• Keine Ergebnisse gefunden

Developmental and Experiencedependent. of a Song-control Brain Region in the Zebra Finch

N/A
N/A
Protected

Academic year: 2022

Aktie "Developmental and Experiencedependent. of a Song-control Brain Region in the Zebra Finch"

Copied!
207
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

ETH Library

Developmental and Experience- dependent Synaptic Remodeling of a Song-control Brain Region in the Zebra Finch

Doctoral Thesis Author(s):

Huang, Ziqiang Publication date:

2017

Permanent link:

https://doi.org/10.3929/ethz-b-000228826 Rights / license:

In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection.

For more information, please consult the Terms of use.

(2)

DISS. ETH NO. 24673

DEVELOPMENTAL AND EXPERIENCE-DEPENDENT

SYNAPTIC REMODELING OF A SONG-CONTROL BRAIN REGION IN THE ZEBRA FINCH

A thesis submitted to attain the degree of DOCTOR OF SCIENCES of ETH ZURICH

(Dr. sc. ETH Zurich)

presented by ZIQIANG HUANG

Dipl. M.Sc., Georg-August-Universität Göttingen

born on 28.04.1986 citizen of

China

accepted on the recommendation of Richard Hahnloser

Sebastian Jessberger Michael Griesser

2017

(3)

2

(4)

3

Abstract

During a developmental critical period, juvenile male zebra finches learn to sing a highly stereotyped song from a tutor bird. Song learning in zebra finches shares many common behavioral, physiological, and neuroanatomical features with other nervous systems that exhibit critical periods. These include development of binocular vision in cat, formation of sensory maps in rodent, and acquisition of language in humans. One important regulatory principle summarized from the various systems is the experience-dependent structural remodeling within the associated brain areas. The brain area HVC, which is the vocal motor cortex analog in the zebra finch brain, plays a crucial role during song learning in juveniles and actively drives production of the learned song in adulthood. Similar to the structural remodeling that has been observed in primary visual cortex (V1) in cat and barrel cortex in rodents, dendritic spines in HVC undergo rapid remodeling in response to critical sensory experience. However, the mechanisms underlying the experience-dependent remodeling at the synaptic level are unclear. In particular, the HVC inhibitory circuit, which is shown to be crucial for song learning, has not been studied anatomically at the synaptic level yet. And the experience-dependent structural changes in HVC has not been carefully decoupled from developmental structural changes. Therefore, I set out to carefully study the structural synapse changes in HVC during song development.

I propose two experimental approaches to investigate the open scientific questions described above. The first experiment was inspired by an in vivo imaging study, which reported initial sensory experience of the tutor song triggered spine stabilization and enhancement overnight in HVC in juvenile zebra finches (Roberts et al. 2010). I further investigated this experience-dependent structural remodeling in HVC in two ways. First, I further manipulated the song sensory experiences in juvenile zebra finches. In different groups of juveniles, I provided short, long, or no tutor exposure, and then harvested their HVCs for neurostructural analysis at the same age. I then examined HVC further with serial-section and block-face electron microscopy, which allowed me to directly and clearly identify, classify, and morphologically reconstruct HVC synapses. The results showed synaptic pruning, and suggested synaptic strengthening in HVC excitatory synapses in a song sensory experience-dependent manner, which was in agreement with previous findings.

Surprisingly, the results also suggested a transient density increase of the inhibitory synapses in HVC, following one day of tutor song sensory exposure. Rapid experience-dependent remodeling of the inhibitory circuit in HVC has not been reported before, but similar has been observed in the mouse somatosensory cortex (Knott et al. 2002). Taken together, these results indicate that tutor song exposure

(5)

4

during song learning has short- and long-term effects on both the excitatory and the inhibitory synaptic networks in HVC.

In the second experiment, I investigated this experience-dependent structural remodeling at different developmental stages. I used two main groups of juvenile male zebra finches. In one group I deprived the juvenile completely from sensory exposure to the tutor song, while in the other group I provided the bird regular and controlled tutor song exposure. I harvested HVC from the birds at different development stages, and examined the synaptic structure in HVC under the electron microscope. I repeatedly observed a peak-decline developmental pattern, in which neurostructural parameters such as brain volume, synapse density, and total synapse number first increase and then decrease. This developmental pattern, which had been observed in different brain regions and species, reflects the overproduction of neural connections in the early stages of postnatal development and structural pruning in the later stages (Herrmann and Bischof 1986; Murphy and Magness 1984). Tutor song exposure lead to relative pruning of excitatory and inhibitory synapses in HVC, in comparison to birds that have not received tutor exposure.

Combined with previous findings, I provide a detailed picture of the developmental structural synaptic remodeling in HVC during song learning. These findings provide clues into how HVC’s synaptic networks might store a memory of tutor song.

(6)

5

Zusammenfassung

Der Zebrafink ist eine Australische Singvogelart bei der die Jungvögel während der sogenannten kritischen Phase einen stereotypischen Gesang von einem männlichen Tutor lernen. Dieses besondere Gesangslernverhalten macht Zebrafinken zu einem guten Modellsystem um vokales Lernen zu untersuchen. Gesangslernen bei jungen Zebrafinkenmännchen teilt viele Charakteristika mit der Entwicklung anderer Lernsystemen mit kritischer Phase, wie der des binokularen Sehens bei Katzen, der Bildung sensorischen Karten bei Nagetieren und des Sprachlernens beim Menschen in Bezug auf Verhalten, Physiologie und Neuroanantomie. Eine wichtige gemeinsame Charakteristik dieser Systeme ist die Erfahrungsabhängigkeit der strukturellen Veränderung in den dazugehörigen Gehirnarealen. Das Gesangsareal HVC im Singvogelgehirn, welches dem Motorkortex von Säugetieren entspricht, spielt eine wichtige Rolle beim Gesangslernen von jungen Zebrafinkenmännchen.

HVC steuert das Singen bei erwachsenen Zebrafinkenmännchen und ähnlich den strukturellen Veränderungen die im V1 von Katzen und im Barrelkortex von Ratten stattfinden, wurde in HVC eine Veränderung der dendritischen Dornfortsätze gefunden, die sich hochdynamisch und schnell ändern, nachdem der Vogel den Tutorgesang zum ersten Mal hört. Jedoch ist nach wie vor unbekannt, wie sich diese strukturellen Veränderungen auf Synapsen auswirken. Speziell die Veränderung der inhibitorischen neuronalen Schaltkreise, für die gezeigt wurde, dass sie die schon gelernten Segmente des Gesangs vor weiteren Veränderungen schützen, wurden auf Synapsenebene noch nicht untersucht. Auch die erfahrungsabhängigen strukturellen Veränderungen in HVC wurden noch nicht unabhängig von entwicklungsstruktureller Reifung mit modernen neuroanatomischen Techniken untersucht. Somit ist in diesem Kontext der Zusammenhang zwischen Gehirnentwicklung und Gesangslernen noch unbekannt.

