The donor to donor proteomic variability documented in this
The donor-to-donor proteomic variability documented in this paper can help to explain previous reports that show functional discordance between cell populations from different donors (Phinney, 2007, 2012). As per our data, only 13% of the total proteins were found universally expressed by all cell lines. Such a high degree of molecular diversity is a little surprising given the fact that samples from all cell lines were treated under similar conditions. The cell lines used in this study were obtained from two commercial sources (Materials and Methods section) and the cell-handling protocols during isolation from bone marrow aspirates could have contributed to the proteomic heterogeneity. To examine the existence of such affect, we compared the proteomic datasets compiled using cell lines from All Cells (n=3) and Lonza (n=3). In total, 6327 and 7015 proteins were identified from All Cells and Lonza, respectively. The two datasets shared at least 5596 proteins between them (72% similarity) and evidently the variability between the two group-proteomic datasets is not typically higher than the variability calculated for proteomes from individual donors (Table 2). The reference proteome we constructed based on this study (Supplement 2, Table S5) is not differentially represented by any proteomes of the two commercial cell sources (respectively, 97% and 99% of the proteins in the reference proteome were identified from hBMSCs obtained from All Cells and Lonza). Furthermore, hBMSCs from the two sources have also been compared in terms of growth kinetics, differentiation capacity, and cell surface markers. Although functional differences can be observed, none of these can be associated with the cell source or can be observed in different patterns of cell surface marker expression. Overall, functional and proteomic variabilities which have been observed between cell lines in this study cannot be linked directly to commercial cell source. Molecular and functional heterogeneities in cell populations is not unique to hBMSCs; it is a widespread phenomenon among stem Otamixaban including embryonic (Canham et al., 2010), hematopoietic (Schroeder, 2010), neural (Suslov et al., 2002), and cancer stem cells (Wong et al., 2012). The analysis of major proteomic studies (n=34) of human and mouse pluripotent stem cells (hPSCs and mPSCs) by Gundry et al. highlighted the degree of heterogeneity among pluripotent stem cell populations (Gundry et al., 2011). According to their analysis, the comparison of nine major human proteomic studies, which together indexed 6966 proteins, revealed that less than 20% of the proteins were identified from at least 50% of the studies. By comparison, our proteomic screening from six cell lines found 62% of the proteins to be shared by at least 50% of the study subjects. The better proteomic overlap achieved in our study does not necessarily mean that hBMSCs are less heterogeneous than hPSCs. Rather, it reflects uniformity in the analytical proteomic approach. The comparison between the hBMSC proteomic dataset with pluripotent stem cells revealed that numerous proteins are commonly expressed between the three cell types. It also has revealed that 28% (2879) of the hBMSC proteins were not identified from the PSC types. However, not all of these proteins were uniformly expressed between cell lines, only 253 have been identified in all six lines (Supplement 2, Table S6). While hBMSCs, hPSCs, and mPSCs expressed at least 41 CDs in common, 39 CDs have been identified exclusively from the hBMSC dataset. Only 15 CDs out of these 39 were identified from at least three of the six hBMSC lines, and only CD351 was identified from all lines. The 15 CDs are mostly receptors (CD5L, CD11D, CD59, CD23, CD85D, CD108, CD123, CD206, CD204, CD247, CD264, and CD351,) and some of them are involved in calcium ion binding (CD437), or are kinases (e.g., CD167, CDw293). The reference proteomic map for hBMSCs, which was constructed by combining proteins identified from at least 50% of cell lines, was used as a tool to get a quantitative sense of the proteomic variability between cell lines. For example, cell line 8F3560 has a proteome set of 4716 proteins, of which 3961 were also found in the reference proteome and the remaining proteins (834) were not part of the 8F3560 proteome. To account for the variability caused by the size of a proteome we used a correction factor, which is the ratio between the proteome size of a given cell line and the reference proteome (in this case 4716/4797=0.98). The reference proteome consists of 4797 proteins (Supplement 2, Table S5). Thus, the percent variability will be 17.1% (834/4795*0.98). Overall, an average variability between lines was 20.9% (Table 2). We also screened the data to determine if highly variably-expressed proteins are uniformly distributed among cellular organelles. For this purpose, the subcellular localizations of the most variably expressed proteins (those expressed in ≤2 cell lines) were compared against the profile obtained using the reference proteome (Supplement 4, Fig. S2). There was no apparent bias towards certain organelles in molecular variability between cell lines, although the plasma membrane and mitochondria contain slightly higher proportions of variably-expressed proteins.