A neural hub for holistic courtship displays

Summary

Courtship displays often involve the concerted production of several distinct courtship behaviors.  The neural circuits that enable the concerted production of the component behaviors of a courtship display are not well understood.  Here we identify a midbrain cell group (A11) that enables male zebra finches to produce their learned songs in concert with various other behaviors, including female-directed orientation, pursuit and calling.  Anatomical mapping reveals that A11 is at the center of a complex network including the song premotor nucleus HVC as well as brainstem regions crucial to calling and locomotion. Notably, lesioning A11 terminals in HVC blocked female-directed singing, but did not interfere with female-directed calling, orientation or pursuit. In contrast, lesioning A11 cell bodies strongly reduced and often abolished all female-directed courtship behaviors. However, males with either type of lesion still produced songs when in social isolation. Lastly, imaging calcium-related activity in A11 terminals in HVC showed that during courtship A11 signals HVC about female-directed calls and, during female-directed singing, about the transition from simpler introductory notes to the acoustically more complex syllables that depend intimately on HVC for their production. These results show how a brain region important to reproduction in both birds and mammals enables holistic courtship displays in male zebra finches that include learning songs, calls and other non-vocal behaviors.

Graphical Abstract

Introduction

Males in many animal species produce elaborate courtship displays that comprise several distinct behaviors. For example, during courtship, male peacock spiders make dance-like locomotor movements while elevating a brightly colored abdominal “fan”^1,2, male bowerbirds generate deep-throated calls in concert with stereotyped postural changes and unilateral wing raises³, and male red-capped Manakins slide backwards down a branch towards a perched female while rapidly clicking their wings to generate audible buzzes and clicks^4,5. While the neural circuits that control individual components of a male’s display are in some cases well-described^6,7, little is known about the neural circuits that enable a wide variety of motor programs to be produced together during courtship. The courtship display of the male zebra finch includes a song comprising a sequence of stereotyped syllables, or motif, that is seamlessly coordinated with a variety of other behaviors, including female-directed orientation and pursuit, postural changes, beak-wipes and calling^8–10 (Video S1). Therefore, male zebra finches offer the potential to identify the neural circuits that help recruit a wide variety of motor programs during courtship. Despite this potential, little is known about the neural circuits that enable male zebra finches to recruit the song motif along with other vocal and non-vocal behaviors to produce holistic courtship displays.

Presumably, the underlying neural circuits must access motor regions for learned song as well as those that control female-directed calls and the head and body movements necessary to orient towards and pursue a female. Studies of zebra finches, as well as other songbirds, have thoroughly characterized the specialized forebrain nuclei, especially the song premotor nucleus HVC, that control and pattern the motif^11–21. In addition, vocal gating circuits have been identified in the avian midbrain periaqueductal gray (PAG^6,22–25) and a region that initiates hopping and flying has been localized to the pontine reticular formation^26–28. How such highly distributed motor regions are recruited to enable the male’s holistic courtship display remains a mystery. One idea is that this recruitment is achieved by a neural circuit that receives information about sexual motivation and communicates with HVC as well as the vocal PAG and the pontine reticular formation.

The midbrain A11 cell group in mammals is implicated in consummatory aspects of male sexual behavior through its descending dopaminergic projections to the spinal cord²⁹ and is part of a larger dopaminergic neuronal network found in all vertebrates^30,31 that is implicated in motor control, motivation, reward and reproduction^32–35. The midbrain A11 in songbirds receives input from the medial preoptic nucleus (POM)³⁶, a region important to appetitive and consummatory aspects of reproduction^37–39, and extends axons into HVC^11,40, raising the possibility that it functions to gate the song motif along with other vocal and non-vocal aspects of the male’s courtship display. In fact, immediate early gene expression in A11 neurons in male finches is elevated in a variety of salient interactions with other zebra finches, including when an adult male engages in female-directed singing^34,41, a sexually-motivated vocal display that is an essential appetitive component of the male’s courtship display. Moreover, infusing muscimol into A11 largely prevents testosterone-induced singing in castrated male canaries, suggesting that the POM to A11 to HVC pathway is essential to sexually-motivated singing⁴². Lastly, tracing studies in quail, which lack an HVC and do not sing, indicate that the central gray region in which A11 is embedded projects to midbrain and brainstem structures that control calling and locomotor motor programs used in courtship⁴³. Here we combine pathway tracing combined with molecular phenotyping, cell- and axon-terminal specific manipulations, vocal analysis and machine learning tools, and in vivo calcium imaging to study how A11 contributes to the production of holistic courtship displays of male zebra finches.

Results

A11 axons target regions that encode a variety of motor programs important to courtship displays

Although retrograde tracers have been used to establish A11’s connections with HVC in canaries and zebra finches^11,40 and anterograde tracers have been used to study POM’s projections to A11 in starlings³⁶, a systematic, A11-centered approach has not been undertaken in any songbird species to analyze its afferents and efferents. To begin to describe these connections, we injected either conventional or genetically encoded (self-complimentary GFP^44,45) anterograde tracers into the A11 of adult male zebra finches (Figure 1A, B; Figure S1A–G). These injections resulted in dense terminal labeling in the intercollicular nucleus of the midbrain (ICo), which controls calling in birds^22–24 (Figure 1C); the gigantocellular part of the reticular nucleus of the caudal pons (RPgc), a region that drives terrestrial and aerial locomotion in birds^26–28 (Figure 1D); and also confirmed A11’s known projection to HVC^11,37 (Figure 1E). Anterograde labeling was densest in the ventrolateral ICo, but sparser labeling was also detected in medial and dorsal ICo (Figure 1C, Figure S1H), which are regions where electrical stimulation can elicit calls in zebra finches and other avian species²⁵. The locations of these three axonal terminations of A11 correspond to three elements of the male zebra finch’s holistic courtship display, namely female-directed calls (ICo), locomotor movements necessary to female pursuit (RPgc), and female-directed song (HVC). We also injected retrograde tracers into A11 of adult males to identify brain regions that provide it with synaptic inputs (Figure 1A). These injections identified four major sources of input to A11: the medial and ventrolateral ICo; the medial preoptic nucleus (POM); the lateral deep cerebellar nucleus (lateral DCN, also known as the dentate nucleus); and the ventral part of the intermediate arcopallium (AIv) (Figure 1F–I; Figure S1H). As such, A11 is situated to receive information about call generation from the ICo, sexual motivation via the POM^36–39,43, sensorimotor integration and motor timing through the DCN^46,47, and song-related information from AIv^48,49 (Table S1).

Figure 1. A11 axons target regions that encode learned and innate motor programs important to courtship.

(A) Schematic of male finch brain in sagittal view showing injection of anterograde (AAV-GFP/dextrans) or retrograde (retrobeads/dextrans) tracers and A11’s efferents (green) and afferents (red). Consensus map is drawn from n = 5 hemispheres from N = 5 birds. (B) Representative injection site of a viral anterograde tracing strategy with GFP (green) and fluorescent antibody labeling of TH+ cells (pseudo-coloured grey). (C-E) Axonal projections from A11 to ICo, RPgc and HVC. (F-I) Cell bodies retrogradely labelled from A11 in ICo, POM, DCN and AIv. AIv, intermediate arcopallium; DCN, deep cerebellar nucleus; ICo, the intercollicular nucleus; MLd, dorsal part of the mesencephalic nucleus; POM, medial preoptic nucleus; RPgc, gigantocellular part of the reticular nucleus of the caudal pons; SNc, substantia nigra pars compacta; VTA, ventral tegmental area. Scale bars in (C-I) are 200 μm. See also Figure S1 and Table S1.