In dieser Doktorarbeit stelle ich zwei Experimente vor, die auf die Beantwortung der oben beschriebenen Fragestellungen abzielen. Das erste Experiment wurde von einer in-vivo Mikroskopiestudie angeregt. Diese Studie (Roberts et al. 2010) zeigte, dass sich dendritischen Dornfortsätze in HVC über Nacht nach Tutorgesangerfahrung stabilisieren. Wir untersuchten diesen erfahrungsabhängigen Strukturumbau in zweierlei Hinsicht. Wir manipulierten die Tutorgesangserfahrung von jungen Zebrafinkenmännchen, indem wir die Zeit des Tutorkontakts variierten und später im selben Alter der Versuchstiere die neuroanantomische Struktur von HVC untersuchten. Dafür passten wir die elektronenmikroskopischen Bildaufnahme und Analyse an, indem wir Aufnahmen serieller Schnitte und FIB-SEM-Tomographien kombinierten. Diese Kombination gestattete HVC Synapsen eindeutig zu erkennen, zu klassifizieren, morphologisch zu rekonstruieren, und quantitativ zu analysieren, um zwischen den Tieren mit unterschiedlicher Tutorgesangserfahrung zu vergleichen. Wir fanden eine erfahrungsabhängige selektive Synapsenreduktion und -verstärkung, was in Übereinstimmung mit früheren Studien ist. Zu unserer Überraschung fanden wir bereits einen Tag nach der ersten Tutorgesangserfahrung auch einen schnellen Anstieg der Dichte und der

(7)

6

Anzahl (Prozentsatz der gesamten Synapsen) der inhibitorischen Synapsen. Dieser schnelle erfahrungsabhängige Umbau des inhibitorischen Netzwerks wurde im HVC von Vögeln bisher noch nicht beschrieben, jedoch im somatosensorischen Kortex der Maus gefunden. Legt man dieses Ergebnis mit weiteren Studien zu inhibitorischen Netzwerken in HVC und in anderen System mit einer kritischen Phase zusammen, deutet unser Resultat darauf hin, dass das inhibitorische Netzwerk eine wichtige Rolle bei der Regulierung des lokalen plastischen Umbaus beim Gesangslernen spielen könnte.

Im zweiten Experiment untersuchte ich die gesangsabhängigen Umstrukturierungen von HVC, zu unterschiedlichen Zeitpunkten der Gesangsentwicklung. Wir zogen dazu zwei Gruppen von junge Zebrafinkenmännchen auf. Einer Gruppe enthielten wir jegliche sensorische Gesangserfahrung vor, während wir der anderen Gruppe kontrolliert regelmässige Tutorgesangserfahrung gewährten. Wir untersuchten HVC elektronenmikroskopisch zu unterschiedlichen Entwicklungszeitpunkten in beiden Gruppen in Bezug auf den strukturellen Umbau der Synapsen. Wir beobachteten einen allgemeinen Trend zur „Höchstwertabnahme“ bei Gehirnvolumen, Synapsendichte und Synapsenanzahl in ganz HVC. Das heisst, diese Grössen stiegen erst auf einen Höchstwert an und nahmen dann ab. Dieses Phänomen der Höchstwertabnahme wurde schon in unterschiedlichen Gehirnarealen unterschiedlicher Arten gefunden und entspricht einer Überproduktion von synaptischen Verbindungen in den frühen postnatalen Entwicklungsphasen und einer selektiven Synapsenreduktion während späterer Entwicklungsphasen (Herrmann and Bischof 1986; Murphy and Magness 1984). Wir fanden auch eine starke Veränderung des Verhältnisses von exzitatorischen zu inhibitorischen Synapsen während der Gesangsentwicklung. Zusammengenommen mit anderen Studienresultaten und dem Resultat des ersten Experiments, bestärkt das unsere Hypothese, dass der erfahrungsabhängige Umbau von inhibitorischen Synapsen in HVC eine wichtige Rolle bei der Regulierung der Plastizität des lokalen HVC Netzwerkes bei Gesangslernen von jungen Zebrafinken spielt.

(8)

7

(9)

8

Contents

Abstract... 3

Zusammenfassung ... 5

Introduction ... 11

1.1. Vocal learning in the zebra finch ... 11

1.2. The song system in the zebra finch brain ... 13

1.3. The function and structure of the song nucleus HVC ... 16

1.4. Critical period learning-related structural changes in the brain ... 18

1.5. Microscopy techniques in structural studies of the brain ... 21

1.6. Two behavioral experiments ... 23

1.6.1. Experiment I ... 24

1.6.2. Experiment II ... 25

Experiment I: Neurostructural changes in HVC in response to song sensory experience ... 26

2.1. Experimental design ... 26

2.2. Methods ... 28

2.2.1. Song tutoring, recordings and analysis ... 28

2.2.2. Brain tissue collection and histology ... 29

2.2.3. Brain tissue preparation for EM ... 30

2.2.4. ssSEM and FIB-SEM imaging ... 33

2.2.5. Measurements of synapse densities ... 36

2.2.6. Classification and quantification of the synapse subtypes ... 38

2.2.7. Segmentation and geometric measurements of synapses ... 41

2.2.8. Linear regression analyses of the neuroanatomical data... 47

2.3. Results ... 49

2.3.1. Song learning performances in the tutored groups ... 49

(10)

9

2.3.2. Synapse density ... 62

2.3.3. Asymmetric synapse density ... 65

2.3.4. Symmetric synapse density ... 67

2.3.5. Percentage of symmetric synapses ... 69

2.3.6. Synapse size ... 72

2.3.7. Synapse Feret diameter ... 78

2.4. Discussion ... 85

2.4.1. Comparison of ssSEM and FIB-SEM datasets... 85

2.4.2. Synapse density in HVC ... 88

2.4.3. Percentage of symmetric synapses in HVC ... 89

2.4.4. Synapse size and diameter in HVC ... 91

2.4.5. Summary of the results of Experiment I ... 94

Experiment II: Neurostructural changes in HVC during song development ... 97

3.1. Experimental design ... 97

3.2. Methods ... 98

3.2.1. Song tutoring, recordings and analyses ... 99

3.2.2. Brain tissue collection and preparation for LM... 100

3.2.3. Volume measurements of HVC with LM histology ... 101

3.2.4. Brain tissue preparation for EM and ssSEM imaging ... 101

3.2.5. Measurements of synapse densities... 102

3.2.6. Measurement of symmetric synapse percentage ... 103

3.2.7. Estimation of synapse total number ... 104

3.2.8. Linear regression analyses of the neuroanatomical data ... 105

3.3. Results ... 107

3.3.1. Song learning performances in the tutored groups ... 107

3.3.2. Volume of HVC ... 118

3.3.3. Synapse density ... 120

3.3.4. Percentage of symmetric synapses ... 129

3.3.5. Total synapse number ... 132

3.4. Discussion ... 137

3.4.1. Song learning performance (Experiment I and II) ... 138

3.4.2. Volume ... 140

3.4.3. Synapse density and total number ... 142

(11)

10

3.4.4. Percentage of symmetric synapses ... 146

3.4.5. Summary of the results of Experiment II ... 148

Summary and Conclusion ... 153

4.1. Experience-dependent synapse remodeling in HVC during song development ... 153

4.2. Developmental synapse remodeling in HVC and the impact of song-experience on it ... 156

4.3. HVC excitatory synapse changes during song development ... 159

4.4. HVC inhibitory synapse changes during song development ... 162

4.5. Conclusion ... 163

Supplementary Method and Data ... 166

S.1. Tissue physical deformation during EM preparation ... 166

S.2. Post-processing of synapse segmentations in MATLAB ... 171

S.3. Data distribution of the ssSEM and FIB-SEM datasets ... 176

S.4. Measurement of symmetric synapse percentage with ssSEM dataset ... 180

S.5. Measurement of synapse density with FIB-SEM... 183

S.6. An approach to estimate the network E/I balance in HVC with neuroanatomical data ... 188

Bibliography ... 191

(12)

11

Chapter 1 Introduction

In the first chapter, I will explain the motivation for studying vocal learning in general, and the importance of understanding the brain’s structural changes associated with vocal learning. I will summarize the current understanding of vocal learning in zebra finches, and examine in detail HVC (acronym used as a proper name), a song-control brain area that is the region of interest (ROI) in the current study. Inspired by relevant studies on neuroanatomy and learning-associated brain structural changes, I will propose two behavior experiments that aim to address some of the unanswered questions in the neurostructural changes in HVC during song learning. The findings will be presented and discussed in later chapters.

