Cell death and the song control system: A model for how sex steroid hormones regulate naturally-occurring neurodegeneration


Author to whom all correspondence should be addressed. Email: ckt@zedat.fu-berlin.de


The production, learning, and perception of song in songbirds are regulated by a series of discrete brain nuclei known as the song control system. In most songbird species, the song control system is sexually dimorphic, and these dimorphisms become more robust after birds have hatched. In seasonally breeding songbirds, the song control system grows and regresses depending upon breeding context. The development and seasonal plasticity of the song control system are dependent upon neurodegenerative processes, which can be ameliorated, at least in part, by circulating sex steroid hormones. I will describe two areas of song control system research that have provided important information about how hormonal control of cell death contributes to the shaping of behaviorally-relevant brain circuits. First, sexual dimorphism in the zebra finch song control system is robust and emerges partially due to substantial regression of female song control system nuclei during development. Second, in seasonally-breeding songbirds, the song control system regresses as birds transition from breeding to non-breeding conditions. In a controlled laboratory setting where hormones can be acutely withdrawn, these brain areas regress in only a matter of hours to days. Taken together, these results demonstrate that the study of cell death in the song control system provides an excellent opportunity for understanding how changes in circulating levels of sex steroids affect the degeneration of hormone-sensitive brain circuits.


The process of programmed cell death, or apoptosis, is a fundamental feature of brain development in all animals. As a brain grows, new neurons are generated in overabundance, and the culling of excess neurons via apoptotic mechanisms shapes brain circuits into its ultimate structure and organization. Such a process also underlies the development of several forms of sex differences in the brains of vertebrates, the emergence of which, in many circumstances, is dependent upon the action of sex steroid hormones such as testosterone (T) and its metabolite estradiol (E2). These hormones can act on brain areas to affect a multitude of attributes, including the rate of neuronal survival, leading to substantial sex differences in the brain.

The supportive roles of T and E2 during the development of sexual dimorphism in certain brain circuits is reflected in the fact that sex steroid hormones can also protect the brain from neurodegenerative processes and diseases in adults, including humans (Pike et al. 2006; Suzuki et al. 2006). For instance, in older men, age-related declines in circulating levels of androgens are associated with cognitive decline and an increased risk of neurodegenerative disease (Beauchet 2006). The active, supportive role of sex steroid hormones is echoed in the fact that several hormone-sensitive brain areas regress in response to hormone withdrawal (Cooke et al. 1999; Panzica et al. 2001; Johansen et al. 2004; MacLusky et al. 2006). It is likely that, at least in some cases, the neuroprotective effects of T are largely due to the conversion of T to E2 by the enzyme aromatase (Pereno & Beltramino 2009; Pereno et al. 2010). Given the clinical ramifications of sex steroid hormone-mediated neuroprotection, there is a significant need for animal models to study this phenomenon, especially in species that undergo substantial neurodegeneration in response to the withdrawal of circulating hormones.


There are many forms of neurodegeneration, including apoptosis, necrosis, autophagy, and retraction/atrophy of dendritic/axonal arbors. Though little is known about the molecular mechanisms that underlie neurodegeneration in the song control system, much of the evidence (described below) indicates that at least the development of sexual dimorphism and seasonal regression of song control system nuclei involve the activation of programmed cell death pathways. Activation of “death receptors” (extrinsic pathway) or cytological factors (intrinsic pathway) can initiate apoptotic mechanisms. Intrinsically, changes in the cytoplasm induce a breakdown of the mitochondrial membrane and a large efflux of cytochrome c into the cytosol (Vayssiere et al. 1994; Liu et al. 1996). This efflux activates a number of caspases, proteases essential for apoptosis (Liu et al. 1996). Caspases are a family of proteases that play an integral role in programmed cell death pathways in vertebrates and invertebrates (Kumar 2006). Activated caspases lyse cytoskeletal and nuclear proteins and ultimately lead to DNA degradation, cell “blebbing”, and membrane fragmentation into pyknotic bodies. Indeed, pyknosis is generally considered a hallmark of apoptosis (Kroemer et al. 2009). Inhibitors that reduce the activity of various caspases have greatly elucidated the mechanisms of caspase-dependent programmed cell death and may even have clinical significance (Ray 2006).

