Protocol for mapping T cell activation using single-cell RNA-seq

Summary

Stimulation of CD4 T cells with anti-CD3/CD28 is a commonly used model to study T cell activation. Here, we present a protocol for investigating T cell activation based on anti-CD3/CD28 bead stimulation and single-cell RNA sequencing (scRNA-seq). We describe the workflow from the isolation of human peripheral blood mononuclear cells (PMBCs) and CD4 T cell enrichment to anti-CD3/CD28 bead stimulation, scRNA-seq, and data analysis.

For complete details on the use and execution of this protocol, please refer to Li et al.¹

Graphical abstract

Highlights

• •Steps for CD4 T cell isolation from human whole blood using a density gradient medium
• •Procedures for anti-CD3/CD28 microbead-based CD4 T cell activation
• •Instructions for single-cell RNA library construction
• •Guidance for data integration and downstream analysis of scRNA-seq data

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

Stimulation of CD4 T cells with anti-CD3/CD28 is a commonly used model to study T cell activation. Here, we present a protocol for investigating T cell activation based on anti-CD3/CD28 bead stimulation and single-cell RNA sequencing (scRNA-seq). We describe the workflow from the isolation of human peripheral blood mononuclear cells (PMBCs) and CD4 T cell enrichment to anti-CD3/CD28 bead stimulation, scRNA-seq, and data analysis.

Before you begin

T cells are essential components of adaptive immunity, playing a key role in the clearance of pathogenic cells and the regulation of autoimmune disease.^2,3,4,5,6 The T cell response is initiated when the T cell receptor recognizes an antigen-loaded MHC molecule, triggering rapid differentiate and proliferate of effector T cells (T_EFF) that recognize the same antigen, a process termed T cell activation. This step is crucial for establishing adaptive immune responses against pathogens. There are several studies focusing on the activation of naïve T cell (T_N). However, in the human peripheral blood, CD4 T cells comprise diverse subsets except for T_N, including central memory T cells (T_CM), effector memory T cells (T_EM) and regulatory T cells (T_reg).^4,7,8,9 The response of CD4 T cells to antigenic stimulation varies due to their intrinsic heterogeneity. For example, T_CM and T_EM are able to proliferate and secrete cytokines upon stimulation. ISAG^high T cells highly express IFN signaling–associated genes and have antiviral properties.¹⁰ HSP^high T cells highly express heat shock proteins and are derived from naive T upon activation.¹ Advances of single cell RNA-sequencing (scRNA-seq) have revealed the dynamics and influencing factors of T cell activation at single-cell resolution.^1,7,11,12,13

Here, we provided a protocol for isolation of human CD4 T cells from the peripheral blood and investigation of T cell activation based on anti-CD3/CD28 stimulation using single-cell RNA-seq. Firstly, peripheral blood mononuclear cells (PMBCs) are isolated from whole blood using density gradient medium Lymphoprep. CD4 T cells are subsequently purified from PBMCs using magnetically labeled CD4 MicroBeads. T cell activation is achieved through in vitro stimulation using Dynabeads human T activator CD3/CD28 for 18 h. Single cell RNA-seq is performed on both resting CD4 T cells and stimulated CD4 T cells. This protocol typically processes 1.6×10⁴ cells per sample for sequencing. The data analysis pipeline includes quality control, normalization, data integration, dimensionality reduction, cell clustering and identify subset-specific genes. Differential expression genes (DEGs) analysis and gene ontology (GO) analysis are conducted to infer the transcriptional alterations and underlying biological significance after T cell activation.

The advantages of this protocol include: (1) the experimental procedures are streamlined with readily available reagents, enhancing both feasibility and success rate of the experiments; (2) the analytical pipeline is straightforward, facilitating efficient data processing. In summary, we provided a systemic and reproducible method for studying the dynamics of CD4 T cell activation at a single-cell level, which is crucial for the understanding of T cell mediated immunity.

