Organogensis

VI.1 COORDINATORS
VI.2 PARTICIPANTS
VI.3 SUMMARY
VI.4 INTRODUCTION
VI.5 SPECIFIC AIMS
VI.6 BACKGROUND AND SIGNIFICANCE
VI.7 THEORETICAL FRAMEWORK
VI.8 PRELIMINARY RESULTS
VI.9 RESEARCH DESIGN AND METHODS

• VI.9.ii Limb Development
  • VI.9.ii.a Overview
  • VI.9.ii.b Computational Approach
  • VI.9.ii.c Subprojects
    • VI.9.ii.c.1 Subproject 1 — Role of fibronectin in precartilage condensations
      • VI.9.ii.c.1.i Experiments
      • VI.9.ii.c.1.ii CPM simulation of condensation
    • VI.9.ii.c.2 Subproject 2 — Activator-inhibitor interactions in skeletal pattern formation
      • VI.9.ii.c.2.i Overview
      • VI.9.ii.c.2.ii Background and significance
      • VI.9.ii.c.2.iii Hypothesis
      • VI.9.ii.c.2.iv Models
        • VI.9.ii.c.2.iv.a Cellular automata simulation of reaction-diffusion plus haptotaxis
    • VI.9.ii.c.3 Subproject 3 —Viscoelasticity of limb bud tissues
    • VI.9.ii.c.4 Subproject 4 — The Genetic Control of Limb Development
      • VI.9.ii.c.4.i Introduction
      • VI.9.ii.c.4.ii Goal
      • VI.9.ii.c.4.iii Animal and Tissue Models
      • VI.9.ii.c.4.iv Experimental Approach
    • VI.9.ii.c.5 Subproject 5 — Complete model for avian chondrogenesis
      • VI.9.ii.c.5.i Overview
      • VI.9.ii.c.5.ii Background
      • VI.9.ii.c.5.iii Hypothesis
      • VI.9.ii.c.5.iv Model and preliminary results
      • VI.9.ii.c.5.v Plan
• [ Complete VI.9 Outline ]

VI.10 RELATIONSHIP TO CYTOSKELETON (PROJECT 2) AND BIOLOGICAL NETWORKS (PROJECT 1)
VI.11 TIMELINE

< Previous | Page 18 of 27 | Next >

VI.9.ii.c.2 Subproject 2 - Activator-inhibitor interactions in skeletal pattern formation:
VI.9.ii.c.2.i Overview:

This Subproject consists of two parts:

1. Experimental and computational research to investigate the activator-Inhibitor interactions to determine a realistic set of parameters which allow formation of typical cell patterns in cartilaginous condensations chick limb.
2. Integrating reaction-diffusion of molecules like TGF-βs and fibronectin, with models for cell dynamics, differentiation, and mitosis within the CPM framework.

VI.9.ii.c.2.ii Background and significance:

Well-defined developmental axes during limb growth and pattern formation led to the suggestion that positional information, in which morphogen gradients create a three-dimensional chemical address is a possible mechanism for skeletal development (Wolpert, 1989; Tickle, 1994). Each cell then differentiates in a manner dependent on its position using a complex genetic program to create the spatial distribution of skeletal elements that appears during development. Another possibility is that self-organization mediated by biologically relevant molecules such as the FGFs, TGF-βs and fibronectin induces patterned cell differentiation. For example, a cell might differentiate if the concentration of a specific chemical species were above some critical level. These two scenarios differ profoundly. Positional information requires very simple morphogen gradients and complex genetic programs, while in self-organization the response of the cell can be very simple and the complexity lies in nonlinear patterning. Hybrid mechanisms are possible (Newman, 1996), but the question remains which primarily determines the skeletal pattern.

While earlier work indicated that one or more TGF-β growth factors is capable of acting as a Turing-type activator which gives rise to quasi-periodic patterns of precartilage condensations (Newman and Frisch, 1979; Newman, 1996), the molecular mechanism of perinodular inhibition, required by this class of models, has been elusive. Our recent experimental work shows that FGF receptor 2 mediates lateral inhibition of chondrogenesis and arises early in condensation, precisely at sites of precartilage condensation (Moftah et al., 2002. As a theoretical approach to these questions we have expressed the basic mechanisms of differentiation patterning in the form of reaction-diffusion equations (Eqn. VI.10). These equations model the interplay of several factors, including activators representing TGF-β and lateral inhibitors such as those FGFs elicit.

