Protocol for identifying immune checkpoint on circulating tumor cells of human pancreatic ductal adenocarcinoma by single-cell RNA sequencing

Summary

Circulating tumor cells (CTCs) are regarded as the “seeds” of tumor metastasis. Identifying immune checkpoints on CTCs is essential for establishing efficient immunotherapies to prevent tumor metastasis. Here, we present a protocol for isolating CTCs and obtaining single-cell suspensions from pancreatic ductal adenocarcinoma liver metastatic patients. We describe steps for biospecimen acquisition, CTC isolation, and tissue dissociation. We then detail procedures for performing single-cell RNA-seq, annotating cell types, and identifying immune checkpoints on CTCs.

For complete details on the use and execution of this protocol, please refer to Liu et al. (2023).¹

Graphical abstract

Highlights

• •Isolation of circulating tumor cells (CTCs) from blood samples
• •Dissociation of single cells from solid tumor primary and metastatic lesions
• •Single-cell RNA-seq data processing and cell-type annotation
• •Identifying immune checkpoints of circulating tumor cells (CTCs), primary tumor cells, and metastatic tumor cells

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

Circulating tumor cells (CTCs) are regarded as the “seeds” of tumor metastasis. Identifying immune checkpoints on CTCs is essential for establishing efficient immunotherapies to prevent tumor metastasis. Here, we present a protocol for isolating CTCs and obtaining single-cell suspensions from pancreatic ductal adenocarcinoma liver metastatic patients. We describe steps for biospecimen acquisition, CTC isolation, and tissue dissociation. We then detail procedures for performing single-cell RNA-seq, annotating cell types, and identifying immune checkpoints on CTCs.

Before you begin

Before collecting the clinical biopsies, the study must be approved by Ethics Committee and comply with all relevant ethical regulations. Informed consent should collect from all patients. Before processing the clinical samples, patients should undergo pre-transfusion testing for hepatitis B, C, syphilis, and HIV, and the results must be negative. All individuals involved in the processing of clinical samples must adhere to safety precautions governing work, which include wearing lab coats, gloves, safety glasses, and other necessary protective gear.

Institutional permissions

All human studies were approved by the Ethics Committee on Biomedical Research of West China Hospital and we complied with all relevant ethical regulations.

Prepare biopsy samples from patients

Timing: 3–4 h

• 1.Obtain biopsy tissue, including primary tumors, metastasis lesions, and hepatic portal vein (HPV) blood, from pancreatic ductal adenocarcinoma (PDAC) liver metastatic patients by laparoscopic surgery. The specimens of primary pancreatic tumor and liver metastasis were collected with Groff electrosurgical knife. The hemostasis of the excision surface was performed by electrocoagulation. Approximately 10 mL of blood sample was collected from the hepatic portal vein (HPV) system with a scalp needle.
• 2.After surgical excision, place the tissue immediately (typically within 15 min) in the Tissue Storage Solution (Miltenyi Biotec, Cat# 130-100-008) and keep it chilled on ice. Then, transport the samples to the laboratory within an hour for subsequent tissue dissociation.

Prepare reagents for CTC isolation and tumor tissue dissociation

Timing: 30–60 min

• 3.Obtain the CTC microfluidic chip system from Hangzhou MerryHealth (Hangzhou, China), which includes a pre-coated microfluidic chip with anti-human EpCAM and anti-human CA19.9 antibodies, elution buffer, cell resuspension buffer, polyethylene tubes, and a syringe pump.
• 4.Ensure the microfluidic chip is full of protection solution and free of air bubbles before use.
• 5.Prepare 0.1% and 0.01% BSA-HBSS solution. 0.1% and 0.01% BSA solutions were prepared by adding 5 mg or 0.5 mg of BSA powder to 50 mL of HBSS buffer, mixing the solution by pipetting it up and down, and then passing it through a filter with a 0.22 μm pore size hydrophilic polyethersulfone (PES) membrane.
• 6.Place the cell resuspension buffer, elution buffer, red blood cell (RBC) lysis buffer, and BSA-HBSS solution on ice or 4°C.

