GABA Regulates the Multidirectional Tangential Migration of GABAergic Interneurons in Living Neonatal Mice

Cortical GABAergic interneurons originate from ganglionic eminences and tangentially migrate into the cortical plate at early developmental stages. To elucidate the characteristics of this migration of GABAergic interneurons in living animals, we established an experimental design specialized for in vivo time-lapse imaging of the neocortex of neonate mice with two-photon laser-scanning microscopy. In vesicular GABA/glycine transporter (VGAT)-Venus transgenic mice from birth (P0) through P3, we observed multidirectional tangential migration of genetically-defined GABAergic interneurons in the neocortical marginal zone. The properties of this migration, such as the motility rate (distance/hr), the direction moved, and the proportion of migrating neurons to stationary neurons, did not change through P0 to P3, although the density of GABAergic neurons at the marginal zone decreased with age. Thus, the characteristics of the tangential motility of individual GABAergic neurons remained constant in development. Pharmacological block of GABAA receptors and of the Na+-K+-Cl− cotransporters, and chelating intracellular Ca2+, all significantly reduced the motility rate in vivo. The motility rate and GABA content within the cortex of neonatal VGAT-Venus transgenic mice were significantly greater than those of GAD67-GFP knock-in mice, suggesting that extracellular GABA concentration could facilitate the multidirectional tangential migration. Indeed, diazepam applied to GAD67-GFP mice increased the motility rate substantially. In an in vitro neocortical slice preparation, we confirmed that GABA induced a NKCC sensitive depolarization of GABAergic interneurons in VGAT-Venus mice at P0-P3. Thus, activation of GABAAR by ambient GABA depolarizes GABAergic interneurons, leading to an acceleration of their multidirectional motility in vivo.


Introduction
The establishment of CNS cytoarchitectures is essential for brain function. To acquire ordered neuronal arrangements, neurons migrate from their birth place to their final positions during development. The function of the cerebral cortex essentially relies on the balance between excitation and inhibition, which is served by two classes of neurons: excitatory glutamatergic projection neurons and inhibitory GABAergic interneurons. Glutamatergic neurons arise in the germinal zones lining the lateral ventricle, and migrate radially within the cortex toward the pial surface [1]. On the other hand, most cortical GABAergic interneurons originate from the ganglionic eminences (GEs), located in the subpallial telencephalon [2]. These GABAergic interneurons migrate tangentially toward the pallium and cortex. They show several migration modalities in the neocortex before they reach their final positions in their corresponding target layers [3]; 1) tangential migration along the intermediate zone (IZ), 2) radial migration from the IZ towards the marginal zone (MZ) via the cortical plate (CP) [4] or from the IZ towards the ventricle surface [5], 3) multidirectional tangential migration in the MZ [6], and 4) radial migration from the MZ toward the CP [7]. The proper regulation of each of these modalities has a critical role in determining the final location of GABAergic interneurons in the neocortex [8]. In particular, disruption of intracortical migration in the MZ causes changes in the distribution of interneurons in the mature cortex [9,10]. This suggests that multidirectional tangential migration in the MZ has a critical role for the balanced distribution of inhibitory neurons throughout the entire cortical areas. It has been reported that the tangential migration of GABAergic interneurons mainly proceeds during embryonic period [11]. In consistent with this finding, the interneurons in the IZ disappear by P0.5 [12]. On the other hand, it has been reported that a significant number of interneurons still exist in the MZ at P0.5 [13]. These reports imply that the MZ is one of the dominant cortical layers for the migration of GABAergic interneurons during early postnatal development. Thus, in vivo examination of migratory behavior and its regulatory mechanism in the early postnatal MZ is important to understand the organizational principles of the neocortex. Extracellular GABA has been reported to have two contrary roles in regulating the migration of cortical GABAergic interneurons. On the one hand, time-lapse imaging in an in vitro embryonic preparation revealed that extracellular ambient GABA promoted tangential migration from the medial GEs towards the neocortex via the tonic activation of GABA A Rs [14]. Alternatively, in vivo local application of GABA A R-agonists and antagonists via Elvax implants induced severe disturbances in cortical layering and focal cortical dysplasias, suggested the modulation of GABA A Rmediated mechanisms impaired radial migration of cortical GABAergic interneurons [15]. An intrinsic change in the responsiveness of migrating neurons to GABA, from acceleration migration to inhibiting migration, mediated via the upregulation of the potassium-chloride cotransporter KCC2 was recently reported using an embryonic slice culture system [16]. Taken together, this suggests that extracellular GABA may promote tangential migration and terminates radial migration in the cortex. However, it is not known exactly whether GABA promotes or inhibits the multidirectional tangential migration in the MZ in vivo, which determines the final distribution of GABAergic neurons across the cortical areas. Furthermore, the underlying GABA action (excitation or inhibition) mediating any affects in vivo, which can depend on extracellular GABA [17] and intracellular Cl 2 concentration [18], is also unclear. It is necessary to elucidate the GABA action on the migrating neurons in intact preparation and thus, in the present study, we carried out in vivo time-lapse imaging of GABAergic interneurons in the MZ using two-photon microscopy. We found in vivo that interneurons migrate in all directions within the tangential plane, and that the activation of GABA A Rs enhances this motility, by increasing the frequency of movements, without affecting the preference in direction and stroke of each movement.