A11 neurons synthesize both dopamine and glutamate

In addition to expressing TH, many DA neurons in mammals also express mRNAs for excitatory and inhibitory neurotransmitters, such as glutamate^50,51 and gamma-aminobutyric acid (GABA)^52,53. To explore whether A11 cells in the zebra finch also might possess such dual transmitter identities, we combined retrograde tracing from HVC along with in situ hybridization chain reaction⁵⁴ for tyrosine hydroxylase (TH, a marker for dopaminergic neurons⁵⁵), the vesicular glutamate transporter 2 (VGLUT2, a marker of glutamatergic neurons⁵⁶) and/or the vesicular inhibitory amino acid transporter (VGAT, a marker of GABAergic neurons⁵⁷). We found that the majority of TH+ A11 neurons, including those that project to HVC, also express VGLUT2, while none were VGAT+ (Figure 2A–F). Thus, A11 neurons are likely to exert fast excitatory effects on their postsynaptic targets in HVC in addition to exerting DA-mediated neuromodulatory effects.

Figure 2. A11 neurons synthesize both dopamine and glutamate.

(A) High-magnification images of A11 show the overlap of TH⁺ cells (green), VGlut2⁺ cells (blue) and VGAT⁺ cells (magenta). Scale bar, 100 μm. (B) Proportion of TH+ neurons in A11 that are also VGlut2⁺. χ² test, χ²₁=23.07, P < 0.001, n = 231 TH+ cells from N = 9 hemispheres from 5 birds. (C) Proportion of TH⁺ neurons in A11 that are also VGAT⁺. χ² test, χ²₁=182, P < 0.001, n = 182 TH+ cells from N = 5 hemispheres from 3 birds. (D) High-magnification images of A11 show that many HVC-projecting A11 neurons (magenta; labeled by retrobead injection in HVC) are also TH⁺ (green) and VGlut2⁺ (blue). Scale bar, 100 μm. (E) Proportion of HVC-projecting cells in A11 that are also TH⁺. χ² test, χ²₁=25.33, P < 0.001, n = 43 retrobead-labeled cells from N = 6 hemispheres from 3 birds. (F) Proportion of TH⁺ HVC-projecting cells in A11 that are also VGlut2⁺ or VGAT⁺. χ² test, χ²₁=13.33, P < 0.001, n = 27 retrobead-labeled cells that are also TH+ from N = 4 hemispheres from 2 birds and χ²₁=11, P < 0.001, n = 11 retrobead-labeled cells that are also TH+ from N = 2 hemispheres from 1 bird, respectively. Data are mean ± s.e.m.

A11 cell bodies and their terminals in HVC are crucial for female-directed songs and calls

The divergent anatomical projections of A11 raise the possibility that it plays an important role in the male zebra finch’s courtship behaviors. Therefore, we studied how the adult male behaved when placed in a chamber where visual access to an adult female could be carefully controlled via an electronic glass partition and where we could record his vocal and non-vocal behaviors (Figure 3A; Methods). Granting the male visual access to the female elicited a range of female-directed behaviors, including female-directed singing (Figure 3B, top panel). Male zebra finches also sing in social isolation, a behavior known as undirected song, which is not sexually-motivated⁵⁸. Males in our experimental chamber readily produced undirected song when visual access to the female was blocked (Figure 3B, bottom panel). This approach allowed us to systematically explore the role of A11 in both vocal and non-vocal aspects of male courtship displays.

Figure 3. A11 cell bodies and terminals in HVC are crucial for female-directed song motifs.

(A) Schematic of the behavioral paradigm. (B) Example spectrograms of female-directed (FD) song bout and undirected (UD) song bout, with introductory notes, calls and motifs underlined. (C) Schematic of 6-OHDA injection to A11 terminals in HVC, with the male finch brain shown in sagittal view. (D) Dopaminergic terminals in HVC labeled with TH antibody (pseudo-colored white) in a brain with 6-OHDA lesion in HVC (left) and an intact brain (right). Scale bar, 100 μm. (E) Schematic of 6-OHDA injection to A11 cell bodies, with the male finch brain shown in sagittal view. (F) Same as D but for 6-OHDA treatment in A11. (G) Mean normalized number of FD motifs before and after 6-OHDA injection in HVC (N=5 birds). (H) Mean normalized number of FD motifs before and after 6-OHDA injection in A11 (N=4 birds). (I-J) Same as (G and H) for undirected song bouts. (K) Mean normalized number of FD motifs before and after sham injection in HVC (top, N=5 birds) and in A11 (bottom, N=6 birds). (L) Mean post-treatment singing rates normalized to pre-treatment singing rates. A linear mixed effects model analysis (see methods) revealed a significant effect of treatment (6-OHDA vs. sham, p<0.001) and context (FD vs. UD, p<0.001). Post-hoc Tukey tests were used to compare singing rates between 6-OHDA treated males and sham injected males and between FD motifs and UD motifs. UD singing rates were significantly higher than FD singing rates (p<0.005 and p<0.001 for HVC and A11 groups, respectively). FD singing rates were significantly higher for the control group than for the treated group (p<0.05 and, p<0.001 for HVC and A11 groups, respectively). Data are mean ± s.e.m. (M) Left: Schematic showing a microdialysis probe used to deliver D1 receptor (D1R) antagonist or D1R and D2 receptor (D2R) antagonists into HVC. Right: Confocal image showing that Fast Green (pseudo-colored magenta) diffused through the probe in HVC. Scale bar, 200 μm. (N) Number of female-directed motifs in saline days (grey bars) and D1R blockers days and D1R+D2R blockers days (green and light blue bars, respectively). DA receptor blockade decreased singing rates by approximately half (paired t-test; p<0.01). See also Figure S2 and Videos S1 and S2. In all panels *, ** denote p<0.05 and p<0.005 respectively.

We first compared each male’s singing behavior before and after we lesioned either A11 terminals in HVC or A11 cell bodies in the midbrain with 6-hydroxydopamine (6-OHDA⁵⁹, Figure 3C–F; desipramine was added to the 6-OHDA solution to increase specificity for DA neurons (see Methods) and we also established that 6-OHDA treatment in either HVC or A11 did not reduce TH+ neurons in the VTA or SNc (Figure S2 A–F)). Either treatment abolished or strongly reduced female-directed singing in adult male zebra finches (Figure 3G, H, Figure S2G, H). Specifically, no female-directed motifs were detected in 3/4 A11 6-OHDA males and 2/5 HVC 6-OHDA males at two weeks post treatment; the remaining 4 birds that did produce female-directed motifs two weeks post treatment did so at rates that were significantly lower than female-directed motif rates produced by sham-treated males (Figure 3L and FigureS2G–L). These effects could also be persistent, as two males from the A11 6-OHDA group and one male from the HVC 6-OHDA group that we recorded two months post-lesion did not produce any female-directed motifs. Surprisingly, male zebra finches with A11 terminal or cell body lesions recovered their ability to sing undirected songs, albeit at reduced rates, even though they failed to sing when presented with a female (Figure 3I, J, L, Figure S2I, J). This recovery of undirected song was achieved within 5–10 days post-lesion. We also noted that motif structure became distorted in males following A11 terminal lesions but remained intact following A11 cell body lesions (Figure S2M–S). This distortion may reflect the effects of increased cell death in HVC that followed 6-OHDA treatment in this nucleus (Figure S2T, U). However, despite this distortion, the component syllables of the motif still remained readily distinguishable from other vocalizations that the male produced, as further detailed in the following paragraph. In summary, A11 plays a particularly important role in generating a female-directed song motif, a learned component of the male’s courtship display, and this role involves its projections to the song nucleus HVC.

