TGF-β2 Regulates Transcription of the K+/Cl− Cotransporter 2 (KCC2) in Immature Neurons and Its Phosphorylation at T1007 in Differentiated Neurons

KCC2 mediates extrusion of K+ and Cl− and assuresthe developmental “switch” in GABA function during neuronal maturation. However, the molecular mechanisms underlying KCC2 regulation are not fully elucidated. We investigated the impact of transforming growth factor beta 2 (TGF-β2) on KCC2 during neuronal maturation using quantitative RT-PCR, immunoblotting, immunofluorescence and chromatin immunoprecipitation in primary mouse hippocampal neurons and brain tissue from Tgf-β2-deficient mice. Inhibition of TGF-β/activin signaling downregulates Kcc2 transcript in immature neurons. In the forebrain of Tgf-β2−/− mice, expression of Kcc2, transcription factor Ap2β and KCC2 protein is downregulated. AP2β binds to Kcc2 promoter, a binding absent in Tgf-β2−/−. In hindbrain/brainstem tissue of Tgf-β2−/− mice, KCC2 phosphorylation at T1007 is increased and approximately half of pre-Bötzinger-complex neurons lack membrane KCC2 phenotypes rescued through exogenous TGF-β2. These results demonstrate that TGF-β2 regulates KCC2 transcription in immature neurons, possibly acting upstream of AP2β, and contributes to the developmental dephosphorylation of KCC2 at T1007. The present work suggests multiple and divergent roles for TGF-β2 on KCC2 during neuronal maturation and provides novel mechanistic insights for TGF-β2-mediated regulation of KCC2 gene expression, posttranslational modification and surface expression. We propose TGF-β2 as a major regulator of KCC2 with putative implications for pathophysiological conditions.


Introduction
K + /Cl − cotransporter 2 (KCC2; SLC12A5) mediates the electroneutral extrusion of K + and Cl − across neuronal membranes and its developmental upregulation controls the developmental "switch" in GABA function from depolarizing and excitatory in immature neurons to hyperpolarizing and inhibitory in mature neurons [1]. Independent of its transport function, KCC2 regulates the development and morphology of dendritic spines [2][3][4] and is also involved in ontogenetic cell death of cortical neuronal subpopulations [5]. Deficits in KCC2 function significantly contribute to the pathogenesis of several neurological disorders such as epilepsy and neuropathic pain, and can occur following traumatic brain injury [6,7]. Consequently, KCC2 is an attractive target for therapeutic treatment.

Animals
All protocols were carried out in accordance with German ethical guidelines for laboratory animals and approved by the Institutional Animal Care and Use Committee of the University of Freiburg and the ethics committee of the City of Freiburg (authorizations: X14/16H, X19/09C and G-21/140). Adult C57BL/6N mice (strain code 027) of either sex were maintained on a 12 h dark/light cycle with food and water ad libitum. Mice were sacrificed by cervical dislocation, and all efforts were made to minimize suffering. Embryos at embryonic day 17.5-18.5 were sacrificed by decapitation and brains were isolated. Subsequently, either hippocampus was excised for generation of primary neuronal cultures, or forebrain and hindbrain/brainstem tissue were separated or acute brainstem slices were cut and processed as described below. Tgf-β2-deficient mice have been described earlier [34].

Quantitative RT-PCR
Total RNA was isolated from DIV4 primary hippocampal neurons and tissue samples (forebrain or hindbrain/brainstem) from Tgf-β2 +/+ and Tgf-β2 −/− mice at E17.5 using TRIzol reagent (Invitrogen; Darmstadt, Germany) according to manufacturer's instructions. Subsequently, 1 µg of RNA was reverse-transcribed into cDNA and analyzed with quantitative real-time PCR. Primer sequences used in QRT-PCR are listed in Table 1. PCR conditions were the following: initial denaturation at 95 • C for 10 min, 40 cycles of denaturation at 95 • C for 15 s, annealing at the appropriate temperature of the primer pairs for 30 s, elongation at 72 • C for 30 s and final elongation at 72 • C for 10 min. All reactions were performed in triplicate and the mean Ct value was used to determine gene expression with the 2 −∆∆Ct method using Gapdh as housekeeping gene.

Chromatin Immunoprecipitation Assay (ChIP)
Chromatin immunoprecipitation assay was performed with an EZ-ChIP kit (Cat# 17-371, Millipore; Darmstadt, Germany) according to manufacturer's instructions with modifications. Briefly, forebrains from E17.5 Tgf-β2 +/+ and Tgf-β2 −/− embryos were dissected and cross-linked in 1% paraformaldehyde for 15 min at room temperature. After homogenization, cell pellets were lysed in SDS lysis buffer and sonicated on ice with 20 cycles of 10 s pulse on/50 s pulse off using a probe sonicator (UW 3100, Bandelin electronic, Berlin, Germany), to obtain 200-1000 bp DNA fragments. Lysates were incubated with anti-AP2β, anti-RNA polymerase II and mouse IgG on a rotating platform overnight at 4 • C. Immunocomplexes were collected with protein G agarose beads, eluates were reverse cross-linked and bound DNA was purified. Immunoprecipitated DNA was analyzed with quantitative real-time PCR using primers specific to the two AP2 binding sequences [20] in mouse Slc12a5 promoter (Genbank accession number NC_000068.8). Primer sequences are shown in Table 1. PCR conditions were the following: initial denaturation at 95 • C for 10 min, 50 cycles of denaturation at 95 • C for 15 s, annealing at the appropriate temperature of the primer pairs for 30 s, elongation at 72 • C for 30 s and final elongation at 72 • C for 10 min. For the quantification, the adjusted input Ct was calculated by subtracting 6.64 cycles (log2 of 100), as 1% of starting chromatin was used as input. Percent of input for each IP (or negative control IgG) was calculated as follows: 100*2 (Adjusted input Ct-IP Ct) .