Development of in vivo imaging for immature brain
In vivo imaging of neuronal population with single-cell resolution became possible with the application of two-photon laser-scanning microscopy [19,20]. While the importance of in vivo experiments in immature animals to understand the development of neural circuit is well accepted, there are also studies which examined the dynamics of neural circuits during development [21,22]. Our initial studies found that the standard approaches to make a cranial window (open skull or thin skull methods) needed for imaging, easily damaged the migrating interneurons in the MZ. Furthermore, the application of room temperature immersion water at the skull -lens interface disturbed the local circulation at the brain surface. Thus, we performed two technical improvements for imaging cortical neurons of immature mice (Fig. 1A). First, we circulated warm water (35uC) above the imaging area to prevent any decrease of cortical temperature. Secondly we made a three-direction restraint bar (back, left, and right side of the head) to alleviate brain movement while avoiding any damage due to pressure. At the same time, the thin skull bone of immature mouse allowed the sufficient transcranial access of the laser without the need for any surgery. With these combined technical improvements, we succeeded in the stable time-lapse imaging of the migration of cortical GABAergic interneurons located in the MZ of immature living mice.

Multidirectional tangential migration of GABAergic interneurons in immature intact cortex
Migratory behavior of GABAergic interneurons in the MZ has been reported using a flat-mount cortical preparation in vitro [6] and in vivo [7]. In the present study, we observed GABAergic interneurons in the MZ in vivo by applying two-photon time-lapse imaging to VGAT-Venus transgenic mice (VGAT-Venus mice) at P0-P3. In the neocortex of VGAT-Venus mice, more than 95% of Venus positive cells expressed GABA and more than 93% of GABA-immunoreactive cells were positive for Venus [23]. Thus it is unlikely that only a specific subtype of GABAergic interneurons were observed in the present experiment.
We performed continuous time-lapse imaging (1 frame/3 min, Fig. 1B) for at least 60 minutes and made a quantitative analysis of the migration direction. By plotting the translocation of a basal point of the leading processes in each frame, we quantified the movement of each GABAergic interneuron (Fig. 1C, n = 61 cells, 5 animals). Neurons that migrated out of the focal plane during the imaging period (less than 10% of neurons) were excluded from subsequent analysis. Migrating neurons were defined as those that migrated more than 5 mm during 1 hour. We subdivided these migrating neurons into 4 groups according to their final position, and counted the proportion of the migrating neurons located in each sector (Fig. 1D These results indicate that GABAergic interneurons in the MZ have no preference in direction of migration within the tangential plane in vivo. The histogram of migration rate showed a broad distribution among all the neurons (Fig. 1F), and nearly half of them did not migrate significantly (Fig. 1G, 59.269.5% for moving cells and 40.863.0% for stationary cells).
When we compared the migration rate between P0 and P3 VGAT-Venus mice including control and ACSF treated groups, there was no significant difference between two groups (30.961.5 mm/hr for P0 and 36.9 mm/hr for P3 VGAT-Venus mice, n = 3 animals respectively). The cell density in the imaging plane decreased between P0 and P3 in VGAT-Venus mice (10636201 cells/mm 2 in P0 and 696646 cells/mm 2 in P3 MZ) without any change in the proportion of stationary cells (35.462.6% for P0 and 37.662.5% for P3 mice).

