Is neurogenesis in two songbird species related to their song sequence variability?

Abstract

Neurogenesis takes part in the adult songbird brain and new neurons are integrated into the forebrain including defined areas involved in the control of song learning and production. It has been suggested that the new neurons in the song system might enable vocal variability. Here, we examined the basal levels of neurogenesis in two songbird species, zebra finch (Taeniopygia guttata) and Bengalese finch (Lonchura striata var. domestica), which do not learn new song elements as adults but differ in the level of song sequence variability. We found that Bengalese finches had less linear and stereotyped song sequence and a higher number of newborn cells in the neurogenic subventricular zone (SVZ) as well as the number of newly born neurons incorporated into the vocal nucleus HVC (used as a proper name) in comparison to zebra finches. Importantly, this vocal sequence variability in Bengalese finches correlated with the number of new neurons in the vocal nucleus HVC and more plastic song was associated with higher neuronal incorporation. In summary, our data support the hypothesis that newly generated neurons facilitate behavioural variability.

1. Background

Neurogenesis is a process which gives rise to new neurons. Both, the neurons born during development and the adult-born neurons originate in the region lining the walls of the lateral ventricle, called a subventricular zone (SVZ) (Dewulf & Bottjer [1]; Scott & Lois [2]). It has been proposed that the adult neurogenesis enables the forming of new memories and that the new neurons enhance behavioural plasticity [3–6].

Songbirds that learn their song represent an excellent model to study neurogenesis in relation to behavioural plasticity. In songbirds, the newborn neurons migrate from the SVZ to the whole telencephalon [7]. The migrating fusiform neuroblasts express the microtubule-associated protein doublecortin (DCX; [8]) which is also expressed in the young round shaped neurons for up to 30 days when they differentiate and start expressing the neuronal marker NeuN [9]. The new cells reach the final destination within 20 days [7]. Among the brain areas which incorporate new neurons are the vocal regions HVC (used as a proper name; [10]) and Area X of striatum [6,11] that are involved in control of song production and learning, respectively [12,13]. In the HVC, the new neurons are already present 8 days after their labelling with [³H]-thymidine [14] and they form efferent projections to the nucleus robustus arcopallii (RA) within 31 days [14]. Species vary in the proportion of new neurons incorporated into the HVC, ranging from 0.1–0.2% in zebra finches (Taeniopygia guttata), 0.4% in Bengalese finches (Lonchura striata var. domestica), to 0.1–0.74% in canaries (Serinus canaria) [5]. The data for zebra finches and Bengalese finches, however, come from deafened birds.

Song learning strategies form a continuum between closed-ended and open-ended ones across bird species. The ‘open-ended' learners (such as European starlings or canaries) can learn new songs or song elements during their entire life [15]. The new neurons added in adulthood might be necessary or permissive for learning the new song motor patterns [6]. On the other hand, the ‘closed-ended' learners (such as zebra finches or Bengalese finches) learn their songs during early development and do not learn new song elements later on [15]. Nevertheless, they also add new neurons into their vocal nuclei in adulthood [16] and their role is not yet understood.

Here, we investigated the naturally occurring rates of cell proliferation in the neurogenic SVZ as well as neuronal incorporation into the vocal regions HVC and Area X in two songbird species, zebra finches and Bengalese finches. We determined syntax variability (i.e. linearity, consistency, stereotypy) of individual birds and tested whether it is related to the neural plasticity both between and within species.

2. Material and methods

(a). Animals and housing

Adult male zebra finches (Taeniopygia guttata) and Bengalese finches (Lonchura striata var. domestica) more than five months and less than 1 year old were bred and housed in the animal facilities of the Centre of Biosciences, Slovak Academy of Sciences. Before the experiments, the males were kept under a 14 L : 10 D cycle in aviaries with dried seed mix and water ad libitum. All procedures followed European Union ethical guidelines and were approved by the State Veterinary and Food Administration of the Slovak Republic.

