Potentiality of multiple modalities for single-cell analyses to evaluate the tumor microenvironment in clinical specimens

ABSTRACT Single-cell level analysis is powerful tool to assess the heterogeneity of cellular components in tumor microenvironments (TME). In this study, we investigated immune-profiles using

the single-cell analyses of endoscopically- or surgically-resected tumors, and peripheral blood mononuclear cells from gastric cancer patients. Furthermore, we technically characterized two

distinct platforms of the single-cell analysis; RNA-seq-based analysis (scRNA-seq), and mass cytometry-based analysis (CyTOF), both of which are broadly embraced technologies. Our study

revealed that the scRNA-seq analysis could cover a broader range of immune cells of TME in the biopsy-resected small samples of tumors, detecting even small subgroups of B cells or Treg

cells in the tumors, although CyTOF could distinguish the specific populations in more depth. These findings demonstrate that scRNA-seq analysis is a highly-feasible platform for elucidating

the complexity of TME in small biopsy tumors, which would provide a novel strategies to overcome a therapeutic difficulties against cancer heterogeneity in TME. SIMILAR CONTENT BEING VIEWED

BY OTHERS SELECTION OF RNA-BASED EVALUATION METHODS FOR TUMOR MICROENVIRONMENT BY COMPARING WITH HISTOCHEMICAL AND FLOW CYTOMETRIC ANALYSES IN GASTRIC CANCER Article Open access 20 May 2022

INTEGRATED SINGLE-CELL AND BULK RNA-SEQ ANALYSIS IDENTIFIES A PROGNOSTIC T-CELL SIGNATURE IN COLORECTAL CANCER Article Open access 30 August 2024 DEVELOPMENT OF A CD8+ T CELL ASSOCIATED

SIGNATURE FOR PREDICTING THE PROGNOSIS AND IMMUNOLOGICAL CHARACTERISTICS OF GASTRIC CANCER BY INTEGRATING SINGLE-CELL AND BULK RNA-SEQUENCING Article Open access 24 February 2024

INTRODUCTION The complexity of tumors arises from various cellular components comprising the tumor microenvironment (TME); these include cancer cells, immune cells, fibroblasts, blood

vessels, and the extracellular matrix1. Each cellular component also internally exhibits heterogeneous profiles with distinct morphology and phenotype, and the intra-tumor heterogeneity

makes tumor contexture more complicated. Tumor heterogeneity, the diversity of cancer cell types in the tumor microenvironment, has recently attracted attention as a burgeoning research

area, which is multi-directionally approached on the basis of cellular morphology, gene expression, metabolism, motility, proliferation, and metastatic potential. The interplay between the

heterogeneous cancer cells with their microenvironments appears to play an important role in not only tumor development, but also therapeutic response/resistance to anticancer drugs2,3.

Tumor-infiltrating lymphocytes (TILs) in TME, for instance, have a prominent role in determining the antitumor activity of immune checkpoint blockades (ICBs) such as anti-PD1 and anti-PD-L1

antibodies4,5,6,7,8,9. As for cancer cells, on the contrary, presentation of neo-antigens and/or PD-L1 expression on the surface of cancer cells apparently contribute to the efficacy of

ICBs10,11. Given that cancer cells exhibit heterogeneous expression levels of these factors, the use of these factors as biomarkers requires further improvements for efficacious predictive

precision12,13,14. To capture a view of tumor complexity more comprehensively and panoramically, single cell analysis, rather than a conventional bulk analysis, is expected to be a powerful

tool since this methodology could profile small heterogeneous populations in the tumor microenvironment (TME)15,16. For decades, flow cytometry is the most widely-used methodology to analyze

single cells, especially immune cells. A novel research platform of flow cytometry equipped with mass spectrometry, termed mass cytometry (CyTOF), has been recently developed. CyTOF could

detect single-cell resolution using ~ 40 simultaneous cellular parameters, which evaluates the complexity of cellular systems and processes17,18,19. Although single-cell analysis with CyTOF

has been well-established, especially for the analysis of peripheral blood mononuclear cells (PBMCs)20,21, it also possesses technical difficulties to overcome. Firstly, analysis for TIL in

a small biopsy sample is technically challenging. Secondly, the number of measured parameters is limited (~ 40 parameters), indicating that only focused cell populations can be detected by

CyTOF18. On the contrary, single-cell RNA-sequencing (scRNA-seq) is based on the expression level of the entire gene in individual cells and is expected to cover various biological pathways

comprehensively22,23. Recently, a droplet-based scRNA-seq successfully characterized various clusters of immune cells based on the gene expression profile; this methodology is gaining

popularity for widespread use24,25. In this study, we investigated immune-profiles using the single-cell analyses of clinical specimens, endoscopically- or surgically-resected tumors, as

well as PBMCs from gastroenterological cancer patients. Furthermore, we technically characterized two distinct platforms of the single-cell analysis, 10 × Genomics Chromium Single cell 3′

v2-based scRNA-seq analysis and Fluidigm Helios-based CyTOF analysis, both of which are broadly embraced technologies for single-cell level analysis. Using the two-distinct methodologies, we

demonstrated that single-cell analysis is a powerful tool for classifying the cell types in clinical specimens as well as to understand the complexity of the TME. RESULTS EVALUATION OF TILS

IN TME BY CYTOF ANALYSIS, COMPARING FRESH AND FROZEN TUMORS We first conducted single-cell analysis in Fluidigm Helios-based CyTOF to evaluate and compare TILs isolated from

freshly-prepared and frozen-stocked tumor samples. For this analysis, surgically-resected gastrointestinal cancer specimens were used (Sup. Table S1 and S2: patient). As shown in Fig. 1a, we

detected significantly higher cell numbers, as well as higher numbers of marker-positive cells, from the fresh samples than the frozen samples (Fig. 1b; upper panel). Similar to the

surgically-resected samples, the endoscopically-resected biopsy samples (Sup Table S1) also exhibited higher cell numbers (Fig. 1a) and more marker-positive cells (Fig. 1b; lower panel) in

the fresh samples than the frozen samples, revealing that CyTOF analysis of freshly-prepared samples worked better for both surgically-resected and endoscopically-resected specimens. These

results indicate that freshly-prepared tumors would be preferred for single-cell analyses to evaluate TILs in TME, especially for analyses of small pieces of the biopsy samples. EVALUATION

