SEARCH

SEARCH BY CITATION

Keywords:

  • GABA;
  • central nervous system;
  • in situ hybridization

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

γ-aminobutyric acid (GABA)ergic neurons in the central nervous system regulate the activity of other neurons and play a crucial role in information processing. To assist an advance in the research of GABAergic neurons, here we produced two lines of glutamic acid decarboxylase–green fluorescence protein (GAD67-GFP) knock-in mouse. The distribution pattern of GFP-positive somata was the same as that of the GAD67 in situ hybridization signal in the central nervous system. We encountered neither any apparent ectopic GFP expression in GAD67-negative cells nor any apparent lack of GFP expression in GAD67-positive neurons in the two GAD67-GFP knock-in mouse lines. The timing of GFP expression also paralleled that of GAD67 expression. Hence, we constructed a map of GFP distribution in the knock-in mouse brain. Moreover, we used the knock-in mice to investigate the colocalization of GFP with NeuN, calretinin (CR), parvalbumin (PV), and somatostatin (SS) in the frontal motor cortex. The proportion of GFP-positive cells among NeuN-positive cells (neocortical neurons) was approximately 19.5%. All the CR-, PV-, and SS-positive cells appeared positive for GFP. The CR-, PV, and SS-positive cells emitted GFP fluorescence at various intensities characteristics to them. The proportions of CR-, PV-, and SS-positive cells among GFP-positive cells were 13.9%, 40.1%, and 23.4%, respectively. Thus, the three subtypes of GABAergic neurons accounted for 77.4% of the GFP-positive cells. They accounted for 6.5% in layer I. In accord with unidentified GFP-positive cells, many medium-sized spherical somata emitting intense GFP fluorescence were observed in layer I. J. Comp. Neurol. 467:60–79, 2003. © 2003 Wiley-Liss, Inc.

Information processing in the central nervous system is based on the balance of the actions of excitatory and inhibitory neurons. Most excitatory neurons use glutamate as their transmitter, whereas most inhibitory neurons use γ-aminobutyric acid (GABA) and some use glycine as their transmitter. GABAergic neurons are organized into highly elaborated neuronal circuits, as are glutamatergic neurons. To investigate the distribution of GABAergic neurons and their processes, antiserum against the GABA synthetic enzyme glutamic acid decarboxylase (GAD) has been used for immunohistochemical analysis (McLaughlin et al., 1975a, b; Ribak et al., 1976). Immunohistochemistry with an antibody directed against GABA has also been used to visualize the GABAergic neurons (Storm-Mathisen, et al., 1983; Hodgson et al., 1985). The availability of antisera to various peptides and calcium binding proteins contained in GABAergic neuron subtypes has made it possible for various research groups to perform routine analyses of GABAergic neuron subtypes (Jande et al., 1981; Celio and Heizmann, 1981; Baimbridge and Miller, 1982; Hendry et al., 1984; Somogyi et al., 1984; Jones and Hendry, 1986). The findings of the immunohistochemical studies in combination with various other techniques have made it possible to understand the rough framework of the distribution and morphology of GABAergic neurons in the cerebral cortex (Freund and Buzsaki, 1996; Gupta et al., 2000).

However, to achieve a more precise understanding of the distribution and morphology of GABAergic neuron, additional new techniques are required. Moreover, as studies of the development or electrophysiological characteristics of GABAergic neurons progress, the necessity for visualizing GABAergic neurons in living tissues becomes more urgent. The most popular method of visualizing living cells in tissues or in vitro is to introduce the cDNA encoding a green fluorescent protein (GFP) obtained from a jellyfish into the cell or organism of interest. The GFP living color system has indeed been applied to GABAergic neuron studies, and transgenic mice harboring GFP cDNA whose expression is regulated by GAD67 promoter (Oliver et al., 2000), potassium channel promoter (Metzger et al., 2002), or calcium binding protein promoter (Meyer et al., 2002) have been produced. These transgenic mice have been useful for visualizing GABAergic neurons and monitoring the dynamic activity of GABAergic neurons in living organisms. However, because the DNA construct for GFP expression was inserted randomly into the mouse genome, the accuracy of GFP expression in GABAergic neurons or in a subtype of GABAergic neurons may be different from case to case, depending on the size of the promoter and enhancer region connected to the GFP cDNA and the location where the DNA construct is integrated in the mouse genome. To overcome these drawbacks, we produced a line of mice in which GFP cDNA is inserted between the GAD67 5′ flanking region and the GAD67 codon start by using a gene-targeting protocol, similar to other studies (Asada et al., 1997; Fuchs et al., 2001). We used the GAD67 promoter to reveal GABAergic neurons with GFP expression, because GABAergic neurons become positive for GAD67 from early developmental stages (DeDiego et al., 1994), while they become positive for vesicular GABA transporter after maturation (Chiu et al., 2002).

First, we describe whether GFP expression is accurately regulated in the knock-in mouse brains by the effects of the GAD67 promoter, enhancer, and suppressor. Second, we describe the distribution of GFP in the adult knock-in mouse brains and in the knock-in mouse embryos, because the GAD67-GFP knock-in mouse will be an important tool for the developmental and neurophysiological studies of the GABAergic neurons in various brain regions.

Finally, we examined the colocalization of GFP with NeuN, calretinin (CR), parvalbumin (PV), and somatostatin (SS) in the frontal motor cortex and estimated the proportion of GFP-positive cells among NeuN-positive cells (neocortical neurons) and CR-, PV-, and SS-positive cells among GFP-positive cells.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The experiments were conducted in accordance with the regulations of animal care of the Institute of Laboratory Animals, Faculty of Medicine, Kyoto University and National Institute for Physiological Science.

GAD67-GFP knock-in mice

The generation of GAD67-GFP knock-in mice will be described elsewhere. In brief, a cDNA-encoding enhanced GFP (EGFP; ClonTech, Palo Alto, CA) was targeted to the locus encoding GAD67 using homologous recombination (Fig. 1A). Homologous recombinant ES cells were used to generate chimeric male mice by 8-cell-stage injection. GAD67-GFP knock-in mice were obtained by breeding the chimeric male mice with C57BL/6 or ICR females. Heterozygous progeny were backcrossed to C57BL/6 and referred to as GAD67-GFP mice. These GAD67-GFP mice retained the EGFP cDNA and the loxP-flanked PGK-neo cassette in the GAD67 locus. The loxP-flanked PGK-neo cassette was used as a positive selection marker for screening homologous recombinant ES cells. However, the PGK-neo cassette may affect the GFP expression in either by reducing or enhancing it. To examine the effect of the PGK-neo cassette on the GFP expression, we generated GAD67-GFP mice lacking the cassette by mating the GAD67-GFP mice with the CAG-cre transgenic mice (Sakai and Miyazaki, 1997). In the progeny of the mating of these two strains of mice, the loxP-flanked PKG-neo cassette was excised in vivo. Offspring with germ line excision of the loxP-flanked PGK-neo cassette were identified by PCR using the following oligonucleotides: primer 1 (5′-TGAAGAACGAGATCAGCAGCCTCTGT-3′) corresponding to 3′ terminus of the neomycin-resistance gene and primer 2 (5′-GCTCTCCTTTCGCGTTCCGACAG-3′) corresponding to intron 1 of the GAD67 gene (Yanagawa et al., 1997). The GAD-GFP mice without the PKG-neo cassette were referred to as GAD67-GFP (Δneo) mice. Because the knock-out of both GAD67 alleles is lethal at birth (Asada et al., 1997), mice heterozygous for the altered GAD67 allele were used for all the observations in this study.

