SEARCH

SEARCH BY CITATION

Keywords:

  • brain-derived neurotrophic factor;
  • chick;
  • critical period;
  • hyperpallium;
  • imprinting behavior

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

With the aim of elucidating the neural mechanisms of early learning, we studied the role of brain-derived neurotrophic factor (BDNF) in visual imprinting in birds. The telencephalic neural circuit connecting the visual Wulst and intermediate medial mesopallium is critical for imprinting, and the core region of the hyperpallium densocellulare (HDCo), situated at the center of this circuit, has a key role in regulating the activity of the circuit. We found that the number of BDNF mRNA-positive cells in the HDCo was elevated during the critical period, particularly at its onset, on the day of hatching (P0). After imprinting training on P1, BDNF mRNA-positive cells in the HDCo increased in number, and tyrosine phosphorylation of TrkB was observed. BDNF infusion into the HDCo at P1 induced imprinting, even with a weak training protocol that does not normally induce imprinting. In contrast, K252a, an antagonist of Trk, inhibited imprinting. Injection of BDNF at P7, after the critical period, did not elicit imprinting. These results suggest that BDNF promotes the induction of imprinting through TrkB exclusively during the critical period.

Abbreviations used
BDNF

brain-derived neurotrophic factor

Cont

control training group

GAD65

glutamate decarboxylase 65

GLAST

glutamate/aspartate transporter

HD

hyperpallium densocellulare

HDCo

core region of the HD

HDPe

periventricular part of the HD

IMM

intermediate medial mesopallium

LTP

long-term potentiation

Ni

nidopallium

NR2A

NMDA receptor subunit 2A

NR2B

NMDA receptor subunit 2B

P

post-hatching

PBS

phosphate-buffered saline

PFA

paraformaldehyde

PS

preference score

rBDNF

recombinant BDNF

Tra

training group

VGLUT2

vesicular glutamate transporter 2

VW

visual Wulst

Imprinting behavior in precocial birds is a good model for learning, memory, and social behavior during infancy (Lorenz 1937). The neural mechanisms of this behavior have been studied extensively using newly hatched chicks (Horn 2004). When baby chicks see a moving two-dimensional object for the first time within a few days of hatching, they visually learn its characteristics and follow it (Maekawa et al. 2006). The neural circuit responsible for both object recognition and subsequent memory, which is therefore indispensable for imprinting, has been described recently (Nakamori et al. 2010): visual cues are recognized and processed in the rostral region of the telencephalon, the visual Wulst (VW), and then the information is conveyed to the intermediate medial mesopallium (IMM) where imprinting memory is stored (Horn 2004). Neurons in the ventroposterior part of the VW, the core region of hyperpallium densocellulare (HDCo), send long axons to cells in the periventricular region of the HD (HDPe), and then neurons in the HDPe project to the IMM. The HDCo is indispensable for visual imprinting and a key region that controls the activity of this VW–IMM circuit. Activation of HDCo neurons expressing NR2B-containing NMDA receptors induces the refinement and consolidation of the VW–IMM circuit (Nakamori et al. 2010). Thus, elucidation of the activation mechanism of HDCo neurons may lead to a clearer understanding of juvenile learning.

As seen in rodent hippocampal neurons during spatial learning tasks (Morris 1989), the increased activity of HDCo neurons in the chick may result from an increase in synaptic efficacy, which correlates with long-term potentiation (LTP), the cellular basis of memory. Several molecules have been implicated in the process of synaptic modification, which is potentially related to the induction of LTP. We have shown that cholecystokinin neurons are indispensable for the acquisition of imprinting memory, probably through the strengthening of appropriate synapses in the VW and/or HD (Maekawa et al. 2007). Other candidate molecules are neurotrophins, including brain-derived neurotrophic factor (BDNF), nerve growth factor, neurotrophin-3, and neurotrophin-4/5. BDNF plays an important role in the induction and maintenance of mouse hippocampal LTP (Korte et al. 1995, 1996; Patterson et al. 1996; Pozzo-Miller et al. 1999). In genetically modified murine models, BDNF and its high-affinity receptor TrkB are involved in hippocampus-dependent learning (Linnarsson et al. 1997; Minichiello et al. 1999; Liu et al. 2004). However, these studies focused on learning without an apparent critical period, and they did not address juvenile learning. Consequently, it is largely unknown whether neurotrophins are involved in learning during infancy, such as imprinting.