1.1. Vocal learning in the zebra finch

Vocal learning, which is the ability to learn and produce vocalizations through imitation, is a behavior exhibited by human and songbird (Petkov and Jarvis 2012).

Although may not convey information as complex as human speech, birdsong also displays syntax and semantic structures (Fujimoto, Hasegawa, and Watanabe 2011;

Petkov and Jarvis 2012), and shares a lot of behavioral, neural, and genetic similarities (Doupe and Kuhl 1999; Moorman, Mello, and Bolhuis 2011). For example, in both human and songbirds, vocal learning is best performed during a juvenile critical learning period (Hensch 2004), which consists of a sensory learning phase that precedes a motor production phase (Bolhuis, Okanoya, and Scharff 2010).

Thus, birdsong has provided neuroscientists with an excellent model to study vocal learning in the laboratory, and can potentially shed light on human language acquisition.

(13)

12

Figure 1. 1: Zebra finches (Taeniopygia guttata): A juvenile male (~ 30 days post-hatching [dph], left) and an adult male (> 120 dph, right) rest next to each other.

The zebra finch (Taeniopygia guttata), which is one of the most studied songbird species displays unique song behavior (Immelmann 1969; Morris 1954). In this species, only the males sing. To attract a female for mating, adult male sings a single stereotyped song. The song is relatively simple in its syntactical structure, which consists of multiple repetitions of a stereotyped song motif. Each motif contains a sequence of syllables in fixed order. An example of a typical adult zebra finch song is shown in Figure 1. 2.

Figure 1. 2: The sound spectrogram of a sample adult zebra finch song. The song consists of several repetitions of the introductory notes i that are followed by several repetitions of the stereotyped song motif, which is outlined by the red rectangles. The repeating song motif in turn consists of individual syllables A, B, and C.

The zebra finch song can be learned only once during a juvenile critical learning period and is thereafter stable (crystallized) throughout a bird’s life (Lipkind et al.

2013). This critical learning period consists of two overlapping phases. During the initial sensory learning phase (15 to 60 days post-hatching, [dph]), the juvenile listens to the songs produced by a tutor, which is a male zebra finch that sings a

(14)

13

crystallized song, and acquires a sensory template of the tutor song. In the subsequent sensorimotor learning phase (25 to 90 dph), the juvenile starts singing and utilizes auditory feedback of its own vocalization to gradually shape its songs into a good imitation of the memorized tutor song (M. S. Brainard and Doupe 2000).

The song crystallizes approximately at 90 dph and remains stable for the rest of the bird’s life. An intact hearing has been shown to be crucial for both song development and maintenance of the adult song (Nordeen and Nordeen 1992;

Tschida and Mooney 2012). Deafening prevents regular song learning in juveniles and results in gradual song deterioration in adults (Lombardino and Nottebohm 2000). Because of the highly stereotyped nature of learning (one target song, learned once), zebra finches have served for decades as one of the most common animal models in the study of vocal learning and related neural mechanisms (Mello 2014).

As mentioned the vocal learning in both human infant and juvenile zebra finches is a form of critical period learning. The critical period is defined as a postnatal neurodevelopment stage during which neuronal circuits are very sensitive to the environment and shaped by critical sensory experiences (Hensch 2004). If the critical sensory experiences do not occur during the critical period, the neuronal circuit will not develop the structure required for its function, as exemplified by vision (Hubel and Wiesel 1970), sensory map formation (Waite and Taylor 1978), human language acquisition (Neville and Bavelier 2002), and birdsong learning (Beecher and Brenowitz 2005). Song learning in zebra finches shares many common regulatory principles to various well studied critical periods (see Hensch 2004 for review) in other species. For example, the song learning process exhibits well-defined brain plasticity windows, and is regulated by critical sensory experience in addition to age. The learning performance depends greatly on the social environment and motivation of the juvenile birds (Chen, Matheson, and Sakata 2016; Derégnaucourt et al. 2013; Tchernichovski et al. 1999; Tchernichovski and Nottebohm 1998). Moreover there is evidence of structural consolidation in the corresponding brain region that is associated with initial song exposure (Roberts et al. 2010), which is observed in other critical periods as well (Lichtman and Colman 2000; Walsh and Lichtman 2003). However, besides this reported structural consolidation, by far few studies have investigated brain structural remodeling during song development especially from song sensory experience. Such studies are needed not only to fill the gaps in the current knowledge regarding birdsong learning but also to shed light on the general regulatory principles of critical period and experience-dependent synaptic rearrangement.

1.2. The song system in the zebra finch brain

As opposed to the six-layered cortical structure of the mammalian brain, the songbird brain is organized into different functional nuclei (Fernando Nottebohm, Stokes, and Leonard 1976). The nuclei involved in song learning and song production have been intensively studied in recent decades (Akutagawa and Konishi

(15)

14

1998; Aronov, Andalman, and Fee 2008; Bottjer, Miesner, and Arnold 1984; F Nottebohm, Stokes, and Leonard 1976). The neural circuits linking these song-related nuclei relay the sensory input and the motor output of the song. These nuclei and their interconnections are crucial for the learning, production, and memory of birdsong (Scharff and Nottebohm 1991), and together they form the song system (Anton Reiner, David J.Perkel, Claudio V. Mello et al. 2004) (Figure 1.

3).

Figure 1. 3: A schematic of the song system emphasizing HVC and its connections. This parasagittal view of the songbird brain shows the major brain pathways involved in song production and song learning. Adapted from (Mooney 2014).

* Note that in this figure, the HVC is colored as if it has distinctive subregions along rostrocaudal axis, that correspond to the red and blue brain pathways, respectively. This is not the case. For a more accurate review of the HVC local architecture, please see section 1.3.

The song system of zebra finches comprises two major brain pathways that are actively involved in song production and song learning. The song motor pathway (SMP, shown in blue in Figure 1. 3), drives the production of the crystallized song in adult birds, and the anterior forebrain pathway (AFP, shown in red in Figure 1. 3), drives song learning and the production of the subsong (vocal babbling) in juveniles (M. S. Brainard and Doupe 2000).

The SMP starts with nucleus HVC and projects to the robust nucleus of the arcopallium (RA), with some reciprocal connections from the RA back to HVC (Roberts et al. 2008; Yip, Miller-Sims, and Bottjer 2012). RA neurons in turn

(16)

15

project to the dorsal medial nucleus (DM) (not shown in Figure 1. 3) and the tracheosyringeal part of the hypoglossal nucleus (nXIIts) (Wild 1993). The motor neurons in the nXIIts then directly innervate the syringeal muscles of the vocal organ, the syrinx, in order to produce the learned songs in adult male zebra finches (Vicario and Nottebohm 1988). Alternatively, the DM nucleus projects to brainstem respiratory areas, the nucleus paraambiguus (PAm) and nucleus retroambiguus (RAm) (not shown in Figure 1. 3) (Vates, Vicario, and Nottebohm 1997), to coordinate breathing during singing. Single-unit recording of the RA-projecting neurons in HVC, denoted as HVCRA neurons, revealed that the neurons fire sparsely to drive RA activity during song production (Hahnloser, Kozhevnikov, and Fee 2002). In contrast, premotor neurons in RA fire relatively dense and short bursts of spikes, with less precision but still reliably associated with the song features during singing (Yu and Margoliash 1996). The motor neurons in the descending target of RA, the nucleus nXIIts, are somewhat somatotopically organized to project to and control different muscles of the syrinx (Vicario and Nottebohm 1988). Thus, these interconnected brain nuclei, HVC, RA, and nXIIts, structurally and functionally form the neural pathway underlying song production in adult male zebra finches (McCasland 1987).

