Can Active Perception Generate Bistability? Heterogeneous Collective Dynamics and Vascular Patterning
Year: 2014
Type: conference-paper
Abstract: During morphogenesis (the generation of form), biological cells, agents or robots must collectively coordinate where and when to move. How to solve such complex, spatial problems in a timely manner, is fundamental to survival in biological organisms, though temporal regulators are largely unexplored. We take the generation of new blood vessel networks (angiogenesis) as our case study system, where tissues low in oxygen stimulate endothelial cells “ECs” (the inner lining of blood vessels) to grow new network branches. This requires ECs to take on heterogeneous states by collectively competing with one another for migratory status via lateral inhibition. We propose here that the traditional “decide then move” perspective of cell behavior in angiogenesis may miss a key temporal regulator as it is too slow to account for the rapid, adaptive assignment of heterogeneous cell states. Here we show that a “move and decide” view may provide a better account. In a study focused on an individual EC in a simulated collective, we find that active perception (sensorimotor feedback) can generate bistability through migration-induced cell shape changes. We further exemplify that when parameters affecting active perception are modulated, bistability is lost in the single cell. As a consequence, active perception can directly modulate collective decision timing. Introduction Collective behavior during morphogenesis requires timely coordination of many autonomous agents, which becomes increasingly complex if the task requires agents have heterogeneous and adaptive phenotypes. Understanding collective coordination mechanisms holds great promise for morphogenetic engineering of well-adapted robot designs (Doursat et al 2012). To this end, the vasculature in living systems is a perfect case study. New vessels grow and remodel in a dynamic adaptive way to maintain a network with near-intimate contact to every cell in the body, required due to the diffusion limit of Oxygen (Aird 2005). When new vessel growth (angiogenesis) is needed, e.g., in development or wound healing, endothelial cells (ECs) lining blood vessel tubes adaptively respond to the release of diffusing growth factors from hypoxic (low in oxygen) tissue. The ECs then collectively coordinate such that some cells migrate and lead new tubular branches (“sprouts”), while others line the tube walls and ensure that the branches are well spaced (Geudens and Gerhardt 2011) (Fig. 1a). The selection of these two states “active” or “inhibited” cell movement is known to be coordinated by Notch-driven “lateral inhibition”, where cells battle to inhibit their neighbors (Hellstrom et al 2007, Jakobsson et al 2009). Lateral inhibition is known to generate stable alternating patterns of on and off cell states in many systems, hereafter referred to as a Salt and Pepper pattern “SP Jakobsson et al 2010; Bentley et al 2014a). This occurs when ECs 1) meet new neighbors during branch fusion (required to form Figure 1. a) Endothelial cells, initially all the same, collectively coordinate to become heterogeneous when stimulated to grow new blood vessel branches. Some cells move, leading new sprouts, others are inhibited, lining the branch and keeping them regularly spaced. The optimal spatial arrangement is a “salt and pepper” (SP Fig 1b) (Bentley et al 2009) and 2) rearrange their positions in the collective (Jakobsson et al 2010). Thus, the cellular collective must be capable of rapidly re-establishing the S&P pattern of behaviors in the face of local neighborhood changes, to keep the network branching structure optimized. Otherwise, cells would drift into a “half-way house” state of homogenous movement (Fig. 1c). Lack of differential movement has been shown to disrupt branching as cells hypersprout instead (Hellstrom et al 2008), and has been implicated in abnormal vessel thickening in disease (Bentley et al 2014a). Here we ask, is there an extra dimension of temporal regulation that keeps the system from lingering in a homogeneous state as the process unfolds. Do cells “Decide then Move” or “Move and decide”? In Biology, signaling cascades regulating cell behavior are usually presented in a feed-forward manner (Fig. 2a). E.g. ECs are selected as migratory collectively, by Notch lateral inhibition and then those selected will migrate, they “decide then move”. However, the relative timing of events indicates that things may, in fact, be more complex. Lateral inhibition communication between the ECs can be summarized as follows: each cell detects a diffusing stimulatory signal from the hypoxic tissue, primarily Vascular Endothelial Growth Factor (hereafter input signal,“I”) by activation of its VEGFR receptors (hereafter sensors, “S”). Sensor activation leads to the up-regulation of the ligand Dll4 (“D”), which then binds to and activates Notch receptors (“N”) on neighboring cells. When N is active, the cell downregulates S. Several amplification cycles of this pathway then leads to one cell inhibiting the other’s ability to sense, more than the inhibition it, itself, receives (see Figs. 2c-e). The same sensors also trigger the migratory response of the cell, by locally activating the actin cytoskeleton to first polymerize, generating long thin membrane protrusions called “filopodia” that reach into the environment. Second, actomyosin contractions and adhesion dynamics propel the cell forward. Thus a cell that is more inhibited by Notch signaling, will have less sensors and so will be less migratory (the “inhibited” state). Lateral inhibition requires gene regulatory changes in the nucleus and protein synthesis, which can take on the order of hours. While faster than multiple cycles required for lateral inhibition, a full migratory response (e.g., where single cells rearrange their positions) also occurs on the time-scale of hours (Jakobsson et al 2010). However, the initial migratory response of filopodia growth occurs more rapidly (on the order of minutes) and locally (at the cell surface), without the need to alter gene expression. It is thus logical to assume that the cell shape changes (filopodia) and early movements of the cells membrane during the initial stages of migration will proceed ahead of the established collective decision making process of lateral inhibition. Consequently, we propose a “move and decide” strategy better captures the process.It is important to note that, the sensors themselves move rapidly during motion, as they reside on the deforming membrane surface. As filopodia extend and retract, sensors are moved to new locations in the environment, continually changing their input (Fig 2c). If the environment contains a gradient, this process has been indicated to speed up collective decision making (Bentley et al 2008). However, the exact reason for this has not been established. Here we propose that filopodia act as a means to perform active perception of the tissue environment, similar to the proposition that flagella motion confer sensorimotor dynamics and indeed minimal cognition to migrating E. coli cells (Van Duijn et al 2006). While the importance of embodiment in vascular morphogenesis has recently been highlighted (Bentley et. al., 2014b), here we further posit that such active perception confers bistability to the active/ Figure 2. a) traditional feed-forward view of cell behavior regulation in biology termed here a “decide then move” strategy. b) The alternative “move and decide” strategy. c) Signaling pathways involved in blood vessel selection of heterogeneous movement behaviors. Green dots indicate that sensors (VEGFR receptors) reside on the cell membrane, which is deformed and moved as actin polymerizes to form filopodia creating a sensorimotor loop (active perception) d-e) simplified model of (c) S = sensors, D = Dll4, N = Notch, I = input signal. Overtime cells battle back and forth to inhibit each other with one eventually acquiring more filopodia and inhibiting its neighbor more. b
Authors: Katie Bentley, Kyle Harrington, Erzsébet Regan
Institution: Harvard University
Language: English
Subfield: Modeling and Simulation
Field: Mathematics
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W2323602828