Prepare reagents for tumor tissue dissociation and FACS isolation

Timing: 30–60 min

• 7.Prepare a digestive enzyme solution containing 0.25% trypsin (Gibco), 0.4 mg/mL collagenase type I and type IV (Gibco). Warm the digestive enzyme solution in a 37°C water bath for 30 min before use.
• 8.Prepare the complete medium by adding 10% fetal bovine serum (FBS) into DMEM basic medium.
• 9.Prepare FACS collection tube: a. Add 100 μL 0.1% BSA solution to a 1.5 mL Eppendorf tube and place on ice.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FITC anti-human CD326 (EpCAM) antibody (working dilution: 5:100)	BioLegend	Cat# 324204; RRID: AB_756078
APC/cyanine7 anti-human CD45 antibody (working dilution: 5:100)	BioLegend	Cat# 304014; RRID: AB_314402
Biological samples
PDAC patient samples	West China Hospital, Sichuan University	N/A
Chemicals, peptides, and recombinant proteins
RPMI-1640 medium	Gibco	Cat# C11875500BT
Fetal bovine serum (FBS)	Gibco	Cat# 12483020
DMEM medium	Gibco	Cat# C11995500BT
Collagenase, type 1	Gibco	Cat# 17100017
Collagenase, type IV	Gibco	Cat# 17104019
Trypsin	Millipore	Cat# SM-2003-C
Hoechst 3342	Sigma-Aldrich	Cat# 14533
Propidium iodide (PI)	BD	Cat# 556463
HBSS	Thermo Fisher Scientific	Cat# 14175079
BSA	Sigma-Aldrich	Cat# V900933
Glucose	Merck	Cat# D9434
Red blood cell (RBC) lysis buffer	Solarbio	Cat# R1010
Cell stain buffer	BioLegend	Cat# 420201
Tissue storage solution	Miltenyi Biotec	Cat# 130-100-008
Trypan blue solution	Thermo Fisher Scientific	Cat# 15250061
0.25% Trypsin	Millipore	Cat# SM-2003-C
Nuclease-free water	Thermo Fisher Scientific	Cat# AM9937
Low TE buffer (10 mM Tris-HCl pH 8.0, 0.1 mM EDTA)	Thermo Fisher Scientific	Cat# 12090-015
QIAGEN Buffer EB	QIAGEN	Cat# 19086
10% Tween 20	Bio-Rad	Cat# 1662404
Glycerin (glycerol), 50% (v/v) aqueous solution	Ricca Chemical Company	Cat# 3290-32
Critical commercial assays
Chromium Single Cell 3ʹ GEM, Library & Gel Bead Kit v3, 16 rxns	10× Genomics	Cat# PN-1000075
SPRIselect Reagent Kit	Beckman Coulter	Cat# B23318
Chromium i7 Multiplex Kit, 96 rxns	10× Genomics	Cat# 120262
Qubit dsDNA HS Assay Kit	Thermo Fisher Scientific	Cat# Q32854
Deposited data
scRNA-seq data of PDAC in this study	NGDC: OMIX (https://ngdc.cncb.ac.cn/omix/releaseList/)	OMIX002487
Software and algorithms
Cell Ranger-3.0.0	10× Genomics	http://10xgenomics.com/
Python-3.7.8	Van Rossum and Drake (2009)	https://www.python.org/
R-4.0.3	R Core Team (2008)	https://www.r-project.org/
Seurat-4.0.1	Seurat et al.2	https://satijalab.org/seurat/
FlowJo version 10.8.0	BD	https://www.flowjo.com/
ggplot2-3.3.3	CRAN	https://cran.r-project.org/web/packages/ggplot2/
ggpubr-0.4.0	CRAN	https://cran.r-project.org/web/packages/ggpubr/
ComplexHeatmap-2.6.2	Gu et al.3	https://www.bioconductor.org/packages/release/bioc/html/ComplexHeatmap.html
CellPhoneDB-2.1.2	Efremova et al.4	https://www.cellphonedb.org/
data.table-1.14.8	CRAN	https://cran.r-project.org/web/packages/data.table/index.html
Psych-2.3.3	CRAN	https://cran.r-project.org/web/packages/psych/index.html
CopyKAT-1.0.5	Gao et al.5	https://github.com/navinlabcode/copykat
Other
Web portal	This manuscript	https://github.com/Jinen22/scRNA-PDAC-CC, https://doi.org/10.5281/zenodo.8151927
CTC microfluidic chip system	Hangzhou MerryHealth	http://merryhealthbio.com/product/
DNA LoBind Tubes, 1.5 mL	Eppendorf	Cat# 022431021
DNA LoBind Tubes, 2.0 mL	Eppendorf	Cat# 022431048
PCR tubes 0.2 mL 8-tube strips	Eppendorf	Cat# 951010022
Pipette tips: 20 μL, 200 μL, and 1000 μL	Rainin	Cat# 30389226, 30389240, and 30389213
Centrifuge tube: 15 mL and 50 mL	Jet Biofil	Cat# CFT011150, CFT011500
Mlllex-GP Filter Unit	Merck	Cat# SLGP033RB
Vacutainer EDTA blood collection tube	BD	Cat# 02-657-32
Falcon round-bottom polystyrene test tubes	Thermo Fisher Scientific	Cat# 352054
Sterile cell strainers	Thermo Fisher Scientific	Cat# 352350
BD FACSAria SORP Flow Cytometer	BD Biosciences	N/A

Materials and equipment

0.1% BSA-HBSS solution
Reagent	Final concentration	Amount
BSA	0.1 mg/mL	5 mg
HBSS	N/A	50 mL
Total	-	50 mL
Storage at 4°C. The medium stored for one week.

0.01% BSA-HBSS solution
Reagent	Final concentration	Amount
0.1% BSA-HBSS solution	10% (v/v)	5 mL
HBSS	N/A	45 mL
Total	-	50 mL
Storage at 4°C. The medium stored for one week.

Cell resuspension buffer (this buffer mainly used in the CTC isolation part)
Reagent	Final concentration	Amount
BSA	0.1 mg/mL	10 mg
Glucose	5 mg/mL	500 mg
HBSS	N/A	100 mL
Total	-	100 mL
Storage at 4°C. The medium stored for one week.

Digestive enzyme solution
Reagent	Final concentration	Amount
Collagenase type I	0.4 mg/mL	20 mg
Collagenase type IV	0.4 mg/mL	20 mg
0.25% trypsin	0.25% (w/v)	50 mL
Total	-	50 mL
Storage at 4°C. The medium stored for one week.

Complete medium
Reagent	Final concentration	Amount
1 × DMEM basic medium	N/A	450 mL
FBS	10% (v/v)	50 mL
Total	-	500 mL
Storage at 4°C. The medium stored for one month.

Step-by-step method details

Isolation of CTCs from blood samples

Timing: ∼2 h

This section will demonstrate capture CTCs from PDAC whole blood samples.

CRITICAL: To obtain high-quality CTCs, it is essential to isolate the CTCs from fresh blood samples. The hepatic portal vein (HPV) blood samples should be collected in EDTA blood collection tubes and processed within 1–2 h after collection to avoid cell death and mRNA degradation. Throughout the protocol, all samples should be kept on ice or at 4°C to maintain their integrity.

• 1.
  Red blood cell lysis (Figures 1A–1C).

  • a.
    Collect HPV blood samples from PDAC liver metastatic patients.

    Note: In this study, we collected 6 HPV blood samples from PDAC liver metastatic patients to collect CTC and perform single-cell RNA-seq.

  • b.
    Transfer the blood samples to a 50 mL centrifuge tube.
  • c.
    Add RBC lysis buffer with a volume ratio of 1:3 (e.g., every 10 mL blood sample add 30 mL RBC lysis buffer), and gently resuspend the samples.
  • d.
    Incubate the samples for 10 min at 4°C.
  • e.
    After step 1d, dilute the samples by adding double volumes of a 0.1% BSA-HBSS solution (e.g., every 10 mL samples add 20 mL BSA-HBSS solution) to stop the reaction.
  • f.
    Centrifuge at 500 g at 4°C for 5 min and carefully aspirate/pour off the supernatant.

    Note: We recommend prepare step 2 and 3 after step 1f.

  • g.
    Add 10 mL of cell resuspension buffer and gently resuspend the cells with a plastic pipette. Then, collect the cells together and transfer them into a 15 mL centrifuge tube.
  • h.
    Repeat step 1f.
  • i.
    Gently resuspend the cell with 5 mL cell resuspension buffer and adjust the concentration to 1–3×10ˆ7 cells/mL.

    Note: If the cell pellet at this step appears light red, repeat RBC lysis steps from step 1c to 1f. In step 1i, we recommend resuspending each 10 mL blood derived cells with 5 mL of cell resuspension buffer.

• 2.After step 1f, prepare the CTC capture chip. The microfluidic chip (Figure 1D) is provided by MerryHealth Ltd (Hangzhou, China).

