3. Results and Discussion

3.1. Cell Culture Alters the Transcriptional Profiles of Human Adipose-Derived Stem Cells
Utilizing a microfluidic-based single-cell gene expression platform previously described by our laboratory [20], the transcriptional profiles of 60–100 individual cells from fresh, P0, and P1 adipose-derived stem cells (ASCs) were simultaneously evaluated for 48 surface-marker coding gene (Table S1). In this analysis, ASCs isolated from all passages displayed significant heterogeneity at the single-cell level, both within and across groups (Figure 1).
Differences in the transcriptional profiles of genes related to cell stemness, proliferation, and tumor biology, such as the CD47, CD151, DPPA3, and IGF2R were also observed at the single cell level (Figure 2), suggesting a potential evolution of these properties throughout expansion in cell culture. These differences are embodied in both the fraction of cells expressing these genes, as well as the characteristics of their distributions across cell types.
Figure 1  Single cell transcriptional analysis of fresh, passage 0, and passage 1 adipose-derived stem cells. Hierarchical clustering of simultaneous gene expression for single cells from fresh (left; black), passage 0 (middle; dark grey), and passage 1 (right; light grey) human adipose-derived stem cells (ASCs). Gene expression is presented as fold change vs. median on a color scale from yellow (high expression, 32-fold above median) to blue (low expression, 32-fold below median). Cell/gene qPCR reactions failing to amplify after 40 cycles are designated as non-expressers and represented in grey.
Figure 2  Cell culture alters the transcriptional profile of adipose-derived stem cells. Differentially-expressed genes among primary, passage 0, and passage 1 cells identified using non-parametric two sample Kolmogorov–Smirnov testing. Twelve genes exhibit significantly different (p < 0.01 following Bonferroni correction for multiple comparisons) distributions of single cell expression between fresh→P0 or P0→P1 populations, illustrated here using median-centered Gaussian curve fits. The left bar for each panel represents the fraction of qPCR reactions that failed to amplify in each group. The curves above the dashed line represent transcriptional comparisons between primarily isolated cells versus cells at passage 0, whereas the curves below the dashed line represent comparisons between cells at passage 1 versus passage 0.   Furthermore, the top molecular networks associated with each set of significantly-altered genes, generated using the Ingenuity Knowledge Base, appear to link key mediators of cell proliferation, embryonic development, and organogenesis (Figure 3).
Collectively, these data support the transcriptional evolution of cultured stem cells toward gene expression profiles characterized by more robust, proliferative, and stem-like properties.
Figure 3  Network analysis of gene expression changes in of fresh, passage 0, and passage 1 adipose-derived stem cells. Top scoring Ingenuity Pathway Analysis (IPA)-constructed transcriptome networks based genes that were significantly up-regulated in fresh→P0 (A) and P0→P1 (B) human adipose-derived stem cells (ASCs). Direct relationships are indicated by solid lines, and dashed lines represent indirect relationships.

3.2. Partitional Cluster Analysis Reveals Subpopulations of Adipose-Derived Stem Cells that Evolve Throughout Cell Culture Passage
To further examine this niche, the super-set of transcriptional profiles for primary, passage 0, and passage 1 cells was evaluated using unbiased partitional clustering of gene expression data, as previously described [20]. This analysis identified three distinct, transcriptionally-defined ASC subpopulations or clusters within each phenotypic group (Figure 4). Interestingly, the first subgroup was considerably enriched for primary cells, the second subgroup dominated by a majority of passage 0 cells and a strong minority of passage 1 cells, and the third subgroup comprised primarily of passage 1 cells.
To predict the functional significance of these subpopulations, we evaluated the top molecular networks associated with genes up-regulated in each cellular subgroup using known transcriptional relationships derived from the Ingenuity Knowledge Base (IPA). The network generated from the first subpopulation (cluster 1), dominated by fresh ASCs, was characterized by non-specific cell-to-cell signaling and interactions (Figure 5A). The second network, derived from cluster 2, was associated with cellular proliferation and inflammatory pathways, which may represent the temporal shift of cells in this subgroup from primary to early passage (Figure 5B), and the third network (corresponding to cluster 3 and dominated by P1/P0 cells) was similarly characterized by cell development, proliferation, and movement (Figure 5C).
To further explore the significance of these signaling pathways, we used IPA to merge these three networks to generate a super-network including all genes associated with either pathway (Figure S1). Interestingly, the central elements linking these networks were extracellular-related kinase (ERK)1/2 and focal adhesion kinase (FAK), both closely implicated in the hyperproliferative response in cutaneous scarring [21], and tumor necrosis factor receptor superfamily member 12A (TNFRSF12A), which is a key regulatory element of cellular invasion in breast cancer [22].
Figure 4  Partitional cluster analysis of single cell adipose-derived stem cell data across cell culture passage. K-means clustering of fresh (black), passage 0 (dark grey), and passage 1 (light grey) human adipose-derived stem cells (ASCs). Gene expression is presented as fold change from median on a color scale from yellow (high expression, 32-fold above median) to blue (low expression, 32-fold below median). Pie graphs represent the fraction of cells comprising each cluster from fresh, passage 0 (P0), and passage 1 (P1) cells, colored in purple, orange, and green in accordance with clusters 1, 2, and 3, respectively.
Figure 5  Network analysis of adipose-derived stem cell subpopulations. Top scoring Ingenuity Pathway Analysis (IPA)-constructed transcriptome networks based genes that were significantly up-regulated in cluster 1 (A; purple), cluster 2 (B; orange), and cluster 3 (C; green). Direct relationships are indicated by solid lines, and dashed lines represent indirect relationships.

4.