The other brain pathway, the AFP, also connects HVC to RA through various nuclei.

It starts with another class of HVC projection neurons, the HVCX neurons that project to area X. Unlike the late-developing projections from HVC to RA, the pathway that links HVC to area X is generated early in development (Mooney and Rao 1994). Songbird area X, which is the homolog of mammalian basal ganglia, in turn projects to the dorsal lateral nucleus of the medial thalamus (DLM). From DLM, the AFP continues projecting back to the lateral magnocellular nucleus of the anterior nidopallium (LMAN) in the forebrain, which then projects back to the motor circuitry at RA and merges again (after HVC) with the SMP pathway. In addition, LMAN sends collateral projections back to area X and forms the recurrent connections of the basal ganglia-thalamocortical loop (Doupe et al. 2005; Goldberg and Fee 2011). Lesions in any component of the AFP pathway disrupt song development in juvenile birds, but do not affect the maintenance or production of song in adults (Bottjer et al. 1984; Goldberg and Fee 2011; Scharff and Nottebohm 1991). The AFP actively drives vocal babbling in juvenile birds, but does not require HVC as shown by a lesion experiment (Aronov et al. 2008).

As highlighted in green and white in Figure 1. 3, auditory information reaches the song system through different afferent pathways (Mooney 2014). The auditory information of the tutor song reaches and is stored in the song system, and actively shapes the song learning related neural circuit in juveniles (Keller and Hahnloser 2009; Roberts et al. 2010; Yanagihara and Yazaki-Sugiyama 2016). Tutor song deprivation in juvenile leads to significant long-lasting structural and physiological changes in HVC, even into adulthood, as observed from a closely related songbird species (Peng et al. 2012). In addition to learning, auditory feedback is also shown to be important for song maintenance in adult zebra finches (M. S. Brainard and Doupe 2000; Tschida and Mooney 2012).

(17)

16

Many brain nuclei in the zebra finch song system show song-selective electrophysiological responses to auditory stimuli (Vicario and Yohay 1993), even in the nuclei that are mainly involved in song motor functions, which suggests that these nuclei have mixed sensorimotor characteristics (Doupe and Solis 1997). HVC, as one of such nuclei located at the origin of the SMP and AFP pathways, shows selective response to the bird’s own song (BOS). Intracellular recordings have shown that all three major neuron types of HVC, HVCRA neuron, HVCX neuron, as well as inhibitory interneurons (HVCI neurons), are BOS selective (Rosen and Mooney 2000, 2003). This selectivity most likely arises within HVC because BOS-selective responses are absent or less selective in the known afferents to HVC (Coleman et al. 2007; Doupe and Konishi 1991; Janata and Margoliash 1999).

The zebra finch HVC is important in both song production and song learning, and it plays an active role in the auditory processing of the tutor song during song development. Therefore, I focused on HVC in the present study.

1.3. The function and structure of the song nucleus HVC

Lesion studies has shown juvenile zebra finches cannot develop the song normally without HVC (Bottjer et al. 1984). When HVC is lesioned in adults, the birds immediately lose the ability to produce the learned song and instead generate subsong-like vocalizations (Aronov et al. 2008). Single-unit recordings have revealed that HVCRA neurons produce premotor bursts that are time locked to the song with millisecond precision (Hahnloser et al. 2002). Different HVCRA neurons are active at different time points within the song and on a population level forming a continuous sequence of firing (Lynch et al. 2016; Troyer 2016). Cooling of HVC slows the temporal features of the song across all timescales, which suggests that the local neural architecture within HVC is essential for the generation of these premotor sequences (Long and Fee 2008). In a word, songbird HVC plays crucial role in both song learning and learned song production.

Studies on the connections of HVC identified two major neural outputs through the two classes of projection neurons that mentioned: the HVCRA neurons and the HVCX neurons. Another class of HVC projection neurons (HVCAv), which are sparsely distributed within HVC and project to the avalanche nucleus (Av), a restricted area of the caudal mesopallium, was subsequently discovered (Akutagawa and Konishi 2010). The HVCAv neurons, which transmit motor-related song information to the auditory systems, play an essential role in juvenile song copying and the adaptive modification of the temporal, but not spectral, features of the adult song (Roberts et al. 2017). HVC receives neural input from many areas including:

the medial magnocellular nucleus of anterior nidopallium (mMAN), the forebrain nucleus interface of the nidopallium (NIf), the nucleus uvaeformis (Uva) in the thalamus (Foster and Bottjer 1998), auditory area Field L (Shaevitz and Theunissen 2007), and HVC shelf (Wang, Sokabe, and Sakaguchi 2001). RA (Roberts et al.

(18)

17

2008) and nucleus Av (Akutagawa and Konishi 2010) also reciprocally project to HVC.

From neuroanatomical aspect, HVC is roughly oval-shaped with an elongated anterior-posterior axis. It is located at the posterior part of the caudal nidopallium and immediately ventral to the lateral ventricle in the zebra finch brain (Poirier et al.

2008). HVC steadily grows in volume and neuron number during postnatal development, reaches close to adult size roughly around 40 dph, and remains stable thereafter (Herrmann and Bischof 1986). In adult male zebra finches, HVC contains on average 43k neurons, which mainly consist of the three classes of neurons that already mentioned. The total number of HVC neurons varies greatly (up to 2.5-fold) among individual birds, but the relative proportion of the two types of projection neurons remains stable, with the majority (42%) comprising HVCRA neurons and a relatively low proportion of HVCX neurons (16%) (Ward, Nordeen, and Nordeen 2001). The HVC interneurons generally express the inhibitory neural transmitter gamma-aminobutyric acid (GABA) and contribute to roughly 10% of the total neuronal population in adults HVC (Scotto-Lomassese et al. 2007). HVC interneurons are suggested to comprise many sub-classes based on their calcium-binding protein expression patterns (Wild et al. 2005). The inhibitory synaptic transmission of HVC is believed to be mostly (if not all) local, without any confirmed inhibitory input to HVC thus far (Pinaud and Mello 2007;

Scotto-Lomassese et al. 2007; Vallentin et al. 2016). All three type of HVC neurons are mainly produced in the embryo or during the first few days after hatching (15 dph) (Alvarez-Buylla, Theelen, and Nottebohm 1988), and they increase in number steadily till 60 dph (Bottjer, Miesner, and Arnold 1986; Cummings 2016; Nordeen and Nordeen 1988). No adult-neurogenesis can be found in most types of HVC neurons (Scotto-Lomassese et al. 2007), except for HVCRA neurons that have been shown to be double in density in zebra finches between 90 dph and 9 years old (Walton, Pariser, and Nottebohm 2012).