In this study, we studied the role of BDNF in imprinting behavior in chicks. We focused on the HDCo region, where synaptic modification is important for the establishment of imprinting. We first examined the location and timing of BDNF and TrkB mRNA expression relative to the critical period and imprinting. Second, using immunohistochemistry with an anti-phosphotyrosine antibody, we examined whether neurons expressing TrkB in the HDCo are activated after imprinting training. Then we tried to clarify the role of BDNF in imprinting memory by injecting recombinant BDNF or Trk antagonist into the HDCo region before training.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Animals

Fertilized eggs of White Leghorn chickens (Gallus gallus domesticus) were obtained from local suppliers (Akebono Farm, Hiroshima, Japan, and Nihon Layer, Gifu, Japan) and incubated at 37.7°C under moderate moisture and quasi-constant darkness. After hatching, the chicks were kept in groups in the same incubator under quasi-constant darkness (Maekawa et al. 2006). This study was carried out in accordance with the Guidelines for the Treatment of Experimental Animals of Tokyo Medical and Dental University and Kitasato University. The experimental protocols described in this article were approved by the Animal Care and Use Committee for Tokyo Medical and Dental University and for Kitasato University.

Histology

Chicks were killed with an overdose of anesthetics and were perfused with 4% paraformaldehyde (PFA). The whole brains were post-fixed in the same fixative for 24 h at 4°C, cryoprotected by immersing in 30% sucrose for 48 h, embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek Japan, Tokyo, Japan), and frozen in powdered dry ice. Sagittal sections of the left hemisphere, cut at 20- to 40-μm thickness, were made with a cryostat (CM1900; Leica, Nussloch, Germany). We analyzed the left hemisphere exclusively, because it has been suggested that NMDA receptors in the left IMM play an important part in imprinting (McCabe and Horn 1988; McCabe et al. 1992). In addition, we had found that restricted ablation of the left HDCo results in the impairment of imprinting, indicating dominance of the left over the right hemisphere in imprinting (Nakamori et al. 2010).

In situ hybridization was performed as described previously (Maekawa et al. 2007). The chick TrkB gene fragment (GenBank accession no. NM_205231; nt 1780–2870 bp), which contains the intracellular tyrosine kinase domain, was amplified from chick brain cDNAs using the following primers: forward 5′-AGT CCT CTC CAT CAC ATC TC-3′ and reverse 5′-TGG AGT TCA GCG GCA GTT GA-3′. The amplified fragment was cloned into the pBSIISK (−) vector (Stratagene, La Jolla, CA, USA). Similarly, the chick glutamate/aspartate transporter gene fragment (GLAST, GenBank accession no. XM_425011; nt 1047–1752 bp; forward primer, 5′-GCT GTG ATC ATG TGG TAT GCT C-3′ and reverse primer, 5′-CAT TCT CCT CTA TCA CAG AAT TCC C-3′) was prepared and cloned. The chick BDNF fragment (GenBank accession no. NM_001031616; nt 1–741 bp; forward primer, 5′-ATA AAG CTT AGA GTG ATG ACC ATC CTT TTC C-3′ and reverse primer, 5′-TAT TCT AGA CTA TCT TCC CCT TTT AAT GGT T-3′) was prepared and cloned into a pGEM-T-easy vector (Promega, Madison, WI, USA). Plasmids containing vesicular glutamate transporter 2 (VGLUT2) and glutamate decarboxylase 65 (GAD65) cDNA were described previously (Maekawa et al. 2007). Gene-specific sense and antisense digoxigenin- or fluorescein-labeled cRNA probes were generated using a Roche RNA labeling kit (Roche Applied Science, Indianapolis, IN, USA).

For immunohistochemistry, sections were treated as described (Maekawa et al. 2007) and mouse anti-phosphotyrosine antibody (clone 4G10, 1 : 1000; Millipore Corporation, Billerica, MA, USA) was applied to the sections for 12–16 h at 4°C. After washing with phosphate-buffered saline (pH 7.0; PBS) with 0.5% triton X-100, the sections were reacted with a polymer reagent including peroxidase and goat anti-mouse IgG antibody (Dako Envision kit/HRP; DakoCytomation, Glostrup, Denmark) for 1 h at 25°C. Then, they were treated with 0.1% DAB to visualize peroxidase. For double fluorescence immunohistochemistry, mouse anti-phosphotyrosine (1 : 500) and rabbit anti-TrkB (794, sc-12, 1 : 500; Santa Cruz, CA, USA) antibodies, and Alexa 488- and 568-labeled secondary antibodies (A11008 and A11031, respectively, 1 : 1000; Invitrogen, Carlsbad, CA, USA) were used. Images were acquired using a confocal laser microscope (LSM710; Carl Zeiss, Oberkochen, Germany).