CRITICAL: During the CTC capture process (from step 2–14), always keep the microfluidic chip wet.

• 3.Fill a syringe with 10 mL 0.1% BSA-HBSS solution. Use a syringe pump to pass the solution through the chip channels at 10 mL/h flow rate to remove the protection solution of the chip and prevent cell adhesion on the inner walls of the tubes.
• 4.Transfer the resuspended cells (about 5 mL) into the syringes and connect polyethylene tubes to the syringe.

Note: We recommend using disposable plastic syringes and removing the needle when aspirating the cells.

• 5.Place the microfluidic chip on a level table.
• 6.Mount the syringes to a syringe pump and connect them to the entrance of microfluidic chips using the polyethylene tubes (Figure 1E).
• 7.Connect the exit ports of the chips to a waste container through another set of polyethylene tubes (Figure 1E).

CRITICAL: Avoid trapping air bubbles get when connecting the syringe to the microfluidic chip.

• 8.Adjust the flow rate of a syringe pump to 5 mL/h and inject the samples completely into microfluidic channels.
• 9.Replace the empty sample syringes with new syringes filled with 10 mL of 0.1% BSA solution.
• 10.Connect the wash syringes to the microfluidic chip and wash channels at 30 mL/h flow rate to remove the unattached cells from the channels.
• 11.Repeat steps 9 and 10 two times and observe under microscope to confirm the unattached cells were removed (Figure 1F).
• 12.Replace the empty wash syringes with new syringes filled with 15 mL of Elution buffer.
• 13.Replace the waste container with a new centrifuge tube to collect the eluted cells.
• 14.Wash the channels with Elution buffer at 50 mL/h flow rate to collect the captured CTCs.
• 15.Centrifuge at 500 g for 5 min at 4°C and resuspend the pellet with 0.01% BSA-HBSS solution.
• 16.To evaluate cell viability (recommended higher than 80%) and concentration, aspirate 5 μL of cell suspension with 5 mL of 0.4% (w/v) Trypan Blue Solution and count the number using hemocytometer under bright field microscope.
• 17.Store the cell suspension on ice before further processing.
• 18.
  To identify CTCs and estimate the abundance of CTCs in the collected cells, by splitting 10% of the collected cells for immunofluorescent staining.

  • a.
    Add TRITC-CD45 and FITC-EpCAM antibodies to the suspension and incubate at 4°C for 30 min in the dark.
  • b.
    After incubating the antibodies for 20 min, add Hoechst dye to stain the nucleus.
  • c.
    Wash the cells with cold 0.1% BSA-HBSS by centrifugation at 500 g for 5 min at 4°C.
  • d.
    Resuspend the cells with 10 μL HBSS and coat the cells on a slide for confocal observation.
  • e.
    Detect and calculate the number of CTCs under fluorescence microscope. The typical CTCs are Hoechst 33342 positive (blue), EpCAM positive (green), and CD45 negative (red) (Figure 1G).
• 19.After step 17, place the cell suspension on ice and without further delays (recommended within 1 h) to perform “droplet-based single-cell RNA-Seq”.

CRITICAL: Maintaining cell viability is essential for scRNA-seq, so it is important to capture the CTCs at 4°C or ice throughout the process.

CRITICAL: Because the number of CTC is crucial for scRNA-seq, it is recommended to calculate the CTC number before being subject to the 10× Genomic process and ensure that at least 100 CTCs is collected.

Figure 1.

The experimental schema for capturing CTC by microfluidic chip

(A and B) The HPV blood sample from PDAC liver metastatic patients were collected in EDTA tube (A) and lysed with RBC lysis buffer (B).

(C) The photograph shows the cells after RBC lysis.

(D) The external view of CTC microfluidic chip.

(E) The photograph presents the experiment procedure of capturing CTC from lysed HPV blood samples by microfluidic chip.

(F) Representative images of CTC captured on the microfluidic chip.

(G) CTCs, which were captured from HPV blood, were verified by multiplex immunofluorescence staining, including EpCAM (green), immunocyte marker CD45 (red), and Hoechst 33342 (blue). Scale bar, 100 μm.

Processing of tumor samples to single-cell suspensions

Timing: ∼1 h

This section will dissociate single-cells from solid tumor primary and metastatic lesions.

• 20.Collect the primary and liver metastatic tumor samples from PDAC samples through laparoscopic surgery.

CRITICAL: To identify the unique immune checkpoint molecules on CTCs, we recommend setting the paired primary and metastatic tumor cells as control.

• 21.Place the fresh surgical specimen in a sterile container filled with 5–10 mL of tissue storage solution and immediately transport it to the laboratory on ice.

Note: Ensure that the biospecimen is completely submerged in tissue storage buffer.

• 22.Transfer the tumor samples into a 15 cm dish and immediately cut them into small pieces of less than 1 mm³ with scissors on ice.
• 23.Transfer the pieces to a 50 mL centrifuge tube with 20 mL digestive enzyme containing (Gibco) 0.25% trypsin, 0.4 mg/mL collagenase type I, and type Ⅳ (Gibco).
• 24.Place the tube in a 37°C water bath and digest for 10–15 min gently with manual shaking. The digestion time can be adjusted according to the size of tissue chunks but should not exceed 15 min.
• 25.Add an equal volume of ice-cold DMEM complete medium (containing 10% fetal bovine serum) to stop the digestion and gently mix the samples up and down 5–8 times with a Pasteur pipette.
• 26.Filter the dissociated cells through a 70 μm cell strainer. If the tissue chunks block the cell strainer, utilize a plunger from a 5 mL syringe to press the samples and rinse the filter with ice-cold RPMI medium.
• 27.Centrifuge the cell suspension at 500 g for 5 min at 4°C and gently aspirate/pour off supernatant to remove debris and dead cells.
• 28.Gently resuspend the cell pellet with 2–3 mL RBC lysis buffer and incubate for 3–5 min on ice.

CRITICAL: This step must be performed even though the red blood cells are invisible to the naked eye.

• 29.Dilute the RBC buffer with 10 mL 0.1% BSA-HBSS solution and centrifuge at 500 g for 5 min to remove the supernatant.
• 30.Rinse the cells again with 1 mL of 0.1% BSA-HBSS solution and transfer the pellet into a new 1.5 mL Eppendorf Protein LoBind Tube.
• 31.Centrifuge the samples at 500 g for 5 min to discard the supernatant.

CRITICAL: All waste solutions and materials should be autoclaved for sterilization.