As previously described, the A11 neurons that project to HVC synthesize and presumably release both DA and glutamate. To test whether DA release in HVC is the major driver of female-directed singing, we used microdialysis methods to reversibly block either D1-type or D1- and D2-type DA receptor signaling in the male’s HVC (Figure 3M, N). Unlike males treated with 6-OHDA in HVC, which largely lost the ability to generate female-directed songs, males treated with DA receptor blockers in HVC readily produced female-directed songs, albeit at a reduced rate (Figure 3N). Furthermore, unlike males treated with 6-OHDA in HVC, males treated with D1 receptor blockers in HVC continued to produce motifs that were not distorted (Figure S2V–Z). Therefore, DA receptor activation in HVC alone cannot fully account for the essential role that A11 axons in HVC play in female-directed singing.

In addition to a song motif, male finches also produce other vocalizations during courtship, including introductory notes, which are repetitive call-like vocalizations that immediately precede the song motif^58,60,61, and female-directed calls. In fact, males treated with 6-OHDA in HVC produced abnormally long strings of introductory notes but failed to produce a motif and still continued to produce female-directed calls at rates similar to or higher than pre-treatment values (Figure 4A (middle panel), B, C). And whereas males treated with 6-OHDA in HVC produced undirected song motifs with spectrally distorted and temporally altered syllables (Figure S2 M–S), the spectral features and durations of their introductory notes and female-directed calls were unaffected (Figure 4D–K). Lastly, the relative amplitudes of introductory notes and female directed calls were unaffected by 6-OHDA treatment in HVC, whereas syllable amplitudes significantly decreased (Figure S3A). Consequently, following 6-OHDA treatment in HVC, males produced female-directed introductory notes and calls that remained readily distinguishable from the distorted and significantly quieter syllables these males produced in social isolation (Figure S3B–E). In contrast, males with A11 cell body lesions failed to produce any introductory notes or female-directed calls (Figure 4A (bottom panel), B, C). Therefore, during courtship, the A11 to HVC projection plays a special role in recruiting the song motif, while A11 cell bodies are essential to engaging all female-directed vocalizations.

Figure 4. A11 cell bodies, but not their terminals in HVC, are important for vocal-related courtship display.

(A) Example spectrograms of female-directed vocalizations for a control male, from the same male after 6-OHDA treatment in HVC and from another male treated with 6-OHDA in A11. Magenta, blue, red and black lines denote introductory notes, motifs, female-directed (FD) calls and female calls, respectively. (B) Number of FD introductory notes before (grey bars) or after 6-OHDA treatment for HVC, HVC control injection, A11, and A11 control injection. A linear mixed effects model analysis (see methods) revealed a significant interaction effect of Treatment (6-OHDA vs. sham) X Group (A11 vs. HVC) X Time (pre- vs. post-lesion) on the production of FD introductory notes. Post-hoc Tukey tests were used to compare notes production rates between sub-groups (* p<0.05, ** p<0.005, *** p<0.0005). (C) Same as B for FD calls. Number of FD calls in the A11 lesioned group significantly decrease following 6-OHDA treatment (* p<0.05) (D) An example spectrogram (top) and amplitude envelopes (bottom traces) of introductory notes of an intact, untreated male zebra finch. (E) Same as in (A), for the same male after treatment with 6-OHDA in HVC. (F) Distributions of introductory note durations for pre-treatment (grey) and post HVC 6-OHDA treatment (green). (G) Pairwise similarity scores for randomly selected introductory notes. Similarity scores were comparable for introductory notes produced pre- and post-6-OHDA treatment (One-way repeated measurements ANOVA p=0.888). (H-J) Same as D-G but for FD calls. (K) Pairwise similarity scores for randomly selected FD calls. Similarity scores were comparable for calls produced pre- and post-6-OHDA treatment (One-way repeated measurements ANOVA p=0.073). See also Figure S3.

A11 is also important for non-vocal courtship behaviors

The courtship displays of male zebra finches also include a variety of non-vocal behaviors, including orientation towards and pursuit of the female, postural changes (standing tall), and head gestures (beak-wipes)^62–66. Movement analysis using either a supervised learning algorithm (DeepLabCut⁶⁷, 3/9 birds, Figure S4) or visual scoring (6/9 birds, see Methods) revealed that males pursued and oriented towards females in a similar manner before and after 6-OHDA treatment in HVC (Figure 5A–H, Video S2). Furthermore, 6-OHDA treatment in HVC did not affect the more erect posture that the male often adopts when in close proximity to the female as an immediate prelude to singing^62,64,65 or alter beak-wipes, which are associated with a female-directed “dance”^63,65,66, Figure 5I, J. Therefore, males with 6-OHDA treatment in HVC still generated a full range of non-vocal courtship behaviors when provided with visual access to a female.

Figure 5. A11 cell bodies are necessary for locomotion-related courtship display.

(A) Change in male’s displacement after female presentation, before and after 6-OHDA treatment. Warm and cool colors represent displacement towards and away from the female, respectively. (B) Displacement of the males towards the female, normalized by cage size, before and after lesion for HVC 6-OHDA (top) and A11 6-OHDA treated birds (N = 5 HVC birds and N = 4 A11 birds). (C) Mean displacement towards the female pre- and post-treatment. HVC 6-OHDA males’ displacements towards females were not affected by the 6-OHDA lesion, while A11 6-OHDA treated males significantly reduced their approach to the female post-treatment. A linear mixed effects model analysis revealed a significant interaction effect between Area (A11 vs. HVC) and Time (pre- vs. post-lesion). Post-hoc Tukey tests were used to compare notes production rates between sub-groups (* p<0.05). (D) Mean overall movement post-treatment, normalized to pre-treatment values did not differ between the two experimental groups (t-test; p=0.91). (E) Example of head orientation distributions after female presentation, before and after 6-OHDA treatment. Pink shading represents the range in which the female was estimated to be visible to the male. (F) Same as B for time looking at the female. (G) Mean normalized time males spent looking at the female for the two experimental groups. A linear mixed effects model analysis revealed a significant interaction effect between Area (A11 vs. HVC) and Time (pre- vs. post-lesion). Post-hoc Tukey tests were used to compare notes production rates between sub-groups. A11 6-OHDA lesion birds spent significantly less time looking at the female (*** p<0.0005). (H) Probability distribution of a random walk simulation of time spent looking at the female, mean chance probability of looking at the female was 25.28±5.81%. Mean of time spent looking at the female for A11 6-OHDA birds (dashed gray and purple, for pre- and post- treatment, respectively) and HVC 6OHDA (gray and green, for pre- and post-treatment, respectively) (I) Mean normalized number of postural changes after 6-OHDA treatment. A linear mixed effects model analysis revealed a significant interaction effect between Area (A11 vs. HVC) and Time (pre- vs. post-lesion). Post-hoc Tukey tests were used to compare number of postural changes between sub-groups (*** p<0.0005). (J) Mean normalized number of beak wipes after 6-OHDA treatment. A linear mixed effects model analysis revealed a significant interaction effect between Area (A11 vs. HVC) and Time (pre- vs. post-lesion). Post-hoc Tukey tests were used to compare number of beak wipes between sub-groups (*** p<0.0005). Data are mean ± s.e.m. See also Figure S4 and videos S1 – S3.