(b). BrdU/EdU injections

Eleven zebra finches and 11 Bengalese finches were transferred into the sound attenuating chambers for 1 month and their songs were recorded using the Sound Analysis Pro 2011 software [17]. They were divided into two groups, BrdU/BrdU (n = 5 zebra finches, n = 6 Bengalese finches) and BrdU/EdU (n = 6 zebra finches, n = 5 Bengalese finches; figure 1a). Neurons in the same animals were birthdated at two different time points. During the first 7 days of the experiment, all birds received once a day the thymidine analogue 5-bromo-2-deoxyuridine (BrdU; Sigma Aldrich; dose 50 mg kg⁻¹, concentration 10 mg ml⁻¹) administered alternatively into the left and right pectoral muscle. The newborn neurons are known to migrate from the SVZ into the whole telencephalon within 20 days [7] and these BrdU injections were intended to detect the migrated and differentiating neurons in HVC and Area X. To also monitor the acute cell proliferation in SVZ, on day 30 of the experiment (29 days after the first BrdU injection) and 2 h before sacrifice, one additional dose of a cell division marker was injected intramuscularly. Originally, it was BrdU (BrdU/BrdU group), however, we found that a significant number of the BrdU⁺ cells did not migrate from the SVZ (see results) and thus the number of these remaining cells interfered with the number of the acutely labelled BrdU⁺ cells born 2 h before sacrifice. Therefore, another group of birds was injected with 5-ethyl-2-deoxyuridine (EdU; Carl Roth; dose 50 mg kg⁻¹, concentration 10 mg ml⁻¹) 2 h before sacrifice and these birds belong to the BrdU/EdU group (figure 1a).

Figure 1.

Schematic drawing of the experimental design. (a) Birds received seven injections of BrdU (1 per day, days 1–7) and on day 30 they received either one more BrdU injection (BrdU/BrdU group) or EdU injection (BrdU/EdU group) 2 h before perfusion. (b) Schematic drawing of the coronal brain sections including regions where the cells were quantified. The regions are highlighted by the red bold line along the ventricle at four rostro-caudal levels of SVZ with clear anatomical markers, or they are shown as rectangles within Area X, HA and HVC. Area X, vocal nucleus; CA, commisura anterior; HA, hyperpallium apicale; HVC, proper name of vocal nucleus; TSM, tractus septopallio-mesencephalicus. The scale bar represents 500 µm. (Online version in colour.)

Three other zebra finches were used for EdU injections (EdU only group; one EdU injection 2 h before sacrifice; dose 50 mg kg⁻¹, concentration 10 mg ml⁻¹). They stayed in the sound attenuating chambers overnight and were injected and sacrificed in the morning.

(c). Perfusion and tissue processing

The birds were sacrificed with a lethal dose of ketamine-xylazine mixture (15 mg ml⁻¹ ketamine; 3 mg ml⁻¹ xylazine). Their brains were perfused using phosphate buffered saline (0.1 M PBS) followed by 4% paraformaldehyde. After dissection, the brains were post-fixed for 12 h in 20% sucrose, for another 24 h in 30% sucrose, and frozen in the Tissue-Tek OCT compound (Sakura, Japan). Then they were cut coronally into 30 µm sections using the Leica 1800 cryocut (Leica, Germany). Slices were collected at four levels of SVZ based on anatomical landmarks (figure 1b): (i) at the level of Area X (about 300 µm caudally from its anterior end; SVZ_X), (ii) at the level of tractus septopallio-mesencephalicus (SVZ_TSM), (iii) at the level of commisura anterior (SVZ_CA), and (iv) at the level of HVC (at its anterior level; SVZ_HVC). The free-floating slices were stored in PBS with addition of 0.1% sodium azide at 4°C.

(d). Immunohistochemistry

Sections were washed 3 × 5 min in PBS (pH = 7.4). For BrdU staining, sections were incubated for 7 min in 2 N HCl at 37°C, then the reaction was blocked for 3 min by 0.1 M borate buffer (pH = 8) at room temperature (RT), and the sections were again washed 3 × 5 min in PBS. Blocking the non-specific binding sites was done by incubating the sections for 1 h in blocking solution (PBS containing 0.1% bovine serum albumin with 0.3% Triton X-100). Then, sections were incubated at 4°C with primary antibodies diluted in blocking solution for two nights for BrdU staining and overnight for stainings without BrdU antibody. We used the rat polyclonal anti-BrdU (OBT 0030, Accurate Chemical, USA, diluted 1 : 500), mouse monoclonal anti-BrdU (MOBu-1, Invitrogen, USA, diluted 1 : 1000), mouse polyclonal anti-NeuN (MAB377, Milipore, USA, diluted 1 : 500), rabbit polyclonal anti-DCX (AV41333, Sigma, USA, diluted 1 : 3000) and rabbit polyclonal anti-glial fibrillary acidic protein (GFAP; AB5804, Millipore, USA, diluted 1 : 500) antibodies. Next, the sections were washed 3 × 5 min in PBS and incubated for 2 h at RT with secondary antibodies in the dark. We used donkey anti-mouse IgG conjugated with Alexa 488, goat anti-mouse IgG conjugated with Alexa 594, goat anti-rabbit IgG conjugated with Alexa 594 and Alexa 488 (all from Invitrogen, USA) and donkey anti-rat IgG conjugated with Cy3 (Jackson ImmunoResearch, USA). The sections were washed 3 × 5 min in PBS, mounted on silanated slides, rinsed in deionized H₂O and coverslipped with the Vectashield mounting medium (Vector Laboratories, USA). It contained 4′,6-diamidino-2-phenylindole (DAPI) for DCX staining. For BrdU and EdU stainings, the mounting medium was without DAPI.