Institutional permissions

Handling patient related experiment requires adherence to local institutional guidelines for laboratory safety and ethics. This study was approved by IRB at Southern University of Science and Technology (SUSTech). The human peripheral blood samples used in this study were obtained from two healthy adult donors. All experiments were conducted following the protocols approved by IRB at SUSTech.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-human CD25-PE (clone: BC96)	eBioscience	Cat #12-0259-42; 1:200
Anti-human CD3-FITC (clone: SK7)	BioLegend	Cat #306510; 1:200
Anti-human CD4-BV510 (clone: RPA-T4)	BioLegend	Cat #300545; 1:200
Anti-human CD45-APC (clone: HI30)	BioLegend	Cat #982304; 1:200
Dynabeads Human T-activator CD3/CD28	Thermo Fisher Scientific	Cat #11131D
CD4 MicroBeads, human	Miltenyi Biotec	Cat #130-045-10
Chemicals, peptides, and recombinant proteins
Lymphoprep	STEMCELL	Cat #07851
MACS BSA Stock Solution (cell culture grade)	Miltenyi Biotec	Cat #130-091-376
UltraPure 0.5 M EDTA, pH 8.0 (cell culture grade)	Invitrogen	Cat #15575020
PBS−/−, pH 7.4 (cell culture grade)	Gibco	Cat #C10010500BT
RPMI-1640 (cell culture grade)	Gibco	Cat #11835830
FBS (cell culture grade)	Gibco	Cat #A5669701
Penicillin/Streptomycin, 100× (cell culture grade)	Gibco	Cat #15140122
Critical commercial assays
Chromium Single Cell 3′ Library & Gel Bead Kit v2, 16 rxns	10× Genomics	Cat #PN-120237
Other
Pipette tips, 10 μL	Rainin	Cat #30520478
Pipette tips, 200 μL	Rainin	Cat #30520439
Pipette tips, 1 mL	Rainin	Cat #30520440
Microcentrifuge tube	LABSELECT	Cat #MCT-001-150
Centrifuge tube, 15 mL	LABSELECT	Cat #CT-002-15A
Centrifuge tube, 50 mL	LABSELECT	Cat #CT-002-50A
96-well cell culture plate	NEST	Cat #701001
Deposited data
Raw and processed scRNA-seq data	This paper	HRA002777
Code	This paper	https://github.com/JinWLab/T_active
Software and algorithms
Cell Ranger v3.0	10× Genomics	https://github.com/10XGenomics/cellranger
Seurat v3.0	Stuart et al.14	https://satijalab.org/seurat/index.html
R	R Core Team	https://www.r-project.org/
ggplot2	Auguie	https://ggplot2.tidyverse.org/
Ggrepel	CRAN	https://github.com/slowkow/ggrepel
Dplyr	CRAN	https://dplyr.tidyverse.org/
Monocle3	Trapnell et al.15	https://cole-trapnell-lab.github.io/monocle3/

Materials and equipment

Timing: 10 min

MACS buffer

Reagent	Final concentration	Amount
MACS BSA Stock Solution	0.5% (w/v)	5 mL
0.5 M EDTA, pH 8.0	2 mM	0.4 mL
PBS−/−, pH 7.4	1×	94.6 mL
Total	N/A	100 mL

Stock at 2°C–8°C for up to 1 month.

The MACS BSA Stock Solution consists of PBS supplemented with 10% bovine serum albumin. Typically, 5 mL of MACS buffer is needed for the magnetic beads-based T cell purification from 10⁷ of PBMCs.

Beads washing buffer

Reagent	Final concentration	Amount
MACS BSA Stock Solution	0.1% (w/v)	1 mL
0.5 M EDTA, pH 8.0	2 mM	0.4 mL
PBS−/−, pH 7.4	1×	98.6 mL
Total	N/A	100 mL

Stock at 2°C–8°C for up to 1 month.

Typically, 5 mL of Beads Washing buffer is needed for the beads-based T cell activation (10⁶ to 10⁷ of purified T cells).

Culture medium

Reagent	Final concentration	Amount
FBS	10% (v/v)	10 mL
Penicillin/Streptomycin, 100×	1×	1 mL
RPMI-1640	1×	89 mL
Total	N/A	100 mL

Stock at 2°C–8°C for up to 3 months.

2% FBS

Add 2 mL of FBS in 98 mL PBS^−/−. Stock at 2°C–8°C for up to 3 months.

Typically, 10–15 mL of 2% FBS is needed for PBMCs isolation from 5 mL of whole blood.

Alternatives: This protocol uses commercial chemicals including 0.5M EDTA (pH 8.0) and PBS (pH 7.4) to make sure the consistence of pH value. The components of these chemicals are provided in the following table. Users can also prepare these solutions using this recipe.

Name	Components
PBS, pH 7.4	1.06 mM KH2PO4, 155.17 mM NaCl, 2.97 mM Na2HPO4·7H2O, pH adjusted to 7.4 using HCl
0.5 M EDTA, pH 8.0	0.5 M Na2EDTA, pH adjusted to 8.0 using NaOH

The following equipment is required:

Centrifugation:

Magnet: (1) MiniMACS Separators or OctoMACS Separators from Miltenyi.

    (2) DynaMag-2 or any magnet with tube racks.

Library QC: Bio-fragment Analyzer or other systems such as Caliper LabChip and Agilent TapeStation.

Cell purity/status examination (optional): Flow Cytometer (BD FACSCanto SORP or other multicolor flow cytometry analysis systems with FITC, PE, APC and BV510 channels).

CRITICAL: Perform regular checks of equipment following the institutional guidelines.

Step-by-step method details

Isolation of human PBMCs

Timing:1h

The following steps describe how to isolate PBMCs from human whole blood using a density gradient medium Lymphoprep in accordance with the user guideline. We point out the precaution to ensure the success of the experiment. The protocol is starting from 5 mL blood and the blood was collected in a sodium heparin vacutainer tube.

