Neuroestrogens rapidly shape auditory circuits to support communication learning and perception: Evidence from songbirds

Abstract

Steroid hormones, such as estrogens, were once thought to be exclusively synthesized in the ovaries and enact transcriptional changes over the course of hours to days. However, estrogens are also locally synthesized within neural circuits, wherein they rapidly (within minutes) modulate a range of behaviors, including spatial cognition and communication. Here, we review the role of brain-derived estrogens (neuroestrogens) as modulators within sensory circuits in songbirds. We first present songbirds as an attractive model to explore how neuroestrogens in auditory cortex modulate vocal communication processing and learning. Further, we examine how estrogens may enhance vocal learning and auditory memory consolidation in sensory cortex via mechanisms similar to those found in the hippocampus of rodents and birds. Finally, we propose future directions for investigation, including: 1) the extent of developmental and hemispheric shifts in aromatase and membrane estrogen receptor expression in auditory circuits; 2) how neuroestrogens may impact inhibitory interneurons to regulate audition and critical period plasticity; and, 3) dendritic spine plasticity as a candidate mechanism mediating estrogen-dependent effects on vocal learning. Together, this perspective of estrogens as neuromodulators in the vertebrate brain has opened new avenues in understanding sensory plasticity, including how hormones can act on communication circuits to influence behaviors in other vocal learning species, such as in language acquisition and speech processing in humans.

Introduction

In nature, animals confront an overwhelming array of sensory cues. Filtering and processing this stream of sensory information is necessary to evaluate potential mates, mediate territory disputes, recognize kin, identify neighbors, and detect predators. Production and perception of air-borne cues manifest across multiple modalities, such as visual displays, tactile/vibrational signals, chemical cues, as well as auditory signals (Ai et al., 2017; Endevelt-Shapira et al., 2018; Mangiamele et al., 2016; Ota et al., 2015; Shamble et al., 2016; Smotherman and Narins, 2000). While most species integrate multimodal information, many rely primarily on acoustic cues for intraspecies communication, i.e., vocal communication.

Vocal communication is widespread among vertebrates. Humans specialize in spoken language. Rodents emit ultrasonic vocalizations (USVs) across many contexts, ranging from mother-pup interactions (Portfors, 2007), courtship and mating (Holy and Guo, 2005), and social play (Knutson et al., 1998). The vast majority of teleosts, such as toadfish and midshipman fish, produce underwater calls (Bass, 2008). But for most acoustically-communicating vertebrates, these vocalizations are innate. Experience-dependent vocal learning is only found in a handful of animals, including songbirds and humans (Petkov and Jarvis, 2012). Thus, hearing serves a unique dual function in songbirds: to both detect and learn their elaborate, species-specific vocal communication signals.

Here, we review the neural circuit modulation of auditory processing in a well-studied songbird species, the Australian zebra finch (Taeniopygia guttata). We suggest that songbirds in general, and zebra finches in particular, offer a unique opportunity to investigate how rapid estrogen signaling in sensory cortex enables both the processing and learning of vocal communication cues across development and in adulthood. Further, we provide suggestions for areas of future research on this topic, and suggest possible clinical implications of this research for understanding human cognition and language.

Neuromodulators that tune neural circuits

For intra-species communication to have adaptive value, an organism must integrate external and internal cues – such as energy reserves, social standing, and reproductive status – and adjust ongoing communication encounters. Such flexibility allows for context (both current and previous) to guide communication for both sender and receiver. In the vocal communication domain, the neural circuits that underlie vocal production as well as hearing must therefore be sensitive to context, by way of neuromodulation.

The recent scientific fascination with neural ‘connectomics’ has produced detailed neural circuit diagrams for a number of organisms. But it has also revealed that a wiring diagram is a useful predictor of behavior only when the dynamic ‘functional connectivity’ of that diagram is taken into account (Bargmann, 2012; Bargmann and Newsome, 2014; Marder et al., 2014). Neuromodulators such as biogenic amines, neurotransmitters, neuropeptides, and even gases like nitric oxide all are produced within neural circuits to exert modulatory effects (Katz and Lillvis, 2014; Nusbaum et al., 2017; Petersen and Hurley, 2017). To momentarily alter the wiring diagram, that is, to shift the functional connectivity of a neural circuit, modulators can influence the efficacy and even the sign (excitation vs. inhibition) of synaptic connections on a minute-by-minute timescale, enabling extraordinary circuit and behavioral flexibility. There is now growing appreciation that steroid hormones can act as neuromodulators via local synthesis and action in neural circuits (Balthazart and Ball, 2006; Kelly and Vitousek, 2017; Remage-Healey, 2014; Rudolph et al., 2016; Woolley, 2007). The emergent perspective that steroids may be genuine neuromodulators of neural circuits and behavior has been useful in guiding the exploration of neuroestrogen synthesis and action in the songbird auditory forebrain, as we describe in detail below.

