Genetic+pattern+formation+and+modelling+morphogenesis+(Paulien's+critters)

Prev: Modeling gene regulaton (lac operon) Next: Multi-level modeling (conclusion)


 * TODO LIST**
 * REF dicty mechanism of aggregation (Savill ?) -> new page?
 * REF Turing systems -> new page? see this years course
 * REF drosophila stripes in embryogenesis -> new page?
 * REF E. coli 10% diff in genes in diff environments
 * REF squeeze cell apoptosis, stretch cell division
 * REF pressure induced local gene expression
 * REF regulation integrin pathway

=Genetic pattern formation and modelling morphogenesis= The previous example was an example of **systems biology** sensu strictu. Now we more to an extremely artificial example and study the interaction between evolution of genetic pattern formation and modelling morphogenesis.

So far we looked at what sort of patterns are generic:
 * e.g. spiral waves as generic property of excitable media
 * cell sorting
 * turing systems (activator / inhibitor): the main aim here was how to get non-uniform states from a uniform initial condition (REF). This was thought to be important for biological systems. This can be achieved by an inhibitor which diffuses faster than an activator and can give stripes.
 * very analogous to zebra stripes, even including interference patterns and unique zebras!
 * however is it relevant for biology to focus on getting inhomogeneity from homogeneity?

Perhaps the turing patterns are an example of **applying a model that we know gives the desired result!** For example in drosophila 7 stripes form during embryogenesis (REF). Are these then turing patterns? Now we know that each stripe is regulated in its own way! And that is probably about as ugly as things can get relative to elegant simple models! (i.e. making 7 stipes in 7 different ways!).
 * 1-dimension: turing patterns work fine to explain the pattern
 * 2-dimensions: here it becomes harder because now spots also become possible and one needs to be more precise
 * 3-dimensions: even harder and then also including tissue growth makes it even harder.
 * No longer such a comfortable model fit!

(//Annekdote from Crick about Turing: "The stipes are easy, but what about the horse part?")//

On the other hand excitable media does play a role in Dichtostelium:
 * shows that generic properties such as spirals can generate a mechanism of aggregation (REF)
 * but what is its importance?
 * **Wow**: we get simple to complex!
 * **But**: perhaps more often we should be focussing on complex to complex (e.g. drosophila prepattern is not achieved in a simple way)


 * Our aim here is therefore to try and get some handle on complex to complex mapping, and therewith get some fundamental insight into complex systems!**

First it is important to note that some defining properties of biotic systems are ignored in conventional modelling:
 * 1) **There are high design degrees of freedom**. Not simple to complex, but complex to complex, e.g. in //E. coli// from different environments have 10% differences in actual genes (REF). We don't have to be finicky about having 1 more molecule, but we do it in modelling!
 * 2) **Biotic systems are evolved!** They are not just arbitrary self-organized (i.e. generic). So in what sense are they "specific cases"? Are they rare cases, i.e. only a small subset of all DNA strings will generate organisms? Can we see evolutionary signatures? And what are their non-generic generic properties: from these rare cases can we still say general things?
 * 3) **Intertwining levels of organization.** Separation may 'beg the question' and be a too free parameter choice. By intertwining we can focus on general properties.

So we study complex to complex mapping:
 * rare cases in high dimensional specification space
 * evolutionary dynamics under dynamic constraints, e.g. genotype-phenotype mapping
 * evolutionary signatures
 * no //a priori// "search image"

Therefore when going from DNA to mutlicellular organism:
 * NOT: a gene "does what it does" (population genetics), i.e. not appointing genes, but evolved.
 * NOT: dynamical system "does what it does" (cf dictostelium morphogenesis) as a generic property.
 * BUT: how does inherited information interact with physical systems with a "dynamics of their own"

So how to find these rare cases?
 * evolve them, but don't know where to go
 * use fitness criterion which is necessary, but sufficient, and //not// on the phenomena of interest, otherwise it becomes trivial
 * study the side-effects of getting that
 * and study what evolves ....