For cell counting, three or more chicks in each condition were used, and three sections that included the VW, HDPe, IMM, hyperpallium apicale, and nidopallium were selected per brain. The HDCo is located about 2.5–3.5 mm caudally from the rostral surface and 2.0–3.0 mm from the dorsal surface. The HDPe is located about 0–1.0 mm rostrally from the lateral ventricle and 2.0–3.0 mm from the dorsal surface. Therefore, with reference to the chick brain atlas (Puelles et al. 2007), the sections corresponding to L1.68–1.92 were selected for analysis. Images were acquired using a Zeiss microscope (Axio Imager A1; Carl Zeiss) equipped with Olympus DP25 digital camera and DP2-BSW software (Olympus Corporation, Tokyo, Japan). The light microscopy images were transferred to a graphics program (Adobe Photoshop CS4; Adobe Systems Incorporated, San Jose, CA, USA) in which the brightness and contrast were adjusted. To compare the number of positive cells between conditions, we first determined the threshold level in 256-shade grayscale, which adequately reflected the positive cells. This threshold level was kept constant for all samples designed to be compared. We manually counted the number of assemblies of black pixels in a 225 × 300-μm square (1040 × 1392 pixels) of the HDCo and a 178 × 534-μm square (680 × 2040 pixels) of the HDPe. Experimenters were not aware of the experimental condition of the chick from which the sections were derived. Considering the size of the cells, assemblies that contained fewer than 300 pixels were not counted. The number of cells involved in each assembly was confirmed and determined by the experimenter with reference to the original photograph. We compared the number of positive cells counted by the experimenter with the number counted as described above, examined the correlation, and confirmed insignificant differences between these two values.

Cell culture and transfection

HEK293T cells were grown in Dulbecco's modified Eagle medium (Sigma-Aldrich, MO, USA) supplemented with 10% fetal calf serum, 2% chick serum, 100 U/mL penicillin and streptomycin (Sigma-Aldrich), and 2 mM l-glutamine (Sigma-Aldrich). An IRES2–EGFP cassette was excised from the pIRES2–EGFP plasmid (PT3267-5; Clontech, CA, USA) with SmaI and NotI, and inserted into the EcoRV/NotI site of pCAGGS (Niwa et al. 1991), yielding pCX-IE. The chick BDNF fragment cloned into the pGEM-T-easy vector (see Histology section above) was excised with EcoRI, blunted, and inserted into the XhoI-cut and blunted pCX-IE, yielding pCX-BDNF-IE. HEK293T cells were then transfected with pCX-BDNF-IE or pCX-IE using FuGene6 (Roche). After 48-h culture, the cells were lysed in the extraction buffer [20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 100 mM NaF, 100 μM phenylarsine oxide, 10 mM Na4P2O4, 1% protease inhibitor cocktail (Sigma-Aldrich), and 1% phosphatase inhibitor cocktail (Sigma-Aldrich)] and stored at −80 °C until use.

Western blot

Chick brains were sagittally sliced at 1-mm thickness using a brain slicer (Muromachi Kikai, Tokyo, Japan) and the left HDCo was dissected from the slice, weighed, and then stored at −80°C until use. Frozen tissue was then homogenized in the extraction buffer, sonicated, centrifuged at 16 000 g for 5 min at 4°C, and the supernatant was collected. Protein content was determined using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Rockford, IL, USA) with bovine serum albumin as a standard. The protein (5 μg) was separated on 15% sodium dodecyl sulfate –polyacrylamide gel by electrophoresis, and blotted onto Immobilon-P membrane (Millipore). Then the membrane was soaked in 0.2% tween 20/tris-buffered saline containing 5% skim milk (Difco, Becton-Dickinson, Franklin Lakes, NJ, USA), and incubated with polyclonal anti-BDNF (N-20, 1 : 500; Santa Cruz, CA, USA) or anti-β actin (sc-47778, 1 : 1000; Santa Cruz) antibody at 4°C overnight. After three washes, the blot was reacted with a secondary antibody conjugated with horseradish peroxidase (1 : 1000). Signals were detected using SuperSignal West Femto (Thermo Fisher Scientific). Images were acquired using a detector (Image Quant LAS 4000 mini; GE Healthcare, Little Chalfont, UK) with Image Quant LAS 4000 Control Software.

Analysis of imprinting behavior

Behavioral imprinting experiments were done as previously described (Maekawa et al. 2006). Briefly, a chick was placed in a running wheel and an imprinting stimulus was presented to the chick by one of the two liquid crystal monitors (15-inch Flex Scan L367, EIZO; Nanao) on either side of the wheel, perpendicular to its axis. The images were generated using a visual stimulus generator system (VSG; Cambridge Research Systems) and each image bounced left and right horizontally on the screen. The chick's movements toward or away from the image presented on the display were recorded (Muromachi Kikai, Tokyo, Japan).

During the training, a chick was placed in the running wheel and exposed to a training image (a red circle, unless otherwise indicated) presented on the monitor. Depending on the experiment, duration of training was selected: 1 h (30 min presentation on one monitor followed by 30 min on the other monitor), 20 min, or 15 min (presentation only on one monitor). Training that lasts 1 h can induce imprinting behavior (Nakamori et al. 2010). For the control training, chicks were put into the apparatus for the same duration as the training group without any image on the black screen. We trained chicks once on P1 (24–48 h after hatching), unless otherwise specified. After the training, they were returned to the same incubator as before the training.