The goal of this review is to describe what is known about how sex steroid hormones regulate programmed cell death during the development and adult plasticity of the song control system, a series of discrete brain nuclei that regulate the production, learning, and perception of song in songbirds. First, I will give a brief description of the song control system and the molecular and cellular pathways that underlie apoptosis. Next, I will review what is known about hormonal control of cell death during the course of development of the song control system. Then I will provide an overview of seasonal regression of the song control system and its regulation by androgens, focusing primarily on the song control system nucleus HVC (used as a proper name). Last, I will argue that the study of such processes in the song control system can provide insights into how sex steroid hormones can support hormone-sensitive brain circuits.

The song control system

Birdsong is a very complex yet highly stereotyped behavior and is used primarily in a reproductive context; either to attract mates or to ward off rivals from a home territory (Catchpole & Slater 1995). Songbirds typically learn their vocalizations early in life, though many species can add additional elements to their repertoire after sexual maturation. Birdsong is regulated by a series of discrete brain nuclei collectively known as the song control system (Fig. 1). Much of the song control system, from auditory perception all the way to motor output, has been well characterized since it was first described in 1976 (Arnold et al. 1976; Nottebohm & Arnold 1976a; Nottebohm et al. 1976). Nomenclature for avian neuroanatomy, including the song control system, was recently revised in 2004 (Reiner et al. 2004a,b; Jarvis et al. 2005). The ascending auditory stream begins with information entering the central nervous system via the 8th nerve to various, interconnected hindbrain and midbrain nuclei, which ultimately project to the thalamic nucleus ovoidalis (Ov). Neurons in Ov project to a region of the telencephalon called Field L, which then projects to several secondary auditory telencephaic brain regions, including caudal medial nidopallium (NCM) and caudal medial mesopallium (CMM). The integration of song perception and motor output is likely to involve the transfer of auditory information from various auditory-related brain areas, including NCM, CMM, and the interfacial nucleus of the nidopallium (Nif), into the song control system via HVC. From HVC, the song control system can be broken into two distinct pathways: the descending motor pathway (DMP) and the anterior forebrain pathway (AFP). The DMP consists of projections from HVC to the robust nucleus of the arcopallium (RA), which projects to the tracheo-syringeal portion of the hypoglossal nucleus (nXIIts), which innervates the muscles of the syrinx, the vocal producing organ in songbirds. RA also projects to other brainstem nuclei, which innervate respiratory muscles, to force air through the syrinx in a precise fashion. There is also a small reciprocal projection from the dorsal portion of RA back to HVC (Roberts et al. 2008). Lesions of nuclei in the DMP result in a severe disruption of song (Nottebohm et al. 1976). The AFP consists of a subset of HVC neurons that project to Area X, a basal ganglia nucleus made up of striatal and pallidal elements, to the dorsolateral nucleus of the medial thalamus (DLM), back to the telencephalon to the lateral portion of the magnocellular nucleus of the anterior nidopallium (lMAN), whose axons bifurcate and synapse on neurons in RA and in Area X. In zebra finches, lesions of nuclei in the AFP disrupt the learning of song in juveniles but have relatively minimal impact on song in adults (Sohrabji et al. 1990; Scharff & Nottebohm 1991; Nordeen & Nordeen 1993). In Bengalese finches, partial lesions of Area X do not abolish song but instead induce a type of stuttering, which is not entirely unexpected given the role Area X may play as a basal ganglia nucleus in regulating timing of precise motor control (Kobayashi et al. 2001). Recent results show that lesions of lMAN in young, juvenile zebra finches leads to an abolishment of song-like vocalizations for which lMAN neurons seem to exhibit pre motor activity (Aronov et al. 2008). Interestingly, HVC lesions have no effect on these kinds of vocalizations. Thus, some parts of the AFP may actually control song production, at least in early developmental stages.

Figure 1.

 A simplified schematic of the avian song control system in parasagittal view. The ascending auditory stream (blue) relays auditory information to the telencephalon. Integration of auditory input and motor control is primarily gated by HVC. The descending motor pathway (green) regulates production of song. The anterior forebrain pathway (red) is essential for song learning.

Sexual dimorphic cell death in song control system nuclei

Of the songbird species that have been examined so far, many exhibit substantial sexual dimorphism in the song control system, especially in those species where only the male sings. To date, the most thoroughly studied species with regard to development of the sexual dimorphism of the song control system is the zebra finch. Outwardly, sexual dimorphism in zebra finches is quite obvious; feather coloration is much more ornate in males. Furthermore, male zebra finches produce an elaborate courtship ritual used to attract females, consisting of bouts of song coupled with a dance, whereas females do not sing at all. This stark contrast in behavior is reflected in the song control system (Nottebohm and Arnold, 1976b). The volume of HVC and RA is several times larger in male zebra finches than in females (Fig. 2). Area X is a particularly robust nucleus in males and is quite easy to see in brain sections stained with conventional techniques, but is virtually impossible to see in females. These differences in brain anatomy between the sexes are some of the most robust seen in any vertebrate species.