Estrogens can be rapidly synthesized within sensory circuits to act as neuromodulators

Estrogens were classically thought to be secreted exclusively from the gonads. However, it is now clear that estrogens and other steroid hormones are also synthesized within the brain (Balthazart et al., 2018; London, 2016). Initial evidence for brain-derived estrogens (neuroestrogens) arose from the discovery of brain aromatase expression in multiple vertebrate species. Aromatase, the enzyme necessary for converting precursor androgens into subsequent estrogens, was previously described solely in peripheral tissue. In the 1970s, the first direct evidence for the capacity of central estrogen production showed neural aromatase in humans (Naftolin et al., 1975a; Naftolin et al., 1975b; Naftolin et al., 1971), and across a diverse range of vertebrate taxa, including reptiles, fish, amphibians, and birds (Callard et al., 1978a; Callard et al., 1978b). Follow-up work in songbirds demonstrated that the brain is the primary source of both local and circulating estrogens (Schlinger and Arnold, 1992), which suggested a novel role for central estrogen synthesis to locally target neural circuits.

We now understand that brain-derived estrogens can also rapidly tune neural circuits and impact a diverse range of behaviors. Initial evidence for rapid effects of estrogens on synaptic physiology came from single-neuron recordings in the preoptic area (POA) of female rats, in which 17β-estradiol (estradiol) altered firing rates within seconds (Kelly et al., 1976). Since then, acute effects of estrogens on neuronal activity states and cellular events have been reported for the hypothalamus, hippocampus, striatum, amygdala, brainstem, and more recently auditory cortex (Abraham et al., 2004; Bryant et al., 2005; Chaban et al., 2003; Dufy et al., 1979; Mermelstein et al., 1996; Nabekura et al., 1986; Remage-Healey and Bass, 2004; Remage-Healey et al., 2010a; Vasudevan and Pfaff, 2008; Woolley, 2007). Functionally, estradiol’s impact on circuit physiology is exceptionally diverse in terms of behavioral actions, timing, and species. In mice, aromatization is key to organize the medial amygdala early in life to selectively respond to opposite-sex olfactory cues (Bergan et al., 2014). At a more acute timescale, testosterone rapidly increases visually-guided responses to a female stimulus in male goldfish, likely via rapid aromatization into estrogens (Lord et al., 2009; Mangiamele et al., 2017). Even nociception within the dorsal horn of Japanese quail is rapidly modulated by acute estrogen actions (Evrard and Balthazart, 2004). Therefore, estrogen synthesis in the brain is important for many behaviors, neural circuits, and species, at a range of timescales.

Classically, steroid hormones like estradiol were thought to exclusively target intracellular nuclear receptors and affect transcriptional changes over the course of hours to days. However, estrogen receptors found on dendritic and axonal processes in guinea pig hypothalamic neurons provided evidence for a non-nuclear site for the neural actions of estrogens (Blaustein et al., 1992). Since then, evidence has emerged that estrogens can rapidly influence neuronal activity through membrane-docked estrogen receptors (both ERα and ERβ) that are associated with metabotropic-glutamate receptors (Mermelstein, 2009; Micevych and Mermelstein, 2008). More recently, rapid actions of estrogens have also been found to act through a G-protein coupled estrogen receptor, GPER1 (formerly the orphaned ‘GPR30’) (Barton et al., 2017; Rudolph et al., 2016; Srivastava et al., 2013). The emergent understanding of these many mechanisms for steroid actions were presented in a recent review previewing the current Special Issue (Balthazart et al., 2018). Below, we describe the contribution of recent work in songbirds testing the role of rapid neuroestrogen signaling in shaping sensory processing, and place this work in a broader context.

The songbird auditory circuit as a model to explore rapid estrogen actions on vocal communication processing and learning

Songbirds are a powerful system to explore neuroestrogen actions in sensory circuits across the lifespan. First, the forebrain circuits that guide auditory-dependent behaviors are enriched with estrogen receptors and estrogen synthase (aromatase), especially as compared to rodent sensory cortices (as reviewed in Vahaba and Remage-Healey, 2015). In agreement with high aromatase concentrations, the brain is the primary site of estradiol synthesis in male zebra finches (Schlinger and Arnold, 1992), so much so that circulating estrogen levels persist in castrated males (Adkins-Regan et al., 1990). The abundance of estrogen production and signaling in the songbird auditory forebrain makes it an attractive system to measure and manipulate neuroestrogen signaling and evaluate its effects on audition and learning. Below, we review how neuroestrogens are generated in the songbird brain, and their rapid effects on physiology and related behaviors in both adult and developing songbirds.

The organization of auditory circuits is relatively conserved across the class Aves. As in other vertebrates, birds perceive acoustic signals beginning at peripheral hair cells in the ear, and these auditory signals reach central cortical regions in the auditory forebrain (reviewed in Jarvis, 2004). As shown in Figure 1, brainstem and midbrain auditory signals are initially relayed from the thalamic nucleus oviodalis (Ov) to the avian auditory telencephalic homologue of primary auditory cortex (Field L complex; Field L2), which sends afferent projections to secondary auditory cortex, including the caudal mesopallium (CM) and the caudomedial nidopallium (NCM) (Jarvis, 2004; Vates et al., 1996; Wang et al., 2010).

Figure 1.