VI.9.ii.c.2.iii Hypothesis:

We believe that the best recent evidence suggests that the early stages of mesenchymal condensation and differentiation into cartilage depend on complex nonlinear dynamics. This dynamics, involving several growth factors, and differentiated cells with different receptors capable of responding to these factors, ultimately produces tissue domains enriched in fibronectin. Within these domains cells accumulate by adhesive interactions. (see reviews in Newman and Tomasek (1996); Hall and Miyake (2000); and in vitro studies Miura and Shiota, 2000a; 2000b; Miura et al., 2000; Moftah et al., 2002).

VI.9.ii.c.2.iv Models:
VI.9.ii.c.2.iv.a Cellular automata simulation of reaction-diffusion plus haptotaxis:

The Alber group has undertaken a lattice gas cellular automaton (LGCA) simulation of limb mesenchymal pattern formation in vitro, as a way of investigating the role of reaction-diffusion mechanisms in a simpler format than the complete limb simulation we are aiming towards (see below).

The experimental system is the same as the one discussed above, in Section VI.9.ii.c.1.i. Investigations using the cellular automata model will be directed towards simulating results such as the fibronectin transfection experiment described above, as well as already established results in this system, such as effect on the condensation pattern of interfering with cell-fibronectin interactions (Frenz et al., 1989a, b; Downie and Newman, 1995), of stimulating the cells with TGF-β (Leonard et al., 1991), of enhancing lateral inhibition by addition of FGF2/FGF8 (Moftah et al., 2002), and of attenuating inhibitor production by antisense blocking of FGF receptor 2 (Moftah et al., 2002). We also plan to investigate the patterning difference between wing and leg mesenchyme in vitro, which we have previously associated with differences in responsivity to grow factors and differences in fibronectin synthesis levels and utilization (Downie and Newman, 1994, 1995).

The model is summarized as follows:

First, cells are modeled as directed points whose default behavior is diffusion by random walk on a square lattice.

Second, a cell-driven reaction-diffusion process that depends on the interaction between two molecular species (a diffusible activator, A, identified with TGF-β, and a faster diffusing inhibitor, B) is simulated on the lattice. To specify the molecular species as cell products, the reaction step of the reaction-diffusion process is permitted to occur only at nodes of the lattice that contain cells.

Third, when threshold levels of activator are encountered, cells respond by producing a secreted, but otherwise immobile, molecular species, identified with fibronectin, to which they attach.

The simulation scenario is as follows: Cells are initially homogeneously distributed and undergo typical random motion. There is initially low basal production of TGF-β (A) and inhibitor (B). These molecules diffuse freely through the extracellular matrix, but the inhibitor diffuses more rapidly. Production and release of TGF-β and inhibitor by cells increases with increased levels of TGF-β. This increase is linear with respect to TGF-β levels until a threshold cell production. Increased levels of inhibitor decreases both TGF-β and inhibitor levels. At a threshold level of TGF-β, cells secrete fibronectin. Fibronectin diffuses for only one time step before being deposited within the extracellular matrix. The relative rates of cell attachment to and detachment from fibronectin favor attachment.

Fig IV.6 shows the results of a cellular automata simulation of chondrogenesis in vitro (right panels) with two different levels of inhibitory response for the reaction-diffusion system (high, upper panel; low, lower panel). These are compared with experiments from Moftah, et al. (2002) (left panels) showing a culture in which the level of lateral inhibition had been increased by treatment with FGF-2, (upper panel) compared to a control culture (lower panel).

In addition to this experiment several other experimental results have been simulated with realistic parameter changes. Once we have arrived at a model that accounts for all known experiments we plan to challenge it by simulating new experiments (e.g., knockdowns of fibronectin by interfering RNA) and testing whether the model has predictive value.