Involvements of GABA A R and [Cl 2 ]i in multidirectional migration
To investigate the regulatory mechanisms of these migratory behaviors of GABAergic interneurons, we developed in vivo drug application system for immature cortex ( Fig. 2A). Drugs were pressure-injected approximately 2 millimeters away from the imaging area via a glass pipette inserted through a small hole in the skull. This procedure did not induce any noticeable damage to neurons in the imaging area, and preliminary data (not shown) indicted no increase in cleaved caspase-3 immunostaining following saline pressure-injection. To examine whether the drugs applied could be delivered to imaging areas, sulforhodamine 101 (SR101), a red fluorescent molecule that has been used to label cortical glial cells [24], was also added to the solution. We confirmed that drugs could access the imaging area by detecting this SR101 fluorescence after similar pressure injection. Application of artificial cerebrospinal fluid (ACSF) had no effect on the migration rate of cortical GABAergic interneurons (Fig. 2B  continuously activate GABA A R to promote increased migration rate. In addition, chelating [Ca 2+ ] i by injecting BAPTA-AM (dissolved in DMSO) virtually abolished migration rate compared with DMSO injection group (Fig. 2C, DMSO 39.565.9 mm/hr, BAPTA-AM 1.260.3 mm/hr. p,0.01). Furthermore, bumetanide, a blocker of the Na + -K + -Cl 2 cotransporters (NKCC), also reduced migration rate significantly (Fig. 2C, DMSO alone; 39.565.9 mm/hr, bumetanide; 18.265.2 mm/hr). Taken together, the data suggest that the activity of GABA A R, [Ca 2+ ] i and [Cl 2 ] i levels regulate the motility of multidirectional tangential migration, either separately or in an associated manner. A previous study reported that the activation of GABA A R could depolarize the membrane potential of immature neurons due to the activity of NKCC1 [25].

Gramicidin patch-clamp recording from GABAergic interneurons in slice preparations
To confirm that GABA results in a depolarization of the GABAergic interneurons during multidirectional tangential migration, we made horizontal brain slices containing the MZ, and measured the GABA A R reversal potential of VGAT-Venus using gramicidin-perforated patch-clamp recordings which maintain [Cl 2 ] i intact [26]. Ramp voltage step were applied to Venus positive cell (Fig. 3A) in the absence and presence of GABA application. Based on the resulting current-voltage relationships, the reversal potential of GABA A R response was used to estimate the Cl 2 equilibrium potential (E Cl ), which was measured as 243.466.1 mV, ( Fig. 3B; n = 5 cells). This E Cl of GABAergic interneurons, was more depolarized than the resting membrane potential (V rest ) measured in the same cells (Fig. 3B, 258.364.4 mV, n = 5 cells, p,0.05). Bumetanide shifted E Cl negatively to 260.365.9 mV (n = 5 cells, Fig. 3B), closed to V rest . This result suggests that NKCC1 activity maintains a high [Cl 2 ] i, which is responsible for GABA induced membrane depolarization in immature GABAergic interneurons.