Two schematized songbird auditory circuits. A–B) Auditory stimuli, such as song, arrives from the brainstem and cochlear nuclei (not shown) into the midbrain MLd and into the thalamic OV (ovoidalis). The primary thalamic recipient of auditory information in the cortex is the Field L complex (L1, L2, and L3), which projects to the caudomedial nidopallium (NCM) and caudal mesopallium (CM), which are themselves reciprocally connected. Auditory information reaches HVC (used as a proper name) by way of the nucleus interface (NIf), which itself receives projections from CM (not illustrated for clarity). Not depicted are the forebrain basal ganglia and song motor circuit pathways that are essential for song learning and production. Adapted from Brenowitz and Remage-Healey, 2016 and Vahaba and Remage-Healey, 2017.

Interestingly, while this auditory pathway is conserved across birds (Bonke et al., 1979; Vates et al., 1996; Wang et al., 2010; Wild et al., 1993), a high concentration of aromatase and estrogen receptors in avian forebrain distinguishes vocal learning birds from other species (Metzdorf et al., 1999; Silverin et al., 2000; Yoder and Vicario, 2012), especially within NCM (Caras and Remage-Healey, 2016). Within the auditory forebrain, aromatase is almost exclusively found in NCM, whereas little to no aromatase has been described in Field L or CM of zebra finches (Ikeda et al., 2017; Peterson et al., 2005; Pinaud et al., 2006; Saldanha et al., 2000). Therefore, NCM provides the predominant source of estrogens to the auditory forebrain circuitry in male and female zebra finches. In addition to being regionally concentrated in NCM, aromatase is also co-expressed in specific cell types, namely parvalbumin-positive interneurons (Ikeda et al., 2017). The exceptional capacity for estrogen synthesis in the songbird brain has led to investigations of its functional significance for cognition and sensory processing.

Acutely synthesized estrogens within songbird auditory forebrain rapidly alter physiology. In zebra finches, estrogens are produced through both de novo steroidogenesis and through aromatization of circulating androgens (London et al., 2006; London et al., 2009; Remage-Healey et al., 2010b). Social interactions and song playbacks rapidly increase neuroestrogen production in the NCM of both adult male and female zebra finches (Remage-Healey et al., 2012; Remage-Healey et al., 2008). Functionally, local increases in neuroestrogens within NCM directly enhance auditory function. In anesthetized zebra finches, perfusing estradiol in NCM rapidly increased auditory-evoked firing rates and bursting in NCM, consistent with a neuroestrogen-dependent enhancement of the representation of communication signals (Remage-Healey, 2012; Remage-Healey et al., 2010a). Similarly, acute peripheral estrogen treatment resulted in a stronger song-evoked fMRI BOLD response bilaterally in the auditory lobule (which includes Field L, CMM, and NCM) of male European starlings (De Groof et al., 2017). Interestingly, global suppression of estrogen synthesis in these same animals specifically reduced auditory responsiveness in the left but not right hemisphere auditory lobule (De Groof et al., 2017). In addition to direct effects in NCM, estradiol in NCM also increases stimulus-selectivity and auditory responsiveness in downstream regions, including the key sensorimotor nuclei HVC and NIf (Pawlisch and Remage-Healey, 2015; Remage-Healey and Joshi, 2012). Taken together, estradiol in NCM enhances central auditory processing; however, it remains to be determined whether these enhanced neural representations translate into improved audition as assessed by psychophysic or behavioral measures.

So far, we have limited information about how neuroestrogen signaling in NCM regulates behavior. Inhibiting local estrogen synthesis in the NCM of male zebra finches rapidly suppressed behavioral preferences for the birds’ own song when delivered to the left, but not right hemisphere (Remage-Healey et al., 2010a). Studies using peripheral administration of aromatase inhibitors also support the general idea of estrogen synthesis and auditory function in songbirds (Alward et al., 2016a; Yoder et al., 2012). One intriguing possibly is that in addition to rapidly guiding auditory encoding, local neuroestrogen production in NCM may also facilitate auditory memory consolidation of recent experiences in adults, as we discuss below.

Neuroestrogen provision may help consolidate recent auditory experiences

In addition to facilitating audition, elevated neuroestrogens in NCM may also rapidly enhance the consolidation of recent experiences. While this idea has been explored to a lesser extent in auditory circuits, accumulating evidence indicates that estrogens enhance cognition in another estrogen-sensitive brain region: the hippocampus (HP). Since the 1990s, exogenous estrogens were known to have mnemonic-enhancing properties in spatial memory tests (Luine and Rodriguez, 1994). Early studies by Packard & Teather provided the first behavioral description of memory-enhancement from post-training peripheral and intra-hippocampal presentations of estradiol in rodents (Packard, 1998; Packard and Teather, 1997a, b). This work built on the emerging idea that estrogens mediate ovarian-cycle dependent changes in dendritic spine plasticity in the hippocampus (Woolley and McEwen, 1992), and provided a potential behavioral/functional consequence of this plasticity. Since then, rapid estrogen synthesis and action in hippocampus has become an active area of investigation (see reviews by Choleris, Frick, Luine, and Korol in this same Special Issue), and a more detailed understanding of these mechanisms has emerged.