OF TWO DISTINCT PLATFORMS FOR SINGLE-CELL ANALYSIS IN PBMCS: SCRNA-SEQ AND CYTOF We evaluated two distinct platforms for single-cell analysis: Chromium v2 (10X Genomics) for scRNA-seq and

Fluidigm Helios-based CyTOF using the same PBMC sample sets. R package Seurat for scRNA-seq and CytoBank for CyTOF were used to classify PBMC populations based on the indicated markers

(Figs. 2a, S1, S2 and Sup Table S3 and S4) 26,27. Both scRNA-seq and CyTOF clearly distinguished the classified immune-cell types such as T cells, NK cells, B cells, and other immune-related

cells in PBMCs. However, scRNA-seq exhibited less accuracy to identify the differences between T cells and NK cells (Fig. 2a, S1, S2 and Sup Table S4). We also compared the abilities of

scRNA-seq and CyTOF single-cell analysis to detect % T + NK cell, % B cell, and % myeloid cell in CD45+ immune cell populations, revealing similar proficiencies between these platforms. The

values of coefficient of determination (R2) were 0.86 in T + NK cells, 0.87 in B cells, and 0.83 in myeloid cells (Fig. 2b). However, one sample (gc_007) showed poor agreement between the

techniques, which could be due to a technical error in the recovery from the frozen stock, since cell viability of this sample was extremely low (Fig. S3a). COMPARISON OF SINGLE-CELL

ANALYSES BETWEEN SCRNA-SEQ AND CYTOF IN THE ENDOSCOPICALLY-RESECTED BIOPSY SAMPLES Compared to the surgically-resected tumor samples, the endoscopically-resected tumor biopsy samples were

much smaller. Thus, to utilize these small numbers of cells in the biopsies as much as possible, we subjected the whole biopsy tumors with no sorting to the single-cell analyses for

scRNA-seq and CyTOF. In this manner, we expected to mitigate the loss of cell numbers that occurs when using small pieces of the biopsies. The unsorted whole tumors in the biopsies should

include not only immune cells, but also non-immune cells, such as epithelial cells, fibroblasts, endothelial cells, and other TME-component cells. As a preliminary study, we first performed

scRNA-seq analyses using this unsorted methodology on the surgically-resected tumors (Fig. 3a). Freshly-prepared tumors were used for the analyses, and the experiments were quickly initiated

within 30 min after surgical resection. The single cells in the tumors were isolated by treatment with the indicated dissociation enzyme and then directly, without the sorting step,

subjected to Chromium v2 scRNA-seq analyses. The gene expression analyses of the scRNA-seq by Seurat clearly identified a number of clusters of the TME components in the surgical tumors—NK +

 T cells, B cells, plasma cells, myeloid cells, epithelial cells, and fibroblast + endothelial cells (Fig. 3b)—revealing that this scRNA-seq with “no sorting” protocol could clearly detect

the clusters of TIL components even though the unsorted tumors would contain various non-immune cells (Figs. 3b, Fig. S4, and Table S4). Next, we conducted scRNA-seq analyses of the

endoscopically-resected tumor biopsies. For this study, we also used freshly-prepared biopsies with the “unsorted protocol” as established above. As shown in Fig. 3c, the gene expression

analyses by Seurat clearly identified a number of clusters of the TME components in the small biopsy samples as well, namely NK + T cells, B cells, plasma cells, myeloid cells, epithelial

cells, and fibroblast + endothelial cells. Taken together, the single-cell isolation protocol in unsorted tumors technically works to evaluate the clusters of TIL components in the

endoscopically-resected tumor biopsies. We also evaluated concordance between the scRNA-seq and CyTOF results in distinguishing % T + NK cells, % B cells, and % myeloid cells in the

surgically-resected tumors (Fig. 3e). The values of coefficient of determination (R2) were 0.99 in T + NK cells, 0.77 in B cells, and 0.60 in myeloid cells. Although the R2 was relatively

low in B and myeloid cells, the correlative results between the two platforms were broadly confirmed in tumors as well as PBMCs (Figs. 2 and 3). Interestingly, scRNA-seq identified the

plasma cells, a CD45-CD138+IGKC+CD20- cluster, while CyTOF did not due to lack of B cell markers in our preset CyTOF panel (only CD20 was represented). Given that the markers of CyTOF must

be selected for the target(s) prior to the experiments, this may be a technical limitation of CyTOF to cover a broader range of (unbiased) heterogeneous populations. COMPARISON BETWEEN

SCRNA-SEQ DATA AND IHC OR CYTOF DATA Next, we evaluated the concordance between scRNA-seq and IHC data. The location of sampling area for CyTOF and scRNA-seq were marked, and the marked

blocks were stained for IHC (Fig. 4a). We analyzed the ratio of immune cells/epithelial cells by IHC staining of CD45 and pan-cytokeratin and compared them with the scRNA-seq data (Fig. 4b).

