HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 68/2/337    most recent biolreprod.102.004531v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Rothchild, I. Search for Related Content PubMed PubMed Citation Articles by Rothchild, I. Agricola Articles by Rothchild, I.

The Yolkless Egg and the Evolution of Eutherian Viviparity

^a 2441 Kenilworth Road, Cleveland Heights, Ohio 44106

One of the principal objects of theoretical research in any department of knowledge is to find the point of view from which the subject appears in its greatest simplicity. (Willard Gibbs, 1881)

Omne vivum ex ovo. (Wm. Harvey, 1651)

	ABSTRACT

TOP ABSTRACT INTRODUCTION BACKGROUND: REPRODUCTION BACKGROUND: HOW THINGS EVOLVE PUTTING IT ALL TOGETHER NOTES TO THE READER REFERENCES

corpus luteum, developmental biology, evolution, follicular development, oocyte development, progesterone

	INTRODUCTION

TOP ABSTRACT INTRODUCTION BACKGROUND: REPRODUCTION BACKGROUND: HOW THINGS EVOLVE PUTTING IT ALL TOGETHER NOTES TO THE READER REFERENCES

 
In spite of the vast amount of information about the ovary that has accumulated during the past century, we are still almost completely ignorant regarding its most fascinating aspects. From what did it originate? What forces drove it to change its structure and function? How universal are any of the explanations for its activities? Is the production of gametes always associated with the production of hormones? How did oocytes become enclosed in follicles? There are many more and many ways of searching for answers [1], but to me, the most intriguing questions are these: How did the mammalian ovary evolve, and how was its evolution connected to that of eutherian viviparity?

Viviparity is a form of reproduction in which the embryo either begins or begins and also completes its development within or on a part of the parent's body. In the eutherians ("placental mammals"), it differs from all other forms of viviparity in at least two essential respects. First, progesterone (P) is indispensable for both its initiation and its maintenance. Second, the blastocyst's trophoblast cells have properties lacking in all other forms of vertebrate embryogenesis.

The vertebrate ovarian follicle—the basic unit of ovarian hormonal activity and the means by which the oocyte reaches maturity—in eutherians is also unique. All known vertebrate follicles have the same basic form, but the mammalian follicle has four unique characteristics: 1) in marsupials and eutherians, the mature follicle is vesicular; 2) its granulosa cells are more highly differentiated at follicular maturity than are those of other vertebrates; 3) it contains proportionately more of these cells than in other vertebrates; and 4) in eutherians, its oocyte is the smallest of all vertebrate oocytes and contains no yolk.

The common ancestor of all vertebrates almost certainly reproduced by spawning. Both in the transformation of the spawner's follicle into that of the eutherians and in the accompanying changes leading to the unique character of eutherian viviparity, two features of the vertebrate follicle stand out above all others: atresia and yolk.

Among the eutherians, the vast majority of all the gametes a female has are lost through atresia before ovulation. Among spawning vertebrates, most of the oogonia produced during a female's life are ovulated as mature oocytes. Yolk is the principal—and sometimes only—source of nutrition through which the zygotes of all nonmammalian vertebrates start embryogenesis and through which most complete it. It is still used during the early stages of embryogenesis in monotremes and marsupials. The eutherian zygote, however, depends directly on maternal nutrients from the very beginning of embryogenesis.

Since the structure of all vertebrate follicles is the same and all follicles function in a way that brings the oocyte to maturity, the fact that atresia is as widespread among the eutherians as it is rare among spawners tells us there must be a connection of fundamental importance between the ability of the oocyte and the follicle to survive and the particular process through which the animal reproduces (*see Notes to the Reader, p. 353). I think the source of the connection between the eutherian follicle and viviparity will be found in genetic changes that affected the viability of oogonia, the stability of the follicle, and the liver's ability to produce vitellogenin (VTG), the principal lipoprotein constituent of yolk. I will try to spell out the basis for this idea in as much detail as it requires and our store of information allows (**see Notes to the Reader, p. 353).

	BACKGROUND: REPRODUCTION

TOP ABSTRACT INTRODUCTION BACKGROUND: REPRODUCTION BACKGROUND: HOW THINGS EVOLVE PUTTING IT ALL TOGETHER NOTES TO THE READER REFERENCES

 
Organization of the Vertebrate Ovary and Follicle

In all known vertebrates, the follicle consists of the primary oocyte surrounded by one or more layers of epithelial cells, called granulosa cells (GCs), enclosed in a spherical envelope, called the basement membrane. The ovarian blood vessels do not penetrate this membrane until ovulation occurs—and even then not in all species. At some point during development of the follicle, a well-vascularized layer of epithelial cells derived from the ovary's interstitial tissue cells surrounds the follicle. This layer is called the theca interna, and its cells are called theca interna cells (TICs). In some, but not all, species, still another layer of smooth muscle-like connective tissue cells, called the theca externa, may surround the TICs [3–6].

The ovary is usually, but not always, paired. In birds, for example, only a single ovary is active; the other remains undeveloped. The ovaries are suspended in a mesovarium from the dorsal wall of the abdominal cavity on either side of the vertebral column in the region of the kidneys; in the embryo, this area is called the germinal ridge. Between the cystic ovary of most fishes and amphibians and the solid ovary of eutherian and marsupial mammals there is an enormous variation in size and appearance of the ovary. The ripe follicles may lie just beneath or bulging from the ovarian surface (eutherians), hang pendulously (birds and most reptiles), or lie in ramified branches of the ovarian stroma, facing the interior of the ovarian cavity (bony fishes and amphibians). Both the amount and kind of stroma also vary enormously [3, 5]. Nevertheless, the common organization of the follicle in all the vertebrates must mean that it was a specific trait of the common ancestor of all vertebrates—and probably of all chordates.

Stages of Oocyte and Follicle Growth and Development

From Oogonia to Oocytes Oogonia, the female descendants of the primordial germ cells, proliferate by mitosis in either a discontinuous or a continuous pattern. The discontinuous pattern, which is practiced by cyclostomes, elasmobranchs, birds, and mammals, results in a single, large population of oogonia, the only source of all the oocytes a female will have during her lifetime. In the continuous pattern, which is practiced by almost all bony fishes, amphibians, and most (or all?) reptiles, oogonial stem cells persist throughout the female's life and produce a batch of new oogonia during each breeding season from which the oocytes come. Each oogonium becomes an oocyte when it begins the first of the two meiotic divisions [7–12].

The Primordial Follicle The GCs arise from a population of stromal cells only when an oogonium becomes an oocyte and in response to the latter [13]. They then form a single layer of small, flat, relatively undifferentiated epithelial cells around the surface of the brand new oocyte; the basement membrane then forms around them and completes the structure of the primordial follicle. The primordial follicle is very small (diameter, 10–20 µm) and does not grow until it is transformed into a primary follicle. Once this happens, the follicle is committed to one of three fates: 1) death through atresia, 2) ovulation, or 3) death through luteolysis if the postovulatory follicle (POF) becomes a corpus luteum (CL).

Once meiosis begins, the oocyte's chromosomes proceed through the leptotene, zygotene, and pachytene to the diplotene stage of prophase, where they remain (meiotic arrest) until ovulation occurs [7, 9, 10, 13, 14].

From Primordial to Primary or Previtellogenic Follicle In response to still unknown intrinsic controls [15–17], growth of the follicle begins with the metamorphosis of the GCs into cuboidal-columnar cells with a reorganized internal structure connected to the oocyte through microvilli [18, 19]. The GCs then begin to proliferate and the oocyte and follicle to grow while the name changes to primary in marsupials and eutherians and to previtellogenic in monotremes and nonmammalian vertebrates. Follicle growth proceeds in two main stages: previtellogenic and vitellogenic in the latter and primary (or preantral) and vesicular (or antral) in the former. Previtellogenic and primary growth are more or less homologous, but vitellogenic and vesicular growth are quite different processes.

