Data from: Integrated single-cell and spatial analysis reveals context-dependent myeloid-T cell interactions in the response to immune checkpoint blockade in head and neck squamous cell carcinoma
Data files
Mar 09, 2026 version files 2.49 GB
-
CosMX_Protein_TMA_Data_Expr.zip
958.27 MB
-
CosMX_RNA_TMA_Data_Expr.zip
1.53 GB
-
CosMX_TMA_Protein_TMA.txt
945 B
-
CosMX_TMA_RNA_TMA.txt
943 B
-
README.md
14.88 KB
Abstract
Purpose: We conduct a systematic evaluation of cell-cell interactions between tumor-infiltrating immune cells in patients with head and neck cancer squamous cell carcinoma (HNSCC) who have been treated with immune checkpoint blockade (ICB) using spatial and single-cell omics data.
Experimental Design: We employ complementary techniques from both Visium spot-based spatial transcriptomics and CosMx SMI single-cell spatial omics, utilizing a 64-plex protein panel and a 1000-gene RNA panel, which includes 435 ligands and receptors. We conducted integrated bioinformatics analyses to identify cellular neighborhoods of co-localizing cell types and Ligand-Receptor interactions across different single-cell and spatial data modalities.
Results: With 522,399 single cells profiled for both RNA and protein from 23 patients, along with spot-resolved spatial RNA-seq data from 8 patients treated with ICB, and through bioinformatics analysis of publicly available single-cell and bulk RNA-seq, we identified a spatial and cell-type specific context dependency in the differences of myeloid and T cell interactions between Responders and Non-Responders samples. We further defined the cellular neighborhood and sources of chemokine CXCL9/10-CXCR3 interactions, emphasizing the specificity of this marker in Responders samples, an emerging target in ICB, as well as other underappreciated markers and targets for ICB response in HNSCC, such as CXCL16-CXCR6, CCL4/5-CCR5.
Conclusions: We have provided a valuable resource for analyzing spatial and cell-cell ligand-receptor interactions, including the cellular and spatial contexts of ICB response markers. Our data suggest that future mechanistic studies should consider this context specificity when evaluating ICB response biomarkers and targets.
https://doi.org/10.5061/dryad.n2z34tn7x
Protein and RNA CosMx TMA samples were generated using CosMx SMI by Nanostring Technologies Inc. (Bruker Spatial Biology), Seattle, WA, using a Human Universal Cell Characterization 1000-plex RNA panel, including cell typing markers and a list of 435 ligand-receptors (https://nanostring.com/resources/cosmx-smi-universal-cell-characterization-panel-gene-list/) and Human Immune Oncology Protein Panel including 64 markers (https://nanostring.com/products/cosmx-spatial-molecular-imager/cosmx-protein-assays/cosmx-human-immuno-oncology-protein-panel/). Target RNA readout onthe SMI instrument was performed following published protocols with a fluorophore-conjugated antibody cocktail against CD298/B2M, PanCK, CD45, and CD3 proteins and DAPI to acquire morphology images. Nine Z-stack images for 5 channels (4 antibodies and DAPI) were captured after washout of unbound antibodies and addition of Imaging buffer.
Description of the data and file structure
Dataset Overview
This dataset contains the Bruker Spatial (formerly Nanostring) CosMX SMI single-cell spatial profiling data from 68 tumor microarray cores from 23 HNSCC patients.
Details
CosMX_RNA_TMA_Data_Expr (CosMX_RNA_TMA_Data_Expr.zip): This folder contains all RNA expression and related spatial data from each of the three TMA runs for the CosMx RNA assay . Each TMA includes the following five files (filenames start with Run5607_TMAA, Run5545_TMAB, Run5668_TMAC)
- (1) Transcript File (_tx_file.csv): Column descriptions
- fov
- The Field Of View (FOV) the transcript is in. All FOVs from the sample are contained within this transcript file.
- cell_ID
- Unique identifier for a single cell within a given FOV. To make a unique identifier for a cell within the whole sample use both the “fov” and “cell_ID” columns. Transcripts not assigned to cells have a value of 0.