thumbnail image

Figure 1. Generation of glutamic acid decarboxylase–green fluorescence protein (GAD67-GFP) (Δneo) mice. A: Schematics of the wild-type, targeted, and recombinant alleles of GAD67. The original GAD67-GFP knock-in mice retained a loxP-flanked neomycin-resistance cassette (PGK-Neo). The knock-in mice were bred with CAG-Cre transgenic mice to eliminate the neomycin-resistance cassette from the GAD67 locus. The recognition sites of EcoRI (E), HindIII (H), and KpnI (K) are indicated. B: Immunodetection of GFP protein. Lysate (10 μg of protein) from the cerebrum was subjected to electrophoresis. Lane 1, lysate from hetero GAD67-GFP mouse; lane 2, lysate from hetero GAD67-GFP (Δneo) mouse. The arrow indicates the band of GFP at 28 kDa. C: GFP fluorescence in a GAD67-GFP mouse. D: GFP fluorescence in a GAD67-GFP (Δneo). The section from the GAD67-GFP (Δneo) mouse emitted more GFP fluorescence than that from the GAD67-GFP mouse. All the settings for confocal microscopy were the same for taking the two photomicrographs. No changes of the contrast or brightness of photographs were made during the figure preparation. E: γ-Aminobutyric acid contents in the forebrain of two knock-in mouse strains at two different stages shown in the ratio to those of wild-type mouse. Columns indicate the ratios in percentage, and bars on top of the columns indicate standard deviations (see also Table 1). Scale bar = 50 μm in C (applies to C,D).

Download figure to PowerPoint

Table 1. GABA Content in the Mouse Brains1
 Newborn (nmol/mg)Ratio (%) t test6–7 weeks old (nmol/mg)Ratio (%) t test
  • 1

    The difference between the GABA content of 7-week-old GAD67-GFP mouse brain and that of 7-week-old wild-type mouse brain was not significant (t test, P = 0.09). The difference between the GABA content of 6-week-old GAD67-GFP (Δneo) mouse brain and that of 7-week-old GAD67-GFP mouse brain was not significant either (t test, P = 0.05). However, the other differences were considered significant. GABA, γ-aminobutyric acid; GAD, glutamic acid decarboxylase; GFP, green fluorescence protein.

GAD67-GFP(Δneo)21.7 ± 1.8 (n = 6)61%19.9 ± 2.0 (n = 6)84% P < 0.05
Wild type35.5 ± 2.6 (n = 7)P < 0.00123.7 ± 2.1 (n = 5)P < 0.05
GAD67-GFP (Δneo)/GAD67-GFP 90% 90%
  P < 0.05 P = 0.05
GAD67-GFP24.7 ± 2.1 (n = 7)68%26.9 ± 2.1 (n = 6)93%
Wild-type36.3 ± 2.3 (n = 8)P < 0.00129.0 ± 1.7 (n = 6)P = 0.09

Western blot analysis

We prepared protein lysates from the cerebrum of the mice as described previously (Asada et al., 1996). We electrophoresed the lysates (10 μg of protein) on 10% polyacrylamide gels and performed Western blot analysis as described previously (Yanagawa et al., 1997), using an anti-GFP rabbit antibody (1/50; Tamamaki et al., 2000).

Measurement of GABA

Samples from mouse brains were homogenized with 0.1 M perchloric acid to extract amino acids and precipitate proteins (Fisher et al., 2001). The homogenates were centrifuged at 10,000 g for 15 minutes at 4°C and then the supernatant was neutralized with 0.1 M sodium carbonate. After addition of sodium carbonate, the samples were filtered by using Ultrafree-MC (Millipore Corp., Bedford, MA). Amino acids in the samples were derivatized with o-phthalaldehyde, and GABA concentrations were measured by using high performance liquid chromatography and fluorescence detection (CMA280, B.A.S., Japan; Asada et al., 1997; Stork et al., 2000). Protein concentrations were determined by using BCA protein assay reagent (Pierce, Rockford, IL) with bovine serum albumin as a standard.

GFP fluorescence observation

Ten adult GAD67-GFP mice and GAD67-GFP (Δneo) mice older than 3 months of age were used for the analysis of GFP-distribution in their brains. They were deeply anesthetized with sodium pentobarbital (50 mg/kg body weight) and fixed as follows: they were perfused transcardially with 30 ml of 5 mM sodium phosphate-buffered 0.9% (w/v) saline (PBS, pH 7.4) and then with 50 ml of 2% (w/v) formaldehyde, 75%-saturated picric acid, and 0.1 M Na2HPO4, pH 7.0 (adjusted with NaOH). After the mice were left thus for 2 hours, their brains were removed, immersed in PBS, and kept in a solution of 30% (w/w) sucrose in PBS for cryoprotection. The blocks were cut into 50-μm-thick frontal sections on a freezing microtome unless otherwise stated. Every sixth section was used for observation of GFP fluorescence.

Three heterozygous GAD67-GFP mouse embryos on embryonic day 15 (E15) were used to analyze the temporal regulation of GFP expression. They were obtained by Caesarean section from an anesthetized pregnant wild-type mouse that had been mated with a male GAD67-GFP mouse, and the embryos were perfused with the same fixative described above. The embryo heads were post-fixed with the same fixative for 2 hours. They were immersed in PBS and kept in a solution of 30% (w/w) sucrose in PBS for cryoprotection, and cut into 50-μm-thick frontal sections. After the observation of GFP fluorescence, all the serial sections of adult and embryo brains were counterstained with cresyl violet for observation of cytoarchitecture.

Fluorescence of GFP or Alexa 594 (see below) in the sections was observed and photographed with a fluorescence digital microscope (VB-6000; Keyence, Japan) at low magnification or with a confocal microscope (LSM Pascal; Zeiss, Germany) at high magnification.

Immunohistochemistry

Fifteen GAD67-GFP (Δneo) mice and GAD67-GFP mice were anesthetized, perfused with the same fixative or a fixative with glutaraldehyde for GABA immunohistochemistry (0.1% glutaraldehyde, 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4), post-fixed with the same fixative for 2 hours, and then sectioned at 25-μm thickness as described above. The sections were incubated overnight with one of the following antibodies: an anti-calretinin (CR) rabbit antibody (1/2,000; Chemicon, Temecula, CA), an anti-GABA rabbit antibody (1/2,000 for Alexa 594-conjugated streptavidin detection, 1/1,000,000 for detection with TSA; Sigma, St. Louis, MO), anti-GAD67 rabbit antibody (1/1,000 for Alexa 594-conjugated streptavidin detection, 1/500,000 for TSA detection; Chemicon), an anti-GFP guinea pig antibody (1/300; Tamamaki et al., 2000), an anti-NeuN mouse monoclonal antibody (1/500; Chemicon), an anti-PV mouse monoclonal antibody (1/1,000; Sigma), or an anti-SS rabbit antibody (1/2000; Peninsula Labs, Belmont, CA). Then the sections were incubated for 2 hours with 10 μg/ml biotinylated anti-rabbit IgG or biotinylated anti-mouse IgG donkey antibody (Jackson Laboratories, West Grove, PA). The incubation was carried out at room temperature in PBS containing 0.3% (v/v) Triton X-100, 0.25% (w/v) λ-carrageenan, and 1% (v/v) normal donkey serum, followed by a rinse with PBS. The sections were further incubated for 2 hours with Alexa 594-conjugated streptavidin (Molecular Probes, Eugene, OR) in the same incubation buffer. In the immunohistochemical detection of GAD67 and GABA, GFP was also detected by immunohistochemistry. After incubation with primary antibody, the sections were further incubated for 2 hours with 10 μg/ml anti-guinea pig IgG conjugated with Alexa 488 (Molecular Probes) and biotinylated anti-rabbit IgG donkey antibody, and then for 2 hours with ABC solution (Vector Laboratories, Burlingame, CA). Finally sections were reacted with TSA Cyanine 3 (Perkin Elmer, Oak Brook, IL). The sections were mounted onto gelatinized glass slides and covered with 50% (v/v) glycerol and 2.5% (v/v) DABCO in PBS for fluorescent microscopic observation.