Figure 2.

 The song control system is sexually dimorphic. Nissl-stained brain sections cut in the coronal plane from an adult male (A) and female (B) zebra finch reveal the robust nucleus of the arcopallium (RA). RA volume, neuron density, and soma area is much greater in males than in females. Arrowheads indicate the border of RA. Scale bar = 200 μm. (C) Data replotted from Bottjer et al. (1985) and Nordeen et al. (1992) illustrating differences in the development of RA, HVC, and the lateral portion of the magnocellular nucleus of the anterior nidopallium (lMAN). The ordinate is calculated as fold change in volume of male song control system nuclei relative to the earliest age measured (12 post hatch days [PHD] for RA and HVC, 25 PHD for lMAN). RA and HVC grow substantially in males over these ages, whereas these nuclei regress in females. lMAN is smaller in adults relative to 25 PHD in both males and females but regresses more in females, primarily mediated by a loss of neurons.

Virtually all of the song control system nuclei in zebra finches are visible after ∼10 post-hatch days (PHD) using conventional staining techniques, though some are visible even earlier, especially when labeling for particular molecules. In general, the volumes of song control system nuclei are slightly larger in males than in females when nuclei are visible, suggesting that the process of sexual dimorphism begins either in the first week of life or even earlier in ovo. The developmental trajectory of song control system sexual dimorphism diverges substantially at about 14–21 PHD. At that point, song control system nuclei become larger and more robust in males. In contrast, there is substantial regression of song control system nuclei in females. Though much study has been dedicated to the enlargement of song control system nuclei in males, I will focus instead on the degeneration seen in females, since it is a very interesting example of naturally-occurring neurodegeneration in the song control system.

HVC  The youngest age at which HVC has been observed in the zebra finch brain is at PHD 4, in both males and females (Kim & Arnold 2005). HVC volume is more or less the same between the sexes at PHD 11 (though slightly larger in males), as defined by expression of retinaldehyde dehydrogenase (Kim & Arnold 2005). HVC volume and neuron number increases in both male and female zebra finches in the second week of life (Bottjer et al. 1985; Konishi & Akutagawa 1985; Kirn & DeVoogd 1989). In females, at about 15 PHD, HVC neuron number begins to decrease, dropping from about 25 000 neurons to about 12 000 neurons in just 5 days (Kirn & DeVoogd 1989). HVC in females continues to regress over the course of development, decreasing to a fraction of its maximal size. In fact, adult female HVC has only 15% of the neurons of what is seen in 15 PHD females (Kirn & DeVoogd 1989). In contrast, male HVC quadruples in volume from 11 PHD to adulthood (Kim & Arnold 2005).

The molecular mechanisms that mediate the degeneration of female zebra finch HVC during development are largely unknown, but evidence indicates that apoptosis may be a key mediator. Kirn and DeVoogd (1989) showed that the rate of pyknosis, a key (though not entirely definitive) feature of apoptosis, increases substantially at the ages when female HVC is regressing. Though both sexes show some rate of pyknosis, the maximal rate of HVC pyknosis in females is nearly two times that seen in males. In addition, the rate of pyknosis in female HVC increases by about 300% from 10 to 25 PHD, the age at which HVC is undergoing maximal regression. From 25 PHD to adulthood, the rate of pyknosis decreases, which is coincident with the steady decline of female HVC neuron number. Given that the pattern of change in the rate of pyknosis matches the pattern in the decrease in neuron number, and given the major differences between the sexes in the rate of pyknosis, these results taken together suggest that the degeneration of HVC is mediated by apoptotic mechanisms. In addition, terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL), a marker for apoptosis which labels cells with degraded DNA, has been observed in HVC (Burek et al. 1997; Tekumalla et al. 2002), providing further evidence that HVC neurons can undergo apoptotic-like cell death.

Robust nucleus of the arcopallium  The RA is observable in the brains of male and female zebra finches at 5 PHD (Kim & Arnold 2005). RA volume and neuron number increases in male and female zebra finches in the second week of life (Bottjer et al. 1985; Konishi & Akutagawa 1985; Kirn & DeVoogd 1989). In males, RA continues to increase in size and neuron number as they develop. The regression of RA in females is slightly delayed relative to HVC, beginning between 20 and 25 PHD but also follows a similar precipitous decline that continues over the course of development. The final neuron number in female RA is only about 40% of the maximum seen at 20 PHD. Sexual dimorphism in RA is not solely restricted to neuron number and nucleus volume; there are sex differences in RA neuron density and soma area as well (Konishi & Akutagawa 1985). These changes are also robust and are likely major contributors to the development of RA sexual dimorphism.