Imprinting performance was evaluated 2 h or 48 h later. The chicks were put into the running wheel again. After a 5-min adaptation period (black monitor, no image presentation), a red circle was presented for 5 min, followed by a blue square for 5 min on the same monitor. The direction and number of wheel revolutions were recorded. We calculated the preference score (PS) as described previously (Maekawa et al. 2006) using the following formula as an index of success of visual imprinting: PS = The number of wheel revolutions toward the display during 5 min/The total number of wheel revolutions during 5 min. Inactive chicks that rotated the wheel fewer than 22.5 revolutions during the 15-min evaluation period were excluded from the analysis (Maekawa et al. 2007). When the PS value for a red circle was significantly larger than chance (0.5) and the difference between the PS for the red circle and the blue square was regarded as statistically significant (p < 0.05), we judged the chick to be imprinted. These criteria are based on the fact that imprinting behavior is a following response specific for the imprinting stimulus.

In this study, the order of image presentation at evaluation was constant. This is based on our observation that showed this order did not affect imprinting performance. When the novel stimulus was presented before the imprinting stimulus, the PS values were 0.40 ± 0.29 and 0.73 ± 0.26, respectively (p < 0.01).

For the analysis of phosphotyrosine and TrkB expression, chicks were trained with a red square for 1 h. Evaluation was performed 1 h later with a red and blue square sequentially, and killed immediately.

Microinjection

Human recombinant BDNF (rBDNF; mature BDNF containing 119 amino acids) was a generous gift from Dainippon Sumitomo Pharma Co. Ltd. (Osaka, Japan). PBS or rBDNF (125 ng/μL in PBS), both containing 0.01% Evans blue dye and 1% bovine serum albumin (Sigma-Aldrich), was injected with a syringe (Hamilton Company, Reno, NV, USA; 2 μL/chick) using the methods described previously (Nakamori et al. 2010). Chick was anesthetized by brief exposure to diethyl ether and the head was held in a horizontal position and a free-hand injection was performed at the left HDCo or HDPe. Left HDCo was positioned 7 or 8 mm rostrally from bregma for P1 and P7 chicks, respectively, 2 mm to the left of midline of the skull, and at a 2-mm depth from the skull surface. The left HDPe was positioned 2.5 mm rostrally from bregma, 2 mm to the left of the midline, and at a 2-mm depth from the skull surface for P1 chicks. These positions were determined by geographic features on the skull surface seen through the skin. The depth of the injection was governed by a plastic sleeve on the 26s gauge needle (Hamilton). Twenty minutes after the injection, the imprinting training for 15 min (weak training) or 1 h was performed for P1 and P7 chicks, respectively, and imprinting behavior was evaluated 2 h after the end of the training.

A Trk receptor family antagonist, K252a (ENZO Life Science, Farmingdale, NY, USA; 50 μM, 50% dimethyl sulfoxide in PBS, 47 ng/chick; Rattiner et al. 2004; Niculescu et al. 2008), or a control solvent, both containing 0.01% Evans blue dye was injected by the same method into the HDCo (2 μl/chick). Five minutes after the injection, imprinting training was performed for 20 min, and 48 h later, the imprinting performance was evaluated. It was shown previously that K252a injected in the brain is no longer effective after 48 h (Rattiner et al. 2004).

At the end of the experiment, the chicks were deeply anesthetized, the brain was dissected, and the blue-stained area in the brain was inspected in each slice. We excluded all data obtained from chicks that lacked staining with the blue dye in the target area, or that showed staining that extended to areas other than the target area.

Statistical analysis

All data in this article are expressed as mean ± SEM. The number of animals used is indicated in each figure or the legends. Normal distribution of the data was checked by chi-square test and confirmed unless otherwise specified below. After confirming equality of variances by the Bartlett test, we used one-way anova, followed by Scheffé's F test to compare the values between conditions (Fig. 2e and g). For Fig. 2d, as data were not normally distributed, Kruskal–Wallis and Bonferroni tests were used. After confirming equality of variances by F-test, a Student's t-test was used to compare the number of positive cells between the control and training groups (Fig. 3d–f). A one-sample t test was used to examine whether or not the PS value was significantly larger or smaller than chance (0.5; indicated by #; Figs 4c, d and 5b). To compare the PS between the red circle and blue square, a paired t-test was used except for one case in which data were not normally distributed and a Mann–Whitney U-test was used (Fig. 4d control). Two-way repeated measure anova was also used to analyze the effects of variables (Fig. 4c, d and 5b). Differences were regarded as statistically significant at < 0.05 (*, #) and < 0.01 (**, ##).