FACS sorting

Timing: ∼1–1.5 h

In this section, we will sort the live cells from the single-cell suspensions derived from primary and metastatic lesions using FACS

• 32.Resuspend the cell pellet with 0.1% BSA-HBSS solution at a density of no greater than 1 × 10ˆ6 cells/100 μL.
• 33.Add 5 μL CD45 antibodies for sorting CD45 positive immunocytes and negative non-immunocytes. Incubate the antibody for 20 min on ice.

Note: If enriching all live cells, CD45 staining was not required. Set the sorting gate negative for PI.

• 34.After antibody incubation, wash the cells with 0.1% BSA-HBSS solution and resuspend them at a density of approximately 5 × 10ˆ6 cells/mL.
• 35.Prior to sorting, add 5 μL/100 μL PI staining solution and incubate for 5 min on ice.
• 36.After PI incubation, filter the cells again through a 70 μm cell strainer to avoid clumps.
• 37.Prepare FACS sorted cell collection tubes by adding 50 μL of 0.1% BSA-HBSS solution to 1.5 mL Eppendorf LoBind Tube and placing them on ice.
• 38.Establish appropriate sorting gates to isolate the desired cell population (Figures 2A and 2B). To sort live non-immunocytes, ensure that the sorting gates are negative for both PI and CD45.

Note: In this protocol, the BD FACSAria SORP Flow Cytometer (manufactured 2016, equipped with a 100 μm nozzle) was used for cell sorting. The sorting parameters were set as follows: threshold (5,000), sheath fluid pressure (30 PSI) and sorting speed (1000 events/s). However, cell sorting can also be performed on any other conventional FACS instrument.

• 39.After setting the sorting gates, initiate cell sorting and collect the target cells into the collection tube.

CRITICAL: Collecting the sorted cells into a 1.5 mL Eppendorf LoBind Tube is recommended due to the low number of sorted cells (commonly is 1–2×10ˆ5 cells per sample). Use filtered 0.1% BSA-HBSS solution as the collection buffer to increase cell viability.

• 40.Centrifuge cells at 500 g for 4 min at 4°C. Slowly remove the supernatant without disturbing the cell pellet.

Note: To avoid aspirate cells, it can leave 100 μL supernatant in this step.

• 41.Rinse the cells once with cold 0.01% BSA-HBSS solution and centrifuge at 500 g for 5 min at 4°C. Carefully aspirate the supernatant without disturbing the cell pellet.
• 42.Based on the number of the sorted cells, resuspend the cells by using 0.01% BSA-HBSS to a concentration of 700–1200 cells/μL. The resuspension volume should be determined based on the sorted cell number. For example, if the FACS count shows 100,000 cells, resuspend the cell pellet with 50–60 μL 0.01% BSA-HBSS solution.
• 43.To evaluate cell viability and concentration, aspirate 5 μL of cell suspension with 5 μL of 0.4% (weight/volume) Trypan Blue Solution (Thermo, 15250061) and count the number using hemocytometer under bright field microscope. Dead cells will appear dark blue.
• 44.Store the cell suspension on ice.

CRITICAL: Maintaining high cell viability is essential for scRNA-seq, therefore it is important to keep the sample on ice during waiting periods and centrifuge always at 4°C.

Figure 2.

FACS sorting strategy for single-cell isolation from solid tumors in this study

(A and B) All lived cells and non-immunocytes were enriched by sorting PI^- (A) and PI^-/CD45^- (B), respectively.

Droplet-based single-cell RNA-Seq

Timing: ∼2 days

In this section, we will perform the single cell RNA-seq by using the 10× genomics Chromium Next GEM Single Cell 3ʹ GEM, Library & Gel Bead Kit (v3).

• 45.After performing cell counting, proceed droplet-based single-cell RNA-Seq using Chromium Next GEM Single Cell 3ʹ GEM, Library & Gel Bead Kit (v3). Detailed information on the protocol can be found at https://www.10xgenomics.com/support/single-cell-gene-expression/documentation/steps/library-prep/chromium-single-cell-3-reagent-kits-user-guide-v-3-1-chemistry-dual-index.
• 46.According to the cell suspension volume calculator table aspirate indicated number of cells and mix them with the reagents. The capture target is approximately 5,000–8,000 cells per sample.

CRITICAL: If the total cell number falls short of the requirement, centrifuge the cell suspension again, remove the supernatant to adjust the final volume to approximately 42 μL, and load all the cells for scRNA-seq.

CRITICAL: The number of loaded cells should not exceed 10,000 cells. Overloading can result in an increase in doublets, with a rate of 0.8% for every 1,000 cells.

• 47.Follow the manual for Chromium Controller to load cell-reagents mix, partitioning oil, and Gel Beads onto the Chromium Chip B and then perform reverse transcription of poly-adenylated mRNA.
• 48.Amplify cDNA and construct the libraries according to the instrument of 10 × Genomics Single Cell Reagent Kit (v3) instructions.
• 49.After library construction, quantify the library concentration and check its quality using a Bioanalyzer 2100 (Agilent) with a High Sensitivity DNA kit.
• 50.Proceed to next-generation sequencing using a mainstream second-generation sequencer, such as the NovaSeq 6000 platform (Illumina) or any other equivalent system at the read depth of 80,000–100,000 reads per cell.

Processing the single-cell RNA-seq data

Timing: 3 days

In this section, we will process the raw data of single-cell RNA-seq.

• 51.Sequence mapping: Align sequence reads to the GRCh38 genome reference and quantify gene expression using the cell ranger software (v.3.0.0, 10× Genomics).
• 52.Import the single cell RNA-seq gene expression data and create a seurat object for each sample. Since the cell barcode of different samples overlap, cell names need to be modified to ensure uniqueness.

install.packages("Seurat")

library(Seurat)

seuread <- function(dir, projectname) {

 data <- Read10X(data.dir = dir)

 colnames(data) <- paste(projectname, colnames(data), sep =

"_")

 seurat <- CreateSeuratObject(counts = data, project =

projectname, min.cells = 3, min.features = 200)

 return(seurat)

}

# data import

Seu.p1hpv <- seuread(dir = "/outs/filtered_feature_bc_matrix/",

projectname = “P1_HPV”)

CRITICAL: Before data import, set the working directory to cellranger output result location, and read the gene expression data from “/outs/filtered_feature_bc_matrix/” folder.

• 53.Merge the data from all samples by importing each single-cell RNA-seq gene expression dataset into a separate Seurat object, following the steps outlined in Step 52. Once all datasets have been imported, combine them into a single Seurat object using the function "merge".