Generally speaking, estrogens acting in HP enhance spatial memory and object recognition (Galea et al., 2017; Srivastava et al., 2013). In rodents, estradiol’s ability to enhance memory consolidation is limited to a time-sensitive window immediately after learning: subsequent recall is unaffected by estradiol treatments if presented >2 hours after initial training (Fernandez et al., 2008). As such, the relatively acute impact on memory consolidation is thought to be mediated by rapid neuroestrogen signaling (Tuscher et al., 2016b). One puzzle associated with these findings is the limited, indirect evidence for aromatase in the rodent HP (Sato and Woolley, 2016; Tabatadze et al., 2014; Tuscher et al., 2016b; Wu et al., 2009). By contrast, the songbird hippocampus is highly enriched with synaptic and axonal aromatase protein, suggesting it is well positioned to facilitate rapid, non-classical steroid signaling (Ikeda et al., 2017; Peterson et al., 2005; Remage-Healey et al., 2011; Rohmann et al., 2007; Saldanha et al., 2004; Saldanha et al., 2000). In agreement with this, hippocampal estradiol typically facilitates spatial cognition in zebra finches (Bailey et al., 2017; Bailey and Saldanha, 2015; Oberlander et al., 2004; Rensel et al., 2013), and blocking GPER1 in HP completely prevents learning a food caching task (Bailey et al., 2017). With these recent findings in mind, neuroestrogens may play a similar role in sensory learning in songbirds, including the processing and consolidation of recent auditory experiences (Vahaba and Remage-Healey, 2015).

In addition to providing a source of estrogens to the auditory system, NCM is also implicated in auditory learning and recognition memory in adult songbirds (Bolhuis and Gahr, 2006; Chew et al., 1995; Gobes and Bolhuis, 2007; Hahnloser and Kotowicz, 2010; Mello et al., 1995). NCM exhibits a seasonal enlargement during breeding photoperiods in European starlings, who are open-ended song learners (De Groof et al., 2009). In adult zebra finches, NCM is considered a focal region for storing recent auditory representations (Chew et al., 1995; Dong and Clayton, 2008, 2009; Kruse et al., 2000; Smulders and Jarvis, 2013; Soyman and Vicario, 2017; Stripling et al., 2001). A single, brief exposure (40 mins) to song results in a short-term memory for the trained song and subsequent recognition in NCM (Dong and Clayton, 2009). While NCM appears to be required for adult auditory memory consolidation and recognition, the molecular mechanisms supporting this are only recently becoming clearer (Ahmadiantehrani and London, 2017; London and Clayton, 2008) and may involve rapid estrogen signaling.

Neuromodulators act within central auditory circuits to enable post-training memory consolidation. Like estrogens, local noradrenergic modulation of NCM is required for both auditory processing (Ikeda et al., 2015; Lee et al., 2017) and memorization (Velho et al., 2012). Moreover, estradiol levels increase in adult NCM during social and song exposure, which may facilitate changes necessary for auditory memory formation (Remage-Healey et al., 2012; Remage-Healey et al., 2008). Auditory memory consolidation in NCM involves epigenetic modifications (Phan et al., 2017), which is also a route by which estradiol mediates spatial learning in rodents (Zhao et al., 2010). In adult songbirds, inhibiting global estrogen synthesis impairs short-term auditory memorization and recognition in NCM (Yoder and Vicario, 2012). While the specific role for neuroestrogens in sensory learning has yet to be directly tested in adult songbirds, local estradiol in the olfactory bulb of mice improves odor memory consolidation (Dillon et al., 2013), providing an intriguing parallel. In the following section, we consider how a similar mechanism may exist for consolidating sensory (tutor) memories across the song learning critical period in juvenile songbirds.

Evidence that estrogens are involved in auditory processing necessary during developmental song learning

Both male and female developing songbirds form an auditory memory of their tutor’s song that is necessary for accurate vocal (song) learning and imitation. In closed-ended learners, song models are acquired across a critical period early in development classically described as occurring across two phases: 1) tutor song memorization (“sensory phase”), and 2) motor rehearsal (“sensorimotor phase”) (London, 2017). During the sensory phase (tutor song memorization), pre-vocalizing songbirds begin encoding/memorizing their father’s or older sibling’s song beginning around 25 days post-hatch (dph) (Deregnaucourt and Gahr, 2013; Immelmann, 1969; Roper and Zann, 2006), or possibly earlier since some embryonic birds are selectivity responsive to adult conspecific song (Colombelli-Negrel et al., 2012; Spencer and Minderman, 2018). Once a tutor song ‘template’ memory is formed, birds begin to evaluate their burgeoning vocal imitations compared to the tutor memory during the sensorimotor phase. The sensorimotor phase (motor rehearsal) is akin to early infant babbling (Aronov et al., 2008; Doupe and Kuhl, 1999; Lipkind et al., 2013; Prather et al., 2017) and begins with emergent vocalizations, followed by song refinement, and eventual song crystallization that coincides with sexual maturation. In the case of zebra finches, the sensorimotor phase ends with a single highly stereotyped song produced throughout adulthood. While the behavioral study of song learning has intrigued scientists as far back as Aristotle, the neural mechanisms enabling song learning has a relatively more recent history beginning around the 1960s.