The local neural architecture of HVC was first thought to be nontopographically organized, as tracer injected to random subregions of HVC resulted in an even distribution of both the retrograde and anterograde labeling in its upstream and downstream nuclei, in both juvenile (20-23 dph) and adult male zebra finches (Foster and Bottjer 1998). Complete transection of the medial and lateral portions of HVC in both hemispheres had little effect on the production of the learned song in adult zebra finches (Poole, Markowitz, and Gardner 2012). However, subsequent studies have suggested that synaptic connectivity within HVC is organized preferentially along the rostrocaudal axis (Stauffer et al. 2012). A more elaborated paired-tracer study further revealed that each afferent nucleus (mMAN, NIf, Uva, Av and RA) has a unique topographical pattern of input to HVC (Elliott et al. 2017), indicating the neural afferents to HVC might be organized non-homogeneously, especially along the medial-lateral axis. Within HVC, the local neuronal population and circuits are suggested to be arranged in functional clusters spanning approximately 100 µm (Kosche, Vallentin, and Long 2015; Markowitz et al. 2015;

Wild et al. 2005). Regarding the local wiring diagram, pair-wise recordings of HVC

(19)

18

neurons in adult zebra finches have revealed that all types of paired connections among the three classes of HVC neurons may exist but with different probabilities (Mooney and Prather 2005). The authors found a major HVC synaptic connection consisted of disynaptic inhibition from the HVCRA neuron to the HVCX neuron through inhibitory interneurons (HVCI neurons). The HVC local circuit was further investigated connectomically, with a focus on the HVCRA premotor neurons and their connections (Kornfeld et al. 2017). By retrogradely labelling HVCRA neurons and examining of the same local structure with both light microscopy (LM) and electron microscopy (EM), the authors found that HVCRA neurons exclusively targeted inhibitory interneurons close to their soma, which was consistent with electrophysiology. Interestingly, far from the soma, the HVCRA neurons mostly targeted other excitatory neurons, with half the postsynaptic population confirmed as other HVCRA neurons. Based on these two observations, the authors hypothesized that the neural sequences underlying the zebra finch song production are generated by interconnected HVCRA neurons that are embedded within local inhibitory networks in HVC. In agreement with this hypothesis, during song learning in juvenile zebra finches, the local inhibitory circuits in HVC has been shown to act as a global mask of activity, to protect specifically the learned portion of the song (Vallentin et al. 2016).

In summary, HVC is critically involved in song learning in zebra finches. The neural connections of HVC with other song nuclei and its local neural architecture mature together with song development. Significant structural remodeling within HVC is expected during song development to eventually form an electrophysiologically and anatomically well-organized neuronal network that drives learned song production. Moreover, the song learning experience might actively shape the local neural circuit in HVC and thus influence both the short- and long-term structural development in HVC (Peng et al. 2012; Roberts et al. 2010).

Currently no study has systematically traced this potential structural remodeling occurs within HVC during song development, and how learning experiences influence this structural development. However the experience-dependent structural remodeling underlying brain functional changes has been extensively studied in other animals that also exhibit critical periods. In the next section, I will summarize the brain structural changes that are associated with functional changes, with an emphasis on the brain structural remodeling during critical period and related findings in the songbird HVC.

1.4. Critical period learning-related structural changes in the brain

Structural changes in the brain have been observed to be tightly associated with brain functional changes, such as learning (Draganski et al. 2006), memory formation (Bailey and Kandel 1993), and pathological changes (Grutzendler and

(20)

19

Gan 2006). In particular, during early postnatal development, structural changes often coincide with the emergence of brain functions such as binocular vision development (Morishita and Hensch 2008), sensory map formation (Lendvai et al.

2000), and birdsong learning (Herrmann and Bischof 1986; Peng et al. 2012). In addition, structural changes in the brain correlate with the acquisition of new behaviors in both juvenile and adult animals, such as the learning of a new motor task (Hayashi-Takagi et al. 2015), fear conditioning (Lai, Franke, and Gan 2012), and birdsong learning (Roberts et al. 2010). A direct link between newly acquired behavior and the coordinated formation of synaptic structures has also been observed in vivo (Fu et al. 2012).

To characterize the brain structural changes during a critical period is of particular interests to neuroscientists, since brain structure is not merely shaped by the regular developmental processes, but also influenced by external experience (Hensch 2005).

In the primary visual cortex of cats and rodents, normal binocular vision can only be established during a short postnatal period. The permanent loss of normal binocular vision and the anatomical remodeling within the visual cortex following monocular deprivation (MD) are classic examples of critical period development (Morishita and Hensch 2008). During the normal maturation of visual cortex in cat, synapse density rapidly increases (up to 70 days postnatal) and then decreases significantly toward the end of the critical period. Excitatory and inhibitory synapses exhibit different developmental patterns, in that the former increases at a declining rate to a stable level at about 70 days and the latter increases approximately linearly to reach a near adult value at around 110 days (Winfield 1981). Such difference would result in an increase in the level of inhibition in the visual cortex near the end of the critical period, which has been demonstrated as a crucial regulating factor of brain plasticity changes (Chen et al. 2011; Fagiolini and Hensch 2000; Iwai et al. 2003).

With MD, the spine density in V1 rapidly decreases, which accurately reflects the competitive interactions between the neural inputs from the two eyes because it does not occur when both eyes are sutured (Majewska and Sur 2003). Spine density largely recovers after prolonged MD, indicating that this competing-input-triggered pruning is mostly transient (Mataga, Mizuguchi, and Hensch 2004).

Similarly, in the rodent barrel cortex, whisker deprivation during the critical period can affect the normal sensory map formation, and produces profoundly long-lasting abnormal receptive fields (Fox 1992; Stern, Maravall, and Svoboda 2001). The cortical circuits of barrel cortex are highly dynamic and regulated by sensory experience during both development and adulthood. During development, whisker sensory deprivation reduces spine protrusive motility (not to be confused with spine turnover) in the deprived region of the barrel cortex (Lendvai et al. 2000). In the adult, the dendritic spines of the barrel cortex are still highly dynamic and can be stabilized by novel sensory experiences (Holtmaat et al. 2006). The sprouting and retraction of these dendritic spines are tightly associated with synapse formation and elimination, which further suggests an adaptive synaptic remodeling mechanism underlying the sensory experience-dependent brain plasticity changes (Trachtenberg et al. 2002). Altering sensory experience by trimming whiskers in adult animals

(21)

20

induces targeted rewiring of the local excitatory connections (Cheetham et al. 2008).

Persistent whisker stimulation resulted in a significant increase in the density of the inhibitory synapses in the corresponding cortical barrel, and persists after 4 days (Knott et al. 2002).

Not surprisingly, trends of similar structural remodeling are also observed in HVC of songbirds. In general, the neurostructural features, such as volume, neuron number, neuron size, and synapse density increase steadily in HVC between 20 and 90 dph, an age range that coincides with the behavioral development of birdsong (Bottjer, Glaessner, and Arnold 1985; Herrmann and Bischof 1986; Nordeen and Nordeen 1988). Sensory deprivation of tutor song by means of song-isolation or deafening lead to a significant increase in HVC synapse density in juvenile Bengalese finches (Peng et al. 2012). In this study, alternatively, the lower synapse count in the normal reared (song exposed) birds could be explained by experience-dependent synapse pruning. Manipulation of inhibitory synaptic transmission in HVC during the song learning period prematurely closed the critical period plasticity window (Yazaki-Sugiyama et al. 2007, 2009). The experience-dependent synapse pruning, and brain plasticity changes regulated by inhibitory synaptic transmission, are also observed in the visual and somatosensory cortex during critical period development, which represents a common phenomenon across many different critical period learning systems (Hensch 2004). The song experience dependent structural remodeling of HVC was also confirmed in a two-photon in-vivo imaging study, which revealed that enhanced spine dynamics in HVC correlated with behavioral song learning and that the initial exposure to the tutor song was the critical instructive experience that rapidly stabilized and strengthened the dynamic spines in HVC (Roberts et al. 2010).

The two-photon microscopy-based techniques are limited by their spatial resolution and cannot resolve the HVC synapse changes during song development. Subsequent EM examination revealed the dendritic spines previously defined at the fluorescent microscopy level do not always exhibit synapses (Blazquez-Llorca et al. 2015), which makes EM identification almost indispensable for synapse centered structural studies. Besides, the structural remodeling of the HVC inhibitory circuits are not accessible with two-photon microscopy, since a great deal of the inhibitory synapses are aspinous (Mooney 2000). Nevertheless the HVC inhibitory circuit is actively involved in the song learning process, as demonstrated in electrophysiology (Vallentin et al. 2016) and pharmacology (Yazaki-Sugiyama et al. 2009) studies.

These facts suggest that EM-based studies are essential to fully explore the developmental and experience-dependent structural synapse remodeling in HVC.