In contrast, males treated with 6-OHDA in A11 lost all of their non-vocal courtship behaviors. These males did not pursue the female, and actually spent more time moving away from the female than they did prior to the lesion (Figure 5A–C; Video S3). They also reduced the amount of time that they spent looking towards the female to a level that was indistinguishable from chance (Figure 5F, H; Video S3; see Methods). Lastly, these males did not display postural changes typically associated with singing or make beak-wipes (Figure 5I, J; Video S3). Notably, 6-OHDA treatments in neither A11 nor HVC affected the male’s mobility, as the overall time males spent moving around the cage while the female was present was unchanged relative to pre-treatment levels (Figure 5D). Taken together, these findings show that A11 plays an essential role in recruiting the song motif along with various other vocal and non-vocal behaviors that constitute the male zebra finch’s holistic courtship display.

Monitoring the activity of A11 terminals in HVC during directed and undirected song

Here we observed that males in which A11 terminals in HVC were lesioned could not produce female-directed songs but still generated introductory notes and female-directed calls (Figure 4A). Furthermore, although HVC is not necessary for the production of calls^12,20, which are instead thought to be gated by the ICo, which we show is reciprocally connected with A11^22–25,68, prior studies have shown that HVC neurons exhibit elevated activity that is time-locked to the production of introductory notes and also female-directed calls^19,69–72. In light of these prior studies, our current observations raise the possibility that, during courtship, A11 transmits information to HVC about introductory notes and female-directed calls. To test this idea, we injected AAV axon-targeted GCaMP6s⁷³ into the A11 of male zebra finches (n = 4) and waited 3–8 weeks for expression of GCaMP6s in A11 axons in HVC (Figure A-D). Post hoc analysis of fixed tissue from these 4 animals revealed extensive GCaMP expression in fibers and synaptic terminals in HVC, along with expression of a nuclear-targeted mRuby tag in TH+ A11 cell bodies (Figure S5A–C). We also confirmed in another subset of adult male zebra finches (n = 2), that injection of AAV axon-targeted GCaMP6s into A11 resulted in GCaMP expression in TH+ synaptic (synapsin+) puncta in HVC (Figure S5D), and that the percentage of mRuby-expressing TH+ cell bodies in A11 overlapped with the percentage of GCaMP-expressing TH+ puncta in HVC (Figure S5E).

We used fiber photometry in HVC to measure calcium signals in these A11 axons as the male interacted with a female partner (Figure 6A; Figure S5F). These imaging experiments revealed that, during female-directed singing, calcium signals in A11 axons in HVC were elevated prior to motif onset, during the period when the male is producing introductory notes (Figure 6B). Alignment of the calcium signal to the male’s vocal display showed that A11 axon activity started to increase above baseline well before the first introductory note (> 1 sec; onset time was calculated at 5% of the peak amplitude, see Methods) and peaked at motif onset (Figure 5E–G, K–M). A similar imaging strategy in which we injected AAV axon-targeted GCaMP6s directly into HVC (Figure 5C, D; n = 4 adult males) was used to detect elevations in vocalization-related calcium signals predominantly in local HVC axons and also in some HVC cell bodies. These experiments showed that, during female-directed singing, vocalization-related increases in calcium signals in local HVC axons were delayed relative to A11 axon activity, with local axonal activity increasing above baseline ~0.5 seconds before the first introductory note and peaking in mid-motif (Figure 5E–G,K–M). In contrast, during undirected song, the peaks in A11 and HVC axon activity were almost simultaneous, a temporal difference from directed song that was accounted for by a rightward shift in A11 axon activity relative to motif onset (Figure 5H–J). Therefore, while A11 axons in HVC are active during both directed and undirected singing, as expected given that HVC and A11 are both nodes within a recurrent network¹⁶, A11 leads HVC activity during directed song whereas these two regions are simultaneously active during undirected song (Figure 5K). In addition to these timing differences, higher levels of activity in A11 also characterize female-directed singing, as indicated by increased expression of the immediate early gene c-fos during female-directed song compared to undirected song (Figure S5G–I). Although the elevated c-fos expression we observed cannot be assigned specifically to the female’s presence versus other stressors created by the serial introduction of different females into the male’s cage, it is in agreement with the results reported in an earlier study⁴¹. Lastly, the calcium-related activity of A11 axons in HVC also increased when the male produced introductory notes that did not lead to a motif and during female-directed calls, but not when the male called in social isolation (Figure7A, B). These differences in calcium signals were not dependent on the amplitudes or durations of calls made in the female-directed versus socially isolated contexts (Figure 7C, D). Taken together with our other findings, these imaging results from males indicate that during their encounters with a female, A11 signals HVC about the generation of female-directed calls and introductory notes, prior to the transition to the song motif.

Figure 6. Song-related activity of A11 terminals in HVC depends on social context.

(A) Left: schematic of male finch brain in sagittal view showing fiber photometry imaging of A11 axons in HVC (A11 axons) using axon-targeted GCaMP6s. Right: GCaMP⁺ A11 axons in HVC labeled in green. (B) Calcium signal recorded from A11 axons in the HVC of a male finch as he called and sang to a female. Overlaid on the calcium signal are red, blue and magenta lines that denote the production of the male’s introductory notes, song motif and female-directed calls, respectively. (C, D) Same as (A, B) for axon-targeted GCaMP6s injection in HVC (HVC axons). mRuby⁺ cell bodies labeled in red and GCaMP+ axons labeled in green. (E) Example calcium traces recorded from HVC during female-directed (FD) singing, aligned to motif onset (black dashed line), for one A11 axon-targeted bird and one HVC axon-targeted bird. (F) Mean normalized calcium activity during FD song motifs for the birds presented in (E). Black dashed line denotes the motif onset. (G) Probability of peak activity time relative to motif onset. For each bird, for each trial we detected the peak of the signal in the time window between −0.5 to 1.5 seconds relative to motif onset. Then we generated a probability histogram of the peak times. We then averaged the histograms of all birds belonging to the same group. Purple and green dashed lines represent the mean peak activity of A11 axons and HVC axons, respectively (N=4 A11 axons birds and N=4 HVC axons birds). (H-J) Same as (E-G), for undirected song motifs. Data in (H-I) are from the same birds as in (E-F). Data in (J) are from three of the same birds as in G (one of the 4 birds in G did not sing undirected song). (K) Mean peak activity latencies for A11 axons (purple bars) and HVC axons (green bars) in female-directed and undirected singing relative to motif onsets. Peak activity of A11 axons during FD singing is significantly earlier than HVC axons peak activity during FD singing and from A11 axons activity during undirected singing (mixed-effects analysis, p<0.05. p<0.0005 and p<0.01 for the two comparisons, respectively). (L) Calcium traces recorded in HVC from one A11 axon-targeted GCaMP bird and one HVC axon-targeted GCaMP bird, aligned to the first introductory note. The A11 axon calcium signal started rising before the production of the first introductory note and preceded the increase in HVC local axon activity. (M) Mean onset time of axon calcium signals relative to first introductory note for the HVC axon-targeted group (N = 4 birds) and the A11 axon-targeted group (N = 4 birds). Calcium signal onset times for the A11 group occurred significantly earlier than for the HVC group (t-test, p<0.005). See also Figure S5.

Figure 7. A11 terminals in HVC are active in other types of social vocalizations.