Seu.p1c <- seuread(dir = "/outs/filtered_feature_bc_matrix/",

projectname = “P1_C”)

# Read samples one by one. .....

seu.all <- merge(x = p1hpv, y =c(Seu.p1c, ...))

Note: In this protocol, data set used for sc-RNA-seq analysis could be obtained from National Genomics Data Center (NGDC) under GSA-human: HRA003672 (https://ngdc.cncb.ac.cn/gsa-human).¹ The combined gene expression matrix of all samples and cell meta data information are available at NGDC under OMIX: OMIX002487 (https://ngdc.cncb.ac.cn/omix/releaseList/). Users can download the combined gene expression matrix and build merged Seurat Objects directly.

• 54.Cell filtering: Cells and genes were evaluated based on three criteria: (1) Cells with gene expression counts over 7,500 or under 200 were excluded, as low-quality cells often have few genes and doublets may show a high gene count. (2) Cells with more than 25% mitochondrial counts were filtered, as dying cells may have high mitochondrial contamination. (3) Genes expressed in fewer than three cells within a sample were removed.

seu.all <- PercentageFeatureSet(seu.all, pattern = "ˆMT-", col.name

= "percent.mt")

seu.all <- subset(x = seu.all, subset = nFeature_RNA > 200 &

nFeature_RNA < 7500 & percent.mito < 25)

• 55.Data normalization and identification of highly variable features: The data were normalized using the "LogNormalize" method, with a scaling factor of 10,000 as the default. Highly variable features were identified using the "FindVariableFeatures" function, with a mean-variance relationship approach.

seu.all <- NormalizeData(object = seu.all, normalization.method =

"LogNormalize", scale.factor = 1e4)

## 3 Detection of variable features across the single cell

seu.all <- FindVariableFeatures(object = seu.all, selection.method

= 'mean.var.plot', mean.cutoff = c(0.0125, 3), dispersion.cutoff

= c(0.5, Inf))

• 56.Scale the data and remove unwanted sources of variation. To remove technical noise and batch effects, consider specifying variables to regress using the "vars.to.regress" parameter.

seu.all <- ScaleData(object = seu.all, vars.to.regress =

c("nCount_RNA", "percent.mt"))

• 57.Perform dimensionality reduction and cluster analysis on cells derived from blood circulation, primary, and metastatic lesions. To avoid overfitting, only consider highly variable genes for the "RunPCA" function. Then, determine the optimal number of principal components to be used for the input to the 'FindClusters' function based on the Elbow plot analysis. Select the first 50 principal components for the example dataset in this protocol. Note that the pc and resolution parameters should be adjusted based on the size of the dataset.

# Perform linear dimensional reduction

seu.all <- RunPCA(object = seu.all, features =

VariableFeatures(object = seu.all), verbose = FALSE)

# Determine statistically significant principal components

print(ElbowPlot(object = seu.all, ndims = 50))

# cluster cells

pc <- 40

res <- 1

seu.all <- FindNeighbors(object = seu.all, dims = 1:pc)

seu.all <- FindClusters(object = seu.all, resolution = res)

# tSNE

seu.all <- RunTSNE(object = seu.all, dims = 1:pc)

• 58.Identify cluster-specific markers using a comparison of each cluster to all other cells, and report only those with positive expression.

seu.all.markers <- FindAllMarkers(object = seu.all, only.pos = TRUE,

min.pct = 0.25, logfc.threshold = 0.25)

• 59.Save data as a separate R object for further processing.

saveRDS(object = seu.all, file = "seuObject.rds")

Identification of CTC from the blood sample

Timing: >1 day

In this section, we will provide the procedures and codes for identifying CTC from the blood samples.

• 60.As circulating tumor cells only exist in the blood, it is necessary to identify and analyze them separately. To achieve this, data from the blood samples need to be specifically extracted from all the collected samples. Subsequently, a dimensionality reduction and cluster analysis need to be performed again (Figure 3A).

seu.b <- subset(x = seu.all, subset = group_tissue == "Blood")

seu.b <- NormalizeData(object = seu.b, normalization.method =

"LogNormalize", scale.factor = 1e4)

seu.b <- FindVariableFeatures(object = seu.b, selection.method =

'mean.var.plot', mean.cutoff = c(0.0125, 3), dispersion.cutoff =

c(0.5, Inf))

seu.b <- ScaleData(object = seu.b)

seu.b <- RunPCA(object = seu.b, features = VariableFeatures(object

= seu.b), verbose = FALSE)

seu.b <- FindNeighbors(object = seu.b, dims = 1:40)

seu.b <- FindClusters(object = seu.b, resolution = 1)

seu.b <- RunTSNE(object = seu.b, dims = 1:40)

• 61.Distinguish CTCs from blood samples by marker gene expression. Firstly, PTPRC-negative cells in Blood samples are considered potential CTCs. Then marker genes of epithelial cells (KRT8), pancreatic cancer cells (CD9), and platelet genes (PPBP, PF4) were used for further calibration. PTPRC^-CD9⁺PPBP⁺ cells of Blood samples were preliminarily defined as CTC cells (Figure 3B).

genes <- c("PTPRC", "CD9", "KRT8", "PPBP", "PF4")

FeaturePlot(seu.b, features = genes, reduction = "tsne")

CRITICAL: Marker genes for malignant epithelial cells can vary across different cancer types. It is important to use markers specific to the tumor type of interest, which can be found in the latest published literature or databases such as CellMarker.

• 62.Distinguish CTC from blood samples by CNV analysis. Here copykat was used to infer the copy number variation of cells in Blood samples. Due to the large difference in the cardinality of the result of copy number variation among different patients, each patient was analyzed separately here, and the analysis results were combined (Figure 3C).

library(copykat)

for (i in unique(unique(seu.b$group_patient))) {

 raw.ctc <- subset(seu.b, subset = group_patient == i)

 exp.rawdata <- as.matrix(raw.ctc@assays$RNA@counts)

 copykat.test <- copykat(rawmat=exp.rawdata,

    id.type="S",

    ngene.chr=1,

    win.size=25,

    KS.cut=0.1,

    sam.name=i,

    distance="euclidean",

    norm.cell.names="",

    n.cores=20)

}

CRITICAL: The raw gene count is used as the input file here, and the “n.cores” parameter should be adjusted based on the server situation. If the gene name format of the input file is not symbol, the “id.type” parameter needs to be adjusted. For details about the adjustment mode, visit “https://github.com/navinlabcode/copykat“.