Image Acquisition and Analysis
Images were acquired with a Leica TCS SP8 confocal microscope using an HC PL APO CS2 40×/1.30 or 63×/1.40 oil objective lens, and immunofluorescence intensity was analyzed as described previously [41]. Within each experiment, the confocal microscope settings (laser power, detector gain and amplifier offset) were kept the same for all scans in which the immunofluorescence intensity was compared. Z-stacks of 16-30 optical sections with a step size of 0.4 µm were taken for at least 3 separate fields of view for each experimental condition. Maximum intensity projections were created from the z-stacks. To quantify the protein expression, LAS X software was used to select the area of interest (neuronal soma) and measure the average fluorescence intensity within this area for each cell. Background subtraction was applied to the images.
The values of KCC2 fluorescence intensity within the soma of each cell were used for the classification of neurons with high/low KCC2. The average KCC2 fluorescence intensity of all wildtype neurons measured per experiment was set as threshold. For each genotype, neurons exhibiting intensity higher than the threshold were classified as 'high KCC2', whereas neurons exhibiting intensity lower than the threshold were classified as 'low KCC2'. The data are presented as percent of total cells counted per genotype.

Immunofluorescence on Tissue Sections
Mouse embryonic brains or acute brainstem slices were fixed with 4% PFA for 24 h or 2 h, respectively, cryoprotected in 15% sucrose/PBS, embedded in cryomatrix and cryosectioned at 10 µm. neurons exhibiting membrane KCC2 were counted in wildtype and Tgf-β2 −/− sections. Membrane KCC2 abundance was judged at high magnification and confirmed by line scans (peaks of fluorescence intensity at the periphery of the cells).

Surface Biotinylation of Acute Brainstem Slices
Acute brainstem slices from wildtype and Tgf-β2-deficient mouse embryos at E18.5 containing the PreBötC were subjected to control or experimental conditions (application of recombinant TGF-β2) and then kept on ice. Isolation of cell surface proteins was performed using the Pierce ® (Thermo Fisher Scientific; Dreieich, Germany) cell surface protein isolation kit following the manufacturer's instructions. Proteins were then processed for immunoblotting with antibodies for KCC2 and Na + /K + -ATPase, as described above.

Statistical Analysis
Statistical tests were performed as indicated in thetext. All tests were performed in GraphPad Prism, Version 7.04, GraphPad, San Diego, USA for Windows. Data were tested for normal distribution, using a Shapiro-Wilk test and subsequently assessed for homogeneity of variance. If the data passed both tests, further analyses were performed using the two-tailed unpaired Student's t-test. For datasets with unequal variances, Welch's correction was applied after Student's t-test. Values are reported as mean ± S.E.M., unless otherwise indicated. For datasets with non-normal distributions, the Mann-Whitney ranksum test was used. For all statistical tests, p ≤ 0.05 was considered statistically significant, and p-values are indicated in the figures as follows: * p ≤ 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, # p < 0.05 and ## p < 0.01.