(A) Example spectrograms of male introductory notes (red), female-directed (FD) calls (magenta) and calls produced in social isolation (yellow). (B) Baseline subtracted calcium signals recorded from A11 axons during these various types of vocalizations in two different males. Data are mean ± s.e.m from >10 vocalizations in each condition. Activity of A11 terminals in HVC recorded during FD introductory notes and calls was significantly higher than during calls produced in social isolation (t-test with Holm–Bonferroni correction for multiple comparisons, p<0.001). (C) The differences in peak calcium signals between introductory notes (red) and FD calls (magenta) versus isolated calls (yellow) plotted as a function of vocal amplitude. (D) Same as (C) but plotted for vocal duration.

Discussion

Here we identify a neural hub that helps to recruit a variety of motor programs to generate a holistic courtship display in the male zebra finch. To successfully court a female, the male zebra finch must seamlessly and rapidly integrate a variety of behaviors, including female-directed head and gaze orientation, locomotor pursuit, postural changes and beak-wiping, female-directed calls, introductory notes, and the song motif. The anatomical studies undertaken here illuminate how the midbrain A11 could enable such seamless and rapid motor integration through its divergent efferent projections that target forebrain, midbrain and pontine regions important to learned song, calling, and locomotion, respectively. The A11-centered anatomical mapping approach employed here complements earlier studies in songbirds that used retrograde tracer injections into HVC and anterograde tracer injections into POM to establish that A11 connects ancestral vertebrate reproductive circuits in the hypothalamus to evolutionarily more recently derived forebrain song circuitry^36,40,43. It also confirms and extends an earlier carbocyanine tracing study⁷⁴ in fixed brain tissue of the quail that concluded that the central gray region in which A11 is embedded projects to the ICo and the pontine reticular formation by showing that A11 makes similar projections in the zebra finch, in addition to the previously emphasized projection that A11 makes to HVC. Our mapping studies also reveal that A11 is reciprocally connected with the ICo, an arrangement that could enable A11 to both influence calling and transmit calling-related information to HVC. Lastly, we found that A11 receives input from the lateral DCN and AIv, two regions that are likely to encode sensorimotor and motor-related information important to singing and other courtship behaviors. Thus, A11 displays hub-like anatomy, integrating a wide range of convergent inputs from the forebrain and brainstem and making divergent projections to forebrain, midbrain and pontine centers that control vocal and non-vocal behaviors produced by the male zebra finch during courtship.

The functional studies executed here also go beyond these prior anatomical studies by showing that A11’s projection to HVC is essential to the male’s ability to produce a female-directed motif, but not other courtship behaviors, whereas the A11 cell group helps the male to generate all aspects of his courtship display. Exactly how A11 recruits the vocal and non-vocal components of the male’s courtship display is unclear. Several functions for A11 include transmitting signals encoding sexual motivation, arousal or drive to HVC and other premotor regions; gating activity in these premotor regions; or even contributing to the precise timing and patterning of different movements used in the male’s display. A recent study in canaries showed that inactivating the region in and around A11 markedly reduced testosterone-induced singing in castrated male canaries⁴². While steroid implants in surgical castrates constitute a physiologically abnormal state, and the unilateral muscimol infusion method used to bilaterally inactivate A11 also inactivated a region beyond A11, a reasonable interpretation is that A11 conveys sexual motivation signals from the POM to HVC. Indeed, male zebra finches are likely to be strongly sexually motivated when singing to a female, but not when they are singing alone. That female-directed singing requires increased levels of sexual motivation-related signals from A11 could explain the loss of directed but not undirected singing following A11 lesions, and this increased motivational signal may be reflected in the earlier peaks in A11 terminal activity and elevated immediate early gene expression in the A11 cell bodies of male zebra finches producing female-directed songs (Figure 6E–K, Figure S5G–I and Bharati & Goodson ⁴¹). In this framework, A11 input to HVC is important to both directed and undirected singing, but the threshold for directed singing is higher, perhaps because singing to an audience is energetically costlier and increases risks from both rivals and predators^75–77. Lastly, adult male finches in which HVC is either inactivated or lesioned still sing at normal rates¹², whereas we found that males treated in HVC with 6-OHDA or dopamine receptor blockers sing at reduced rates, further emphasizing a motivational role for A11 in singing.

Either additionally or alternatively to conveying signals related to sexual motivation, A11 could precisely gate, time or otherwise influence song patterning. Because many TH+ A11 neurons also express VGLUT2, they are likely to rapidly excite song premotor neurons in HVC in parallel to the slower modulatory effects of DA. Therefore, important goals of future studies will be to determine how activity in A11 terminals excites or modulates the activity of neurons on which they synapse⁷⁸, including in HVC, where neurons express various types of DA and glutamate receptors^79–82. Another important future goal will be to use optogenetic methods to directly test whether systematically activating or silencing A11 or its various terminals in HVC, ICo or RPgc can elicit or suppress specific components of the male’s courtship display. If A11 functions as a gate, then stimulating its various projections should rapidly evoke component courtship behaviors, such as song, calls, or locomotor activity. In contrast, if A11 directly contributes to patterning, such stimulation should alter specific components of naturally generated courtship behavior.

During courtship, the male produces both female-directed calls and song bouts comprising extended and tightly interwoven sequences of introductory notes and polysyllabic motifs, generated with millisecond precision over many seconds^58,83. While introductory notes are often regarded as innate vocalizations similar to certain female-directed calls⁸⁴, a more recent study indicates that the number and spectral features of introductory notes can be influenced by the juvenile male’s experience with a singing tutor⁸⁵. Similarly, while zebra finch calls are mostly innate, spectral features in the male zebra finch’s long call can also be influenced by early experience⁸⁶. Nonetheless, our experiments distinguish the selective role of A11’s projections to HVC in the production of female-directed motifs. When presented with a female, males with A11 terminal lesions in HVC still produced calls and long strings of introductory notes that failed to transition to the motif. Moreover, while these female-directed introductory notes and calls were acoustically unaltered by 6-OHDA treatment, the syllables of undirected song motifs produced by these A11 terminal-lesioned males were spectrally distorted and markedly quieter (Figure S2M–S and Figure S3). Therefore, the male’s ability to produce female-directed motifs but not introductory notes or female-directed calls depends on an intact A11 to HVC pathway. Furthermore, fiber photometry shows that A11 terminals in HVC increase their activity before and during the production of both introductory notes and female-directed calls. One possibility is that A11 receives information about female-directed calls and introductory notes from ICo and transmits this information to HVC (Figure 1F) to help mediate the transition from female-directed calls and introductory notes to motifs. While A11 is active during both female-directed and undirected singing, A11 terminal activity in HVC peaks earlier relative to motif onset and c-fos expression in A11 cell bodies is more highly elevated during female-directed song. This shift in the timing and magnitude of A11 activity parallels a shift in A11’s functional influence on song, with A11 playing an obligatory, leading role to drive or gate female-directed song motifs.

Prior studies have shown that the often-subtle acoustic features that distinguish female-directed song from undirected song greatly enhance its salience as a courtship signal^83,87. These subtle acoustic differences likely signal the male’s sexual motivation and vigor, information that would be useful to the female when selecting a mate⁵⁸. Our finding of the selective role that A11’s projections to HVC play in driving or gating the directed but not undirected song motif (Figure 3) underscores that the circuitry that promotes these two types of singing is at least partially distinct⁸⁸. A reasonable idea is that the female’s presence heightens the male’s sexual motivation and arousal⁸⁹, activating ascending pathways, including the POM, A11 and HVC, to drive or gate the female-directed song^37,43. While the pathway that drives or gates the undirected song is unknown, singing in socially isolated males is strongly reduced by administering D2-receptor antagonists either systemically or specifically in a song-specialized region of the basal ganglia (sBG)^90,91. More generally, because A11 and HVC are part of a recurrent network that involves the sBG¹⁶, our findings support the idea that song production shifts from bottom-up control (i.e., POM > A11 > HVC) during encounters with a female to top-down control (i.e., HVC and its other afferents besides A11) when the male sings in social isolation.