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

More HDCo cells express BDNF mRNA during critical period for visual imprinting

Distribution of either BDNF or TrkB mRNA-expressing cells in the telencephalon was examined by in situ hybridization using P1 chicks, hatched and reared in quasi-constant darkness. The chick BDNF genomic DNA consists of at least four exons (exons I~IV) and the sequence that corresponds to the start codon resides in exon IV. Exon I, II, or III alternatively splices to exon IV, and as a consequence, at least four variants of mRNA can be generated (Yu et al. 2009). To detect all variants of mRNA, we designed and used a probe for exon IV. BDNF and TrkB (Fig. 1a upper and lower panel, respectively) mRNAs were expressed in the wide area of the pallium including VW, HDCo, HDPe, and IMM, which are reported to be important regions for imprinting memory. In the subpallium, BDNF mRNA was rarely detected, whereas TrkB mRNA was detected in the medial striatum. No signal was obtained by using BDNF and TrkB sense probes (data not shown). Double in situ hybridization revealed that all BDNF mRNA-positive cells were TrkB mRNA positive (Fig. 1b), and about 10% of TrkB mRNA-positive cells were also positive for BDNF mRNA. BDNF mRNA-expressing cells were positive for VGLUT2 mRNA, a marker of excitatory neurons, but negative for both GAD65 mRNA, a marker of inhibitory neurons, and GLAST, a marker of astrocytes (Fig. 1c). These results show that in the VW, HDCo, HDPe, and IMM, BDNF mRNA-positive glutamatergic neurons were distributed and they also expressed TrkB mRNA.

image

Figure 1. BDNF and TrkB mRNA expression in the telencephalon of dark-reared P1 chicks. (a) Schemas (left) show distribution of BDNF or TrkB mRNA-positive cells in sagittal section. One dot indicates about 10 BDNF mRNA-positive or 60 TrkB mRNA-positive cells respectively. Box indicates the HDCo. HDCo, core region of the hyperpallium densocellulare; HDPe, periventricular part of the HD; IMM, intermediate medial mesopallium; M, mesopallium; Ni, nidopallium; VW, visual Wulst. Photographs (right) show BDNF or TrkB mRNA-positive cells in the HDCo. (b) All BDNF mRNA-positive cells (blue) in the HDCo coexpressed TrkB (red). White arrowhead, BDNF and TrkB mRNA double-positive cell; red arrowhead, TrkB mRNA single-positive cell. (c) All BDNF mRNA-positive cells (blue) expressed VGLUT2, a marker for glutamatergic neurons (red; left), but not glutamate decarboxylase 65 (GAD65) (red; middle) or GLAST (red; right). White arrowhead, BDNF and VGLUT2 mRNA double-positive cell; blue arrowhead, BDNF mRNA single-positive cell; red arrowhead, VGLUT2 (left), GAD65 (middle) or GLAST (right) mRNA single-positive cell. Scale bars: (a) 100 μm; (b, c) 10 μm.

Download figure to PowerPoint

In our experimental protocol, the critical period for imprinting was from P0.5 to P4 (Nakamori et al. 2010). To see whether BDNF or TrkB mRNA expression is related to the critical period, we studied the temporal changes in their expression from E19 chick embryos to P7 chicks incubated and reared in quasi-constant darkness (Fig. 2a). The cells expressing BDNF mRNA at substantial levels in the HD layer, which includes the HDCo and HDPe, appeared to be more numerous at P3 compared with E19 (Fig. 2b). Then we focused on the HDCo region because this is a key area linking VW and IMM, and because the activity of HDCo neurons is important for imprinting memory (Nakamori et al. 2010). As shown in Fig. 2c and d, the number of BDNF mRNA-positive cells was low at E19 (9.7 ± 2.7), and reached its maximum level at P0 (36.1 ± 5.3). During the critical period, the number of cells expressing BDNF mRNA remained rather high, but it gradually decreased with age and reached the lowest levels at P5 (7.7 ± 2.8) and P7 (4.5 ± 1.5). In contrast, the number of TrkB mRNA-positive cells was unaltered throughout this period (Fig. 2c, e; 193.0 ± 23.1 in E19, 168.7 ± 19.0 in P1, 183.9 ± 21.9 in P3 and 140.9 ± 18.4 in P5). Thus, our data indicate that the change in the number of BDNF mRNA-positive cells in the HDCo is closely related to the onset and closure of the critical period.

BDNF mRNA is initially translated into a 32-kDa BDNF precursor (proBDNF), and then pro-BDNF is proteolytically cleaved to form 14-kDa mature BDNF (Mowla et al. 2001). Mature BDNF activates TrkB, whereas pro-BDNF activates p75, a neurotrophin receptor, and their effects are opposite for cell survival and synaptic plasticity (Teng et al. 2005; Woo et al. 2005). Therefore, mature BDNF must be produced to facilitate the synaptic response. To test for the presence of mature BDNF, we used western blotting using anti-BDNF antibody (Fig. 2f). We first analyzed the lysate prepared from HEK293T cells transfected with a BDNF expression vector, and verified that the 14-kDa (mature) and 32-kDa (pro-BDNF) products were detected. Mature BDNF was detected in the HDCo at every time point from E19 to P7. The lysates of P0-P5 brains exhibited significantly higher expression levels of mature BDNF than other time points (Fig. 2f and g). These results show that BDNF mRNA is expressed at high levels in the HDCo during the critical period, and this produces higher levels of mature BDNF.