EdU labelling was performed using an azide-click reaction using a commercially available kit, the Azide-AF 488 dye (Carl Roth, Germany). The sections were washed 3 × 5 min in PBS and incubated for 30 min in EdU labelling cocktail (deionized water, reaction buffer, catalyst solution, 6-FAM-Azide and buffer additive) in the dark at RT.

(e). BrdU/EdU cross-reactivity

Possible cross-reactivity of EdU and BrdU was assessed first by EdU labelling in BrdU/BrdU injected birds (n = 3 zebra finches and n = 3 Bengalese finches, the protocol of injections described above). Second, BrdU immunohistochemistry was performed in EdU only injected birds (n = 3 zebra finches). The mouse monoclonal anti-BrdU MOBu-1 was used (the choice of the antibody was based on personal communication with Dr Tracy Larson). The dilution 1 : 200 in blocking solution led to positive bright labelling. In order to reduce the cross-reactivity, the BrdU antibody was diluted 1 : 200, 1 : 500, 1 : 750 and 1 : 1000. At the end, brain sections from BrdU/EdU birds (n = 2 zebra finches and n = 2 Bengalese finches) were labelled for EdU and then for BrdU with the dilution of the anti-BrdU antibody 1 : 200, 1 : 500, 1 : 750 and 1 : 1000. One section including SVZ and one section including HVC per bird was used for each staining.

(f). Cell counts and statistical analysis

We used images taken using a Leica DFC340 FX camera attached to a Leica DM5500 microscope with the Leica microsystems LAS AF6000 software. Quantification was performed using Photoshop CS 6 software. Cell counts were performed blind to the experimental protocol. To quantify cell proliferation in SVZ, 10× lens was used, the number of BrdU⁺ and/or EdU⁺ cells was counted along the ventricle (figure 1b) in 2–3 images for every hemisphere and this count was expressed as the number of cells per millimeter (length) of SVZ. In the BrdU/EdU labelling, the BrdU antibody dilution was adjusted so that there was a minimal reactivity of the BrdU antibody to EdU⁺ cells (0–20% BrdU⁺ cells were also EdU⁺, see results). These double labelled cells could be distinguished as the BrdU signal was notably low (while the BrdU signal in single labelled cells was bright). To avoid double affiliation of these cells, the EdU⁺ cells (all showing strong signal) were counted first and BrdU⁺ cells were counted after that. The cell counts did not differ between hemispheres (p = 0.11–0.87, paired t-tests) and thus, the data from left and right hemispheres were merged. Cell counts from the BrdU/BrdU and BrdU/EdU experiments were compared by ANOVA with species (zebra finch, Bengalese finch) and brain region (SVZ_X, SVZ_TSM, SVZ_CA, SVZ_HVC) as factors. Later, statistical analyses were performed separately within each experiment (BrdU/BrdU and BrdU/EdU) using ANOVA with species and brain region as factors, followed by the Fisher's protected least significant difference (Fisher's PLSD) post hoc test. A t-test was used to compare cell counts in the whole SVZ between zebra finches and Bengalese finches.

Since we wanted to compare the number of new cells in SVZ between zebra finches and Bengalese finches across the two experiments (BrdU/BrdU and BrdU/EdU), but the numbers of BrdU⁺ cells in the BrdU/BrdU birds and BrdU⁺ and EdU⁺ cells in the BrdU/EdU birds significantly differed (p < 0.001; see results), we normalized the data. In each group separately, we divided every value by the zebra finch mean in the SVZ area and then used ANOVA on Ranks with factors species and brain region, followed by a Student–Newman–Keuls post hoc test.