In situ hybridization histochemistry

A DNA fragment corresponding to a portion of the mouse GAD67 cDNA (nucleotides 43-661, GenBank Y12257) was cloned into the vector pBluescriptII SK(+) (Stratagene, La Jolla, CA). Using these plasmids as templates, sense and antisense single-strand RNA probes were synthesized with a digoxigenin labeling kit (Roche Diagnostics, Japan).

Wild-type adult mice and embryos were perfused with a fixative containing 4% formaldehyde and 0.1 M phosphate buffer (pH 7.4). The brains were removed, post-fixed with the same fixative for 2 hours, immersed in diethylpyrocarbonate-treated 30% sucrose solution in PBS overnight, and then sectioned at 20 μm thickness for adult brains and at 15 μm for embryo brains. Adult brain sections were processed for in situ hybridization under free-floating conditions.

The free-floating sections were washed in PBS for 5 minutes and acetylated in freshly prepared 0.25% (v/v) acetic anhydrate in 0.1 M triethanolamine for 10 minutes with vigorous shaking. After rinsing in PBS for 5 minutes twice, the sections were hybridized with 500 ng/ml digoxigenin-labeled sense or antisense RNA probes for GAD67 in a mixture of 50% (v/v) formamide, 5× standard saline citrate (SSC), 5× Denhardt's solution, 250 μg/ml yeast tRNA, and 500 μg/ml salmon sperm DNA for 16 to 24 hours at 70°C. After two washes in 0.2× SSC for 5 minutes each at 70°C, some sections were incubated with 10 μg/ml RNase A for 30 minutes at 37°C, and subjected to two high-stringency washes at hybridization temperature in 0.2 × SSC for 1 hour. Because there was no difference in the signal distribution pattern observed with and without RNase treatment, RNaseA treatment was omitted in most cases. Subsequently, the sections were incubated with 1/2,000-diluted alkaline phosphatase-conjugated anti-digoxigenin antibody Fab fragment (Roche) in PBS containing 0.3% (v/v) Triton-X 100, 0.25% (w/v) λ-carrageenan, and 1% (v/v) normal donkey serum (PBS-XCD), and the bound phosphatase was visualized by reaction for 24 to 36 hours with 0.375 mg/ml nitroblue tetrazolium and 0.188 mg/ml 5-bromo-4-chloro-3-indolyl phosphate in 0.1 M NaCl, 5 mM MgCl2 and 0.1 M Tris-HCl (pH 9.5). The sections were mounted onto the 3-aminopropyltriethoxy silane–coated glass slides (Matsunami, Japan), washed in water, dried, cleared in xylene, and cover-slipped.

Photographs

Digital images of sections obtained from the GAD67-GFP knock-in mice were processed by using Adobe Photoshop. The contrast and brightness of figures except Figure 1C,D was adjusted as needed for each section to show the GFP-positive structures clearly. Cytoarchitectonic areas were generally determined based on the atlas of Franklin and Paxinos (1997). Boundaries between neocortical layers in the fluorescence photomicrographs were determined on careful comparison with counterstained sections with cresyl violet obtained from similar points.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Two strains of GAD67-GFP knock-in mice

To evaluate the expression of GFP in GAD67-GFP mice and GAD67-GFP (Δneo) mice (Fig. 1A), Western blot analysis was performed using an anti-GFP antibody (Fig. 1B). The antibody detected a band of GFP of the expected size of 28 kDa in the lysates from both a GAD67-GFP mouse and a GAD67-GFP (Δneo) mouse. The band representing GFP in the lysate from the GAD67-GFP (Δneo) mouse was much denser than that from the GAD67-GFP mouse, which indicates that the cerebral cortex of the GAD67-GFP (Δneo) mouse contained more GFP than the cerebral cortex of the GAD67-GFP mouse. We also compared the GFP contents in the brains of the two strains of knock-in mouse directly by fluorescent microscopy and examined the correlation of the GFP contents with GAD67 immunoreactivities and GABA contents. Figure 1C,D shows green fluorescence observed in the neocortex of a GAD67-GFP mouse and that in a GAD67-GFP (Δneo) mouse, respectively. The sections were obtained by the same fixation procedure. Their digital images were printed without any change in either the brightness or the contrast. Comparison of these images revealed that the section obtained from the GAD67-GFP (Δneo) mouse emitted more intense fluorescence than the section obtained from the GAD67-GFP mouse did, whereas the green fluorescence emitted from the sections of wild-type mouse brain was at a background level. The same sections obtained from the two strains of GAD67-GFP mice and wild-type mice were processed for GAD67 or GABA immunohistochemistry in the same bottle. The GAD67 immunoreactivity of the section obtained from the GAD67-GFP (Δneo) mouse brains and the GAD67-GFP mouse brain seemed to be lower than that from the wild-type mouse (data not shown). Thus, we speculated that the deterioration of the GAD67 immunoreactivity correlated with the increase of GFP expression after deletion of neo cassette. To prove our speculation, we investigated the GABA contents in the forebrain of these mice with a high performance liquid chromatography (HPLC). The results of GABA contents measured by HPLC were shown in Table 1 and in Figure 1E by their ratios or the concentrations of GABA. The forebrain of newborn GAD67-GFP (Δneo) mouse and that of newborn GAD67-GFP mouse (including the cerebral cortex and the basal ganglia) contained GABA at the concentration of 21.7 ± 1.8 nmol/mg (n = 6) and 24.7 ± 2.1 nmol/mg (n = 7), respectively. The GABA contents at postnatal day (P) 0 in the two strains of knock-in mice were significantly lower than that in wild-type mice (t test, P < 0.001). Six weeks later, the GABA contents in the forebrain of GAD67-GFP (Δneo) mouse (19.9 ± 2.0 nmol/mg; n = 6) was still significantly lower than that of wild-type mouse (23.7 ± 2.1 nmol/mg; n = 5; t test, P < 0.05). However, the difference between the GABA content in the forebrains of GAD67-GFP (Δneo) mouse and that of wild-type mouse at 6 weeks old was small. Moreover, the difference between that of GAD67-GFP mouse (26.9 ± 2.1 nmol/mg; n = 6) and that of wild-type mouse (29.0 ± 1.7 nmol/mg; n = 6) at 7 weeks old was no longer significant (t test, P = 0.09). The difference between the GABA contents of the newborn GAD67-GFP (Δneo) mouse and that of the newborn GAD67-GFP mouse was also significant (t test, P < 0.05), but it became barely significant (t test, P = 0.05) 6–7 weeks later. The Western blotting study and the fluorescent microscopic observation revealed that removal of the PGK-neo cassette enhanced GFP expression in the GAD67-GFP (Δneo) mice. In return, as the GFP expression increased, GAD67 expression seemed to be reduced and the difference between the GABA contents in the forebrains of the two strains was significant at birth. However, the GABA contents in the brains of the two strains were increased as they grew. GAD67-GFP (Δneo) and GAD67-GFP mice were seizure-free. They exhibited normal growth, normal reproductive behavior, and no abnormality in the brains at macroscopic level.