image

Figure 2. The number of BDNF mRNA-positive cells in the core region of the hyperpallium densocellulare (HDCo) was high during the critical period, notably at its onset. (a) The critical period starts at P0 and ends at P4. The brains were analyzed at a defined age between E19 and P7. (b) In the HD layer, BDNF mRNA-positive cells were more prominent at P3 (right) than at E19 (left). (c) Brain sections from chicks at indicated age showing the expression of BDNF or TrkB mRNA in the HDCo. (d) Comparison of the number of BDNF mRNA-positive cells in the HDCo across ages. The number of BDNF mRNA-positive cells was high during the critical period, and that of P0 was significantly higher than that of E19, P5, and P7. n = 4 for E19 and P7, n = 6 for others. (e) Comparison of the number of TrkB mRNA-positive cells in the HDCo across ages. The number of TrkB mRNA-positive cells showed no difference among the four groups. n = 3 for each group. (f) Western blot analysis of HDCo protein extracts prepared from chicks at indicated age (5 μg/lane). Blots were reacted with antibodies specific to BDNF (upper panel) or β actin (gray triangle; lower panel). The extracts prepared from BDNF-expressing cells (pCX-BDNF-IE; 500 ng/lane; lane C) were analyzed similarly to confirm the size of pro-BDNF (black triangle) and mature BDNF (open triangle). Mature BDNF and smaller amounts of pro-BDNF were detected in the HDCo prepared from E19 to P7 chicks. (g) Expression level of mature BDNF was normalized with the β actin level. n = 3 for each age. *p < 0.05; **p < 0.01. Scale bars: (b) 1 mm; (c) 100 μm.

Download figure to PowerPoint

Number of BDNF mRNA-positive cells in the HDCo was elevated 1 h after the imprinting training, and TrkB tyrosine residues were phosphorylated in the HD layer

Next, we studied whether the number of BDNF mRNA-positive cells increased in the HDCo and/or HDPe after imprinting training. One hour of training or control training (without image presentation) was performed in P1 chicks and they were processed for histology 1 h afterward (Fig. 3a). This training procedure, but not control training, is shown to induce imprinting behavior (Nakamori et al. 2010). Cells expressing BDNF mRNA in the HD layer appeared to be more prominent in the training group (Tra) than in the control training group (Cont) (Fig. 3b). The number of BDNF mRNA-positive cells in the HDCo in the training group was higher than that of the control training group (Fig. 3c and d; Tra, 28.4 ± 4.4; Cont, 16.9 ± 2.9; p < 0.05). However, such a difference was not observed in the HDPe (Fig. 3c and d; Tra, 79.6 ± 15.2; Cont, 68.5 ± 15.3; p = 0.62). All of the BDNF mRNA-positive cells expressed TrkB mRNA (Fig. 3e). As expected, the number of BDNF and TrkB mRNA double-positive cells increased significantly in the training group (Fig. 3f; Tra, 29.8 ± 1.8; Cont, 16.1 ± 1.4; p < 0.01), but the total number of TrkB mRNA-positive cells remained unchanged (Tra, 161.8 ± 7.4; Cont, 153.5 ± 4.4; p = 0.33). Phosphotyrosines were detected in the HD layer in three of six trained chicks, and in none of the five control trained chicks (Fig. 3g). When double immunocytochemistry was performed, some of TrkB-positive signals colocalized with anti-phosphotyrosine in TrkB-positive cells, indicating that a subpopulation of TrkB was phosphorylated at tyrosine residues (Fig. 3h). These results indicate that a subset of TrkB mRNA-positive cells in the HDCo is activated and is induced to express BDNF mRNA by the training.

image

Figure 3. After the imprinting training, BDNF mRNA-positive cells in the core region of the hyperpallium densocellulare (HDCo) increased in number and phosphorylated TrkB was detected. (a) Chicks were trained with a red circle presented on the screen (training, Tra) or exposed to the screen without any image (control training, Cont) for 1 h. Chicks were killed 1 h after each training. (b) Sections from Tra and Cont chick brains showing the expression of BDNF mRNA. In the HD layer, BDNF mRNA-expressing cells were more prominent in Tra than Cont. Box indicates the HDCo. (c) Brain sections from Tra and Cont chicks showing the expression of BDNF mRNA in the HDCo (left) and periventricular part of the HD (HDPe) (right). (d) Comparison of the number of BDNF mRNA-positive cells between Tra and Cont. The number of BDNF mRNA-positive cells was higher in the HDCo of Tra compared to Cont, but a difference was not found in the HDPe. (e) All BDNF mRNA-positive cells (blue) in the HDCo coexpressed TrkB mRNA (red). Arrowhead, double-positive cell. (f) The number of BDNF and TrkB mRNA double-positive cells in the HDCo of Tra was higher than that of Cont, whereas that of total TrkB mRNA-positive cells did not differ between Tra and Cont. (g) Brain sections from Tra and Cont chicks showing the expression of phosphotyrosine. For each condition, left panels are in lower magnification, and the right panels are enlargements of the HDCo region. In the HD layer, phosphotyrosine-positive cells were prominent in Tra, but they were not detected in Cont. (h) Double-labeling of HDCo cells with antibodies specific to TrkB (green) and phosphotyrosine (magenta). Merged confocal image revealed that the phosphotyrosines localize with TrkB (white). N is the number of animals used in each group. *p < 0.05; **p < 0.01. Scale bars: (b) and left panels in (g), 1 mm; (c) and right panels in (g), 100 μm; (e) 10 μm; (h) 2 μm.

Download figure to PowerPoint

Human recombinant BDNF injection into the HDCo before training facilitated imprinting during the critical period

Elevated expression of BDNF mRNA and mature BDNF during the critical period suggests that the secretion of a substantial level of BDNF from HDCo cells during the early phase of imprinting training is necessary for chicks to be imprinted. To test this idea, we microinjected rBDNF into the HDCo (Johnston et al. 1999) at P1, and then examined its effect on imprinting. For this purpose, a ‘weak’ training protocol was used, which normally does not induce imprinting behavior (Fig. 4a). Imprinting was observed in the rBDNF group (Fig. 4b and c; PS for red circle, 0.74 ± 0.05; < 0.01 against PS = 0.5; for blue square, 0.36 ± 0.08; < 0.05 between red and blue), suggesting a facilitatory effect of BDNF on imprinting. The main effect of the presented images was significant [F(2, 20) = 13.147, p < 0.01], and a significant interaction was detected between the injection and the presented images [F(2, 20) = 3.834, p < 0.05]. On the other hand, rBDNF injection into the HDPe did not affect imprinting (PS for red circle, 0.50 ± 0.03; p = 0.90 against PS = 0.5). Only the main effect of the presented images was significant [F(2, 26) = 15.301, p < 0.01]. To test the contribution of TrkB receptors to the formation of imprinting, we used a Trk tyrosine kinase inhibitor K252a, which antagonizes the BDNF effect by inhibiting receptor tyrosine kinase autophosphorylation. Evaluation of imprinting was performed 48 h after the training to completely exclude the effect of K252a at the evaluation (Fig. 4a). As expected, injection of K252a into the HDCo before training impaired imprinting behavior (Fig. 4d; PS for red circle, 0.48 ± 0.04). Significant main effects of injection and of the presented images were detected [injection, F(1, 24) = 5.902, p < 0.05; images, F(2, 24) = 26.504, p < 0.01]. Thus, our data suggest that BDNF in the HDCo during the early phase of training promoted the induction of imprinting through TrkB.

image

Figure 4. Recombinant BDNF infusion into the core region of the hyperpallium densocellulare (HDCo) at P1 facilitated imprinting, whereas K252a infusion inhibited imprinting. (a) Human recombinant BDNF (rBDNF) or vehicle (control) was injected 20 min before the weak training at P1. K252a or vehicle was injected 5 min before imprinting training at P1 (using a red circle as the stimulus). The evaluation was performed 2 or 48 h after the end of training. (b) Sagittal brain sections of P1 chicks showing successful injection into the HDCo (left, arrow) or periventricular part of the HD (HDPe) (right, arrow). Scale bar: 2 mm. (c) Injection of rBDNF into the HDCo facilitated imprinting. When rBDNF was injected before the 15-min training, chicks showed significant preference for the imprinting stimulus (red circle), but not for the novel stimulus (blue square) at evaluation. This effect of BDNF was not observed with the injection into the HDPe. (d) Chicks injected with K252a into the HDCo before training failed to show imprinting behavior, unlike the control chicks. N is the number of animals used in each group. *p < 0.05; **p < 0.01; #p < 0.05 from PS = 0.5; ##p < 0.01 from PS = 0.5.

Download figure to PowerPoint

Injection of rBDNF could not facilitate imprinting after the critical period

We then tested whether the increased amount of mature BDNF in the HDCo was sufficient for imprinting even after the critical period. Vehicle or rBDNF was injected into the HDCo of P7 chicks, after the critical period, and 1 h of training was performed (Fig. 5a). There was no difference between vehicle- (PS for red circle, 0.51 ± 0.02; p = 0.64 against PS = 0.5) and rBDNF-injected groups (0.51 ± 0.02, p = 0.70) in imprinting (Fig. 5b). A significant main effect was only detected for the presented images [F(1, 12) = 35.186, p < 0.01]. Thus, elevated levels of BDNF in the HDCo are not sufficient to facilitate the acquisition of imprinting memory after the critical period.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Our results confirm the involvement of BDNF in visual imprinting in chicks. As the chicks hatched in a dark incubator and light exposure was constant across groups, the effect of BDNF described in this article cannot be related to experience (light)-dependent plasticity. We focused on the HDCo, an important relay site situated at the center of the VW–IMM circuit. During the critical period, BDNF mRNA expression in the HDCo was high compared with before or after the critical period. When imprinting training was performed, the number of BDNF mRNA-positive cells in the HDCo was significantly elevated 1 h after the training. The expression level of phosphotyrosine of TrkB was increased following training, indicating that TrkB in this circuit was activated. With the injection of rBDNF into the HDCo at P1, even weak training was sufficient to induce imprinting behavior. An inhibitor for Trk kinase, K252a, abolished imprinting. It has been shown that the memory of 1-day-old chicks requires BDNF, but not nerve growth factor or NT3 when assessed by passive avoidance learning (Johnston et al. 1999; Johnston and Rose 2001). Therefore, it is conceivable that K252a impaired imprinting through inhibition of BDNF signaling. These results indicate that for the establishment of imprinting, BDNF action is necessary during the early phase of training, and it may also be important after the training. However, BDNF injection before training was not sufficient to induce imprinting behavior at P7.

To our knowledge, this is the first report demonstrating that BDNF is indispensable for imprinting behavior, one of the best-known models for juvenile learning. The importance of BDNF in learning and memory has been shown in other models such as water maze and contextual fear conditioning in mice, and passive avoidance learning in chicks (Linnarsson et al. 1997; Johnston et al. 1999; Johnston and Rose 2001; Liu et al. 2004). In addition, it has been shown that BDNF is required both in the acquisition and consolidation phases of learning in a radial arm maze by using an antisense BDNF oligonucleotide in rats (Mizuno et al. 2000). Our present data are in accord with these results, but none of the previous reports focused on learning with a defined critical period. As we concentrated on imprinting behavior, we evaluated preference for the imprinted stimulus versus a novel one. As shown in Fig. 4c, BDNF injection also seems to reduce preference for the novel stimulus. It will be interesting to study the role of BDNF in fear response in future studies.

Our results indicate that an important property of the critical period is high BDNF mRNA expression in the HDCo, which enables the immediate secretion of BDNF upon training. The increase in BDNF mRNA after training suggests that secreted BDNF further stimulates HDCo cells to express more BDNF mRNA. This activity-dependent increase in BDNF mRNA is a well-known phenomenon and may contribute to the consolidation of memory through modification of synaptic structure and function (Poo 2001; Lu et al. 2008; Kuczewski et al. 2010). Thus, prolonged effects of BDNF may support neuronal plasticity during the critical period.

image

Figure 5. Effect of BDNF on imprinting was not observed at P7, after the critical period. (a) At P7, chicks were injected with recombinant BDNF (rBDNF) or vehicle in the core region of the hyperpallium densocellulare (HDCo) and trained with a red circle for 1 h. Two hours later, the evaluation was performed. (b) The chicks did not follow the imprint stimulus even when rBDNF was injected. This behavior was similar to the control group injected with vehicle (control). N is the number of animals used in each group. #p < 0.05 from PS = 0.5; ##p < 0.01 from PS = 0.5.

Download figure to PowerPoint

The number of BDNF mRNA-expressing cells in the HDPe did not change upon imprinting. Moreover, injection of rBDNF in the HDPe did not facilitate imprinting. These results indicate that BDNF does not regulate neural plasticity in the HDPe. Further studies are required to elucidate different features of the HDCo and HDPe. We exclusively analyzed the left hemisphere in this study because of reasons described above. Therefore, the role of BDNF in the right hemisphere in imprinting remains unknown.

Electrophysiological evidence shows that BDNF enhances LTP in the mouse hippocampal CA1 (Korte et al. 1995, 1996; Patterson et al. 1996; Pozzo-Miller et al. 1999). In a previous study, we found that the activation of NR2B-containing NMDA receptors in HDCo cells is required for imprinting learning (Nakamori et al. 2010). BDNF may be involved in this process, because it has been shown to rapidly phosphorylate NR1 and NR2B subunits in the cortical post-synaptic density, as well as increase activity of NR2B-containing NMDA receptors (Suen et al. 1997; Lin et al. 1998; Levine and Kolb 2000). Thus, LTP may occur in the chick HDCo cells by imprinting training, and BDNF might facilitate LTP as in the mouse hippocampus.

After closure of the critical period, we could not induce visual imprinting by 1 h training combined with infusion of rBDNF. The HDCo cells may no longer have the capacity to respond to BDNF stimulation. It has been shown that the recruitment of TrkB into the lipid rafts is necessary for BDNF signaling (Suzuki et al. 2004). Possible reasons for this unresponsiveness may be a change in the property of the extracellular matrix, or the maturation of inhibitory neurons that prevent this process (Berardi et al. 2003; Dityatev and Schachner 2003; Hensch 2005; Jiang et al. 2005). Dissection of molecules involved in the establishment of imprinting may further clarify the mechanism of juvenile learning, which has an obvious critical period.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

This work was supported by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan. We thank Dainippon Sumitomo Pharma Co. Ltd. for human recombinant BDNF, Kitasato University School of Medicine for the experimental facility, and laboratory members of Tokyo Medical and Dental University and Kitasato University for technical help and discussion. The authors declare no conflicts of interest.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References