Figure 4. Fate trajectories predicted based on a Continuous State Hidden Markov Model.
(A) Schematic summarizing the CSHMM method starting detailed in the STAR methods section.
(B) The resulting CSHMM model for lung directed differentiation. Each dot represents a cell, color denotes the time point in which the cell was sampled. Nodes are denoted by N0, N1 etc. while branches (paths) are denoted by P0, P1 etc. (note that several branches can share a node). Names next to paths are the transcription factors (TFs) that are differentially expressed for these paths.
(C, D) Alignment of bulk RNA-Seq data from 4 in vivo time points (Figure 1) to the CSHMM model. The correlation of expression values between the bulk time series data and all possible set of paths in the model was computed.
(E) Representative confocal fluorescence micrograph of epithelial sphere outgrowth from NKX2-1GFP+ progenitors sorted on day 14 and cultured until day 35 of differentiation. Immunostaining for cytoplasmic GFP (green) and nuclear CDX2 (red) protein indicates distinct cells express these lung vs hindgut markers (Blue, Hoechst DNA counterstain; Scale bar = 200 μm).
(F) Expression of specific markers in cells assigned by CSHMM to different branches. For example, SFTPB expressing cells are mainly assigned to P6 whereas NKX2-1 cells are assigned to all paths leading to P6.
(G) Relative expression levels of each indicated AEC2 or endodermal transcript (RT-qPCR) in iAEC2s after knockdown of CEBPδ by siRNA.
(H) First panel: sort gates for iPSC line carrying a CDX2GFP reporter, on Day 15 were used to purify CDX2GFP+ vs CDX2GFP- cells. Second panel: after outgrowth of each sorted population in identical media, the relative gene expression levels on day 33 are shown for CDX2 and NKX2-1. Representative brightfield and fluorescence microscopy overlays indicate levels of CDX2GFP fluorescence on day 33 of outgrowth resulting from each indicated population sorted on day 15.