While a similar tendency was observed, the immune cell ratio was remarkably high in scRNA-seq data compared with that in IHC data, indicating that a considerable number of epithelial cells,

which include cancer cells, can be damaged and lost during the procedure. Next, to evaluate further detailed concordance, we compared scRNA-seq data with CyTOF data, focusing on immune

cells and observed poor concordances in some samples (Fig. S3b). Populations of B cell and myeloid cell had poor concordances as well (Fig. 3e). These findings suggest that there are the

technical limitations of the experimental procedure in addition to data analyses, and its influence on the analysis of tumor tissue is greater as compared with PBMC (Fig. S3). As with the

result by scRNA-seq data (Fig. 3b), CD138+CD79a+CD20- plasma cells were also identified by IHC (Fig. 4c). Tumor-infiltrating B cells highly expressed HLA class II and CD40 (Fig. 4d), were

also identified, suggesting these cells play as antigen-presenting cells. A fraction of B cells expressed PDCD1, IL10, and TGFB1. In addition, PRDM1 was also expressed, similarly to plasma

cells (Fig. 4d). These suggest that regulatory B cells infiltrated into the TME28. We also found this plasma-cell population in biopsy samples by scRNA-seq (Fig. 3c). Taken together, our

established scRNA-seq technique enabled us to find novel cell populations, although there were some technical limitations. CLASSIFICATION OF REGULATORY T CELLS (TREG CELLS) IN TILS To

investigate the detail of Treg function, T + NK-cell scRNA-seq datasets were re-analyzed (Fig. 5a). Treg datasets were extracted based on FOXP3 and IL2RA gene expressions among CD4+ T cells

(Fig. 5b). As in Fig. 5b, we classified the Treg cell into 6 clusters. Top 10 genes (average log2 fold change > 0.5 compare to other clusters and adjusted p-value < 0.05) are

summarized in Sup Table S6. Cluster 0 and 4 (35.5% ± 4.7%) were characterized by CTLA4, TNFRSF4, TNFRSF18, and TIGIT, which are highly expressed by activated Treg cells in general (Fig. 

5c)29. By contrast, these molecules were low, and MKI67 were highly expressed by cluster 5 cells (7.3% ± 4.9%), which were considered as proliferative Treg cells (Fig. 5c). Treg cells in

cluster 1 (24.1% ± 4.9%) expressed KLRB1 (CD161) and CCL20, which are specific for Th17 (Fig. 5c). Accordingly, CCR6, a receptor of CCL20, and IL17A were also highly expressed by this

cluster (Fig. 5c). Thus, this cluster indicated CD161+ Th17-like Treg cells produce proinflammatory cytokines. CCR7 and SELL (CD62L) were highly expressed by cluster 2 cells (19.0% ± 7.1%),

which were considered as naïve Treg cells (Fig. 5c). Cluster 3-specific molecules, including IFI44L, STAT1, ISG15, and IFITM1, are generally induced by interferon response, suggesting that

cluster 3 cells (14.1% ± 5.1%) are interferon-related Treg cell (Sup Table. S6). STAT4 did not express by all clusters, including cluster 3, as previously reported (Fig. S6)30. DISCUSSION In

this study, we evaluated the two distinct single-cell-based platforms, scRNA-seq and CyTOF, for single-cell level immune profiling using surgery-/biopsy-resected tumors and PBMC from

gastric cancer patients. The number of previous studies focusing on gastric cancer biopsy in single-cell level is limited31,32,33. Our study revealed that the scRNA-seq analysis could cover

a broader range of immune cells of TME in the endoscopically-resected small biopsy tumor samples, detecting even small subgroups of B cells or Treg cells in the tumors. These findings

demonstrate that scRNA-seq analysis is a highly-feasible platform for elucidating the complexity of the tumor microenvironments in small biopsy tumors. Recent advances in single-cell level

analysis, such as scRNA-seq and CyTOF, allowed us to investigate the complexity of heterogeneity in TME in more detail, detecting small heterogeneous populations (e.g., TILs) at a high

dimensional level34,35; however, each of these platforms has strengths and weaknesses in their technical performance36,37. CyTOF, which has an advantage in higher throughput compared to

scRNA-seq, could detect the targeting immune cell subsets more clearly using the ~ 40 selected antigen markers19,38. However, since these markers must be selected for target detection prior

to experiments, the parameters that CyTOF can measure are technically limited to cover a broader range of heterogeneous populations18. In addition, there is the technical challenge of

generating marker antibodies conjugated with metal-isotopes, which also results in narrowing down the populations that CyTOF can detect. On the contrary, scRNA-seq is a transcriptome-based

platform, which detects wider unbiased-populations without marker selection. However, one limitation of droplet-based scRNA-seq is that its transcription-based data are relatively shallow

and sparse39,40; the numbers of transcriptomes in every single cell are in the range of several thousands. Our study also demonstrated the features of these two distinct platforms clearly:

CyTOF is “narrow and clear,” whereas scRNA-seq is “wide and indistinct.” CyTOF distinguished NK cells and T cells more accurately, while scRNA-seq detected plasma cells that CyTOF markers

did not cover (Figs. 2 and 3). Taken together, we need to choose the best platform of single-cell analysis for the purpose and/or the combination of multiple platforms to compensate for

their individual limitations. We also optimized and established the protocols of scRNA-seq analysis for small clinical specimens of the biopsy-resected tumors. The same procedure used for

the surgically-resected tumors could be applied to the protocols for the endoscopically-resected small tumor biopsies as well. Keys to success for single-cell analysis in the small biopsies

appear to be (1) to use freshly-prepared tumor biopsies, and (2) to run the unsorted cells for the analysis. We initiated the single cell isolation as quickly as possible (~ 30 min) after

receiving the tumor samples. The “no sorting” technique appears to mitigate physical damage to the primary cells, thus making it more suitable for analyses of small numbers of cells. The

optimized protocols allowed us to identify small heterogeneous populations of Treg in TME, suggesting the scRNA-seq analysis appears to be a highly-feasible platform to understand the

complexity of the tumor microenvironments in the small biopsy tumors (Fig. 5). A number of scRNA-seq studies in the surgery-resected clinical specimens have been reported, providing a large

amount of comprehensive information to deeply understand the clinically-relevant cancer biology, invasion, metastasis, and cancer evolution41. However, since the surgery-resected tumors are

basically at the earlier stages of tumor development, the information obtained from the surgery-resected tumors may be restricted to the earlier-stage biological events of tumorigenesis,

