Vocal control area-related expression of neuropilin-1, plexin-A4, and the ligand semaphorin-3A has implications for the evolution of the avian vocal system

Animals

Bengalese finches, budgerigars and quails were bought from local breeders or were bred in our lab facilities. We used four postnatal 30 days (P30), three P60, and four adult (from P120 to 3 years old) male Bengalese finches, and three P30 and three adult female Bengalese finches, and three postnatal 2 weeks and three adult (1–3 years old) budgerigars, and three postnatal 2 weeks and three adult (about 3 months old) male quails in this study. Even though some differences were observed in expression level, we got similar expression patterns between juvenile and adult birds in each species (n = 3/3 or 4/4). All birds were deeply anesthetized with an intramuscular injection of sodium pentobarbital (50 mg/kg) and then killed. After decapitation, their brains were embedded in an OCT compound (Tissue-Tek) and frozen on dry ice for cryosectioning. Frozen sections for in situ hybridization or thionine staining for neuroanatomical references were cut serially in 20 or 30 µm thicknesses by using a cryostat (Leica). Before they were killed, all birds were kept under similar conditions in which they could move freely, sing, and listen to other birds singing. To extract the total RNA, brain tissues were dissected and placed in Qiazol Lysis reagent (Qiagen), and the RNA was purified using an RNeasy Lipid Tissue Mini Kit (Qiagen). The sex of the birds was determined either by looking for testis or by extracting genomic DNA from a portion of a digit with a DNeasy tissue kit (Qiagen) and using a polymerase chain reaction (PCR) with primers that amplify the chromo-helicase-DNA binding (CHD) gene (Ellegren 1996). Research protocols were approved by the Animal Care and Use Committee of RIKEN and conformed to National Institutes of Health Guidelines.

Isolation and cloning of cDNA

For the target genes, cDNA fragments were isolated from the brains of adult Bengalese finches (Npn-1 [GenBank Acc. No. AB468949], Npn-2 [AB468950], Sema3A [AB468946], plexin-A1 [AB468693] and plexin-A4 [AB468692]), budgerigar (Npn-1 [AB468951], Sema3A [AB468947], plexin-A1 [AB468695] and plexin-A4 [AB468694]), and quail (Sema3A [AB468948], plexin-A1 [AB468697] and plexin-A4 [AB468696]) using reverse transcription–PCR (RT–PCR). The following primers were used: Npn-1, 5′-ACCAGGTGGATCTGAAGTGG-3′, 5′-GCTTGGCTCAGTGTCATCAA-3′; Npn-2, 5′-GCGTCCTGTCTCTCACCTTC-3′, 5′-TCCAGACGCATTCCTATTCC-3′, Sema3A, 5′-GGAGAGCACACTGGAAAAGC-3′, 5′-TAACAGTGGGTTCCCGAAAG-3′; Plexin-A1, 5′-CGGCAGCAGATTGACTACAA-3′, 5′-GGATGTCCTTGGCGTACAGT-3′; Plexin-A4, 5′-CACCATCACACAGGTCAAGG-3′, 5′-CATATGCATGCGAGACTGCT-3′. For the quail brain, we used chicken neuropilin-1 (Takagi et al. 1995) and neuropilin-2 (Watanabe et al. 2004); the primer sequences were based on the chicken cDNA sequences. Each cDNA fragment was inserted in the pGEM-T Easy vector (Promega). The plasmids were digested with enzymes to release the fragment, and probes were synthesized using SP6, T3, or T7 RNA polymerase (Roche) with digoxigenin (DIG)-labeling mix (Roche).

In situ hybridization

Tissue sections were postfixed for 10 min and then washed in phosphate-buffered saline (PBS) three times for 3 min. The slides were delipidated with acetone and then acetylated and washed in PBS with 1% Triton-X100 (Wako). The slides were incubated at room temperature with hybridization buffer, which contained 50% formamide (Wako), 5 × standard saline citrate (SSC), 1 × Denhart's solution (Sigma), 250 µg/mL yeast tRNA (Roche), and 500 µg/mL DNA (Roche). Then, the sections were hybridized at 72°C overnight in hybridization buffer with RNA probes. The sections were rinsed in 0.2 × SSC for 2 h, and then blocked for 2 h in a solution of 0.1 m Tris (pH 7.5) and 0.15 m NaCl with 10% sheep serum. The slides were incubated overnight with alkaline phosphatase (AP)-conjugated anti-DIG antibody (Roche). After washing, AP activity was detected by adding 337.5 mg/mL nitroblue tetrazolium chloride (NBT) and 175 mg/mL 5-bromo-4-chloro-3-indolyl phosphate (Roche). All of the sections were visualized with an Eclipse E600 microscope (Nikon), and all images were scanned with a computer-based image processing system (Neurolucida; MicroBrightField). Images were processed with Photoshop CS2 software (Adobe Systems).