Like HVC, it is likely that apoptotic mechanisms mediate the developmental degeneration of RA in female zebra finches. The rate of pyknosis in RA increases at the ages when female RA is regressing and is nearly three times higher in females than in males (Kirn & DeVoogd 1989). The maximal rate of pyknosis in female RA occurs at 30 PHD, which is consistent with the fact that RA neuron number begins to decline after HVC, providing further evidence that the reduction in RA neuron number is mediated by programmed cell death pathways.

Though the details of the kind of apoptotic mechanism that mediates degeneration of female RA have not been thoroughly examined, there is a substantial body of literature on how afferent input to RA influences the development of RA sexual dimorphism. First, afferent input from HVC is necessary in order to prevent regression of RA during this critical period of development. HVC axonal terminals lie in wait outside of RA in young male and female zebra finches. Between 25 and 35 PHD males, HVC axons invade RA and make connections onto RA neurons (Konishi & Akutagawa 1985; Holloway & Clayton 2001). This does not happen in females, however. Instead, most HVC axons stay outside of RA while RA degenerates (Konishi & Akutagawa 1985). In addition, lesions of HVC prevent the normal growth of male RA (Akutagawa & Konishi 1994). Lesions of HVC also prevent the E2-induced growth of RA in females (Herrmann & Arnold 1991). Afferent input from lMAN also regulates the development of RA; lesions of lMAN, which also projects to RA, leads to a substantial reduction in RA volume and neuron number (Akutagawa & Konishi 1994; Johnson & Bottjer 1994). Lesions of lMAN have effects on RA only in young juveniles; lesions in older juveniles (>40 PHD) or adults have a negligible effect on RA morphometry (Akutagawa & Konishi 1994; Johnson & Bottjer 1994). It should be noted, though, that lesions of lMAN and HVC do not completely abolish the development of sexual dimorphism in RA; males with HVC and RA lesions still have a larger RA than females, albeit not nearly as large as RA in intact adults (Burek et al. 1995). This result suggests that afferent input is not the sole ingredient responsible for a larger RA in males.

Other factors are known to modulate neuron loss in RA as it develops. First, infusion of neurotrophins into RA following the lesion of lMAN prevents the loss of RA neurons, and neurotrophins are transported anterogradely from lMAN to RA, which expresses neurotrophin receptors (Johnson et al. 1997). A second factor that may influence differential regression of RA is the number of glia present in RA. Indeed, the number of non-neuronal cells in RA is much higher in males prior to the onset of the regression of female RA (Nordeen & Nordeen 1996). Infusion of fibroblast growth factor-2, a glial trophic factor, into RA reduces regression, including the rate of pyknosis, in female zebra finch RA (Nordeen et al. 1998). Thus, degeneration of RA in female zebra finches may be due to, at least in part, a lack of trophic support from upstream nuclei or differential expression of non-neuronal cell types such a glia.

lMAN  Several studies have reported that lMAN decreases substantially in neuron number in males as they mature (Bottjer et al. 1985; Bottjer 1987; Nordeen et al. 1987a; Nordeen & Nordeen 1988; Bottjer & Sengelaub 1989; Burek et al. 1991). Yet all of these studies relied upon conventional staining techniques to elucidate the border of lMAN, which can be difficult to determine over the course of development. To overcome this issue, Nordeen et al. (1992) backfilled lMAN neurons that project to RA and examined developmental changes in lMAN in males and females. They found that lMAN neuron number does not decrease in males aged from 25 PHD to adulthood. In females, on the other hand, lMAN undergoes substantial degeneration, losing as many as 50% of its neurons between the same ages. The nature of this degeneration, i.e. whether it is an apoptotic-like neuronal loss, is not known.

Area X  Though the steroidal mechanisms that underlie sexual differentiation of Area X has been thoroughly researched (Grisham & Arnold 1995; Gong et al. 1999; Grisham et al. 2002), there has been very little focus on the degeneration of Area X in females since females fail to develop a visible Area X from a very early age. Interestingly, while conventional staining and other techniques fail to elucidate an Area X in females, in situ hybridization for 17α-hydroxylase/17,20 lyase, an enzyme that is part of the steriodogenic pathway, reveals an Area X-like region in the striatum in female zebra finches (London et al. 2003). Though it is possible that similar degenerative mechanisms occur in female Area X as in HVC and RA, given the paucity of data it will not be considered further.

