr cxcl13nonexpr cxcl13expr 0.0 FKBP1AFKBP1A RAB27A ZEB2 RAB27A -0.2 RAB27ARAB27A 1515 mean.gene 1515 mean.gene 25 0 2 4 6 8 0 2 4 6 8 CXCL13+ CXCL13‾ GZMM 10 10 GZMMGZMM CCL4 5 5 log2(CPM+1) CCL4 log2(CPM+1) % of expressing of % cells CCL4 0 0 0 0 cxcl13nonexpr cxcl13expr cxcl13nonexpr cxcl13expr ZEB2ZEB2 GZMM GZMB 1515 mean.gene 1515 mean.gene 0 2 4 6 8 GZMBGZMB 0 2 4 6 8 0 GZMB 10 10 ‾ + 5 5 log2(CPM+1) log2(CPM+1) Expression(normalized) 0 0 0 0 CXCL13 CXCL13 Mean expression cxcl13nonexpr cxcl13expr cxcl13nonexpr cxcl13expr ‾ + ‾ + 0 4 8 CXCL13mean.geneCXCL13 CXCL13mean.geneCXCL13 E 0 2 4 6 8 0 2 4 6 8
D TFH SH2D1A(SAP) CXCR5+ CXCR5‾
2.3 0.1 14.9 2.1 Cytotoxicity-related KLRB1
CD4+help-related IFNG
Cytotoxicity-related APC - GZMB ZEB2 CCL4
CXCL13 17 0.5 54.2 4.6 + GZMB-PE CD4 help-related TNFRSF18 (GITR) TNFRSF4 (OX40)
Cytotoxicity-related FKBP1A CCL3 SOD1 PFN1 PRDM1
Scale Cell cycle STMN1 Bright Field CD4 CXCR5 CD25 CXCL13 GZMB 10 µm TOP2A Invasive margin CDK1 NUSAP1 BIRC5
CXCL13 CD4 CXCL13‾
CXCL13+CXCR5+
CXCL13+CXCR5‾CD25‾
GZMB Merge CXCL13+CXCR5‾CD25+CD127‾
z-score -2 0 2 37
+ Figure 3. Highly functional TFH-like CD4 T cells were CXCR5 negative. (A) Canonical pathways
(horizontal axis; bars in plot) for which CXCL13-expressing TILs show enrichment, presented as the frequency of differentially expressed genes encoding components of each pathway that are upregulated
(key) in CXCL13-expressing TILs relative to their expression in CXCL13-non-expressing TILs (left vertical axis), and adjusted P values (right vertical axis; grey squares; Benjamini-Hochberg test); P < 0.05. (B)
Percentage (left margin) of CXCL13-non-expressing or CXCL13-expressing cells that express the indicated
CD4+ help-related genes (left plot) and violin plots of expression of the same genes in CXCL13-non- expressing or CXCL13-expressing cells (right), presented as in Figure 2D. (C) Percentage (left margin) of
CXCL13-non-expressing or CXCL13-expressing cells that express the indicated cytotoxicity-related genes
(left plot) and violin plots (middle) of expression of the same genes in CXCL13-non-expressing or CXCL13- expressing cells, presented as in Figure 2D. GSEA (right) of cytotoxicity signature in the transcriptome of
CXCL13-expressing versus CXCL13-non-expressing cells, presented as in Figure 1F. (D) Flow-cytometric analysis (top panel) shows expression of CXCL13 and granzyme B in CXCR5+ and CXCR5– subsets in live, singlet-gated, CD45+CD3+CD4+ TILs (n = 9) from patients with NSCLC; numbers in quadrants indicate percentage of CD4+ TILs in each. ImageStream analysis (middle panel) shows expression of CXCL13 and granzyme B in live, singlet-gated, CD3+CD4+CXCR5– TILs (n = 3). Pan-CytoKeratin(CK) (DAB) IHC staining
(bottom panel, left) shows invasive margin (black frame) of tumors (n = 41); scale bar represent 200µm.
PseudoIF image of MxIHC staining (bottom panel, right) shows CXCL13 (red), CD4 (green), granzyme B
(cyan) in region indicated by arrow; scale bars represent 10µm. (E) Expression of transcripts differentially expressed in CXCL13-expressing versus CXCL13-non-expressing TILs, in various sorted subsets (above heatmap, right key). Each column represents the average expression (CPM) in a particular subset. Left margin, vertical colored lines indicate subset in which the genes are differentially expressed. Right margin, examples of key transcripts expressed uniquely or shared by corresponding subsets.