Previtellogenic and Vitellogenic Follicle Growth During previtellogenesis, as the result of GC proliferation and the increase in cytoplasmic elements in the oocyte, the follicle and oocyte grow in direct proportion to one another. During vitellogenesis, however, the follicle grows mainly through the oocyte's accumulation of yolk and other cytoplasmic constituents. The GCs continue to proliferate. However, the rate of increase in their number falls behind that of the oocyte volume, and at maturity, they form only a single layer of somewhat flattened cells between the oocyte and the basement membrane [12, 20–24].

Primary and Vesicular Follicle Growth in Mammals As in previtellogenic follicles, and for the same reasons, the oocyte and primary follicle grow in direct proportion to one another until the very first sign of follicular fluid (FF) appears and vesicular follicle growth begins [7, 15, 19, 25]. At this point, the eutherian oocyte is almost mature and virtually full size [17], while the marsupial oocyte continues to grow through yolk accumulation—although at a slower rate than before—until ovulation occurs [26, 27]. In both, the vesicular follicle grows as the result of GC proliferation [26–28] and accumulation of FF in the antrum (the cavity of the follicle). The vesicular, or antral, follicles are usually referred to as small, medium, large, and preovulatory; the largest are also called Graafian follicles.

Differences among Vertebrates In the nonmammalian vertebrates and monotremes, yolk is the main determinant of oocyte size; a few examples will illustrate its importance. The smallest nonmammalian oocyte (in a viviparous fish) has a volume more than 4-fold that of an average marsupial oocyte, and the latter's volume is 4-fold that of an average eutherian oocyte [26, 29].

The growth stages vary enormously, because they depend on the pattern of the ovulation cycle (OC) in each species. However, they bear the same general relation to each other in all. Primordial follicles do not grow; they also live the longest. Primary and previtellogenic growth may last for weeks or months, whereas vesicular follicle growth is limited to the duration of the follicular phase (maximum, 4 weeks) of a marsupial or eutherian OC [26, 30]. Vitellogenesis consists of several stages and may even extend over more than one breeding episode, as in some spawners [12, 23, 31]. In birds and reptiles, it may consist of a slow phase lasting several weeks and a rapid phase lasting no longer than the time limits set by the ovulation and oviposition of a clutch [22, 24, 32]. Growth of the vesicular follicle and at least the rapid phase of vitellogenesis are strictly dependent on the effects of the pituitary gonadotrophic hormones.

Vesicular follicles vary between 1 millimeter and several centimeters in diameter, depending on body size, but a whale's oocyte differs from a shrew's by only a few micrometers in diameter [29]. Marsupial oocytes also tend to be unrelated to body size [26]. In addition, the follicles and oocytes of birds and most oviparous reptiles tend to be proportional to body size. In viviparous nonmammalian vertebrates, little or no relation is found between body and oocyte size, with the primary determinant being the egg's yolk content. Among spawners, outside rather narrow limits, the size of the oocyte tends to be inversely related to the number of eggs that are shed [23, 33].

The Oocyte and Its Follicular Environment

The oocyte becomes fully mature only after a series of developmental changes. The first is breakdown of the germinal vesicle (GVB) (dissolution of the nuclear membrane) and resumption and completion of meiosis I; the next is progression to metaphase of meiosis II. These primarily nuclear events are followed by cytoplasmic changes that lead to the ability to be ovulated, to be fertilized, and to form a viable zygote [17, 25, 34]. In the nonmammalian vertebrates, full maturity also includes the accumulation of yolk and maternal mRNAs, proteins, and other substances on which at least the early stages of embryogenesis depend [35]. In the eutherians, however, the fully mature oocyte is devoid of yolk and almost devoid of maternal mRNAs, proteins, and associated substances [16, 35–37]. By the end of primary follicle growth, the oocyte can complete meiosis I independently of the follicle, but its ability to reach this stage—as well as the remaining ones to full maturity—can be fully expressed only in the environment of the follicle [15, 17, 25, 34, 35, 38].

The follicle is thus an essential element of the processes that allow the oocyte to become mature. Until the beginning of growth of the vesicular follicle in mammals and of vitellogenesis in nonmammalians, the steps to maturity are largely—or entirely—independent of the internal environment of the organism [17, 20, 39]. However, the latter environment is indispensable for completion of the later stages of maturation, because these depend not only on intrinsic follicular factors but also on the effects of the pituitary gonadotrophic hormones and other extrinsic factors.

Oocyte/GC Symbiosis When Nalbandov and his colleagues [40] claimed that the oocyte prevented luteinization of rabbit follicles, these findings were greeted more with skepticism than with excitement, because during the early 1970s, the oocyte was still seen only as a passive occupant of the follicle. Things have changed since then. It is easy to understand that the oocyte's ability to reach full maturity would depend crucially on its GCs [17, 19, 25, 31, 34, 35, 38, 41]. Other investigators, however, including those led by Eppig, have helped us also understand that the oocyte is not the passive recipient of GC attention but, instead, plays an equally crucial role in its own development by regulating the growth, activity, and differentiation of its GCs [16, 25, 42–50]. The detailed evidence will be found in the references included with this statement. It is unnecessary to repeat it, for its significance is clear: the oocyte and its GCs are in constant communication with each other from the time of formation of the follicle. In other words, the oocyte reaches maturity through its ability to induce the formation of GCs and to regulate their activity, but in a way that causes both the oocyte and the GCs to depend on one another for their continued existence.

The oocyte's relation to its GCs is thus a mutualistic (obligatory) symbiosis. With the addition of TICs to the follicle and the beginning of vesicular follicle growth, the oocyte's ability to reach final maturity depends on its control of two conflicting drives in the GCs: proliferation and differentiation. The follicle's development to ovulability depends on how effective the oocyte is in controlling its GCs' behavior (see also [42]); I will discuss these aspects of the follicle in connection with atresia.

A very significant difference between nonmammalian follicles and those of marsupials and eutherians is the morphologic differentiation of GCs, especially during the phase of rapid growth. In mammals, the accumulating FF separates the GCs surrounding the oocyte from those lining the antrum; these anatomical differences lead to physiologic ones, depending on how close a GC is to the oocyte (the GCs of the cumulus oophorus) or to the basement membrane or antrum (the mural GCs). The absence of yolk in the eutherian oocyte may also underlie a further difference between its GCs and those of marsupial follicles (see p. 351, Loss of the genes for VTG changed the character of the vertebrate CL). In birds (if the chicken is typical), the GCs lying directly over the germinal disc (the part of the oocyte containing the nucleus and almost all the cytoplasm) are connected by microvilli to the oocyte. Throughout the rapid phase of vitellogenesis, these GCs continue to proliferate, spreading over the surface of the expanding oocyte. They stop dividing as they do so and dedifferentiate morphologically into a single layer of flattened cells as they differentiate physiologically into P secretors [47]. Except for this instance in birds, no other example of spatial differentiation of GCs is known among the nonmammalian vertebrates, since the yolk-filled oocyte occupies the entire volume of the follicle. Specialized GCs may develop during previtellogenesis or the slow phase of vitellogenesis, such as in reptiles, but as in birds, they become dedifferentiated during the final stages of vitellogenesis into a single layer of flattened cells around the oocyte [22, 24, 32].

Vertebrate oocytes thus do not mature independently of GCs, and GCs do not form follicles independently of oocytes. In the absence of oocytes, as, for example, in Turner's syndrome or after busulfan treatment, "GCs" also do not produce steroids, or if they do, in amounts much smaller than, and in patterns totally different from those of normal ovaries [51–53].

