Cell Death and Sexual Differentiation of Behavior: Worms, Flies, and Mammals

Summary

Sex differences in the nervous system are found throughout the animal kingdom. Here, we discuss three prominent genetic models: nematodes, fruit flies, and mice. In all three, differential cell death is central to sexual differentiation and shared molecular mechanisms have been identified. Our knowledge of the precise function of neural sex differences lags behind. One fruitful approach to the “function” question is to contrast sexual differentiation in standard laboratory animals with differentiation in species exhibiting unique social and reproductive organizations. Advanced genetic strategies are also addressing this question in worms and flies, and may soon be applicable to vertebrates.

Introduction

In many animal species, males and females live in different social worlds, or at least play according to different social rules. Sex differences are commonly seen in courtship, communication, copulatory behaviors, and the processing of opposite-sex sensory cues [1,2,3]. Neural sex differences presumably allow for such behavioral differences and, indeed, are being found throughout the animal kingdom. This review compares sexual differentiation of the nervous system in three major genetic models of neural development: the nematode (Caenorhabditis elegans), the fruit fly (Drosophila melanogaster), and the laboratory mouse (Mus musculus).

In principle, neural sex differences could result from differences in any of the major neurodevelopmental events: neurogenesis, migration, neurite outgrowth, the differentiation of chemical phenotype, or cell death. Interestingly, however, in all three species(C.elegans, D. melanogaster, and M. Musculus), cell death is the best-understood cellular mechanism underlying sexual differentiation of neural tissue. This suggests that cell death may be a highly efficient strategy for building a sex-specific nervous system. Alternatively, we may know so much about this mechanism because differences in cell number are often easier to detect than more subtle changes in connectivity or gene expression. In either case, as we describe below, recent studies have provided a fine-grained molecular analysis of how sex differences in cell number develop in worms, flies, and mice. Cell death has also been linked to neural sex differences in other animals, such as birds and frogs [4,5], although due to the more limited ability for genetic manipulation, our understanding of underlying molecular mechanisms is less complete for these species.

In contrast to the progress made in understanding the mechanisms of neuronal cell death, much less is known about the function of sex differences in neuron number. This is particularly true in mammals where neural circuits are extremely complex and sex differences tend to be quantitative rather than qualitative. One strategy for circumventing this problem is to study simpler nervous systems such as those of worms and flies, where cells can be individually identified and circuitry is better known. Another avenue for understanding the function of sex differences in mammals, especially those related to sociosexual behaviors, is to expand our world-view beyond laboratory rodents to include species with different social and reproductive strategies. A good example is recent studies on naked mole-rats, which are highly social, cooperatively breeding rodents that forego sexual differentiation of behavior or postpone it until well into adulthood[6,7]. Here, we first review the basics of sex determination and differentiation in C. elegans, Drosophila and rodents. Next, examples of sex differences in cell number in each species are discussed, along with what is known about underlying molecular mechanisms and the role in sociosexual behavior. We conclude with ongoing efforts to understand the meaning of neural sex differences.

Two sexes, three mechanisms?

Although C. elegans, Drosophila and mice all come in two sexes and rely on chromosomal sex determination, there are some important differences. The two sexes in C.elegans are males (XO) and hermaphrodites (XX; basically, modified females that produce and store sperm for self-fertilization), whereas fruit flies and mammals are male (XY) or female(XX). Sex determination in C. elegans depends on the ratio of X chromosomes to autosomes. In brief, the lower ratio in males leads to the expression of a secreted factor (her-1) that acts via a cell surface receptor (tra-2) to repress tra-1. Tra-1 is a master sexual regulator, which is active in hermaphrodites and blocks male cell fates (Table 1) [8].

Table 1.

Sex determination mechanisms, examples of sexual dimorphisms in the nervous system, and core cell death pathways in nematodes (Caenorhabditis elegans), fruit flies (Drosophila melanogaster) and mice (Mus musculus).

	C. elegans	Drosophila	Mus musculus
Sexes	Male (XO), Hermaphrodite (XX)	Male (XY), Female (XX)	Male (XY), Female (XX)
Sex determining and differentiating mechanisms	X chromosome to autosome ratio determines production of a secreted protein (Her-1) that in turn regulates the activity of Tra-2 and Tra-1.	X chromosome to autosome ratio determines Sxl and Tra production leading to the alternate splicing of fruitless and doublesex.	Testis-determining gene on the Y chromosome causes formation of testes leading to the secretion of hormones that act throughout the body. Also direct sex chromosome effects.
Neural sex differences	HSN (egg-laying), CEM, RN (chemo- and mechanosensory detection of hermaphrodite), CP motoneurons (male copulatoy behavior).	fru-mAL (pheromone processing), P1 (courtship initiation), MIND (innervate male-specific muscle)	SNB (innvervate male-specific muscle), BNSTp (pheromone processing, sex behavior, stress), AVPV (control of ovulation).
Canonical cell death genes	egl 1/ced 9/ced 4/ced 3	hid/grim/reaper	Pro-survival and pro-death members of the Bcl-2 family

Abbreviations: AVPV, anteroventral periventricular nucleus; BNSTp, principal nucleus of the bed nucleus of the stria terminalis; CEM, cephalic companion neurons; fru-mAL, fruitless expressing neurons medially located above antennal lobe; HSN, hermaphrodite-specific neurons, MIND, muscle of Lawrence-inducing motoneuron; P1, fruitless expressing neurons in posterior region-1, RN, ray sensory neurons; SNB, spinal nucleus of the bulbocavernosus; Tra, transformer; Sxl, sex lethal.

Sex determination in Drosophila also depends on the ratio of X chromosomes to autosomes, but there is no known role for a secreted protein. Instead, the chromosomal sex of each cell determines the splicing within that cell of two key genes, doublesex and fruitless, which confer maleness or femaleness (Table 1) [9]. Thus, sexual differentiation is cell-autonomous. The alternative splicing of the fruitless gene is thought to be especially important for sex-specific cell fates in the nervous system, and the forced expression of male fruitless proteins can elicit male courtship behavior in otherwise normal female flies [10].

The predominant mechanism for sexual differentiation in mammals involves a gene on the Y chromosome that triggers testis development in males. Hormones produced from the testes (most notably, testosterone) then circulate throughout the body, differentiating the periphery and brain in a male direction [11*,12]. This occurs around the time of birth in mice and rats; thus, administering testosterone to genetic females perinatally masculinizes not only the body, but also many neural and behavioral features.

Although nematodes, fruit flies, and mice use different mechanisms to generate the sexes, there are important commonalities. In each case the sex determining molecular cascade relies on members of the DM(Doublesex/mab-3) family of transcription factors [13]. Moreover, there is growing evidence that in addition to hormones sex chromosomal genes exert direct effects on sexual differentiation of the brain and behavior in mammals [11*,14–16], bringing them closer to what is seen in worms and flies. Finally, in each species developmental cell death leads to important sex differences in neuron number. Neural development in all animals involves the overproduction of neurons, followed by a period of pruning by apoptosis [17], and sexual differentiation apparently taps into this mechanism, as detailed below.

Sex differences in neurogenesis also have been reported [e.g., 18], but one may wonder why examples are not more common. In mammals, the answer at first appears straightforward. Most neurons are born well before the differentiating surges of testosterone. Moreover, a neuroblast born at a proliferative zone is not fully determined; its final destination and phenotype will depend on environmental cues, so it would be difficult to target cells to precise sexually dimorphic cell groups with a change in neurogenesis. These explanations may be misleading, however, when viewed through a wider lens: species such as worms and flies also rely on developmental cell death for neural sex differences, yet cell fate is more strictly determined and the overlap with hormones is not a consideration. Presumably, the sculpting of neuron number by cell death allows for more precision, or was easier to evolve, than the tweaking of cell number by neurogenesis. Once established, the mechanism of apoptosis was conserved, as the cell death cascades in worms, flies and mammals all converge on the activation of caspases, cysteine proteases that cleave key cellular proteins to cause the cell’s demise [19]. Moreover, homologues of each of the members of the canonical cell death pathway in nematodes can be found in humans (20).

Sexual differentiation of neuron number

1. Roundworms

In C.elegans, the identity and number of cells that die is invariant between individuals. During development of the hermaphrodite, for example, exactly 131 cells die and many of these are neurons, or would become neurons if spared [21,22]. The adult nervous system consists of a series of ganglia containing exactly 294 neurons shared by both sexes, 8 hermaphrodite-specific neurons and about 90 male-specific neurons [23]. Only hermaphrodites have the so-called hermaphrodite-specific neurons (HSNs), a pair of bilaterally symmetric motoneurons that innervate vulval muscles required for egg laying behavior [24,25]. Other hermaphrodite-specific cells include the VC motoneurons that inhibit the vulval muscles [24]. The cephalic companion cells (CEMs) are specific to males. These four chemosensory neurons (together with cells of the core nervous system) are important for chemotaxis to hermaphrodite pheromones during courtship [26*]. Most other male-specific neurons also are implicated in sexually dimorphic functions, namely, the pheromonal or mechanical detection of hermaphrodites, and male-specific mating postures [23]. How sex differences in these other cells develop is less well known, however.

Painstaking visual observation of developing worms have revealed that HSNs and CEMs are born, migrate and to some extent differentiate normally before being eliminated in one sex by programmed cell death [22]. Subsequently four genes were identified that comprise the core cell death pathway in C. elegans: ced-3 (a caspase), ced-4 (homologous to the apoptosis adaptor molecule, Apaf1, in mammals), ced-9 (homologous to the mammalian pro-survival molecule Bcl-2), and egl-1 (homologous to pro-death members of the Bcl-2 family) [27]. Mutations in any of the four genes affect most of the 131 cells that die during development, including the sexually dimorphic HSNs. Interestingly, the pro-death egl-1 gene contains a binding site for the sexual regulator, tra-1 [28]. When tra-1 binds, egl-1 transcription is repressed, thereby allowing the survival of HSNs in hermaphrodites [28]. Thus, a direct link has been established from the sex-determining pathway to sexually dimorphic cell death of HSN neurons.

Recent work has also elucidated the molecular basis for the sex difference in CEM survival. The mechanism again involves tra-1, but in this case may be independent of the core cell death pathway. CEH-30 is an anti-apoptotic factor acting within CEM neurons of males [29*, 30*]. Hermaphrodites express tra-1, which binds and transcriptionally represses CEH-30, leading to CEM cell death. There is a mammalian homolog of CEH-30, Barhl1, and it can substitute for CEH-30 in protecting CEMs from death [29*]. Mutations of Barhl1 lead to increased cell death in some neural regions of mice [31], but whether this gene plays a role in sexually dimorphic mammalian cell death has not yet been examined.

2. Fruit flies

As noted above, alternative splicing of two key sex-determining genes – doublesex and fruitless -underlies sexual differentiation in Drosophila. Until recently there was thought to be a tidy dichotomy, with doublesex directing differentiation of the body and fruitless that of brain and behavior. This turns out to be an oversimplification, as several recent studies have demonstrated essential roles for the sex-specific splicing of doublesex as well as fruitless in sexual differentiation of neuroanatomy and behavior [32*–34*]. Male-specific fruitless proteins (Fru^M) are made in a few thousand neurons (about 3% of the total) of the nervous system [35]. Most Fru^M producing neurons in males have counterparts in females [36,37], but there are several important exceptions. For example, males have about 30 Fru^M -expressing interneurons in the cluster fru-mAL (medial cells above the antennal lobe), whereas females have only five [38]. This sex difference is apparently due to cell death because when the core Drosophila cell death genes (hid, grim, and reaper) are deleted, females have nearly as many mAL neurons as males [38]. The fru-mAL interneurons contribute to processing pheromonal signals that drive sexual behavior, so the sex-specific elimination of these cells may account for sex differences in the response to these cues.

Male Drosophila also have a cluster of about 20 neurons known as P1 that is completely absent in females [33*]. Differential cell death is again implicated in this difference because P1 cells are present in hid/grim/reaper mutant females. However, the P1 cluster still forms in males with an inactivating fruitless mutation and not in females expressing Fru^M proteins, indicating that in this case sexual differentiation is independent of fruitless. Many Fru^M cells co-express doublesex [34*] and, indeed, the female splice variant of the doublesex protein turns out to be responsible for the death of P1 cells [33*]. Importantly, mosaic females with a masculinized P1 cluster exhibit the initial steps of male courtship behavior, indicating that sexual differentiation of this single cell group can account for a prominent difference in sociosexual behavior.

3. Mice (and Rats)

In contrast to the situation in nematodes, in which lineage determines which cells will die, developmental neuronal cell death in vertebrates appears to be stochastic: a pool of essentially equal cells competes for trophic support, and one cannot predict a priori which will survive [17]. There are a number of well-studied sex differences in neuron number in rodents that depend on testosterone levels during critical periods in development and are due to cell death[Figure 1; reviewed in 39]. As in worms and flies, these have been linked to detection of opposite sex mates, copulatory behavior and “egg laying” (in this case, ovulation). For example, most nodes of the neural circuit from the accessory olfactory bulb to the hypothalamus that processes pheromonal cues are larger in male than in female rats [40].

Figure 1.

Sexual differentiation of neuron number in worms, flies and mice. The columns represent four different possibilities: male-specific cell groups (cells completely absent in females), female-specific (absent in males), as well as male-or female-biased (greater cell numbers in one sex). In worms, sex differences in neuronal cell number are absolute. Male-specific neurons (e.g., CEMs) die in XX hermaphrodites (first column; healthy cells represented by colored circles and dead cells by stippling) while female-specific neurons (e.g. HSNs) die in XO males (second column). Flies provide examples of male-specific neurons (first column; e.g. P1 cluster) as well as quantitative sex differences favoring males (third column; e.g. mAL cluster). Compared to the vast literature on male behavior, much less is known about the neural substrate of female behavior; we are not aware of examples of female specific or female-biased cell groups in flies (question marks in columns two and four). In mice, sex differences are not absolute and cell numbers are vastly greater. Mice have no known examples of male-or female-specific neurons (columns one and two), unless one considers the subset of spinal motoneurons in the SNB that innervate only the bulbocavernosus muscle. There are examples of male-biased (e.g., BNSTp) and female-biased (e.g. AVPV) sex differences (columns three and four); in both cases, cell groups are comprised of many different phenotypes (different colored cells in columns three and four). Sex differences may be seen in overall cell number, in specific cell types, or both. As discussed in the text, the magnitude of the differences and the molecular mechanisms underlying sex differences in cell number can vary for different cell phenotypes within single sexually dimorphic nuclei.

In addition, male rats and mice have more neurons than females in the spinal nucleus of the bulbocavernosus (SNB) and principal nucleus of the bed nucleus of the stria terminalis (BNSTp), whereas the opposite pattern (female > male) is seen in the anteroventral periventricular nucleus of the hypothalamus (AVPV)[39]. SNB motoneurons innervate muscles that attach to the phallus and are active during copulation. The BNSTp, with over 20,000 cells and many cell types, is clearly multifunctional and has been linked to pheromone processing, male sex behavior, gonadotrophin secretion, and the response to stressors[41]. AVPV is a region with direct projections to gonadotrophin releasing hormone cells that control ovulation [41]. In each case, the sex difference in cell number is due to testosterone acting during early development: testosterone treatment of females around the time of birth increases cell number in the SNB and BNSTp but reduces it in AVPV [39].

Similar to the strategy used in worms and flies, mice with mutations in cell death genes have been used to test the role of cell death in neural sex differences. Pro-survival and pro-death proteins of the Bcl-2 family critically regulate the death of developing neurons in mammals. The pro-death protein Bax, in particular, is required for most developmental neuronal cell death and apoptosis is nearly eliminated in many neural regions of bax knockout mice [42]. In mice with a targeted deletion of bax, cell number in the SNB, BNSTp and AVPV are increased and, importantly, sex differences in cell number are eliminated [43,44]. Sex differences in some behaviors (e.g. female sexual behavior and olfactory preference), also are eliminated in bax knockout mice [45; M Holmes, L Niel, D Monks, N Forger, unpublished].

The story becomes more complex when specific subtypes of neurons are examined, however. For example, the sex difference in vasopressin neurons in the BNSTp (male > female) is not eliminated in bax knockout mice, apparently because in this case testosterone directs the differentiation of neuronal phenotype [46]. Similarly, a small subset of neurons in AVPV is dopaminergic and females have about three times as many of these cells as do males [47]. Although this sex difference is due to the hormonal control of cell death [48], it is independent of Bax or Bcl-2 [43,49]. Neurons within AVPV that express kisspeptin, a peptide that stimulates GnRH secretion and regulates puberty, also are more numerous in females than in males and this sex difference is not altered by deletion of the bax gene [50]. The bulk of neurons in AVPV are GABAergic, and recent work demonstrates that sexual differentiation of these cells involves the tumor necrosis factor α-NF κB cell survival pathway as well as a bax-dependent mechanism[51*]. Thus, even within a single, relatively small nucleus such as AVPV, multiple sex differences in cell number due to different molecular mechanisms coexist. Interestingly, work on worms and flies contributed to what we know about bcl-2-realted and tumor necrosis factor α-NF κB cell survival pathways, and therefore differentiation of the mammalian brain.

Function of differential developmental cell death: What naked mole-rats have to tell us

Part of the problem in ascribing function to sex differences in the mammalian brain is that nuclei contain thousands of cells comprising multiple cell types (Figure 1). The BNSTp alone, for example, contains almost 100 times as many cells as in the entire C. elegans nervous system. In addition, sex differences in mammals typically are quantitative rather than absolute. Precisely what the “extra” cells in one sex buy the animal in terms of function is not well understood [52]. If the purpose of neural sex differences is to generate sex differences in behavior, then one might predict that species with minimal sex specific roles might lack sex differences. This seems to be true for naked mole-rats(Heterocephalus glaber). These small rodents live in underground colonies that include on average 70–90 and in some cases over 200 individuals [6, 7]. Only a single female breeds and she mates with between one and three males. All other colony members are siblings or offspring of the breeders and are socially subordinate to the breeders. The subordinates provide pup care, foraging, and colony defense but never exhibit sexual behaviors. They also show no sex differences in behavior, and the genitalia, associated muscles, and nervous system are also remarkably monomorphic [53,54]. Even the sex difference in AVPV kisspeptin neurons, so prominent in other rodents, is absent in naked mole-rats [S Zhou, M Holmes, N Forger, B Goldman, M Lovern, A Caraty, C Faulkes, CW Coen, abstract in Soc Neurosci Abstr 2010, 594:16]. Subordinates can become breeders, and this switch is accompanied by neural changes, including an increase in volume of several brain nuclei [54] and kisspeptin cell number [S Zhou abstract cited above]. These changes are seen in both sexes, however, so they correlate with breeding status and not with sex.

The relative lack of sex differences in naked mole-rats presumably is related to their social structure. Sex differences in cell number, in particular, are generally “permanent,” and may not be a good strategy for generating the sex-specific social and reproductive behaviors required of the very small percentage of animals that will ever become breeders. In support of this argument, a solitary mole-rat species is more “traditionally” dimorphic [55]. Naked mole-rat breeders may rely on more subtle neural sex differences for sex-specific roles, such as changes in gene expression or synapse number. Indeed, the only sex difference found thus far in the naked mole-rat brain concerns the expression of androgen receptors [56]. Analagously, in Astatotilapia burtoni, a chiclid fish with a social structure that involves dominant and subordinate individuals with different reproductive strategies, social status changes are associated with altered neural expression of receptors for sex steroids, kisspeptin, and GnRH [57–59].

As in naked mole-rats, some functions that differ between male and female mice may not depend on changes in gross neural morphology. While under normal conditions only male mice display male sexual behavior, females exhibit high levels of this behavior if gonadectomized in adulthood and treated with testosterone or following a single gene deletion of a vomeronasal receptor [60–62]. The most parsimonious explanation is that the basic circuitry does not differ much between males and females whereas sensitivity to elicit these behaviors does. Although differential sensitivity could be based on differences in cell number, it could also be due to the differentiation of synapse number or neuronal and glial gene expression [12].

Future Strategies

A time-honored approach to the problem of ascribing functions to neural sex differences is to turn to simpler systems. In mammals, it can be argued that we know more about the function of the difference in the SNB, a simple neuromuscular system, than any other neural sex difference [63], although even in this case the afferent inputs are not completely known. The lower complexity of flies and worms combined with a rich arsenal of genetic methods make their “brains” more accessible to functional analysis. Optogenetics has recently been used to directly compare the neural circuitry in male and female flies at both functional and anatomical levels. For example, by using light-activated ion channels to stimulate fru-expressing neurons in the relevant ganglia, it was demonstrated that flies of both sexes have the basic circuitry for male courtship song, although it normally lies dormant in females [64*]. A combined genetic and optical approach also recently identified sex specific axonal projections that may explain how the activation of a single olfactory receptor on the same set of sensory neurons can lead to very different behavioral outcomes in male and female flies [65*]. With these [66,67] and other [68–69] genetic tools now being developed for mammalian behavioral systems, there is good hope that our understanding of the function of sexually dimorphic cell types in mammals will soon increase as well.