Note Smit's proper conceptual separation of initial cooptation from later retention by natural selection. Following both Darwin and Nietzsche, the secondary adaptive enhancement by natural selection (in this case, perhaps, merely a selective retention without further structural change) — that is, the promotion of the current utility — does not permit a conclusion about the different cause of historical origin (in this case, presumably as an exaptation [Page 1242] from repeated copies of a transposon, replicated and amplified for quite different reasons, and probably initially nonadaptive at the primary Darwinian level of the organismal phenotype).
The concepts of functional shifting, the structuralist implications thereof, the classical examples (especially the exaptation of feathers for flight), and the terminology of cooptation and exaptation, were all worked out on the conventional “playing field” of anatomies and behaviors of complex multi-cellular organisms. But this rubric of theory and argument will surely enjoy its greatest application in the domain of molecular evolution, where the functional redundancy of multiple usage for most gene products and the structural redundancy of duplications and repeated elements enlarge the scope of functional shifting away from mere striking illustrations, and towards ubiquity. (Note how this pairing, at the molecular level, of several functions for one gene with several genes for one function precisely matches the logic of Darwin's anatomical argument for the structural prerequisites of quirky functional shift — see p. 107 for my discussion in terms of Darwin's favorite example of lungs and swim bladders in the evolution of fishes.) Classical cases are already beginning to emerge from this level of analysis, perhaps none more complex and fascinating, or more widely cited, than the work of J. Piatigorsky, G. Wistow and many others on the eye-lens crystallins of both vertebrates and invertebrates.
The crystallins are structural elements that constitute about 90 percent of the total soluble protein of eye lenses in most vertebrates. Most crystallins are found in lens fibers, which lose their nuclei (and other organelles) and must therefore, since they cannot replenish by division, remain stable through the organism's life (Piatigorsky, 1992). In beginning an invited review to a society of ophthalmologists, Piatigorsky (1993a) explained (to a group of professional biologists who generally lack specialized training in evolutionary theory) how functionalist and adaptationist biases had led to longstanding assumptions that must now be discarded (1993a, p. 283):
As scientists and physicians we are accustomed to seeking order and purpose in the world. It is commonplace for us to think that specialized tasks require custom made instrumentation, the more sophisticated the mission, the more honed the result . . . And so for approximately 100 years vision scientists have considered the crystallins as a very limited set of highly specialized proteins especially chosen and designed for their ability to confer the required refractive properties onto the transparent lens. We have grown up with the idea that crystallins are as specialized as the eye itself.
Molecular and genetic studies of the 1980's and 1990's have dramatically reversed this view by identifying crystallins as a diverse set of exapted enzymes and proteins with strikingly different original functions, often still maintained. Piatigorsky continued (1993a, p. 283): “Recent studies, however, have changed this restricted view and have shown that crystallins are essentially borrowed proteins of diverse origins. These lens structural proteins [Page 1243] not only play a refractive role in the lens, but they have important non-refractive functions within and outside of the eye.”
Vertebrate crystallins have been divided into two groups (Wistow, 1993; Lee et al., 1994): the structural stress-proteins of the alpha and beta/gamma crystallin group found in most vertebrate lenses; and the highly diverse, so-called “taxon specific” crystallins, generally found in more restricted lineages, and exapted from enzymes that continue to operate in their earlier manner elsewhere in the body (and often in the lens as well).
The structural proteins of the first group also represent exaptations, rather than direct adaptations, for vision. This cascade of reinterpretation began in the early 1980's with the discovery that alpha crystallins are homologs of the small heat shock proteins of Drosophila. One of the two alpha crystallin genes continues to produce a heat shock protein (Piatigorsky et al., 1994), while the other has become more specialized for lens functions, although both also continue to act as molecular chaperones. The beta/gamma crystallins (Piatigorsky and Wistow, 1991) are more distantly related to microbial dormancy proteins, also inducible by osmotic shock and other stresses.