The active role of the oocyte in its own development should not surprise us, because it is both the least and the most specialized cell in the body. After all, the follicle is only the oocyte's way of getting to where it can become a zygote. Within the borders of this symbiosis, however, each symbiont retains a small but distinct amount of autonomy, which can be affected—but never abolished—by the other. For example, the oocyte can actually be fertilized at the end of primary follicle growth, even if not very well [17]. Under a variety of in vitro conditions, GCs can produce steroids for much longer durations than normal [54, 55]. Within the confines of the intact follicle, in the intact organism in most known vertebrates, however, the symbiotic arrangement affects the oocyte and its GCs in a way that allows only a small proportion of the ovary's vast stock of gametes to express their potential for maturity by becoming zygotes or the GCs to express their potential for differentiation by becoming a CL. The vertebrate follicle is, in a word, unstable. With the possible exception of losses among oogonia, this instability is almost certainly the basis for most of the gametes lost through atresia. Although factors outside the environment of the follicle (e.g., the pituitary gonadotropins) may be considered as causes of atresia, they probably act ultimately by contributing to the follicle's basic instability.

Atresia of Gametes and Follicles

Like many biologic terms, atresia, which literally means a narrowing or closure of a body orifice, has acquired many wider meanings. I will use it here to mean female gamete losses at all stages of development.

Atresia among the Eutheria Gamete losses in the human female may be somewhat above average for all eutherian species, but they emphasize the great difference between mammalian reproduction and that of spawning vertebrates. In round figures, oogonial proliferation peaks during midgestation at a population of 6 million. During the next 4.5 months, as each oogonium enters meiosis I, the female fetus loses gametes at the rate of 1 million per month, and the baby girl is born with only approximately 2 million primordial follicles. During the next 12 years, she loses follicles at the rate of approximately 12 000 per month and enters puberty with no more than 300 000 follicles. Between puberty and climacteric, a span of approximately 38 years, she can ovulate no more than 400–500 eggs and so loses all the others at the rate of approximately 700 per month (calculated from the data of Baker [56]).

Atresia among other mammals differs from the human pattern only in minor details [57, 58] and, indeed, is not much different from that among other vertebrates in which the parental energy used for reproduction is invested in a relatively small number of zygotes. A peculiar aspect of atresia is that most of what we know about its causes comes from the study of follicles in which the smallest number of gamete losses occurs! As a step at least toward a better understanding of all its causes and its significance in vertebrate reproduction, let's examine five major transitions or nodal points in gamete development (see also [35, 58, 59]).

Nodal Points in Development of the Oocyte and Follicle The five nodal points in development of the oocyte and follicle are: 1) transition from mitosis to meiosis, 2) formation of the primordial follicle, 3) metamorphosis of the GCs, 4) initiation of growth of the oocyte and follicle, and 5) transition from primary to vesicular follicle growth in mammals or from previtellogenesis to vitellogenesis in nonmammalians. The transition of the ripe follicle to ovulability can also be thought of as a nodal point, but I will discuss this in connection with instability of the antral follicle. Each of these nodal points distinguishes between gametes and follicles that can pass to the next stage of development and those that, for the lack of whatever quality, fall into the discard pile of atresia [58].

Transition from mitosis to meiosis No one knows how this switch occurs, except that it must be intrinsically controlled [59, 60]. Mitosis is a critical event in the life of any cell. However, meiosis must bring even more hazards, because it includes fundamental changes in how the chromosomes are duplicated and segregated [14, 15, 60, 61]. For example, one-third of all losses at this transition occur during the pachytene stage [60].

Formation of the primordial follicle The first step is the formation of GCs among the neighboring somatic cells, a form of embryonic induction with the oocyte being its specific agent [13]. Even if an oocyte reaches the diplotene stage but stumbles along the way, it may not acquire adequate competence to induce GCs or to participate with them in formation of the primordial follicle. Losses could be caused by failure to form a follicle or by formation of a defective one [59].

Metamorphosis of GCs and initiation of growth of the oocyte and follicle The lack of any sign of atresia among primordial follicles, in spite of a steady drop in their number, as among the eutherians, may be because they do not undergo atresia. Their number falls almost certainly because of losses during the transition from quiescence to activity and may be traced to defects in GC metamorphosis and/or in the signals arising in both the oocyte and the GCs as the changes occur that allow the oocyte to start growing and the GCs to resume proliferation. The survival of the symbiotic arrangement that follows these changes depends on how effectively the oocyte's and the GC's intrinsic signals are coordinated. When we consider the many steps in converting GCs to proliferating cells and in the processes that permit the oocyte to grow [19, 42, 59, 62, 63], it is no wonder that this nodal point is the second most frequent site of gamete loss.

Transition to rapid growth of the follicle A major difference exists between the eutherian and marsupial mammals as a group and all the other vertebrates in what happens at this point. It will be easier, perhaps, to explain this difference if I consider the mammals first.

In contrast to the first four nodal points, in which the danger of atresia arises almost entirely from the transition itself, here the danger also includes all the phases of rapid growth that follow. Let's look at the hazards of becoming a rapidly growing vesicular follicle and then at the hazards of being one.

The basement membrane separates the GCs from their stem cells. As each cell divides, therefore, the daughter cells form a population of increasingly more differentiated GCs, which also become spatially differentiated as the antrum enlarges (see above). This process ultimately leads to ovulation and luteinization [28].

As the GCs begin to increase in number and differentiation, the interstitial cells form a layer of nascent TICs around the basement membrane, and soon after this, the intrinsic controls of the follicle's activity are supplemented and then largely replaced by extrinsic controls as the TICs and, through them, the GCs become responsive to the pituitary's gonadotrophic hormones [13, 30, 59, 62, 63]. All further development toward ovulability depends on the follicle's ability to maintain this responsiveness. Thus, if communication between the primary follicle's GCs and oocyte is defective, if GC differentiation is less than ideal, if the number of GCs is less than a critical level, induction of the TIC layer may be defective. Because all further maturation of the oocyte and of proliferation and differentiation of the GCs depends crucially on the rich blood supply of the TICs, through which they receive all their nourishment and extrinsic controls, any faltering of the TICs to join the follicle will make atresia inevitable.

Instability of the Vesicular Follicle Even if an oogonium had survived all the hazards of these transitions, it is still faced with those of being an oocyte in a vesicular follicle, for it is now developing to full maturity as its environment is changing from moment to moment. In contrast to the preantral follicle, the vesicular follicle now also responds to much of its external environment, and this is loaded with potential danger. More than any other characteristics, meiotic arrest, GC proliferation and differentiation, estrogen and P production, and the feedback arrangement between the follicle and the gonadotrophins contribute to the antral follicle's instability.

Meiotic arrest Because the oocyte is already partially mature at the end of primary follicle growth, it has at least the potential ability to complete meiosis I spontaneously and prematurely [17, 35, 64]. Should this occur, its life expectancy would be shortened, because it cannot be fertilized inside the follicle. The GCs will also die as a result, even if they show some signs of luteinization in the process [64].

Proliferation and differentiation of GCs The rapid growth of the vesicular follicle is the result mainly of FF production and a great increase in GC proliferation. The FF is itself a product of GC activity as well as a serum exudate passed through them to the antrum [30]. Both activities are stimulated by estrogens made by the GCs. With the acquisition of TICs, the GCs also acquire FSH receptors and make the oocyte maturation inhibitor (OMI) in response to FSH; estrogens and the OMI help to maintain meiotic arrest [35, 64]. However, FSH also promotes the GCs' innate capacity for P production, and large-scale P production is the ultimate expression of GC differentiation (see p. 350, The follicular GC is the larval form of luteal GC). The GCs' ability to proliferate thus falls as their state of differentiation rises [28, 62, 64, 65]. The oocyte also encourages GC proliferation and suppresses GC differentiation [46].

Obviously, then, the precarious balances between GC proliferation and differentiation in general, between the effects of FSH on estrogen production and on P production, between the ability of OMI and estrogen to maintain meiotic arrest and of P to end it, and between all these and the oocyte's effect on proliferation and differentiation of GCs can be easily upset and lead to atresia. The number of GCs in mitosis at any one time may also affect these balances, because secretory activity stops during mitosis [37, 66]. The lack of stem cells also may affect their number, because losses cannot be replaced. Both the total number of GCs and their state of differentiation [64] are critical factors in regulating ovulability and the oocyte's attainment of full maturity [17, 42]. The wonder is, on the one hand, that any vesicular follicle in any mammal ever reaches ovulation and, on the other, that all of them do in a few (see below)!