Other recent reviews have provided excellent coverage of the role of motor and cortical-basal ganglia pathways in sensorimotor learning (Brainard and Doupe, 2013; Mooney, 2009), and here we restrict our discussion on neuroestrogens and song learning by focusing on tutor memorization during the sensory phase and the contributions of auditory forebrain circuits, namely NCM (Bolhuis and Gahr, 2006; Bolhuis et al., 2010). While other auditory forebrain regions are likely involved in auditory memory acquisition for learned song (e.g. CMM, Jeanne et al., 2011; Terpstra et al., 2006), as well as other auditory-responsive regions (Adret et al., 2012; Mandelblat-Cerf et al., 2014; Piristine et al., 2016; Roberts et al., 2017), these areas are largely devoid of aromatase in cell bodies as well as neurites, compared to the high expression found in NCM (Ikeda et al., 2017; Saldanha et al., 2000), and thus direct roles for neuroestrogens are unlikely.

NCM is considered a primary site required for tutor song memorization and representation (Bolhuis and Gahr, 2006; Bolhuis and Moorman, 2015; Clayton, 2013) but see (Canopoli et al., 2016, 2017). Compared to sensorimotor-aged males, auditory-evoked firing rates and the coding accuracy of single neurons for individual song stimuli in NCM are both elevated in pre-singing, sensory-aged zebra finches that are beginning to form auditory memories of their tutor song (Vahaba et al., 2017). In developing songbirds, tutor song playback evokes higher immediate-early gene expression in NCM than does a novel male’s song (Gobes et al., 2010). Behavioral preferences for tutor song in adults is abolished when NCM is bilaterally lesioned (Gobes and Bolhuis, 2007). Further, like adult songbirds, habituation to specific vocalizations occurs in NCM early in development, suggesting a role in encoding recent/familiar auditory experience (Miller-Sims and Bottjer, 2014; Stripling et al., 2001). Transcript levels for the plasticity-related immediate early gene (IEG) egr-1 (also known as zenk) peak in male NCM during the onset of sensory learning/opening of the song learning critical period (Jin and Clayton, 1997). In agreement with this, blocking plasticity-related MAPK signaling pathway in the auditory lobule (including both NCM and CMM) specifically during developmental tutoring prevents accurate tutor song imitation in adulthood (London and Clayton, 2008). Furthermore, tutoring naïve juvenile songbirds rapidly biases a subpopulation of single NCM neurons towards selectivity for the tutor’s song (Yanagihara and Yazaki-Sugiyama, 2016). These findings together suggest that NCM is required for accurate tutor song encoding, memorization and imitation, yet the molecular mechanisms enabling putative NCM-dependent auditory memory consolidation are less well known (Moorman et al., 2011). Neuronal cell density is adult-like by 20 dph in NCM (Stripling et al., 2001), and auditory responsiveness is markedly enhanced in sensory-aged songbirds compared to sensorimotor-aged males (Vahaba et al., 2017). Therefore, it may be that age-dependent changes in steroid hormones and their cognate receptors across the song learning critical period in development can partially explain NCM’s role in song learning/tutor song memorization.

Steroid hormones can limit song learning critical period plasticity during development, such as androgens which, like estrogens, also exert fast-actions on neural circuits (Bass and Remage-Healey, 2008; Foradori et al., 2008; Kelley and Bass, 2010; Wu et al., 2001). In developing songbirds, administering androgens, such as testosterone (T), to closed-ended learners before adult-like song is achieved leads to premature song and circuit crystallization (Bottjer and Johnson, 1997; Korsia and Bottjer, 1991; Whaling et al., 1995) but see (Templeton et al., 2012). One explanation for this is that testosterone acts to crystallize the song circuit and enable adult-like courtship, binding together the timing of sexual maturation with song maturation. Accordingly, circulating T peaks towards the tail-end of the song learning/sexual maturation period, potentially signifying the ‘closure’ of the critical period for song learning (Marler et al., 1987). In addition to pre-maturely crystallizing plastic song production, androgen implants early in development also lead to parallel premature ‘adult’-like physiology in the song motor pathway (Livingston and Mooney, 2001). According to this model, androgens impede motor variability by blocking vocal exploration, leading to stereotyped/crystallized song during development. In agreement with this interpretation, androgens continue to exert profound influence over song motor circuits in adulthood (Alward et al., 2013, 2017; Alward et al., 2016b; Alward et al., 2014).

Song learning experiments based on circulating levels and peripheral hormone manipulations are confounded by the fact that brain is the main source of circulating estrogens in songbirds (Schlinger and Arnold, 1992). For example, if testosterone acts as a cue to end song learning plasticity once adequate song is achieved, one would expect that peripheral T levels correspond to song learning fidelity. However, peripheral T levels measured at 100 dph in male zebra finches do not correlate with the degree of tutor song imitation (Deregnaucourt et al., 2013). Moreover, circulating T levels do not change in male zebra finches between the sensorimotor phase (50–60 dph) and the closing of the song learning critical period (105 – 130 dph), suggesting peripheral androgen levels are stable across development (Mori and Wada, 2015). Unlike androgens, circulating estradiol levels during the sensory phase of song learning are a more reliable predictor of eventual song similarity in adulthood (Marler et al., 1988). As such, estrogens are a prime candidate for the neuromodulation of tutor song encoding and memorization.

