Phosphoregulation of the intracellular termini of K+-Cl− cotransporter 2 (KCC2) enables flexible control of its activity

The pivotal role of K+-Cl− cotransporter 2 (KCC2) in inhibitory neurotransmission and severe human diseases fosters interest in understanding posttranslational regulatory mechanisms such as (de)phosphorylation. Here, the regulatory role of the five bona fide phosphosites Ser31, Thr34, Ser932, Thr999, and Thr1008 was investigated by the use of alanine and aspartate mutants. Tl+-based flux analyses in HEK-293 cells demonstrated increased transport activity for S932D (mimicking phosphorylation) and T1008A (mimicking dephosphorylation), albeit to a different extent. Increased activity was due to changes in intrinsic activity, as it was not caused by increased cell-surface abundance. Substitutions of Ser31, Thr34, or Thr999 had no effect. Additionally, we show that the indirect actions of the known KCC2 activators staurosporine and N-ethylmaleimide (NEM) involved multiple phosphosites. S31D, T34A, S932A/D, T999A, or T1008A/D abrogated staurosporine mediated stimulation, and S31A, T34D, or S932D abolished NEM-mediated stimulation. This demonstrates for the first time differential effects of staurosporine and NEM on KCC2. In addition, the staurosporine-mediated effects involved both KCC2 phosphorylation and dephosphorylation with Ser932 and Thr1008 being bona fide target sites. In summary, our data reveal a complex phosphoregulation of KCC2 that provides the transporter with a toolbox for graded activity and integration of different signaling pathways.

K ϩ -Cl Ϫ cotransporter 2 (KCC2) 2 plays a pivotal role in inhibitory neurotransmission. Under normal physiological conditions, KCC2 mediates outward transport of K ϩ and Cl Ϫ , thereby lowering the intracellular Cl Ϫ concentration ([Cl Ϫ ] i ) in neurons (1)(2)(3). This renders the action of GABA or glycine hyperpolarizing as their receptors are ligand-coupled Cl Ϫ channels. KCC2 exists in two isoforms, KCC2a and KCC2b, which differ in their N termini due to alternative promotors and first exon usage (4). In the adult, KCC2b is the most prominent isoform (5). Mice with disruption of the gene Slc12a5 encoding both KCC2a and KCC2b die shortly after birth due to motor deficits (6). KCC2b-deficient mice survive up to 3 weeks postnatally (7), whereas KCC2a-deficient mice show no obvious phenotype (8).
Dysregulation of KCC2 is associated with several neurological and psychiatric disorders, including epilepsy, neuropathic pain, spasticity, ischemic insults, brain trauma, schizophrenia, and autism (9 -19). This renders KCC2 a prime pharmacotherapeutic target and fosters interest in understanding posttranslational mechanisms of its regulation (20 -25). Among those, phosphorelated mechanisms are most intensively scrutinized. The broad-spectrum kinase inhibitor staurosporine enhances KCC2 transport activity in hippocampal neurons (26). N-Ethylmaleimide (NEM) enhances KCC transport activity as well and is thought to act on the same regulatory kinases as staurosporine (27)(28)(29)(30). In contrast, the protein phosphatase inhibitors calyculin and okadaic acid block KCC activation by cell swelling (31,32).
Here, we analyzed the impact of bona fide phosphosites identified in large-scale phosphoproteomics studies on KCC2 transport activity. These residues were substituted by either alanine or aspartate in mouse KCC2b to block or mimic phosphorylation, respectively. Subsequently, transport activity measurements were performed in HEK-293 cells to determine KCC2 transport activity in the presence of various agents.

Database mining and evolutionary conservation of KCC2 phosphosites
The databases PhosphoSitePlus and PHOSIDA subsume all experimentally observed phosphosites that were identified by MS-based proteomics studies (51,52 The remaining phosphosites are located in the shared C terminus of both KCC2 isoforms. In the following, we focus our analyses on the two N-terminal Ser 31 and Thr 34 and the three C-terminal Ser 932 , Thr 999 , and Thr 1008 phosphosites as they were so far not characterized. Multiple sequence alignment of vertebrate KCC subfamily members revealed that these phosphosites cover different patterns of phylogenetic conservation (Fig. 1). The phosphosite Thr 1008 is highly conserved throughout all vertebrate KCC isoforms. Thr 999 is moderately conserved throughout KCC2 and KCC3 and partially conserved in KCC1 and KCC4. The N-terminal phosphosite Ser 31 is highly conserved in orthologous KCC2 members and is rarely observed in KCC1 and KCC4, whereas Thr 34 is mainly present in therian KCC2. Ser 932 is highly conserved in vertebrate KCC2 and nontherian KCC4. This phosphosite is located in exon 22, which also harbors the previously analyzed and highly conserved phosphosites Thr 934 , Ser 937 , and Ser 940 (33).

