Zygotic transcriptional activation elevates initial blastomere-to-blastomere biases. (A) Illustration of the strategy to analyze separately the two sources of blastomere-to-blastomere heterogeneity, namely partitioning errors and transcriptional noise, in an early-to-late two-cell embryo in a stage-specific manner. (B) Blastomere-to-blastomere expression noise calculated with a previously published formula using a single-blastomere transcriptome from an early-to-late two-cell embryo as well as using technical repeats of RNA-seq data from early blastomeres (Eqn 2 in Materials and Methods). (C) The extent of blastomere-to-blastomere gene expression asymmetry is represented by E_higher/E_lower. If the gene expression in the two blastomeres is more similar, the value of E_higher/E_lower will be closer to 1. ***P<0.001; different letters indicate statistical difference, P<0.01; same letters indicate P>0.05; one-way ANOVA.

Multiplied partitioning errors and elevated overall blastomere-to-blastomere biases during the two- to 16-cell stages. (A) Illustration showing that multiplied partitioning errors at each cleavage division drive the continuously increasing cell-to-cell bias, each time creating one daughter cell bearing a higher quantity of a specific substance and another with a lower quantity of that substance (Eqns 3-5 in Materials and Methods). The graph shows the amplification effect of a small bias (10% co-efficiency) during the first rounds of cleavage division; G0-3 indicates generations 0-3. (B) Single-blastomere RNA-seq and blastomere-to-blastomere expression noise in two- to eight-cell embryos in both mouse and human. Each dot represents a comparison of two blastomeres within one embryo. Datasets from mouse (Deng et al., 2014) and human (Yan et al., 2013) were used for the calculations. Different letters indicate statistical difference, P<0.01; same letters indicate P>0.05; one-way ANOVA.

Trends of transcriptional asymmetry during the two- to 16-cell embryo stages reveals monostable and bistable patterns. (A-D) The trends of blastomere-to-blastomere asymmetry for genes with high expression level (top 20%) and strong bias (top 20%) at the two-cell (A), four-cell (B), eight-cell (C) and 16-cell (D) stages. (E) The expression asymmetries of Tubb2c, Carm1 and Pou5f1 were analyzed using single-blastomere expression profiles from two-cell to 16-cell embryo stages in mouse. The average expression level of each gene in each embryo is normalized to 1, and the relative expression level of each gene in each blastomere is shown using different shapes/colors, as illustrated. The dispersion degree along the y-axis can be directly visualized in each embryo and represents the overall extent of expression asymmetry between different blastomeres. Each dot represents the relative value of each blastomere to the average RPKM of all blastomeres within one embryo. (F,G) Dynamics of single gene expression systems with negative/positive feedback regulation. The change of protein X over time (dX/dt) can be described by a differential equation which is synthesis rate S⁺ minus degradation rate S⁻. The top two panels are the plot of the functions S⁺ and S⁻ within the same graph. The lower two panels illustrate the corresponding potential function of the system. From these graphs, there are one (F) or three (G) intersection points, indicating the steady states in which dX/dt is equal to 0. The directions of the arrows indicate the movement towards equilibrium. If two arrows move away from each other, this represents an unstable steady state. Therefore, a single gene expression system with negative feedback regulation (F) has only one stable steady states and a single gene expression system with positive feedback regulation (G) has two stable steady states: one in which the level of X is low and one in which the level of X is high; the system could shift into either of these two states from the unstable point.

Cleavage history and de novo transcription contribute to the relative ratio of opposing lineage specifiers. (A) Illustration of the tilted ratio of two substances as a result of partitioning error. (B) Colors indicate the probability of tilting an initial ratio (10% threshold) of a pair of lineage specifiers with counteracting functions (Eqn 6 in Materials and Methods). (C) The stage-specific expression level (RPKM value) of Carm1 and Cdx2 during the one- to eight-cell embryo stages. n=3 for each column. Data used to generate these columns are derived from a previous publication (Xie et al., 2010). (D,E) The RPKM values of Carm1 and Cdx2 in each eight-cell blastomere examined from mouse (D) and human (E) embryos. (F,G) The relative ratios of Carm1 and Cdx2 in different eight-cell blastomeres form distinct lineage strengths in both mouse (F) and human (G) embryos. The graphs in F and G were generated from the raw numbers in D and E, respectively. Previously published mouse (Deng et al., 2014) and human (Yan et al., 2013) datasets were used to generate the table and figure.