Expression level of cation-chloride transporters
The [Cl 2 ] i has been reported to be predominantly regulated by two cation-chloride cotransporters in immature and mature neurons [27]. NKCC1, as mentioned previously, transport Cl 2 into neurons while the K + -Cl 2 cotransporter 2 (KCC2) excludes Cl 2 from neurons [28]. A developmental change in the expression and function of these transporters switches the GABA response from depolarization to hyperpolarization [29]. We therefore examined the protein level of NKCC1 between wild type and VGAT-Venus. NKCC1 was equally expressed in the two strains (Fig. 3C). The expression levels of KCC2 have been reported to be elevated in cortical GABAergic interneurons during early developmental stages, with a critical role in regulating the termination of migration [16]. In the present experiment, however, the expression level of KCC2 at P0 was very low, under detectable levels, while marked expression was observed in more mature neurons from wild type and VGAT-Venus mice under the same conditions (Fig. 3C, 3D). To further determine the extent of NKCC1 and KCC2 expression in individual interneurons in the MZ, we immunostained for these transporters in slices from fixed Making a hole at 1-2 mm away from the observed area has no effect on the morphology of the cells within imaging area. Scale bar = 50 mm. Lower left: Two-photon image of sulforhodamine 101 (SR101) showing that the drug does not reach the imaging area immediately after injection. Upper right: Schematic drawing of the drug injection procedure; making a hole and injecting the drug conjugated with SR101. The drug diffuses into the imaging area after 30-60 minutes of injection. Lower right: Two-photon image of Venuspositive cells and SR101 showing that the pipette solution spread around the cells without injuring them. Scale bar = 50 mm (B) Cell displacement during 1 hour of migration for neurons in VGAT-Venus without drug application (VGAT), after injection of ACSF (+ ACSF) and after injection of SR95531 (+SR95531). The selective GABA A R antagonist (SR95531, 1 mM, 1 ml) reduced the migration rate significantly (*p,0.05) whereas ACSF (1 ml) had no effect. (C) DMSO (1 ml) had no effect on cell displacement. Chelating intracellular Ca 2+ by BAPTA-AM (2 mM, 1 ml) conjugated with DMSO almost completely inhibited the migration (**p,0.01) and blockade of NKCC by bumetanide (1 mM, 1 ml) significantly reduced the migration rate (*p,0.05). VGAT-Venus; n = 5 animals, ACSF; n = 5 animals, DMSO; n = 4 animals, SR95531; n = 5 animals, bumetanide; n = 4 animals, BAPTA-AM; n = 4 animals. All animals were used at P0-P3, except the bumetanide group which were used at P1-P3. Each column in B and C indicates mean 6 SEM. doi:10.1371/journal.pone.0027048.g002 brains from P0-P3 VGAT-venus mice (Fig. 3E-G). No GABAergic interneurons in the MZ showed any expression of KCC2 whereas a substantial proportion of GABAergic interneurons in the CP showed staining for KCC2 (Fig. 3F, 3G). In contrast, most GABAergic interneurons in the MZ stained positive for NKCC1.

Different migration rate between VGAT-Venus and GAD67-GFP mice
To investigate the effect of extracellular GABA concentration on multidirectional migration rate, we employed another transgenic strain, the GAD67-GFP knock-in mice (GAD67-GFP mice), which specifically express GFP in GABAergic interneurons. In this strain, there is a decrease of GABA content in the immature cortex, caused by the inactivation of a single GAD67 gene allele [30].
We performed an in vivo observation of the movements of GABAergic interneurons in GAD67-GFP mice and quantified their migratory behavior as done in VGAT-Venus mice (Fig. 4A). Migration of GABAergic interneurons in the MZ showed a similar lack of preference for direction, as seen in VGAT-Venus mice (Fig. 4B, 4C, medial; 24.961.8%, rostral; 28.765.0%, lateral; 22.563.0%, caudal; 24.961.8%). Furthermore, the histogram of migration rate also showed a broad distribution among this population and again nearly half of them did not migrate significantly (Fig. 4D, n = 4 animals, 55.067.7% for moving cells and 45.068.4% for stationary cells). However, there was a significant difference in the average migration rate between the two strains in vivo (Fig. 4E, 41.561.8 mm/hr for VGAT-Venus; n = 61 cells, 5 mice and 22.862.1 mm/hr for GAD67-GFP; n = 45 cells, 4 mice, p,0.05). In contrast, when we analyzed migration rate in vitro using brain slices from each strain, both species showed a reduced migration and there was no significant difference between the two (Fig. 4E, n = 139 cells, 7 cortices for VGAT-Venus, n = 122 cells, 7 cortices for GAD67-GFP). To further characterize the behavior of each cell that undergoes multidirectional migration, we analyzed the movement in vivo as a function of time using time-lapse sequences (Fig. 4F). GABAergic interneurons in both GAD67-GFP and VGAT-Venus mice showed discontinuous, saltatory movements. There was no significant difference in the distance of each single movement phase between the two strains (Fig. 4G, 6.4760.28/hr for VGAT-Venus; n = 61 cells, 5 mice and 5.4460.36/hr for GAD67-GFP mice; n = 45 cells, 4 mice). However, the frequency of these movement phases was significantly less in GAD67-GFP mice than that in VGAT-Venus mice in vivo (Fig. 4H, 5.9060.27 for VGAT-Venus and 4.2060.27 for GAD67-GFP mice). This result suggests that the lower level of ambient GABA reported for the GAD67-GFP mice (and also expected in the in vitro conditions), decreases migration rate by disturbing the trigger of movement phase.
To confirm this correlation between the GABA content and migration rate, we measured the levels of GABA in the cerebral cortex of both strains using high performance liquid chromatography (HPLC). The forebrain of P0 mice, including the cerebral cortex, contained GABA at a concentration of 2.5460.47 mmol/ 100 mg in VGAT-Venus (Fig. 4I, n = 9), 0.9160.12 mmol/ 100 mg in GAD67-GFP (n = 9), and 1.8360.19 mmol/100 mg protein in wild-type mice (n = 10). The GABA content in GAD67-GFP mice was significantly lower than those in wild type and VGAT-Venus mice (p,0.05, one-way ANOVA followed by posthoc Tukey-Kramer test). In contrast, the GABA content was not different between wild type and VGAT-Venus mice (p.0.05).
These results confirmed the correlation between the GABA content in VGAT-Venus and GAD67-GFP mice in vivo, and the migration rate. To further confirm that the activation of GABA A R by ambient GABA in GAD67-GFP mice has a critical role in migration rate, we applied diazepam, which increases the activation by GABA of GABA A Rs containing a/c2 subunits [31,32]. Diazepam increased the migration rate (Fig. 5A) to 33.764.6 mm/hr for the diazepam treated group from 16.161.8 mm/hr for the DMSO control group, and reduced the duration of the stationary phase between each movement in GAD67-GFP mice (Fig. 5B). Furthermore, we found that the application of the GABA A R antagonist, SR95531, also shortens the stationary phase in VGAT-Venus mice (Fig. 5C).