Expression analyses of KCC2 phosphomutants
Tostudytheposttranslationalregulatoryimpactofphosphorylated Ser 31 , Thr 34 , Ser 932 , Thr 999 , and Thr 1008 for KCC2b (Mus musculus (mm)KCC2b) function, we generated two mutants for each phosphosite mimicking the phosphorylated (mutation into aspartate) or dephosphorylated (mutation into alanine) status. This resulted in a total of 10 mutants (KCC2 S31A/D , KCC2 T34A/D , KCC2 S932A/D , KCC2 T999A/D , and KCC2 T1008A/D ). To analyze whether these mutations affect the expression pattern of KCC2, the constructs were transiently expressed in HEK-293 cells. All mutants showed a transfection rate in HEK-293 cells similar to KCC2 WT (Fig. S1). Furthermore, immunoreactivity against all mutants was detected at the plasma membrane and in the cytosol (Fig. 2). Only the nucleus was spared. Thus, all mutants were well expressed in HEK-293 cells and therefore suitable for transport activity measurements.

Regulatory role of S932D and T1008A
To investigate the regulatory relevance of the five bona fide phosphosites, we determined the transport activity of phosphomimetic (aspartate) and dephosphomimetic (alanine) mutations of mmKCC2 by Tl ϩ flux measurements. All mutants as well as the KCC2 WT displayed at least a 2.9-times increased transport activity compared with mock-transfected cells (Fig. 3, Table 2, and Fig. S2). The loop diuretic furosemide, which specifically inhibits the function of cation-chloride cotransporters (53,54), blocks most of the flux, demonstrating that the transport activity was largely mediated by KCC2 (see Fig. 5, Table 2, and Fig. S2).
Analogous to the previously described N-terminal Ser 25 and Ser 26 phosphosites (27), substitution of Ser 31 and Thr 34 by alanine or aspartate did not alter transport activity compared with KCC2 WT . Similar results were obtained for the C-terminal mutants S932A, T999A/D, and T1008D. All four mutants showed a transport activity indistinguishable from KCC2 WT ( Fig. 3 and Table 2). In contrast, substitution of Ser 932 by aspartate and Thr 1008 by alanine significantly enhanced KCC2 transport activity, albeit to a different extent ( Fig. 3 and Table 2). To conclude, phosphorylation of the C-terminal residue Ser 932 and dephosphorylation of Thr 1008 result in an increased KCC2 transport activity in HEK-293 cells.

No change in abundance and cell-surface expression of KCC2 S932D and KCC2 T1008A
To analyze whether the enhanced activity of the two phosphomutants is due to increased abundance at the cell surface, we performed biotinylation assays for KCC2 WT , S932D, and T1008A (Fig. 4). KCC2 WT displayed a cell-surface localization of 16.7 Ϯ 1.6% compared with the total KCC2 protein amount.

Regulation of KCC2 by multiple phosphosites
transport activity of S932D and T1008A is caused by intrinsically kinetic effects and not by changes in KCC2 cell-surface abundance.

Staurosporine and NEM differentially affect KCC2 phosphomutants
Staurosporine is a broad kinase inhibitor that generally activates KCCs (27,57,58). To investigate whether Ser 31 , Thr 34 , Ser 728 , Ser 932 , Thr 999 , or Thr 1008 is an indirect target of staurosporine, we determined the transport activity of the corresponding mutants upon application of 8 M staurosporine. S31A, T34D, and T999D mutants were stimulated by staurosporine to a similar extent as KCC2 WT (Fig. 5 and Table 3). In contrast, mutations S31D, T34A, S932A/D, T999A, and T1008A/D abrogated stimulation by staurosporine. These data indicate that all five amino acid residues tested, Ser 31 , Thr 34 , Ser 932 , Thr 999 , and Thr 1008 , are involved in the staurosporinemediated effect.
As described previously, NEM, which generally activates KCCs (27,57,58), closely mimics the effect of staurosporine (27). To investigate whether this assumption holds true, we analyzed the impact of NEM on the transport activity of KCC2 S31A/D , KCC2 T34A/D , KCC2 S728A/D , KCC2 S932A/D , KCC2 T999A/D , and KCC2 T1008A/D . As observed for staurosporine, NEM still stimulated transport activity of the mutant T999D and failed to enhance the transport activity of S932D.
Contrary to staurosporine, NEM also enhanced the transport activity of S31D, T34A, T932A, T999A, and T1008A/D, whereas it did not affect the transport activity of S31A, T34D, or S932D. Thus, we demonstrate for the first time that staurosporine and NEM can differentially affect KCC2-mediated transport activity.