Regulators of sexual dimorphism in zebra finches  Unlike the development of sexual dimorphism in the brains of mammals, the emergence of sexual dimorphism in the song control system in zebra finches does not completely depend upon gonadally-secreted sex steroid hormones (Arnold & Schlinger 1993; Wade & Arnold 2004). This is best illustrated by the fact that castration of young male zebra finches (which has been done only as early as 9 PHD, given the technical difficulty of doing such a surgery in younger animals) has virtually no impact on either the development of song or production of song in adults (Arnold 1975a,b; Adkins-Regan & Ascenzi 1990). Several studies indicate that there are brief increases in circulating levels of sex steroid hormones in juvenile zebra finches and that the timing of these peaks varies significantly between the sexes (Prove 1983; Hutchison et al. 1984; Adkins-Regan et al. 1990; Schlinger & Arnold 1992). One problem, though, is that the details (e.g. the timing and/or presence of a peak in circulating sex steroid hormone levels) are not consistent across these studies. Thus, there is no clear picture of how circulating levels of sex steroid hormones vary across development or between sexes, if they do at all. Nevertheless, it may be the case that brief increases play a role in mediating the development of sexual dimorphism in the song control system, leading to growth in males and regression in females. Alternatively, sex steroid hormones generated in the brain may regulate this process.

Regardless, there is substantial evidence that steroid hormones can affect the development of song control system nuclei in females. For instance, early administration of sex steroid hormones, in particular E2, protects the female song control system from regression (Gurney & Konishi 1980; Gurney 1981; Pohl-Apel 1985; Simpson & Vicario 1991; Adkins-Regan et al. 1994; Grisham & Arnold 1995). E2 treatment in females at 45 PHD or older fails to protect RA from regression (Konishi & Akutagawa 1988). Some evidence indicates that steroid-mediated neuroprotection of female RA and lMAN is mediated via HVC. For instance, an implant of E2 near HVC masculinizes various aspects of the ipsilateral RA and lMAN (Grisham et al. 1994). Interestingly, the antiandrogen flutamide prevents the neuroprotective effects of E2 on the female song control system (Grisham et al. 2002), which indicates that androgens must play some role in mediating the neuroprotective effects of E2. The amount of E2 may also play a role; higher levels of circulating E2 are required for the rescue of RA neurons, whereas lower levels are sufficient in protecting HVC from neuron loss (Grisham et al. 2008). In cultured brain sections from both males and females containing HVC and RA, E2 promotes the innervation of HVC axons into RA (Holloway & Clayton 2001). Direct effects of sex steroid hormones on some song control system nuclei are to be expected because several nuclei are acutely sensitive to steroid hormones. Neurons in RA, HVC, and lMAN express androgen receptors, and neurons in HVC express estrogen receptors (although the expression of estrogen receptors in zebra finch HVC is quite low, especially in comparison to other songbird species) (Arnold et al. 1976; Nordeen et al. 1987a,b; Gahr 1990; Balthazart et al. 1992; Brenowitz & Arnold 1992; Gahr et al. 1993; Johnson & Bottjer 1995; Gahr & Wild 1997; Fusani et al. 2000). In addition, the level of expression of androgen receptor in some brain areas is higher in males than in females (Arnold & Saltiel 1979).

Despite the vast array of data indicating that sex steroid hormones can protect the female song control system from degeneration and can influence the growth of the male song control system, other results make it clear that sexual dimorphism is not solely mediated by circulating levels of sex steroid hormones (Wade et al. 1999; Wade & Arnold 2004). Instead, genetic factors unrelated to circulating sex steroids play a clear role, as is illustrated in one particular gynandromorph zebra finch, which was genetically male on one side of the body and genetically female on the other side (Agate et al. 2003). The song control system in this bird was larger on the male side of the brain, despite the fact that both sides were exposed to the same gonadal hormone environment. Interestingly, though, the song nuclei in the genetically female half were larger than what is normally observed in normal adult females. It may be the case that the genetically male half of the brain generated or at least influenced the local hormonal environment, which means this phenomenon still may be due to the action of sex steroid hormones. Though the results from this study clearly show that gonadal steroids alone cannot account for the sexual dimorphism of the song control system, this does not dismiss the fact that sex steroid hormones play a supportive and neuroprotective role in preventing degeneration of the developing song control system in zebra finches. Indeed, the results summarized in this section underscore the fact that sex steroid hormones contribute to the development of sexual dimorphism in the song control system and can regulate neuronal death that is likely mediated by apoptotic mechanisms.