The number of fusiform, round, and all DCX⁺ cells in HVC, Area X, and in a region not related to song, hyperpallium apicale (HA; [10]), was quantified using a 40× lens. DCX⁺ cells were counted from three sections and in each section from 2 to 3 counting fields measuring 240 × 322 µm (figure 1b). The cell counts were recalculated per mm². Since the values did not differ between the two hemispheres (p = 0.25–0.94, paired t-test), the cell counts from left and right hemispheres were merged. Two Bengalese finches were not included in the analyses because of technical difficulties. Statistical analyses for DCX⁺ cells were performed using ANOVA with factors species, brain region and cell type, followed by the Fisher's PLSD post hoc tests. To estimate the association between song parameters (linearity, consistency and stereotypy) and DCX⁺ or EdU⁺ cell counts, linear regressions were performed.

The number of BrdU⁺/NeuN⁺ cells in Area X, HVC and HA was counted in three sections per bird (in the whole vocal nucleus appearing on the section; for HA in the area covering the rectangles shown in figure 1b) using 10× lens. The values were recalculated per mm². Because the values of BrdU⁺/NeuN⁺ neurons did not differ between hemispheres (p = 0.06–0.98, paired t-tests), they were merged for each bird. Statistical differences between zebra finches and Bengalese finches were assessed by ANOVA with factors species and brain region, followed by the Fisher's PLSD post hoc test. The association between song parameters and the number of BrdU⁺/NeuN⁺ new neurons was assessed by linear regression.

We also estimated the percentage of BrdU⁺/NeuN⁺ newborn neurons added to HVC and Area X per day in six zebra finches and six Bengalese finches. The counts were obtained using 10× lens from four to six sections per bird and averaged. Since we used seven BrdU injections within three to four weeks prior to sacrifice and BrdU is circulating in the blood of songbirds longest for 1 h [18], we recalculated the percentage values for a 24 h period (no. of neurons × 24 / 7). Differences between zebra finches and Bengalese finches were assessed by t-test.

(g). Song analysis

Song usually started with a rendition of the same introductory note and it was followed by one or more song motifs. Each motif consisted of a sequence of different syllables. The song syllables were defined as acoustic elements separated by silence gaps longer than 10 ms. As we intended to calculate the variability of the song sequence, the repetitive introductory notes were not included in the analyses. Song sequence variability was determined according to the protocol used by Scharff & Nottebohm [13]. Song linearity, consistency and stereotypy were calculated in the first 10 songs sung after the onset of light. The song linearity score was expressed as the number of different syllable types divided by the number of syllable transition types in a song. Transitions included only transitions from one syllable to another; they did not include transitions from the start to the first syllable and from the last syllable to end. The sequence consistency score was expressed as the number of typical transitions divided by the number of total transitions in the song. Song sequence stereotypy was calculated as a mean of song linearity and song consistency. To identify the difference between species, t-tests were used.

3. Results

(a). BrdU/EdU cross-reactivity

Because there are several reports of cross-reactivity in detecting thymidine analogues [19,20], we first checked this possibility in our experiments. EdU was not detected in any bird injected with BrdU. Initially, BrdU (with the antibody dilution 1 : 200) was detected in EdU injected birds. The higher dilutions of the BrdU antibody 1 : 750 and 1 : 1000 led to low and faint signals in EdU injected birds while providing a strong signals in BrdU/BrdU injected birds. In BrdU/EdU injected birds, high cross-reactivity was observed with the dilution of the BrdU antibody 1 : 200. The lowest reactivity of the BrdU antibody to EdU⁺ cells was detected when it was diluted 1 : 1000. With this dilution, from 0 to 20% of BrdU⁺ cells were also EdU⁺ and these double labelled cells could be clearly distinguished as they were notably less intensively labelled for BrdU (electronic supplementary material, figure S1).

(b). Bengalese finches had less linear and stereotyped song than zebra finches

There was a significant effect of species on the song sequence of zebra finches (n = 11, figure 2a) and Bengalese finches (n = 11, figure 2b). The songs of Bengalese finches were less linear and stereotyped than the songs of zebra finches (p < 0.001, figure 2c). The mean number of syllables in Bengalese finches was 7.93 ± 0.47 (mean ± s.e.m.) and it was significantly higher (p < 0.001; t-test) than in zebra finches with 5.31 ± 0.29 syllables. In line with the above, there were more syllable transitions in Bengalese finch songs (13.04 ± 0.77) as compared to the zebra finch songs (6.01 ± 0.39; p < 0.001; t-test).

Figure 2.