Discussion
The main function of KCC2 is to establish a low intracellular [Cl Ϫ ] as this is required for the hyperpolarizing action of GABA and glycine. Decreased Cl Ϫ extrusion by dysregulated KCC2 activity has been associated with a variety of disorders (12,18,19,23,25). One potent mechanism to rapidly and reversibly regulate the intrinsic transport rate and cell-surface abundance of KCC2 is phosphorylation and dephosphorylation of serine, threonine, and tyrosine residues (23). Recently, some progress in our understanding of the regulatory impact of specific phosphosites (27,34,35,39,41,43) and underlying kinases and phosphates (35)(36)(37)(43)(44)(45)(46) has been made.
Here, we focused on the functional characterization of bona fide phosphorylation sites of KCC2 that were identified by MS analysis of brain tissue. Site-directed mutagenesis in combination with transport activity measurements identified a potent role of Ser 932 and Thr 1008 in phosphoregulation. Phosphorylation of Ser 932 (mimicked by S932D) and dephosphorylation of Thr 1008 (mimicked by T1008A) increase KCC2 activity. In principle, such an increase can arise from an increased cell-surface abundance of the protein or from an intrinsic conformational change (20,23). Surface biotinylation experiments demonstrated that surface abundance of S932D and T1008A were indistinguishable from KCC2 WT . Thus, their increased transport activity resulted from intrinsically kinetic changes. Similar results were reported previously for the mutants T934D and S937D (27).
Ser 932 is located within exon 22, which is exclusively present in vertebrate KCC2 and nontherian KCC4. This exon also harbors the previously characterized phosphorylation sites Thr 934 , Ser 937 , and Ser 940 (27,43). Dephosphorylation of Thr 934 and Ser 937 , mimicked by mutation to alanine, results in transport activity similar to KCC2 WT , and phosphorylation of both residues, as mimicked by mutation to aspartate, intrinsically increases KCC2 transport activity (27). Furthermore, PKC-directed phosphorylation of Ser 940 enhances KCC2 cell-surface stability and increases ion transport, whereas mutation of Ser 940 to alanine results in transport activity that is equal to or decreased compared with KCC2 WT (34,43,59). Thus, phosphorylation of each phosphosite so far described in exon 22 stimulates KCC2 function. Exon 22 phosphosites are therefore in strikingly contrast to non-exon 22 phosphosites such as Thr 906 and Thr 1006 . The latter two are highly conserved throughout KCCs, and their dephosphorylation increases KCC2 transport activity (27,34,37,39). In addition, Thr 1008 , analyzed in the present study, that is near Thr 1006 conforms to this observation as dephosphorylation augments KCC2 activity as well. Thus, KCC2 harbors two main regulatory principles in which phosphorylation of exon 22-specific phosphosites and dephosphorylation of highly conserved KCC phosphosites outside exon 22 increase KCC2 transport activity. Therefore, con- Table 2 Transport activity under basal conditions and in the presence of furosemide **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant.