Seasonal regression of the song control system

Songbirds that reproduce in temperate latitudes breed in a seasonal fashion in order to maximize reproductive output when conditions are optimal. The major stimulus that indicates the breeding season should begin is an increase in day length, which stimulates seasonally-breeding songbirds to enter breeding condition, primarily characterized by an increase in circulating levels of T. While under breeding conditions, male songbirds increase their song rate and stereotypy. These changes in photoperiod and circulating levels of T promote the growth of the song control system, which underlies the changes in song output. Much is known about the seasonal growth of the song control system, which has served as an excellent model of adult brain plasticity (Nottebohm 1981; Ball et al. 2004; Brenowitz 2008). In contrast, much less is known about the mechanisms regulating the transition from breeding to non-breeding conditions. In wild male Gambel’s white-crowned sparrows, for example, once they mate, establish nests, and start brooding, T levels decline to basal levels within a matter of weeks (Wingfield & Farner 1978). This decline happens prior to the summer solstice while photoperiods are still increasing. Thus, the transition to non-breeding conditions is not actively driven by decreasing photoperiod. Instead, male white-crowned sparrows, like other seasonally-breeding songbird species undergo absolute photorefractoriness, i.e. their reproductive axis becomes insensitive to the stimulating effects of long-day photoperiods (Dawson et al. 2001). Ultimately, photorefractoriness is characterized by a regression of testes size and a decrease in circulating T.

Studying such processes in the wild can be challenging, however. Instead, changes in the song control system and singing behavior can be induced in the laboratory using the appropriate environmental and hormonal cues, allowing for carefully controlled experiments. To mimic the spring and summer breeding season, birds are exposed to long-day (LD) photoperiod (for white-crowned sparrows, 20 h light, 4 h dark; typical of their Alaskan breeding grounds) and a systemic T implant. The T implant ensures that plasma T levels are within the physiological breeding range. To mimic the transition to non-breeding conditions, birds are exposed to short-day (SD) photoperiod (8 h light, 16 h dark). In addition, subcutaneous pellets are removed and the birds are castrated to ensure rapid and synchronous withdrawal of circulating sex steroids. This “seasonal-like” plasticity provides controlled conditions under which proximate mechanisms underlying changes in behavior and morphology can be carefully studied.

Changes in neural morphology that underlie song control system regression  Previous results have shown that long-term seasonal changes in HVC volume coincide with changes in neuron number. In other words, birds that are sacrificed while under breeding conditions (LD and high levels of circulating T) have a large HVC with many more neurons than birds under non-breeding conditions (SD and low/undetectable levels of T). Immediate changes in HVC volume following an acute transition from breeding to non-breeding conditions do not fit this pattern, however. Instead, the initial decrease in HVC volume within 12 h of withdrawal of circulating T is accompanied by a 50% increase in neuron density with no immediate change in neuron number (Thompson et al. 2007). HVC essentially collapses in on itself as the space between neurons decreases. Over the next 4 days, neuron number and soma area significantly decrease (Thompson et al. 2007). At this point, the rate of neuron loss offsets the decrease in neuron spacing, and neuron density significantly decreases. Thus, the increase in neuron density lasts only a few days.

The downstream nuclei of HVC, RA and Area X, regress more slowly than HVC, taking days to weeks instead of hours to days. The volumes of Area X and RA are not significantly regressed until 7 and 20 days, respectively, following the transition to non-breeding conditions (Thompson et al. 2007). Seasonal change in Area X seems to be mediated primarily by changes in neuron density and soma area with no change in neuron number (Thompson & Brenowitz 2005). In RA, soma area and neuron density significantly regress by 2 days after the transition to non-breeding conditions and continue to regress for at least 20 days (Thompson et al. 2007). The neuronal mechanism mediating regression of RA stands in contrast to developmental regression of RA in female zebra finches; substantial changes in soma area and neuron density with no change in neuron number mediate seasonal regression, whereas a massive neuronal die-off (along with changes in RA neuron density and soma area) contributes to developmental regression.

