Brief Report: A Differential Transcriptomic Profile of Ex Vivo Expanded Adult Human Hematopoietic Stem Cells Empowers Them for Engraftment Better than Their Surface Phenotype

Summary

HTSeq gene count matrix (htseq_counts_small_molecules_psatha_2017.txt.gz) and sample information (sample_info_small_molecules_psatha_2017.txt). Samples are CD34+ HSPC healthy donor derived populations sorted by flow cytometry based on the expression of CD38 and CD90 markers. Each population is treated with one or the combination of the following chemical small molecules (UM171, SR1, LY) and DMSO was used as vehicle control.

Sample Info column information

• AG: Sample name
• Kit: RNAseq library kit
• SampleTaxonomy: taxonomic classification of samples (human CD34+ cells)
• Treatment: Small molecule treatment (UM171SL: UM171 + SR1 + LY; SR1LY: SR1 + LY)
• Population: FACS sorted population
• Day: Number of days treated

Materials and Methods

For the transcriptomic analysis, total RNA was collected using the mirVana RNA isolation kit (ThermoFisher Scientific) or RNeasy Mini Kit (Qiagen) from sorted and bulk cultured cells (>20,000 cells). Illumina libraries were constructed using the TruSeq RNA Library Prep Kit v2 (Illumina) and samples were sequenced on HiSeq 4000 (Illumina) with 36 base paired-end read length. Adaptor trimming and paired alignment was performed using TopHat2 (2.0.13). Gene counts from sequencing reads were calculated using htseq-count. Differential expression analysis was performed in R (3.4.1 release) using the package DESeq2, and gene set enrichment analysis was performed with Enrichr available at (http://amp.pharm.mssm.edu/Enrichr/). The loadings PCA was generated by applying Principal Component Analysis on the centered log-transformed expression values of the top 1000 genes ranked by their coefficient of variation as returned by DESeq2 default function. The rotation of each vector represents the eucledian distance between each treatment. The gene clusters result from K-means clustering after 1000 iterations of the aformentioned 1000 most variable genes and the number of clusters (K = 5) was empirically estimated by the number low-order branches of a hierarchical clustering tree. Differential expression was called upon change (positive or negative) above 1.5-fold (0.58 log₂-fold) and significance below 0.05 of the FDR adjusted p-values. Pathway maps were generated using GAGE⁷ with annotations retrieved from KEGG pathway database (http://www.genome.jp/kegg/pathway.html).