• 1.Withdraw 5 mL of Lymphoprep using a syringe, and transfer to a clean 50 mL Falcon.
• 2.Dilute 5 mL blood with 5 mL of 2% FBS.
• 3.Layer 10 mL of diluted blood on top of 5 mL Lymphoprep using a sterile dropper in a dropwise manner, avoiding to disturb the blood:Lymphoprep interface.

Note: Volumes can be adjusted depending on the amount of blood. See Table 1 for the recommended volumes.

• 4.Centrifuge the 50 mL Falcon at 800×g for 20 min at 20°C–25°C with brake off.
• 5.Discard the upper plasma layer without disturbing the plasma:Lymphoprep interface. Collect the PBMCs at the interface and transfer them to a new 15 mL centrifuge tube.

Note: PBMCs appear as a distinct white layer after centrifugation, as shown in the previous work.¹⁶ Carefully collect the PBMC layer without disturbing the erythrocytes/granulocyte pellet.

• 6.Pellet PBMCs by centrifugation at 300×g for 10 min at 4°C.
• 7.Remove the supernatant completely and resuspend the cell pellet with 1 mL MACS buffer. Pipette up and down for 3 times to generate single cell suspension.
• 8.Determine cell viability and cell number using a hemocytometer.

Note: A total of 10⁷ PBMCs is needed to further purification. Typically, 5 mL of whole blood can yield around 2 to 5×10⁷ of PBMCs.

Table 1.

Recommended volumes of blood and Lymphoprep

Blood (mL)	2% FBS (mL)	Lymphoprep (mL)	Tube size (mL)
1	1	1.5	5
2	2	3	14
3	3	3	14
4	4	4	14
5	5	10	50
10	10	15	50
15	15	15	50

This table is adopted from the manufacturer’s instrument.

Magnetic bead-based T cell enrichment

Timing: 1 h

The following steps describe how to enrich CD4 T cells from PBMCs using Miltenyi magnetically labeled beads in accordance with the manufacturer’s instruction. Cell purity can be determined using flow cytometry analysis after cell enrichment.

Alternatives: Fluorescence-Activated Cell Sorting (FACS) can be used to isolate CD4 T cells from PBMCs.

CRITICAL: Keep the cells cold, and pre-cool all the reagents to 4°C. All the procedures should be performed in a biosafety cabinet.

• 9.Resuspend 10⁷ of PBMCs with 80 μL of MACS buffer in a 1.5 mL Eppendorf.
• 10.Add 20 μL of CD4 Microbeads and mix well by pipetting up and down three times, and incubation for 15 min on ice.

Note: For cell number larger than 10⁷, scale up the volume of MACS buffer and Microbeads accordingly.

• 11.Add 500 μL of MACS buffer into the Eppendorf and mix well. Pass the cell suspension through a 40 μm strainer. Rinse the strainer with 500 μL of MACS buffer and combine the flow-through.

Note: Cell clumps will affect the performance of column purification. This step is to remove cell clumps before loading onto column.

• 12.Centrifuge at 300×g for 5 min at 4°C. Discard the supernatant completely.
• 13.Resuspend the cells with 500 μL of MACS buffer.
• 14.Place a MS column in a suitable MACS Separator, and rinse with 500 μL of MACS buffer.
• 15.Apply cell suspension onto the column after rinsing medium has completely pass through.
• 16.Wash the column with 500 μL of MACS buffer for three times.
• 17.Remove the column from the separator and place it on a suitable collection tube. Apply 1 mL of MACS buffer and immediately flush out the magnetically labeled cells using the column’s plunger (troubleshooting 1).
• 18.Determine cell viability and cell number using a hemocytometer.
• 19.
  Transfer 100 μL of cell suspension (around 10⁵ of CD4 T cells) to determine cell purity.

  • a.
    Add 2 μL of Fc receptor blocking solution and incubate for 5 min on ice.
  • b.
    Prepare and add an antibody cocktail as listed in Table 2, incubate for 20 min on ice.
  • c.
    Wash the cell by adding 500 μL of PBS and pipette up and down. Pellet the cells by centrifugation.
  • d.
    Resuspend the cell pellet with 200 μL of PBS. Add DAPI and incubate on ice for 1 min.
  • e.
    Analysis with flow cytometer (Figure 1A). Typically, purity is over 95%. If the purity is less than 95%, another round of enrichment is suggested.

Table 2.

Summary of data sources and experimental conditions for scRNA-seq data

Dataset	Stimulation	GEO accession ID	Data origin
Resting_1	no	-	generated in this study
Activated_DB1	stimulated with CD3/CD28 Dynabeads	-	generated in this study
Resting_2	no	GEO: GSM4450387	downloaded from GEO
Activated_DB2	stimulated with CD3/CD28 Dynabeads	GEO: GSM4450386	downloaded from GEO

Figure 1.

Flow cytometry analyses of CD4 T cells

(A) Gating strategies for CD4 T cells.