KCC2 Expression in Immature Neurons Is Regulated by TGF-β/Activin Signaling
During maturation of most central neurons, the expression of KCC2 is upregulated, resulting in an intracellular Cl − concentration below its electrochemical equilibrium, thereby rendering GABA A responses to hyperpolarizing and inhibitory [1,42]. Early studies have shown that the mechanisms regulating KCC2 expression appear to involve signaling induced by trophic factors [23][24][25][26] rather than GABA A -receptor-mediated or ionotropic transmission or neuronal activity [42]. We have previously shown TGF-β-dependent and Rab11b-mediated KCC2 trafficking to the plasma membrane in differentiating and mature neurons [33]. Considering the context-dependent action of TGF-β, we first investigated the effect of TGF-β on KCC2 expression and/or localization in immature neurons.
Mouse hippocampal neurons were isolated at E17.5-18.5, cultured for 4 days in vitro, and treated either with 2 ng/mL TGF-β2 or with 10 µM SB431542,-an ALK4/5/7 inhibitor, thus, an inhibitor of TGF-β/activin signaling [35]-for 60 min. Subsequently, transcript expression of the two splice variants of Kcc2, Kcc2a and Kcc2b [18], protein abundance of total and phosphorylated (p)KCC2 as well as subcellular localization were assessed by QRT-PCR, immunoblotting and immunofluorescence microscopy, respectively. Figure 1A shows that exposure of the cultures to exogenous recombinant TGF-β2 had no effect on Kcc2a or Kcc2b transcript expression (1.35 ± 0.33 fold and 1.15 ± 0.27 fold for Kcc2a and Kcc2b, respectively, not significant using two-tailed unpaired Student's t-test with Welch's correction, n = 6), whereas treatment with SB431542 significantly downregulated Kcc2b (0.69 ± 0.06 fold, # p < 0.05 using two-tailed unpaired Student's t-test with Welch's correction, n = 4) but not Kcc2a (0.93 ± 0.20 fold) compared to untreated controls. Immunoblotting for total KCC2 ( Figure 1B) and pKCC2 at S940 and T1007 revealed prominent immunoreactive bands at~135-140 kDa in whole-cell homogenates from controls. Exposure of the cultures to exogenous recombinant TGF-β2 had no effect on total KCC2 or pKCC2 at S940 and T1007 protein expression ( Figure 1B,C; 0.93 ± 0.06 fold, 1.01 ± 0.08 fold and 1.04 ± 0.07 fold for total KCC2, pKCC2 at S940 and at T1007, respectively; not significant using two-tailed unpaired Student's t-test with Welch's correction, n = 5) compared to untreated controls. Furthermore, treatment with SB431542 did not affect protein expression of total KCC2 or on cDNA from primary immature mouse hippocampal neurons treated with either TGF-β2 (2 ng/mL) or 10 µM SB431542 (an ALK4/5/7 inhibitor) for 60 min ( # p < 0.05 for significant decrease compared to controls, using the two-tailed unpaired Student's t-test with Welch's correction, n = 4-6). (B) Representative blots and (C) quantification of protein abundance of total and phosphorylated (p)KCC2 at S940 and T1007 by immunoblotting in control immature hippocampal neurons and following treatment with 2 ng/mL TGF-β2 or the ALK4/5/7 inhibitor SB431542 (10 µM); not significant after densitometric analysis of the signal ratio protein of interest: GAPDH and two-tailed unpaired Student's t-test with Welch's correction, n = 5). A total of 20 µg protein was loaded per lane. (D) or 10 µM SB431542 (an ALK4/5/7 inhibitor) for 60 min ( # p < 0.05 for significant decrease compared to controls, using the two-tailed unpaired Student's t-test with Welch's correction, n = 4-6). (B) Representative blots and (C) quantification of protein abundance of total and phosphorylated (p)KCC2 at S940 and T1007 by immunoblotting in control immature hippocampal neurons and following treatment with 2 ng/mL TGF-β2 or the ALK4/5/7 inhibitor SB431542 (10 µM); not significant after densitometric analysis of the signal ratio protein of interest: GAPDH and two-tailed unpaired Student's t-test with Welch's correction, n = 5. A total of 20 µg protein was loaded per lane. (D) Immunofluorescence confocal microscopy for KCC2 (red) and pKCC2 at S940, T1007 or T906 on mouse primary immature hippocampal neurons. Asterisks indicate intracellular KCC2 localization. Scale bar: 20 µm. (E) Quantification of immunofluorescence intensity of total KCC2 and pKCC2 at S940, T1007 and T906 following treatment with TGF-β2 (2 ng/mL) or SB431542 (10 µM) (** p < 0.01 for significant increase and # p < 0.05 for significant decrease compared to controls using the two-tailed unpaired Student's t-test with Welch's correction, n = 3-7). All data are shown as mean ± S.E.M. The values of the respective controls were set to 1.
These results suggest that TGF-β/activin signaling is involved in regulation of KCC2 and provide first hints for a contribution of the ligand TGF-β2 in this process.
These results suggest that TGF-β/activin signaling is involved in expression of transcription factors with potential to regulate Kcc2 transcript on cDNA from primary immature mouse hippocampal neurons treated with either TGF-β2 (2 ng/mL) or 10 µM SB431542 (an ALK4, 5 and 7 inhibitor) for 60 min ( # p < 0.05 and ## p < 0.01 for significant downregulation compared to controls using the two-tailed unpaired Student's t-test with Welch's correction, n = 4-6). All data are shown as mean ± S.E.M. The values of the respective controls were set to 1.

TGF-β2-Dependent Regulation of Kcc2 Gene Expression and KCC2 Phosphorylation State in the Developing Mouse Forebrain
Since SB431542 inhibits ALK4/5/7 receptors, several ligands that signal through these receptors are putative candidates for regulating the expression of Kcc2 transcript ( Figure 1A) and/or transcription factors ( Figure 2) potent to regulate Kcc2 transcription. TGF-βs signal through ALK5, and the isoform TGF-β2 has been considered relatively more relevant during neuronal development compared to TGF-β3 [28]. With this background in mind, we investigated a putative relative contribution of the isoform TGF-β2 in regulating Kcc2 in immature neurons. Therefore, forebrain tissue from wildtype (Tgf-β2 +/+ ) and Tgf-β2 −/− embryos at E17.5 was isolated and, subsequently, Kcc2a and Kcc2b transcript expression, KCC2 whole protein and phosphorylation at S940 and T1007 was determined. Figure 3A shows that transcript expression of both Kcc2a and Kcc2b was significantly downregulated in the forebrain of Tgf-β2 constitutive null mutants (0.80 ± 0.05 and 0.83 ± 0.06 fold for Kcc2a and Kcc2b, respectively), compared to that of wildtype (wt) littermates (1.00 ± 0.02 and 1.00 ± 0.04 fold for Kcc2a and Kcc2b, respectively; ## p < 0.01 using two-tailed unpaired Student's t-test with Welch's correction; # p < 0.05 using two-tailed unpaired Student's t-test; n = 7 for wt and n = 6 for mutants). In line with these results, total KCC2 protein in homogenates from forebrain ( Figure 3B,C) was significantly downregulated in Tgf-β2 −/− embryos, compared to wt littermates (0.85 ± 0.04 vs. 1.00 ± 0.01 fold, ## p < 0.01 using twotailed unpaired Student's t-test with Welch's correction, n = 7 for wt and n = 6 for mutants).
These results demonstrate TGF-β2-dependent regulation of KCC2 T1007 phosphorylation. Since T1007 phosphorylation has inhibitory effects on KCC2 activity, these results implicate putative TGF-β2-dependent regulation of KCC2 activity through modulating T1007 phosphorylation state.