The midbrain A11 cell group in mammals is the likely homologue of the songbird midbrain A11^30,42,43,92. In mammals, A11’s dopaminergic projections to the spinal cord are thought to play an important role in male sexual function and dopamine more generally contributes to appetitive, pre-copulatory behaviors^29,93,94. The present results illustrate how this ancestral reproductive structure has expanded to incorporate the learned song motif into the male’s holistic courtship display via A11’s ascending projections to song premotor circuitry in the forebrain. Indeed, because the learned syllables in the zebra finch song motif are speculated to have evolved from innate, female-directed calls⁸⁴, an attractive idea is that A11’s ancestral role in calling expanded to encompass singing as calls evolved into syllables. Moreover, the POM of birds is homologous to the mammalian preoptic area (POA), which in mice drives courtship vocalizations through its connections to vocal-gating neurons in the caudolateral periaqueductal gray (PAG), a region that corresponds to call-related regions in the avian ICo^{6,7,92,95–99}. The evolutionarily ancient and conserved nature of these hypothalamic and midbrain structures raises the possibility that A11 also helps to recruit a wide variety of movements in other species where such complex motor integration occurs during courtship, including our own.

STAR methods

RESOURCE AVAILABILITY

Lead contact

Further information and data requests should be directed to the lead contact, Richard Mooney (mooney@duke.edu).

Materials availability

This study did not generate any new reagents.

Data and code availability

• Data and code reported in this paper can be found in the following data repository: https://doi.org/10.7924/r4hd80x9b.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Adult males and adult females (>105 days old) zebra finches (Taeniopygia guttata) were obtained from the Duke University Medical Center breeding facility. All experimental procedures were in accordance with the NIH guidelines and approved by the Duke University Medical Center Animal Care and Use Committee. Birds were kept under a 14:10-hr light:dark cycle with free access to food and water. Data were collected from 50 adult male zebra finches.

METHOD DETAILS

Tissue collection.

Birds were deeply anesthetized with intramuscular injection of 20 μl Euthasol (Virbac), and transcardially perfused with 0.025 M phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were removed, post-fixed in 4% PFA at 4°C overnight and moved to cryoprotective 30% sucrose PFA solution for two days. Frozen sagittal sections (thickness of 50 or 75 μm) were prepared with a sledge microtome (Reichert) and collected in PBS.

Immunofluorescence.

Floating sections were washed three times in PBS, permeabilized with 0.3% Triton X-100 in PBS (PBST) for 10 minutes, blocked in 10% blocking reagent for 1 hour (Nacalai Tesque, 06349–64), and incubated with either primary antibody against TH, GFP, mCherry, or synapsin (1:1000, AB152; Millipore/Sigma; A11120, Invitrogen; ab167453, Abcam; 106 011, Synaptic Systems) at 4°C overnight. Sections were then washed three times in PBS and incubated with a fluorophore conjugated secondary antibodies (1:500; Invitrogen, A-21245, A21207, A11001) in PBS at room temperature for 2–4 h, followed by three washes in PB. Sections were coverslipped with Fluoromount-G (SouthernBiotech), and then imaged with a confocal microscope (Zeiss) through a 20x or 10x objective lens controlled by Zen software (Zeiss). To label A11 projections and inputs, AAV2/9.CAG-scGFP (made in Duke’s Viral Vector Core), dextrans (Alexa Fluor 488, D-22910, ThermoFisher) or retrobeads (LumaFluor) were injected into the A11 of adult male birds. Incubation times before perfusion were 4 weeks for viral expression and 4–7 days for tracers. Images were processed with ImageJ to adjust for brightness and contrast. For the analysis of TH⁺ neurons in A11 after 6-OHDA treatments, neurons with diameter greater than 10 μm were counted manually.

Floating section in situ hybridization chain reaction (HCR).

Birds were perfused with 4%PFA/PBS and post-fixed in the same solution overnight and then in 30% sucrose in RNAse-free PBS for 2 overnights at 4°C. Brains were then sectioned at 75 μm and collected into 0.5–1% PFA (4% PFA diluted in RNAse-free PBS). At room temperature, slices were first washed twice in PBS for 3 min, incubated in 5% SDS/PBS for 45 min, rinsed twice with 2x sodium chloride sodium citrate 0.1% Tween 20 (2x SSCT), and put in 2x SSCT for 15 min on a shaker. Then, they were pre-incubated in probe hybridization buffer for 30 min at 37°C, and later hybridized in 2.5 μL probe set/500 μl probe hybridization buffer overnight at the same temperature. The probes were custom made by Molecular Instruments to detect zebra finch isoforms of VGluT2 (NCBI Reference Sequence: NM_001309508.1), TH (XM_002198931.3), and VGAT (XM_002189664.2). The next day, slices were washed four times for 15 min with 500 μL of probe wash buffer at 37°C and twice in 2x SSCT for 5 min at room temperature on a shaker. Then they were incubated in 500 μl of HCR amplification buffer for 30 min at room temperature on a shaker. Last, slices were incubated in a solution containing 300 μL HCR amplification buffer and fluorescent hairpins for the HCR initiator probe for 2 overnights, in the dark at 25°C. On the last day, at room temperature, slices were washed twice with 2x SSCT for 5 min, stained with Neurotrace for 2 hours (1:500, N21479; Invitrogen), rinsed twice with 2x SSCT and mounted on a slide with Fluoromount-G. Hairpins, probe sets and probe hybridization buffer were created by Molecular Instruments. HCR for TH and VGluT2 was performed on sections from 4 birds, HCR for TH and VGAT was performed on sections from 2 birds, and HCR for TH, VGluT2 and VGAT was performed on sections from 1 bird.

Lesion experiments.

A pair of male and female adult zebra finches were housed in an isolated soundproof box. The male and the female were separated by electronic glass (HOHOFILM Electronic PDLC, Smart Film) that was connected to an external switch. When powered off, this glass is opaque, preventing the male and the female from seeing each other; they could hear each other in either condition. In order to record female-directed singing, the experimenter powered the glass on, rendering it transparent and enabling the birds to see each other. The use of electronic glass eliminated the need to handle the birds, increasing the probability that the males would sing to the female. Video recordings started approximately 20 seconds prior to visibility onset. The glass remained transparent for 1–7 minutes. We recorded female-directed singing 2–5 times a day, allowing enough time between exposures to let the main return to baseline levels of arousal (at least 40 minutes between exposures; with a mean interval of 97±54 minutes). Video recordings stopped approximately 20 seconds after visibility offset. Baseline singing rates were recorded for 5–7 days, after which birds were divided into 4 experimental groups, HVC 6-OHDA, HVC sham, A11 6-OHDA, and A11 sham. Videos of the birds were recorded using webcams (Genius WideCam F100). Songs were automatically recorded with Sound Analysis Pro (SAP2011 (Tchernichovski et al., 2000)). Singing rates were calculated manually by counting all female-directed songs produced during the first minute of female presentation and by counting all undirected songs produced during a 4-hour period each day. Female-directed and undirected song rates were calculated for five days prior to surgery. An average singing rate over this baseline period was calculated and then used to normalize singing rates for each day pre- and post-surgery.

Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay.

Birds injected in HVC with 6-OHDA were perfused either 3 or 6 days post-injection, and brain sections were collected as described in the previous sections. Detection of apoptosis was carried out with some modifications to the DeadEndTM Fluororimetric TUNEL System (Promega). Briefly, free-floating sections were fixed in 4% PFA for 15 mins., and washed three times with PBS at room temperature. The sections were digested in a Proteinase K solution (20ug/ml) for 15–20 min., followed by three PBS washes and an additional fixation step. The samples were placed on a shaker at room temperature in an equilibration buffer for 10 min., followed by an incubation step with a fluorescent nucleotide mix and rTdT enzyme in a 37°C humidified chamber for 1 hour. The sections were washed once with 2xSSC for 15 min., and three times with PBS. Then, they were incubated for 90 min. at room temperature with Neurotrace (1:500, N21479; Invitrogen), and washed 3 times with PB before mounting.

General surgery.

Adult male birds were anesthetized with 1%–2% isoflurane inhalation and placed in a stereotaxic device on a heat blanket. Stereotaxic coordinates and multiunit recordings were used to localize target sites for injection and implantation. Stereotaxic coordinates, measured from the bifurcation of the midsagittal sinus, were 0.0 mm rostral, 2.4 mm lateral and 0.5 mm ventral (head angle of 18°) for HVC and 3.4 mm rostral, 0.7 mm lateral and 6.1 mm ventral (head angle of 62°) or 0.35 caudal, 0.7 mm lateral and 5.15 ventral (head angle of 22°) for A11. Reagents (Dextran, Alexa Fluor 488, 594 or 647, Invitorgen; RetroBeads 590, Lumafluor) or viruses were injected using Nanoject-II (Drummond Scientific). Viral injections were performed bilaterally with a volume of 300–1000 nL per hemisphere. Viruses and plasmids were obtained from the Duke Viral Vector Core (Durham, USA) and Addgene.

Injection of 6-OHDA.

Adult male birds received bilateral injections of either 400nl 6-OHDA solution into HVC or 80–100nl 6-OHDA solution into A11 (N=4 for A11 and N=5 HVC). The solution was PBS-based and included 10–60 mM 6-OHDA hydrochloride (Tocris, 2547), 10 μM l-ascorbic acid (Millipore/Sigma, A92902), and 1 μM desipramine hydrochloride (Tocris, 3067), which was included as an inhibitor for noradrenaline and serotonin transporters to protect noradrenergic and serotonergic neuron terminals at the injection site. Control birds received an injection of PBS with 10 μM ascorbic acid and 1 μM desipramine (N=6 for A11 sham group and N=5 for HVC sham group). Drugs were dissolved in PBS immediately before injection in place of equimolar NaCl (working solution: ~300 mOsm, pH 7.3). After injection, birds were returned to their original home cage until approximately 14 days post injection.

Microdialysis infusion of drugs.

Adult male birds were implanted bilaterally with custom-made microdialysis probes and then housed individually in an acoustic sound-proof box. Beginning two days later, saline or dopamine blocker were infused on alternate days into HVC at light onset (D1-type blocker: 5 mM R(+)-SCH-23390 hydrochloride, Millipore/Sigma, D054, D2-type blocker: 5 mM S-(−)-sulpiride, Tocris, 0895). Female-directed songs were collected by introducing a female for one-minute sessions, 4–5 sessions per day, starting an hour after the infusion, up to four hours after the infusion. Song counts were calculated for 7 birds (D1-type blocker) and 2 birds (D1-type and D2-type blockers), in 3 saline days and 2 dopamine blockers days.

Fiber photometry imaging.

Adult male birds were injected with pAAV-hSynapsin1-axon-GCaMP6s-P2A-mRuby3 (axon-target GCaMP6s) bilaterally into A11 or HVC. After waiting a minimum of 3 weeks for viral expression, birds were anesthetized and placed in a stereotaxic apparatus. Bilateral craniotomies were made over HVC and fiber optic ferrules (200 um core, 0.37 NA, Neurophotometrics) were implanted. For all recordings, axon-targeted GCaMP6s was excited at two wavelengths (470nm for imaging of calcium-dependent signals and 415nm for an interleaved isosbestic control to eliminate motion artifacts). An sCMOS camera was used to capture fluorescence (FP3001, Neurophotometrics) at 30 Hz. Synchronized video and sound recordings were acquired using a webcam (Logitech). Data acquisition was performed with custom Bonsai code and data analyzed using custom-written Matlab scripts. In each imaging session, the signal from the isosbestic control channel was first smoothed and then regressed to the signal from the calcium-dependent channel. To calculate the calcium-dependent signal, first the linear model generated from regression was used to generate a predicted control signal. Then the calcium-dependent signal was calculated by subtracting the predicted control signal from the raw calcium-dependent signal (Figure S5F). Audio recordings were filtered using a third-order median filter. Introductory notes, syllables and different types of calls were labeled manually in Matlab. Calcium signals were aligned to the audio recordings and then z-scored to normalized changes in fluorescence across animals. Onset times were defined as the time for which the signal reached 5% from its peak value.

Hybridization chain reaction for identification of Fos expression.

In order to minimize off-target Fos mRNA detection, birds were perfused 30 minutes after cage lights first turned on in the morning. For female-directed singing (n = 4 birds), 4–5 females were presented sequentially to maximize motif amounts. Live video was monitored to ensure no significant undirected (facing away from the female, disengaged) singing occurred during female presentations. For the undirected condition (n = 4 birds), birds were allowed to sing freely in the morning for the 30-minute window. Lights were then turned off and birds were immediately perfused. Brains were processed for HCR as described in a previous section, the custom probes designed by Molecular Instruments targeted Fos (XM_002200534.5), TH and VGAT. For each bird, a z-stack encompassing A11 was collected at 40x power to accurately visualize Fos signal, along with a TH channel. All image processing was done with ImageJ. First, the TH and Fos channels were noise-subtracted (20 μm rolling ball radius), and TH channel was smoothed, automatically thresholded (otsu method), and converted to a binarized mask. The mask was then transferred to the Fos channel, which was then also thresholded. Fos particles within the TH mask were then quantified for intensity (expressed as a fraction of the TH mask).

QUANTIFICATION AND STATISTICAL ANALYSIS

Data reporting.

No statistical methods were used to predetermine sample size. The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.

Song analysis.

Vocalizations of >5 ms were detected by thresholding of the recorded sounds. Pairwise similarity was quantified as percentage of similarity (asymmetrical similarity) between song motif and/or introductory notes of different experimental groups using SAP2011 (Tchernichovski et al., 2000) with default parameters for zebra finches, and reported as similarity score. The percentage of similarity was calculated for representative song motifs and introductory notes randomly chosen from treated birds, before and after 6-OHDA treatment, and averaged across all comparisons. Individual syllables and introductory notes were determined using custom MATLAB graphical user interfaces (Tumer & Brainard, 2007) to further measure their duration and inter-note gap duration.

Video analysis.

Videos were analyzed for 20 seconds prior to female presentation and 30 seconds after female presentation. For each video, the bird’s position and head orientation were measured using either DeepLabCut (3/9 birds, Mathis et al., 2018) or custom MATLAB graphical user interfaces (6/9 birds, M. Ben-Tov) that enable the marking of the bird’s body position and head orientation across video frames. Two experimenters scored the videos (the scores given by the two experimenters were not significantly different (paired t-test, p=0.54). For each video, the position was normalized to the cage size, to allow a comparison between different videos and birds.