Vocal area-related neuropilin expression in songbird

We examined Npn-1 expression in juvenile and adult Bengalese finch brains. We detected Npn-1 expression in many brain areas, including vocal control areas such the high vocal center (HVC), AreaX, and the arcopallium, including the robust nucleus of the arcopallium (RA), and tracheosyringeal hypoglossal nucleus (nXIIts), by postnatal 30 days (P30) (n = 4/4) (Fig. 2A–C). The vocal control area-related Npn-1 expression was increased during development (Fig. 2D–G). Interestingly, corresponding to previous reports that no structure of HVC and AreaX are seen in the female Bengalese finch anatomically (n = 3/3) (Tobari et al. 2005), the strong Npn-1 expression in HVC and AreaX was male-specific, reflecting the sexual dimorphism at the molecular level (Fig. 2H,I).

Next, we examined Npn-2 expression. Although strong Npn-2 expression was detected in the hippocampus, as with Npn-1 expression, no Npn-2 expression was detected in the HVC and RA, suggesting less involvement in the vocal system (n = 4/4) (Fig. 3A,B). However, in some vocal control areas, such as AreaX and nXIIts (data not shown), both neuropilins were expressed.

Semaphorin and Plexin expression in songbird

Since Npn-1 expression is more related to vocal control areas than Npn-2, we examined expression of Sema3A, a specific ligand of Npn-1. Interestingly, Sema3A showed complementary expression to Npn-1 in HVC and the arcopallium both in juvenile and adult stages (n = 4/4) (Fig. 3D–H). Interestingly, strong Sema3A expression was detected in the surrounding area of arcopallium (n = 3/3) (Fig. 3H).

It has been suggested that neuropilins make receptor complexes with four type-A plexins and mediate semaphorin signals (Takahashi et al. 1999; Tamagnone et al. 1999). However, the combinations of functional receptor complexes in vivo remain unclear. Several works have found direct evidence of the in vivo function of neuropilin-plexin complex. Analysis of mutant mice revealed that Plexin-A4 interacts with Npn-1 rather than Npn-2, mediating semaphorin signaling into the growth cone (Yaron et al. 2005). In dorsal root or sympathetic ganglia, the repulsion of Sema3A was largely inhibited while the repulsion of Sema3F was intact in plexin-A4 mutant mice, suggesting that plexin-A4 is a primary mediator of Sema3A signaling via Npn-1 in these neurons but not a mediator of Sema3F signaling with Npn-2 (Suto et al. 2005). Therefore, we examined plexin-A4 expression. In the Bengalese finch brain, plexin-A4 was expressed in the HVC (n = 4/4) (Fig. 3F), as Npn-1 was in its serial section (Fig. 2F). Plexin-A4 was also expressed in RA (n = 4/4) (Fig. 3I,J), and nXIIts (data not shown). By contrast plexin-A1, which is another type of plexin that is involved in both Npn-1 and Npn-2 signaling in vitro (Takahashi & Strittmatter 2001), showed complementary expression to plexin-A4 in the RA, although both plexins were expressed in another area of the arcopallium (n = 3/3) (Fig. 3K).

Npn, Plexin, and Sema expression in the budgerigar and quail brains

Next, we examined neuropilin-1, semaphorin, and plexin expression in another vocal learner, the budgerigar, and a-non-vocal-learner, the quail.

In the budgerigar brain, Npn-1 was expressed in vocal control areas such as the central nucleus of the lateral nidopallium (NLC), anterior arcopallium (AAC), and nXIIts (n = 4/4) (Fig. 4A,D). Interestingly, expression of Sema3A in vocal areas was not seen in the budgerigar brain as in the Bengalese finch (n = 4/4) (Fig. 4A,B). Strong plexin-A4 expression in NLC and AAC was also seen, as in the Bengalese finch. However, in contrast to RA, both plexin-A4 and plexin-A1 were expressed in AAC (n = 4/4) (Fig. 4C,D).

In the quail brain, strong Npn-1 expression in the arcopallium and Sema3A expression outside the arcopallium were detected, as in the Bengalese finch and budgerigar (n = 3/3) (Fig. 4E,F). However, no strong aggregate HVC-like Npn-1 or plexin-A4 expression was seen in the NCL, the area including the corresponding area to the HVC, in the quail brain (data not shown), although the Npn-1 expression in nXIIts was similar (n = 4/4) (Fig. 4J). Furthermore, no strong plexin-A4 expression was seen in the arcopallium (n = 3/3) (Fig. 4G), although strong plexin-A1 expression was seen in the arcopallium (Fig. 4H) and plexin-A4 expression was detected in nXIIts, as in the other two species (n = 3/3) (Fig. 4L).

Possible roles of semaphorin/neuropilin signals in vocal system development

This study focused on the vocal control area-related expression of three genes: Npn-1, Sema3A, and plexin-A4. Interestingly, in the songbird brain, Npn-1 and its coreceptor plexin-A4 were expressed in the HVC, RA, and nXIIts, while its ligand Sema3A was expressed outside vocal control areas. Though further confirmation analysis should be necessary, as double staining for Npn-1 and plexin-A4 or their proteins, expression pattern of Npn-1 and Plexin-A4 in serial sections (2, 3) suggests that many cells in HVC express both Npn-1 and Plexin-A4. Since a high level Sema3A expression was detected just beside the margin of the arcopallium, particularly in the lateral region and in the anterior region of the nidopallium, it appears that Sema3A repulses Npn-1/Plexin-A4-positive RA-projecting neurons, preventing them from entering these areas. Neuropilin knockout mice show axon defasciculation (Cloutier et al. 2002; Kawasaki et al. 2002). Npn-1 may be involved in axon bundle formation of RA-projecting neurons.