But the second group of «taxon specific» crystallins shows far more diversity in their multiple routes of exaptation from previous functions (often still retained) as enzymes. For example, delta crystallin of chickens is arginino-succinate lyase; epsilon crystallin of ducks is identical with the metabolic enzyme lactate dehydrogenase; tau crystallin of turtles is alpha-enolase; and mu crystallin, found in many marsupials, is ornithine cyclodeaminase (Piatigorsky, 1993b). In a proof of multiple recruitability in independent events across great phyletic distances, the eta crystallin that constitutes more than 25 percent of soluble proteins in the lens of elephant shrews is the enzyme cytoplasmic aldehyde dehydrogenase (Piatigorsky and Wistow, 1991), whereas the omega crystallin of octopuses has also been exapted from aldehyde dehydrogenase in a separate cephalopod event (Piatigorsky et al., 1994). The theme of lens proteins as exapted enzymes then extends to further phyletic diversity, for the major lens component of squid S crystallin, is related to the detoxification enzyme glutathione S-transferase.
Piatigorsky (1993a, p. 284) summarizes the dominant role of exaptation for the origin and status of lens crystallins: “A number of the crystallins have been shown to be expressed outside of the lens and to possess its original nonrefractive activity. Indeed, a hallmark of an enzyme-crystallin is that it is expressed at high concentration in the refractive lens and at a much lower concentration in other tissues, where it has at least one other non-refractive role.”
Since examples of exaptation always raise the structuralist theme of preconditions for recruitment, we must ask what common properties of proteins and enzymes in this large array of highly disparate sources prompts or facilitates cooptation as lens crystallins. Some evident requirements — with transparency as the most obvious property — probably represent merely incidental and nonadaptive consequences of molecular structures evolved for other reasons, in the same evident sense that natural selection did not make our blood [Page 1244] red or our bones white for any directly adaptive reason rooted in the colors themselves. But although transparency surely stands as a primary prerequisite, many other enzymes and proteins share this necessary property, but have never been recruited as lens crystallins — so more specific preconditions must be sought (Wistow, 1993; Piatigorsky, 1993a and b). Piatigorsky (1993a, p. 285) lists “high solubility in water to achieve the high concentrations necessary to attain the appropriate refractive index and thermodynamic stability, since loss of cell nuclei in the fiber cells prevents turnover in this region of the lens.”
Even more specifically, Wistow (1993, pp. 303-304) notes that all cellular lenses require unusually elongated cells as building blocks — a common property or potential, as Wistow argues, of the original utilities from which lens crystallins have been exapted: “As the lens evolved, the necessary refractive power must have been achieved by recruiting genes that are active under the prevailing conditions of cell elongation and whose protein products fit the broad requirements of their new role. Osmotic stress proteins, cytoskeleton chaperones and easily inducible detoxification enzymes would have been good candidates. Such an origin could have engendered underlying similarities in gene expression for groups of crystallins.”
But the case of crystallins owes its emerging status as a “classic” of exaptation largely to the strong evidence gathered for a range of structural prerequisites and preconditions that can facilitate such functional shifts. Ar
nold (1994) has proposed a set of subcategories for sources and styles of exaptation (see also Gould and Vrba, 1982), and I shall devote the final section of this chapter (pp. 1277–1294) to the further development of such taxonomy and to exploring its implications for macroevolutionary patterns and possibilities. The subject has assumed some urgency in studies of molecular evolution because the crucially important mechanism of gene duplication has frequently been overextended and interpreted as virtually the only possible basis for exaptation — when a gene with an important function duplicates (Ohno, 1970, for the classic statement), thereby “freeing” one copy for cooptation to a different utility. Exaptation does occur by duplication in the evolution of some lens crystallins, but other exapted crystallins are products of a single gene that continues to make the critical enzyme of its presumably original function — a process that Piatigorsky and Wistow (1989) called “gene sharing,” and that Darwin explicitly recognized in citing organs with two distinct functions as good candidates for quirky functional shift (see p. 1223).
