Precision of morphogen-driven tissue patterning during development is enhanced through contact-mediated cellular interactions

Cells in developing embryos reliably differentiate to attain location-specific fates, despite fluctuations in morphogen concentrations that provide positional information and in molecular processes that interpret it. We show that local contact-mediated cell-cell interactions utilize inherent asymmetry in the response of patterning genes to the global morphogen signal yielding a bimodal response. This results in robust developmental outcomes with a consistent identity for the dominant gene at each cell, substantially reducing the uncertainty in the location of boundaries between distinct fates.

Figure 1

Cell fate determination through a morphogen concentration gradient needs to be robust against stochastic fluctuations. (a) A morphogen gradient across a cellular array results from the processes (shown in the inset) of secretion of molecules from a source located at the boundary of the domain, their diffusion across space, and degradation over time such that the decay rate is linearly proportional to its concentration. (b) Schematic representation of a Xenopus embryo where the differentiation of the cells of the mesoderm into dorsal and ventral fates (represented by blue and orange, respectively) is guided by the concentration gradient of the morphogen activin between the dorsal (D) and ventral (V) ends (displayed below the embryo). (c) The steady-state expression of patterning genes $A,B$ across a 1D array comprising $N$ cells, with the indices of the cells indicated by $i=1,...,N(=50)$, subject to a noisy morphogen gradient in the absence of interaction between the cells. Results of 300 different realizations are shown. (d) While in the absence of noise the boundary separating the regions with the two different fates corresponding to $B>A$ (blue) and $A>B$ (orange) is expected to occur at the same position across all realizations (the idealized situation shown at left), fluctuations in the morphogen concentration and gene expression dynamics results in variations across realizations (shown at right) if fate determination occurs only on the basis of positional information provided by the morphogen gradient. (e) Interactions between neighboring cells mediated by the notch signaling pathway (shown here schematically) can aid in the robust determination of fate boundaries in the presence of noise. Genes $A$ and $B$ comprising the morphogen interpretation module affect the expression of genes coding for notch receptors. The notch intracellular domain (NICD), released from the bound notch complex that results from the trans-activation of notch receptors, in turn up- or downregulates the expression of $A$ and $B$ (depending on the type of interaction).

Figure 2

Robust determination of cell fates results from interaction between stochastic gene expression dynamics and contact-mediated signaling. The intercellular interactions mediated by the notch downstream signal ($S$) can be classified into four types, determined by which of the patterning genes ($A$ or $B$) is up/downregulated by $S$, as represented by the motifs shown beside each panel (a)–(d) [arrows as in Fig. 1]. For each type, the spatial pattern formed by cells adopting distinct fates $A$, $B$ in a 1D domain comprising $N(=50)$ cells subject to a morphogen gradient is characterized by the location $l_B$ of the boundary [$\sim 20$, in absence of any interactions between the cells, see Fig. 1] demarcating the segments expressing the two fates. The variance in $l_B$ across 300 stochastic realizations is shown for each choice of the pair of parameters quantifying the strength of intercellular coupling, viz., $J$ representing critical value of patterning gene expression segregating low/high receptor production and $Q$ representing critical signal intensity that distinguishes between weak and strong regulation of patterning gene expression. The continuous curves in each panel are contours indicating the variance in $l_B$ in the absence of intercellular interactions ($\simeq 1.38$). The regions in the $J\text{−}Q$ plane above the broken curves (shown in white) correspond to the mean value of $l_B$ lying within [10,30], i.e., $50\%$ of its value in the uncoupled case. Note that, for coupling types in which $S$ upregulates $A$ either directly (b), or indirectly via suppression of its inhibitor $B$ (a), fluctuations in $l_B$ are markedly reduced over a wider range of $J$ and $Q$. (e, f) Temporal evolution of the expression of $A$ and $B$ shown for cells around $l_B$ for the uncoupled case, contrasting (e) the dynamics seen in absence of any intercellular interactions, with (f) that obtained when $S$ inhibits $B$ [as in panel (a)]. While the uncoupled cells exhibit large fluctuations in expression levels with uncertainty in $l_B$ sustained for a long time, in the presence of intercellular interactions cells rapidly converge to their eventual fates.

Figure 3

Reduction in variability of response to fluctuating morphogen concentrations by contact-mediated signaling. (a), (b) Steady-state distributions for the expression levels of the patterning genes $A$ and $B$ shown for cells located around the respective positions of the fate boundary when (a) intercellular interactions are absent, or (b) the notch downstream signal $S$ suppresses expression of $B$ [as in Fig. 2]. In the uncoupled case, the distributions are extremely broad with a high degree of overlap close to the fate boundary. Interactions, on the other hand, result in sharply defined peaks at very low and high values, with the dominant gene at a given cell clearly identifiable. This leads to a steady-state expression of the patterning genes [shown in panel (c) for a 1D array of 50 cells] that exhibits a robust, sharply defined cell fate boundary (at $i\approx 12$) even in the presence of a noisy morphogen gradient. Results of 300 different realizations are shown. (d), (e) Time-series (left) and corresponding distributions (right) of patterning gene expressions for a cell placed in a morphogen gradient when it is (d) isolated or (e) coupled to a neighboring cell. The coupling-induced separation of $A,B$ reduces the uncertainty in fate determination. (f)–(i) The numerical results (f, uncoupled; g, coupled) are supported by analytical calculations for a pair of coupled cells using linear noise approximation where the high degree of uncertainty as measured by the normalized Fano factor $f_A$ (broken curve) in the absence of coupling (h) decreases markedly on including contact-mediated signaling between the cells (i). Shaded regions represent the position-dependent steady-state expression levels for $A$ and $B$.