Discussion
In the present study, we developed in vivo imaging for immature mouse cortex with two-photon microscopy and quantified the migratory behavior of GABAergic interneurons in the MZ. To specifically identify the GABAergic interneurons, we employed two strains of mice, VGAT-Venus transgenic and GAD67-GFP knock-in mice in which the fluorescence is specifically expressed in GABAergic interneurons. Furthermore, we utilized the fact that the two strains have significant differences in their GABA content within the cerebral cortex (Fig. 4). Although each individual neuron seemed to move in a specific direction, the population of migrating neurons equally moved in all directions. Nearly half of neurons were stationary during the recording session. These features were similar in both strains. In contrast, the migration rate of VGAT-Venus mice was significant higher than that of GAD67-GFP mice, suggesting that ambient GABA mediates the multidirectional tangential migration. Previous reports have demonstrated an increased expression of specific GABA A receptor subunits (a1, a2, a5, c2, and c3) in mouse neocortical interneurons during tangential migration, coupled with altered pharmacological properties [32]. There remains the possibility that different properties of GABA A R receptors in the two mice strains may also contribute to the reduced migration rate in the GAD67-GFP mice observed here.
A role for GABA in the migration of cortical GABAergic interneurons has been previously discussed [14][15][16][17]33,34]. In this study, we investigated the role of GABA for multidirectional tangential migration in the MZ by combining in vivo and in vitro preparations. In vivo imaging of genetically defined GABAergic interneurons in the developing cerebral cortex revealed that multidirectional migration rate was enhanced by the membrane depolarization resulting from activation of GABA A Rs. GABAinduced depolarization is caused by NKCC1 activity causing a high intracellular Cl 2 concentration. We also show that intracellular Ca 2+ was critical for migration, and the GABA A R induced depolarization may enhance migration by increasing intracellular Ca 2+ .
In vivo imaging for immature brain Some previous papers have also taken up the challenge to undertake in vivo time-lapse imaging of neuronal migration [7,35]. Now we have designed an improved experimental configuration to archive stable in vivo time-lapse imaging to quantify the role of GABA signaling on multidirectional tangential migration of immature cortical GABAergic interneurons. On average, the in vivo population of migrating interneurons has no preference in direction and this was roughly similar to that in previous observations employing in vitro preparation [6]. However, the in vivo migration rate was much higher than that found in vitro (Fig. 4E), suggesting that some intrinsic factor(s) that regulate migration are disturbed in the slice preparation. Multidirectional tangential migration in the postnatal cortex Whereas many studies have examined interneuronal migration in the embryonic cortex [1,2,8], few studies have reported on migration in the postnatal mouse cortex. Indeed, tangentially migrating interneurons are depleted from the IZ by P0.5 [12] while the density of interneurons in the MZ decreases between P0.5 and P7.5 in a non apoptotic manner [13], which implies that the MZ is a dominant cortical layer for GABAergic interneurons migration during early postnatal development. Consistent with the previous report [13], we observed that the cell density in the MZ imaging plane decreased between P0 and P3 in both control and ACSF treated VGAT-Venus mice groups. Despite a reduction of the total number of cells, the proportion of stationary cells was preserved, indicated that the most stationary cells observed at P0 in the MZ eventually migrated towards the deep cortical plate and their final positions. These results suggest two distinct hypotheses. Firstly, there is a waiting period for entering the CP. Several types of developing axons are also known to stop growing and wait for the maturation of their target [36,37]. By analogy, GABAergic interneurons in the MZ may require further maturation before continuing their migration. Secondly, the migrating GABAergic interneurons in the MZ may release signals which suppress the migration of neighboring cells. As a result, the proportion of activating (moving) cells in the MZ is maintained as constant.

