Aligning single-cell developmental and reprogramming trajectories identifies molecular determinants of reprogramming outcome. Aligning single-cell developmental and reprogramming trajectories identifies molecular determinants of reprogramming outcome

Cellular reprogramming through manipulation of defined factors holds great promise for large-scale production of cell types needed for use in therapy, as well as for expanding our understanding of the general principles of gene regulation. MYOD-mediated myogenic reprogramming, which converts many cell types into contractile myotubes, remains one of the best characterized model system for direct conversion by defined factors. However, why MYOD can efficiently convert some cell types into myotubes but not others remains poorly understood. Here, we analyze MYOD-mediated reprogramming of human fibroblasts at pseudotemporal resolution using single-cell RNA-Seq. Successfully reprogrammed cells navigate a trajectory with two branches that correspond to two barriers to reprogramming, with cells that select incorrect branches terminating at aberrant or incomplete reprogramming outcomes. Differential analysis of the major branch points alongside alignment of the successful reprogramming path to a primary myoblast trajectory revealed Insulin and BMP signaling as crucial molecular determinants of an individual cell’s reprogramming outcome, that when appropriately modulated, increased efficiency more than five-fold. Our single-cell analysis reveals that MYOD is sufficient to reprogram cells only when the extracellular milieu is favorable, supporting MYOD with upstream signaling pathways that drive normal myogenesis in development. Overall design: scRNA-seq experiment for MYOD-mediated human fibroblasts reprogramming. HSMM derivation, expansion and differentiation was as previously described (Trapnell et al., 2014). Foreskin Human Fibroblasts obtained from commercial vendor (Stemgent) were expanded in alpha-mem supplemented with glutamax, 10% FBS and 16ng/ul of FGF-b (ThermoFisher Scientific). The fibroblasts were then infected with a mixture of lentiviruses encoding hTERT (and Puromycin resistance gene), Tetracycline Repressor (and Geneticin resistance gene) and TR-controlled hMYOD (and Blasticidin resistance gene) (ThermoFisher Scientific). All the resistance genes are constitutively expressed and therefore triple selection was performed and maintained using 1ug/ml of Puromycin, 500ug/ml of Geneticin and 2ug/ml of Blasticidin to generate the hFib-MYOD line. To perform reprogramming and differentiation experiments HSMM and hFib-MYOD cells were plated in 24-well formats at a density of 50.000 cells per well. Gelatin was sometime used as a coating agent with no significant difference with respect to uncoated dishes. Differentiation and reprogramming was induced using a differentiation media containing alpha-mem supplemented with glutamax and 2% HS (ThermoFisher Scientific), supplemented with 2ug/ml of doxycycline to enact MYOD expression. When indicated, Insulin was used at 8ug/ml (ThermoFisher Scientific), BMP inhibitor LDN-193189 was used at 0.1uM (Stemgent), LSD1 inhibitor RN-1 was used at 1uM (EMD Millipore). At the indicated time points, cells were harvested by gentle dissociation using TrypLE (ThermoFisher Scientific) and processed for full-length transcriptome sequencing using the Fluidigm C1 Single Cell or bulk mRNA sequencing as previously described (Trapnell et al., 2014). Cells were loaded onto the microfluidic chip at a concentration of ~250 cells/μl to maximize the number of single cells captured and minimize the occurrence of doublets. In some instances, cells were reloaded if the majority of capture sites were not occupied on the first attempt. As in Trapnell et al, 2014, captured cells were scored by manual on-chip microscopic inspection to determine if they were singletons and free of other debris.