For example, the duck genome includes only a single gene to code for both the exapted lens crystallin and the original enzyme in at least two cases: epsilon crystallin (lactate dehydrogenase B) and tau crystallin (alpha-enolase). In some cases, the lens crystallins even retain their enzymatic activity within the eye. The zeta crystallin of several hystricomorph rodents is quinone oxido-reductase, and may protect the eye against oxidation “or even filter UV radiation” (Wistow, 1993, p. 301). The amount of epsilon crystallin in many birds, produced by the same single-copy gene that codes for the enzyme lactate dehydrogenase B, also correlates well with exposure to light, and may [Page 1245] provide enzymatic protection as well as visual refraction (Wistow, 1993, p. 301).
As an example of exaptation associated with duplication, two genes produce delta crystallin in chickens and ducks, with the deltai gene specialized for lens expression, and the delta2 gene producing the same enzyme, arginino-succinate lyase, in non-lens tissue, but also generating some lens crystallin as well. Interestingly, both genes are equally active in the duck lens, which thus includes ASL enzyme activity (through the enhanced action of the delta2 gene) at a 1500 fold higher level than in chicken lenses (for no understood function as yet).
A presumably much older duplication occurred in the alpha crystallins present in most vertebrates, with the alphaA and alphaB genes now residing on different chromosomes. The alphaA gene has specialized for production of its lens crystallin, but maintains some activity in other organs of some species. However, the alphaB gene has retained more of its original function in generating a heat shock protein, while also coding for lens crystallin.
Interestingly, and beginning to unite crucial themes of the last two chapters, the alphaA crystallin gene of chickens is regulated by at least 5 control sites (Cvekl et al., 1994). Sites C and E bind Pax-6 (the famously homologous “master regulator” of eye development in squids, arthropods, and vertebrates — see pp. 1123–1132) in the lens to stimulate alphaA crystallin promotor activity, thus controlling high expression of this gene in the lens and repression in fibroblasts (Cvekl et al., 1994, p. 7363). These authors also report that Pax-6 binds to the lens-specific regulatory enhancer of the delta! crystallin gene of chickens, and to the lens-specific regulatory sequence of the zeta crystallin gene of guinea pigs.
Finally, some lens crystallin genes undergo more extensive duplication, usually followed by a further specialization of some copies for lens functions, as expected. “For the beta and gamma crystallins, multiple gene duplications have led to gene families with six or more members that seem to be specialized for lens” (Piatigorsky and Wistow, 1991, p. 1079). In the most extensive example of duplication, the squid genome includes at least 10 S-crystallin genes, all derived from the gene that produces the glutathione S-transferase (GST) enzyme. These S-crystallin genes are expressed only in the lens and cornea and now lack enzymatic function, with one exception of “very slight GST activity” in a single S-crystallin (Piatigorsky et al., 1994, p. 243): “The S-crystallin genes encoding the inactive enzyme derivatives have acquired an additional exon which probably contributes to the loss of enzyme activity of the crystallin.”
To close this long section, and these details of a developing classic, with a lovely corroborative tale in the venerable tradition of natural history, the squid Euprymna scolopes collects phosphorescent bacterial symbionts in a “light organ” located in the center of its mantle cavity. “The squid uses the light emitted by the symbiotic bacteria in its behavior, presumably in anti-predatory displays and/or intraspecific communication” (Montgomery and McFall-Ngai, 1992, p. 21000). The light organ also includes a lens formed as [Page 1246] a thick pad of transparent tissue, and built by the squid as a derivative of muscle from its hindgut. “The tissue functions as a convex lens to refract the light from the localized bacterial source over the ventral surface of the squid” (Montgomery and McFall-Ngai, 1992, p. 21000). (Incidentally, I shall never forget the kindness of these authors, or the eerie fascination of this system in these remarkable animals, when I had the privilege of visiting their lab in the early 1990's.)
The ocular lens of squid is epidermally derived and used for sight. This second and entirely different kind of lens, both in development and function, is built from muscle tissue and operates to enhance and refract the light generated by symbiotic bacteria! Yet the lens crystallin of the muscle-derived light enhancer, called L-crystallin by Montgomery and McFall-Ngai, is apparently exapted from an ALDH-like enzyme, as is the eta crystallin of the ocular lens of elephant shrews and the omega crystallin of the ocular lens of octopuses, both discussed previously. Montgomery and McFall-Ngai (1992, p. 21003) argue that enzymatic activity of ALDH may be preserved for protection against peroxidative damage. They end their paper with both an observation and a challenge: “Possibly, ALDH was first recruited for such a purpose, and then secondarily converted in some species to serve a largely structural role. However, as is the case with all other enzyme/crystallins discovered, why ALDH was selected [I would say exapted] as a structural protein is unknown.”
