First Release, Latest online

A reservoir of stem-like CD8+ T cells in the tumor-draining lymph node preserves the ongoing anti-tumor immune response

Okinawa Flat Belly Tonic Okinawa Flat Belly Tonic

INTRODUCTION

Non-small cell lung cancer (NSCLC) is amongst the deadliest cancers (2), but immune checkpoint inhibitors (ICIs), like anti-PD-1 and anti-PD-L1, have provided durable responses in ~20% of treated NSCLC patients (3). Several parameters correlate positively with response, including the presence of an immunologically “hot” tumor microenvironment (TME; contains infiltrating CD3+ T cells, also called T-cell inflamed), infiltration of PD-1+ CD8+ T cells, PD-L1 immunostaining on tumor and immune cells, and increased tumor mutational burden/neoantigens (47). These observations are in line with the idea that PD-1 blockade potentiates the function of “exhausted” PD-1+ tumor-infiltrating CD8+ T cells in hot tumors (8). By contrast, patients with immunologically “cold” tumors (also called “immune excluded” or “non-T-cell inflamed”) respond poorly to immunotherapy, and the status of their anti-tumor CD8+ T cell response is uncertain (47). As response rates to immunotherapy are low, particularly for patients with cold tumors, a better understanding of the CD8+ T cell biology associated with hot and cold tumors is essential.
CD8+ T cell exhaustion is a progressive process of terminal differentiation (912). Exhausted T cells (TEX) can be characterized by the loss of proliferative potential and effector functions (ability to produce TNFα and IFNγ), as well as increased expression of several inhibitory receptors (e.g., PD-1 and Tim3) and transcription factors (e.g., Blimp-1, Tox, and Eomes) (9, 1230). TEX cells are derived from less differentiated precursors, including PD-1mid CXCR5+ SLAMF6+ TCF1+ “stem-like” T (TSL) cells (12, 21, 24, 3140). TCF1+ TSL cells have at least two functions in chronic immune responses: maintaining the ongoing T cell response and mediating therapeutic responses to PD-1 blockade (24, 31, 33, 34, 41). Expression of TCF1 is necessary for both functions (12, 34), and thus, provides an important tool for identifying TSL cells.
TCF1+ CD8+ T cells are present in tumors, and their presence correlates with better outcomes following immunotherapy (10, 12, 21, 24, 31, 33, 34, 37, 4043). Yet, tumors are rich in signals that promote the exhaustion of CD8+ T cells, like persistent antigen exposure (15, 44). This raises a fundamental question: how are intratumoral TCF1+ CD8+ T cells maintained over the months-to-years of natural tumor development? Moreover, because TCF1 expression is required for maintenance of T cell populations, do immunologically cold tumors result from a loss of intratumoral TCF1+ CD8+ T cells.
The KP (Kraslox-stop-lox (lsl)-G12D/+;p53flox/flox) model is a genetically engineered mouse (GEM) model of cancer that faithfully recapitulates the histological, transcriptomic, epigenomic, and genetic features of a developing human lung adenocarcinoma and has played a fundamental role in our understanding of how human lung cancer develops (4548). Tumors in the KP model can be programmed to express neoantigens, which allows for the investigation of tumor-specific T cell responses in developing tumors (4959). Early neoantigen-expressing KP lung tumors are infiltrated by tumor-specific CD8+ T cells and have an immunologically hot TME. However, as tumors develop, they take on an immunologically cold TME, with T cells being excluded from the tumor parenchyma and restricted to tertiary lymphoid structures (TLS) (49, 60). Thus, the KP model provided us with an opportunity to investigate the differences between T cells from hot and cold tumor microenvironments, and to assess the mechanisms for how tumor-specific CD8+ T cells are maintained over the course of tumor growth.