__//Studying the evolution of morphogenesis (i.e. the horse part of the zebra) as side-effect of cell differentiation and differential adhesion//.__ Combine: Then:
 * within cell dynamics (gene regulation)
 * between cell dynamics (signal transduction + adhesion)
 * evolutionary dynamics (fitness as how many cells differentiated: i.e. not morphogenesis directly!)
 * evolve these **critters**
 * generic form: **blob.** So how not to be a blob?

We know from experimental evidence:
 * when we **squeeze** a cell we get **apoptosis** (Chen et al 1997, Rurosalthi 1997)
 * **stretch**: cell division and growth
 * **pressure** induced localized gene expression via beta-ceterin? pathway (adhesion molecule) (Scot et al. 200?).
 * regulation of integrin patchway: all kinds of things regulated by that: i.e. very fundamental role in regulation.

Model

 * simplification of the above, but what kind of simplification? (cf Einstein)
 * CPM formalism: which is easily extended
 * increave V (target volume)
 * squeeze induced apoptosis (lambda too small): too far from target V, the cell dies for free in model
 * stretch induced growth (if v > V + tau then V++)
 * cell differentiation (Jij to J'ij)
 * polarity induction (if not update Ok++, Ok=Ok/SumOk???)

In the model we want:
 * development -> gene regulation -> evolution (**+ feedback from evolution to gene regulation and development and gene regulation to development)
 * cell same type adhere and to other cell differently: **adhesion dynamic and evolvable**
 * start with one big zygote and initialize with a few fixed cleavage processes
 * after that dynamics of system can continue to proceed: only induced by growth (mechanical instabilities)

To get away from blob, cell differentiation is a necessary condition, but it is not sufficient! Selection is only on cell types, i.e. attractors of gene regulation:
 * gene regulation has 2 inputs: from own cell or neighbouring cell

Results

 * Various types of critters evolve (Plate 1)!
 * Their form is not deterministic, but common features arise:
 * elongated stalks (2 different cell types and gene regulation attractors). Differentiation depends on neighbourhood.
 * turns inside out and final state (nice for animal not for plant!)
 * So different morphs arise from different evolutionary runs (different random initial conditions): //independent planets//.

So how do we get different cell types from one zygote and one cell attractor?
 * flip maternal factors to allow differentiaiton.

Moreover: Morphogenesis as a sustained transient of energy minimization
 * **rare cases**: are not found without evolution (i.e. from random regulation network don't get them), but only sometimes with evolution (10-20% of evolutionary runs)
 * **non-generic**: never see the same things happening again!
 * However: many morphemes but just a small number of (generic) mechanisms which make non-blobs
 * engulfing
 * meristematic growth
 * intercalation
 * convergent extension
 * budding
 * i.e. can give these things a name relative to processes that have been identified in actual morphogenesis!
 * these are coupled to type of cell differentiation
 * stable attractors (memory)
 * signal dependent: Wolpert (like extreme statements): //Plants are not interesting because cells don't have memory.// (Can very easily redifferentiate)
 * history dependent cell differentiation (long state cycles)
 * this can be seen as a quantitative difference: how hard to disturb or not.
 * intrinsic conflict mainted by cell differentiation

In development:
 * So what about evolutionary history?**
 * we see many different forms of critters on the way
 * early evolution: just cell differentiation, but still blobs and then reach maximal fitness
 * then evolution on neutral path: which is where nice shapes are produced
 * highly differentiated shapes are evolutionary quite close together
 * initial cell differentiation configuration is quite similar (i.e. primary cell differentiation is maintained, cf zootype)
 * then different cell layers develop in different ways
 * also similar shapes are revisisted: reinvention withn runs (NB: in other runs such morphs are never even seen!), i.e. some kind of evolutionary memory and information intergration
 * also re-invented in independent runs from same semi-evolutionary initial conditions

Next: Multi-level modeling (conclusion)

**Hogeweg P** (2000) Evolving mechanisms of morphogenesis: on the interplay between differential adhesion and cell differentiation. //J. theor. Biol.//, **203**: 317-333. [|MEDLINE]. [|DownLoad PDF]. **Hogeweg P** (2000) Shapes in the shadow: Evolutionary dynamics of morphogenesis. //Artif. Life//, **6**: 85-101. [|MEDLINE]. [|DownLoad PDF].
 * References**