Estrogen production by the vesicular follicle More than 60 years ago, P.C. Williams [67] found that treating rats with high doses of estrogens prevented posthypophysectomy atresia. Since then, many others have confirmed Williams' finding and extended it to the current concept of estrogen as a powerful GC mitogen [62, 63, 68, 69], probably a reflection of the effect of a small amount of autonomously produced estrogens limited to the GC itself (autocrines). Autocrine estrogens are also made by the GCs in response to FSH in cooperation with growth factors and other intracellular substances; the mysteries of their production and interactions are the subject of very active current research [16, 59, 70, 71], most of which is outside the scope of this paper.

How the ovary makes large amounts of estrogen in response to both gonadotrophins, with effects almost entirely outside the ovary (hormonal estrogens), is much more well known. Its most popular explanation is the so-called "2-cell theory," the essence of which is that TICs can make androgens but cannot convert them to estrogens, whereas GCs cannot make androgens but can convert them to estrogens [72, 73]. The theory fits neatly with two sets of facts: First, FSH acts primarily on GCs and LH on TICs, and second, the amount of estrogen made by the ovary in response to both gonadotropins is much greater than the sum of the amounts made in response to either one alone [30]. Nevertheless, the theory is misleading, because both GCs and TICs can—and do—make estrogens independently of one another [74]. However, because the synergistic effect of the two gonadotropins is also indisputable, we must conclude that whatever the reasons may be for GCs and TICs to cooperate in making estrogens, an idiosyncratic deficiency of specific steroidogenic enzymes in either cell is not among them. In other words, the two cells are not obliged to make estrogens together; they just do.

TICs are not really an integral part of the follicle, but like the hermit crab's empty cockle shell's relation to the hermit crab, their addition to the follicle's immediate environment had a profound influence on what it could do. About the time this change in the follicle's economy was evolving, FSH and LH were probably evolving from the pituitary's original single gonadotrophin into separate but related protein hormones [75–77]. These two sets of changes also allowed estrogens to evolve from their autocrine state (as one of several intracellular regulators of GC mitosis) into a hormone with effects on growth and development of the body as a whole, on behavior, on the pituitary, on the female reproductive tract, and through some or all of these, on the stability of the vesicular follicle.

Hormonal estrogens, however, are not necessary for either maturation of the oocyte or ovulability of the follicle, for pure FSH in the virtual absence of LH can induce both effects without more than a barely detectable increase (or none at all) in estrogen production (e.g., [78, 79]). The importance of hormonal estrogens in vertebrate reproduction is in the coordination of their effects on the central nervous system (making the female interested in coitus) [80], on the pituitary (facilitating the surge of LH that induces ovulation) [81], and on the female reproductive tract (making it a zygote-friendly place), with the intrafollicular events leading to oocyte maturation and ovulability. This coordination increased enormously the chances for successful fertilization of the ovum and survival of the zygote. Accompanying these selective advantages was a continuation of gamete losses into the phase of rapid follicle growth itself, a phenomenon that may be unique to mammals, because it does not seem to occur in birds or reptiles or, probably, in amphibians and fishes (see below, Gamete Losses among All Vertebrates).

Aside from the negative feedback of estrogens and several nonsteroidal GC products on gonadotropin secretion, the main reason why hormonal estrogens make the follicle unstable comes from the way that GCs and TICs make them. TICs respond to LH by becoming the main source of androgen production, and androgens are the principal—or sometimes only—precursor of estrogen production. Regardless of what proportion of these androgens is converted to estrogens by TICs, by GCs, or by the combined activity of both cell groups, their moment-to-moment conversion must affect the androgen : estrogen (A:E) ratio in the GCs, and atresia is associated more with high A:E than otherwise [58]. The ability of any follicle's GCs to convert whatever androgens it is exposed to into an optimal amount of estrogens therefore becomes a measure of its ability to survive because atresia, in essence, involves only the oocyte and its GCs [58]. (TICs, unlike GCs, remain in contact with their stem cells [the interstitial cells] and can survive even when the rest of the follicle dies.) If the rate of conversion to estrogens falls below a critical level or the rate of androgen production rises above one, atresia will become almost inevitable [64, 82].

Production of P by GCs A capacity for autonomous P production by the mammalian CL, and perhaps even of all vertebrate CL [83–85], may be an attribute of follicular GCs in a nascent state, but it may also be dependent on FSH for its expression. This small amount of P may fit into the follicle's economy only as a precursor for estrogens made in the GCs, but the potential capacity for large-scale P production may lead to atresia. Only proliferating GCs will maintain meiotic arrest and the oocyte's progression to full maturity [17, 42, 64], but GCs will luteinize spontaneously if removed from contact with the oocyte [44, 46]. Normal luteinization during the periovulatory period may even depend on P itself [86]. If, for example, the GCs escape from the influence of the oocyte and begin to luteinize prematurely within the intact follicle, meiotic arrest will be broken, and atresia will become inevitable [64]. It is conceivable, in fact, that one of the connections between androgens and atresia may come from their stimulation of P production by GCs [64, 82].

Negative-feedback control of gonadotropin secretion During the normal OC, the vesicular follicle, in response to both FSH and LH, increases its secretion of hormonal estrogens in direct proportion to its increase in size [63] until just before ovulation, when estrogen and androgen production switch to P production [58, 87]. Less consistent changes in production of the nonsteroidal hormones inhibin, activin, and folistatin by the follicle [59, 71] accompany the increase in hormonal estrogens. At their peak, these estrogens facilitate the sudden surge of LH and FSH secretion that leads to ovulation [81]. The ability of the follicle to become ovulable, however, depends on GC proliferation (probably to an optimal number), and in turn, this depends on autocrine estrogens, which in turn depend on FSH. The complex of hormonal estrogens and nonsteroidal hormones depresses FSH secretion and, to a lesser extent, LH secretion. Only those follicles, therefore, that can remain responsive to the steadily falling level of circulating FSH will continue growing to ovulability. Most of the total cohort of vesicular follicles will be unable to do so, and the resulting imbalance between oocyte and GCs will drive them into atresia.

These losses, although only a tiny proportion of all gamete losses among mammals, probably are only the fine-tuning of a process that ends with the production of a very small number of viable zygotes (see below).

Gamete Losses among All Vertebrates

I have restricted the examination of atresia until now almost entirely to its manifestations in the eutherians. However, among all vertebrate species in which the amount of energy expended in reproduction is distributed among only a very small number of zygotes, a high incidence of gamete losses always precedes ovulation. We can only speculate about how these losses, in most cases, are connected to the way in which a particular species reproduces.

Take, for example, the process usually referred to as selection of the dominant follicle. Among the eutherians, this takes place during the rapid growth of a cohort of vesicular follicles, as just described, with 1 oocyte left to be ovulated in primates and approximately 5 in dogs, 10 in rats, etc. [2]. In birds and reptiles, however, in which the ovulated eggs form a clutch (homologous to a eutherian litter), selection through atresia is completed before, not during, the phase of rapid vitellogenesis that precedes ovulation [22, 32, 57]. When atresia does occur among these follicles, it is only after ovulation of the entire clutch or in response to random environmental events, such as poor nutrition, stress, disease, etc., regardless of the pattern of oviposition [22].

Another especially pertinent example occurs among mammals. In the elephant shrew, an insectivore [88], and the plains viscacha, a rodent [89], among eutherians and in the American opossum and the dunnart among marsupials [29], almost all of the cohort of rapidly growing vesicular follicles—which may number 50–100—are ovulated, and their eggs are fertilized! Only those blastocysts that can find room in the uterus or neonates that can find a free teat have a chance to survive, and these are never more than about a dozen [90]. Thus, selection in these species occurs neither before nor during rapid growth but only after fertilization!

Among spawners, selection is preovulatory and is similar in principle, but with one very significant difference. Losses at each of the nodal points in the oocyte's development are minimal, so the "cohort" or "clutch" of ovulable follicles comes from almost all the primordial oocytes with which the animal started an OC [12, 21, 23, 57].