Songs of zebra finch and Bengalese finch. Sonograms of (a) zebra finch song and (b) Bengalese finch song. The individual syllables are colour-coded and labelled as A, B, C, etc. Introductory notes at the beginning of the song are not colour-coded and are labelled as i. The bold line at the bottom of the zebra finch sonogram represents the duration 1 s. The scores for linearity, consistency, and stereotypy are above each of the songs. (c) Song scores for linearity, consistency and stereotypy for n = 11 zebra finches and n = 11 Bengalese finches. Each bar represents mean and standard error of the mean (SEM). ***p < 0.001. ANOVA with Fisher's LSD test. (Online version in colour.)

(c). Bengalese finches showed more newborn cells in the subventricular zone than zebra finches

Two experiments were performed, BrdU/BrdU (seven BrdU injections followed by one BrdU injection 30 days later) and BrdU/EdU (seven BrdU injections followed by one EdU injection 30 days later; figure 1a). Therefore, we first examined if the number of BrdU⁺ cells in SVZ in the first experiment was the same as the number of BrdU⁺/EdU⁺ cells in SVZ in the second experiment. We found that the cell counts differed (p < 0.001) and therefore we present the results related to BrdU and/or EdU from these experiments separately (figure 3a–e).

Figure 3.

Cell proliferation in SVZ. The number of BrdU⁺ cells from the experiment BrdU/BrdU (a) in four regions of SVZ and (b) in the whole SVZ. (c) The number of BrdU⁺ and EdU⁺ cells from the experiment BrdU/EdU in four regions of SVZ. The number of EdU⁺ cells (d) in four regions of SVZ and (e) in the whole SVZ. Each bar represents mean and SEM. (f) Normalized values for the amount of new cells in birds from both groups (BrdU/BrdU and BrdU/EdU). Each circle represents mean and SEM. **p < 0.01, ***p < 0.001. ANOVA followed by Fisher's LSD test for (a), (c), (d), t-test for (b), (e), and ANOVA on Ranks followed by Student–Newman–Keuls post hoc test for (f).

In both experiments, we found that there was a significant effect of species on the number of new cells in SVZ where Bengalese finches showed higher proliferation then zebra finches (p < 0.001 for BrdU/BrdU, figure 3a; p = 0.08 for BrdU/EdU, figures 3c and 4). Specifically, in the BrdU/BrdU experiment, Bengalese finches showed a significantly higher number of BrdU⁺ cells in SVZ_X, SVZ_TSM and SVZ_CA (p = 0.009, 0.003, 0.001, respectively; figure 3a) and also overall in the whole SVZ (figure 3b). When stained only for EdU (EdU and NeuN staining) in the BrdU/EdU experiment, the number of EdU⁺ cells was significantly higher in Bengalese finches than in zebra finches (p = 0.002, figure 3d). The acute cell proliferation, i.e. the number of newborn cells less then 2 h old (EdU⁺) was higher in Bengalese finches in SVZ_CA (p = 0.008) and tended to be higher also in SVZ_HVC (p = 0.09; figure 3d). To be able to combine the data from both experiments, we normalized the values so that we divided the number of BrdU⁺ cells from the BrdU/BrdU experiment and the number of BrdU⁺ plus EdU⁺ cells from the BrdU/EdU experiment obtained from Bengalese finches with the mean value in a given brain area in zebra finches. We again found significantly higher values in Bengalese finches in SVZ_X, SVZ_TSM and SVZ_CA (p is less than 0.001, less than 0.001, 0.003, respectively; figure 3f). In both experiments, there was also a significant effect of brain region on the number of new cells (p < 0.001). The highest numbers were found in SVZ_CA and SVZ_TSM and the lowest numbers were in SVZ_HVC.

Figure 4.

Representative images showing newborn cells in SVZ_CA in coronal brain sections of zebra finch (left) and Bengalese finch (right). New cells arising latest 2 h before the sacrifice are labelled by EdU (green, first row). New cells arising 3 to 4 weeks before the sacrifice are labelled by BrdU (red, second row). Merged images are in the third row. White rectangles show the locations of the inserts in left bottom corners. The fields in the inserts are magnified three times. The Bengalese finch has higher numbers of both EdU⁺ and BrdU⁺ cells in comparison to the zebra finch. There is no co-localization of EdU and BrdU. The scale bar represents 100 µm.

Interestingly, we found not only EdU⁺ cells but also BrdU⁺ cells in SVZ in the BrdU/EdU experiment (figure 4), i.e. three to four weeks after BrdU injections. The proportion of the new EdU⁺ cells was 51.16 ± 2.94% from all the newborn cells found in SVZ and there was no difference between the species (p = 0.12, t-test). The remaining 48.84 ± 2.94% new cells in SVZ, that were BrdU⁺/EdU⁻ and thus three to four weeks old, did not express GFAP or DCX (electronic supplementary material, figure S2).