Neurodegenerative mechanisms of HVC regression and steroidal neuroprotection  In male white-crowned sparrows, the loss of HVC neurons during song control system regression is mediated by caspase-dependent cell death pathways. Apoptotic-like processes are to be expected, given that HVC neuron number decreases by nearly 30% following a seasonal-like transition to non-breeding conditions (Thompson et al. 2007). HVC cells positive for activated caspase-3 and cleaved PARP (Poly ADP-ribose polymerase) are apparent 3 days after the transition to non-breeding conditions (Thompson & Brenowitz 2008, 2010). In addition, infusion of a cocktail of caspase inhibitors near HVC in male Gambel’s white-crowned sparrows transitioned to non-breeding conditions prevents a decrease in HVC volume, neuron number, neuron spacing, and soma area (Thompson & Brenowitz 2008). Such an infusion also prevents an increase in caspase-3 activation (Fig. 3, Thompson & Brenowitz 2008).

Figure 3.

 Seasonal-like regression of HVC is mediated by caspase-dependent cell death. Infusion of caspase inhibitors on one side of the brain prevents regression of the ipsilateral HVC. Activated caspase-3 expression increases in the contralateral HVC (Control) 3 days following the transition to non-breeding conditions, whereas expression in the ipsilateral HVC is significantly reduced. Arrowheads delineate the ventral border of HVC. Scale bars = 100 μm. Reprinted from Thompson and Brenowitz (2008) with permission.

As discussed above, steroid hormones protect song control system nuclei from regression during development. This suggests that T and/or its metabolites could play a neuroprotective role in rescuing HVC from the regression induced by the withdrawal of systemic T and exposure to SD photoperiod, not unlike the hormone-mediated neuroprotection that is seen in several in vivo animal models of neurodegenerative insult (Ramsden et al. 2003; Pike et al. 2006; Saldanha et al. 2009). Consistent with this, direct intracerebral infusion of T near HVC unilaterally in castrated male white-crowned sparrows transferred to SD photoperiod and systemic T-withdrawal protects the ipsilateral HVC from neurodegeneration (Thompson & Brenowitz 2010). In addition, T reduces the number and density of activated caspase-3 and cleaved PARP positive cells in the ipsilateral HVC. These results strongly suggest that T and/or its metabolites directly act on HVC neurons to protect them from degenerative mechanisms induced by withdrawal of circulating sex steroids and photoperiod shift. Seasonal-like rapid regression of the song control system therefore may serve as an excellent model to further elucidate the molecular mechanisms that underlie hormone-mediated neuroprotection.

Can changes in photoperiod drive seasonal regression independent of gonadal steroids? Seasonal plasticity of the song control system that is independent of circulating levels of sex steroids has been observed several times (Bernard & Ball 1997; Smith et al. 1997; Bentley et al. 1999; Dloniak & Deviche 2001; Strand & Deviche 2007). For instance, exposing castrated Gambel’s white-crowned sparrows to long day conditions induces partial growth of song control system nuclei, despite the fact that there is no rise in circulating levels of sex steroid hormones (Smith et al. 1997). Photoperiod can modulate the effects of an increase in circulating T as well. In white-crowned sparrows with high levels of circulating T, RA neurons grow faster and have higher spontaneous firing rates when exposed to LD than to SD (Meitzen et al. 2007). Nevertheless, photoperiod-induced growth may be mediated by sex steroid hormones; LD photoperiod may induce an increase in neurogenic sex steroid hormone synthesis that stimulates growth of the song control system.

It is not yet known if changes in photoperiod are sufficient to induce steroid-independent regression of the song control system in photostimulated males with high levels of circulating T. Despite the fact that the onset of photorefractoriness is not dependent upon a transition to SD photoperiod, Thompson et al. (2007) and Thompson & Brenowitz (2008, 2009) exposed birds to SD photoperiod at the same time circulating T is withdrawn to ensure that any stimulatory signal LD photoperiod may contribute to maintaining a breeding-state song control system is removed (Thompson et al. 2007). This manipulation is not the same as a transition to photorefractoriness, because male white-crowned sparrows are probably still photostimulated after just 3–4 weeks of LD and high levels of circulating T. Yet the manipulation abruptly removes two factors known to promote enlarged song control system nuclei; high levels of circulating T and LD photoperiod. The purpose of the manipulation is to remove as much exogenous trophic support for the song control system as possible so that it resembles that which is seen in birds held in non-breeding conditions. Acute withdrawal of circulating T induces a significant decrease in HVC volume within 12 h (Thompson et al. 2007). This decrease occurs before the birds experience a change in photoperiod, which demonstrates that, at the very least, the initial regression of HVC volume is driven by the withdrawal of circulating and not by physiological changes driven by photoperiod. This suggests that regression of the song control system in male white-crowned sparrows is likely to be driven by changes in circulating levels of sex steroids. It remains to be tested whether a transition to SD with no withdrawal of circulating T can induce regression of song control system nuclei, however. Similarly, withdrawal of T while birds are maintained on LD may slow the degeneration of the song control system.