Impaired KCC2 Membrane Expression in Pre-Bötzinger Complex of Tgf-β2 −/− Mice
Based on the results in the hindbrain/brainstem showing that KCC2 phosphorylation at T1007 is considerably increased in Tgf-β2 −/− ( Figure 5E,F) together with previous reports showing that mice expressing homozygous phosphomimetic KCC2 mutations at T906/T1007 die early postnatally due to impaired rhythmogenesis in respiratory networks [14,15]-a phenotype that is similar to Tgf-β2 −/− mutants [34]-we examined the KCC2 phenotype in a distinct region of the ventrolateral brainstem that is considered to be the site for respiratory rhythm generation, namely the pre-Bötzinger complex (preBötC) [46]. This nucleus is one of the first to develop a functional neuronal network requiring GABAergic KCC2-mediated inhibition [47,48]. Moreover, we have previously reported that Tgf-β2 −/− mice show impaired synaptic neurotransmission, implying GABAergic and/or glutamatergic mechanisms [49]. Here, we tested the hypothesis that KCC2 membrane abundance may be impaired in preBötC neurons of Tgf-β2 −/− mutants.
A subset of neurons within the preBötC co-express the neurokinin 1 receptor (NK1R) and somatostatin (SST) [50]. Phox2b is expressed in the parafacial respiratory group within the ventrolateral medulla [51] and, hence, serves as a marker to distinguish the neurons of the preBötC from neighboring neuronal subpopulations ( Figure S1) in tissue sections from fixed mouse brains at E18.5. Figure 5. TGF-β2-dependent regulation of KCC2 in the mouse hindbrain/brainstem. (A) Quantitative RT-PCR analysis for Kcc2a and Kcc2b on cDNA from either mouse forebrain or hindbrain/brainstem at embryonic (E) day 17.5 (**** p < 0.0001 for significant upregulation compared to forebrain using two-tailed unpaired Student's t-test with Welch's correction, n = 10). (B) Immunoblot analysis for total KCC2 in tissue homogenates from mouse forebrain and hindbrain/brainstem at E17.5. (C) * p < 0.05 for significant increase after densitometric analysis of the signal ratio KCC2: GAPDH and Student's t-test, n = 6). Data are shown as mean ± SEM relative to the mean of forebrains. The blots are representative of six different litters. A total of 30 µg protein was loaded per lane. (D) Quantitative RT-PCR analysis for Kcc2a and Kcc2b on cDNA from hindbrain/brainstem at embryonic day (E) 17.5 of Tgf-β2 +/+ and Tgf-β2 −/− mice ( ## p < 0.01 using Mann-Whitney test, n = 7 for wildtype and n = 6 for mutant). (E) Immunoblot analysis for total KCC2 and for phosphorylated (p)KCC2 at S940 or T1007 in tissue homogenates from hindbrain/brainstem of Tgf-β2 +/+ and Tgf-β2 −/− mice. (F) *** p < 0.001 for significant increase after densitometric analysis of the signal ratio pT1007: GAPDH and twotailed unpaired Student's t-test, n = 6 for wildtype and n = 6 for mutant). The blots are representative for four different litters. A total of 30 µg protein was loaded per lane. Data are shown as mean ± SEM relative to the mean of wildtype littermates.