Colocalization of GFP with GAD67 and GABA

We investigate the colocalization of GFP with GAD67 and GABA in adult male GAD-GFP mice. Although the GAD67-GFP mouse brains contained less GFP than GAD67-GFP (Δneo) mouse brains did, we used GAD67-GFP mouse brains for the GAD67 and GABA immunohistochemistry, because the GABA content in GAD67-GFP mouse brains was not significantly lower than that in wild-type mice. Thus, the GFP-positive cellular structures appeared in green by GFP fluorescence or by Alexa 488 fluorescence and GAD67- and GABA-immunoreactive structures in red by TSA cyanine 3. The level of green fluorescence depended on the amount of GFP accumulated in the somata, dendrites, axons, and axon terminals. Although both GFP and GAD67 are soluble proteins, their distribution in subcellular structures were different. We examined the difference in the GAD67- and GFP-distributions by confocal microscopic observation in the neocortex (Fig. 2A–F). GFP was present throughout the GFP-positive somata, including in their nuclei, whereas GAD67 was present preferentially in the perikarya of the GAD67-positive somata. The vast majority of the GFP-positive cells had GAD67 immunoreactivity in their perikarya (156 of 196). However, all the GFP-positive cells were not concluded as GAD67-positive. Some of them with GAD67 immunoreactivity close to that in the neuropil (arrows in Fig. 2A,C) were regarded as negative for GAD67. If we included them as GAD67-positive cells, almost all of the GFP-positive cells would be positive for GAD67. Most GAD67-positive cells were identified by GFP fluorescence higher than that in the neuropil. However, some of the GAD67-positive cells were counted as GFP-negative, because they had GFP immunoreactivity close to that in the neuropil. If we included them as GFP-positive cells, basically all the GAD67-positive cells were positive for GFP. Overall, we did not encounter any ectopic GFP expression in apparently GAD67-negative cells, whose red-fluorescence level was lower than that of the neuropil, or any lack of GFP-expression in apparently GAD67-positive neurons in the knock-in mouse. In a double-labeling study of GFP and GABA immunohistochemistry (Fig. 2G–I), most GFP-positive cells were regarded as positive for GABA (141 of 159), but some of them had GABA immunoreactivity similar to that of neuropil and regarded as GABA-negative (arrows in Fig. 2G,I). If we included them as GABA-positive cells, basically all the GFP-positive cells were positive for GABA. All the investigated GABA-positive cells seemed to have a certain amount of GFP immunoreactivity at least similar or higher than that of the neuropil.

thumbnail image

Figure 2. A–C: Double labeling of green fluorescence protein (GFP) immunoreactivity (green) and glutamic acid decarboxylase (GAD)67 immunoreactivity (red) in the neocortex. Arrowheads in A and C indicate the cells regarded as positive for GFP but negative for GAD67. D–F: Double labeling of GFP fluorescence and GAD67 immunofluorescence in the ventrolateral thalamic nucleus. GAD67 was distributed preferentially in the axon terminals, whereas GFP was distributed ubiquitously within axons. GFP and GAD67 colocalized in the axon terminals. G–I: Double labeling of GFP immunoreactivity (green) and GABA immunoreactivity (red) in the neocortex. Arrowheads in A and C indicate the cells regarded as positive for GFP but negative for GABA. Scale bars = 10 μm in A (applies to A–C,G–I); 10 μm in D (applies to D–F).

Download figure to PowerPoint

Distribution of GFP and GAD67 in the axons was investigated in the thalamus where we found many GFP-positive afferent fibers but fewer GAD67-positive cells (see below). GFP was distributed in the axons and axon terminal-like puncta (Fig. 2D,E) as well as in dendrites (Fig. 2A), whereas the GAD67 immunoreactivity was preferentially found in the axon terminal-like puncta (Fig. 2E,F). The double-labeling study of GFP and GAD67 revealed that GFP and GAD67 were preferentially colocalized in the perikarya and in the axon terminal-like puncta (Fig. 2A–F). In the rest of the subcellular structures, GFP and GAD67 would be present in different amounts, depending on the thickness of dendrites and that of axon fibers. GFP expressed under the control of the GAD67 promoter would be a good tracer for detecting somata, dendrites, and axon fibers of GAD67-positive GABAergic neurons but not a good tracer for detecting the intracellular distribution of GAD67 in the knock-in mouse brain.

Temporal regulation of GFP expression

GFP expression in the GAD67-GFP knock-in mouse seemed to be regulated accurately not only spatially but also temporally. Some panels in Figure 4 show negatives of the original fluorescent photomicrographs taken of sections of GAD67-GFP mice. In the E15 embryo head, GFP fluorescence was found not only in the brain but also in the ectodermal placodes of developing whiskers (Fig. 3). In frontal sections through the tooth buds, it was apparent that odontoblasts emitted GFP fluorescence (Fig. 4A–C). All these tissues have been reported to be positive for GABA and GAD67 during embryonic stages (Asada et al., 1997; Maddox and Condie, 2001). In sections through the ganglionic eminence, intense GFP fluorescence was seen to fill the area (Fig. 4D). An in situ hybridization study at E15 revealed a high level of GAD67 mRNA in the ganglionic eminence, especially in the subventricular zone (Fig. 4E). GFP-positive cells were also distributed in the neocortex (Fig. 4D,F) and the olfactory bulb (Fig. 4G). Each GFP-positive cell in the neocortex had a long, thick process and a short tail, which corresponded to the leading and trailing processes (Rakic, 1990) of tangentially migrating GABAergic neurons (Anderson et al., 1997; Tamamaki et al., 1997). The ventricular zone of the neocortex contained a small number of GFP-positive cells (Fig. 4F), which had an appearance typical of migrating cells and were regarded to be GABAergic neurons migrating into the ventricular zone from the intermediate zone (IZ) (Nadarajah et al., 2002). The other ventricular zone cells were negative for GFP and for GAD67 mRNA.

thumbnail image

Figure 4. Green fluorescence protein (GFP) fluorescence observed in serial frontal sections obtained from a glutamic acid decarboxylase (GAD)67-GFP knock-in mouse embryo at embryonic day (E) 15, and GAD67 in situ hybridization signal in a frontal section obtained from a wild-type mouse embryo brain at E15. A,B,D,F,G show negatives of original fluorescent light photomicrographs. A: GFP fluorescence in a frontal section through the tooth bud. B: GFP fluorescence in the odontoblast layer (Ob) at higher magnification. C: Counterstained odontoblast layer and enameloblast layer. The section shown in A and B was stained with cresyl violet. D: GFP fluorescence in a frontal section through the lateral ganglionic eminence (LGE). Arrows in D,E indicate the GFP-positive cells in the intermediate zone and marginal zone. The double arrow in D,E indicates the ventricular edge of the neocortex (Cx). E: In situ hybridization signal for GAD67 antisense probe in a frontal section through the lateral ganglionic eminence. F: GFP-positive cells in the neocortex. G: GFP-positive cells in the olfactory bulb. MZ, marginal zone; IZ, intermediate zone; V, ventricle. For other abbreviations , see list. Scale bars = 1 mm in A (applies to A,D,E); 100 μm in B (applies to B,C); 100 μm in G (applies to G,F).