Seasonal regression and neurogenesis  The birth and death of neurons are fundamental processes in the development and adult plasticity of the vertebrate brain. Though ongoing neurogenesis in the postnatal brains of endothermic vertebrates was first suggested by Altman and colleagues over 40 years ago (Altman & Das 1965) and conclusively demonstrated by Nottebohm and colleagues (Goldman & Nottebohm 1983; Paton & Nottebohm 1984), many fundamental questions about the mechanisms of neuronal replacement remain. Of particular importance is the relationship between the rate of neuron death and the rate of neurogenesis. Given that the brains of endothermic vertebrates do not continue to grow without limit, the rate of new neuron addition to the adult brain must be offset by the rate of death of mature neurons. During seasonal growth and regression of HVC, however, the rates of neurogenesis and death are not in constant balance. HVC neuron number increases after birds transition to breeding conditions and decreases after they transition to non-breeding conditions. Ironically, though, the rate of neuronal recruitment is lowest in birds under non-breeding conditions, which suggests that there must some change in the balance of cell death and neurogenesis across breeding states (Kirn et al. 1994; Tramontin & Brenowitz 1999). Kirn et al. (1994) found that the increases in the rate of neuronal recruitment into HVC are preceeded by increases in the rate of pyknosis. This strongly suggests that neuron death in HVC induces enhanced neuronal recruitment by creating “vacancies” in HVC (Alvarez-Buylla & Kirn 1997; Wilbrecht & Kirn 2004). This hypothesis has been tested experimentally twice in the song control system. Scharff et al. (2000) selectively killed a distinct population of HVC neurons using targeted-photolysis. Culling of HVC neurons induced an increase in neuronal incorporation, demonstrating that an increase in neuronal death is sufficient in enhancing the rate of neuronal recruitment. To determine whether HVC neuron death is necessary for enhanced neuronal recruitment, Thompson & Brenowitz (2009) infused caspase inhibitors near HVC on one side of the brain in adult male white-crowned sparrows previously treated with BrdU to label new cells. They found that the blockade of HVC neuron death reduced the number and density of new neurons recruited to the ipsilateral HVC compared with contralateral HVC, providing further evidence that neuronal death is a necessary precursor to the recruitment of new neurons in the adult brain. Thus, an increase in the rate of cell death is necessary and sufficient in promoting neuronal recruitment into HVC. One hypothesized scenario of how these various rates change, taking into account what is known about seasonal changes in HVC neuron number, the rate of apoptosis and neurogenesis, and the relative homeostasis of these two processes over breeding and non-breeding conditions, is illustrated in Figure 4.

Figure 4.

 A schematic illustrating hypothesized seasonal changes in the balance of the rates of apoptosis and neurogenesis in HVC in order mediate seasonal changes in neuron number. When HVC is neither growing nor regressing, the rate of apoptosis and neurogenesis must be in balance. At the onset of breeding conditions (grey box), HVC neuron number increases. This may be mediated by a decrease in the rate of apoptosis. In addition, breeding conditions leads to a decrease in the relative rate of neurogenesis, which may include changes in ventricular zone proliferation, migration and/or incorporation into HVC. After the transition to non-breeding conditions, there is a wave of apoptosis, which likely induces an increase in neuronal recruitment. Once HVC regresses, apoptosis and neurogenesis again achieve balance.


The song control system is an excellent model for understanding how cell death shapes behaviorally-relevant brain circuits. Developmental degeneration of the female song control system reveals that steroid hormones can prevent the collapse of HVC and RA, and that afferent input from HVC and lMAN is necessary for this protection. What is not known, though, is what molecular mechanisms underlie this regression. With the publication of zebra finch genome (Warren et al. 2010) and the development of zebra finch-specific microarray technology (Replogle et al. 2008), this issue is much more amenable to study than in years past.

Seasonal regression of HVC provides an opportunity to understand what makes neurons susceptible to death following a decrease in circulating levels of androgens. Because the laboratory-based study allows for a synchronized and massive culling of HVC neurons, this phenomenon is amenable to revealing the molecular underpinnings to this process. As noted above, there is a need for an animal model that is acutely sensitive to circulating levels of sex steroids, because age-related declines in levels of circulating sex steroid hormones may play a role in neurodegenerative disease in humans. Studies of naturally-occurring neurodegeneration in the song control system can elucidate how sex steroid hormones promote the survival of neurons that would otherwise die and how the withdrawal of circulating hormones leads to degeneration of hormone-sensitive brain areas.