In the next section, I will elaborate on the current approaches used in neuroanatomy research, including different microscopy techniques and their applications. At the end, I will provide the rationale of the techniques of choice, justified by the specific biological question of the current study.

(22)

21

1.5. Microscopy techniques in structural studies of the brain

Both LM and EM techniques have been applied to visualize the fine structure of the brain. In the history of neuroanatomy, the most remarkable research was done by Spanish neuroscientist Santiago Ramón y Cajal in the nineteenth century. By applying Golgi’s silver staining method to stain neurons in the central nervous system, he created an extensive catalogue of detailed hand drawing of the dark-stained nerve cells under LM that are still being used in neuroscience textbooks today (Kandel, Schwartz, and Jessell 2013). His work expanded the understanding of the nervous system structure, laid the foundation of modern neuroscience (Finger 2010), and recognized by the Nobel Prize in 1906 (Ramon y Cajal 1906). The silver staining technique and LM-based neuroanatomical techniques finally led to the development of Golgi/EM techniques, whereby individual Golgi-impregnated neurons was first characterized by LM and then thin-sectioned for detailed analyses with EM (Peters 2007). While LM-based studies are designed to examine only single nerve cells, or specific parts of nerve cells, the EM-based technique allows neuroanatomist to study profiles of all parts of neurons and synapses between the cells. EM-based studies enabled or tremendously advanced several subfields of neuroscience research such as neural circuit mapping (Briggman and Denk 2006;

Helmstaedter et al. 2013), neuroanatomical tract tracing (Köbbert et al. 2000;

Zaborszky, Wouterlood, and Lanciego 2006), cellular or membrane proteins localization (Baumert et al. 1989; Gerfen and Sawchenko 2016), and fine morphological characterization of the ultra-structure of the nervous system (Gray 1959). It is worth noting that, in the latter study, Gray found at least two kinds of synapses in the cerebral cortex, distinguished by their ultra-structural morphologies (see Figure 2. 6, inset): the Gray type-I synapses and Gray type-II synapses. The clearly different appearance between these two types have been further confirmed in other studies and other systems, and their distinctive functions were characterized:

the Gray type I, or the asymmetric synapses are excitatory, and the Gray type II, or the symmetric synapses are inhibitory (Gray 1969; Klemann and Roubos 2011).

The LM techniques in neurostructural researches have been improved tremendously in the last century. The ever-increasing use of fluorescence-based microscopy have enabled the visualization of specifically targeted (fluorescently marked) neuron groups as well as neuronal processes such as axons and dendrites (Fuxe et al. 1970).

Particularly, with the two-photon laser scanning fluorescent microscopy technique, neuroscientists can monitor neural spine dynamics that are several hundred microns deep inside the living brain over several hours to several months (Denk, Strickler, and Webb 1990; Helmchen and Denk 2005). However, to date, LM-based techniques still cannot unambiguously resolve the fine structure of the nervous system such as synapses. For example, the two-photon microscopy technique, without additional synaptic markers, can only detect dendritic spines that are associated with putative synapses. The solid identification of putative synapses on the two-photon microscopy imaged spines or between two targeted neurons, is always facilitated with subsequent EM examination on the same brain tissue

(23)

22

(Blazquez-Llorca et al. 2015). Besides, spine growth has been observed to precedes synapse formation, which means that the growing spines observed with two-photon microscope are not necessarily associated with synapses but rather predicating their formation (Knott et al. 2006). When combined with synaptic markers, it is still fairly difficult to provide a precise quantification of synapses in a given brain volume with only the LM-based techniques, because of prevalence of multiple synapses on the same boutons (termed as multiple synaptic boutons, MSB) in the brain (Villa et al.

2016). Moreover, symmetric (inhibitory) synapses tend to form axosomatic synapses (Beaulieu and Colonnier 1985; Colonnier 1968), and these synapses are not associated with a spine structure thus systematically ignored by the two-photon microscopy technique. To unambiguously identify, quantify and analyze neural ultrastructural features in dense neuropil, a spatial resolution up to several nanometers is required, and thus EM is the method of choice.

Volume EM techniques such as serial-section transmission or scanning EM (ssTEM or ssSEM) (Harris et al. 2006), serial block-face scanning EM (SBFSEM) (Denk and Horstmann 2004), and focused ion beam scanning EM (FIB-SEM) (Knott et al.

2008) had allowed neuroscientist to reconstruct the three-dimensional (3D) structure of the brain tissue in great details (Helmstaedter, Briggman, and Denk 2008).

The lateral spatial resolution of the volume EM techniques are normally restricted by the quality of the preparation process of the brain tissue. The resolution can easily reach down to 5 nm, which is sufficient to clearly resolve the fine structures of brain tissue such as tiny circle-shaped synaptic vesicles and apposing pre- and post-synaptic membranes separated by the synaptic cleft (Griffiths 1993). However, when it comes to the axial resolution, different volume EM techniques have different limitations. In the serial section based EM techniques (ssTEM and ssSEM), axial resolution is determined by the thickness of the ultrathin section acquired from the ultramicrotomy process, and can reaches down to 40 nm (for experienced EM technician, in my hands it can reaches down to 70 nm). In the SBFSEM technique where the ultramicrotomy is performed with a diamond knife, which is installed inside the microscope chamber, simultaneously with the EM imaging process, the axial resolution can reach down to 25 nm (Helmstaedter et al. 2008). In FIB-SEM imaging of brain tissue, thin tissue sections in the range of several nanometer thickness are repeatedly removed and discarded from the brain block surface automatically using a focused gallium ion beam. Thus, the axial resolution of FIB-SEM is determined by the thickness of the ion-beam milling of the brain tissue, which can be controlled by adjusting the beam current and can reach down to 3 nm (Wei et al. 2012). Compared with the aforementioned other volume EM techniques, only FIB-SEM is capable of generating high-resolution isotropic EM datasets, with which a saturated reconstruction of local brain circuit with every single synapse inside can be performed.

The good spatial resolution in both the lateral and axial axes of the acquired FIB-SEM dataset also makes it possible to perform automatic neural tracing, synapse identification, synapse segmentation, and even brain connectome mapping with computer vision algorithms (Becker et al. 2013; Hayworth 2012; Kreshuk et al.

(24)

23

2011). Due to these advantages, FIB-SEM is a suitable tool for precise measurement of the synapse number, size, and morphology, regardless of the synapse type, in a given local volume of brain tissue. However, the FIB-SEM technique is currently limited by its small dimension of imaging (in the lateral plane, ~ 40 µm) (Knott et al.

2008). Serial section based EM techniques are still the methods of choice for estimating the number or the density of synapses in a given volume of brain in the range of cubic micrometers to millimeters (Reichmann et al. 2015)

As mentioned, the zebra finch HVC is a roughly flat oval shaped brain nucleus that is located right below the ventricle in the posterior part of the caudal nidopallium (Poirier et al. 2008). The HVC of the male adult zebra finch is around 0.55 ± 0.01 mm3 (Airey et al. 2000), with an anterior-posterior length of roughly 1.5 mm and medial-lateral and caudal-ventral lengths of approximately 0.5 to 1 mm (Nixdorf-Bergweiler and Bischof 2007). To representatively quantify the structural synapse changes in HVC associated with song development, I chose to use both the ssSEM and FIB-SEM techniques. I located both the ssSEM and the FIB-SEM imaging sites to the neuropil region near the center of HVC in every brain sample (Figure 2. 3, Figure 2. 4). To estimate the synapse density of HVC, I used ssSEM to image a relatively large 2D area in the para-sagittal plane near the center of HVC on several serial sections of 70 nm thickness. I used FIB-SEM to image a smaller volume in the neuropil region near the center of HVC, to precisely identify, classify, and 3D-reconstructed all synapses contained inside. Detailed descriptions of the ssSEM and FIB-SEM procedures are provided in section 2.2.4. Comparison and evaluation of the use of these two techniques are provided in the discussion sections 2.4.1, and supplementary method and data sections S.4 and S.5.