The distribution of degrees of instability among a group of follicles could be the basis of the selection process—that is, a reflection of the number of the group's least unstable follicles, genetically determined for each species. In all the departures from the spawning pattern, and in association with the change in the oocyte/GC relationship that was the basis of the new form of reproduction, the hazards in the oocyte's transition from one developmental stage to the next were undoubtedly intensified, but where most gamete losses occurred seems to have depended more on which nodal point proved to be the most susceptible to breakdown rather than on any taxonomic relation between species or rule applicable to all taxa.

The difference between the shrew and most mammals illustrates one example of this guess. Another may be the transition from mitosis to meiosis, which in mammals is the site of greatest loss, but is this true of other vertebrates? Still another is the difference in the incidence of atresia between previtellogenic follicles in birds and reptiles and their homologue, primary follicles in mammals. More than 90% of these follicles in birds and reptiles are lost during their growth, while in mammals most losses occur in the transition to and from rather than during primary follicle growth [58, 59, 64, 87, 91].

What is more interesting than all the attempts to understand the specific causes of atresia of vesicular follicles in all their biochemical, physiologic, and genetic variety is how such enormous losses of female gametes became such a dominant theme in most forms of vertebrate reproduction. The next step in our search for the origins of the eutherian follicle and viviparity, of which this question is an essential element, is to examine how parental energy is invested in the production of offspring.

How Parental Energy Is Distributed among Offspring

Of all the energy living things expend in maintaining themselves, the portion invested in reproduction varies enormously in amount and character, for it can be affected by even the smallest details of an organism's existence. This is a subject of absorbing interest, complexity, and importance (e.g., [92, 93]), but my concern is limited exclusively to how the amount invested in reproduction (regardless of its proportion of the total energy expenditure) becomes distributed among an organism's potential offspring. This is a critical factor in the origin of any system of reproduction. In particular, I will focus on a problem that does not seem to have interested most students of the subject: Faced with the compromise between size and number of offspring, how did those vertebrates that opted for a small number of large offspring (K-selection to ecologists) reduce the number of oocytes available for fertilization?

I will use parental energy invested per zygote (PEI/Z) to describe how this distribution is related to atresia and to evolution of the mammalian ovary and form of viviparity. The term zygote is used rather than gamete, because the postfertilization part of reproduction can be as important as what went before. In addition, the term parental is used rather than maternal because of the importance of the male in some forms.

Spawning as a System of Low PEI/Z In typical spawning vertebrates (most bony fishes and most anurans), the majority of the oogonia a female produces during a breeding episode become oocytes, and almost all of these reach maturity, are ovulated and fertilized, and become viable zygotes [21, 94]. Nearly all the energy a mating pair expends in reproduction is invested in producing sperm, producing and yolking eggs, and migrating to the breeding and/or nesting site, where the eggs and sperm are shed in a freshwater or marine environment. The investment in most cases ends at this point, but in a few species, a relatively small amount of energy may also be invested in parental care or other variants of the spawning pattern [21, 94–96].

Since the number of zygotes, especially in broadcast spawners, typically runs into the thousands or even millions, each one gets only a very small part of its parents' total investment. The success of spawning rests on the likelihood that at least one pair of breeding adults will be produced by each pair of parents. Cod fishing off the Grand Banks—before human greed turned it into a disaster area—tells us that it is a very good way to reproduce if you are a fish [97].

High PEI/Z Alternatives to Spawning In high PEI/Z reproduction, all of a mating pair's reproductively applied energy is invested in very few zygotes. The male's production of sperm is not different from that of spawners, but in some species, the male's participation in postfertilization reproduction is as important as the female's participation.

Although one cannot make an absolute distinction between low and high PEI/Z or among the many variants of either, it is convenient to fit systems of high PEI/Z into three very broad categories: 1) cleidoism, 2) viviparity, and 3) brooding (i.e., parental care).

Cleidoism Joseph Needham [98] coined the term cleidoic egg (from the Greek cleid, meaning key and implying locked or enclosed) to describe an egg containing all the ingredients necessary for completion of embryogenesis outside the mother's body, enclosed in a protective shell (e.g., the hen's egg). He originally described the shell as being firm and permeable to gases but not to water; the ovum itself was always megalecithal. Cleidoism applies the principle of the cleidoic egg to eggs with a protective shell that may also be permeable to water, as in some reptilian species; whether the water is included among the egg's ingredients or is available from its environment is only of secondary importance. In spawners, the egg is mio- or mediolecithal, and the shell is gelatinous [21, 94, 95]. The spawners' hatchling is usually a free-swimming, independently feeding larva; the cleidoic egg hatchling may be either precocial (i.e., advanced enough in development to feed itself, such as chickens or sea turtles) or altricial (i.e., almost fully developed but dependent on parental care for full development, such as robins or gulls). Although the cleidoic egg is clearly a change that made terrestrial life for vertebrates possible, it also occurs in somewhat different form in elasmobranchs (sharks, rays, and skates) [99] and in principle but totally different form in parasitic wasps. Parental energy is thus invested mainly in producing a few heavily yolked eggs and mating behavior, including migration to the breeding and/or nesting site. It may end with oviposition (e.g., elasmobranchs and most oviparous reptiles). It may also include brooding eggs until they hatch and feeding of altricial young (e.g., birds). Cleidoism is typical of all birds and oviparous reptiles. It does not occur among bony fishes, amphibia, or mammals.

Viviparity Oviparity, ovoviviparity, and viviparity are terms used to describe offspring at the moment of separation from a parent, but even the worthwhile effort by Wourms [96] to distinguish between different forms of oviparity by use of the terms ovuloparity, zygoparity, and embryoparity is of limited value. Isn't it more interesting to know how a pair of gametes became an offspring rather than whether it left its parent inside a shell or wrapped only in its own skin? Cleidoism includes oviparity in its meaning, but it tells us much more about how an organism reproduces. Viviparity has so many meanings that biologists do not agree on precisely how to define it [100–103], probably because its very variety obscures the fact that it is only the flip side of cleidoism. It differs from cleidoism mainly in the fact that an embryo develops inside or on a parent's body instead of inside a shell and, in addition to or in place of yolk, it feeds on parental nutrients that reach it through a variety of delivery systems. Aside from several fascinating exceptions, like the rockfish Sebastes or the cichlid Tilapia [96], the essence of viviparity is that each zygote receives a very high proportion of the parental energy invested in reproduction, because the limited amount of available space determines how many can be accommodated.

It is of only minor importance, therefore, whether the conceptus develops in a shark's oviduct and feeds on yolk, its siblings, or through trophonemata [99, 100, 104, 105]; inside its shell in a lizard's oviduct [103, 106, 107]; within a bony fish's ovarian follicle, where it started life as an oocyte [108]; in the marsupium of a male sea horse [109]; in the stomach of a frog [95]; or in a rabbit's uterus—all are versions of viviparity (see also [110]). We can better appreciate the unique quality of eutherian viviparity in the context of this broader definition than in that of the classical definition (i.e., "bearing of live young").

Brooding and other forms of parental care Parental care refers to behavior of either or both parents that protects and, in many cases, also feeds the young during their development. It may be expressed only as nest building, which can also be part of mating behavior, or also as brooding, which can vary in extent and complexity from simple guarding (as in several families of bony fishes [33]) to lactation (as in mammals). Brooding may also be associated with spawning, cleidoism, and viviparity. In some rockfish species, it is even associated with both spawning and viviparity, for here an entire brood, numbering in the thousands, may be spawned as free-living larvae [96]. Viviparity in some reptiles is little more than the brooding of an embryo in the mother's oviduct [103, 107, 111]. Mouth brooding, even of fairly large clutches, is common among cichlids and a few other families of bony fishes [33, 94, 112], and skin pits or pouches in either sex are used as brooding sites in several species of frogs and toads [95, 113]. Parental care, however, even in the rockfishes may add little to the low PEI/Z of typical spawners when compared to the cost of yolking thousands of eggs, but in birds and, most of all, in mammals, brooding is a very large item in the total cost of reproduction. In the marsupials and monotremes, as lactation, it is the main site of the investment of reproductive energy.