38
Figure 4 A B Invasive margin
Invasive margin and TLS CD8 CD103 GZMB
Invasive margin
CXCL13 CD4 Merge CXCL13 CD4 PanCK CD8+ CD4+ CD103+ + TLS CXCL13 Tumor core
CD8 CD103 GZMB
CXCL13 CD4 PanCK Tumor core CXCL13 CD4 Merge CD8+ CD4+ CD103+ CXCL13+
CXCL13 CD4 PanCK
C
15 Tumor core r = 0.72 100 r = 0.58 P < 0.0001 P < 0.0001
) Invasive Margin 2
(log 10 2 cells) +
cells/mm 5 50 + (% of CD4 of (% + CXCL13
+ 0 CXCL13 CD4
-5 0 -5 0 5 10 -5 0 5 10 + + 2 CD8+CD103+ cells/mm2 (log2) CD8 CD103 cells/mm (log2)
39
+ Figure 4. TFH-like cells infiltrate tumor and associate with CD8 TRM cells. (A) Pan-CK (DAB) IHC staining shows TLS (blue frame), invasive margin (black frame)(above, left image) and tumor core (red frame)(below, left image). PseudoIF image of MxIHC staining shows pan-CK (white), CD4 (green), CXCL13
(red), nuclei (blue) in regions indicated by arrows; scale bars represent 200µm. (B) Pan-CK (DAB) IHC staining shows invasive margin (black frame)(above, left image) and tumor core (red frame)(below, left image); scale bars represent 200µm. PseudoIF image of MxIHC staining shows CD8 (magenta), CD103
(blue), granzyme B (cyan), CXCL13 (red), CD4 (green) in regions indicated by arrows; scale bars represent
10µm. (C) Correlation of the number of CD4+CXCL13+ cells and CD8+CD103+ cells in lung tumors
(quantified by IHC)(left). Correlation of the percentage of CXCL13+ cells in CD4+ cells and the number of
CD8+CD103+ cells in lung tumors (quantified by IHC)(right). Each symbol (key) represents an image; 3 images analyzed per patient (n = 41); r values indicate the Spearman correlation coefficient; P values,
Spearman correlation.
40
Figure 5 A
250 10 * )
B16F10- TEFF/TFH 3 No adoptive transfer 8 OVA adoptive transfer +TEFF +TFH 6 125 4 0 11 13
Days volumeTumor (mm 2 Foldincrease involumetumor 0 0
4 8 11 13 FH EFF +T +T Days B16F10 B
3000 D+4OTII OTII (n=10) (n=10) * )
B16F10- OTII Immunization 3 OVA transfer with OVA in alum D+4OTII OTII, & immunization D+5,7,9 IMM (n=15) 2000 *
1000 * 0 4 5 7 9 20 *
Days volumeTumor (mm
0 10 12 14 16 18 20 Days Days C ** 6 *** 100 ****** 100 * 100 ** T cells) T T cells) T T cells) T + T cells) T + + + + 3 50 50 50 (% of CD8 of (% (% of CD3 of (% (% of CD8 of (% + (% of CD4 of (% + + FH T CD8 KI67 0 0 0 GZMB 0
41
Figure 5. Induction of TFH cells in tumor-bearing mice promotes anti-tumor immunity. (A) Mice were
+ + s.c. inoculated with B16F10-OVA and received adoptive transfer of OT-II CD4 TEFF cells or OT-II CD4 TFH cells at the indicated time point. Tumor volume (left) and fold increase in tumor volume (right) from day 11 to day 13 in mice (n = 4-8/group) treated as indicated (error bars are mean ± SEM); * P < 0.05 (Mann-Whitney test); data representative of two independent experiments. (B,C) Mice were s.c. inoculated with B16F10-
OVA and received adoptive transfer of naïve OT-II CD4+ cells followed by immunization with OVA in alum at the indicated time points. Tumor volume (B) and tumor-infiltrating cell frequencies assessed by flow cytometric analysis (C) in mice (n = 10-15/group) treated as indicated (error bars are mean ± SEM); * P <
0.05, ** P < 0.005, *** P < 0.0005 (Mann-Whitney test); data representative of two independent experiments.
42