The contribution of GABA for intracellular Ca 2+ and multidirectional migration
It is well established that activation of GABA A R has a depolarizing effect in immature cortical neurons [38]. In rodents, this depolarizing action of GABA changes at the 2nd postnatal week into a hyperpolarization [39] due to a developmental upregulation of the neuronal chloride-extruding molecule, KCC2 [28] and a downregulation of inward Cl 2 transporter, NKCC1 [40]. Although the exact downstream mechanisms have not been fully elucidated, previous observations [41] suggests that GABA A R activation profoundly regulates the migration process by altering [Ca 2+ ] i levels. Similarly, GABAergic interneurons in the MZ during multidirectional migration in vivo responded to the specific GABA A R antagonist, SR95531, and to the intracellular Ca 2+ chelator BAPTA-AM with a reduction in migration rate (Fig. 2). This suggests that the activation of GABA A R in immature neurons produces a depolaration-induced increment of [Ca 2+ ] i , secondary to the more depolarized Cl 2 equilibrium potential at this developmental stage [39]. A release from intracellular Ca 2+ store, e.g. Ca 2+ induced Ca 2+ release, has also been reported as crucial for migration in cerebellar granule cells [42].

Multidirectional dispersion enhanced by ambient GABA
Previous studies reported that GABAergic interneurons wait in the MZ for 24-48 hours before entering the cortical plate [10,43]. These reports suggested that multidirectional tangential migration in the MZ contributes to the ultimate dispersion of GABAergic interneurons throughout the cortical area. In the present study, we proposed that, in living mice, ambient GABA accelerates the rate of multidirectional tangential migration caused by its depolarizing effect. Indeed, several studies have reported that the activation of GABA A R regulates neuronal proliferation [44], migration [16,41] and differentiation [45] in the developing cerebral cortex. Thus, the function of GABA signaling during development serves to balance the spatial arrangement of glutamatergic and GABAergic neurons which compose mature cortical neural circuits. In addition, the in vivo imaging techniques for immature brain used in this study will strongly contribute to further research into the development of neuronal circuits.

Materials and Methods
The experimental protocols in this study were approved by the Animal Research Committee of the National Institutes of Natural Sciences (approval number 10A179). All efforts were made to minimize the suffering and number of animals used in this study.