Download figure to PowerPoint

thumbnail image

Figure 3. An embryonic day 15 embryo head. Arrows indicate green fluorescence protein–positive hair roots. Scale bar = 1 mm.

Download figure to PowerPoint

GFP distribution in the knock-in mice

The double-labeling study and the examination of GFP expression at an embryonic stage indicated that the GFP expression is regulated by GAD67 promoter properly in the knock-in mice and convinced us of the usefulness of the knock-in mice for the studies of GABAergic neurons in various brain regions. Thus, we describe the GFP distribution in the knock-in mouse brain for future use by other researchers.

Although brain sections obtained from GAD67-GFP (Δneo) mice emitted more intense fluorescence than those obtained from GAD67-GFP mice, the distribution pattern of GFP fluorescence in the GAD67-GFP (Δneo) mice was indistinguishable from that in GAD67-GFP mice in all parts of the brain. Therefore, we used GAD67-GFP mice to make a map of GFP distribution in the knock-in mouse brain. Figures 5 and 6 show negatives of the original fluorescent photomicrographs taken of sections of the brains of GAD67-GFP mice. Thus, areas with intense GFP fluorescence appear as dark areas in the figures.

thumbnail image

Figure 5. Green fluorescence protein (GFP) fluorescence in a parasagittal section (A) and frontal sections (B–L) obtained from glutamic acid decarboxylase (GAD)67-GFP knock-in mouse brains. All the figures are negatives of the original fluorescent light photomicrographs. For abbreviations , see list. Scale bars = 1 mm in A, C (applies to B–F), G (applies to G–J), K, L.

Download figure to PowerPoint

thumbnail image

Figure 6. Green fluorescence protein (GFP) fluorescence observed in various brain regions of the glutamic acid decarboxylase (GAD)67-GFP knock-in mouse. All the figures are negatives of the original fluorescent light photomicrographs. A: Cortex of the olfactory bulb. B: Neocortex. Arrowheads indicate the GFP-positive somata. The double arrowhead indicates a large GFP-negative hole, and the arrow indicates a small GFP-negative hole. C: Neocortex, white matter, and caudate putamen. D: Hippocampus. E: Caudate putamen. F: Globus pallidus. G: Reticular thalamic nucleus. H: Ventrolateral thalamic nucleus. I: Dorsal lateral geniculate nucleus. J: Superior colliculus. K: Lateral lemniscus. L: Cerebellar cortex. M: Substantia nigra pars reticulata. N: Ventral tegmental nucleus. O: Inferior olive. P: Outer edge of the inner granular layer of the retina. An arrow indicates the area out of focus. Q: Center of the inner granular layer of the retina. R: Inner edge of the inner granular layer of the retina. For other abbreviations , see list. Scale bars = 100 μm in A (applies to A,C,D), B (applies to B,E,F), G (applies to G,H,L), I (applies to I,J), K, M (applies to M–O), P (applies to P–R).

Download figure to PowerPoint

GFP fluorescence was found throughout the adult GAD67-GFP-mouse brain, from the olfactory bulb to the spinal cord, as shown in the serial frontal sections made from a mouse brain (Fig. 5B–L). The intensity of GFP fluorescence in the knock-in mouse brains varied from site to site. The strongest intensity was seen in the olfactory bulb. When the whole brain was observed under a fluorescence microscope, the fluorescence emitted from the olfactory bulb was most obvious, and that from the cerebellum and cerebral cortex followed. Thus, GFP fluorescence from the olfactory bulb was a good marker to identify knock-in mice. At birth, we could identify such mice by the GFP fluorescence emitted from the olfactory bulb through the meninges, the neurocranium, and the skin. Therefore, phenotyping of the knock-in mouse could be accomplished with a glance through a fluorescence microscope, and there was no need for genotyping of the knock-in DNA sequence. The GFP fluorescence was emitted from the granule cell layer and molecular layer of the olfactory bulb (Fig. 5A,B) (Shipley et al., 1995; Shepherd and Greer, 1998). The distribution was even clearer at higher magnification (Fig. 6A). It is well known that the olfactory glomeruli are the structures with GABAergic neuron processes (Kosaka et al., 1997); the relative intensity of GFP fluorescence was weaker here than in the periglomerular region and, furthermore, weaker than the granule cell layer. Thus, the glomerular layer appeared weak in GFP fluorescence.

In the frontal sections obtained from the caudal level to the olfactory bulb, the anterior part of the caudate putamen and olfactory tubercle emitted GFP fluorescence at moderate intensity. In the olfactory tubercle, the fluorescence of the islands of Calleja was clearly higher than the surrounding structures (Fig. 5C). The GFP fluorescence seen in the olfactory tubercle was continuous with that in the medial septum and that in the diagonal band. The caudate putamen contained medium-sized GFP-positive cells and had higher background fluorescence than the neocortex. The high background of this region was due to small GFP-positive cells and neuropil packed among the fascicles of the internal capsule (Fig. 6E). The globus pallidus emitted more GFP fluorescence among the fascicles of the internal capsule (Fig. 5D). The strong intensity of GFP fluorescence was due to the highly packed GFP-positive cells and GFP-positive neuropil (Fig. 6F).

The neocortex showed less GFP fluorescence than the area described above (Fig. 5C,D). At higher magnification, GFP-positive cell somata, dendrites, and axon terminal-like boutons could be identified in a thin confocal plane (Fig. 6B). The low fluorescence in the matrix suggests that non-GABAergic cells predominate in the neocortex, whereas GABAergic cells predominate in the caudate putamen and globus pallidus. At higher magnification, the non-GABAergic neurons appeared as empty holes, which were often surrounded by GFP-positive puncta. By counting GFP-positive somata (arrowheads in Fig. 6B) and large GFP-negative somata (double arrowhead in Fig. 6B) in areas of the same size as that shown in the Figure 6B, we tried to estimate the proportion of GFP-positive neurons among the neocortical neurons. However, we might have underestimated the number of GFP-positive neurons by regarding the somata with weaker GFP-fluorescence than the fluorescence of the neuropil as GFP-negative neurons. In addition, we might have overestimated or underestimated the GFP-negative neurons (the arrow in Fig. 6B) considered as glial cells. We estimated that the proportion of GFP-positive neurons among the neocortical neurons was 12% (12 ± 1.8%, n = 5), a far lower proportion than that estimated by another method (see below). The white matter contained GFP-positive cells (Fig. 6C), which correspond to the white matter neurons (Clancy et al., 2001), the remnant of the GABAergic neurons found in the IZ at the embryonic stage (Tamamaki et al., 1997, 2002). The hippocampus had weak GFP fluorescence, comparable to that in the neocortex. However, because of the stratified structures, GFP-positive cells in the pyramidal cell layer and the granule cell layer were obvious, especially in the hilus, even at low magnification (Figs. 5E,F, 6D).