(B) Flow cytometry analysis of CD25 expression level in resting CD4 T cells and stimulated CD4 T cells.

T cell activation by anti-CD3/CD28 stimulation

Timing: 18 h

The following steps describe how to activate CD4 T cells using Dynabeads Human T-Activator CD3/CD28 in culture plate. The recommended cell number is 8×10⁴ cells per well of a 96-well plate.

CRITICAL: All the procedures should be performed in a biosafety cabinet to avoid contamination.

• 20.
  Prepare Dynabeads according to the manufacture’s instruction.

  • a.
    Resuspend the Dynabeads by vortex for 1 min.
  • b.
    Transfer 10 μL of Dynabeads to a 1.5 mL Eppendorf.
  • c.
    Add 1 mL of bead wash buffer and vortex for 5 s to mix completely.
  • d.
    Place the Eppendorf on a magnet for 1 min and discard the supernatant without disturbing the beads.
  • e.
    Remove the Eppendorf from the magnet and resuspend the beads with 5 μL of culture medium. Keep on ice before use.
• 21.Pellet the CD4 T cells by centrifugation at 300×g for 10 min. Discard the supernatant completely.
• 22.Resuspend the cell with culture medium and perform cell counting using a hemocytometer.
• 23.Dilute the cell to reach a final concentration of 400 cells/μL. Add 200 μL of cell suspension (8×10⁴ cells) per well of a 96-well tissue culture plate.
• 24.Add 2 μL (8×10⁴) pre-washed Dynabeads per well and mix completely.
• 25.Set one well of cells without Dynabeads as a negative control.
• 26.Incubate the tissue culture plate in a humidified CO₂ incubator at 37°C for 18 h, or adjust the duration according to the experiment design.
• 27.Exam the T cell’s statues post activation.

CRITICAL: It is important to exam the T cells’ statues after activation. In this protocol, we aim to achieve the maximal activated T cells. This protocol is standardized at 18 h of stimulation according to our previous studies.^1,4 We investigate the expression of CD25 (a marker for activated T cells) using flow cytometer (Figure 1B). Users can follow this framework to investigate the expression levels of other marker genes or T cell subsets of interest and to optimize stimulation time.

• 28.Pipette up and down to generate a single cell suspension after incubation.
• 29.Harvest cells with beads in a 1.5 mL Eppendorf and place the tube on a magnet for 1–2 min.
• 30.Transfer the supernatant containing activated CD4 T cells into a new Eppendorf. Centrifuge at 300×g for 10 min.
• 31.Resuspend the cell pellets with 0.5 mL of cold 1× PBS and perform cell counting using a hemocytometer (troubleshooting 2).
• 32.Quality control: check for cell viability, decries or cell aggregates under microscopes.

scRNA-seq library preparation and sequencing

Timing: 2 days

The following steps describe how to generate a 10x scRNA-seq library using 10× Genomics Chromium (v2) from stimulated T cells. All procedures strictly follow the recommended user guide (CG00052RevB).

• 33.Gently mix the single cell suspension from step 31.
• 34.Add 33.8 μL of single cell suspension (containing 1.6 × 10⁴ cells) to 66.2 μL of Master Mix.
• 35.Load 90 μL of the Master Mix with cell suspension without introducing bubbles.
• 36.Load the Master Mix with cell suspension, the Gel Beads and Partition Oil onto Chromium Next GEM Chip G.
• 37.The extracted RNA was reverse transcribed to generate cDNA.
• 38.Analyze the cDNA quality using a FragmentAnalyzer System (Figure 2A).
• 39.Break oil droplets and generate sequence libraries using the Chromium Single Cell 3′ Library Construction Kit v2 according to the recommended user guide.
• 40.Perform quality control of library using a FragmentAnalyzer System (Figure 2B).
• 41.Perform sequence on the Illumina NovaSeq 6000 with the final diluted pool at 2.5 pM, using the following sequencing run settings: 28 bp Read1, 8 bp I7 Index, 8 bp I5 Index and 91 bp Read2.

Figure 2.

Typical fragment size distribution of cDNA and library

(A) The fragment size distribution of cDNA.

(B) The fragment size of library. Fragment size is analyzed using Qsep 100.

Data processing and analysis

Timing: 5–7 days

The following steps outline the procedures for processing and analyzing scRNA-seq data derived from different sources, as specified in Table 2. “DB1” refers to scRNA-seq data generated using the protocol described in this manuscript. “DB2” indicates datasets produced by Ding et al.¹⁷ and were downloaded as RDS files from the GEO database. Datasets are categorized as “Resting” for non-stimulated cells and “Activated” for cells stimulated with CD3/CD28 Dynabeads. We first performed sequence alignment for the DB1 datasets. We than used Seurat (v3.0) to integrate DB1 and DB2, normalize data, reduce dimension and cluster cells. We performed downstream analyses on this integrated data.