Impaired KCC2 Membrane Expression in Pre-Bötzinger Complex of Tgf-β2 −/− Mice
Based on the results in the hindbrain/brainstem showing that KCC2 phosphorylation at T1007 is considerably increased in Tgf-β2 −/− (Figure 5E, F) together with previous reports showing that mice expressing homozygous phosphomimetic KCC2 mutations at T906/T1007 die early postnatally due to impaired rhythmogenesis in respiratory networks [14,15]-a phenotype that is similar to Tgf-β2 −/− mutants [34]-we examined the KCC2 phenotype in a distinct region of the ventrolateral brainstem that is considered to be the site for respiratory rhythm generation, namely the pre-Bötzinger complex (preBötC) [46]. This nucleus is one of the first to develop a functional neuronal network requiring GA-BAergic KCC2-mediated inhibition [47,48]. Moreover, we have previously reported that Tgf-β2 −/− mice show impaired synaptic neurotransmission, implying GABAergic and/or glutamatergic mechanisms [49]. Here, we tested the hypothesis that KCC2 membrane abundance may be impaired in preBötC neurons of Tgf-β2 −/− mutants. of Tgf-β2 +/+ and Tgf-β2 −/− mice ( ## p < 0.01 using Mann-Whitney test, n = 7 for wildtype and n = 6 for mutant). (E) Immunoblot analysis for total KCC2 and for phosphorylated (p)KCC2 at S940 or T1007 in tissue homogenates from hindbrain/brainstem of Tgf-β2 +/+ and Tgf-β2 −/− mice. (F) *** p < 0.001 for significant increase after densitometric analysis of the signal ratio pT1007: GAPDH and two-tailed unpaired Student's t-test, n = 6 for wildtype and n = 6 for mutant. The blots are representative for four different litters. A total of 30 µg protein was loaded per lane. Data are shown as mean ± S.E.M. relative to the mean of wildtype littermates.
In fixed brainstem tissue sections from wt mouse embryos, KCC2 immunoreactivity is located in the periphery of preBötC neurons. Figure 6(A1-A3) shows line scans for KCC2 (red), actin (green) and DAPI (blue) for three randomly depicted neurons (numbered 1, 2 and 3). Peaks for KCC2 immunolabeling (arrows) are present at the periphery of neuronal cell bodies, suggesting labeling of the plasma membrane. In contrast, representative line scans for three neurons from the Tgf-β2 −/− mice ( Figure 6(B1-B3)) reveal a predominantly intracellular distribution of KCC2 immunoreactivity. We further quantified the KCC2 distribution pattern in somatostatin-positive preBötC neurons by analyzing 600 neurons/animal/genotype ( Figure 6C) and revealed that in Tgf-β2 −/− mice at E18.5, the number of cells exhibiting membrane KCC2 was significantly reduced (53.76 ± 3.28%; ## p < 0.01 using two-tailed unpaired Student's t-test with Welch's correction, n = 3) compared to wildtype littermates.
These results demonstrate reduced KCC2 membrane abundance in the preBötC of Tgf-β2 −/− mutants and implicate that this phenotype might contribute to the respiratory failure observed in Tgf-β2 −/− mice [34]. cell bodies, suggesting labeling of the plasma membrane. In contrast, representative line scans for three neurons from the Tgf-β2 −/− mice (Figure 6(B1-3)) reveal a predominantly intracellular distribution of KCC2 immunoreactivity. We further quantified the KCC2 distribution pattern in somatostatin-positive preBötC neurons by analyzing ~600 neurons/animal/genotype ( Figure 6C) and revealed that in Tgf-β2 −/− mice at E18.5, the number of cells exhibiting membrane KCC2 was significantly reduced (53.76 ± 3.28%; ## p < 0.01 using two-tailed unpaired Student's t-test with Welch's correction, n = 3) compared to wildtype littermates.

Rescue of KCC2 Phenotype in Tgf-β2 −/− Neurons by Exogenous TGF-β2
Impaired developmental dephosphorylation of KCC2 at T1007 in Tgf-β2 −/− hindbrain/brainstems can be partly rescued by exogenous TGF-β2, as illustrated in Figure 7A,B. Acute brainstem slices containing the preBötC from wt and Tgf-β2 mutant embryos at E17.5 were exposed to exogenous TGF-β2 (5 ng/mL) for 60 min and T1007 phosphorylation was subsequently determined by immunoblotting. The results show that when wt slices were exposed to exogenous TGF-β2, pKCC2 T1007 was comparable (1.05 ± 0.17 fold) to the untreated wt controls (1.00 ± 0.13 fold). Interestingly, in untreated Tgf-β2 −/− brainstem slices, pKCC2 T1007 was significantly upregulated (1.50 ± 0.08 fold) compared to wt, an effect that was rescued (1.22 ± 0.06 fold) following exposure of the slices to exogenous TGF-β2 for 60 min at the levels of the treated Tgf-β2 +/+ slices (* p < 0.05 and # p < 0.05 for significant upregulation and downregulation, respectively, using two-tailed unpaired Student's t-test; n = 3 for wt, n = 4 for mutants). Immunoblot analysis for phosphorylated (p)KCC2 at T1007 in homogenates from preBötC brainstem slices of Tgf-β2 +/+ and Tgf-β2 −/− mutant mice (B); * p < 0.05 for significant increase and # p < 0.05 for significant decrease after densitometric analysis of the signal ratio pT1007: GAPDH and two-tailed unpaired Student's t-test, n = 3 for Tgf-β2 +/+ and n = 4 for mutants. A total of 10 µg protein were loaded per lane. Data are shown as mean ± SEM relative to the mean of untreated wt controls. (C,D) Fixed cryosections immunolabeled for KCC2 (red). Nuclei were labeled with DAPI (blue). Representative line scans (out of three animals/genotype) for KCC2 (red) and DAPI (blue) illustrate KCC2 labeling pattern from randomly selected neurons for each experimental condition. Scale bar: 30 µm. (E,F) Immunoblot analysis following biotinylation of surface proteins of acute brainstem slices containing the preBötC from Tgf-β2 +/+ and Tgf-β2 −/− mutant mice in the presence or absence of exogenous TGF-β2 (* p ≤ 0.05 for significant increase after densitometric analysis of the signal ratio KCC2: Na + /K + -ATPase and two-tailed unpaired Student's t-test, n = 5/genotype; 50 µg protein was loaded per lane). The data are shown as mean ± SEM, relative to the untreated controls.