The GFP fluorescence in the areas described above was distributed in the somata and in the neuropil (dendrites and axon branches). However, GFP fluorescence in the reticular thalamic nucleus was found predominantly in the large somata (Fig. 6G), whereas GFP fluorescence in the thalamic relay nuclei was found predominantly in the axons and terminal branches with boutons (Fig. 6H). Even if the GFP fluorescence was of similar intensity, the structures emitting the fluorescence might have been different. The rostral part of the thalamus showed fluorescence less than that of the caudate putamen (Fig. 5A,E) and did not show any GAD67 mRNA signal (Fig. 7C). The lack of GAD67-positive GABAergic interneurons in the thalamic relay nuclei is characteristic to rodent thalamus (Ottersen and Storm-Mathisen, 1984). The caudal part of the thalamic relay nuclei emitted a comparable level of GFP fluorescence to the caudate putamen (Fig. 5A,F), and GAD67 mRNA signal was increased in the pretectal area and the lateral geniculate nucleus (Fig. 7C). Figure 7 shows serial sections processed for GAD67 in situ hybridization. In this staining, only the GAD67 mRNA-positive somata appear as dark-colored dots. The distribution of GFP-positive somata coincided well with the distribution of the GAD67 in situ hybridization signal in all parts of the brain. The reticular thalamic nucleus was the area with the most conspicuous GFP-positive and GAD67 mRNA-positive somata in the thalamus. GFP-positive somata were also found in the ventral and dorsal lateral geniculate nuclei (Figs. 5F, 6I) and the pretectal areas (Fig. 5F). In the lateral geniculate nucleus, both the GFP in the somata and neuropil contributed equally to the GFP fluorescence, which was higher than that in other thalamic nuclei (Figs. 5F, 6I). The medial geniculate nucleus contained sparse GFP-positive somata.

thumbnail image

Figure 7. In situ hybridization signal for glutamic acid decarboxylase (GAD)67 antisense probe in serial frontal sections obtained from a wild-type mouse brain. Scale bars = 1 mm in A (applies to A–G), H; 100 μm in I.

Download figure to PowerPoint

In the midbrain, GFP fluorescence was conspicuous in three sites (Fig. 5G). The most conspicuous site was the substantia nigra pars reticulata, the second-most conspicuous site was the superior colliculus, followed by the oculomotor nucleus. GFP fluorescence in the substantia nigra pars reticulata was emitted from many weakly GFP-positive somata and the strongly GFP-positive neuropil (Fig. 6M). Because the intensities of the fluorescence from the somata and neuropil were similar, they were not distinguishable in Figure 6M. The superior colliculus emitted a high level of GFP fluorescence that was conspicuous even at low magnification (Fig. 5A,G,H). At higher magnification, GFP-positive somata were seen to be abundant in the superficial layers of the superior colliculus (Fig. 6J). GFP expression was also conspicuous in the oculomotor nucleus (Fig. 5G) and ventral tegmental nucleus (Fig. 5I). Nuclei related to the auditory system also showed GFP fluorescence at various levels. The medial geniculate nucleus showed GFP fluorescence at a low level, comparable to that of the neocortex (Fig. 5G). The inferior colliculus showed fluorescence at a moderate level (Fig. 5I). Somata displaying a GAD67 mRNA signal were also contained by each nucleus at a level corresponding to the level of fluorescence (Fig. 7F,G). The lateral lemniscus contained both large GFP-positive somata and GFP-positive thick projection axons (Fig. 6K). The cerebellar cortex also emitted GFP fluorescence, but the fluorescent intensity varied from lobule to lobule (Figs. 5A,J, 6L). Ventral lobules in the vermis emitted intense GFP fluorescence. In the confocal image (Fig. 6L), Purkinje cell somata, Golgi cell somata, basket cell somata, and stellate cell somata were seen to be fluorescent (Palay and Chan-Palay, 1974). The GFP fluorescence of the inferior olive was conspicuous because of its intensity and the shape of the nucleus (Fig. 5A,J). At higher magnification, it was obvious that the fluorescence was emitted from the fine axon terminals surrounding cells in the inferior olive (Fig. 6O).

GFP-positive cells were also found in the retina. Figure 6P–R shows confocal images taken by setting a focal plane parallel to the retina and by shifting the focal plane from the outer edge of the inner granular layer (Fig. 6P) to the center of the inner granular layer (Fig. 6Q), and then to inner edge of the inner granular layer (Fig. 6R). The GFP-positive cells at the outer edge of the inner granular layer, which correspond to GABAergic horizontal cells, were aligned in a plane and were difficult to focus on precisely (the area with a faint pattern in Fig. 6P). GFP-positive cells at the inner edge of the inner granular layer, which correspond to GABAergic amacrine cells, were distributed in a layer with a regular, uninterrupted geometric pattern. This pattern of GFP-positive cell arrangement mimicked the GABA-immunoreactive site reported previously in the same layer (Brandon et al., 1979; Famiglietti and Vaughn, 1981; Vaughn et al., 1981; Wu et al., 1981). The similarity between the distribution patterns of GFP-positive cells and those of GABAergic horizontal cells and amacrine cells suggests that GFP expression occurs consistently in the GABAergic neurons in the retina. The center of the inner granular layer emitted intense GFP fluorescence, which was probably due to the GFP contained in the axons and dendrites of horizontal cells and amacrine cells.

Colocalization of GFP with NeuN and neuropeptides

We used the GAD67-GFP (Δneo) mice to investigate the colocalization of GFP with NeuN and several neuropeptides that characterize GABAergic neuron subtypes in the frontal motor cortex. According to the data obtained by anti-NeuN immunohistochemistry, basically all the GFP-positive cells in the neocortex appeared positive for NeuN. The colocalization of GFP with NeuN reconfirmed that the GFP is a good marker of GABAergic neurons in the adult neocortex. We counted the GFP-, NeuN-, CR-, PV-, and SS-positive somata in slits (460 μm wide) set at the frontal motor cortex in 25-μm-thick frontal sections. The slits were divided into zones of layer I, layer II–III, layers V, and layer VI plus white matter, respectively. The motor cortex in mice basically lacks layer IV. The proportion of GFP-positive cells among NeuN-positive cells (neocortical neurons) was 19% (291 of 1,496) in a slit (Fig. 8A–C).

thumbnail image

Figure 8. Colocalization of green fluorescence protein (GFP) with NeuN, calretinin (CR), parvalbumin (PV), and somatostatin (SS) in glutamic acid decarboxylase (GAD)67-GFP (Δneo) mice. A–C: Colocalization of GFP with NeuN. All the GFP-positive cells were positive for NeuN and confirmed to be neurons. A–C show GFP fluorescence, the merged image, and NeuN-immunoreactive sites (red) in layers I–II of the frontal motor cortex, respectively. Bars in C indicate the pia mater and the boundary between layers I–II. D–F: Colocalization of GFP with CR. D–F show GFP fluorescence, the merged image, and CR-immunoreactive sites in layers II–III of the frontal motor cortex, respectively. Arrowheads in D, G, and J indicate double-labeled cells. G–I: Colocalization of GFP with PV. G–I show GFP fluorescence, the merged image, and PV-immunoreactive sites in layers II–III of the frontal motor cortex, respectively. J–L: Colocalization of GFP with SS. J–L show GFP fluorescence, the merged image, and SS-immunoreactive sites in layers V–VI of the frontal motor cortex, respectively. Scale bar = 100 μm in A (applies to A–L).

