Chlck I Hoxc6 \ I Hoxa9 \ Hoxd9

Hoxc8 Hoxb9 Hoxc9

C1 T1

00000000000000000000000000000/ Goose 2 Hoxc6

□ □□□□ // □□□□□□□□□□□□□□□□□□□□□/100s of thoracic vertebrae ooooooooo /ooooooooooooooooooooo/

5 6 Hoxc6 Pyth°n Hoxc8

Legend: Q Vertebrae Q Somites

| Spinal nerves contributing to brachial plexus

Figure 9-8. Functional conservation of axial Hox expression boundaries in vertebrates. Despite differences in detail, such as the number of vertebrae, gene expression patterns along AP axis are conserved relative to major homologous boundaries in the organism, such as the origin of nerves to the brachial plexus (that services the forelimbs). For corresponding body regions in other animal taxa, see (Carroll, Grenier et al. 2001). Redrawn from (Burke et al. 1995).

them (vertebrae and ribs, for example, that themselves vary and/or have patterned subunits).The homologous boundary of Hox gene expression switches shown in the figure indicates that an earlier broader patterning process is in place as the Hox system engages. Such layering of what may be rather similar kinds of patterning process may be involved in the structure of most meristic traits (and that means a high fraction of all traits in complex organisms).

Each of the roughly 800 ommatidia in the compound eyes of fruit flies has a basically fixed number of cells that are programmed, or patterned, by a reaction-diffusion-like activation-inhibition system. Each has about 20 cells of seven different types, including eight photoreceptor cells, seven of which are arranged hexagonlike around the other (known as R8); six surrounding accessory cells; and four lens-generating cone cells. The eye develops from an imaginal disk of about 20,000 cells. The cells in the eye imaginal disk are apparently equivalent and equipotent (e.g., all expressing the Eyeless/Pax6 TF). These cells are induced to form omma-tidia by a sweep of activation-inhibition activity induced by an indentation called the morphogenetic furrow that moves wavelike from posterior to anterior across the disk.

As the furrow passes it induces cells to express Hh (a Hedgehog gene family homolog of the vertebrate Shh), which then induces Dpp (the Bmp homolog) in the furrow to drive it anteriorly. In the R8 cell, the SF Boss (Bride of sevenless) is released and induces the Sevenless gene in the neighboring R7 cell in the little prephotoreceptor cluster. This signal to R7 initiates a cascade of differentiation of the other photoreceptors and surrounding supporting cells. The Wingless SF is also active in the disk as an inhibitor expressed ahead of the morphogenetic furrow, which helps establish anatomic polarity in the eye. Again the result is a somewhat elaborate pattern of structures generated by a single and relatively simple quantitative, nested signaling process.

Growth and inhibition in Patterning

By themselves, pattern-generating processes might produce continually changing patterns, but the system can be stopped in various ways including apoptosis and mineralization (e.g., calcification of teeth and vertebrae, the moving away of shell material from the active mantle cells). One mechanism that establishes inhibition zones surrounding initiation sites is the Delta-Notch system (Table 7-2). This is used in the establishment of sensory bristles and ommatidia in insects and in feathers, teeth, and other structures in vertebrates. In the case of bristles (whose structure and nature will be described in Chapter 12), activation zones arise in a set of comparably prepared cells; cells expressing high Delta levels inhibit surrounding cells, and a few of those cells become the precursor of bristles. The successful cell (if being a bristle is more "successful" than being the surrounding tissue!) arises by chance.

The similarity of biological patterns to patterns simulated by computer (noted above) justified the inference that activation-inhibition processes were responsible for such traits in the real world. This was confirmed with the examples we have discussed. However, as so often happens, evolution frustrates what experience excites. The early Drosophila egg manifests stripes of expression of various genes, including Even-skipped (Eve) (the gene was named for the effect of its mutants on the stripes of early fly embryos). This seemed a clear example of a reaction-diffusion system, yet experimental work has shown that there are separate stripe-specific Eve expression mechanisms (Carroll, Grenier et al. 2001; Ludwig et al. 2000; Ludwig and Kreitman 1995; Ludwig et al. 1998). There is no rule without an exception—or is it that there is no exception without a rule?

Important Aspects of Quantitative Patterning

Turing's and other similar processes can be modeled mathematically in terms of the relative properties of interacting factors. For many years this was thought to be fanciful but rather meaningless (or even mystical and nonscientific) biological theorizing, because it could not be operationalized (proven in experiments, for example). As we have seen, that is certainly now done successfully. In a sense these processes can be viewed as "mathematical" in that it is not necessary to know what the factors are. If they have the specified properties, the effects are seen automatically. This can be said whether the process is by diffusion signaling, mechanical interactions, or temporal or physiological interactions among molecules rather than among physical cells. The result will be pattern; it can be ever-changing, divergent, or steady state depending on the parameters of the interactions.

This is an example of what we mean by the logic of gene use and mechanism in biology. We referred earlier to the functional arbitrariness but logical necessity of mechanism, meaning that it did not matter for the specific structural attributes of a trait what patterning process brought it about. If we think of processes in terms of their logic, it is the interaction of entities that is the process and, in a sense, not the entities themselves. Unfortunately, mathematical verisimilitude does not imply identity. There may be multiple mathematical ways to generate a similar outcome, which can be expressed one in terms of the other. In terms of the logic of patterning, it may not matter which equations are used; but for discovering the actual mechanism in any given case it matters considerably how many and which genetic factors are involved—and this becomes especially important in trying to understand homolo-gies in species or processes we can't study directly from ones we have. Phenogenetic drift shows this as well.