DISCUSSION

PD-1+ TCF1+ CD8+ T cells are present in human and murine tumors and are necessary to sustain both the anti-tumor T cell response and responses after immunotherapy, but the mechanisms for their maintenance remained unclear. Using an autochthonous model of lung adenocarcinoma, we found that the population of intratumoral PD-1+ TCF1+ CD8+ T cells was maintained by migration from the tumor-draining lymph node (dLN). Most tumor-specific CD8+ T cells in dLNs expressed PD-1, TCF1, and SLAMF6 and had transcriptional patterns that were more similar to canonical TSL cells seen during chronic LCMV infection. By contrast, while some intratumoral CD8+ T cells were PD-1+ and TCF1+, many did not express SLAMF6 or other genes associated with TSL cells. Moreover, as the tumor microenvironment shifted from hot to cold, the intratumoral T cell population became more differentiated, while the population in the dLN was unchanged at the transcriptional and phenotypic levels. These findings, in combination with the shared TCR sequences found between tissues and pseudotime analyses, present convincing evidence for tumor-specific CD8+ T cells in the dLNs being developmentally related to (and likely the developmental precursors of) more differentiated intratumoral T cells in early and late tumors. Future studies will be needed to test this directly. While most studies of anti-tumor CD8+ T cell function and differentiation have focused on tumor tissues (8290), our data demonstrate that the process of differentiation for tumor-specific CD8+ T cell begins in the dLN, with the dLN serving as a reservoir for maintaining T cells in a stem-like state throughout the course of tumor development.
The question of whether patients with cold tumors can respond to immunotherapeutic intervention has remained uncertain. Cold tumors have poor infiltration of T cells and/or T cell exclusion, which is thought to reflect a diminished or absent anti-tumor immune response, consistent with their poor response to checkpoint therapies (4). Yet, cold tumors have similar mutational burdens and antigen presentation capacity as hot tumors, suggesting they both have the capacity to initiate and drive anti-tumor T cell responses (53). These findings are in line with the cold tumors in our model, which maintain neoantigen expression and in vivo presentation of neoantigens (49, 60, 62, 91, 92). Given these observations, we hypothesized that the cold tumor microenvironment was a result of global exhaustion of tumor specific CD8+ T cells throughout the host. Surprisingly, many tumor-specific CD8+ T cells in late tumors were TCF1+, although pseudotime analyses demonstrated that these T cells were more differentiated than T cells from early tumors. These data are consistent with the possibility that the increased differentiation state of intratumoral T cells could account for the cold phenotype of late tumors. Additional possibilities include the inability of migrating T cells to physically enter tumors or defects in DC migration or function (53, 9395). We also found that a cold TME is not indicative of global exhaustion of tumor-specific CD8+ T cells as cold tumors were associated dLNs containing TSL cells that were transcriptionally similar to TSL cells in dLNs associated with hot tumors. Thus, the distal location of dLN TSL cells likely protects them from changes that occur within the TME over the course of tumor development.
While many signals could promote the differentiation of intratumoral CD8+ T cells, TCR signals are prime candidates. TCR signals drive terminal T cell differentiation in chronic infection, and CD8+ T cells that recognize more abundant antigens are subject to more severe exhaustion (15, 25, 27, 33, 44, 54, 7679). Likewise, antigenic peptides that deliver weaker TCR signals are less potent drivers of T cell exhaustion (44). Tcf7 is required to sustain CD8+ T cells during chronic infection (31, 40, 44), but TCR and inflammatory signals promote TCF1 down-regulation (96). We found that intratumoral T cells had high levels of transcripts associated with downstream TCR signaling, while dLN T cells had low expression of these transcripts. These data are consistent with the idea that the dLN may protect TSL cells from persistent antigen exposure. We hypothesize that intratumoral CD8+ T cells are unable to escape persistent antigen and that without migration from the dLN, the pool of tumor-specific CD8+ T cells would become exhausted. Moreover, because T cell clones that recognize tumor antigens with higher avidity (so called “best fit” clones) are more prone to exhaustion, the dLN likely plays an important role in preventing the loss of best fit clones over the course of tumor development. The splenic white pulp may play a similar role during chronic LCMV infection (31, 38), although both the white and red pulps are sites of LCMV Clone 13 infection (97). Our data also raise the question of whether intratumoral niches (like TLS) could exist in tumors to protect TSL cells from differentiation. We previously showed that TLS associated with late tumors in our models were sites for antigen presentation (60), but it remains to be seen whether tumor-proximal niches such as TLS could protect resident T cells from persistent antigen exposure (60, 91, 92).
Our data highlight the critical role of migration in the maintenance of TSL cells in tumors but are less consistent with the idea that T cells differentiate in the lymphoid tissue prior to migration. The latter has been seen in chronic LCMV infection (31, 38) and may be due to the ongoing infection in the tissue. By contrast, we observed that TCF1hi T cells differentiate within tumors, and that migration was required for the presence of TCFhi T cells in tumors. We cannot rule out the possibility that a low number of naïve-like T cells from other extratumoral tissues besides the dLN, such as the spleen or thymus, may contribute to the ongoing anti-tumor immune response over the time course studied, as FTY720 treatment could potentially block migration from these tissues as well. However the concordance of TCR sequences between dLN and tumor suggest that these contributions would be minimal. It is not clear what drives the migration of TSL cells from dLNs to tumors. DC migration from tumors to dLNs is important for priming T cells, but the role of DCs in maintenance of TSL cells in dLNs is uncertain (98, 99). One simple model is that periodic signals from migrating DCs are also necessary for maintaining the migratory T cell population. Critically, while we did not see the accumulation of less-differentiated T cells in dLNs upon FTY720 treatment, it is possible that FTY720 also blocks the migration of antigen-presenting DCs to LNs, which could impact the differentiation of TSL cells in the dLN. Further studies will be needed to test what signals are necessary for maintenance and migration of T cells in dLNs.
The role of the dLN in immunotherapy remains uncertain. Expression of PD-L1 on DCs is important for responses to anti-PD-L1 in some tumor models, and migratory DCs in tumor dLNs express both PD-L1 and the costimulatory receptor B7-2 (100). Moreover, PD-1 blockade can act in dLNs in transplant tumor models (101, 102). Whether PD-1 blockade acts outside the TME in humans is not known, but therapeutic efficacy after anti-PD-1 treatment in patients is associated with changes in immune cell populations in the peripheral blood (103, 104) and with the appearance of new T cell clones in the tumor after therapy (105107). Our analyses of CD8+ T cells from humans showed the presence of TSL-like cells in LNs and tumors. However, as PD-1 blockade functions poorly in patients with cold tumors, this suggests that these patients either lack LN TSL cells or that PD-1 blockade is insufficient to drive therapeutic responses in LNs of these patients. Thus, identifying novel therapeutic strategies directed toward tumor-specific T cells in the dLN may be a means toward improving outcomes for cancer patients with cold tumors.