1.6. Two behavioral experiments

I focused on examining the structural remodeling in HVC associated with song sensory experience and age in the current PhD project. Specifically, I aim to study the structural synapse remodeling in HVC under different learning conditions (i.e.

different level of tutor song exposure) and track this structural remodeling during song development.

On the basis of the previous findings on structural remodeling during the critical period, especially those in the local networks of the songbird HVC, and the changes in spine dynamics during song learning in zebra finches, I conducted two experiments. The first experiment was designed to study the structural synapse remodeling that occurred in HVC in juvenile zebra finches in response to sensory exposure to the tutor song. And, the second experiment was designed to further study and decouple the developmental and song experience-dependent structural synapse remodeling in HVC of juvenile zebra finches learning song.

Considering the findings of the common phenomenon of structural remodeling in response to different types of learning during critical periods, I hypothesize tutor

(25)

24

song experience would trigger local structural remodeling in HVC, such as the formation or removal of synapses, changes in the structural balance of excitatory and inhibitory synapses, and/or changes in synaptic morphology. Moreover, I expect that the excitatory and inhibitory synapses in HVC would change in number and proportion significantly during song behavior development, and would be influenced by song sensory experience of the bird.

1.6.1. Experiment I

The primary goal of this experiment is to quantify the synapse density and the percentage of synapse subtypes (symmetric synapses for example) in HVC of juveniles subjected to different tutoring paradigms. I control the experimental variables so that different groups in Experiment I differ only in their sensory experience of the tutor song exposure. Therefore, I chose to harvest HVC samples from birds of all groups at the same development stage during the song learning period of zebra finches. As described in previous sections, the synapses in the corresponding brain region could react quickly and long-lastingly to critical sensory experiences. Thus, it was interesting for me to learn about how the synapses in HVC react to the tutor song exposure, in short term, and in long term. Whether the synapses in HVC of song-tutored birds differ from those of song-isolated birds, at the same developmental stage? Given the potential different roles during song learning, I wanted to learn about how the excitatory and inhibitory synapses in HVC react to such critical experiences. I proposed the following specific questions to be addressed by Experiment I:

1. Does the synapse density in HVC change as a result of sensory exposure to tutor song? And if so, how does the synapse density change in the short term (acute effect), and in the long term (chronic effect)? Do the excitatory and inhibitory synapses in HVC follow the same structural dynamics upon tutor song exposure?

2. Does the percentage of inhibitory synapses in HVC change as a result of sensory exposure to tutor song? And if so, how does the percentage of the inhibitory synapses react in short-term and long-term to the sensory exposure to the tutor song?

3. Does the size of the synapses in HVC change as a result of sensory exposure to tutor song? And how does the synapse change in size upon short-term and long-term sensory exposure to the tutor song?

(26)

25

1.6.2. Experiment II

Experiment II was proposed as a follow-up experiment to Experiment I. HVC sample were collected from birds of all groups in Experiment I at the same age to eliminate developmental differences in the result. In experiment II, I further investigated both the developmental and experience-dependent structural remodeling in HVC of juvenile zebra finches. The primary goal of this experiment was to quantify the structural synaptic differences in HVC of juvenile birds subjected to different development stages (i.e. 30 dph, 60 dph, and 90 dph), with different tutor song experience (i.e. song-tutored or song-isolated). In the findings summarized in section 1.4, there were quite a number of experience-dependent structural synapse changes inside the adult brain, regulated similarly as in the juvenile brain. This made it more important to decouple the developmental structural changes from the experience-dependent structural changes of the brain, since the latter one seems persist into adulthood and could be governed by more general regulatory principles. Therefore, I wanted to take advantage of the well-defined and highly stereotyped song learning behavior in juvenile zebra finch as a model, to separate and investigate both the developmental and experience-dependent structural synapse remodeling. Specifically, with Experiment II I wanted to address the following questions:

1. Does the volume of HVC, synapse density, and synapse number in HVC change as a result of development while deprived from the critical sensory experience (song-isolated birds), and does it change as a result of normal song development (song-tutored birds)? What would be the difference resulted from song experience at given developmental stage (60 or 90 dph)?

2. Do the excitatory and inhibitory synapses grow simultaneously or follow different developmental patterns? Does the percentage of inhibitory synapses in HVC change during song development, and how would the song experience impact such changes?

(27)

26

Chapter 2

Experiment I: Neurostructural changes in HVC in response to song sensory experience

2.1. Experimental design

Experiment I was designed to investigate whether sensory experience of the tutor song alters HVC neurocircuitry.

Three experimental groups were used in the study (Figure 2. 1).

Figure 2. 1: Timelines of the experimental groups used in Experiment I. For all groups, the experiments began at 35 dph and continued until 59 dph when all the birds were sacrificed to collect HVC tissue samples. The tutoring of the SHORT and LONG groups started at 58 and 35 dph respectively. Each group consisted of 4 birds.

(28)

27

This experiment (as well as Experiment II) was conducted in accordance with the Veterinary Office of the Canton of Zürich (Veterinäramt Kanton Zürich, Zürich, Switzerland).

The common treatments of the three groups were as follows:

All birds used, including juveniles (<90 days old; n = 12), female companion (adult female) and tutors (adult male who sang a crystallized zebra finch song), were raised in a local animal facility (Laboratory Animal Services Center, University of Zürich, Zürich, Switzerland). The juvenile birds were isolated from their fathers beginning on 15 dph and fed only by their mothers until 35 dph. Between 25 and 35 dph, sexing was performed by external morphology and genotyping. At 35 dph, selected birds were separated from siblings and proceeded to experimental treatments.

Siblings from the same nest never ended in the same treatment group to minimize the influence of genetic background on the final results.

Each experimental bird resided in two cages with inner sizes of 39 × 23 × 39 (length

× width × height in cm) joined together by the doors, and thus provided a movement range of approximately 39 × 46 × 39-cm. To isolate the bird from external sound sources, the cages was placed inside a soundproof box with inner size of 50 × 60 × 50-cm.

At 59 dph, the juvenile birds of all groups were sacrificed in the morning and their brains were extracted. Brain samples were then treated for histology and EM.

Birds were provided food and water ad libitum, and a 14-hour/10-hour light/dark cycle.

The different treatments among the three groups were as follows:

Fully isolated group (ISO): From 35 until 59 dph, the juveniles were housed with a female companion. Female zebra finches do not sing, thus the juvenile bird was deprived from sensory exposure to any zebra finch song. At 59 dph, the juveniles were sacrificed.

Short tutored group (SHORT): From 35 until 58 dph, the juveniles were housed with a female companion. At 58 dph, the female was replaced by a tutor. Tutor introduction took place in the morning 30 – 60 minutes after the lights turned on to allow for the juvenile to fully wake up. At 59 dph, the juveniles were sacrificed exactly 24 hours after introduction of the tutor.

Long tutored group (LONG): At 35 dph, the juveniles were housed together with a tutor. At 59 dph, after 24 days of tutor exposure, the juveniles were sacrificed.

The singing of the tutors were monitored daily. The experiment would be stopped and discarded from the result if the tutor failed to sing or sang too little (less than 20 song motifs in a day) to the juvenile.

