sustaining or changing in a contingent way. The state of differentiation is not washed out by being blended with that of neighboring cells. One cell or few cells can be the progenitor of a whole complex structure even if isolated from the rest of the organism (e.g., grown in vitro). As cancer shows, and may be important to many normal traits as well, a mutant somatic cell can generate a clone of cells with altered function. Cellular commitments are achieved in various ways; for example, the differentiation state can affect chromatin structure that leaves some genes open for TF binding, whereas others are closed to regulatory machinery (Davidson 2001; Davidson, Rast et al. 2002; Davidson, Rast et al. 2002; Lieb et al. 2001).

In the chicken-and-egg discussion raised in Chapter 4 about RNA and DNA, the regulation of an organism is in this view necessarily hierarchical, all kick-started by parental mRNA and proteins produced in the zygote. The increasingly committed nature of cells that occurs during development is the result. The new zygote is but another form of single-cell precursor to an autonomously developing tissue cascade. Of course, this is hierarchical with respect to the zygote but not the process itself. The process is circular, since it was the same genome that made the maternal mRNA and will eventually make that of the germ cells of the next generation.

Because sequestration is necessarily incomplete, cells are able to continue differentiating in contingent, context-dependent ways, but in some cases can re- or de-differentiate. The degree of reversibility that occurs in nature varies. In a mature animal embryo, most cells are unable to differentiate into more than a limited set of tissue or cell types; artificial cloning from somatic (body, as opposed to germ) cells in mammals, for example, requires special treatment of the cells to dediffer-entiate them. However, the nature of commitment depends on species and even context, and not all cells are fully committed to some end state. Reptiles can regenerate entire tails or limbs, and humans can regenerate blood vessels and skin, from cells that are partially differentiated even in adults. Cells in a developing brain may in some instances be entrained to behave like cells amongst which they intercalate (see Chapter 15). Plants are rather different in this respect, with much more flexibility built into the basic way they do business (for example, in some, the tip of any branch able to form flowers).

Implications of Pleiotropy in Complex Regulatory Circuits

If the pleiotropic use of a limited set of regulatory genes makes complex evolution easier to explain, it entails the potentially serious problem that mutation in a TF could disrupt all of its different functions simultaneously, which could be a disaster

Figure 7-4. Aspects of regulation by and of genes. (A) Hox combinatorial expression figure; similar anterior-posterior colinear expression affects both invertebrates and vertebrates; gray shade is used to indicate the gene class and the most anterior location of its expression along the axis where the gene has its predominant effect, as shown by corresponding shading in the two types of animal; for example, darker-shaded genes are expressed posteriorly and affect caudal regional development; (B) schematic of regulatory circuits, hierarchies, and batteries, illustrated by a schematic of the use of the TF Cdx to regulate Hoxc8 that in turn is a selector for many other genes, in patterning early caudal vertebrate development; the diamond symbol represents an enhancer sequence that flanks Hoxc8 (e.g., see Shashikant et al. 1998).

for the organism (e.g., Losick and Sonenshein 2001; Struhl 2001). However, because transcription factors work via regulatory elements, there is a ready escape: mutations that add, delete, or modify a particular RE will only affect the expression of its nearby gene and hence only that particular use of the TF that recognizes the RE.

This provides room for tremendous evolutionary flexibility, but there is also a potential flip side. If the same TF is used in many contexts, an organism is vulnerable to a high degree of multiple jeopardy. A given TF may use hundreds of REs scattered across the genome in its different functional contexts. Relying on one sequence in too many ways raises the likelihood that one or more of these REs will be mutated, and that could be harmful. There are at least two ways to protect a gene against this kind of mutational regulatory "noise" (Sengupta et al. 2002). One is to flank it with multiple copies of the relevant RE; Figure 4-7 provided an example of enhancers upstream of lens protein genes (this shows between-species variation, but the principle is the same). As was shown in Table 4-5, another mutational buffer is for a TF to evolve tolerance for variation in its recognition sequence. Not all the sequences are equally efficient at binding the TF, but they are at least recognized.

However, this same robustness that buffers a gene's enhancers from being erased by mutation makes it more difficult for evolution to remove that gene from the TF's set of regulated genes—it could take many mutations to delete all the relevant REs for a given gene to completely destroy the binding site (as opposed just to making it less efficient and, for example, lowering the expression level of the gene). This could tie functions together over evolutionary time once the use of REs for a common TF is established, yielding suites of traits that evolve together, at least for a period until mutational variation accumulates. During that time, traits might appear to share adaptive constraint, and could possibly be wrongly interpreted as of apparent canalization of traits, of selectively constrained variation, evolutionary stability, or range of response to environmental changes (e.g., Siegal and Bergman 2002). This is speculative, but a balance between robustness to mutation and stabil-

Figure 7-5. Schematic of regulatory element evolution by duplication and modification by mutation (lightning). Short elements can also be created or destroyed by mutation in otherwise random sequence. Bent arrow indicates start and direction of gene transcription.

ity on the one hand and evolutionary flexibility on the other may be the consequence of the network or mesh-like nature of developmental and homeostatic controls.

0 0

Post a comment