MATERIALS AND METHODS

Study Design

The aim of this study was to investigate mechanisms by which CD8+ TSL cells are maintained in tumors over the course of cancer progression. We utilized an autochthonous model, as well as an orthotopic transplant mouse model, of lung adenocarcinoma in which tumor cells express the neoantigen GP33 from LCMV. We evaluated the presence of tumor-specific CD8+ TSL cells in various lymphoid and non-lymphoid tissues by FACs using tetramer-specific cell staining. We analyzed single-cell RNA-sequencing to assess the differentiation state and trajectory of the cell subsets present after FACs sorting on endogenous tetramer-specific CD8+ T cells from tumors and draining lymph nodes of tumor-bearing mice, comparing them to tetramer-specific CD8+ T cells from spleens of mice infected with acute or chronic LCMV. In order to determine the clonal relationship of these cells from tumors and draining lymph nodes, we analyzed single-cell TCR sequencing. We tested the effect of blocking lymphocyte migration into tumors after three weeks of FTY720 treatment in autochthonous mice by FACs. Lastly, to assess whether similar phenomena occur in humans, we analyzed a publicly available single cell RNA-sequencing dataset from lymph nodes and lungs of non-small cell lung cancer patients.

Mice

C57BL/6J mice (Jackson Laboratories) were used for all transplant experiments. KP x CCSP-rtTA mice, referred to here as KP mice were obtained from Tyler Jacks lab (50) and crossed to NINJA mice (59) to obtain KP-NINJA (KraslslG12D/+, p53fl/fl, R26-NINJA/NINJA, CCSP-rtTA+) mice. KP (KraslslG12D/+, p53fl/fl, CCSP-rtTA+) mice were used as controls in some cases. 6+ week-old male and female mice were used for all experiments and were sex-matched and age-matched for each individual experiment. All studies were carried out in accordance with procedures approved by the Institutional Animal Care and Use Committees of Yale University. All mice were bred in specific pathogen-free conditions.