Note: Cells with more abnormal copy number events were defined as malignant cells. Combine the marker gene definition results in step 61. CTCs were eventually identified from the Blood sample originated cells.

• 63.Visual t-SNE dimension reduction result according to the cell types (Figure 3D).

Figure 3.

Identifying CTCs from HPV-derived cells by scRNA-seq

(A) The t-SNE plot displays all cells from HPV blood samples based on their respective clusters.

(B) The dot plot illustrates the expression levels of marker genes in each cluster. The color reflects the z-score of normalized expression, and dot size corresponds to the percentage of marker gene positive cells in each specific cluster.

(C) The heatmap exhibits the inferred CNV results of CTCs and immunocytes in blood circulation, with immunocytes serving as the normal cell control. The chromosome scale is labeled at the top.

(D) The t-SNE plot presents all cells from HPV blood samples based on cell type.

Identify cell types

Timing: >1 day

In this section, we will perform the reduction and cluster analysis for the cells from primary and metastatic lesions, separately.

• 64.Coarse cell typing: Determine the cell type of each cluster based on the expression of key gene markers (Table 1). Cells are divided into eight large sub-types: Epithelial, Fibroblast, Endothelial, CTC, B cells, Myeloid, NK, and T cells (Figures 4A and 4B).
• 65.Copykat is used to infer the copy number variation of cells in primary and metastatic samples as described in step 62. All the cells in the primary and metastatic samples from each patient are analyzed (Figures 4C and 4D).
• 66.Immune cell typing: All the immune cells identified in step 64 is subjected to dimensionality reduction and cluster analysis, as described in step 57. Then, immune subtypes are defined based on the expression of key gene markers (Figures 4E–4G and Table 2).

Table 1.

Marker genes of each major cell type

Cell types	Genes
Epithelial	EPCAM, CDH1, MUC1, KRT8
Fibroblast	FAP, COL1A1, DCN
Endothelial	VWF, CDH5, ENG, PECAM1
CTC	PTPRC, CD9, TIMP1, PPBP, PF4
B cells	CD79A, CD79B, MS4A1
Myeloid	AIF1, CD14, LYZ, FPR1
NK	KLRF1, KLRD1, FGFBP2, NKG7, GNLY
T cells	CD3D, CD3E, CD3G

Figure 4.

The single-cell transcriptional atlas of PDAC primary tumors and metastatic lesions

(A and B) The UMAP plots present all sequenced cells based on their respective cell types (A) and tissue origins (B).

(C and D) The UMAP plots present epithelial cells based on the CopyKat prediction results (C) and tissue origins (D).

(E and F) The UMAP plots present all immune cells based on cell type (E) and tissue origins (F).

(G) The dotplot shows the expression level of marker genes in each immune cell type. The color indicates z-score of normalized expression, and dot size represents the percentage of marker gene positive cells.

Table 2.

Marker genes of each immune cell sub type

Cell types	Genes
NK/NKT	KLRB1, KLRF1, KLRD1, NCAM1, CD3D, CD3E, CD3G
CD8 Ex	CD8A, LAG3, TIGIT, CTLA4, PDCD1
CD8 EFF	GZMA, GZMK
Treg	FOXP3, IL2RA, IKZF2
Memory T	IL7R, LTB
Naive T	CCR7, TCF7, LEF1
B cell	CD79A, CD79B, MS4A1
Neutrophil	FCGR3B, FPR1
Monocyte	CD14, FCN1
M1	CD68, ITGAX, ITGAM, CD86, IL1B
M2	CD163, MRC1, MSR1
cDC	CD1C, FCER1A, CLEC10A
pDC	LILRA4, CCDC50, IL3RA
Mast cell	MS4A2, TPSAB1, KIT

Cell-cell communication analysis

Timing: >4 h

In this section, we will analysis the cell-cell interactions between tumor cells (CTC) and immunocytes, and provide the relevant codes.

• 67.Generate the input file required for CellPhoneDB in R. We applied CellPhoneDB (v2.0) to analyze the ligand-receptor interactions between tumor cells and immune cells. As the pancreatic primary tumor, portal venous blood, and liver metastatic lesion are located at different sites, we analyzed cells from Primary, Blood, and Metastatic samples separately. The input file should contain gene names as symbols. For more information on the gene name format, please refer to "https://github.com/Teichlab/cellphonedb".

library(Seurat)

library(data.table)

for (i in c("Primary", "Metastasis", "Blood")) {

 seu.sub <- subset(seu.all, subset = group_tissue == i)

 # cell type

 group <- seu.sub$label

 df <- as.data.frame(as.matrix(GetAssayData(seu.sub, slot =

"data")))

 df <- df[rowSums(df) > 10, ]

 Gene <- rownames(df)

 df <- cbind(Gene, df)

 celltype <- data.frame("Cell" = colnames(seu.sub), "Cell_type" =

group)

 dir.create(i)

 fwrite(df, file = paste0(i, "/cellphonedb_count.txt"), row.names

= F, col.names = T, sep = "\t", quote = F)

 fwrite(celltype, file = paste0(i, "/cellphonedb_celltype.txt"),

row.names = F, col.names = T, sep = "\t", quote = F)

}

• 68.Perform cell-cell communication analysis in Python. It is important to note that the input file used for the analysis should contain gene names as symbols. In this step, we used CellPhoneDB to identify ligand-receptor interactions between tumor cells and immune cells, which can provide insights into the complex cellular interactions within the tumor microenvironment. For more information on the analysis procedure and results interpretation, please refer to the CellPhoneDB documentation.

$ pip install cellphonedb

$ cellphonedb method statistical_analysis cellphonedb_celltype.txt \

> cellphonedb_count.txt --iterations=100 \

> --threads=12 --counts-data=gene_name --output-path=outs

Identifying the immune checkpoint molecules between tumor-cells/CTCs and immunocytes

Timing: 3 h

In this section, we will identify the immune checkpoint molecules between tumor cells (CTC) and immunocytes, and provide the relevant codes.

• 69.To determine the interaction weight score between cells, a cell-cell communication network analysis was performed, using the sum of the gene expression count of all significant ligand-receptor pairs between two cell types. Only significant ligand-receptor pairs containing the cell connect gene are retained (Figures 5A–5C). The cell connect gene file is available at "Github: https://github.com/Jinen22/scRNA-PDAC-CC/blob/main/Cellchat_cell_contact_gene_20211213.csv".

library(psych)

library(tidyverse)

# Take Primary for example

files = "./Primary/"

df.p = read.table(paste0(files, "/outs/pvalues.txt"), header=T,

stringsAsFactors = F, sep = '\t', comment.char = '', check.names=F)

df.m <- read.table(paste0(files, "/outs/means.txt"), header=T,

stringsAsFactors = F, sep = '\t', comment.char = '', check.names=F)

rownames(df.p) <- df.p$interacting_pair

rownames(df.m) <- df.m$interacting_pair

df <- read.csv(file = "Cellchat_cell_contact_gene_20211213.csv",

header = T)

keep <- rownames(df.p) %in% df$interaction_name

df.p <- df.p[keep, -(1:11)]

df.m <- df.m[keep, -(1:11)]

df.plot <- c()

for (cell in colnames(df.p)) {

 pval <- df.p[, cell]

 mea <- df.m[, cell]

 mea <- mea[pval < 0.05]

 df.plot <- c(df.plot, sum(mea))

}

names(df.plot) <- colnames(df.p)

name1 <- unlist(lapply(strsplit(names(df.plot), split = "[|]"),

function (x) x[1]))

name2 <- unlist(lapply(strsplit(names(df.plot), split = "[|]"),

function (x) x[2]))

df.plot <- data.frame(source = name1, target = name2, count =

df.plot)

Figure 5.

The cell-cell interaction between CTC/tumor cells and immunocytes in blood circulation, primary tumor, and metastasis lesion

(A–C) The potential cell-cell communication network based on the calculation of interactive ligands-receptors between tumor cells/CTCs and immunocytes in the primary tumor (A), blood circulation (B), and liver metastasis (C). The lines connect the cell types expressing cognate receptors or ligands, and the thickness of the line represents the number and expression level of ligand-receptor pairs.

(D–F) The dotplots present the expression levels of significant immune checkpoint-related ligand-receptor pairs between tumor cells/CTCs and immunocytes in the primary tumor (D), blood circulation (E), and liver metastasis (F). The color represents the mean expression of ligand and receptor genes, and the dot size represents the statistical significance of interactive molecular pairs. The p value presents - Log₁₀ (P value).

Detailed steps for data visualization are available at "https://github.com/Jinen22/scRNA-PDAC-CC/blob/main/Figure2.R".

• 70.
  Identifying the immune checkpoint molecules.

  • a.
    Import the CellPhoneDB output files “pvalues.txt” and “means.txt”in R software.
  • b.
    Import the file of the immune checkpoint related genes, the file is available at “https://github.com/Jinen22/scRNA-PDAC-CC/blob/main/cellphonedb_gene.csv”. Only immune checkpoint related interaction pairs were retained.
  • c.
    Filter low expression interaction pairs. Only significant immune check-point related interaction pairs (p < 0.05 & mean (Molecule 1, Molecule 2) > 0.5) were retained.

library(data.table)

cellorder <- c("B cell", "Naive T", "CD8 EFF", "CD8 Ex", "Memory

T", "Treg", "NKT", "NK", "cDC", "pDC", "M1", "M2", "Monocyte",

"Neutrophil", "Mast")

cellread <- function(files, tissue, cellorder) {

 df <- read.table(files, header=T, stringsAsFactors = F, sep =

'\t', comment.char = '', check.names=F)

 df <- df[, -c(1, 3:11)]

 df <- df[, c(1, grep(tissue, colnames(df)))]

 colnames(df)

 filter <- which(colnames(df) == paste0(tissue, "|", tissue))

 df <- df[, -filter]

 colorder <- c(colnames(df)[1], paste0(tissue, "|", cellorder),

paste0(cellorder, "|", tissue))

 colorder <- colorder[colorder %in% colnames(df)]

 df <- df[, colorder]

 return(df)

}

# pvalue

pvalue.p <- cellread("cellphonedb/Primary/outs/pvalues.txt", tissue =

"Primary", cellorder = cellorder)

pvalue.m <- cellread("cellphonedb/Metastasis/outs/pvalues.txt",

tissue="Metastasis", cellorder = cellorder)

pvalue.b <- cellread("cellphonedb/Blood/outs/pvalues.txt", tissue =

"CTC", cellorder = cellorder)

df.p <- full_join(pvalue.b, pvalue.p, by="interacting_pair")

df.p <- full_join(df.p, pvalue.m, by="interacting_pair")

df.p[is.na(df.p)] <- 1

mean.p <- cellread("cellphonedb/Primary/outs/means.txt", tissue =

"Primary", cellorder = cellorder)

mean.m <- cellread("cellphonedb/Metastasis/outs/means.txt",

tissue="Metastasis", cellorder = cellorder)

mean.b <- cellread("cellphonedb/Blood/outs/means.txt", tissue =

"CTC", cellorder = cellorder)

df.m <- full_join(mean.b, mean.p, by="interacting_pair")

df.m <- full_join(df.m, mean.m, by="interacting_pair")

df.m[is.na(df.m)] <- 0

## keep immune interaction pairs

df <- read.csv(file = "cellphonedb_gene.csv", header = T)

gene.group <- "immune"

genes <- unique(df$genes[df$group == gene.group])

keep <- c()

for (i in genes) {

 pair <- grep(i, df.p$interacting_pair)

 keep <- c(keep, pair)

}

keep.row <- unique(keep)

# 2 filter

df.p.use <- df.p[keep.row, -1]

df.m.use <- df.m[keep.row, -1]

CRITICAL: The gene list in step 70 is summarized from the latest published research articles and reviews.^6,7,8,9 The deadline for the collection is 2022, so some novel immune checkpoints may be missed. If some new immune checkpoint molecule pairs are identified in the future, it can be added to the gene list and followed up for analysis.

• 71.Visualize the results of the remaining immune check-point related interaction pairs in R (Figures 5D–5F). The script including the detailed steps is available at “https://github.com/Jinen22/scRNA-PDAC-CC/blob/main/Figure 2.R”.

Expected outcomes

Following this protocol, the tumor-immunocyte interactions in blood circulation and solid (primary and metastatic) malignant lesions can be dissected at single-cell scale. By processing steps 1–19, CTCs from blood samples of PDAC metastatic patients can be efficiently isolated and subjected to scRNA-seq analysis. After immunofluorescence staining, the CTC can be identified when they are positive for nuclear (Hoechst 3342⁺), EpCAM, and negative for CD45 (Figure 1). Commonly, several CTCs can be detected per milliliter of HPV blood from patients with metastatic PDAC. However, due to the high heterogeneity of PDAC patients, it is difficult to predict the amount of CTC in a blood sample before the trial. Following steps 20–42, approximately 300,000–500,000 FACS-sorted cells can be obtained from PDAC primary and metastatic lesions for scRNA-seq (Figure 2). After scRNA-seq, CTCs are expected to be identified from blood samples by canonical cell markers and CNV variability at the transcriptome level (Figure 3). This protocol is also expected to profile the cellular ecosystem in PDAC primary, CTC, and metastatic lesions at the single-cell transcriptome level (Figure 4).