Animal surgery for in vivo imaging
Animals were anesthetized by i.p. injection of ketamine/ xylazine (130/19.5 mg/kg of body weight). Depth of anesthesia was assessed by monitoring the tail-pinch reflex and respiration rate. After removing the skin from the temporal and parietal sides of the head, the mice were transferred into the set-up and placed onto a warming plate (37uC). The stainless steel restraint bars were then fixed on the skull with dental cement and cyanoacrylate adhesive (Vetbond; 3M, St. Paul, MN, USA). A custom-made recording chamber with a hole in the center was placed on the restraint bars, so as to not exert any pressure on to the skull and brain. The gap between the recording chamber and the skull was filled with low melting-point agarose gel (SeaPlaque, Lonza Rockland, ME, USA) made at 2% w/v with saline. The recording chamber was perfused with a warm (35uC) distilled water to help maintain the body temperature at the imaging area.
In vivo two-photon imaging Animals were anesthetized either by inhalation of isoflurane (1.0% in pure O 2 ) or i.p. injection of ketamine/xylazine during imaging. Imaging was performed by using a custom built twophoton laser-scanning microscope based on a mode-locked laser system operating at 950 nm, 80-MHz pulse repetition rate, 100-fs pulse width (Mai Tai HP, Spectra Physics, Mountain View, CA, USA) and a laser scanning system (Olympus FLUOVIEW, Olympus, Tokyo, Japan) coupled to an upright microscope (BX61WI, Olympus) and equipped with water-immersion objectives: 625, 1.05 numerical aperture (Olympus). The coordinate of the imaging area was as follows: 4.5 mm posterior to the rostral end of the brain, 1.0 mm lateral to midline. This region develops into the mediomedial secondary visual cortex (V2MM) in adult mouse, which was identified based on the comprehensive atlas of developing mouse brain [46]. Images were taken from the MZ (at a depth of 0-50 mm under the skull) at 3 min intervals. A stack of images for tangential migration was the sum of 1060.5 mm sections (total thickness 5 mm), advancing in the Z axis. The average power delivered to the brain was ,30 mW. Images were analyzed off-line with MetaMorph software (Molecular Devices, Downingtown, PA, USA).

Measurement of GABA content
To measure GABA levels in the cerebral cortex of immature mice, high performance liquid chromatography (HPLC) was employed. Cortical tissues from P0 mice were dissected and homogenized with 0.1 M perchloric acid to destroy the three dimensional structures of proteins and make amino acids soluble. The homogenates were centrifuged at 10000 g for 15 min at 4uC and then the supernatant was neutralized with 0.1 M sodium carbonate. After addition of sodium bicarbonate, the samples were filtered by using Ultrafree-MC filters (Millipore Corp., Bedford, MA, USA). Amino acids in the samples were derivatized to ophthalaldehyde. An octadecylsilyl column was used and kept at 30uC in a column oven (ATC-300; Eicom, Kyoto, Japan). The mobile phase consisted of (in mM) 50 NaH 2 PO 4 2H 2 O at pH3.5, 0.015 EDTA 2Na, and 50% methanol. A flow rate of 0.5 ml/min was maintained using an isocratic pump (EP-300, Eicom). Protein concentrations were measured by a Quick Start protein assay kit (Bio-Rad Laboratories, Hercules, CA, USA) which employed bicinchoninic acid protein assay reagent with bovine serum albumin as a standard.

Electrophysiology
Animals were anaesthetized with 130/19.5 mg/kg ketamine/ xylazine (i.p.) and sacrificed by decapitation, and brains were rapidly removed and transferred into ice-cold ACSF (124 mM NaCl, 2.5 mM KCl, 1.2 mM KH 2 PO 4 , 1 mM MgSO 4 , 2 mM CaCl 2 , 10 mM glucose, 24 mM NaHCO 3 ). This solution was adjusted to pH 7.4 with tris(hydroxymethyl)aminomethane (Trisbase). Horizontal slices were cut by a vibratome (VT1000S; Leica, Nussloch, Germany) in ice-cold ACSF bubbled with 95% O 2 and 5% CO 2 . The neocortical 300 mm-thick slices were kept in control incubation solution (124 mM NaCl, 2.5 KCl, 1.2 mM NaH 2 PO 4 , 24 mM NaHCO 3 , 2.4 mM CaCl 2 , 1.3 mM MgCl 2 and 10 mM glucose) saturated with 95% O 2 and 5% CO 2 at room temperature (21-24uC) for at least 1 hour before being transferred into the recording chamber. Gramicidin-perforated patch-clamp recordings were done on Venus-positive neurons in the MZ at room temperature in an incubation solution bubbled with 95% O 2 /5% CO 2 (solution flow rate 0.8 ml/min) and with slices anchored with silver bars. Patch pipettes were made from borosilicate capillary glass in two stages on a vertical pipette puller (PC-10, Narishige, Tokyo, Japan). The resistances between the recording pipettes filled with internal solution and the reference electrode in the bubbled external ACSF solution were 4-6 MV. Neurons were visualized under phase contrast on an in upright microscope (ECLIPSE E600FN, Nikon, Japan). Current and voltage were continuously monitored on a pen recorder (WR3320, Graphtec, Tokyo, Japan) and stored on a computer equipped with pCLAMP8.2 software (Axon Instruments). Venus or GFP positive cells were excited by a mercury lamp (HBO MERCURY SHORT ARC, OSRAM, Germany) and recordings were only done on these fluorescently labeled cells. GABA-activated currents were measured with a patchclamp amplifier (Multi Clamp 700B, Axon CNS, Molecular Devices), and recorded with a sampling frequency of 10 kHz after low-pass filtering at 2 kHz. The neurons were voltage-clamped at a holding potentialof 260 mV. E GABA was measured by using voltage ramps (from 280 to 220 mV, 3 s duration), from a holding potential (V H ) of 260 mV, applied before and during 10 mM GABA application. GABA was puffed onto the soma using a microinjector (IM 300, Narishige, Tokyo, Japan). The internal (patch pipette) solution for perforated patch clamp recordings contained 60 mM potassium methanesulfonate, 90 mM KCl, 10 mM HEPES (pH 7.2). Gramicidin was dissolved in dimethyl sulfoxide (DMSO) to prepare a stock solution of 10 mg/ml, and then diluted to a final concentration of 30 mg/ml in the pipette solution. The gramicidin solution was prepared just before use. The series resistance typically reached a steady level below 30 MV within 40 mins after making the GV seal. During recordings, the series resistance errors were electronically compensated for by 70-80%. To minimize series resistance errors, current responses were kept small by restricting the voltage range and using a submaximal GABA concentration. With a peak current of 100 pA, the uncompensated voltage drop across a 30 MV series resistance would be less than 1 mV.