Note: Sequence alignment of DB1 dataset was performed by generating a Feature Count Matrix. Reads were mapped to reference genome (hg19) using Cell Ranger v3.0, following standard protocols provided by 10× Genomics (https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger). Data from non-stimulated cells are stored in the directory named “cd4r_filtered_feature_bc_matrix”, and data from stimulated cells are stored in the directory named “cd4a_filtered_feature_bc_matrix”.

• 42.
  Process scRNA-seq data of DB1.

  • a.
    Read the data using “Read10×” function.
  • b.
    Create a Seurat project using “CreateSeuratObject” function.
  • c.
    Perform quality control: remove genes expressed by less than 5 cells, cells with less than 200 unique feature counts (nFeature_RNA) or mitochondrial counts larger than 7.5%.
  • d.
    Normalize the feature expression levels by the total expression for each cell using “NormalizeData” function.

>library (Seurat)

# Import 10x scRNA-seq data

>Resting_1.data <- Read10X(data.dir = "./cd4r_filtered_feature_bc_matrix/")

>Activated_DB1.data <- Read10X(data.dir = "./cd4a_filtered_feature_bc_matrix/")

# Create Seurat project and QC

>Resting_1 <- CreateSeuratObject(counts = Resting_1.data, project = "Resting_1", min.cells = 5)

>Activated_DB1 <- CreateSeuratObject(counts = Activated_DB1.data, project = "Activated_DB1", min.cells = 5)

>Resting_1[["percent.mt"]] <- PercentageFeatureSet(Resting_1, pattern = "ˆMT-")

>Resting_1 <- subset(Resting_1, subset = nFeature_RNA > 200 & percent.mt <7.5)

>Activated_DB1[["percent.mt"]] <- PercentageFeatureSet(Activated_DB1, pattern = "ˆMT-")

>Activated_DB1 <- subset(Activated_DB1, subset = nFeature_RNA > 200 & percent.mt < 7.5)

# Normalization

>Resting_1 <- NormalizeData(Resting_1, verbose = FALSE)

>Activated_DB1 <- NormalizeData(Activated_DB1, verbose = FALSE)

• 43.
  Integrate DB1 and DB2.

  • a.
    Load datasets of DB2.
  • b.
    Add metadata for DB1 and DB2 to indicate experimental condition and data source.
  • c.
    Integrate DB1 and DB2 using “FindIntegrationAnchors” and “IntegrateData” functions (troubleshooting 3).

# loading GEO datasets and updating Seurat objects

>Resting_2 <- readRDS("./GSM4450387_unstimulated_full_seurat.rds")

>Activated_DB2 <- readRDS("./GSM4450386_stimulated_full_seurat.rds")

>Resting_2<-UpdateSeuratObject(Resting_2)

>Activated_DB2<-UpdateSeuratObject(Activated_DB2)

# Add metadata to indicate data source

>Resting_1$sample <- "Resting_1"

>Activated_DB1$sample <- "Activated_DB1"

>Resting_2$sample <- "Resting_2"

>Activated_DB2$sample <- "Activated_DB2"

# Add metadata to indicate experimental condition

>Resting_1$activated <- "Resting"

>Activated_DB1$activated <- "Activation"

>Resting_2$activated <- "Resting "

>Activated_DB2$activated <- " Activation "

# Integrate DB1 and DB2

>cd4.anchors <- FindIntegrationAnchors(object.list = c(Resting_1,Activated_DB1) , dims = 1:20, anchor.features = 600)

>cd4.combined <- IntegrateData(anchorset = cd4.anchors, dims = 1:20)

Note: After this step, we have a Seurat object named “cd4.combined”, which containing scRNA-seq datasets of non-stimulated and stimulated T cell from DB1 and DB2.

• 44.
  Filter out cells of low quality, red blood cells and CD8 T cells.

  • a.
    Identify erythrocyte.
  • b.
    Identify CD8 T cells.
  • c.
    Filter using 'subset' function.

# ldentify erythrocyte

>HB.genes <- c('HBA1','HBA2','HBB','HBD','HBE1','HBG1','HBG2','HBM','HBQ1','HBZ')

>HB_m <- match(HB.genes, rownames(cd4.combined@assays$RNA))

>HB.genes <- rownames(cd4.combined@assays$RNA)[HB_m]

>HB.genes <- HB.genes[!is.na(HB.genes)]

>cd4.combined[["percent.HB"]]<-PercentageFeatureSet(cd4.combined, features=HB.genes)

# Identify CD8 T cells

>CD8.genes <- c('CD8A', 'CD8B')

>CD8_m <- match(CD8.genes, rownames(cd4.combined@assays$RNA))

>CD8.genes <- rownames(cd4.combined@assays$RNA)[CD8_m]

>CD8.genes <- CD8.genes[!is.na(CD8.genes)]

>cd4.combined[['percent.CD8']]<-PercentageFeatureSet(cd4.combined, features=CD8.genes)

# Filter

>cd4.combined <- subset(cd4.combined,subset = nFeature_RNA > 200 & nCount_RNA < 60000 & percent.CD8 < 0.01 & percent.HB<0.1)

>DefaultAssay(cd4.combined) <- 'integrated'

Note: Data quality, including the number of Features, count numbers, and the proportion of mitochondrial genes for each dataset is shown (Figure 3).