- x_global_px
- See “x_local_px” description. The global positions describes the relative position of this transcript within the large sample reference frame.
- y_global_px
- Same as “x_global_px” but for the y dimension
- x_local_px
- The x position of this transcript within the FOV, measured in pixels. To convert to microns multiply the pixel value by 0.168 um per pixel.
- y_local_px
- Same as “x_local_px” but for the y dimension
- z
- The Z-slice image number. This has an arbitrary unit.
- target
- The gene detected. HUGO gene symbols are used.
- CellComp
- The subcellular compartment (Nuclear, Membrane, or Cytoplasmic. “0” denotes Extracellular) in which the transcript was detected as determined by cell segmentation algorithm.
- fov
- (2) Cell Expression File (_exprMat_file.csv): Column descriptions
- fov
- The Field Of View (FOV) the transcript is in. All FOVs from the sample are contained within this Cell Expression file.
- cell_ID
- Unique identifier for a single cell within a given FOV. To make a unique identifier for a cell within the whole sample use both the “fov” and “cell_ID” columns. All transcripts not assigned to a cell are show with a “cell_ID” value of 0.
- Gene expression target (AATK, ABL1, …, etc)
- The number of transcripts observed for a given gene target for a given cell.
- Negative probes (i.e. NegPrb##)
- Negative probes, which do not match any sequence within the transcriptome or genome, are included. These can be used to assess background levels.
- fov
- (3) Cell Metadata File (_metadata_file.csv): Column descriptions
- fov
- The Field Of View (FOV) the transcript is in. All FOVs from the sample are contained within this Cell Metadata file.
- cell_ID
- Unique identifier for a single cell within a given FOV. To make a unique identifier for a cell within the whole sample use both the “fov” and “cell_ID” columns.
- Area
- Number of pixels assigned to a given cell.
- Aspect Ratio
- width divided by height.
- CenterX_global_px
- See “CenterX_local_px” description. The global positions describes the relative position of this transcript within the large sample reference frame.
- CenterY_global_px
- Same as “CenterX_global_px” but for the y dimension
- CenterX_local_px
- The x position of this transcript within the FOV, measured in pixels. The pixel edge length is 168nm. Thus, to convert to microns multiply the pixel value by 0.168 um per pixel.
- CenterY_local_px
- Same as “CenterX_local_px” but for the y dimension
- Width
- Cell’s maximum length in x dimension (pixels)
- Height
- Cell’s maximum length in y dimension (pixels)
- Mean.MembraneStain
- Mean fluorescence intensity within a given cell for the channel detecting the membrane stain.
- Max.MembraneStain
- Maximum fluorescence intensity within a given cell for the channel detecting the membrane stain.
- Mean.PanCK
- Mean fluorescence intensity within a given cell for the Green Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Max.PanCK
- Maximum fluorescence intensity within a given cell for the Green Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Mean.CD45
- Mean fluorescence intensity within a given cell for the Yellow Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Max.CD45
- Maximum fluorescence intensity within a given cell for the Yellow Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Mean.CD3
- Mean fluorescence intensity within a given cell for the Red Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Max.CD3
- Maximum fluorescence intensity within a given cell for the Red Channel.
- As the morphology markers are customizable this column will change to match the requested markers in each experiment.
- Mean.DAPI
- Mean fluorescence intensity within a given cell for the DAPI Channel.
- Max.DAPI
- Maximum fluorescence intensity within a given cell for the DAPI Channel.
- fov
- (4) FOV Positions File (_fov_positions_file.csv): Column descriptions
- fov
- The Field Of View (FOV) number.
- x_global_px
- The relative x position of this the FOV, measured in pixels. To convert to microns multiply the pixel value by 0.168 um per pixel. All FOVs are the same dimension, 4480 x 4480 pixels.
- y_global_px
- See “x_global_px” description, but in for the y dimension.
- fov
- (5) Cell Polygons File (-polygons.csv)
- fov
- The Field Of View (FOV) number.
- cell_ID
- Unique identifier for a single cell within a given FOV. To make a unique identifier for a cell within the whole sample use both the “fov” and “cell_ID” columns.