Lung Tumor Initiation

Autochthonous tumor generation: KP-NINJA mice were infected intratracheally with 2.5 × 107 PFU Ad5mSPC-Cre (Dr. Anton Berns, Netherlands Cancer Institute), after precipitation with 10mM CaCl2 for 20-60 min, or 5 × 104 PFU Lenti-cre. To induce expression of NINJA neoantigen in infected cells, mice were given doxycycline hyclate chow (625mg/kg; Envigo cat. TD.09628) days 7-11 post infection (p.i.) and concomitantly treated with 4.4mg tamoxifen (MP Biomedicals cat. MP215673894) in corn oil (ThermoFisher Scientific cat. S25271) by gavage on days 8-10 p.i. To induce neoantigen expression via lentivirus, KP mice were infected with 2.5 × 104 PFU mClover-GP33-80-Cre lentivirus and assessed at 8 weeks p.i. Orthotopic KPN1 tumor transplants: Established KPN1 cells were maintained in complete DMEM (10% HI-FBS, 55μM beta-mercaptoethanol, 1x Pen/Strep and 1x L-Glut). Prior to injection, cells were washed 3x with 1xPBS and 200,000 cells were injected intravenously via tail vein injection. Subcutaneous KPN1 transplants: Established KPN1 cells, sorted for GFP+ (NINJA-expressing) cells, were maintained in complete DMEM (10% HI-FBS, 55μM beta-mercaptoethanol, 1x Pen/Strep and 1x L-Glut). Prior to injection, cells were washed 3x with 1xPBS and 500,000 cells were injected s.c. and measured using standard caliper measurements. Tumor volume = (LxW2)/2.

Tissue processing for flow cytometry

Prior to sacrifice, mice were injected retro-orbitally with 200uL anti-CD45-PECF594 in 1X PBS (1:200; BD Biosciences Cat# 562420). After 2-3 min, lungs (or thymus) were harvested at various time points ranging from 8-25 weeks p.i. into Collagenase IV (Worthington Biochemical, cat. LS004189) Buffer (1x HEPES buffer, 0.5mg/mL Collagenase IV, 20μg/mL DNase in 1x HBSS with MgCl2 and CaCl2) and run on the default Lung_01 protocol on a gentleMACS Dissociator instrument (Miltenyi Biotec). Samples were then incubated at 37°C for 30 min and further dissociated with default Lung_02 protocol. Digestion was quenched by adding 500 μL FBS. Bone marrow was collected from femurs and processed into single cell suspensions. Samples were then strained through 70 μm cell strainers, washed with 1% HI-FBS RPMI-1640 (ThermoFisher Scientific cat. 11875085) and red blood cells were lysed using 1x RBC Lysis Buffer (eBioscience, cat. 00-4333-57). Cells were counted using a hemocytometer for absolute number calculations. Lymph nodes (as well as spleens) were concomitantly harvested from tumor-bearing mice, and processed as described in (108). Single cell suspensions were stained using one of two antibody panels (see Flow cytometry section) in addition to tetramer for H2Db/GP33-43-specific CD8+ T cells (NIH Tetramer Core Facility). For intracellular staining, FoxP3/Transcription Factor Staining Buffer set (eBioscience cat# 00-5523-00) was used as per manufacturer’s protocol. Cells were washed and resuspended in FACs Buffer (0.5% FBS, 20% sodium azide in water, PBS 1X without Mg2+/Ca2+) until analysis on a BD LSRII flow cytometer (BD Biosciences).

Ex Vivo IFNγ expression

Single cell suspensions were obtained as described above. The number of cells from draining lymph nodes and tumors was determined using hemocytometer. Samples were then plated in 96-well flat bottom plates at a ratio of 25:75 with CD45.1 splenocytes and stimulated in 10% HI-FBS RPMI-1640 (Thermo Fisher Scientific cat. 11875085) containing Brefeldin A (eBioscience cat. 00-4506-51), and LCMV GP33-41 peptide (AnaSpec cat. AS-61296), or left unstimulated in 10% HI-FBS RPMI-1640 (ThermoFisher Scientific cat. 11875085) containing Brefeldin A (eBioscience cat. 00-4506-51). Plates were incubated for 4-6 hours at 37°C, and samples were then transferred to 96-well round bottom plates. Samples were stained for extracellular markers (see Flow cytometry section), fixed with BD Cytofix/Cytoperm Fixation/Permeabilization Solution kit (BD Biosciences cat. 554714), and stained with anti-IFNγ for intracellular cytokine assessment (see Flow cytometry section) in BD Perm/wash Buffer (BD Biosciences cat. 554714) as per manufacturer’s protocol.

Flow cytometry

Cells were prepared from various tissues and stained with extracellular antibodies in FACs Buffer (0.5% FBS, 20% sodium azide in water, PBS 1X without Mg2+/Ca2+). Staining reagents included PECF594 anti-CD45 (30-F11) and FITC anti-IFNγ (XMG1.2) from BD Biosciences; PERCP anti-CD90.2 (30-H12), BV421 anti-CD279 (PD-1; 29F.1A12), APCFire750 anti-CD90.1 (THY1.2; OX-7), BV605 anti-CD90.2 (30-H12), PECY7 anti-CD366 (TIM3; RMT3-23), BV421 anti-CD279 (29F.1A12), PE anti-SLAMF6 (330-AJ), BV421 anti-CD8α (53-6.7), BV711 anti-CD44 (IM7), APC/Fire750 anti-CD90.1 (THY1.1; IM7), and APC-Cy7 anti-CD45.1 (A20) from Biolegend, FITC anti-CD8α (CT-CD8α), and PeCy5 anti-CD8α (CT-CD8α) from Thermo Fischer Scientific; PE TCF1/7 (C63D9) from Cell Signaling Technologies. H-2D(b) LCMV GP 33-41 tetramer-KAVYNFATM-APC was provided by the NIH tetramer core. Cells were stained at 4°C for 30 min followed by fixation and permeabilization with apropriate intracellular staining kit. For intracellular staining of cytokines, the Cytofix/Cytoperm Fixation/Permeabilization Solution Kit from BD Biosciences was used as per manufacturer’s protocol. For intracellular staining, FoxP3/Transcription Factor Staining Buffer set (eBioscience cat# 00-5523-00) was used as per manufacturer’s protocol. Data were collected on LSRII cytometer (BD Biosciences). For sorting, indicated populations were sorted to >90% purity with FACSAria III cytometer (BD Biosciences).

Histology and IHC staining

Tumor-bearing lungs of KP-NINJA mice were fixed in 1x Formalin solutions in PBS (Millipore-Sigma) for 24 hours at 4°C, switched into 70% ETOH, and submitted to Yale histology core for paraffin embedding, sectioning, and hematoxylin and eosin (H&E) staining. Unstained slides of KP-NINJA autochthonous lung tumors were stained with anti-CD3 (ab5690) using the ImmPACT DAB Peroxidase kit (Vector Labs) for immunohistochemistry. H&E and anti-CD3 IHC stained sections were imaged on a Nikon TE2000 microscope (Micro Video Instruments, Inc. Avon, MA) using a 20x objective.

FTY720 Treatment

KP-NINJA mice were infected intratracheally with 2.5 × 107 PFU Ad5mSPC-Cre (Dr. Anton Berns, Netherlands Cancer Institute) and treated with tamoxifen and doxycycline as previously described. From 6 to 9 weeks following intratracheal infection, mice were treated with 0.3 mg/kg FTY720 or vehicle (saline) i.p. every other day.

Cell line generation

The generation of the KPN1 cell line has been described (62). Briefly, KP-NINJA mice were infected intratracheally with 5 × 104 PFU of lentiviral vector LV-rtta3-Cre. KP-NINJA mice were treated with doxycycline and tamoxifen as described to induce NINJA expression in transformed cells. Tumor-bearing lungs of all mice were harvested 20 weeks p.i., minced with scissors, and rotated at 37 C for 40 min in Collagenase IV Buffer + 2 mg/mL Dispase II (Sigma Aldrich cat. 04942078001). Homogenate was filtered through a cell strainer (Corning cat. 352340) and centrifuged at 200xg for 4 min at room temperature. Pellet was resuspended and cultured at 37°C and 5% CO2 in complete DMEM (DMEM + 10% FBS + 1% P/S), + 1x Gentamicin for the first 2 passages. After 6+ passages fibroblasts were visually undetectable and cell lines were verified to be 100% Kras-transformed by treating with puromycin (unrecombined Kras in this mouse confers puromycin resistance).

LCMV-Clone 13 and -Armstrong infections

For Chronic and acute LCMV infections, 7-10 weeks old C57BL/6 mice were infected intraperitoneally with 2×106 PFU/mouse of LCMV-Clone 13 or LCMV-Armstrong (Figure S3I-K), respectfully. Mice were euthanized 28 days after infection to collect and process spleens as previously described (108).

Sorting and single cell RNA- and TCR- sequencing of GP33-specific CD8+ T cells

KP-NINJA mice were infected with Ad5mSPC-Cre, treated with doxycycline and tamoxifen, and lungs and dLN were harvested 8 and 17 weeks p.i. after i.v. injection of anti-CD45-PECF594 antibody (clone 30-F11, BD Biosciences), as described. Spleens were harvested from C57BL/6 mice 28 days following infection with LCMV-Clone 13 (or LCMV-Armstrong – Figure S3I-K). Tissues were dissociated as previously described and GP33-specific CD8+ T cells were sorted (i.v.CD45CD8+GP33-loaded MHC I tetramer+) and submitted to the Yale Center for Genome Analysis for single-cell RNA and TCR sequencing. Single cell RNA-sequencing data was demultiplexed using Cell Ranger 3.0 Software and then further analyzed using Python. Data represents cells from n=3 pooled at each time point. Pooled GP33-specific endogenous cells from tumors and matched dLNs, as well as from spleens, were submitted for 10X single cell RNA-and TCR-sequencing to Yale Center for Genome Analysis (YCGA).

Motif Analysis

Consensus motifs in grouped CDR3 amino acid sequences were identified using two motif based sequencing analysis tools: Multiple Em for Motif Elicitation (MEME) and Gapped Local Alignment of Motifs (GLAM2) (1). The motif analysis across all four samples (early and late dLN and tumor) was performed separately for TCR alpha and TCR beta chain, including clones with ≥ 2 cells only. At first, fasta files were created separately, including TCR alpha or TCR beta CDR3 amino acid sequences for the clones with ≥ 2 cells, using Biostrings package. These fasta files were used as input files for motif analysis, separately for each chain. Filtration of CDR3 amino acid sequences were performed based on low alignment scores by GLAM2. From there, CDR3 amino acid sequences for each chain were further sub-grouped into separate fasta files based on similarity in sequences and alignment scores. Each sub-group of sequences for each chain were run for motif analysis using GLAM2 function and a position weight matrix as an output to define the motif for each sub group of sequences, either for TCR alpha or beta chain. The contribution of clones from each of the four samples to each motif (either for TCR alpha or TCR beta chain), were traced back using the clone IDs. Following this, consensus motif for each chain was defined as having clones shared by all four samples and ranked in an order based on the number of clones giving rise to each motif (Table S1).

Bioinformatics analysis of GP33-specific CD8+ T cells

Single cell RNA- and TCR-sequencing data from LCMV-Clone 13, LCMV-Armstrong, early and late dLNs and tumors was processed with CellRanger 3.1 using the mm10 mouse genome indices from 10x genomics. Number of cells analyzed and genes detected for each sample: Chronic LCMV(1,185 and11,595), Acute LCMV(10,768 and 12,960), early dLN (1,742 and 12,116), late dLN (876 and 11,595), early tumor (806 and 11,749), and late tumor (731 and 11,150). The libraries were further pre-processed in Python using the scprep package (github.com/krishnaswamylab/scprep). Cells with library size below 1000 UMI/cell and rare genes (genes detected in fewer than 5 cells) were removed. The data was then normalized by library size to 1,000 counts per cell and square-root transformed.
For visualization, PHATE (72) was used to embed the cells into two dimensions based on transcriptional profiles, allowing for visual comparison of global and local similarities between cells. Groups of similar cells were identified by running spectral clustering on our input data. For visualizing gene expression, we imputed missing and dropped-out values with MAGIC and visualized on PHATE (109). A small percentage of the cells were found to have low expression of CD8a after de-noising, and excluded from further analysis. Cell clusters for GP33-specific CD8+ T cells from chronic LCMV Clone 13 infection were visualized using PHATE maps and colored based on clustering into arbitrary 7 clusters. Similar to UMAP projections, the organization of clusters and the relative distances between clusters on PHATE embeddings have meaning (i.e., closely related clusters are located in closer physical proximity).
To analyze the cellular trajectories and infer pseudotime, we used the scVelo stochastic model (75, 110) stochastic model. Pseudotime was computed on the basis of the inferred velocity graph with scVelo.

Single group TCR sequence diversity was calculated using Simpson’s index based on number of clones and number of cells with clonal sequences (shared by 2 or more cells). Number of clonal sequences and number of cells with clonal sequences, respectfully: Early tumor (448 and 767), Early dLN (1098 and 1734), Late tumor (216 and 675), Late dLN (346 and 886). Morisita-Horn index between samples was calculated to compare overlap between samples. These calculations were conducted using R program.

Human CD8+ T cell single cell RNA-sequencing analysis

Single cell data was obtained from Gene Expression Omnibus (GEO) by accession code GSE131907 (81). Data processing, analysis and visualization were conducted using R program with package Seurat (v 3.1.0). Only CD8+ T cells (original labels “CD8 low T”, “Cytotoxic CD8+ T”, “Naive CD8+ T” and “Exhausted CD8+ T”) with tissues from tumor or normal lungs, and metastatic or normal lymph nodes (original labels “tLung”, “nLung”, “mLN” and “nLN”) were used for the analysis. Raw gene counts were log-normalized by Seurat function NormalizeData with parameter normalization.method set to “LogNormalize”. Cell clusters were identified from the normalized data using functions FindNeighbors and FindClusters on top 20 PCs and resolution 0.75. UMAP was used to visualize cell clusters based on top 20 PCs. For the marker gene expression heatmap, relative abundances for each gene were calculated as Z-scaled average of log2(RC+1). Here RC are relative counts calculated by Seurat function NormalizeData with parameter normalization.method set to “RC”.

Statistical analyses

All statistical analyses were performed using Prism V8.3.0 software.

Figure Design

Acknowledgments

We thank Joshi lab members for reviewing the manuscript. We also thank the Yale Cancer Center (P30 CA016359 40), Yale Flow Cytometry Core, Yale Center for Genomics Analysis, and Yale School of Medicine Histology Facility. For Ad5mSPC-Cre we thank Dr. Anton Berns (Netherlands Cancer Institute). We also thank Dr. John Wherry (University of Pennsylvania) for the generous gift of LCMV clone 13. Funding: This work was supported by grants from the NCI K22CA200912 (N.S.J.), Young Investigator Award- Melanoma Research Alliance (N.S.J.), Career Enhancement Award from Yale SPORE in lung cancer 1P50CA196530 (N.S.J.), a grant from the Lung Cancer Research Foundation (LCRF) (N.S.J.), NCI 1RO1CA237037-01A1 (N.S.J.), an American Lung Association Discovery Award (N.S.J.), the Yale Cancer Center Leslie Warner Postdoctoral Fellowship (K.A.C.), the Interdisciplinary Immunology Training Program NIH AI07019 (K.A.C.), AI125741 (W.C.), AI148403 (W.C.), American Cancer Society Research Scholar Grant (W.C.), Novo Nordisk grant NNF20OC0063436 (S.K.), and HIPC NIH grant AI089992 (S.K.). M.Y.K. is a member of the Medical Scientist Training Program at the Medical College of Wisconsin, which is partially supported by a training grant from NIGMS (T32-GM080202). This work was also funded in part by the NHLBI-funded postdoctoral fellowship: T32 HL007974 (G.F.). GF is a PhD Student in the Investigative Medicine Program at Yale which is supported by CTSA Grant Number UL1 TR001863 from the National Center for Advancing Translational Science (NCATS), a component of the NIH. Author contributions: K.A.C. and N.S.J. conceptualized study. K.A.C. designed and analyzed experiments. I.W., B.F., E.F., I.M., A.H., J.F.C., G.G.F, D.L.M, M.D., N.H. and M.N. performed and analyzed experiments. N.H. provided blinded tissue infiltration scores. M.K., A.V., and S.K. performed formal bioinformatics analyses and visualizations and S.K. supervised these analyses. A.K. and M.Y.K. edited manuscript and performed motif analysis. W.C. supervised motif analyses. J.W. performed statistical analyses of TCR diversity. C.C. and J.W. performed bioinformatics analysis of human dataset GEO: GSE131907 and H.Z. supervised these analyses. N.S.J. supervised, designed, and funded this study. K.A.C. and N.S.J. prepared initial draft. All authors critically reviewed the paper and agreed on the final form. Competing interests: S.K. is a paid advisor for ImmuneAI. This relationship did not influence the work performed in this study. The other authors declare that they have no competing interests. Data and materials availability: RNA- and TCR-seq data from this study are deposited in GEO under accession number GSE182509. Ad5mSPC-Cre was obtained from Dr. Anton Berns at the Netherlands Cancer Institute ([email protected]). All data needed to evaluate the conclusions in this paper are present in the paper or the Supplemental Materials.

Source link

Keravita Pro-To Support Good Nails and Hair Keravita Pro-To Support Good Nails and Hair

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button