• 45.Identify the top 2000 high variable genes (HVGs):

cd4.combined <- FindVariableFeatures(cd4.combined, selection.method = 'vst', nfeatures = 2000, assay = 'RNA')

• 46.Perform data scaling to make each gene equal weight:

cd4.combined <- ScaleData(cd4.combined, verbose = FALSE)

Note: To avoid the domination of highly expressed genes in later analysis, data scaling is performed by setting the mean and variances of each feature to 0 and 1, respectively.

• 47.Perform principal component analysis (PCA) for linear dimension reduction using 'RunPCA' function and return 30 PCs. Investigating the contribution of the 30 PCs to the standard deviation through an elbow plot (Figure 4).

# PCA analysis and cell clustering

>cd4.combined <- RunPCA(cd4.combined, npcs = 30, verbose = FALSE)

>ElbowPlot(cd4.combined, ndims = 30)

Note: Although “ElbowPlot” was performed to infer the dimensionality of the data, we suggest exploring different number of PCs to choose suitable number of PCs for cell clustering.

• 48.Select top 20 PCs for cell clustering base on the observation through the elbow plot.

>cd4.combined <- FindNeighbors(cd4.combined, reduction = 'pca', dims = 1:20)

>cd4.combined <- FindClusters(cd4.combined, resolution = 1.4)

Note: Different resolutions would result in different number of clusters (Figure 5). Generally, setting resolution between 0.4 and 1.2 usually returns nice results for datasets of 3k cells.⁶ Higher resolution values lead to greater number of clusters. We recommend that users try different resolutions to achieve optimal clustering and biological insight.

• 49.Visualizing cells of different status (Figure 6A): DimPlot(cd4.combined, reduction = 'umap', group.by = 'activated', label = F, cols = c('#add8e6′,'#f08080'), pt.size = 0.1) + theme(legend.position = c(0.6,0.8)) + labs(title = element_blank()).
• 50.Define cell clusters using T cell subset specific marker genes from reference article^1,4 and visualization of scRNA-seq data on UMAP (Figures 6B and 6C).

# Annotation

>markers.to.plot <-c('HSPA6','HSPA1A','HSPB1', # HSPhi

'TUBB','TUBA1B','FABP5','MIR155HG', # proliferation T

'IL2','IFNG', # Th1

'IFNG','IL2','CCL20', # cytokine T

'IL2RA','CTLA4','FOXP3', #Treg

'GZMA','GZMK','CST7', #TEMRA

'IL7R') # TM

>DotPlot(cd4.combined, features = rev(unique(markers.to.plot)),dot.scale = 8, cols =c('gray','red'), col.min = 0) + RotatedAxis() + theme(axis.title=element_blank(),legend.text = element_text(size = 12), legend.title=element_text(size=12), axis.text = element_text(size=18), legend.position = 'right')

# Identifying the cell types based on marker genes

>celltype <- c('TN','TEM','TN','TCM','TCM','TCM','CTLA4hi TEFF','Treg','Conv TEFF','TEM','Conv TEFF','cytohi TEFF','TEMRA','Prolif TEFF','TEMRA','Treg','TEMRA','TN','TEMRA','HSPhi','ISAGhi','Prolif TEFF')

>names(a) <- levels(cd4.combined)

>cd4.combined <- RenameIdents(cd4.combined, a)

# Visualization

cd4.combined <- RunUMAP(cd4.combined, reduction = 'pca', dims = 1:30)

set3 = brewer.pal(n = 12,name = 'Set3')

set3 = c(set3, '#A6CEE3')

>DimPlot(cd4.combined, reduction = 'umap', label = F,cols = set3) + theme(legend.position='none')

• 51.Compare the distribution of cell subsets before and after stimulation (Figure 6D).

> library(ggalluvial)

>df <- table(Idents(cd4.combined), cd4.combined$activated)

>df[,1] <- df[,1]/21573 # convert the cell number to percentage of total cells

>df[,2] <- df[,2]/10242

>cell.prop<-as.data.frame(prop.table(df))

>names(cell.prop) <- c('cluster', 'sample', 'proportion')

>cell.prop[1:11,3] <- cell.prop[1:11,3]/sum(cell.prop[1:11,3])

>cell.prop[12:22,3] <- cell.prop[12:22,3]/sum(cell.prop[12:22,3])

>ggplot(data = cell.prop, mapping = aes(x = sample, y = proportion,fill=cluster,alluvium = cluster)) +

geom_bar(stat = 'identity', position = 'fill',alpha = 0.8, colour = 'black', size = 0.1, width = 0.6) +

theme_bw() + geom_alluvium(alpha = 0.4, colour = 'black', size = 0.1, width = 0.6) +

labs(x = '',y = 'Relative Abunance(%)',title = '') +

scale_fill_manual(values=set3) + RotatedAxis() +

scale_y_continuous(expand = c(0,0)) +

theme(axis.title = element_text(size = 13, face = 'bold'), axis.text = element_text(size=13), legend.text = element_text(size = 11))

• 52.Infer the pseudotime trajectories during T cell activation using monocle3 (Figure 6E).