(d). The number of young neurons in HVC was higher in Bengalese finches

Next, we counted the number of young neurons expressing DCX. There was a significant effect of species (p < 0.01), brain region (p < 0.001) and cell type (p < 0.001) on the number of DCX⁺ neurons. Then we analysed separately the two morphological states, DCX⁺ fusiform neurons suggesting their migratory state and DCX⁺ round neurons that represent their early differentiation state. The number of DCX⁺ fusiform neurons in HVC was higher in Bengalese finches in comparison to zebra finches (p = 0.0005) and it was not different between the species in Area X and in the brain region HA that is not related to song (figure 5a). Similarly, Bengalese finches in comparison to zebra finches showed a higher number of DCX⁺ round neurons in HVC (p = 0.04, figure 5b). Together, Bengalese finches had a higher number of all DCX⁺ neurons in HVC (p = 0.003) but not in Area X or HA (figure 5c). The fusiform neurons expressed DCX and were co-stained with DAPI but not NeuN (figure 6). On the other hand, the round DCX⁺ neurons expressed NeuN, although sometimes it appeared punctate as at the beginning of the NeuN expression (see the highlighted neuron in zebra finch in figure 6).

Figure 5.

Incorporation of new neurons into HVC, Area X and HA. Quantification of DCX⁺ (a) fusiform, (b) round, and (c) all neurons. (d) Quantification of BrdU⁺/NeuN⁺ neurons. Each bar represents mean and SEM. *p < 0.05, **p < 0.01, ***p < 0.001. ANOVA followed by Fisher's LSD test.

Figure 6.

Representative images of young new neurons labelled as DCX⁺ (red, first row), neurons labelled as NeuN⁺ (green, second row), all DAPI⁺ cells (blue, third row) and merge (last row) in zebra finch (left column) and Bengalese finch (right column). Red arrows point to new fusiform neurons, yellow arrows point to new round neurons that start to express NeuN. The scale bar represents 100 µm.

Similarly to the results with DCX, the number of new neurons shown as BrdU⁺/NeuN⁺ in HVC was higher in Bengalese finches than in zebra finches (p < 0.05, figure 5d). There was no statistically significant difference in Area X or in HA.

(e). Higher song sequence variability was not associated with higher cell proliferation in the subventricular zone

The higher song sequence variability (i.e. lower linearity and stereotypy) and cell proliferation in Bengalese finches in comparison to zebra finches might imply that these traits are related. To explore this possibility, we looked to see if there is an association between song sequence variability measures (linearity, consistency, stereotypy) and the number of new cells in SVZ. We performed the analyses separately for the BrdU/BrdU and BrdU/EdU birds and found that the regression analyses were not significant when performed within zebra finches or Bengalese finches for any of the song parameter in any of the part of SVZ or together in the whole SVZ (figure 7a for the comparison of song with the number of EdU⁺ cells in the whole SVZ). Thus, cell proliferation in SVZ was not related to song sequence variability.

Figure 7.

Correlations of the parameters of song variability (linearity, consistency, stereotypy) in Bengalese finches with (a) the number of new EdU⁺ cells in SVZ, (b) the number of new DCX⁺ neurons in HVC, and (c) the number of new DCX⁺ neurons in Area X. In (a), p = 0.53, r = 0.38 for consistency, p = 0.98, r = 0.01 for stereotypy, p = 0.71, r = 0.23 for linearity; in (b), p = 0.96, r = 0.02 for consistency, p = 0.03, r = 0.70 for stereotypy, p = 0.03, r = 0.73 for linearity; in (c), p = 0.60, r = 0.21 for consistency, p = 0.16, r = 0.51 for stereotypy, p = 0.09, r = 0.60 for linearity. *p < 0.05. Linear regressions.

(f). Higher song sequence variability was associated with a higher number of young neurons

Then we looked at the number of new DCX⁺ neurons in the vocal regions Area X and HVC. Within zebra finches, there was no significant correlation of any song variability parameter (linearity, consistency, stereotypy) with the number of DCX⁺ neurons in HVC or Area X. Within Bengalese finches, there were significant negative correlations of song linearity and stereotypy with the number of DCX⁺ round and all DCX⁺ neurons in HVC (p = 0.01–0.04, r = 0.68–0.78; figure 7b for song versus all DCX⁺ neurons). We did not find a relationship of any song sequence parameter with DCX⁺ neurons in Area X, only a trend to negative correlation of song linearity with the number of all DCX⁺ neurons (p = 0.09, r = 0.58; figure 7c).