Western blotting
Proteins were separated using 7.5% acrylamide gels by SDS-PAGE. The gels were transferred to an Immobilon-P (Millipore). The blots were blocked in 1% bovine serum albumin and incubated overnight with primary antibody at 4uC. They were then incubated with horseradish peroxidase-conjugated secondary antibody (GE Healthcare UK Ltd., Buckinghamshire, England) for 1 hour at room temperature. Enhanced chemiluminescence (ECL, GE Healthcare) exposure on instant film and an ECL mini-camera luminometer were used to visualize labeled protein. The following primary antibodies (and their dilutions) were used: anti-NKCC1 (1:200, Millipore), anti-KCC2 (1:1000, Millipore). The optical densities of each band were analyzed with Scion Image software with a gel-plotting macro program. The total NKCC1 and KCC2 band densities were normalized to the b-actin band density. Western blottings were repeated at least three times to ensure the reproducibility of results. Immunohistochemistry P0 VGAT-venus mice were anaesthetized with 130/19.5 mg/kg ketamine/xylazine (i.p.) and sacrificed by decapitation. Brains were immediately excised and fixed overnight in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) at 4uC. Thereafter, 20 mm horizontal slices of the neocortex were cut using a vibratome (VT1000S; Leica, Nussloch, Germany). After washing in PB, slices were incubated in blocking solution (10% normal goat serum, 2% bovine serum albumin, 0.5% Triton X-100 in 0.1 M PB) containing primary antibodies overnight at 4uC. NKCC1 was stained with a monoclonal mouse antibody (T4, Developmental Studies Hybridoma Bank, Iowa City, IA, 1:800 dilution) for 24 h at 4uC. KCC2 immunostaining used a polyclonal rabbit antibody (Upstate, Temecula, CA, 1:1000 dilution) for 24 h at 4uC. Alexa633-bound goat anti-mouse and Alexa594-bound goat anti-rabbit fluorescent secondary antibodies (1:200 dilution; Invitrogen, San Diego, CA) were then applied for 3 hrs at room temperature. Fluorescent images were acquired with a confocal laser scanning microscope (Leica TCS SP2, Wetzlar, Germany).

Data analysis and statistics
All values are given as means 6 SEMs. A two-tailed Student's t test was used to determine significant differences between different conditions. For multiple groups, one-way ANOVA with post-hoc Tukey-Kramer test was used. The level of significance was set at p,0.05 for all tests.

Drugs
Drugs used in the present study were SR95531, BAPTA-AM, gramicidin D, bumetanide, diazepam and GABA from Sigma (St. Louis, MO, USA) and sulforhodamine 101 from Invitrogen (Eugene, OR, USA). Final drug concentrations were made up from stock solutions just before use.