>library(monocle3)

>library(ggplot2)

>library(patchwork)

>library(dplyr)

# create CDS obeject and pre-processing data

>sdata <- cd4.combined

>data <- GetAssayData(sdata, assay = 'RNA', slot = 'counts')

>cell_metadata <- sdata@meta.data

>gene_annotation <- data.frame(gene_short_name = rownames(data))

>rownames(gene_annotation) <- rownames(data)

>cds <- new_cell_data_set(data,

cell_metadata = cell_metadata,

gene_metadata = gene_annotation)

>cds <- preprocess_cds(cds, num_dim = 50)

>cds <- reduce_dimension(cds, preprocess_method = 'PCA')

# Load integrated UMAP from Seurat object

>cds.embed <- cds@int_colData$reducedDims$UMAP

>int.embed <- Embeddings(sdata, reduction = 'umap')

>int.embed <- int.embed[rownames(cds.embed),]

>cds@int_colData$reducedDims$UMAP <- int.embed

>cds <- cluster_cells(cds, reduction_method = 'UMAP')

#Learn the trajectory and order the cells in pseudotime

>cds <- learn_graph(cds)

>cds <- learn_graph(cds)

>cds <- order_cells(cds)

#Visualization

#p <- plot_cells(cds,

color_cells_by = 'pseudotime',

label_cell_groups=FALSE,

label_leaves=FALSE,

label_branch_points=FALSE,

graph_label_size=1.5)

ggsave('lineage.png',p,width = 6,height = 5)

• 53.Perform differential expression genes (DEGs) analysis between the “resting” and “activated” T subsets using 'FindMarkers' function. The genes with adjusted p value (p_val_adj) < 0.05 and log Fold Change (avg_logFC) > 1.5 were treated as DEGs. DEGs are visualized in the volcano plot (Figure 7A).

>Dat <- FindMarkers(object = cd4.combined , ident.1 = 'HSPhi', assay = 'RNA', logfc.threshold = 0)

>Dat <- cbind(allele = row.names(Dat), Dat)

>Dat$threshold <- factor(ifelse(Dat$p_val_adj < 0.05 & abs(Dat$avg_log2FC) >= 1.5, ifelse(Dat$avg_log2FC >= 1.5 , 'Up','Down'), 'NoSignifi'), levels = c('Up', 'Down', 'NoSignifi'))

>Dat[Dat$threshold == 'Up',] %>% write.csv(file = 'gene_hsphi.csv')

>g <- ggplot(Dat, aes(x = avg_log2FC, y = -log10(p_val_adj), color = threshold)) +

geom_point() +

scale_color_manual(values = c("#DC143C", "#00008B", "#808080")) +

theme_bw() + ylab('-log10 (p-adj)') + xlab('log2(FoldChange)') +

geom_vline(xintercept = c(-1.5,1.5), lty = 3, col = "black", lwd = 0.5) +

geom_hline(yintercept = -log10(0.05), lty = 3, col = "black", lwd = 0.5) +

ggtitle(label = 'HSPhi' )

>g <- LabelPoints(plot = g, points = Dat[Dat$threshold != 'NoSignifi', 'allele'], repel = T)

Note: We present an example of the differential expression genes (DEGs) analysis between resting and activated HSP^hi T cells.

• 54.Perform Gene Ontology (GO) enrichment analysis on Metascape website. The top 15 most up-regulated pathways are downloaded from Metascape website (http://metascape.org/gp/index.html#/main/step1) (Figure 7B).

Figure 3.

Representative violin plots showing data quality after datasets integration

Figure 4.

Representative elbow plot showing the standard deviation across the first 30 PCs

Figure 5.

UMAP visualization of clusters obtained with different resolution parameters

Figure 6.

Dynamics of CD4 T cells pre-and post-stimulation

(A) UMAP projection of CD4 T cells pre-and post-stimulation, colored by resting T cells and stimulated T cells (stim).

(B) UMAP projection of CD4 T cells pre-and post-stimulation, colored by cell subsets.

(C) Marker genes used to annotate CD4 T cell subsets.

(D) The fraction of T cell subsets in resting T cells and stimulated T cells (stim).

(E) Trajectory of T cell activation on UMAP plot.

Figure 7.

Differential analysis and gene set enrichment analysis of HSP^hi T cells relative to all cells

(A) Volcano plot displaying differential genes of HSP^hi T cells.

(B) Bar chart displaying pathways enriched with high expression genes in HSP^hi T cells.

