RevGel-seq: instrument-free single-cell RNA sequencing using a reversible hydrogel for cell-specific barcoding

RevGel-seq sample preparation workflow

Experiments were performed with the RevGel-seq protocol, capable of analyzing 10,000 input cells per sample with the specially designed gelation device (Fig. S3). The individual steps shown in Fig. 1A, from cell-barcoded bead coupling to library preparation for sequencing, are more fully described below:

Cell labeling

Cells are first labelled with a bifunctional chemical linker used to tether individual cells to barcoded beads. The cells are incubated with this bifunctional linker (polyA at one extremity and a hydrophobic moiety at the other extremity) in DPBS for 5 min. The labelled cells are washed by 0.1% BSA in DPBS twice and resuspended at a density of 100 labelled cells/µL with the same buffer for a sample preparation with 10,000 input cells.

Cell coupling

For preparation of the cell-bead complexes, 100 µL of the barcoded bead suspension1 (1000 beads/µL, 20% PEG8000 in DPBS) are pipetted into the bottom of gelation tube (see Fig. S3), followed by transfer of 100 µL of the labelled cell suspension into the suspension of barcoded beads. The combined suspension is homogenized by micropipette to couple the individual labelled cells to individual barcoded beads upon their collisions. The number of barcoded beads is in large excess to the number of labelled cells to avoid that one bead is bound to more than one cell. Quantification of cell-bead complexes can be achieved at this point by microscopy, if evaluation of coupling is required.

Dilution into gel solution

Cell-bead complexes suspension (200 µL) are diluted into 10 × volume of hydrogel solution (200 µL). The hydrogel solution contains thermal sensitive polymers and is isotonic with a density matched to barcoded beads to prevent cells from osmotic shock as well as beads from precipitation prior to gelation. The diluted cell bead complexes are then homogenized by simple rotation of the gelation tube. After this homogenization, the gelation piston is slowly inserted into the gelation tube while keeping both vertical, as the contents rise smoothly along the wall of the gelation tube with a level horizontal meniscus.

Gelation

The assembled gelation device is then placed vertically on ice and incubated for 20 min for complete gelation to immobilize the cell-bead complexes. This resulting thin hydrogel layer has a high surface to volume ratio to make the subsequent cell lysis efficient and synchronous for all cells. While maintaining verticality, the gelation piston is slowly removed from the gelation tube. The gel remains in the tube, partially collapsed, with no gel remaining on the piston. Then 7.5 mL of lysis buffer (sarkosyl 0.2% w/v, 50 mM dithiothreitol, 20 mM EDTA and 200 mM Tris pH 7.5) are added to the gelation tube. If necessary, the procedure can be interrupted at this two-hour stopping point by placing the gelation device vertically in a freezer at − 80 °C or flash freezing in dry ice, to resume the experiment at a later time by thawing the frozen device at 25 °C water bath for 15 min.

Cell lysis and mRNA capture

The hydrogel sheet in the lysis buffer is incubated at room temperature for 1 h with a gentle orbital shaking. Cell lysis reagents diffuse into the hydrogel and lyse cells in the gel, triggering release of polyA tailed RNA and their hybridization on the 3′-polyT extremities of the barcoding oligonucleotides on the capture bead coupled to each cell. The released mRNA molecules are confined in the vicinity of each cell by the hydrogel polymer mesh.

Gel dissolution

When lysis is completed, the degelation step is immediately initiated by adding 3.75 mL of degelation buffer (3 M guanidine thiocyanate and anti-foam reagent supplied to the reverse transcription reaction buffer (RT buffer)) in the gelation tube and sealed with the gelation tube cap. The gelation tube is then held in the vertical position and vortexed for 5 min at maximum speed. The barcoded beads with hybridized mRNA are now free from hydrogel and centrifuged at 1000×g for 3 min.

RT, PCR, and preparation for DNA sequencing

The RNA-loaded barcoded beads are washed once with 5 mL of ice-cold RT buffer, and transferred into 1.5 mL microtubes, then washed two additional times to prepare for subsequent enzymatic steps.

  1. (i)

    Reverse transcription: The pellet of beads is resuspended in 20 µL of bead wash buffer and supplemented by 75 µL of the RT supermix composed of 1.3 × RT buffer supplied with Maxima H Minus reverse transcriptase (ThermoFisher, EP0751), 1.3 mM each dNTP, with 100 U of NxGen RNase inhibitor (LGC Biosearch technologies, 30281), 2 U of beta-agarose I (NEB, M0392), and 5 µL of Maxima H minus reverse transcriptase. The sample is then placed in a heat block and incubated 10 min at 25 °C followed by 90 min at 42 °C. Next, the sample tube is placed on ice for 2 min and then spun briefly on a benchtop centrifuge. The procedure can be interrupted at this point and the sample stored overnight at 4 °C or up to one week at − 20 °C. When ready to continue, the sample is thawed on ice before proceeding to the next steps.

  2. (ii)

    Endonuclease I treatment: 5 µL of Exonuclease I is added to the sample and homogenized by pipetting. The remaining capture oligos non-hybridized with captured RNA are digested by incubation for 50 min at 37 °C. After this incubation the enzymes are removed by washing with 250 µL of TE (10 mM Tris, 1 mM EDTA, pH 8) with 0.5% (w/v) SDS.

  3. (iii)

    Alkaline denaturation: The beads are washed once with 100 µL 0.1 M NaOH solution and resuspended in 50 µL 0.1 M NaOH solution. RNA molecules are then removed from cDNA by 5 min incubation on a microtube rotator/wheel. The bead washing is then carried out with 250 µL of TE-TW (10 mM Tris, 1mM EDTA, pH 8; 0.01% Tween 20) and adjusting the residual volume to 20 µL using RT buffer.

  4. (iv)

    Second strand synthesis: The S3 supermix (1.1 × RT buffer, 1.3 mM each dNTP, 12% PEG8000, 13 µM second strand primer) is first incubated for 5 min at 70 °C, then immediately placed on ice for 1 min, spun, and homogenized. Then 77.5 µL of the S3 supermix is added to the bead suspension, followed by addition of 2.5 µL of the S3 enzyme. The sample is thoroughly homogenized by pipetting and incubated 1 h at 37 °C. Two washes with 250 µL of TE-TW are carried out, followed by a final resuspension in 160 µL with nuclease-free water. Relevant oligonucleotides are described in Table S1.

  5. (v)

    Polymerase chain reaction: 240 µL of the PCR supermix (1.66 × KAPA HiFi HotStart readymix (Roche, 07958935001), 1.33 µM PCR primer) is added to the 160 µL of the bead suspension. The beads in the reaction mix are homogenized by pipetting and split into 8 PCR tubes, with 50 µL in each tube. The PCR program is described in Table S2.

  6. (vi)

    PCR strip tubes are spun for 20 s using a benchtop centrifuge, and 45 µL of bead-free supernatant from each tube is transferred to another clean PCR tube separately. Size selective purification is performed at 0.6 × volume of SPRIselect reagent (Beckman Coulter, B23317) to the transferred PCR product (27 µL SPRIselect to 45 µL of the PCR product) and following the manufacturer’s procedure. Elution is performed in 20–40 µL of nuclease-free water. Quality check of the purified cDNA amplicons is made by Agilent’s TapeStation HSD5000. The purified cDNA can be stored at − 20 °C, then thawed on ice before proceeding to the next steps.

  7. (vii)

    Library preparation: 600 pg of the purified cDNA amplicons from each PCR tube are sampled and processed according to Illumina Nextera XT library preparation kit (illumina, FC-131-1024) using a custom library prep primer for the P5/read1 side of the library amplification PCR.

Mixed species sample preparation

Both human cell line HEK293 (DSMZ, ref. ACC305) and mouse cell line NIH3T3 (ECACC, ref. 93061524) were cultured separately in complete DMEM media. Both cell lines were harvested, and the cell density was adjusted to 100 cells/µL. Each cell suspension was mixed at 1:1 ratio to obtain mixed cell suspension at 50 human cells and 50 mouse cells/µL. This mixed cell suspension was used as start material of RevGel-seq sample prep for species mix experiment.

PBMC sample preparation

Whole human blood from an anonymous healthy donor was provided from the Etablissement Français du Sang (EFS) under a convention for research use only. PBMCs were isolated from fresh blood by HistoPaque-1077 (Sigma) according to the instructions. The purified PBMC from the same blood sample was used for both the RevGel-seq method described above (with doubling of the ratio of beads per cell to increase coupling yield), and the 10 × Genomics Chromium single cell 3′ reagent kit (v3.1), and both scRNA sample preparations with 10,000 input cells were started at the same time.

Pancreatic islet cells

Pancreas islets were isolated from mouse (8 weeks male C57BL/6J) pancreas by collagenase (1 mg/mL, Sigma-Aldrich) injection in the bile duct and handpicked under a binocular microscope (Leica). Islet cells were dissociated with 0.05% trypsin–EDTA (Gibco Thermo Fisher), incubated at 37 °C for 5 min and resuspended in PBS with 2% fetal calf serum and 0.5 mM EDTA. Two technical replicates of 10,000 input cells from this pancreas islet were prepared following RevGel-seq method described above.

Cardiomyocytes, including an early checkpoint

Human cardiomyocyte cells were purchased from Takara Bio (ref. Y10060). The cells were cultured according to the instructions and harvested by trypsin just before processing with the RevGel-seq method described above with 10,000 input cells. After the cell-bead coupling step, 20 µL of the cell-bead complex suspension was sampled with a wide-bore pipette tip and transferred to a cell counting chamber for microscopic examination of the cell-bead complex state. The remaining cell bead complex suspension was processed according to the RevGel-seq procedure and sequenced.

Early stopping point

To confirm the possibility to safely stop the experiment at an early stopping point in the RevGel-seq procedure, three samples of 10,000 input cells of HEK293 and NIH3T3 in a 1:1 ratio were prepared from the same mixed cell suspension and their preparation was stopped at two different times. One sample was stopped after reverse transcription as a known stopping point1. The two other samples were stopped immediately after applying lysis buffer to the hydrogel (2 h from the start of sample preparation) by freezing, one in a − 80 °C freezer, and the other on dry ice for 20 min then transferred to − 80 °C freezer. The following day, the two frozen samples in the cell lysis buffer were thawed at 25 °C in water bath for 15 min prior to resuming the remaining sample preparation procedure.

Sequencing

All sequencing of RevGel-seq samples was performed on a NovaSeq 6000 with spiking custom primer for read1. The read length was 30 bases for read1, 8 bases for i7 index, and 70 bases for read2. Sequencing of 10 × Chromium 3′ RNA samples was performed according to the instructions and NovaSeq 6000 was used as well.

Data analysis with Cytonaut, the Scipio bioscience cloud-based bioinformatics platform

The end-to-end data analysis from raw sequencing data to differential gene expression is performed by Cytonaut (v1.2) cloud software (www.cytonaut-scipio.bio), which integrates data pre-processing (v6.7) as well as data post-processing (v1.5) and interactive data visualization (v1.2). The same end-to-end data analysis methodology based on Cytonaut was applied to all samples except for RevGel-seq and 10 × PBMCs samples and for RevGel-seq Pancreatic islet cells samples for which Cytonaut was used for pre-processing analysis but a separate script was used for post-processing analysis based on Seurat (v4.1.0) and automated cell annotation based on SingleR (v1.8.1). The workflow of the end-to-end data processing pipeline integrated in the Cytonaut platform is summarized in Fig. 1B, with details on individual steps presented below:

Pre-processing pipeline

The data pre-processing pipeline of Cytonaut takes as input the two FASTQ files of the sample (R1 and R2), leverages the specific barcode pattern of RevGel-seq capture beads, and successively performs read quality control, detection of cell-associated barcodes, read alignment based on exons only, read deduplication, and read assignment. Detection of cell-associated barcodes was performed for each sample by applying the distance-based knee method of UMI-tools v1.1.2. The reference genome sequences and annotations are based on Ensembl release v99 for human species (GRCh38; ftp.ensembl.org//pub/release-99/fasta/homo_sapiens/dna/; ftp.ensembl.org//pub/release-99/gtf/homo_sapiens/) and mouse species (GRCm382; ftp.ensembl.org/pub/release-99/fasta/mus_musculus/dna/; ftp.ensembl.org//pub/release-99/gtf/mus_musculus/).

The pre-processing pipeline provides as output:

  1. (i)

    The count matrix of the input sample, which contains the number of transcripts detected for each gene in each detected cell;

  2. (ii)

    A set of quality indicators, which include in particular the median number of genes per cell, the median number of transcripts per cell, the median mitochondrial transcript rate per cell, as well as their distribution statistics;

  3. (iii)

    For the NIH3T3-HEK293 sample in Fig. 2A, the quality indicators also include statistics on cell barcode purity. Cell barcode purity is the probability that a transcript captured by the cell barcode has been expressed by the main cell coupled to the bead of this barcode. Assuming a simple case of equal number of total human and mouse transcripts in the sample, the cell barcode purity is provided by the formula 1—impurity where impurity is twice the ratio of transcripts belonging to the minority species among the transcripts captured by the cell barcode. Cell barcodes are assigned to each species, mouse or human, if they have a cell barcode purity of 95% or more; for cell barcode purity lower than 95% the cell barcodes are represented as gray dots in the Barnyard plot. The total cell multiplet rate is extrapolated from the measured hetero-species cell multiplet rate (twice the rate in case of an equal number of human and mouse cells), where a hetero-species cell multiplet is defined as a cell barcode for which cell barcode purity is lower than 2/3 (i.e., for which cell impurity is higher than 1/3).

Post-processing pipeline

The data post-processing pipeline of Cytonaut takes as input the sample count matrix provided by the pre-processing pipeline and performs the following steps:

  1. (i)

    Filtering of cells and/or genes is achieved according to application-dependent parameter values which are selected by the user (e.g., minimum number of cells expressing a gene set to 3, minimum number of genes per cell set to 200, maximum mitochondrial transcript rate per cell set to 20% for the presented studies);

  2. (ii)

    Cell data normalization and log-transformation are applied using the formula log2(10,000 * X + 1)), where X is the percentage of transcripts detected in the cell for the gene of interest and log2() is the logarithm in base 2;

  3. (iii)

    Highly Variable Gene (HVG) detection is performed according to application-dependent parameter values which are decided by the user (e.g., number of top genes set to 2000 for the presented studies);

  4. (iv)

    Principal Component Analysis (PCA) is run on the top HVG genes, according to application-dependent parameter values which are fixed by the user (e.g., number of PCA dimensions set to 50 for the presented studies);

  5. (v)

    2D Embedding of cells is carried out according to parameter values that are decided by the user (e.g., number of neighbors set to 10 for the presented studies, UMAP method with minimum distance set to 0.3 for the presented studies, or t-SNE method with perplexity set to 30);

  6. (vi)

    Cell Clustering is performed according to parameter values which are set by the user (e.g., Louvain method with resolution set to 0.8 for the presented studies, or Leiden method with resolution set to 0.8);

  7. (vii)

    Differential Gene Expression (DGE), according to a statistical test method established by the user (e.g., Wilcoxon–Mann–Whitney test for the presented studies).

Post-processing pipeline output

The following items are provided:

  1. (i)

    The DGE matrix, which contains for each cell cluster and each gene the z-score, the log fold change, and the p-values (non-adjusted and adjusted) of the expression of the gene in the cells belonging to the cluster compared to the cells belonging to the other clusters.

  2. (ii)

    The attributes of each cell, including the cell quality indicators and the cluster ID of the cell;

  3. (iii)

    The attributes of each gene, including the cell quality indicators and the gene variability indicators;

  4. (iv)

    The AnnData object (h5ad format), which contains all the data allowing to further reproduce the post-processing results.

Interactive data visualization

The Cytonaut Rover module provides an interactive visualization of data generated by the post-processing pipeline, allowing exploration of gene expression and cell attributes in cell clusters and application of customized filters leading to customized distribution statistics (heatmap, dot plots, violin plots). In particular, the visualization of the expression of the top differentially expressed genes in each cell cluster facilitates the annotation of the identified cell types.

Read more here: Source link