Operating Principles For Regulatory Systems That Mediate Developmental Specification Events

Specification processes are those by which initially pluripotent embryonic cells choose among diverse fates, and are allocated to a given spatial territory or anlage. To understand the role of cz's-regulatory elements in specification it is necessary to think of them as devices that make choices: they produce alternative outputs depending on the sets of inputs with which they are confronted in different cells. As discussed in Chapter 1, in each cell the cz's-regulatory inputs consist of the concentrations and the activities of the nuclear transcription factors for which the element includes target sites. Regional differences in transcription factor presentation that are set up in different spatial domains of a pluripotential field of cells usually depend on intercellular signaling, or very early in development, on the location of the nucleus with respect to the spatial coordinates of the egg.

The primary principle of the mechanisms by which specification functions are executed is that specification depends on cz's-regulatory transformation of input patterns into spatial domains of differential gene expression. This is the essential job performed by cz's-regulatory elements at the initiation of each phase of development.

A second principle is that as-regulatory elements that function in specification always consist of assemblages of diverse target sites, because they always require multiple inputs. No such module consists of sites for a single "master regulator," a fantasy of earlier days. This is because as-regulatory elements that make specification choices work by binding sets of factors, the presence of which individually may depend on signaling events, cell cycle activity, temporal state, lineage, or spatial position (Fig. 1.3). as-Regulatory elements that execute specification functions work by interpreting multiple inputs of these and other kinds, to produce a single output in each nucleus.

A third principle is that in specification processes czs-regulatory output is novel with respect to any one of the incident inputs, and is more precise in space and time than these inputs. It is because of this feature that we use terms such as "information processing" in respect to as-regulatory specification events.

A fourth principle is that if a high specificity binding interaction between cis-regulatory DNA and nuclear proteins from the relevant cells can be detected in vitro by a physical measurement, then this interaction will have some function in vivo. If there are multiple sites for a given factor it may (or may not) be difficult to distinguish the roles of individual sites, but that species of interaction will in some manner be necessary for the overall function of the element. Thus the qualitative complexity of the internal functional elements of the as-regulatory module is given by the complexity of the assemblage of target sites within the module. That every specific type of interaction within a regulatory system that can be detected in vitro is fundamentally significant was shown explicitly for two sea urchin as-regulatory systems, both discussed below. These are the cyllla gene, which encodes a cytoskeletal actin, and the endol6 gene, which encodes a secreted embryonic gut protein. In both cases all sites of interaction were mapped for which the affinity of the regulatory protein for the specific site, as opposed to its affinity for a double-stranded synthetic polydeoxynucleotide, exceeded 5-10 x 103 (for cyllla, Calzone et al., 1988; Theze et al., 1990; for endol6, Yuh et al., 1994). Detailed gene transfer studies subsequently revealed that each of the nine species of interaction in the cyllla regulatory system, and each of the ten species of interaction in the region of the endol6 system so far examined contributes to the regulatory performance of these systems. A philosopher (or an evolutionist) might with some justification remark that it had better be so. It is very unlikely that highly specific site clusters, which are of improbable random occurrence within a few hundred base pairs of DNA, would have no function. In fact it is so, and there are some practical consequences. The most important is that at our present level of knowledge it is difficult to identify all the subfunctions executed within a as-regulatory element, except by first mapping its target sites.

A fifth principle is that as-regulatory elements which execute specification functions usually utilize negative as well as positive inputs, i.e., interactions with transcriptional repressors as well as with activators. The repressors set spatial boundaries of expression, while the activators are spatially more widespread

(and are thus used for multiple genes and multiple purposes). Like much else that we have learned by probing within the czs-regulatory module, this aspect of DNA design logic has been more easily appreciated in hindsight, and was not initially anticipated.

+2 0

Post a comment