Expected outcomes

Based on our experience, the typical cell yield rate and recovery rate are as follows:

We general isolate 2 to 5×10⁷ of PBMCs from 5 mL of whole blood using the gradient centrifugation method. However, the cell yield rate may vary between different samples, depending on the quality of blood sample.

The cell purity needs to be investigated after magnetic beads-based enrichment. According to our experience, the cell purity will be > 95%. High purity is crucial for the success of downstream experiments. If the purity is below 95%, we recommend either conducting an additional enrichment or repeating this experiment.

Recovery rate after cell activation: Based on our experience, cell viability is around 95% after 18 h of stimulation. However, cell loss is a common issue due to the binding between cell and beads. The recovery rate is around 70%–80% of the initially input cells. This should be considered to ensure that enough number of cells are available for sequencing.

Limitations

Firstly, this protocol on a single time point (18 h) for T cell activation. This time point was chosen based on previous study. However, T cell activation is a dynamic process. This protocol provides a systemic pipeline for studying T cell activation at single-cell level. Multiple time points could be considered in the future study.

Secondly, this protocol employs beads-based positive selection to isolate CD4 T cells. This method generates CD4 T cell with high viability and purity. Positive selection involved the direct binding of antibodies to CD4, which could potentially alter the cell status. Alternative experimental designs such as negative selection could be considered.

In addition, we employed 10× Genomics Chromium (v2) for scRNA-seq in the protocol. The capture efficiency of v2 platform is relatively lower than the lasted version (e.g.,v3).

Troubleshooting

Problem 1

Column clogging (related to step 17).

Potential solution

There are three possible causes for column clogging. First, the column may be clogged by cell clumps, dead cells or cell debris. Second, column will be clogged when excessive cells are loaded onto the column. Third, column drying out can also lead to clogging.

• •To ensure there is no cell clumps, pass the cell suspension through a stainer before loading.
• •Check the cell aggregates and cell debris under the microscope.
• •Always keep the column wet during experiment.
• •Flush out cells stuck in the column with MACS buffer. Apply the flush-out cells onto a new MS column if the flow remains slow.

Problem 2

Cell loss after bead-based activation (related to step 31).

Potential solution

Some T cells may adhere to the beads or the bottom of 96-well plate, leading to cell loss during cell activation experiment. There are some tips to minimize cell loss and enhance cell recovery.

• •Pipette the cell suspension up and down to detach any cells from the bottom of plate.
• •Mix the cell suspension gently and thoroughly to disperse cell-bead aggregates.
• •Rinse the beads 1–2 times with PBS on the magnet to increase the recovery of cells.

Problem 3

Batch effect and data integration (related to step 43).

Potential solution

In the analysis of scRNA-seq data, batch effects are introduced by various factors such as differences in sample handling, experimental procedures, or sequencing depth. These effects can result in biases in the expression patterns of cells across different batches. Recognizing and correcting these batch effects is essential for data integration. There are several methods available for integrating single-cell data. In Seurat v3, CCA dimensionality reduction is applied to better reflect the correlations of transcriptional features in cell distributions. The MNN algorithm is then utilized to identify the closest cell pairs between two datasets, called anchors. The discrepancies between these anchors are used to represent the technical biases between datasets, thereby effectively reducing the impact of batch effects.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Wenfei Jin (jinwf@sinh.ac.cn).

Technical contact

Technical questions on executing this protocol should be directed to and will be answered by the technical contact, Hui Li (lih8@sustech.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The single cell RNA-seq data has been deposited in Genome Sequence Archive in BIG Data Center and is publicly available as of the date of publication. Accession number is HRA002777.
• •The original code of this paper is available at GitHub (https://github.com/JinWLab/T_active) or Zenodo (https://doi.org/10.5281/zenodo.13859304) and is publicly available as of the date of publication.
• •Any additional information required to reanalyze the data in this paper is available from the lead contact upon request.

Acknowledgments

We thank all members of the Jin lab for the help with this experiment. This study was supported by the National Key R&D Program of China (2021YFF1200900 and 2021YFA0909300), the National Natural Science Foundation of China (32170646), the Guangdong Basic and Applied Basic Research Foundation (2023A1515011908 to N.H.), the Shenzhen Science and Technology Program (JCYJ20220818100401003 and KQTD20180411143432337), and the open project of BGI-Shenzhen. We thank all members of the Jin lab for the helpful discussion. We acknowledge the assistance of Core Research Facilities of SUSTech. We thank Xibin Lu for the excellent support of FACS. The computational work was supported by the Center for Computational Science and Engineering at SUSTech.

Author contributions

W.J. conceived the project. X.W. collected peripheral blood. H.L. sorted the cells and conducted the experiments. Y.L. analyzed the data with help from S.Y., J.W., and Y.H. W.J. and N.H. supervised the project. H.L., Y.L., and W.J. prepared the manuscript, with all authors’ contribution.

Declaration of interests

The authors declare no competing interests.