The Unique Character of Eutherian Viviparity

Diversity of Its Expression Eutherian pregnancy occurs in so many patterns that it does not seem possible for a single thread of uniformity to connect them. For example, it can last as little as 16 days (hamster) or as long as 2 years (elephant). Litters can consist of one (human) or a dozen offspring (pig). The interval from fertilization to implantation is usually approximately 4 days (rat and monkey) or up to 2 weeks (pig), but it can also be 10 months (European badger). (The long interval is the result of delayed implantation [i.e., embryonic diapause], a condition in which the zygote, on reaching the blastocyst stage, remains free in the uterine lumen; it is usually associated with a lack of estrogen and a long period of CL quiescence [2].) Decidualization of the endometrium is an essential part of implantation (rat and human) but is absent in ungulates; where it does occur, it may involve only the epithelium (monkey) or only the stroma (rat). The blastocyst, when it implants, may be only slightly larger than it was as a zygote (rat and monkey) or large enough to almost fill the lumen of a uterine horn (sheep and horse). It can implant anywhere on the endometrial surface (human), only on the antimesometrium (rat), or only on the mesometrium (some bats). The placenta may be diffuse (sheep), discoid (human), or bracelet-shaped (dog). Its trophoblast can be in direct contact with maternal blood (human) or separated from it by one to three layers of maternal tissue (sheep, pig, and dog). The ratio of zygotes to implanting blastocysts can range from 1:1 (human) to 100:1 (elephant shrew). The uterus can be two completely separate tubes opening through individual cervices into the vagina (rat) or a single body, the lumen of which is connected to the vagina through a single cervix (primates). Progesterone secreted by the CL may be necessary throughout pregnancy (pig) or only during the first one-sixth of pregnancy (human). The placenta may secrete only trivial amounts of P (rat) or enough to maintain the last five-sixths of pregnancy in the absence of the CL (human). Pregnancy can last as long as the luteal phase of the OC (dog) or 10-fold longer or more (human and elephant). Newborns can be altricial (mouse and dog) or precocial (sheep and guinea pig) [2, 29, 90, 105, 114].

Six Common Characteristics of Eutherian Viviparity As a group, six common characteristics define the uniqueness of eutherian viviparity. 1) From the blastocyst stage to term, all of the conceptus' nutrition, respiration, and disposal of metabolic waste occur chiefly through a chorioallantoic placenta supplemented to a varying but minor extent by a choriovitelline (i.e., yolk sac) placenta, depending on the species. 2) The eutherian trophoblast is unique in its derivation from the blastomeres of the morula, its duration and growth, and its function as a medium of exchange and protection of the embryo against immune rejection by the mother. 3) The conceptus develops at a higher and more constant temperature than does that of any other vertebrate. 4) P secreted by the CL is essential for the establishment of pregnancy, and P secreted by the CL and/or the placenta is essential for its maintenance to term. 5) Except for disease, the uterus is the only site of pregnancy. 6) In the most altricial eutherian neonate, the major organ systems are much more advanced in development than they are in either the least altricial monotreme hatchling or marsupial neonate.

A Key to Eutherian Viviparity? It is by now a truism that viviparity has appeared in many forms among many members of every vertebrate class, except birds, and that virtually every case of similarity is probably more an instance of convergence or parallelism than of an evolutionary trend [107, 110, 115]. Even among reptiles, evidence for a progression from oviparity to viviparity is questionable [103]. The eutherian style is only another such instance, distinct in its own way even in the quality of those characteristics it shares with other viviparous vertebrates, such as the association with P secretion, the uterus as the site of gestation, or trophoblast (which it shares with marsupials). Nevertheless, a definite trace of evolution from reptilian to eutherian viviparity is not only likely but is the key to its uniqueness.

Two great leaps separate the eutherian from a reptilian form of viviparity The first great leap was from a reptilian to a marsupial-like form, and the second great leap was from the latter to the eutherian form. The uniqueness of the eutherian form of viviparity is in the extent to which it still resembles that of the reptile but differs from that of the marsupial.

In all known marsupial species, pregnancy, from implantation to parturition, lasts no longer than between 4 and 10 days, and in none does its maintenance depend on P [26, 116–118]. These differences from the eutherian pattern stem from the nature of the marsupial CL and uterus [83].

The relative uniformity of pregnancy length, the similarity between the CL of the OC and of pregnancy in life span and pattern of P secretion, and the lack of effect of hysterectomy in either pregnant or nonpregnant animals or of hypophysectomy on CL activity make it safe to assume that with two exceptions, the marsupial CL secretes P autonomously, is relatively insensitive to extrinsic controls, and has a life span limited roughly to approximately 2–4 weeks [26, 83, 118]. One exception is a form of embryonic diapause in kangaroos and wallabies [26] that occurs with seasonal anestrus or lactation. The blastocyst lies free in the uterine lumen for weeks or months, but resumption of the OC, or removal of the pituitary (anestrus) or of the nursing young (in lactation), activates the CL and induces implantation [26]. The pituitary thus seems to maintain the CL in a viable but dormant state. The other exception occurs in bandicoots; their CL do not regress at parturition, as in other known marsupials, but instead secrete P for several weeks afterward, especially during lactation. In fact, parturition occurs at the peak of P secretion, 11 days after fertilization [26, 119]. These exceptions, however, do not affect the basic differences between marsupial and either reptilian or eutherian CL and pregnancy. In five species, ovariectomy or luteectomy after Days 8–10, as the unilaminar blastocyst is transformed into the trilaminar blastocyst, does not prevent pregnancy from proceeding to term [118, p. 706]—that is, presumably in the absence of P. Luteectomy after implantation may even delay parturition by a day [118].

Treatment with P also does not prolong pregnancy [118]. In both marsupials and monotremes, the interval from fertilization to parturition (discounting diapause, if it occurs) or to hatching, respectively, is approximately 2–4 weeks—only long enough to produce a tiny (<1 g), extremely altricial neonate [26, 116, 120]. The inability of the marsupial uterus to retain an embryo beyond a brief period of development and of its CL to secrete P for longer than approximately 2–4 weeks differs strikingly from reproduction in modern reptiles and, probably, also in their Permian and Triassic ancestors. Even in oviparous reptiles, the CL's life span and egg retention in the oviduct tend to coincide, and in viviparous reptiles, the duration of gestation and of P secretion are also similar [110, 111, 121, 122], even though P is probably unnecessary for its maintenance [110]. The first two pivotal events, therefore, in the switch from reptilian to eutherian reproduction were the loss of the uterus' ability to retain the embryo, except only briefly, and the loss of the CL's ability to secrete P beyond approximately 2–4 weeks. Such a drastically truncated viviparous reproduction of the first mammals succeeded only because it happened in endothermic brooders, the skin of which responded to prolactin (PRL) by producing nutrients on which an embryo could feed and develop [123, 124].

In the switch from marsupial to eutherian viviparity, the pivotal events resemble, but are not the exact reverse of, the previous ones: The uterus regained, or now definitively acquired, responsiveness to P long enough to retain an embryo until it had completed most of its development, with only the finishing touches being left for lactation. At the same time, the CL recovered its ability to respond to extrinsic stimuli, including a greater variety than before, and thus the ability to secrete P well beyond the short period of autonomous production.

Key to eutherian viviparity The eutherians seem to have returned to a reptilian mode of viviparity through a marsupial detour, but the resemblance is one of parallelism, not of homology. Behind it lie fundamental changes from reptilian reproduction, the most critical of which are in the nature of the CL and of placentation. Let's look first at placentation.