Download figure to PowerPoint

The distribution and features of CR-positive, PV-positive, and SS-positive cells have been well described in many publications (DeFelipe, 1993; Kubota et al., 1994; Gonchar and Burkhalter, 1997). In brief, CR-positive cells were preferentially found in layers I–II and the deeper part of layer VI. The CR-positive cells had small somata. Other layers contained sparsely distributed CR-positive cells. PV-positive cells were distributed from layers II to VI but not in layer I. PV-positive cells had the largest somata among the GABAergic neuron subtypes. SS-positive cells were found in layers II–VI, but were more abundant in the deeper layers. They had medium-sized somata.

The double-labeling study with GFP revealed that basically almost all of the CR-positive cells were positive for GFP (122 of 131). The intensity of the GFP fluorescence emitted from their small somata varied from moderate to very low (Fig. 8D). The proportion of CR-positive cells among GFP-positive cells was 6.5 ± 4.0% (5 of 89) in layer I, 22.8 ± 1.0% (72 of 341) in layer II–III, 5.2 ± 2.3% (9 of 163) in layer V, and 12.5 ± 5.4% (45 of 358) in layer VI plus white matter. Overall, the proportion of CR-positive cells among GFP-positive cells was 13.9 ± 1.6% (131 of 951) in the neocortex. All the PV-positive cells were positive for GFP (397 of 397). The level of GFP fluorescence emitted from their large somata varied from high to moderate (Fig. 8G). The proportion of PV-positive cells among GFP-positive cells was 0% (0 of 65) in layer I, 44.3 ± 4.7% (198 of 443) in layer II–III, 44.5 ± 6.4% (63 of 146) in layer V, and 41.8 ± 11.3% (136 of 331) in layer VI plus white matter. Overall, the proportion of PV-positive cells among GFP-positive cells was 40.1 ± 7.5% (397 of 985) in the neocortex. Basically almost all of the SS-positive cells were also positive for GFP (187 of 198). The intensity of the GFP fluorescence emitted from their medium somata varied from high to very low (Fig. 8J). The proportion of SS-positive cells among GFP-positive cells was 0% (0 of 52) in layer I, 17.7 ± 3.5% (68 of 381) in layer II–III, 31.4 ± 4.7% (44 of 141) in layer V, and 31.6 ± 5.7% (86 of 275) in layer VI plus white matter. Overall, the proportion of SS-positive cells among GFP-positive cells was 23.4 ± 3.2% (198 of 849) in the neocortex. The proportions identified by colocalization of GFP with CR, PV, and SS in layers were shown in Table 2.

Table 2. GABAergic Neuron Subtypes1
 Layer I (%) Layer II-III (%) Layer V (%) Layer VI-WM (%) 
  • 1

    The table shows the proportion of calretinin-positive, parvalbumin-positive, and somatostatin-positive cells among GFP-positive neurons in each layer of the motor cortex. The proportion is also shown as the ratio of the actual cell numbers counted in each slit (460 μm wide). GABA, γ-aminobutyric acid; WM, white matter.

Calretinin6.5 ± 4.0(5/89)22.8 ± 1.0(72/341)5.2 ± 2.3(9/163)12.5 ± 5.4(45/358)
Parvalbumin0(0/65)44.3 ± 4.7(198/443)44.5 ± 6.4(63/146)41.8 ± 11.3(136/331)
Somatostatin0(0/52)17.7 ± 3.5(68/381)31.4 ± 4.7(44/141)31.6 ± 5.7(86/275)
Total6.5 84.8 81.1 85.9 

Thus, the major three subtypes of GABAergic neurons accounted for 6.5% of the GFP-positive cells in layer I, 84.8% in layer II–III, 81.1% in layer V, and 85.9% in layer VI plus white matter. Overall, the major three subtypes of GABAergic neurons accounted for 77.4% of the GFP-positive cells in the neocortex. Unidentified GFP-positive cells accounted for 22.6% of the GFP-positive cells, most notably for 93.5% of the GFP-positive cells in layer I. In accord with unidentified GFP-positive cells, which were GAD67-positive as revealed by in situ hybridization (Fig. 7I), many medium spherical somata emitting intense GFP fluorescence were observed in layer I (Fig. 8A–C).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

We focused the purposes of this study on three points. The first point was to characterize the GFP expression in the two stains of GAD67-GFP knock-in mouse. The second point was to describe the distribution of GFP in the entire mouse brain and build a map for future studies. The third point was to report the results obtained from the double labeling study of GFP with NeuN, CR-, PV-, and SS.

The mice used in this study had a single-copy gene for GAD67 expression and a single-copy gene for GFP expression, and both of them were connected to the same promoter, enhancer, and suppressor in the introns or in the 5′ and 3′ flanking region in the GAD67 allele. The similarity of the genetic control region structure is thought to account for the parallel patterns of expression of these two different proteins. Thus, we investigated the colocalization of GFP with GAD67 and GABA in the GAD67-GFP mouse brain by immunohistochemistry, because the GAD67-GFP mouse brain contains GABA at a level almost similar to that in the wild-type mouse brain. We used the TSA system to visualize the immunoreactive sites of GAD67 and GABA because of its high sensitivity. We found that 80% of the GFP-positive cells were positive for GAD67 and 88% of the GFP-positive cells were positive for GABA. However, we do not think these data indicate that the 20% or 12% of the GFP-positive cells were negative for GAD67 or negative for GABA, respectively. The 20% or 12% of the GFP-positive cells had a GAD67 or GABA immunoreactivity at a similar level to that of the neuropil. These cells have been regarded as negative for GAD67 and negative for GABA, because there was no way to reveal them in the sections. Now there will be no reason to regard these cells as negative for GAD67 and negative for GABA. Because it is apparent that the neuropil of the neocortex contain GAD67 and GABA immunoreactivities, we believe the cells with immunoreactivity levels similar to those of the neuropil should be regarded as GAD67- and GABA-positive cells. Thus, we considered that basically all the GFP-positive cells were positive for GAD67 and GABA. During the several trials of improvement in fixation condition, we speculated that the knock-in mouse brains contained less GAD67 than the wild-type mouse brains did. Because GAD67 is a synthetase of GABA, GABA immunohistochemistry in the knock-in mouse brain also failed to reveal GABA immunoreactive sites satisfactory. Thus, we measured the GABA contents in the two strains of knock-in mouse brains and in the wild-type mouse brains at birth and at 6- and 7-week-old stage, respectively. As shown in Table 1 and Figure 1E, GABA contents in the knock-in mouse brains were significantly reduced at birth. However, the GABA contents in the knock-in mouse brains seemed to be compensated by another factor such as GAD65 expression as the knock-in mouse grew. Because the GAD65 is the other GABA synthetase coexpressed with GAD67 in similar temporal and spatial pattern (Dupuy and Houser, 1996), GAD65 would be able to compensate for the reduction of GABA content in most of GABAergic neurons. Moreover, the GAD67-GFP (Δneo) and GAD67-GFP mice were similar to the other GAD67 knock-in mice (Asada et al., 1997; Fuchs et al., 2001) and seizure-free. They exhibited normal growth, normal reproductive behavior, and no abnormality in the brains at the macroscopic level. To judge if the reduction of GABA content affected the anatomic aspect of the knock-in mice at microscopic level, such as GABAergic neuron number, axon terminal number, or axon-branching pattern, we had to wait for results from studies in progress by several research groups.