Discussion
The molecular mechanisms responsible for KCC2 regulation during maturation of neuronal networks are not fully elucidated. In the present study, we investigated the impact of TGF-β2 on regulating KCC2 in neurons at different developmental stages. We used primary immature neurons from mouse hippocampi at E17.5 and mouse forebrain and Figure 7. Application of exogenous recombinant TGF-β2 regulates phosphorylation and membrane expression of KCC2 in brainstem of Tgf-β2 −/− mutants. Acute slices from preBötC complex of E17.5-E18.5 from wildtype (Tgf-β2 +/+ or Tgf-β2 +/− ) and Tgf-β2 −/− mutants were cultured in the presence or absence of recombinant human TGF-β2 (5 ng/mL) for 60 min. (A,B) Immunoblot analysis for phosphorylated (p)KCC2 at T1007 in homogenates from preBötC brainstem slices of Tgf-β2 +/+ and Tgf-β2 −/− mutant mice (B); * p < 0.05 for significant increase and # p < 0.05 for significant decrease after densitometric analysis of the signal ratio pT1007: GAPDH and two-tailed unpaired Student's t-test, n = 3 for Tgf-β2 +/+ and n = 4 for mutants. A total of 10 µg protein were loaded per lane. Data are shown as mean ± S.E.M. relative to the mean of untreated wt controls. (C,D) Fixed cryosections immunolabeled for KCC2 (red). Nuclei were labeled with DAPI (blue). Representative line scans (out of three animals/genotype) for KCC2 (red) and DAPI (blue) illustrate KCC2 labeling pattern from randomly selected neurons for each experimental condition. Scale bar: 30 µm. (E,F) Immunoblot analysis following biotinylation of surface proteins of acute brainstem slices containing the preBötC from Tgf-β2 +/+ and Tgf-β2 −/− mutant mice in the presence or absence of exogenous TGF-β2 (* p ≤ 0.05 for significant increase after densitometric analysis of the signal ratio KCC2: Na + /K + -ATPase and two-tailed unpaired Student's t-test, n = 5/genotype; 50 µg protein was loaded per lane). The data are shown as mean ± S.E.M., relative to the untreated controls.
Notably, WNK1-mediated inhibitory phosphorylation of KCC2 at T1007 regulates KCC2 activity by altering its surface expression [11]. Therefore, we next assessed whether TGF-β2 treatment can rescue the Tgf-β2 −/− phenotype ( Figure 6). We used acute brainstem slice cultures containing the preBötC from Tgf-β2 −/− at E18.5 and cultured them in the presence or absence of exogenous TGF-β2 for 60 min. Subsequently, slices were fixed and processed for KCC2 immunofluorescence. Line scans from randomly depicted neurons are shown in Figure 7C,D. In Tgf-β2 +/− acute brainstem slices, KCC2 immunolocalization ( Figure 7C) was similar to wt ( Figure 6A), showing distinct peaks of line scans at the cell periphery (arrows). In contrast, consistent with the in vivo analysis ( Figure 6B), line scans in Tgf-β2 −/− acute brainstem slices revealed a predominantly intracellular localization of KCC2 immunoreactivity ( Figure 7D). Treatment of Tgf-β2 −/− acute slices with exogenous TGF-β2 resulted in a shift of KCC2 immunoreactivity to the cell periphery, as illustrated in the respective line scan ( Figure 7D). Figure 7C also reveals that treatment of Tgf-β2 +/− slices with TGF-β2 did not modify the localization pattern of KCC2 immunoreactivity.
Incorporation of KCC2 to the membrane in Tgf-β2 −/− following TGF-β2 treatment of acute brainstem slices was further determined by surface biotinylation experiments. As shown in Figure 7E,F, biotinylation of cell surface proteins showed a significant increase for the KCC2~135kDa band in Tgf-β2 −/− acute slices, treated with exogenous recombinant TGF-β2 for 60 min (1.24 ± 0.1 fold; * p ≤ 0.05 using two-tailed unpaired Student's t-test, n = 5), compared to the untreated slices. In contrast, KCC2 surface expression was not altered in wt acute slices following application of TGF-β2 (0.95 ± 0.05 fold).
These data highlight the biological significance of TGF-β2 for the subcellular distribution of KCC2 in vivo.

Discussion
The molecular mechanisms responsible for KCC2 regulation during maturation of neuronal networks are not fully elucidated. In the present study, we investigated the impact of TGF-β2 on regulating KCC2 in neurons at different developmental stages. We used primary immature neurons from mouse hippocampi at E17.5 and mouse forebrain and hindbrain/brainstem tissue at E17.5 from TGF-β2 mutants. The advantage, thereby, is that at this developmental stage, these brain areas reveal distinct timetables for neurogenesis. Whereas brainstem neurons show the earliest maturation, in the forebrain, neuronal development is largely delayed.
Our results show multiple and divergent roles of action of TGF-β2 on KCC2 during neuronal maturation. The main novel findings are: (1) TGF-β2 positively regulates KCC2 transcription in immature neurons, possibly via binding of AP2β in the Kcc2 promoter; (2) TGF-β2 contributes to the developmental regulation of the inhibitory phosphorylation at T1007; and (3) TGF-β2 regulates membrane expression of KCC2 in the differentiated neurons of the preBötC.