Estrogens are well-positioned to regulate song learning due to the unusual abundance of aromatase and estrogen receptors in vocal learners, as well as the role of estrogen in masculinizing vocal circuits in females. Although NCM is a highly conserved auditory forebrain region across Aves (Wang et al., 2010; Wild et al., 1993), as mentioned above there is a unique abundance of estrogen receptors and aromatase distribution in the avian forebrain of vocal learners, including NCM, compared to innately vocalizing birds (Metzdorf et al., 1999; Silverin et al., 2000; Yoder and Vicario, 2012). It is interesting to note that unlike songbirds, innately vocalizing male ruffed grouses have somatic aromatase protein expression in the Field L complex (Corfield et al., 2013). One role for local estradiol may be to establish song learning neural circuits. Exogeneous estradiol exposure in female zebra finch chicks, who do not normally sing in adulthood, masculinizes the neural song circuit by enlarging song nuclei, and enables male-like vocal learning and production (Gurney and Konishi, 1980). Follow-up studies demonstrated that brain-derived estrogens could account for the masculinization of the song motor pathway in zebra finches (Holloway and Clayton, 2001). Taken together, these studies suggest that neuroestrogens are required for vocal learning (motor) circuits in songbirds. Therefore, estradiol may be important across development for song learning, and perhaps specifically within the sensory phase, given that estradiol enhances auditory processing in adult songbirds.

Peripheral levels of estrogens and membrane estrogen receptors in the auditory cortex peak during the sensory phase of song learning, suggesting local neuroestrogens in NCM may influence tutor song memory consolidation. Sparrows, zebra finches, and canaries all have elevated levels of circulating estradiol exclusively during the sensory phase of song learning, a period critical for encoding and consolidating the model song (Marler et al., 1988; Marler et al., 1987; Pröve, 1983; Weichel et al., 1986), however see (Adkins-Regan et al., 1990). In swamp sparrows, this sensory phase estradiol peak is a reliable predictor for eventual tutor song imitation in adulthood (Marler et al., 1987). Alongside changes in local and global estradiol, GPER1 transcript levels peak at 30 dph in male telencephalon (which includes NCM) and are 5-times higher at that age than in adult males (Acharya and Veney, 2011). As GPER1 is one putative mechanism by which neuroestrogens rapidly enhance auditory processing (Krentzel et al., 2018; Remage-Healey et al., 2013), a coincident peak in circulating estradiol levels and cortical GPER1 expression suggests a role for estradiol in auditory memory consolidation in NCM.

Neuroestrogens in NCM may be important during development for modulating online auditory processing to guide tutor song memory consolidation. In contrast with adults in which song exposure leads to increased neuroestrogen production, juvenile zebra finches have reduced estradiol levels in NCM during social tutoring, and this is followed by a sharp rise one hour post-training (Chao et al., 2015). The functional role of these dynamics during tutoring is unclear. One hypothesis is that acute changes in neuroestrogens within NCM modulates online auditory processing, which may be important for tutor song memory consolidation. Recently, it was revealed that locally delivered estradiol within NCM rapidly transforms auditory encoding in a lateralized, and age-dependent fashion in developing male zebra finches (Vahaba et al., 2017). Therefore, since both adult and developing NCM is left-lateralized for auditory processing and memory consolidation (reviewed above for adults; Chirathivat et al., 2015; Moorman et al., 2012; Moorman et al., 2015), neuroestrogens in NCM may guide tutor song memorization by impacting sensory coding in a hemisphere-specific manner. The extent of interactions between neuroestrogens and established cell-signaling and molecular mechanisms enabling auditory processing and memory consolidation in songbirds remain to be tested (Ahmadiantehrani and London, 2017; London and Clayton, 2008; Moorman et al., 2011). In the sections that follow, we suggest future research directions to elucidate our understanding of how neuroestrogens mediate cognitive and sensory processes (Figure 2).

Figure 2.

Current understanding of rapid neuroestrogen signaling in songbird auditory forebrain and proposed future investigations. A) Top: Experimental timeline from previous studies on awake or anesthetized songbirds “Auditory experience” is often conspecific song playback through speakers, whereas social exposure is always a live, adult conspecific (male tutor, or male/female) presentation. Bottom: The physiological effects and temporal fluctuation of neuroestrogen production depends on age. In juvenile male zebra finches, song tutoring leads to an immediate decline in local estrogen content, whereas sixty minutes after the offset of tutoring, there is a rapid elevation in estrogen production that may be important for consolidating tutor song memories. When estrogens are presented in NCM, there are age- and hemisphere-dependent effects on spontaneous and auditory-evoked physiology. In adults, song playbacks and social presentations both elicit immediate elevations in neuroestrogen production, which rapidly enhance auditory responses both physiologically and behaviorally. We propose that the rapid production and actions of neuroestrogens B) change across the critical period for song learning, depending on age and hemisphere; C) impact local inhibitory circuits to modulate auditory signal detection; and D) regulate dendritic spine plasticity necessary for sensory memory consolidation. E2 = 17-beta-estradiol; Exc. = excitatory projection neuron; PV = parvalbumin-expressing interneuron. Portions of this figure adapted from Vahaba and Remage-Healey (2017).

Outlook & future directions

1) How does the expression of aromatase and estrogen receptors change across development?

Presently, much of what is known about estrogen receptor and aromatase expression in the songbird brain is from studies on adult songbirds. Thus, relatively little is known about how the synthesis and action of estrogens may change in the songbird brain across development. Previous work describing songbird aromatase expression in the brain has been limited by age (pre-CP aged subjects, Saldanha et al., 2000), inference of protein expression via mRNA measurement, neuroanatomical spatial resolution and antibody specificity (Palkovits punches: (Balthazart et al., 1990; Schumacher and Balthazart, 1987; Vockel et al., 1990), focusing on non-sensory cortices (Vockel et al., 1988), or limited point-sampling during vocal learning (typically ~45 dph only, Saldanha et al., 1999). While neuroestrogen levels gradually increase within NCM across development, it will be important to verify and expand on this by quantifying bilateral aromatase protein expression, as peripheral hormone changes may also impact these findings (Chao et al., 2015). Further, as neuronal cell density is adult-like by 20 dph in NCM, developmental changes in intrinsic synaptic physiology, and auditory-evoked extracellular activity (Jin and Clayton, 1997; Kudo and Yazaki-Sugiyama, 2017; Vahaba et al., 2017) may in part be explained by changing estrogen production across ontogeny (Chao et al., 2015). Acute estrogen action and/or synthesis may explain developmental changes in auditory properties during development, as GPER1 transcript levels are 5-times higher in sensory-aged male telencephalon compared to adults (Acharya and Veney, 2011). Therefore, as recently suggested by physiological recordings (Vahaba et al., 2017) hemispheric- and age-dependent changes in sensory coding may be accounted for by the expression and/or activity of neuroestrogen-related signaling molecules like aromatase.

2) Do neuroestrogens interact with inhibitory microcircuits to shape developmental song learning?

Sensory circuits primarily consist of interconnected excitatory and inhibitory neurons. Excitatory neurons receive and transmit signals within and across brain regions, whereas local inhibitory interneurons shape the gain, coding, selectivity, and modulation state of local cortical networks (Natan et al., 2017; Pi et al., 2013; Vallentin et al., 2016). Inhibitory neurons therefore shape specific auditory response states, making them primary regulators of processing and plasticity (Blackwell and Geffen, 2017). There is a diverse set of cortical GABAergic interneurons involved in auditory encoding (Tremblay et al., 2016; Wood et al., 2017), including the widely studied parvalbumin (PV) expressing neurons.

Parvalbumin is a protein directly important for calcium buffering and is a reliable marker for a subtype of inhibitory cortical interneurons (reviewed in Aizenberg et al., 2015). In mammalian cortex, parvalbumin-positive neurons are the primary inhibitory cell type, including auditory cortex (Xu et al., 2010). Specifically found within layers 2–6 of mammalian auditory cortex, parvalbumin-positive (PV+) neurons are required for encoding amplitude, frequency tuning, and sensorimotor integration, as well as auditory discrimination and adaption (Aizenberg et al., 2015; Cruikshank et al., 2001; Moore and Wehr, 2013; Natan et al., 2017; Schneider et al., 2014). As such, sharper frequency tuning is associated with recruitment of PV+ cells in auditory cortex (Li et al., 2014).

Tuning by PV+ cells in auditory brain regions may be regulated in part by rapid estrogen synthesis and signaling. In mammals, estrogen receptors are exclusively and highly expressed in PV+ interneurons (≥80%) (Blurton-Jones and Tuszynski, 2002; Higaki et al., 2012). Moreover, peripheral estradiol administration increases PV+ neurons in the arcuate nucleus of adult female rats (Sotonyi et al., 2010), a hypothalamic brain region in which estradiol acts rapidly via membrane-bound estrogen receptors (Roepke et al., 2009). In addition to rapid estrogen actions targeting PV cells, aromatase itself is highly and consistently co-expressed in PV+ neurons within human and nonhuman primate temporal cortex (Azcoitia et al., 2011; Yague et al., 2010; Yague et al., 2006; Yague et al., 2008). Taken together, PV+ neurons are critical for sensory coding in mammalian auditory circuits, and rapid estrogen actions on and synthesis within PV+ cells likely participate in the integration of auditory signals.

In songbirds, inhibitory neurons in NCM likely contribute to auditory learning and processing. Nearly half of all neurons In NCM are GABAergic, which are activated by song presentations (Pinaud et al., 2008; Pinaud et al., 2004), and are necessary for shaping auditory processing, selectivity, and memorization (Pinaud et al., 2008; Yanagihara and Yazaki-Sugiyama, 2016). By rapidly tuning inhibitory neurotransmission necessary for auditory-evoked neural activity, neuroestrogens may modulate auditory physiology in NCM. As with human and nonhuman primate temporal cortex (Yague et al., 2006; Yague et al., 2008), PV and aromatase are co-expressed in neurons within adult songbird auditory brain regions, including NCM (Ikeda et al., 2017). Song learning during development also provides a unique opportunity to explore how estrogens and inhibitory circuits in NCM may regulate critical period plasticity. Like aromatase and estrogen receptors, parvalbumin is uniquely expressed in forebrain song nuclei of avian vocal learners (Hara et al., 2012), and higher activation of PV cells corresponds to ‘better’ visual learning in an avian association cortex-like brain region (Ambalavanar et al., 1999). Songbirds thus offer a powerful model to explore natural mechanisms gating critical period plasticity for learned complex vocal signals in auditory forebrain (London, 2017), as well as testing the role for rapid estrogen actions in PV cells on auditory encoding.

3) Do neuroestrogens acutely remodel dendritic spines in NCM to facilitate auditory plasticity?

Estrogens enhance cognition via fast-actions on dendritic spines (Luine and Frankfurt, 2012; Srivastava, 2012). Peripheral estrogen treatment improves learning and memory, and rapidly (within 30 – 40 mins) increases hippocampal synaptogenesis and dendritic spine density (Jacome et al., 2016; MacLusky et al., 2005; Phan et al., 2012). Supporting the role of local and fast actions of estradiol mediating synaptic plasticity, estradiol rapidly (after 30 mins) increases dendritic spine densities in cortical neurons via nongenomic mechanisms, (Srivastava et al., 2008), and blocking in vivo estrogen synthesis centrally, within HP, prevents estradiol-dependent circuit plasticity (Vierk et al., 2015). Together, estrogens quickly modify dendritic spine dynamics that are functionally and behaviorally necessary for improved memory raising the prospect of similar mechanisms for auditory memory consolidation in songbirds.

In adult male zebra finches, dendritic spine densities in NCM rapidly double soon after brief (30 mins) exposures to novel song, an effect which is suppressed when endocannabinoid signaling is blocked (Gilbert and Soderstrom, 2013; Holland and Soderstrom, 2017). Intriguingly, acute estrogen treatment rapidly suppresses inhibitory synaptic transmission in rodent HP via an interaction with the cannabinoid receptor type 1 (CB1) (Huang and Woolley, 2012). As NCM is thought to integrate auditory information in adult songbirds by modulating inhibitory activity, rapid estrogen signaling in NCM may help encode and consolidate auditory experience by increasing dendritic spine density.

Developing songbirds may also undergo similar estradiol-dependent spine remodeling for tutor song memorization. In developing zebra finches, experience- and age-dependent changes in dynamic spine stabilization are critical for song learning and HVC circuit development (Roberts et al., 2010). As estradiol rapidly modulates spine dynamics in mammalian neural circuits, post-tutor neuroestrogen elevations in NCM may be important for consolidating recent tutor experience (tutor song) through acute dendritic spine alterations in developing auditory forebrain. Interestingly, both extracellular signal-regulated kinase (ERK) and mammalian target of rapamycin (mTOR) signaling cascades are required in the auditory forebrain of developing male songbirds for tutor memorization and imitation (Ahmadiantehrani and London, 2017; London and Clayton, 2008), which are two intracellular routes of action required for estradiol-induced memory consolidation and related synaptic plasticity modifications in adult rodents (Fortress et al., 2013; Tuscher et al., 2016a).

Does work on neuroestrogens in songbirds and other species have clinical implications for human cognition and communication?

In humans, an association between circulating hormones and hearing has been established most convincingly for women across the menstrual cycle and during pregnancy. There is an abundance of studies showing that hormonal cycles can shift the behavioral threshold to detect sounds, verbal memory (Fernandez et al., 2003; Zimmerman et al., 2011), as well as the otoacoustic emissions detected from women (Al-Mana et al., 2010; Caras, 2013). There is now increasing interest in the role of estrogens in mediating the pathophysiology of auditory dysfunction as well, and the role of hormone-replacement therapy (HRT) in changing auditory function (Frisina and Frisina, 2016). More work is therefore needed in non-human animal models to understand the basic mechanisms of how hormones like neuroestrogens can impact vocal communication processing and memory.

There is also evidence that hormones are important for speech perception and language learning during development in humans. As with songbirds (Marler et al., 1987), elevations in circulating estradiol during development are a positive predictor of future language success in children (Schaadt et al., 2015; Wermke et al., 2014). Children with social and sensory processing difficulties, such as autism, have difficulties with voice processing and recognition, as well as underconnected auditory circuits (Abrams et al., 2013; Gervais et al., 2004). Autism and related speech language disorders may be due in-part to estrogen abnormalities, such as aromatase gene mutations (Anthoni et al., 2012). Therefore, work in animal models such as songbirds will help elucidate how estrogens transform auditory circuits in development, especially as it relates to learned vocal communication.

Conclusions

Studies on songbirds have provided critical progress toward understanding the rapid, nonclassical effects of neuroestrogens on physiological and behavioral endpoints. The current perspectives that: 1) local neuroestrogen production increases in the songbird auditory forebrain during social encounters, and 2) neuroestrogens enhance auditory processing, now direct future studies to address the broader functional significance of this modulation. Songbirds will continue to provide valuable contributions to further reveal how brain-generated estrogens interact with sensory circuits to enable natural vocal communication perception and learning across the lifespan. Going forward, disentangling whether estrogens improve auditory learning due to improved hearing alone, or whether neuroestrogens enhance both hearing and learning via parallel mechanisms, will require new experimental and theoretical approaches. Studies of songbirds should continue to enable such investigations in the future (Vahaba et al., 2017). As songbirds and humans share the rare and remarkable suite of traits for learned vocalizations (Chakraborty and Jarvis, 2015; Prather et al., 2017), understanding the role of neuroestrogens in communication learning and processing may lead to important translational discoveries.

Highlights.

• -Neuroestrogens rapidly modulate songbird sensory circuits.
• -Songbird vocal learning mechanisms implicate estrogens and cell-signaling pathways.
• -The songbird auditory cortex is key for song learning and tutor memory formation.
• -Estrogens acting in songbird auditory cortex may enhance memory consolidation.