Similarly, we did not find any correlation of song parameters with the number of new BrdU⁺/NeuN⁺ neurons in HVC or Area X in zebra finches. In HVC of Bengalese finches, there was a significant negative correlation of the number of newborn neurons with song consistency (p = 0.004; r = 0.84) but not song linearity or stereotypy. A similar relationship was found in Area X of Bengalese finches where the number of newborn neurons correlated with song consistency (p = 0.02; r = 0.73).

In summary, more variable song was associated with higher numbers of DCX⁺ and BrdU⁺ neurons in HVC of Bengalese finches.

(g). Percentage of newborn neurons

Finally, we estimated the percentages of new BrdU⁺/NeuN⁺ neurons incorporated into HVC and Area X per day. After seven injections of BrdU, 0.11 ± 0.02% of new HVC neurons were recruited in zebra finches and 0.19 ± 0.05% in Bengalese finches. When recalculated per day (assuming that one injection of BrdU is available for 1 h [18]; we did not take into account a possible circadian rhythm of cell incorporation), there was 0.37 ± 0.06% of new neurons incorporated in HVC of zebra finches and 0.65 ± 0.18% of new neurons incorporated in HVC of Bengalese finches. In Area X, the percentages after seven BrdU injections were 0.05 ± 0.01% for zebra finches and 0.06 ± 0.02% for Bengalese finches. The recalculated values per day were 0.18 ± 0.04% in zebra finches and 0.21 ± 0.07% in Bengalese finches. Although the percentage of new neurons in HVC in Bengalese finches was almost double of that in zebra finches, the differences between species were not statistically significant.

4. Discussion

It has been proposed that newborn neurons in vocal nuclei in adult songbirds permit or facilitate forming new memories and are required for song maintenance [5,6]. Here, we studied their function in two songbird species that do not add new elements to their songs as adults and are called ‘closed-ended' song learners, zebra finch and Bengalese finch [15].

The songs of Bengalese finches in comparison to zebra finches have shorter distribution of syllable durations and the silent gaps between syllables include larger numbers of longer gaps [21]. However, differences between the two species in syllable sequencing have not, to our knowledge, been published. Here, we showed that song sequences of Bengalese finches include more syllables, they are clearly more variable, and it is owing to lower linearity but not consistency in comparison to the songs of zebra finches. Although the difference between the number of syllables was not as dramatic as reported in some colonies [21], Bengalese finches in our experiment used twice as many syllable transition types as zebra finches.

Our study showed that the highest cell proliferation in the SVZ occurred at the level of commusiura anterior and the lowest at the level of HVC that is similar to previous reports in zebra and Bengalese finches [22,23]. To our knowledge, this is the first comparison of cell proliferation and/or neurogenesis levels in the SVZ and vocal regions between zebra and Bengalese finches at normal conditions. The higher numbers of newborn cells throughout SVZ is in line with the higher neuronal addition to HVC and/or Area X in Bengalese finches.

Deafening-induced song deterioration is faster in Bengalese than in zebra finches [24–26]. The higher addition of newborn neurons into HVC in the absence of auditory feedback can contribute to the faster degradation of song in Bengalese finches in comparison to zebra finches [23,24]. In our study we show that Bengalese finches have a higher rate of cell proliferation in SVZ and new neuron incorporation into HVC even naturally and without any intervention. This might be or might be not associated with the higher song variability of Bengalese finches in comparison to zebra finches. Importantly, however, we also found a correlation within this species. Bengalese finches with more variable (less linear and stereotyped) song had higher number of newborn neurons in HVC than Bengalese finches with less variable song. We did not find such a relationship within zebra finches, perhaps because their song is very stereotyped in all individuals.

The HVC neurons projecting to Area X show a syllable-selective and a transition-selective activity [27,28]. Both, zebra finches and Bengalese finches with damaged Area X have impaired initiation or transition within a song sequence [29,30]. Although we did not find significant differences between these two species in the number of newborn neurons in Area X, we found a negative correlation of the number of BrdU⁺/NeuN⁺ neurons in Area X with song consistency within Bengalese finches and it further supports the role of Area X in song syntax.

We did not find a correlation of the number of new cells in SVZ with song sequence linearity, consistency, or stereotypy in zebra finches or Bengalese finches. Although we used 11 birds from each species, the experiments in SVZ were performed in two groups, the data could not be combined, and the regression analyses could be done separately with only half of the birds. Therefore, we suppose that the number of birds used for correlations with song sequence variability measures might be not sufficient and we cannot absolutely exclude that such a relationship exists.

Neurogenesis is studied mostly using [³H]thymidine and BrdU which are estimated to be available in the adult mammalian brain for up to 6 h [31]. A recent study, however, shows that BrdU is circulating in the blood of songbirds for only up to 1 h [18]. According to the 1 h period of BrdU availability, we show that 0.37% and 0.65% of all HVC neurons might be added to the nucleus per day in zebra finches and Bengalese finches, respectively. The precise values may also be affected by the circadian rhythm of cell proliferation. The fold difference between the peak and the trough is 1.6 in the subgranular zone of hippocampus in mice [32] and even higher in zebrafish, depending on the brain region [33]. The results in other avian studies do not seem to be recalculated depending on the time of day or number of injections used. Nevertheless, our data correspond with the previous reports in canaries (0.5–0.74% after two [³H]thymidine injections; [34]) and in deafened zebra finches (1.9%) and Bengalese finches (3.84%; both after 20 [³H]thymidine injections; [24]). The recruitment in the vocal region Area X is 0.18% neurons in zebra finches and 0.21% in Bengalese finches which is lower than in HVC.

The thymidine analogue EdU was introduced relatively recently and has proven to be a very useful tool for studying cell proliferation [35–37] although it should be used shortly before sacrifice owing to its toxicity [19]. In birds injected with BrdU 30 days prior the perfusion and then 2 h before the sacrifice, we observed a relatively large amount of BrdU⁺ cells remaining in SVZ. In birds injected with EdU instead of BrdU before sacrifice, we were able to distinguish that the amount of new cells in SVZ labelled by one EdU dose was about the same as the amount of new cells 3 to 4 weeks old that were labelled by seven doses of BrdU. It means that there is a significant portion of the newborn cells in SVZ that do not migrate to other brain regions. In rodents, radial glia in the ventricular zone give rise to SVZ stem cells that maintain the neurogenic lineage and striatal radial glia give rise to SVZ astrocytes that continue to generate neurons in the adult brain [38]. However, we found that these BrdU⁺ cells in SVZ did not express GFAP. Although there were many DCX⁺ neurons in SVZ, these were not the remaining BrdU⁺ cells either. Thus, the BrdU⁺ cells might be neuroblasts that do not yet express DCX or glial cells that were not stained positively. Differences in specificity of GFAP antibodies was previously reported [39].

The numbers of newborn cells in the BrdU/EdU group in SVZ were higher than those in the BrdU/BrdU group, regardless of the species. Although we performed the second experiment 13 months later, the difference in the number of newborn cells in SVZ should not be owing to a seasonal difference. Moreover, neurogenesis does not differ throughout the year in zebra finches such as in canaries [34]. It was not caused by false double counting either as each cell was counted only once as either an EdU⁺ or BrdU⁺ cell. The EdU kit is very sensitive and did not react with BrdU. Although higher concentration of the BrdU antibody led to double labelling with EdU, the decreased BrdU antibody concentration greatly improved the cross-reactivity. In comparison, the equimolar injections of thymidine analogues did not label the same number of neurons even in other studies [40,41]. A possible source of the difference between the two experiments (BrdU/BrdU and BrdU/EdU) might be a different sensitivity of the techniques of labelling. The detection of EdU using the commercially available kit was much faster and the signal was very strong. In comparison, immunohistochemical detection of BrdU is more complicated and requires more time. Alternatively, the difference between the two experiments may be caused by different clearance times of these thymidine analogues. While BrdU is detectable in the blood of canaries for up to 60 min post-injection [18], the time of EdU availability or clearance has not been determined yet.

Most of the cell proliferation and neurogenesis comparative studies have been focused on differences between male and female zebra finches. Here we made a comparison between two species with different song syntax. However, further work on other species should give us a better view about interspecies neuroplasticity and its relation to behaviour.

Ethics

All procedures were approved by the State Veterinary and Food Administration of the Slovak Republic (permit numbers 2982/17-221/3 and 3569/16-221).

Data accessibility

Data are provided in the electronic supplementary material.

Authors' contributions

L.K. and J.P. designed the study, analysed the results and wrote the paper, J.P., K.L. and B.B. performed the research. All authors discussed the data. All authors gave final approval for publication.

Competing interests

We have no competing interests.

Funding

The study was supported by a grant from the Slovak Research and Development Agency APVV-15-0077 to L.K.