We also detected Npn-1/Plexin-A4 expression in the RA and nXIIts. Semaphorin/neuropilin/plexin systems may also be at work in RA-nXIIts and nXIIts-syrinx connections. Although we did not detect related Sema3A expression in the brainstem (Matsunaga & Okanoya unpubl. data, 2008), other type-III semaphorins may function as the repellant to guide RA axons to nXIIts. Otherwise, cooperative action with other guidance molecules, such as other Sema/Plexins, Netrin/DCC, or Slit/Robo, may be involved in making these connections. Indeed, Slit/Robo is expressed in various regions of the brainstem in the zebra finch (Taeniopygia guttata; Hara et al. 2008).

It is also possible that Npn-1 and Sema3A interaction is involved in the formation of the intra-area circuit. In the budgerigar brain, Sema3A expression is very weak in the NLC and AAC, although Sema3A is expressed strongly in the areas surrounding both vocal control areas. Sema3A expression may block axonal extension or cell migration of Npn-1/Plexin-A4-positive NLC and AAC neurons outside the vocal control area. Interestingly, in the Bengalese finch, complementary expression of cadherin-6B and R-cadherin is prominent in the HVC, and neurons form cluster-like structures. By contrast, in the budgerigar brain, both cadherins are expressed moderately in the NLC, and neurons are broadly distributed in the NLC area (Matsunaga et al. 2008), although some molecular mechanisms are similar between the two species (Jarvis 2004; Matsunaga & Okanoya 2008b). In parrots, the Sema/Npn interaction rather than cadherins may be used for formation of this area.

The real situation seems to be much more complex. Sema3A/Npn-1/Pleixin-A4 is not the sole ligand-receptor combination. In fact, neuropilins are neither the sole receptors of type-III semaphorins nor the sole coreceptors of plexins. Neuropilins also make complexes with the vascular endothelial growth factor (VEGF) receptor to transmit VEGF signals (Klagsbrun et al. 2002; Neufeld & Kessler 2008). Similarly, Plexin-A4 is not the sole coreceptor for Npn-1. For example, Plexin-A4 binds to Sema6A directly and meditates its action without neuropilins (Suto et al. 2007). VEGF is also expressed in the HVC nucleus of the canary (Serinus canarius) brain, and it has been suggested that VEGF induced by testosterone is necessary for neuronal recruitment into the HVC nucleus (Louissaint et al. 2002). Interestingly, neuropilins are expressed in the HVC and AreaX, where neurogenesis continues throughout life (Wilbrecht & Nottebohm 2004). There is much evidence that axon guidance molecules also function in various aspects of neural development, such as cell survival, neurogenesis, synaptic function, and the inhibition of regeneration (Dalva et al. 2000; Moreau-Fauvarque et al. 2003; Matsunaga et al. 2004, 2006; Bredesen et al. 2005; Chédotal et al. 2005; Depaepe et al. 2005). The VEGFR-Npn-1 complex may regulate neurogenesis directly. To explore neuropilin function in vocal system development and vocal learning, it is necessary to analyze the expression of related genes, such as VEGF, VEGFR, other plexins, and semaphorins, as well as the function of Npn-1.

Implication of semaphorin/neuropilin signaling in the evolution of the avian vocal system

Interestingly, vocal control area-related expression of Npn-1, Sema3A, and plexin-A4 is seen in the budgerigar brain and even, to an extent, in the quail brain. Based on our results for Npn-1, plexin-A4, and Sema3A expression in the three species, we propose a hypothetic model for the evolution of the avian vocal system (Fig. 5).

Similarity between vocal learners and non-learners has been demonstrated in anatomical and gene expression analyses, and it has been suggested that the HVC-RA connection has diverged from the NCL-Ai general motor pathway (Farries 2001; Feenders et al. 2008). Despite some differences, the expression patterns of Sema3A, Npn-1, and plexin-A4 are similar between vocal learners and non-learners. Npn-1 and Sema3A expression were similar in or surrounding the arcopallium and conserved in the nXIIts. Even in the quail brain, at P0, many Npn-1-positive cells are broadly distributed in the NCL, although no aggregate HVC-like expression was detected and its expression is downregulated during development (Matsunaga & Okanoya unpubl. data, 2008). Therefore, the Sema3A/Npn-1 system may be conserved among avian species to form both NCL-Ai and HVC-RA connections. Some subtle genetic change, such as persistent or stronger Npn-1 and plexin-A4 expression or a change in the expression of downstream signaling molecules such as collapsin-response-mediator protein (CRMP) (Goshima et al. 1995; Zhu et al. 2008) or G-proteins at the telencephalic level, might have happened during evolution. Diverse combinations of various ligand/receptor complexes or their downstream signaling mechanisms might have triggered the evolution of the avian vocal system.