Placentation in eutherian viviparity Among viviparous fishes and amphibia, placentation of any kind is either absent if the site of gestation is in the oviduct, as in the toad Nectophrynoids occidentalis [125] and among the caecillians [21, 39, 95], or is rare, usually being only a minor source of nutrients, and the organ systems used are, in most cases, not at all homologous with the eutherian placenta [21, 39, 95, 105, 108, 113, 126]. In some elasmobranchs, placentation occurs only during the later stages of gestation. However, it can be as complex as in any eutherian arrangement, although the organs used as trophotaenia and trophonemata bear no homology to a eutherian placenta [99, 100, 104]. Among reptiles, only the squamates (snakes and lizards) contain viviparous species. They also use the oviduct as the only site of gestation, and their placentas are somewhat homologous to those of the eutherians. Yolk, however, continues to be the principal form of embryonic nutrition for most species. Placentation is not usually a part of viviparity, and in the vast majority of cases, it serves primarily for respiration, excretion, and as a supplement to yolk [103, 107, 111]. Only among the Scincidae (skinks), a family of lizards, does placentation occur in any form at all resembling that of the eutherians [127, 128]. For example, Chalcides and Mabuya have very small oocytes, and in Chalcides, placentation accounts for 99% of the embryo's nutrition [128]. However, these oocytes are, respectively, 200- and approximately 800-fold greater in volume than the eutherian oocytes; the difference is due to yolk, on which the skink embryo's earliest stages of development depend [128]. Both skinks form chorioallantoic placentas with an intimate connection to maternal tissue. In Chalcides, however, it is the last of a series of five placentas that form during development, and it does not appear until the embryo has reached the limb bud stage. It also has no trophoblast. In a sense, skink placentation is as unique among reptiles as eutherian placentation is among all vertebrates.

The marsupial embryo does not implant until it is past the pharyngula stage, at 20 somites or more; it uses a yolk sac placenta and, in bandicoots, a chorioallantoic one as well, although mainly for excretion [26]. In eutherians, the blastocyst implants even before the primary germ layers have formed (a unique eutherian trait), and the chorioallantoic placenta is the principal one used, with the yolk sac placenta being used as a supplementary one in some species. The eutherian trophoblast also differs from the marsupial trophoblast in how it is formed, intimacy of contact with the endometrium, growth and differentiation, duration of function, and amount and kind of hormonal activity [26, 29, 105, 118, 129, 130]. It probably protects the fetus against immune rejection, but whether this is also true of marsupials is still unknown, except that the very short period of placentation is not an expression of immunity-based rejection [26]. Eutherian trophoblast and placentation, in short, may be improvisations on the marsupial form in a way that made prolonged intrauterine gestation possible (see below).

The Unique Quality of the Eutherian CL

How the CL Is Formed Four specific, coordinated changes in the eutherian follicle cause it to become a CL. The first is luteinization of the GCs (and, to a lesser extent, of the TICs), which begins just before ovulation and is completed several hours to days later, depending on the species [83]. Then, beginning with or just after ovulation, the basement membrane disappears, the gap junctions [18, 19] between the GCs close [131], and endothelial cells invade the mass of luteinizing GCs (angiogenesis) and quickly invest each cell in a dense network of capillaries [83, 132–134]. To my knowledge, these four steps are typical of postovulatory CL formation in all eutherian species in which it has been studied to date.

Luteinization is the second metamorphosis in the life of a GC. Mainly through an increase in cytoplasm, each cell becomes several times larger than before, and its more or less spherical shape changes to a polyhedron with distinct borders. As the GCs hypertrophy and become vascularized, profound changes in the cytoskeleton, endoplasmic reticulum, mitochondria, Golgi, nucleoli, and nucleus complete the metamorphosis. Luteinized GCs are the primary functional element of the mature CL and the main reason why it is larger than the follicle it came from, although TICs, blood vessels, and connective tissue also contribute significantly to its enlargement. The eutherian CL goes through a typical pattern of growth and regression; P secretion [83] and angiogenesis of the GCs [135–137] keep pace with this pattern. At the peak of its activity, the CL's blood circulation is among the highest of any organ in the body [132, 133].

CL Are Not Peculiar to Eutheria or to POFs Postovulatory follicles, containing enough luteinized GCs to look like CL, occur among several species of elasmobranchs, a few species of bony fishes and amphibia, all reptiles [85, 115, 138], and according to some scientists, even in birds [85]. Atretic follicles with varying stages of GC and/or TIC luteinization may also look enough like CL to be labeled as CL atretica [58, 87, 139]. The resemblance to the eutherian CL, especially among those formed from POFs, in the characteristics of luteinization itself, in GCs as the predominant cell type, in secretion of P or of ^4–53-one pregnenes related to P, and in a tendency to be associated with some form of viviparity [85, 125, 138] blurs the distinction between eutherian and noneutherian CL. Nevertheless, the differences are important.

Examples of crucial differences between eutherian and other CL In no case, even including marsupials, do the CL of other vertebrates form exactly as they do in eutherians. Marsupials ovulate naked oocytes; the cumulus is not only included in the CL, it does not form a corona and does not mucify [140]. In almost all known eutherians, the oocyte is ovulated while still embedded in its mucified cumulus [42], and the CL is formed only from mural GCs [139], which are physiologically different from cumulus GCs [17, 25, 28, 46]. A few insectivores ovulate naked oocytes, but their follicles become filled with GCs rather than FF [139], and no differentiation into cumulus and mural GCs is possible, although they all acquire the characteristics of both cell types. The CL of nonmammalians is formed from GCs that are not spatially differentiated and, in general, less physiologically differentiated at ovulation than eutherian GCs [22, 32, 35].

Although luteinization of nonmammalian GCs is more or less like that of eutherian GCs, they do not hypertrophy as much, and they tend to form a luteal cell mass (LCM) surrounded by a distinct TIC and, in some species, theca externa layers [125, 138]. In most species, the LCM remains avascular, and angiogenesis, when it occurs, is quite different from the eutherian pattern. For example, in several reptilian species, finger-like wedges of thecal tissue with intact capillaries penetrate the LCM, isolating and vascularizing it into units [138] rather than investing each GC with a capillary network of its own as in eutherians. The endothelial investiture of individual cells can significantly stimulate their potential for development [141]. More important still is that vascularization occurs not as the LCM is forming but well after it has been established, and the intrusion of blood vessels looks more like a response to a moribund tissue than an accompaniment of healthy growth. This fits with the persistence of the basement membrane, at least during the early stages of CL development, and its fragmentation rather than complete dissolution when the thecal capillaries invade the LCM [125, 138]. The nonmammalian CL also often tends to be smaller than its follicle, especially if it is megalecithal, but also because the follicle has fewer GCs than the eutherian follicle when it ovulates.

Aside from a brief secretion of P or of related pregnenes around the time of ovulation and an equally brief and only partial luteinization of the GCs after ovulation, the POFs of birds and of broadcast-spawning fish and amphibians do not resemble the CL of eutherians or even of other vertebrates. Even their contribution to reproduction is different from that of the CL, because they are physiologically related to the comparatively short interval between ovulation and oviposition rather than to keeping an offspring in a parent's body during its development. Their short life and their importance as agents of oviposition, as in birds [83], sets them apart from the CL, except in relation to the evolution of CL in general and of eutherian CL in particular.

Uniqueness of the eutherian CL Like the uniqueness of eutherian viviparity, that of the CL lies not so much in its particulars—except for angiogenesis of the GCs, which is uniquely mammalian—as in all of them as a group. The most important of all the CL's aspects that emerges from this point of view is the relation between the CL itself and of P secretion to the regulation of viviparity. In eutherians, the CL's secretion of P became an indispensable factor in the establishment of pregnancy, and P secreted by the CL, the placenta, or both became an integral component of the complex of factors that maintained pregnancy to term. In the nonmammalian vertebrates, and even in marsupials (except for the establishment of pregnancy), this essentiality is either lacking or questionable [110, 125]; their CL and P secretion seem to be only another expression of the viviparous process itself rather than one of its essential causes or, at most, only a nonessential accessory in its maintenance. In many cases, the CL is dispensable [125, 142]; (see also p. 350, The follicular GC is the larval form of luteal GC).