- x_local_px
- The x position of this transcript within the FOV, measured in pixels. To convert to microns multiply the pixel value by 0.168 um per pixel.
- y_local_px
- Same as “x_local_px” but for the y dimension
- x_global_px
- The relative x position of this the FOV, measured in pixels. To convert to microns multiply the pixel value by 0.168 um per pixel. All FOVs are the same dimension, 4480 x 4480 pixels.
- y_global_px
- See “x_global_px” description, but in for the y dimension.
- fov
CosMX_Protein_TMA_Data_Expr (CosMX_Protein_TMA_Data_Expr.zip) This folder contains all data from each of the three TMA runs for the CosMx Protein assay. Each TMA (subfolder TMA_A, TMA_B, TMA_C) includes two sub folders (CellStatsDir and ProteinDir) with the following structure and format.
- CellStatsDir (csv files) for each TMA FOV (filenames Run5605_20221222_182336_S1_Cell_Stats_Fxxx.csv, with xxx is FOV_ID for TMA_A, Run5605_20221222_182336_S2_Cell_Stats_Fxxx.csv with xxx is FOV_ID for TMA_B, Run5673_20230215_163321_S2_Cell_Stats_Fxxx.csv with xxx is FOV_ID for TMA_C ): including the following columns - for cell ID, spatial coordinates and intensity of morphological markers (PanCK - tumor epithelial marker, CD298_B2M (combines antibodies for CD298 (ATP1B3) and B2M (Beta-2-microglobulin),) - cell membrane for segmentation , CD45 (immune cell marker), DAPI (4′,6-diamidino-2-phenylindole) is a popular blue-fluorescent dye for nuclear counterstain in fixed cells).
- CellId: A unique identifier for the cell within that specific FOV.
- Area: Total number of pixels assigned to the cell.
- AspectRatio: The ratio of the cell's width to its height.
- CenterX, CenterY: The X and Y coordinates (in pixels) of the cell's centroid relative to the individual FOV image.
- Width,Height: The maximum dimensions of the cell's bounding box.
- Mean-PanCK,Max-PanCK: Average and Maximum intensity for epithelial cells.
- Mean-Yellow,Max-Yellow: Average and Maximum intensity value for the yellow signal (CD3 - T cell marker) in any single pixel within that cell.
- Mean-CD298_B2M,Max-CD298_B2M: Average and Maximum pixel intensity value for this stain found within the cell boundary.
- Mean-CD45,Max-CD45: Average and Maximum intensity for immune cells.
- Mean-DAPI,Max-DAPI: The average & maximum pixel intensity value for the DAPI signal within the cell's boundary.
- ProteinDir: PerCellStats the individual FOV level ( TMA_A/ProteinDir/FOVxxx/PerCellStats/20221222_182336_S1_C001_Fxxx_yyy_perCell_1ChStats.csv with xxx is FOV_ID and yyy is marker name for TMA_A, TMA_B/ProteinDir/FOVxxx/PerCellStats/20221222_182336_S2_C001_Fxxx_yyy_perCell_1ChStats.csv with xxx is FOV_ID and yyy is marker name for TMA_B, TMA_C/ProteinDir/FOVxxx/PerCellStats/20230215_163321_S2_C001_Fxxx_yyy_perCell_1ChStats.csv with xxx is FOV_ID and yyy is marker name for TMA_C), for each 62 markers ("4-1BB" "B7-H3" "BCL2" "Beta-catenin" "CCR7" "CD11b" "CD11c" "CD123" "CD127" "CD138" "CD14" "CD15" "CD16" "CD163" "CD19" "CD20" "CD27" "CD3" "CD31" "CD34" "CD38" "CD39" "CD4" "CD40" "CD45" "CD45RA" "CD56" "CD68" "CD8" "CTLA4" "EGFR" "EpCAM" "FABP4" "FN1" "FOXP3" "GITR" "GZMA" "GZMB" "HER2" "HLA-DRA" "ICAM1" "ICOS" "IDO1" "IL-18" "IL-1B" "IgD" "KI67" "LAG3" "LAMP1" "NF-kBp65" "PD1" "PDL1" "PDL2" "SMA" "STING" "TCF-1" "TIM3" "VIM" "VISTA" "iNOS" "p53" "panRAS"), plus two control markers (MsIgG1, RbIgG): including the following columns from the cell segmentation results using CosMX SMI pipelines - for cell ID, spatial coordinates, segementation, and intensity across the cell area and within subcellular compartment (Nuc - Nucleus, Cyt-Cytoplasm, Mem-Membrane).