THE COMPLETE VERSION, REPLETE WITH SPANDRELS: EXAPTATION AND THE TERMINOLOGY OF NONADAPTIVE ORIGIN
The more radical category of exapted features with truly
nonadaptive origins as structural constraints
Throughout the previous section, I emphasized how the theme of quirky functional shift, and the resulting discordance between reasons for historical origin and the adaptive basis of current utility, introduced an important structuralist component into the otherwise functionalist logic of Darwinian theory. In particular, the developmental prerequisites and structural potentials of any ancestral state — and not only the adaptive pressures emanating from present environments — must be factored in as both limits and facilitators for evolutionary change, thereby acting as constraints (in both positive and negative senses) upon phylogenetic pathways.
Nonetheless, as also emphasized throughout (and in the subsection's title of “the restricted Darwinian version”), the basic concept of exaptation remains consistent with orthodox Darwinism (while expanding its purview and adding some structural clarification and sophistication) for an obvious reason: the principle of quirky functional shift does not challenge the control of evolution by natural selection as an adaptational process. Unpredictable shift of function may establish the ground of contingency, and may imply a role for structural constraints upon phyletic pathways. But this principle does not undermine the functionalist basis of evolutionary change because features so affected [Page 1247] remain adaptive throughout: they originate for one function (presumably by natural selection), and then undergo quirky shift to a different utility.
However, the principle of functional shift, combined with Nietzsche's argument about the invalidity of inferring historical origin from current utility, implies a disarmingly simple and logical extension that does challenge the rule of Darwinian mechanics an
d functionalist control over evolutionary change. Ironically, the very simplicity of the argument has often led to its dismissal as too obvious to hold any theoretical importance — a “feeling” that I shall try to refute in this section, and whose disproof represents an important step in the central logic of this book.
The deeper challenge posed to orthodox Darwinism by the principle of functional shifting flows from the implication that, if current utility does not reveal reasons for historical origin, then these initial reasons need not be adaptational or functional at all — for features with current adaptive status may have originated for nonadaptive reasons in an ancestral form. In other words, and in the terminology of Table 11-1, when current aptations rank as exaptations rather than adaptations, their coopted source will be identifiable as an ancestral structure with either adaptive origins (for a different function) or nonadaptive origins (for no function at all). (I do accept the standard view that strongly wadaptive features hold little prospect for an evolutionary legacy because natural selection must soon eliminate them. But raoraadaptive — that is, effectively or nearly neutral — features may persist for several reasons, including the “invisibility” of true neutrality to pressures of selection, and the status of many nonaptations as automatic architectural byproducts, as in Darwin's “correlations of growth” or Gould and Lewontin's “spandrels.”)
The logical validity and evident application of this simple argument cannot be gainsaid. Indeed, several examples, mentioned inter alia in the preceding section on exaptation, fall into this category of features with current utility exapted from a nonadaptive ancestral status. The optical property of transparency, shared by all the diverse proteins and enzymes that have been exapted as lens crystallins, may represent a trivial and automatic consequence of physical and chemical structures evolved for other reasons. But this purely derivative and nonadaptive feature still stands as a gatekeeper and prerequisite for exaptation to vision. We may regard Darwin's example of non-fusion of skull sutures in mammalian neonates as a far richer and less obvious case. For we need, in this example, to unravel enough specific history of mammalian descent to know that this property arose in ancestors born from eggs, and therefore cannot be a direct adaptation, initially evolved to compress the head and permit passage through the narrow mammalian birth canal. (We should also remember that Darwin explicitly declined to call non-fusion an “adaptation” for this reason, even while he acknowledged the functional necessity for such a property in the evolution of mammalian live birth.)
The Structure of Evolutionary Theory Page 198