It is the essentiality itself of the eutherian CL and P that gets in the way of seeing noneutherian CL other than as playing the same role but not doing it as well. If, instead, we could look on noneutherian CLs simply as POFs, no matter how much any of them may resemble the eutherian CL, we may see that their relation to viviparity can be altogether different from that of a eutherian CL. Perhaps then the concept of the latter as a mutant form of the vertebrate POF will emerge and its uniqueness become clear.

	BACKGROUND: HOW THINGS EVOLVE

TOP ABSTRACT INTRODUCTION BACKGROUND: REPRODUCTION BACKGROUND: HOW THINGS EVOLVE PUTTING IT ALL TOGETHER NOTES TO THE READER REFERENCES

 
Darwin's theory of natural selection [143], even since its synthesis with genetics [144, 145], still makes the origin of major differences (genera, families, orders, etc.) hard to understand if, indeed, it was the only agent of change. An examination of three aspects of evolution—adaptations, improvisations, and saltations—may help to lessen this difficulty.

Adaptations

An adaptation is a trait that because of its particular form and function is useful to its possessor in its way of life, including the ability to reproduce [146]. For example, a streamlined body adapts fish to life in water, lungs adapt terrestrial vertebrates to life in air, and sticky tentacles adapt the sea anemone to a sessile life among moving nutrients. Although all parts of an organism are not adaptations, they are what give an organism selective value. Biologists seem to have two conflicting views about how they arise. One is that usefulness makes an adaptation appear and that uselessness makes it go away. I'll call this the use/disuse viewpoint. The other is that an already existing trait becomes an adaptation when a change in its possessor's way of life happens to make that trait useful. I'll call this the improvisation viewpoint. The difference between them rests on where and how change occurs.

According to the use/disuse viewpoint, natural selection very slowly transforms a minutely useful trait into an adaptation. According to the improvisation viewpoint, a relatively rapid change in a trait's environment can transform it into an adaptation almost immediately; natural selection may then lead to further modifications.

Although the use/disuse principle differs from Lamarck's, its adherents too often see it as if usefulness itself was the cause rather than the result of each small change. Even Darwin believed (among several other such Lamarckianisms) that the teats of dairy cattle were larger than those of wild cattle because of more frequent use [143, pp. 10–11]. He also attributed the blindness of animals living in darkness "wholly to disuse" [143, p. 113]. The much more reasonable reason, of course, is that whatever caused the animal's blindness made darkness a more favorable habitat than a well-lit one. Yamamoto and Jefferey [147], for example, found in a species of teleost that the lens of embryos of blind fish living in darkness was unable to induce normal eye development, whereas members of the same species living at the surface had normal eyes. When they transplanted the lens of sighted embryos into the optic cups of embryos of the blind fish, the latter developed into normally sighted fish.

Distortions of the relation between use/disuse and the origin of adaptations are frequent even today—see any issue of Science!—and they make excellent material for parody, with Dr. Pangloss' opinion that the nose was designed to support spectacles as the high point [148]. Adaptationism, however, does not mean that natural selection cannot lead to adaptations. The fish's streamlining is a typical example, and many others can be cited [146]. However, it is the aleatoric quality of improvisation that is the much richer source of adaptations.

Improvisations

I improvise when I use a dinner knife as a screwdriver. A pianist improvises when he or she plays a dozen ad lib variations of Yankee Doodle. Improvisation (from Latin roots meaning "not seen before") is an unplanned, usually sudden change that makes a thing become or do something it could not be or did not do before while still retaining enough of its original character to be recognizable, even if only by an expert. For example, some viviparous sharks, by continuing to ovulate well-yolked eggs while gestating a few embryos in the oviduct, are using these eggs as embryonic nutrient [99, 100] instead of as gametes. The first vertebrate to switch its habitat from water to land already had a bony infrastructure in its fins and an organ with properties similar enough to a lung's to function as one in air [149]. However these attributes may have arisen and contributed to their possessor's economy during its life in a watery environment, even if they included locomotion and air breathing, they had nothing to do with how they were eventually used on land. What is of first importance is that these attributes could be improvised into being used for something that could not have been foreseen [150].

The biologic improvisation is analogous to the technical change in design that, for example, led to the airplane. Here, the "already existing trait" was the internal combustion engine, the change in its environment was the new form of vehicle, and the combination of these two elements produced a vehicle that could fly. The analogy of the diversification of flying machines to that of terrestrial vertebrates that followed is too obvious for comment.

Darwin certainly recognized the principle of improvisation, for he wrote: "Moreover ...a modification of structure ...may at first have been of no advantage to the species, but may subsequently have been taken advantage of by descendents of the species under new conditions of life and with newly acquired habits" [143, p. 160]. He also knew about mutants ("sports") and admitted not knowing how they or a trait "of no advantage" could arise. (If only he had known about genetics!) Improvisations arise most often from mutations in a trait's environment but also from mutations in the trait itself, and the essence of a mutation, no matter how small, is a sudden change—in plain words, a leap or saltation. What else but a leap was the switch from an exchange of genes between haploid unicellular organisms, to the union of haploid gametes and the formation of a diploid zygote (i.e., sexual reproduction)? How else did sex itself or multicellularity originate? Natura non facit saltum may have the flavor of authority, but it does not keep Dame Nature from taking leaps.

Saltations

Saltation is a dirty word in evolutionary biology largely because of its misuse by pre-Darwinian naturalists and the strong saltationist flavor of Goldschmidt's idea of the "hopeful monster" starting a new species (see Schwartz [151]). Nevertheless, the most important form of biologic change is through saltations. Natural selection would be practically useless if it acted solely on what Darwin called a trait's "infinitesimally small variations." It may have had little difficulty with a trait's extremes of normal variation and none with small mutations, but it could not at all account for the emergence of neutral—or even harmful—traits.

Saltations are coming back into polite biologic society disguised as punctuated equilibrium, the history of which (1972–1999) has been well described [151, 152]. Its basic premise is that rapid, major changes are a valid alternative to Darwinian gradualism. This premise, in one form or another, is being accepted by many biologists [107, 150–156]; mutations are the richest source of saltations.

Mutations are the result of the flexible, versatile, but inherently uncertain system through which genetic information is converted into the working parts of an organism during embryogenesis or is passed on to the organism's offspring during gametogenesis [36, 37, 66, 144, 150]. The most remarkable thing about the system is that with so many possibilities for error, mistakes are so rare. Mutations can be caused by substitution of one base pair by another or by deletion of a base pair (point mutations); gene duplication; deletion of a nucleotide sequence or an entire gene; change in the position of a gene or group of genes in relation to other genes or the DNA strand as a whole; change in the order of expression or repression of a series of genes controlling development or operation of a trait; errors in transcription, in translation of the transcript, in molding of the amino acid sequence into a working protein, in how a chromosome break occurs and how it is repaired, in inversions, deletions, translocations, or insertions of chromosome parts or of entire chromosomes; change in the amount, complexity, or pattern of the nongenomic DNA; etc. [144, 157–159]. Many, and especially large mutations are lethal, but many are neutral (i.e., harmless but useless). Some are even useful. The neutral ones can lead through improvisation—and the useful ones directly—to major changes in evolution.

Improvised Adaptations as Saltations

The term preadaptation is often applied to traits that became adaptations, but this implication of prepared uniqueness draws our attention away from the most outstanding feature of all adaptations: any trait, especially during its development, or any part of a trait, down to any of the base pairs of any of the genes encoding its formation, has the potential ability to be improvised into an adaptation, depending on its structure and on whether and how its environment changes [36, 144, 150, 151, 154]. These changes are usually, if not always, associated with or expressed as mutations.

The mutation connection may not always be easy to see if the primary change is in the organism's external environment, as in the first vertebrate's transition from water to land (unless its bony fins and "lung" were themselves mutations, which is likely). The increased incidence of the four-winged fruit fly mutant in the presence of ether (the "Waddington fly"), however, shows clearly that ether affected the genes controlling appendage development [36, p. 592]. If the primary change is in the trait's internal environment, the improvisation