Although the reduction in GABA content is a significant phenotype of the knock-in mouse at birth, we took the phenotype as a good sign, because the reduction in GABA content seemed to correlate with the increase in GFP expression. The correlation might indicate how tightly the GFP expression was regulated by the GAD67 promoter. We did not encounter any apparent ectopic GFP expression in GAD67-negative cells or any apparent lack of GFP expression in GAD67-positive neurons in the two GAD67-GFP knock-in mouse lines. The distribution of GFP-positive somata mimicked the distribution of the in situ hybridization signal for GAD67. GFP fluorescence was continuous across the neocortex without any interruptions or patches. GFP-positive cells in the inner granular layer of the retina had a distribution pattern that was very similar to that of GABAergic horizontal cells and amacrine cells. Thus, the GFP expression was assumed to be regulated properly in GAD67-positive GABAergic neurons.

GFP was also expressed temporally in parallel with GAD67 expression. GAD67 expression occurs in not only the neuronal tissue but also in developing non-neuronal tissues such as tooth buds and whisker roots. The GFP expression in non-neuronal tissues suggests that our GAD67-GFP knock-in mouse line will also be useful for studying the development of GAD67-positive non-neuronal tissues. In the murine telencephalon, most of the GABAergic neurons originate in the ganglionic eminence. The ganglionic eminence was one of the structures with the strongest GFP fluorescence and GAD67 in situ hybridization signal in the embryo brains (DeDiego et al., 1994). These GFP-positive cells seemed to spread into the neocortex and the olfactory bulb. The cell shape revealed by GFP fluorescence was typical of migrating cells (Rakic, 1990). The ventricular zone in the ganglionic eminence contained some GFP-positive cells, which may indicate that GABAergic neuron production occurs by means of division of the ventricular cells in the ganglionic eminence. The ventricular zone of the neocortex contained a small number of GFP-positive cells, but the shapes of these cells were also the same as those of migrating cells (Nadarajah et al., 2002). The other ventricular zone cells were negative for GFP and the GAD67 in situ hybridization signal. All these observations may support the idea that the most of the GABAergic neurons in the murine neocortex originate in the ventral telencephalon and are supplied by tangential cell migration (Anderson et al., 1997; Tamamaki et al., 1997; Lavdas et al., 1999). Thus, we concluded that the GAD67 promoter regulated the GFP expression both spatially and temporally in the knock-in mouse.

In subcellular levels, GAD67 and GFP showed different distributions. Because GAD67 has the important function of synthesizing GABA, the distribution of GAD67 would not depend on diffusion; rather, GAD67 would be sorted into its proper functional site by axonal transport. Because GFP probably has no functional interaction with other proteins present in GABAergic neurons, GFP would diffuse randomly and fill the nucleus, perikarya, dendrites, axons, and axon terminals. Overlapping GAD67 and GFP distributions were preferentially found in perikarya and axon terminal-like boutons. The intensity of GFP fluorescence emitted from the dendrites and axon branches does not reflect the amount of GAD67 contained in these structures. We speculate that the intensity of GFP fluorescence shown in Figure 5 would indicate the density of GABAergic neuron somata, dendrites, axons, and axon terminals in each point. Because the sections treated for GAD67 in situ hybridization will show the distribution of GAD67-positive somata (Fig. 7), we can estimate the density of other subcellular structures of GABAergic neurons by comparison of Figures 5 and 7. In experiments using our GAD67-GFP knock-in mice, such as neurophysiological experiments, we expect that most GABAergic neurons in a brain slice are detectable at a glance through a fluorescence microscope, whereas we cannot detect GABAergic neurons with low GFP fluorescence in the GFP-positive neuropil (Fig. 6M). To identify more GABAergic neurons in living brain slices, we plan in future experiments to increase the level of expression of GFP in GABAergic neurons by mating a GAD67-Cre recombinase knock-in mouse with a transgenic mouse carrying a cre-mediated GFP reporter gene under a constitutive promoter (Kawamoto et al., 2000). In another experimental plan, it may be possible to reduce the GFP fluorescence in the neuropil by adding a dendrite-targeting signal (West et al., 1997; Stowell and Craig, 1999) to the GFP cDNA.

Because the GFP expression correlated only with GAD67 expression and not with the expression of neuropeptides, we do not anticipate any difficulties in the immunohistochemical studies of neuropeptides in the knock-in mice. Therefore, the GAD67-GFP (Δneo) mouse is a useful mouse, in which GAD67-positive neurons were indicated by intense fluorescence and the expression of neuropeptides was at a normal level, for double-labeling studies. Therefore, we used the knock-in mice for the analysis of GABAergic neuron subtypes found in the motor cortex. We estimated the proportion of GABAergic neurons in the neocortical neurons to be 19.5% by double labeling of GFP and NeuN. The proportion thus estimated by us is the highest yet obtained in the murine neocortex (Lin et al., 1986; Meinecke and Peters, 1987) and is close to the proportions obtained in the monkey and cat (Gabbott and Somogyi, 1986; Hendry et al., 1987). We also estimated the proportions of CR-, PV-, and SS-positive cells among the GABAergic neurons to be 13.9%, 40.1%, and 23.4%, respectively. Most of these estimated values were lower than those previously reported (Kubota et al., 1994; Gonchar and Burkhalter, 1997). That our estimates were higher or lower than those made previously was due to a larger numerator or larger denominators than those used previously, respectively. The larger numerator and larger denominators were probably due to the higher chance of detecting GABAergic neurons with the assistance of GFP fluorescence. The detection of more GABAergic neurons in the neocortex revealed previously unidentified GABAergic neuron by using the three markers, CR, PV, and SS. Although the neocortical GABAergic neurons contain various neuropeptides, it was speculated that CR-positive, PV-positive, and SS-positive GABAergic neurons would account for most of the GFP-positive neurons (Gonchar and Burkhalter, 1997), and the rest of the GABAergic neurons would be a minority. However, the GABAergic neurons that were not included in the three major GABAergic neuron subtypes were not a minority but the majority in layer I. The 93.5% of the GFP-positive cells in layer I, 15.5% in layer II–III, 18.9% in layer V, and 14.1% in layers VI plus white matter of the neocortex remained an unidentified population. The density of GABA-positive neurons in layer I was comparable to that in layer VI (Goncha and Burkhalter, 1997). In accord with the high rate of unidentified GABAergic neurons in layer I, we found conspicuous GFP-positive neurons. Most of the unidentified GFP-positive neurons seemed to have medium spherical somata emitting GFP fluorescence at the highest level and were apparently different from CR-positive GABAergic neurons.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

The authors thank Mr. Akira Uesugi in Kyoto University for photographic assistance.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED