(re)freezing PBMC

This study investigates the impact of cryopreservation and freezing media on RNA quality and gene expression profiles of peripheral blood mononuclear cells (PBMC). No significant difference in cell viability or RNA quality was observed between freshly isolated and cryopreserved cells, even after two freeze-thaw cycles. Transcriptome analysis revealed no significant differences in alignment rates or read counts across various conditions. Differential gene expression analysis identified changes between fresh and thawed samples, with a consistent downward trend of mRNA abundance in the frozen samples, but minimal changes were observed between the first and second freeze-thaw cycles. Gene set enrichment analysis showed only minor significant differences, and stress-related gene sets were not upregulated. Freezing medium appears to be the safer option compared to freezing in trizol. These results confirm that cryopreservation and thawing processes do not compromise RNA integrity or sequencing reliability in PBMC, supporting the use of cryopreserved samples for gene expression studies. Study samples Blood was collected from three healthy donors (one male and two female adults) following informed consent with Ethical Committee approval of the University Hospital of Ghent (reference number 2017/1207). Blood samples were collected in K2 EDTA-tubes (BD Vacutainer) and four 10mL tubes were processed per donor. PBMC isolation, cryopreservation and thawing Within one hour after blood collection, PBMC isolation was performed simultaneously for all three donors. The PBMC were isolated by density gradient centrifugation using Ficoll-Paque PLUS density gradient media. Samples were centrifuged at 800g for 18min without brake at room temperature. Isolated PBMC were washed twice in PBS at 400g for seven minutes at room temperature. The cells were resuspended in 0,5mL PBS and concentration was determined using the BioRad TC20 Automated Cell Counter in combination with trypan blue staining. PBMC were aliquoted at 5 million cells per vial. One vial per donor was directly processed for RNA extraction (these samples will be named “fresh”). For cryopreservation, the remaining PBMC were diluted in freezing medium (RPMI + 1% penicillin/streptomycin + 10% FCS plus 10% dimethyl sulfoxide). PBMC were transferred to isopropanol containers (Mr. Frosty™ Thermo-Fischer Scientific) and stored at -80 °C for 24 h, before being transferred to -150°C freezers for one week. PBMC were thawed at room temperature before transferring to warm (37°C) complete medium (RPMI + 1% penicillin/streptomycin + 10% FCS plus 10% dimethyl sulfoxide). After centrifugation at 400g for 7min the supernatant was discarded, and the cells were resuspended in PBS. One vial was directly processed for RNA extraction (these samples will be named “defrosted”), the cells from one other vial was transferred to complete freezing medium (these samples will be named “FM” or “freezing mix”) and from a third vial the cells were directly refrozen in trizol reagent (these samples will be named “trizol”). These two refrozen samples per donor (FM and trizol) were preserved for an additional week in -150°C before thawing, and subsequent RNA extraction was performed. A graphical summary of the workflow is provided in Figure 1. The viability of the cells was assessed using the BioRad TC20 Automated Cell Counter in combination with trypan blue staining.RNA extraction and RNA sequencing. RNA from PBMC was extracted using the trizol-based miRNeasy Mini kit (Qiagen, Hilden, Germany) following the manufacturers’ instructions, including the on-column DNase digestion step. RNA concentrations were quantified using Nanodrop (ThermoFisher Scientific, Waltham, MA, USA) spectrophotometry. Further characterization and quality assessment was carried out using the Fragment Analyzer (Agilent Technologies, Santa Clara, CA, USA). RNA sequencing was performed after library preparation according to the TruSeq mRNA protocol (Illumina, San Diego, CA, USA) in a single flowcell on the NovaSeq 6000 instrument following these parameters: Illumina NovaSeq 6000 v1.5 sequencing kit 100 cycles single-end reads (101-10-10-0), 1%PhiX. Sequencing yielded high-quality data, with 94.02% of bases exceeding Q30 and 85.38% of clusters passing the quality filter and a mean read depth of 27,8 million reads (18,4M to 32,7M). RNA sequencing preprocessing To achieve a uniform sequencing depth for all samples, allowing a fair comparison, random subsampling was performed using setqk (version 1.4) and adjusted to the minimum read count (18 million). Pseudo-alignment and transcript quantification was obtained by Kallisto13 (version 0.48.0) using a GRCh38 Ensembl v 100 index. The length of reads was set to 200 base pairs, and 20 bootstraps were set as parameters during the (pseudo-)alignment. Plots, statical analysis, differential expression analysis and GSEA All analyses were performed using the R statistical software (version 4.4.1). For statistical comparisons between different sample conditions, the Mann-Whitney U test was performed, and data visualization was conducted using the ggplot2 package (version 3.5.1). Before performing the differential expression analysis, transcript-based counts were summed up to gene counts whereafter genes were filtered to retain those with minimum sum of 100 over all samples. The DESeq2 package (version 1.44.0) was used, and genes were considered to be significantly differentially expressed when the adjusted p-value was below 0.05 and the absolute value of the log2 fold change was larger than 2. For the gene set enrichment analysis (GSEA), fgsea14 package (version 1.30.0) was used to perform the analysis. Firstly, a list of marker genes for each cell type was constructed on Azimuth scRNA-seq database15 by using “FindMarkers” function in Seurat package (version 5.1.0) with the following parameters: logFC > 0.25, min.pct > 0.25. Additionally, Wilcoxon Rank Sum test was selected to identify differentially expressed genes for one cell type versus the rest of the cell types. Subsequently, the Cell Type Signature (C8) pathways were retrieved from the Molecular Signatures Database (MSigDB), and the immune-related gene sets were selected for analysis. For stress specific pathways, chemical stress, fluid shear stress, oxidative stress, osmotic stress, stress in general and stress response signaling signatures were selected from the biological process ontology gene sets (C5, subcategory = GO:BP). Deconvolution of RNA sequencing data To estimate the cell type fractions and compositions from the transcriptome data, reference-based deconvolution was performed using the Dampened Weighted Least Squares (DWLS) (v.0.1.0) method16. We selected the method based on our previous benchmarking studies on RNA-seq based deconvolution approaches17. As reference, we used a publicly available scRNA-seq data generated on PBMC from healthy donors (EGAS00001004571)18. The scRNA-seq data labelled with seven different cell types (T cells, B cells, neutrophils, natural killer (NK) cells, monocytes, erythrocytes, and eosinophils) were used to construct a signature matrix, that was subsequently used as an input for deconvolution. The signature matrix for the different cell types was built in R (v4.3.2) using the function ‘buildSignatureMatrixMAST’ with default parameters. Afterwards, common genes between the bulk data and the signature matrix were filtered with the function ‘trimData’. Subsequently, deconvolution was performed using the function ‘solDWLS’, resulting in inferred cell type proportions from the bulk data. To assess significant differences in deconvolved proportions between the fresh condition and the freezing conditions, paired t-tests were performed, and the FDR method was used for the multiple hypothesis testing. *************************************************************** Raw files for human/patient samples are being made available in EGA (https://www.ebi.ac.uk/ega/) for controlled access to the personally identifiable sequence data. ***************************************************************