(29)

28

Experiment I was such designed that only the lengths of the sensory exposures to the tutor song were different among different groups. Thus, this experiment aimed to answer the question of how synaptic parameters change in HVC because of sensory exposure to the tutor song. Comparisons of the ISO and SHORT groups will reveal potential short-term effects of sensory exposure to the tutor song. Comparisons of the ISO and LONG groups will reveal potential long-term effects of sensory exposure to the tutor song. And comparisons of the SHORT and LONG groups will reveal potential differences between short- and long-term sensory exposures to the tutor song.

2.2. Methods

The collected samples of HVC were studied with EM techniques and stereologically quantified. The neurostructural data were analyzed with statistical methods.

2.2.1. Song tutoring, recordings and analysis

Vocalization of the birds were recorded with a microphone (PRO 42; Miniature Cardioid Condenser Boundary Microphone, Audio-Technica Deutschland, Germany) placed inside the recording chamber. The sound signals were processed with a real-time song detection and recording system implemented in LabVIEW software (National Instruments, Austin, TX, USA; Appendix A, Herbst 2013) at a sampling rate of 32 kHz with 16-bit precision. The recorded songs were visualized as sound spectrograms with MATLAB (The MathWorks, Inc., Natick, MA, USA; Figure 2. 2).

The song motifs were extracted as .wav (waveform) audio files with custom MATLAB scripts, in order to perform the song analysis.

Figure 2. 2: Representative spectrograms of a tutor song and a juvenile song (at 59 dph,

(30)

29

after 24 days of tutoring with the song showed above). The color denotes the amplitude at a given frequency with low to high value indicated with blue to red, respectively. The syllables of the tutor song are labeled A, B, and C above the tutor song spectrogram, with i indicating the introductory note. The corresponding syllable structures in the juvenile song are labeled a, b, and c.

The similarities between the juvenile song and the tutor song were analyzed to assess song learning performance at the end of the experiments. 20 juvenile songs and 20 tutor songs were randomly selected from the last day of the recording. The following precautions were taken during song selection to minimize subjective biases: the selections were restricted to only clear recordings of either the juvenile or the tutor song without background noise or vocalization of the other bird; attempts were made to include the most representative motifs of the tutor and juvenile birds, or all identifiable syllables of the juvenile birds if motif structure is not clearly presented; proper selections of juvenile songs that had not yet developed clear motif or syllable structures were determined by consulting local and external experts (Dr.

Alessandro Canopoli, Birdsong group, Institute of Neuroinformatics, ETH Zürich;

Prof. Dr. Sharon M H Gobes, Assistant Professor of Neuroscience, Wellesley College).

Pairwise comparisons of the selected juvenile and tutor songs were performed with Sound Analysis Pro software (SAP 2011, Tchernichovski et al. 2000) in batch comparison mode. The computed similarity scores were exported and analyzed in MATLAB.

The distributions of the similarity scores for each bird are shown with histograms in Figure 2. 11 - Figure 2. 18. The histograms were differently skewed across birds, and did not all followed a normal distribution. Thus I took the median value of the 400 scores to represent each bird. I further calculated the median absolute deviation (MAD) to represent the variability within the data around median of each bird. For

𝑛

similarity scores measured in bird (

𝑋

1

, 𝑋

2

, … , 𝑋

𝑛

)

, the MAD is defined as the median of the absolute deviation from the population median, as given by:

Equation 2. 1: 𝑀𝐴𝐷𝑋= 𝑚𝑒𝑑𝑖𝑎𝑛 (|𝑋𝑖− 𝑚𝑒𝑑𝑖𝑎𝑛(𝑋)|)

The group mean similarity score was calculated by averaging the similarity scores across all of the birds belonging to the same group, and intra-group variability was represented with the standard error of the mean (SE) (see Table 2. 1 and Figure 2.

19).

2.2.2. Brain tissue collection and histology

(31)

30

At 59 dph, the birds were sacrificed in the morning to collect the HVC samples. The birds were euthanized with intramuscular injection with an overdose (normally 50 µL is quite enough for adult zebra finch) of a pentobarbital-sodium solution (Streuli Pharma AG, Uznach, Switzerland) in the pectoral muscles. When pain reflexes were absent, the wings and legs were fixed onto a polystyrene board, and the thorax was cut open with surgical scissors. Upon exposing the heart in the pericardium area, a small opening was cut into the right atrium with fine surgical scissors, and 20 µL of a heparin-sodium solution (B. Braun Medical AG, Sempach, Switzerland) was injected into the left ventricle to prevent blood coagulation. A butterfly needle (Venofix®, B. Braun Medical AG) that was connected to the tubing system (Materialdienst, University of Zürich, Swizterland) containing the perfusion solutions were then inserted into the left ventricle. The perfusion solutions were kept at body temperature (40°C ~ 50°C, Burness et al., 2010; Griffiths 1993) to minimize temperature-induced tissue shrinkage. First, 50 ml of 0.9% sodium chloride, to wash out the blood, followed by 500 ml of the fixative, which consisted of 4%

paraformaldehyde and 0.1% glutaraldehyde that were dissolved in 0.1 M phosphate buffer (PB, pH 7.4) to hydrophobically cross-link the tissue proteins. This procedure was adapted from (Knott, Rosset, and Cantoni 2011) with a slightly reduced glutaraldehyde concentration in order to allow for ideal brain tissue conditions for consequent EM preparation and imaging. The speed of the perfusion was controlled so that the fixative flowed through the bird for at least 15 min.

The brain was then carefully dissected out of the skull with fine forceps (Dumont Tweezers, Electron Microscopy Science, Switzerland). The dura was removed from the brain surface during the brain dissection. Depending on the quality of the perfusion, the brain was either wet sectioned immediately or in the next day after postfixation in the same fixative solution (4% paraformaldehyde and 0.1%

glutaraldehyde in 0.1 M PB) overnight at 4°C. Prior to wet sectioning, the brain was briefly washed in 0.1 M PB and then cut along the midline to separate the hemispheres. The two hemispheres were then glued with the medial surface facing down onto a metal holder and embedded in 3% agar at 40°C, to provide stability during sectioning. Consecutive 100-µm-thick parasagittal sections were then cut with a vibrating microtome (Microm HM 650 V; Thermo Fisher Scientific Inc., Waltham, MA, USA) in a 0.1 M PB bath (CU 65, Thermo Fisher Scientific Inc) at 4°C. Wet brain sections were collected in their cutting order (from lateral to medial) and stored in 0.1 M PB. The sections containing HVC were then imaged in bright-field mode with a wide-field light microscope (BX61, Olympus Schweiz AG, Volketswil, Switzerland) at two different magnifications (1.25X and 4X). The LM images of the wet brain sections allowed for the identification of the HVC region in each of the EM preparation and imaging steps (see Figure 2. 3).

2.2.3. Brain tissue preparation for EM

Referenzen

ÄHNLICHE DOKUMENTE

occasional textual references, musical quotations, stylistic imitations, biographical relationships of Israeli art music composers to the folk or popular repertoire and vice

The microbial composition of samples collected from the same individual (Defined by the Host ID) at different time points were on average more similar than those of the

When such structurally distinct ways of meaning-making that change in an individual’s lifetime are regarded to present a hierarchical order, we speak of religious

The state-nature relations in a developmental state: The South Korean water resource policy The East Asian ‘developmental state thesis’ hereafter, DST, which has been offered by

Projections of the different parts of the rotundus are similar in our results and those obtained by other authors~: medial regions of the rotundus project to

and Sossinka, R., Influence of external stimulation on the quality of the song of the zebra finch (Taeniopygia guttata castanotis Gould), Z.. 4 Immelmann, K.,

Thus, myelination might be a crucial developmental step, as it determines the time when the neuronal substrate is no longer plastic (but see Jacobson, '80). As could be

There does seem to have been a move toward trendier contemporary poets in Athens during the last quarter of the fifth century bce, but to see this phenomenon in terms of a