But not all repetitive patterns can be simulated by a single set of equations, and in fact, meristic structures raise all sorts of subtle questions about homology (e.g., Hall 1999; Wagner 1996). The usual description of meristic traits is that they are represent serial homology. The notion there is that each unit is a separate use of the same process, not among species but among regions of the same embryo.

Reaction-diffusion-like processes can generate repeated structures as described, and it is tempting to explain any such pattern by these elegant processes. When true, each iterate of the structure (e.g., each hair or scale) is part of a single continuous process. However, by a different route, each unit might more literally be a repeated but independent use of the same process, something more akin to the idea of homol-ogy between structures. As with reaction-diffusion processes, the same genes would be used in each unit. Mutations in those genes would therefore affect all units, demonstrating the unity of the overall system.

Another way to generate repeat units is for the same genes to be used, but to be invoked by separate means in each part of the embryo. In such case, unlike a standard reaction-diffusion patterning system, the separate domains could be sequestered from each other. For example, the regulatory region of a gene that is involved in the cascade that develops each unit can have different enhancer cassettes, each responding to different TFs and in a specific compartment. This is the case, for example, with the Eve gene in Drosophila, where enhancers for different TFs control expression in different stripes in the egg and this helps establish subsequent AP segmentation. Here, mutation within the regulatory region will affect only those units that use the specific enhancer that was mutated.

There are probably instances of all three of these phenomena, and the latter two require that in some way a prepattern is laid down so that the repeated invocation of the same expression cascade takes place. Local reaction-diffusion like processes may be involved.

The patterning of butterfly wing spots seems to be a good example of these phenomena (e.g., Carroll, Grenier et al. 2001;McMillan et al. 2002; Nijhout 1991;Nijhout 1994). Wings are divided into cells (compartments, separated by veins), within which quantitative activation-inhibition patterning involving the development of expression sites ("organizers") of the genes Dll and En and others, leads to a ray of Dll that grows from a stripe of expression along the distal wing margin to the middle of each wing segment. There, it becomes a focal location of expression for a color spot. Other genes, involving the veins separating the segments, participate, and the genes include some of our familiar early polarity-specifying genes (Dpp, Wg, Rhomboid, and others) (Keys et al. 1999; McMillan, Monteiro et al. 2002).

Artificial selection experiments in which butterflies were bred to have larger or more, or fewer or smaller spots, have found that the effects are correlated among spots in the same individual (Beldade and Brakefield 2003; Monteiro et al. 1994). The effect of selection was to favor whatever combinations of genes might affect spotting activity. The result shows that in a sense the wing spots together constitute a single trait. However, other experiments have shown that mutations can affect (e.g., delete) individual spots within the row of spots (each in a different wing compartment), without affecting the rest of the spots (Monteiro et al. 2003). This reveals a strong modular element of control than normally achievable by a single reaction-diffusion-like process, and suggests that a complex enhancer was mutated. Some prepatterning process must, however, still be present, even if moved earlier in development.

Here we have an example of multiple components generating a serially homologous trait. The interpretation of the various results is not yet unambiguous, but begins to show how a complex meristic patterning process can evolve and vary (and scales are developmentally related to sensory bristles, and the focal spot of Dll is produced by the coopting of an evolutionary prior use of Hh signaling to separate anterior and posterior wing compartments) (Carroll, Grenier et al. 2001; Keys, Lewis et al. 1999). It may not be a simple, single patterning process. But it is a series of familiar processes, occurring apparently separately, within the cells of the wing, whose nature and origin are as we would expect.


Most animals are hollow tubes relative to their overall AP, DV, and left-right symmetries, but they have structures like limbs and antennae that grow outward from placodes or imaginal disks along the main axis, as we described above. A variety of new patterning signals are then expressed in spatiotemporal order, to establish the new main proximodistal (PD; sometimes called mesiodistal) axis as well as DV and AP axes within the outgrowing structure.

The same genes can be involved in multiple axes even within the same structure. A good and well-studied example is the formation of tetrapod limbs (Carroll, Grenier et al. 2001; Davidson 2001; Gilbert 2003). Genes from all four Hox clusters are used first to establish the overall AP axis. Then, locally, HoxD9 and 10 are expressed as the most proximal region, or stylopod (e.g., humerus), develops in the outgrowing limb bud in a summed-sequential way, roughly from anterior to posterior. Along the PD axis, HoxA9-13 are expressed. In the formation of the next segment, or zeugopod (e.g., radius, ulna), HoxD10-13 are expressed in an AP summed-sequential combinatorial manner. Finally, as the autopod (hand) forms, HoxA13 as well as HoxD10-13 are expressed again in an AP manner as the digits form (in an axis separate from the PD axis, or in a modified, curved single axis that wraps around the end of the limb to induce the digits).

In each case, Hox expression is associated with the formation of cartilage models that later ossify as the bony elements. HoxC genes (along with other known genes) are expressed differently between forelimb and hindlimb. In this way, the same genes expressed at different times control different aspects of the hierarchical, regionalized, multiple-axis morphogenesis (see Figure 9-9).

0 0

Post a comment