TGF-β2-Dependent Regulation of KCC2 Transcription in Immature Neurons
The results show that in immature mouse hippocampal neurons, a time point when KCC2 is not yet functional [52,53], inhibition of the ALK receptors 4/5/7 significantly downregulates Kcc2b transcript, whereas in gain-of-function experiments-by exposing the immature neurons to exogenous TGF-β2-no effect on Kcc2 transcript expression was detected ( Figure 1A). The latter observation can be explained by the presence of endogenously expressed TGF-β2. Indeed, endogenous TGF-β2 may saturate signaling so that the application of exogenous TGF-β2 has no further effects while the receptor inhibitor does. However, in gain-of-function experiments, KCC2 immunofluorescence is significantly upregulated. With regard to KCC2 whole-protein and pKCC2 T1007 abundance following application of exogenous TGF-β2, there is an inconsistency between the immunoblot data ( Figure 1C) that show no protein expression changes and the immunofluorescence data ( Figure 1E) that suggest altered protein expression levels. However, it is reasonable to assume that there may well be differences within the cellular populations in the hippocampus that give rise to differences between gross protein expression levels and immunofluorescence in single cells. Moreover, quantification of KCC2 immunofluorescence intensity was performed exclusively in the neuronal soma and not in the dendrites; therefore, the effect could be "diluted" in the whole-protein samples used for Western blot analysis. In forebrain tissue of E17.5 Tgf-β2 mutants, KCC2 expression was found significantly decreased at both the transcript and protein level, and in primary hippocampal neuronal cultures, KCC2 immunofluorescence intensity was significantly decreased as well, accompanied by a reduced number of hippocampal neurons expressing high KCC2 abundance (Figure 3), suggesting that TGF-β2 may contribute to the generation of an initial KCC2 vesicular pool that can potentially be recruited at later stages of development. Consistent with this view, the KCC2 protein was present in the somata and dendrites of immature neurons in an intracellular distribution manner rather than plasma membrane-bound. The nature of "high KCC2"-and "low KCC2"-expressing neurons is not clear; however, these experiments were performed in mouse primary hippocampal cultures, containing a mixture of neurons at different (not synchronized) developmental stages. Our results also demonstrate that the effects observed after pharmacological blocking of the TGF-β/activin pathway can be at least partly attributed to TGF-β2. It should be noted, however, that other ligands that signal through ALK4/5/7 may regulate Kcc2 transcript as well. With regard to other TGF-β isoforms, TGF-β3,-an isoform expressed in the cells as well-cannot compensate for the loss of TGF-β2, demonstrating that the observed actions are isoform-specific. These results extend the molecular network responsible for triggering-and thereafter, progressively upregulating-KCC2 gene expression during development by introducing TGF-β2 as a novel molecular determinant and highlighting a growth-factor-dependent increase in Kcc2 transcription.
Indeed, to date, only neurotrophins, namely BDNF and neurturin, are identified as inducing significant ERK1/2-dependent upregulation of the Egr4 transcription factor and increasing Kcc2 mRNA and protein levels in immature cultured hippocampal neurons [23,24]. However, the effects of BDNF in immature neurons are controversial. Increased Kcc2 mRNA levels were found in embryos overexpressing BDNF [54], and KCC2 expression was decreased in early postnatal TrkB-deficient mice [55]. On the other side, surprisingly, Bdnf −/− mice do not exhibit any phenotype for either developmental upregulation or function of KCC2 [56]. Besides EGR4, the transcription factors USF1 and USF2 activate [19], whereas REST represses Kcc2 transcription [21,22]. Here, we show that expression of the transcription factors Ap2β, Sp1, Egr4 and cFos was significantly downregulated in immature neurons following blocking of the TGF-β/activin pathway (Figure 2), but only Ap2β expression was downregulated in forebrains from Tgf-β2 null mice. Subsequent ChIP assay revealed that the transcription factor AP2β binds to the Kcc2 promoter. Notably, this binding was absent in Tgf-β2 mutants, implicating that a TGF-β2-dependent increase in Kcc2 transcription may be mediated via the non-canonical pathway and subsequent binding of AP2β to the Kcc2 promoter. Further functional studies are needed to provide conclusive evidence for activation of Kcc2 promoter through TGF-β-mediated AP2β binding. Moreover, the observation that inhibition of the ALK4/5/7 pathway downregulated several transcription factors with putative binding sites on Kcc2 promoter implies that other ligands may regulate Kcc2 transcript by unknown TGF-β-independent-signaling as well. Both BDNF and TGF-β2 exert different actions relative to KCC2 in immature and differentiating/mature neurons. BDNF downregulates KCC2 transcript and protein in an activity-dependent manner [25,26], whereas TGF-β2 has an impact on KCC2 posttranslational modifications, as will be discussed below.