Behavioral simulations.

To test whether time spent looking at the female is significantly bigger than chance value, we conducted a simulation in which we used random walk to determine the bird’s position in each time point. Then, for each location, a head angle was drawn from a normal distribution (𝜇=0°, 𝜎=40°) around the previous head angle. For each time point, we measured whether the male’s head angles fell within the range for which the female is visible to him. We determined that the average portion of time the simulated bird will by chance look at the female was 25.28±5.81%. The 31% female-directed gaze time we observed in the A11 6-OHDA group is not significantly different from this random value calculated using this simulation (p = 0.15). We ran the simulation 1000 times; in each iteration we used 1000 steps.

Statistics.

Data are shown as mean ± s.e.m., unless otherwise noted. One-way ANOVAs, two-ways ANOVAs, and their corresponding post-hoc comparisons were performed on Prism (GraphPad). T-tests were performed in Matlab, linear mixed effects models (LME) were performed in R. For LME analysis, birds were used as random effects. Day (as day after the lesion), Time (as pre- and post-treatment), Context (UD and FD), Group (HVC and A11) and Treatment (lesion or Sham) were used as fixed effects. We also included 3-way and 2-way interactions of the fixed effects. To assess the validity of the mixed effects analyses, we performed likelihood ratio tests comparing the model with fixed effects to the null model with only the random effects. We rejected results in which the model including fixed effects did not differ significantly from the null model. To examine the different proportion of labelled neurons in the A11, χ2 tests were performed (Figure 2B, C, E, F). LME analysis was used to assess whether 6-OHDA treatment and context affected the production of songs. Post-hoc Tukey tests were used to compare singing rates between 6-OHDA treated males and sham injected males and between FD motifs and UD motifs (Figure 3L). A paired t-test was used to test the effect of D1R blocker on FD singing compared to singing in saline days (Figure 3N). One-way ANOVAs were performed to examine whether 6-OHDA injections abolished A11 neurons or VTA/SNc neurons. (Figure S2 D, E). One-way repeated measurement ANOVAs were performed to examine the effect of 6-OHDA treatment on the pairwise similarities between song motifs for the HVC groups and the A11 group, followed by post-hoc Tukey test to compare between conditions (Figure S2S). A t-test was used to compare numbers of TUNEL+ cells in 6-OHDA lesions in HVC and control hemispheres (Figure S2U). One-way repeated measurement ANOVA was performed to examine the effect of D1 receptor blocker infusion on the pairwise similarities between song motifs (Figure S2Z). LME analysis was used to assess whether 6-OHDA treatment affected the production of introductory notes and female directed calls. Post-hoc Tukey tests were used to compare between sub-groups (Figure 4B, C). One-way repeated measurement ANOVA was performed to examine the effect of 6-OHDA treatment on the pairwise similarities between introductory notes, and between female-directed calls, for the HVC group (Figure 4G, K). One-way repeated measurement ANOVA was performed to examine the effect of 6-OHDA treatment on the pairwise similarities between song motifs, introductory notes, and female-directed calls, followed by post-hoc Tukey test to compare between conditions (Figure S3C). One-way repeated measurement ANOVA was performed to examine the effect of 6-OHDA treatment amplitude, duration and fundamental frequency of song motifs, introductory notes, and female-directed calls, followed by post-hoc Tukey test to compare between conditions (Figure S3I, D, E). LME analysis was used to assess whether 6-OHDA treatment affected the male’s movement towards the female, time spent looking at the female, changes in vertical posture changes and number of beak wipes. Post-hoc Tukey tests were used to compare between sub-groups (Figure 5C, G, I and H, respectively). A t-test was used to test for a significant effect of 6-OHDA treatment on overall movement of the male (Figure 5D). A mixed-effect analysis was used to test for a significant difference in peak response time in calcium signals between A11 terminals in HVC and HVC local axons, in female-directed and undirected singing (Figure 6K). A t-test was used to test the Onset time relative to first introductory note difference between the A11 axons group and the HVC axons group (Figure 6M). A t-test was used to test the ratio of Fos positive TH positive cells in A11 during female-directed and undirected song (Figure S5I).

Supplementary Material

Video S1. Video S1. Controlled social interactions between a male and female zebra finch. Related to Figure 5.

Side view of a male and a female zebra finch with an electronic glass separating them. When the frame is red, the glass is opaque, and the birds cannot see each other. When the frame turns green., the glass turns clear and the female is visible to the male. The male readily sings to the female as soon as she becomes visible, while also producing other courtship behaviors. On the bottom: a spectrogram of the audio recorded during the interaction.

Video S2. Video S2. Social interaction of a male after 6-OHDA treatment in HVC with a female. Related to Figure 5.

On the bottom: a spectrogram of the audio recorded during the interaction. Vocalizations consist of repeated introductory notes with no motifs.

Video S3. Video S3. Social interaction of a male after 6-OHDA treatment in A11 with a female. Related to Figure 5.

On the bottom: a spectrogram of the audio recorded during the interaction. Sounds consist of cage noises.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-TH	Sigma/EMD Millipore	ab152
Mouse monoclonal anti-TH	Invitrogen	MA1 24654
Mouse monoclonal anti-GFP	Invitrogen	A11120
Rabbit polyclonal anti-mCherry	Abcam	ab167453
Mouse monoclonal anti-synapsin	Synaptic Systems	106 011
Goat anti-rabbit 647	Invitrogen	A-21245
Donkey anti-rabbit 594	Invitrogen	A21207
Goat anti-mouse 488	Invitrogen	A11001
Blocking One Histo	Nacalai tesque	06349–64
Bacterial and virus strains
pAAV-hSynapsin1-axon-GCaMP6s-P2A-mRuby3	Addgene (Broussard et al., 2018)	Addgene_111262
AAV2/9.CAG-scGFP	Duke Viral Vector Core	N/A
Biological samples
N/A
Chemicals, peptides, and recombinant proteins
NeuroTrace 455	Invitrogen/Thermo Fischer Scientific	Cat# N21479
Dextran, Alexa Fluor™ 488	Thermofisher-Invitrogen	D22910
Dextran, Alexa Fluor™ 647	Thermofisher-Invitrogen	D22914
Retrobeads	Lumafluor	Red Retrobeads™ IX (100 μl)
6-OHDA	Santa Cruz Biotechnology	28094–15-7
R(+)-SCH-23390 hydrochloride	Millipore/Sigma	D054
S-(−)-sulpiride	Tocris	0895
Critical commercial assays
TUNEL	Promega	G3250
Deposited Data
Data and code	This paper	https://doi.org/10.7924/r4hd80×9b
Experimental models: Cell lines
N/A
Experimental models: Organisms/strains
Zebra Finch (Taeniopygia guttata)	Duke University Medical Center breeding facility	N/A
Oligonucleotides
HCR probe: VGlut2	Molecular Instruments	NCBI Reference Sequence: NM_001309508.1
HCR probe: VGAT	Molecular Instruments	XM_002189664.2
HCR probe:TH	Molecular Instruments	XM_002198931.3
HCR probe: Fos	Molecular Instruments	XM_002200534.5
Recombinant DNA
N/A
Software and algorithms
MATLAB	MathWorks	RRID:SCR_001622
Bonsai	https://bonsai-rx.org/	RRID: SCR_017218
ImageJ	NIH	RRID:SCR_003070
ZEN	Zeiss	RRID:SCR_013672
Other