- CellId: A unique identifier for the cell within that specific FOV.
- Area: Total number of pixels assigned to the cell.
- AspectRatio: The ratio of the cell's width to its height.
- CenterX, CenterY: The X and Y coordinates (in pixels) of the cell's centroid relative to the individual FOV image.
- Width,Height: The maximum dimensions of the cell's bounding box.
- Min,Avg,Max: The min, mean, max intensity of the stain across all pixels in the cell.
- Sum: The total integrated intensity (sum of all pixel values) for that channel in the cell
- Cnt (Count): The total number of pixels that make up the segmented cell area.
- MinSeg,AvgSeg,MaxSeg:The minimum, median, maximum intensity value found strictly within the boundaries defined by the segmentation method.
- SumSeg: The total intensity summed only over the pixels included in the final segmentation mask.
- Nuc-Area: The physical size of the cell's nucleus (in pixels)
- Nuc-Min,Nuc-Avg,Nuc-Max: The minimal, mean, and maximal intensity values within the nuclear boundary.
- Nuc-Sum: The total integrated intensity of the signal within the nucleus.
- Nuc-Cnt: The total pixel count assigned to the nucleus.
- Nuc-MinSeg,Nuc-AvgSeg,Nuc-MaxSeg,Nuc-SumSeg: The minimal, mean, maximal and total signal intensity within that refined nuclear mask generated by segmentation method.
- Cyt-Area: The physical area of the cytoplasm (Whole Cell Area minus Nuclear Area).
- Cyt-Min,Cyt-Avg,Cyt-Max,Cyt-Sum: The minimal, mean, maximal and total integrated intensity of the signal within the cytoplasmic boundary.
- Cyt-Cnt: The total pixel count assigned to the cytoplasmic region.
- Cyt-MinSeg,Cyt-AvgSeg,Cyt-MaxSeg,Cyt-SumSeg: The minimal, mean, maximal and total signal intensity within the refined cytoplasmic segmentation mask
- Mem-Area: The physical area (in pixels) of the identified membrane.
- Mem-Min,Mem-Avg,Mem-Max,Mem-Sum: The minimal, mean, maximal and toal intensity within the membrane boundary.
- Mem-Cnt: The total pixel count assigned specifically to the membrane region.
- Mem-MinSeg,Mem-AvgSeg,Mem-MaxSeg,Mem-SumSeg: The minimal, mean, maximal, and total signal intensity summed across the refined membrane pixels.
TMA cores map to Patient ID
File CosMX_TMA_RNA_TMA.txt has the map of FOV ID for RNA-profile TMA data, including two columns (TMA FOV ID and Patient ID, note that multiple FOVs are from one patient)
- TMA FOV ID: TMA_x_FOVy (x is A/B/C, y is FOV ID number), 68 TMA cores.
- Patient ID: letter (A-Z) for 23 patients.
CosMX_TMA_Protein_TMA.txt has the map of FOV ID for Protein-profile TMA data, including two columns (TMA FOV ID and Patient ID, note that multiple FOVs are from one patient)
- TMA FOV ID: TMA_x_FOVy (x is A/B/C, y is FOV ID number), 68 TMA cores.
- Patient ID: letter (A-Z) for 23 patients.
Human subjects data
The clinical sample collection and analyses were conducted in accordance with the Declaration of Helsinki and received institutional ethics approval from the University of Wisconsin-Madison (UW18144 with IRB#2018–1510, subproject number 2022-009), under a waiver of informed consent. Samples are de-identified by the clinical research team before data profiling and analysis.