Importantly, this protocol also provides scripts for the identification of immune checkpoint molecules between tumor cells and immunocytes in primary tumors, circulating blood, and metastatic lesions. As expected, we identified several unique immune checkpoints between CTCs and immunocytes in the blood circulation, such as CD94-NKG2A: HLA-E, CD94-NKG2C: HLA-E, SIPRA_CD47, LGALS9_CD47, etc. (Figure 5).

Limitations

The number of CTCs and cell viability are important factors to consider in scRNA-seq. However, the number of CTCs varies from patient to patient due to the severity of metastasis and heterogeneity of patients. To ensure acquire adequate CTCs, we recommend collecting no less than 10 mL of HPV blood from PDAC liver metastatic patients. To improve CTC viability, we recommend isolating them from fresh blood and keep the samples on ice or at 4°C throughout the protocol. For primary and metastatic tumor samples, we process the tissue within 1.5 h and encapsulate the cells immediately after FACS sorting.

Another limitation is that although cell-cell communication analysis identifies immune checkpoint molecular pairs between CTCs and immunocytes, it cannot confirm whether these pairs enable CTCs to evade immune surveillance. Validating the biological function of these immune checkpoints in vivo is necessary to understand their role in mediating immune escape and metastasis of CTCs. Furthermore, investigating the expression levels of paired receptors and ligands through immunofluorescence staining can validate realistic physical interactions. However, it is beyond the scope of this protocol to investigate the function of each immune checkpoint molecular pair on CTCs.

Troubleshooting

Problem 1

There are not enough CTC to perform scRNA-seq (related to steps 1–19).

Potential solution

• •The CTCs are captured by EpCAM and CA199 antibodies coated on the microfluidic chip. However, some CTCs may not bind strongly to these antibodies, and during the washing process (steps 9–11), the washing buffer may flush away loosely bound CTCs. Therefore, we propose reducing the wash volume and rate in steps 9 and 11 to minimize CTC loss. Additionally, capturing CTCs with larger blood samples can increase the chances of capturing more CTCs.
• •The number of collected CTCs is often small, and the cell pellet may not be visible after centrifugation. To reduce CTC loss, we recommend using a horizontal rotor and leaving a small amount of supernatant in the tube instead of removing all of it. This technique can also be applied during all centrifugation steps (such as steps 18, 27, 31, and 40) to minimize cell loss.

Problem 2

Cells are sedimented in the syringe during the process of CTC capture (related to step 8).

Potential solution

There are several potential strategies to prevent cell sedimentation during the CTC capture process. One option is to position the syringe vertically to maintain cell suspension. Alternatively, using multiple microfluidic chips in parallel can reduce capture time and prevent cell deposition.

Problem 3

CTCs are not eluted effectively in the process of elution (related to steps 12–14).

Potential solution

A potential strategy increasing is to increase the elution volume or flow rate.

Problem 4

Poor tissue dissociation efficiency (related to steps 21–26).

Potential solution

Pancreatic cancer tissues are characterized by a high abundance of extracellular matrix and fibroblasts, which can impede the efficiency of tissue dissociation and enzymatic digestion. Potential strategies for improving dissociation efficiency include increasing the concentration of digestion enzymes or extending digestion time. However, these approaches may negatively impact cell viability. Another effective approach is to cut the tumor tissue into smaller pieces, which can enhance the dissociation efficiency while minimizing the detrimental effects on cell viability.

Problem 5

Low cell viability after cell dissociation and FACS sorting (related to steps 31 and 41).

Potential solution

Primary and metastatic lesions-derived cells often suffer from low cell viability after cell dissociation and FACS sorting due to various reasons. To enhance the viability of these cells, several strategies can be implemented. Firstly, the timespan from excision to processing in the laboratory should be minimized. Secondly, the samples should be kept on ice throughout the procedure. Thirdly, immediate processing of the biospecimens after surgery is recommended. Fourthly, sorting gates should be set strictly to exclude dead cells, debris, and doublets during FACS sorting. Additionally, reducing the sample processing and FACS sorting time could further improve cell viability.

Problem 6

There is a limited amount of expressed epithelial cell genes in CTCs due to the epithelial-interstitial transition (EMT) process and the limitations of the drop-based single-cell library construction method, which may result in a dropout of gene expressions (related to steps 61 and 62).

Potential solution

To detect epithelial genes in CTCs, it is recommended to increase the sequencing depth or utilize the smart-seq method. In addition, considering the close association of CTCs with platelets,¹⁰ it may be beneficial to incorporate platelet marker genes into the marker gene panel for CTC detection. Although CTCs undergo EMT, they are still malignant cells, and their copy number variation events differ from those of normal cells in the blood. To assist in defining CTCs, inferCNV can be used to infer copy number variation in CTCs.

Problem 7

The results obtained from the interaction network analysis were inconsistent with the direct output of CellPhoneDB (related to step 69).

Potential solution

The default output of CellPhoneDB calculates the number of interacting molecular pairs between each cell subtype in the cell communication network analysis. However, this approach may not accurately reflect true cell communication since gene expression levels significantly affect molecular pair functional performance. To address this issue, we propose to calculate the interaction weight score, which is the sum of the gene expression count of all significant ligand-receptor pairs between two cell types. This approach considers gene expression levels and provides a more accurate representation of cell communication.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Hubing Shi (shihb@scu.edu.cn).

Materials availability

All reagents generated in this study are available from the lead contact upon request with a completed Materials Transfer Agreement.

Data and code availability

• •The raw sequencing data for this study have been deposited at National Genomics Data Center (NGDC, https://ngdc.cncb.ac.cn/) under GSA-human: HRA003672 (https://ngdc.cncb.ac.cn/gsa-human), processed data of scRNA have been deposited at NGDC under OMIX: OMIX002487 (https://ngdc.cncb.ac.cn/omix/releaseList/). Accession numbers are listed in the key resources table.
• •The original code for visualizing scRNA-seq data has been on GitHub and is publicly available: https://github.com/Jinen22/scRNA-PDAC-CC. DOIs are listed in the key resources table.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.