TGF-β2-Dependent Regulation of KCC2 Phosphorylation at T1007
A further novel finding of the present work is the TGF-β2-dependent regulation of KCC2 phosphorylation at T1007. The following lines of experimental evidence support this view: (1) exposure of immature neurons to exogenous TGF-β2 significantly reduced T1007 phosphorylation ( Figure 1E); (2) in hindbrain/brainstem tissue of Tgf-β2-deficient mice T1007 phosphorylation was significantly increased ( Figure 5F); and (3) increased T1007 phosphorylation in acute brainstem slices containing the preBötC in the Tgf-β2 mutants was prevented following treatment with exogenous TGF-β2 ( Figure 7B). To our knowledge, this is the first demonstration of growth-factor-dependent regulation of posttranslational modification of KCC2. It is unequivocally demonstrated that alterations in KCC2 phosphorylation contribute in regulating functional KCC2 expression, and even the developmental upregulation of KCC2 activity is at least partially mediated through (de)phosphorylation events. However, it has also been appreciated that phosphoregulation of KCC2 is complex, and likely more than one phosphosite contributes in stimulus-induced activation or inhibition of KCC2, as reflected in the observation that staurosporine, a broad kinase inhibitor, increases KCC2 activity in immature neurons [12]. T906/T1007 phosphorylation is decreasing in the course of neuronal maturation, is mediated by the WNK-SPAK/OSR1 cascade and inversely correlates with KCC2 function [11][12][13]. Along this line, in mice with constitutive dephosphorylation of T906/T1007, the onset of seizures is delayed and the severity of seizures attenuated [16]. Our results demonstrate increased T1007 phosphorylation in hindbrains of Tgf-β2 mutants, suggesting that TGF-β2 signaling, either by activation of a phosphatase and/or inhibition of WNK1-SPAK/OSR1 or other kinase, regulates (de)phosphorylation of KCC2 at T1007. The exact signaling cascade or cross-talk between pathways needs to be elucidated.
In contrast, PKC-mediated phosphorylation at S940 enhances KCC2 function [10], and in mice harboring the S940 mutation, the development and lethality of status epilepticus was accelerated [17]. Our results show that phosphorylation of KCC2 at S940 is neither regulated by TGF-β2 in immature (Figures 1B-E and 3B-E) nor mature ( Figure 5F) neurons. Taken into consideration the numerous putative KCC2 phosphosites and the several noncanonical TGF-β pathways, it is likely that additional phosphosites may be regulated by TGF-β2, thereby contributing to dynamic activating and inhibitory (de)phosphorylation events that are required for fine tuning of KCC2 transport capacity. Since the necessary tools are, therefore, lacking, a phosphoproteomic approach in wt and Tgf-β2 mutants could be a way forward.
Phosphorylation-induced transport of KCC2 to the cell membrane versus intrinsic properties, by means of transport activity, has been controversial. In mice harboring S940 mutation, surface expression of KCC2 was comparable between wild type and mutants [17], another study however, showed decreased rate of KCC2 internalization from the plasma membrane due to PKC-dependent phosphorylation at S940 [10]. Mice with constitutive dephosphorylation T906A/T1007A reveal no difference on KCC2 surface expression [16], whereas T906 and T1007 phosphorylation was shown to regulate KCC2 activity by altering its surface abundance [11]. We tested the hypothesis that increased T1007 phosphorylation in hindbrain/brainstem of Tgf-β2 mutants is linked to decreased membrane KCC2 expression.

TGF-β2-Dependent Regulation of Membrane KCC2 in the PreBötC
In the context of the present work it is important that in vivo, mice expressing homozygous phosphomimetic KCC2 mutations at T906/T1007 die early postnatally due to impaired rhythmogenesis in respiratory networks [14,15]. Strikingly, Tgf-β2 −/− and Kcc2 −/− mice reveal similar phenotypes, exhibiting impaired respiratory rhythmogenesis, and both mutations are lethal immediately after birth due to respiratory failure [34,47]. The preBötC neurons in the brainstem are responsible for respiratory rhythmogenesis [48] and show the earliest neuronal maturation. Here, we show that in Tgf-β2 −/− mutants about half of preBötC neurons lack membrane KCC2 compared to the wildtype littermates ( Figure 6), an observation matching increased T1007 phosphorylation in brainstem slices containing the preBötC (Figure 7A,B). Interestingly, exogenous TGF-β2 rescued the increased T1007 phosphorylation and increased membrane KCC2 in brainstem slices containing the preBötC in Tgf-β2 mutants ( Figure 7C-F) as determined by immunofluorescence and following biotinylation of surface proteins. In a previous study, we have also shown that TGF-β signaling regulates trafficking of KCC2 to the neuronal membrane in differentiating and mature neurons via Rab11b [33]. Taken into consideration that phosphorylation events have potential to rapidly modulate KCC2, we suggest that TGF-β2 is able to regulate KCC2 by several modes of action that may occur at the same time or sequentially in a context-dependent manner. We can even speculate that deficits in membrane-bound KCC2 alone-either through T1007 phosphorylation-dependent reduced membrane KCC2 abundance in preBötC neurons and/or impaired Rab11b-mediated KCC2 trafficking, or in conjunction with other synaptic impairments-could contribute to the respiratory failure associated with the Tgf-β2 −/− phenotype [49].
In summary, the present work has provided new mechanistic insights for TGF-β2mediated regulation of KCC2 gene expression, posttranslational modification and membrane expression. We therefore propose TGF-β2 as a major regulator of KCC2 with putative implications during pathophysiological conditions.