presumably missing the later-stage events. The small pieces of biopsy-resected tumors, on the contrary, could be technically collected from the later-stage cancer patients; thus, the

single-cell analysis using these biopsy-resected tumors is expected to cover the valuable information from later-stage cancer. In addition, if the biopsy samples could be collected before

and after the drug treatment in the same patients, the single-cell analysis with these pre-/post-treatment specimens should provide a great advantage for a deep understanding of the

mechanisms of actions and/or biomarker development for cancer therapeutic drugs41,42,43. In fact, the single-cell analyses in the biopsy samples, including the scRNA-seq analysis in gastric

cancer, have received a lot of attention in recent years, while the number of reports is extremely limited31,32,33. Although several technical challenges still need to be cleared for

scRNA-seq of the biopsy-resected tumors, especially the cell isolation steps to miss certain cell populations by “bottle-neck effects”33,44, this research platform of scRNA-seq should have

potential to open the doors to the new generation of cancer biology, overcoming a number of difficulties that currently-used conventional methodologies are facing. In conclusion, we

demonstrated two-distinct methodologies for single-cell analysis as powerful tools to clarify the subpopulations of clinical specimens. Although deeper analysis is required, the methods of

the single-cell analysis showed the potential to identify various cell populations, which could not be identified by other modalities providing novel insights into the tumor

microenvironment. Taken with the technical advantages for each methodology, the single-cell analysis would be more powerful tools to understand the complexity of the TME. METHODS PATIENTS

Patients with gastrointestinal cancer, who underwent surgical resection or endoscopic biopsy at National Cancer Center Hospital East in 2017, were enrolled in this study (Sup Table S1 and

S2). All patients provided written informed consent before sampling, according to the Declaration of Helsinki. This study was performed in a blinded manner and was approved by the National

Cancer Center Ethics Committee. THE PROCEDURE OF SAMPLE PROCESSING TUMOR SAMPLE PROCESSING Single fraction (~ 5 mm square size) from surgically-resected tumor samples or three fractions (~ 3

 mm square size for each) from endoscopically-resected tumor biopsy samples were subjected to CyTOF and scRNA-seq analyses. The freshly resected surgery or biopsy samples were kept in ~ 3 ml

of cold saline solution, to start single-cell isolation in ~ 30 min after the tumor dissection. The tumor samples were substantially minced with a surgical blade and scissor into small

pieces, and then isolated into single cells using gentleMACS Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany) at 37 °C for 30 min digesting with ~ 5 ml of enzyme mixture, which

includes 4 ml of DMEM medium with 10% FBS (DMEM/FBS), 1 ml of Collagenase P (final conc. 2 mg/ml, cat# 11213865001, Merck KGaA, Darmstadt, Germany) or Dispase (final conc. 2.5 mg/ml, cat#

4942078001, Merck KGaA, Darmstadt, Germany), and 50 μl of DNase I (final conc. 0.1 mg/ml, Qiagen, Venlo, Netherlands). The digested tumors were filtrated through 40 μm and 100 μm nylon mesh

to remove cell aggregates, and cell viability was determined by microscopy (> 80% viability is preferable). Collecting the cell pellets by spin-down at 300 g for 10 min at 4 °C, the cells

were suspended with 1 ml of Red Blood Cell Lysis Solution (cat# 130-094-183, Miltenyi Biotec B.V. & Co. KG, Bergisch Gladbach, Germany). After incubation for 2 min at 4 °C, the cells

were suspended with 20 ml of DMEM/FBS, spinning down the cell pellets at 300 g for 5 min at 4 °C. The cells were substantially suspended in 1 ml of PBS with 10% FBS (PBS/FBS) and were

filtrated through 40-μm nylon mesh. The resuspended cells with PBS/FBS at 1 × 106 cell/ml were subjected to the single-cell analysis. The remaining portion of the isolated tumors were

stocked in CELLBANKER (cat# CB011, Nippon Zenyaku Kogyo, Tokyo, Japan) according to the manufacturer’s instruction, which were used as “frozen samples”. PBMC SAMPLE PROCESSING PBMC isolation

by density gradient centrifugation with Ficoll-Paque was performed according to the manufacturer’s instruction (cat#17-1440-03, GE Healthcare Bio-Science AB, Uppsala, Sweden). Briefly, 4 ml

of blood samples were carefully layered on to 3 ml of Ficoll-Paque media to the centrifuge tube, and then centrifuged at 400×_g_ for 30 min at room temperature. The PBMCs were collected

from the interface layer. After washing with DMEM/FBS, PBMCs were suspended in 1 ml of PBS/FBS and were filtrated through 40-um nylon mesh. The resuspended cells with PBS/FBS at 1 × 106

cell/ml were subjected to the single-cell analysis. IHC Surgically resected samples were formalin-fixed, paraffin-embedded, and the blocks which we marked before sampling for CyTOF and

scRNA-seq were sectioned onto slides for IHC, which was conducted using monoclonal antibodies against CD20 (L26, Roche, Basel, Switzerland), CD45 (2B11 + PD7/26, DAKO, Agilent Technologies,

Santa Clara, CA the USA), pan-cytokeratin (AE1, AE3, PCK26, Roche, Basel, Switzerland), CD79a (SP18, Roche, Basel, Switzerland), and CD138 (M115, DAKO, Agilent Technologies, Santa Clara, CA

USA). CD45 and pan-cytokeratin staining were counted in five high-power microscopic fields (× 400; 0.0625 mm2), and their averages were calculated. Two researchers (Y.T. and T.K.)

independently evaluated the stained slides. CYTOF PROCEDURE CyTOF staining and analysis were performed as described20. The antibodies used in CyTOF analyses are summarized in Table S3. The

cells were subjected to staining after washing with PBS supplemented with 2% fetal calf serum (FCS, Biosera, Orange, CA, USA) (washing solution) followed by incubation in 5 μM of Cell-ID

rhodium solution (Fluidigm, South San Francisco, CA, USA) in PBS, washed using the washing solution, and stained with a mixture of surface antibodies. After washing, the cells were fixed and

permeabilized using Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions. The fixed and permeabilized cells

were stained with intracellular antibodies. After washing twice, the cells were incubated overnight in 125 nM MaxPar Intercalator-Ir (Fluidigm) diluted in 2% paraformaldehyde PBS solution at

4 °C. The cells were then washed once with the washing solution and twice with MaxPar water (Fluidigm), distilled water with minimal heavy element contamination, to reduce the background

level. The cells were then suspended in MaxPar water supplemented with 10% EQ. Four Element Calibration Beads (Fluidigm) were applied to the Helios instrument (Fluidigm), and data were

acquired at speed below 300 events/s. CYTOF DATA PROCESSING Using Cytobank45, a manual-gating scheme was processed to remove doublet cells, dead cells, and beads. After the cleanup

processes, multidimensional data were clustered using R package FlowSOM46 and reduced dimension using R package Rtsne. After the visualization, cells were annotated by the expression of the

following representative cell surface markers; T cell (CD3+ and CD8a+, or CD3+ and CD4+), B cell (CD19+), NK cell (CD56+), and myeloid cell (CD11b+ or CD11c+). SCRNA-SEQ PROCEDURE Samples

were processed using the Chromium Single Cell 3′ Solution v2 chemistry (10 × Genomics, CA, USA) as per the manufacturer’s recommendations24. Briefly, cell suspension is resuspendeed at 1 × 

106 cells per ml. To generate GEMs, master mix with cell suspension, gel beads and partioning oils are loaded on Chromium Chip. GEM-RT reaction, cDNA amplification, gene expression library

generation were followed using Chromium kits and reagents. After library generation, sequencing was performed using Illumina HiSeq 2500 Rapid run with 98-bp pair-end reads. Using Cell Ranger

(version 2.0, 10 × Genomics), the fastq files were generated from the bcl files. The sequence reads were aligned to UCSC hg38 and UMIs (Unique Molecular Identifiers) were counted for each

gene in each cell barcode using Cell Ranger count (option: –expect_cells = 6000). Then, the data were polished by R package Seurat as below26,27. SCRNA-SEQ DATA PROCESSING FOR PBMC SAMPLES

Using Cell Ranger output files, barcodes.tsv, genes.tsv, and matrix.mtx, Seurat objects were created. Cells with UMI < 1500, expressing < 100 genes, and with > 10% mitochondrial

genes were removed from PBMC datasets using Seurat v2.3.4. Then, the UMI counts were normalized and scaled. Clustering and 2D projection by t-distributed Stochastic Neighbor Embedding

(t-SNE), was also performed after dimensional reduction using the first 10 Principal Components (PCs). For the feature plot, the dataset was updated and plotted using Seurat v3. The number

of cells after the process was shown in Table S5. Cells were annotated with cell types by the expression levels of 40 markers in Table S4. SCRNA-SEQ DATA PROCESSING FOR TIL SAMPLES Using

Cell Ranger output files, barcodes.tsv, genes.tsv, and matrix.mtx, Seurat objects were created. The number of cells with UMI ≥ 1500 were counted using the gene-cell matrix of Cell Ranger. To

extract the data of single cells with UMI ≥ 1500, Cell Ranger count was re-conducted with the option “–force_cells” with the number of cells with UMIs ≥ 1500 in each sample. Then, the data

was aggregated using Cell Ranger aggr with the option “normalize = none” for combining the data from the same patients. Using Seurat v2.3.4, cells with > 10% mitochondrial genes and

expressing < 500 genes were discarded from the datasets. The total reads and the number of detected cells were shown in Table S7. Then, the UMI counts were normalized and scaled.

Clustering and 2D projection by t-SNE were also performed after dimensional reduction using the first 20 PCs using Seurat. Using representative marker genes of each cell type in Table S4,

the cell clusters were annotated22. As gc_003 differs from other samples in total reads, it was removed for the analysis of Figs. 4d and 5. For the analysis of Fig. 4d, the cell clusters,

which were defined as B or plasma cells, were extracted from four patients and individually gathered into B-cell and plasma-cell groups. Then, the expression levels of the representative

B-cell and plasma-cell related genes were plotted using Seurat v3. For the analysis of Fig. 5, the cell clusters, which were defined as regulatory T cells, were extracted from the datasets

of each patient. Briefly, T + NK clusters were first extracted from the TIL datasets and the T + NK cells were re-clustered after cell cycle regression and dimensional reduction using the

first eight PCs. Then, clusters were extracted according to the expression of CD3, CD4, and FOXP3 as Tregs. The extracted Treg clusters of all the cases were combined using the first eight

PCs by performing Seurat RunMultiCCA and AlignSubspace. The Tregs were re-clustered into six sub-clusters (clusters 0–5). For each cluster, the top 10 marker genes were identified, as shown

in Table S6. DATA AVAILABILITY The datasets generated and/or analyzed during the current study will be available on Database of National Biosceience Database Center before this manuscript is

published. REFERENCES * Balkwill, F. R., Capasso, M. & Hagemann, T. The tumor microenvironment at a glance. _J. Cell Sci._ 125, 5591–5596 (2012). Article  CAS  PubMed  Google Scholar  *

Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. _Cell_ 144, 646–674 (2011). Article  CAS  PubMed  Google Scholar  * Meacham, C. E. & Morrison, S. J. Tumour

heterogeneity and cancer cell plasticity. _Nature_ 501, 328–337 (2013). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Zou, W., Wolchok, J. D. & Chen, L. PD-L1 (B7–H1) and

PD-1 pathway blockade for cancer therapy: Mechanisms, response biomarkers, and combinations. _Sci. Transl. Med._ 8, 3284 (2016). Article  CAS  Google Scholar  * Pardoll, D. M. The blockade

of immune checkpoints in cancer immunotherapy. _Nat. Rev. Cancer_ 12, 252–264 (2012). Article  CAS  PubMed  PubMed Central  Google Scholar  * Sharma, P. & Allison, J. P. The future of

immune checkpoint therapy. _Science_ 348, 56–61 (2015). Article  ADS  CAS  PubMed  Google Scholar  * Hodi, F. S. _et al._ Improved survival with ipilimumab in patients with metastatic

melanoma. _N. Engl. J. Med._ 363, 711–723 (2010). Article  CAS  PubMed  PubMed Central  Google Scholar  * Topalian, S. L. _et al._ Safety, activity, and immune correlates of anti-PD-1

antibody in cancer. _N. Engl. J. Med._ 366, 2443–2454 (2012). Article  CAS  PubMed  PubMed Central  Google Scholar  * Brahmer, J. R. _et al._ Safety and activity of anti-PD-L1 antibody in

patients with advanced cancer. _N. Engl. J. Med._ 366, 2455–2465 (2012). Article  CAS  PubMed  PubMed Central  Google Scholar  * Rizvi, N. A. _et al._ Cancer immunology: Mutational landscape

determines sensitivity to PD-1 blockade in non-small cell lung cancer. _Science_ 348, 124–128 (2015). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Reck, M. _et al._

Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer. _N. Engl. J. Med._ 375, 1823–1833 (2016). Article  CAS  PubMed  Google Scholar  * McGranahan, N. _et al._

Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. _Science_ 351, 1463–1469 (2016). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  *

McLaughlin, J. _et al._ Quantitative assessment of the heterogeneity of PD-L1 expression in non-small-cell lung cancer. _JAMA Oncol._ 2, 46–54 (2016). Article  PubMed  PubMed Central 

Google Scholar  * Carbone, D. P. _et al._ First-line nivolumab in stage IV or recurrent non-small-cell lung cancer. _N. Engl. J. Med._ 376, 2415–2426 (2017). Article  CAS  PubMed  PubMed

Central  Google Scholar  * Wang, Y. & Navin, N. E. Advances and applications of single-cell sequencing technologies. _Mol. Cell_ 58, 598–609 (2015). Article  CAS  PubMed  PubMed Central

  Google Scholar  * Kashima, Y. _et al._ Single-cell sequencing techniques from individual to multiomics analyses. _Exp. Mol. Med._ https://doi.org/10.1038/s12276-020-00499-2 (2020). Article

  PubMed  PubMed Central  Google Scholar  * Bendall, S. C. _et al._ Single-cell mass cytometry of differential immune and drug responses across a human hematopoietic continuum. _Science_

332, 687–696 (2011). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Spitzer, M. H. & Nolan, G. P. Mass cytometry: single cells. Many features. _Cell_ 165, 780–791 (2016).

Article  CAS  PubMed  PubMed Central  Google Scholar  * Newell, E. W. & Cheng, Y. Mass cytometry: blessed with the curse of dimensionality. _Nat. Immunol._ 17, 890–895 (2016). Article 

CAS  PubMed  Google Scholar  * Takeuchi, Y. _et al._ Clinical response to PD-1 blockade correlates with a sub-fraction of peripheral central memory CD4+ T cells in patients with malignant

melanoma. _Int. Immunol._ 30, 13–22 (2018). Article  CAS  PubMed  Google Scholar  * Wistuba-Hamprecht, K. _et al._ Establishing high dimensional immune signatures from peripheral blood via

mass cytometry in a discovery cohort of stage IV melanoma patients. _J. Immunol._ 198, 927–936 (2017). Article  CAS  PubMed  Google Scholar  * Tirosh, I. _et al._ Dissecting the

multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. _Science_ 352, 189–196 (2016). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Fan, H. C., Fu, G. K. &

Fodor, S. P. A. Expression profiling. Combinatorial labeling of single cells for gene expression cytometry. _Science_ 347, 1258367 (2015). Article  PubMed  CAS  Google Scholar  * Zheng, G.

X. Y. _et al._ Massively parallel digital transcriptional profiling of single cells. _Nat. Commun._ 8, 14049 (2017). Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Gubin, M. M.

_et al._ High-dimensional analysis delineates myeloid and lymphoid compartment remodeling during successful immune-checkpoint cancer therapy. _Cell_ 175, 1014-1030.e19 (2018). Article  CAS

  PubMed  PubMed Central  Google Scholar  * Li, H. _et al._ Reference component analysis of single-cell transcriptomes elucidates cellular heterogeneity in human colorectal tumors. _Nat.

Genet._ 49, 708–718 (2017). Article  CAS  PubMed  Google Scholar  * Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across

different conditions, technologies, and species. _Nat. Biotechnol._ 36, 411–420 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Matsumoto, M. _et al._

Interleukin-10-producing plasmablasts exert regulatory function in autoimmune inflammation. _Immunity_ 41, 1040–1051 (2014). Article  CAS  PubMed  Google Scholar  * Togashi, Y., Shitara, K.

& Nishikawa, H. Regulatory T cells in cancer immunosuppression: Implications for anticancer therapy. _Nat. Rev. Clin. Oncol._ 16, 356–371 (2019). Article  CAS  PubMed  Google Scholar  *

Cuadrado, E. _et al._ Proteomic analyses of human regulatory T cells reveal adaptations in signaling pathways that protect cellular identity. _Immunity_ 48, 1046-1059.e6 (2018). Article  CAS

  PubMed  Google Scholar  * Sathe, A. _et al._ Single-cell genomic characterization reveals the cellular reprogramming of the gastric tumor microenvironment. _Clin. Cancer Res._ 26,

2640–2653 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  * Zhang, M. _et al._ Dissecting transcriptional heterogeneity in primary gastric adenocarcinoma by single cell RNA

sequencing. _Gut_ https://doi.org/10.1136/gutjnl-2019-320368 (2020). Article  PubMed  Google Scholar  * Zhang, P. _et al._ Dissecting the single-cell transcriptome network underlying gastric

premalignant lesions and early gastric cancer. _Cell Rep._ 27, 1934-1947.e5 (2019). Article  CAS  PubMed  Google Scholar  * Wu, K. _et al._ Redefining tumor-associated macrophage

subpopulations and functions in the tumor microenvironment. _Front. Immunol._ 11, 1731 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  * Zhu, Y. P. _et al._ Identification of

an early unipotent neutrophil progenitor with pro-tumoral activity in mouse and human bone marrow. _Cell Rep._ 24, 2329-2341.e8 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar 

* Marr, C., Zhou, J. X. & Huang, S. Single-cell gene expression profiling and cell state dynamics: collecting data, correlating data points and connecting the dots. _Curr. Opin.

Biotechnol._ 39, 207–214 (2016). Article  CAS  PubMed  PubMed Central  Google Scholar  * Zhang, Z. _et al._ SCINA: A semi-supervised subtyping algorithm of single cells and bulk samples.

_Genes._ 10, 531 (2019). Article  CAS  PubMed Central  Google Scholar  * Simoni, Y., Chng, M. H. Y., Li, S., Fehlings, M. & Newell, E. W. Mass cytometry: a powerful tool for dissecting

the immune landscape. _Curr. Opin. Immunol._ 51, 187–196 (2018). Article  CAS  PubMed  Google Scholar  * Bacher, R. & Kendziorski, C. Design and computational analysis of single-cell

RNA-sequencing experiments. _Genome Biol._ 17, 63 (2016). Article  PubMed  PubMed Central  CAS  Google Scholar  * Ren, X., Kang, B. & Zhang, Z. Understanding tumor ecosystems by

single-cell sequencing: promises and limitations. _Genome Biol._ 19, 211 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Navin, N. E. The first five years of single-cell

cancer genomics and beyond. _Genome Res._ 25, 1499–1507 (2015). Article  CAS  PubMed  PubMed Central  Google Scholar  * Shalek, A. K. & Benson, M. Single-cell analyses to tailor

treatments. _Sci. Transl. Med._ 9, 4730 (2017). Article  CAS  Google Scholar  * Park, S. _et al._ Paired whole exome and transcriptome analyses for the immunogenomic changes during

concurrent chemoradiotherapy in esophageal squamous cell carcinoma. _J. Immunother. Cancer_ 7, 128 (2019). Article  PubMed  PubMed Central  Google Scholar  * Haque, A., Engel, J., Teichmann,

S. A. & Lönnberg, T. A practical guide to single-cell RNA-sequencing for biomedical research and clinical applications. _Genome Med._ 9, 75 (2017). Article  PubMed  PubMed Central  CAS

  Google Scholar  * Kotecha, N., Krutzik, P. O. & Irish, J. M. Web-based analysis and publication of flow cytometry experiments. _Curr. Protoc. Cytom._ 10, 17 (2010). PubMed  Google

Scholar  * Van Gassen, S. _et al._ FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data. _Cytometry. A_ 87, 636–645 (2015). Article  PubMed  Google

Scholar  Download references ACKNOWLEDGEMENT This research was supported by AMED under Grant Number JP19ak0101058, Astellas Pharma, Inc., Daiichi Sankyo Co., Ltd., and Takeda Pharmaceutical

Company Ltd. The super-computing resource was provided by Human Genome Center (the Univ. of Tokyo). We thank Dr. Susumu Kobayashi, Dr. Akito Nakamura, Dr. Saomi Murai for valuable comments

on the manuscript. AUTHOR INFORMATION Author notes * These authors contributed equally: Yukie Kashima and Yosuke Togashi. AUTHORS AND AFFILIATIONS * Division of Translational Genomics,

Exploratory Oncology Research & Clinical Trial Center, National Cancer Center, Chiba, 277-8577, Japan Yukie Kashima & Akihiro Ohashi * Division of Cancer Immunology, Exploratory

Oncology Research & Clinical Trial Center, National Cancer Center, Chiba, Japan Yosuke Togashi, Shota Fukuoka, Takahiro Kamada, Takuma Irie & Hiroyoshi Nishikawa * Department of

Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Chiba, Japan Ayako Suzuki & Yutaka Suzuki * Department of Gastroenterology and

Gastrointestinal Oncology, National Cancer Center Hospital East, Chiba, 277-8577, Japan Yoshiaki Nakamura & Kohei Shitara * Department of Gastroenterology and Endoscopy, National Cancer

Center Hospital East, Chiba, Japan Tatsunori Minamide * Drug Discovery Research, Astellas Pharma, Inc., Ibaraki, Japan Taku Yoshida, Naofumi Taoka & Tatsuya Kawase * Oncology Research

Laboratories I, DaiichiSankyo. Co., Ltd., Tokyo, Japan Teiji Wada & Masataka Chihara * Biomarker & Translational Research Department, DaiichiSankyo. Co., Ltd., Tokyo, Japan Koichiro

Inaki * Takeda Pharmaceutical Company Ltd., Fujisawa, Japan Yukihiko Ebisuno * Axcelead Drug Discovery Partners Inc., Fujisawa, Japan Sakiyo Tsukamoto & Ryo Fujii * Division of

Translational Informatics, Exploratory Oncology Research & Clinical Trial Center, National Cancer Center, Chiba, Japan Katsuya Tsuchihara * Experimental Therapeutics, National Cancer

Center Hospital East, Chiba, Japan Toshihiko Doi Authors * Yukie Kashima View author publications You can also search for this author inPubMed Google Scholar * Yosuke Togashi View author

publications You can also search for this author inPubMed Google Scholar * Shota Fukuoka View author publications You can also search for this author inPubMed Google Scholar * Takahiro

Kamada View author publications You can also search for this author inPubMed Google Scholar * Takuma Irie View author publications You can also search for this author inPubMed Google Scholar

* Ayako Suzuki View author publications You can also search for this author inPubMed Google Scholar * Yoshiaki Nakamura View author publications You can also search for this author inPubMed

 Google Scholar * Kohei Shitara View author publications You can also search for this author inPubMed Google Scholar * Tatsunori Minamide View author publications You can also search for

this author inPubMed Google Scholar * Taku Yoshida View author publications You can also search for this author inPubMed Google Scholar * Naofumi Taoka View author publications You can also

search for this author inPubMed Google Scholar * Tatsuya Kawase View author publications You can also search for this author inPubMed Google Scholar * Teiji Wada View author publications You

can also search for this author inPubMed Google Scholar * Koichiro Inaki View author publications You can also search for this author inPubMed Google Scholar * Masataka Chihara View author

publications You can also search for this author inPubMed Google Scholar * Yukihiko Ebisuno View author publications You can also search for this author inPubMed Google Scholar * Sakiyo

Tsukamoto View author publications You can also search for this author inPubMed Google Scholar * Ryo Fujii View author publications You can also search for this author inPubMed Google

Scholar * Akihiro Ohashi View author publications You can also search for this author inPubMed Google Scholar * Yutaka Suzuki View author publications You can also search for this author

inPubMed Google Scholar * Katsuya Tsuchihara View author publications You can also search for this author inPubMed Google Scholar * Hiroyoshi Nishikawa View author publications You can also

search for this author inPubMed Google Scholar * Toshihiko Doi View author publications You can also search for this author inPubMed Google Scholar CONTRIBUTIONS Y.K. analyzed the scRNA-seq

dataset, interpreted dataset and write the manuscript. Y.T. designed the study, developed the methods and wrote the manuscript. S.F. and T.Kamada developed the methodology and revised the

manuscript. T.I. analyzed the CyTOF datasets. A.S. analyzed the scRNA-seq datasets. Y.N. and K.S. managed the enrollment of patients, developed the methodology, and revised the manuscript.

T.M. revised manuscript. T.Kawase revised the manuscript. T.Y, T. W, and Y. E. designed the study, developed the methodology, revised the manuscript and supervised the study. N.T. and M. C.

developed the methodology and revised the manuscript. K. I., R. F., and S. T. designed the study, developed the methodology, and revised the manuscript. A.O. designed the study. Y.S. and

K.T. revised the manuscript. H.N. supervised the study. T.D. designed the study, developed the methodology, reviewed and revised the manuscript and worked as principal investigator. All

authors read and approved the final paper. CORRESPONDING AUTHORS Correspondence to Akihiro Ohashi or Toshihiko Doi. ETHICS DECLARATIONS COMPETING INTERESTS Y. Togashi has received honoraria

from Ono, Bristol-Myers Squibb, Chugai, MSD and AstraZeneca, and research funding from KOTAI Biotechnologies. Y. Nakamura reported the research funding from Taiho Pharmaceutical. K. Shitara

reported paid consulting or advisory roles for Astellas, Lilly, Bristol-Myers Squibb, Takeda, Pfizer, Ono and MSD; honoraria from Novartis, AbbVie, and Yakult; and research funding from

Astellas, Lilly, Ono, Sumitomo Dainippon, Daiichi Sankyo, Taiho, Chugai, MSD and Medi Science. T. Yoshida, N. Taoka and T. Kawase are employed at Astellas Pharma, Inc. M.Chihara and T.Wada

are employed at Daiichi Sankyo Co., Ltd. K.Inaki is employed at Daiichi Sankyo RD Novare Co., Ltd. Y. Ebisuno is employed at Takeda Pharmaceutical Company Ltd. S. Tsukamoto and R.Fujii are

employed at Axcelead Drug Discovery Partners Inc. A.Ohashi was an employee of Takeda Pharmaceutical Company Ltd. from 2006 to 2018, and reported paid consulting or advisory roles for Ono

Pharmaceutical Company Ltd. out of this study. H. Nishikawa. has received honoraria and research funding from Ono Pharmaceutical, Bristol-Myers Squibb and Chugai Pharmaceutical outside of

this work as well as research funding from Taiho Pharmaceutical, Daiichi-Sankyo, Kyowa Kirin, Zenyaku Kogyo, Astellas Pharmaceutical, Sumitomo Dainippon Pharmaceutical, Asahi-Kasei, Sysmex,

and BD Japan outside of this study. No potential conflicts of interest were disclosed by the other authors. ADDITIONAL INFORMATION PUBLISHER'S NOTE Springer Nature remains neutral with

regard to jurisdictional claims in published maps and institutional affiliations. SUPPLEMENTARY INFORMATION SUPPLEMENTARY LEGENDS. SUPPLEMENTARY FIGURE 1. SUPPLEMENTARY FIGURE 2.

SUPPLEMENTARY FIGURE 3. SUPPLEMENTARY FIGURE 4. SUPPLEMENTARY FIGURE 5. SUPPLEMENTARY FIGURE 6. SUPPLEMENTARY TABLE 1. SUPPLEMENTARY TABLE 2. SUPPLEMENTARY TABLE 3. SUPPLEMENTARY TABLE 4.

SUPPLEMENTARY TABLE 5. SUPPLEMENTARY TABLE 6. SUPPLEMENTARY TABLE 7. RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International

License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source,

provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative

Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not

permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Kashima, Y., Togashi, Y., Fukuoka, S. _et al._ Potentiality of multiple modalities

for single-cell analyses to evaluate the tumor microenvironment in clinical specimens. _Sci Rep_ 11, 341 (2021). https://doi.org/10.1038/s41598-020-79385-w Download citation * Received: 17

June 2020 * Accepted: 07 December 2020 * Published: 11 January 2021 * DOI: https://doi.org/10.1038/s41598-020-79385-w SHARE THIS ARTICLE Anyone you share the following link with will be able

to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing

initiative