Regulation of KCC2 by multiple phosphosites
trary to other KCC isoforms, which are only up-regulated upon dephosphorylation (39,60), posttranslational regulation of KCC2 is more complex. This likely equips KCC2 with a more fine-grained phosphoregulatory mechanism as well as an enlarged capacity to integrate multiple signaling cascades compared with other KCC isoforms. So far, the underlying regulatory mechanisms that lead to phosphorylation and dephosphorylation of specific KCC2 phosphosites have only be partially identified. Ser 940 is phosphorylated via PKC (43). The phosphosites Thr 6 of KCC2a and Thr 906 andThr 1006 areregulatedbytheWNK-SPAK/OSR1phosphorylation cascade (35-37, 39, 47). Toward the identification of kinases involved in the phosphosites analyzed in this study, we treated HEK-293 cells with staurosporine or NEM. In general, both enhance KCC2 transport activity (27, 28, 56, 61, 62). Recent analyses by Deeb and co-workers (28) demonstrated Figure 5. Effect of furosemide, staurosporine, or NEM on transport activity of KCC2 phosphomutants. HEK-293 cells were transfected with KCC2 WT or KCC2 phosphomutant constructs. Transport activity was determined by performing Tl ϩ flux measurements. Furosemide, staurosporine, or NEM was added 15 min prior to measurement. KCC2 WT and all phosphomutants were significantly inhibited by furosemide. KCC2 WT , KCC2 S31A , KCC2 T34D , and KCC2 T999D were significantly activated by staurosporine. KCC2 WT , KCC2 S31D , KCC2 T34A , KCC2 S932A , KCC2 T999A/D , and KCC2 T1008A/D were significantly activated by NEM. The graph represents data of at least four independent measurements (each consisting of two (treated cells) or three (untreated cells) technical replicates) normalized to KCC2 WT . Statistical analysis is presented in Table 3. **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant. F, furosemide; U, untreated cells; S, staurosporine; N, NEM. Squares indicate outliers with respect to the 1.5 ϫ interquartile range of the box plot. Error bars represent S.E.

Regulation of KCC2 by multiple phosphosites
that NEM leads to increased phosphorylation of Ser 940 and decreased phosphorylation of Thr 1007 (numbering according to rat KCC2b). The latter might be explained by the observation that NEM also reduces the phosphorylation level of SPAK, which is involved in phosphorylation of Thr 1007 (28). Interestingly, NEM cannot stimulate the transport activity of T1007E, indicating that the induced activation of KCC2 is mediated by dephosphorylation of Thr 1007 (28). Here, we show that the phosphomutants S31A, T34D, and S932D also abolish the NEM-mediated increase in KCC2 transport activity. Our results reveal that more than one phosphosite partakes in NEM-induced activation of KCC2.
Previous data indicate that NEM and staurosporine act via the same regulatory mechanism (27). This actually holds true for Thr 934 and Ser 937 (27). Our analyses demonstrate for the first time that NEM and staurosporine can differentially impact KCC2 transport activity. NEM did not enhance transport activity of S31A and T34D, whereas staurosporine did not stimulate S31D, T34A, S932A/D, T999A, and T1008A/D. This points to partially different regulatory mechanisms being involved in stimulation of these phosphosites.
Furthermore, two categories of phosphosites were identified. The first category consists of the transport activity-regulatory phosphosites Ser 932 and Thr 1008 where mutation to either alanine or aspartate abrogated stimulation by staurosporine. This is expected for a site directly targeted by the staurosporinemediated action. In line with this result, alteration of the phosphorylation status increased activity (S932D and T1008A). Together, these data strongly suggest that both amino acid residues are bona fide target sites of proteins affected by the kinase inhibitor. Additionally, these results reveal that staurosporine acts by triggering both phosphorylation (S932D) and dephosphorylation (T1008A). In the case of Thr 1008 , this kinase inhibitor might inhibit a kinase that opposes KCC2 activation by phosphorylation of this site. In the case of Ser 932 , the kinase inhibitor might indirectly inhibit a phosphatase that dephosphorylates this site and thereby opposes KCC2 activation. It is currently unclear why mutation of either site alone abolished significant stimulation by staurosporine. One explanation is a functional cross-talk of both sites. Furthermore, we still noted a trend toward stimulated activity of the single mutants.
The second category consists of Ser 31 and Thr 34 for staurosporine and NEM, Thr 999 for staurosporine, and Ser 932 for NEM. In these cases, only mutation into one amino acid residue (alanine or aspartate) abrogated stimulation, whereas mutants containing the other amino acid residue replacement were still sensitive to the respective agent. One explanation is that the kind of substitution at these sites defines accessibility to other phosphosites such as Ser 932 or Thr 1008 in the case of staurosporine. One conformational state occludes hidden sites (S31A, T34D, and S932D for NEM and S31D, T34A, and T999A for staurosporine), resulting in no further activation of KCC2. The other conformational state deoccludes hidden sites (S31D, T34A, and S932A for NEM and S31A, T34D, and T999D for staurosporine), leading to simulation upon staurosporine or NEM treatment, respectively. Thus, (de)phosphorylation of specific phosphosites likely results in different conformational states of KCC2 termini that have a long-range effect on other phosphosites. They therefore indirectly regulate the transport activity of KCC2. This hypothesis is in line with previous work of Forbush and co-workers (63,64). They provided evidence that phosphorylation of N-terminal NKCC1 phosphosites leads to movements of transmembrane domains 10 and 12 relative to each other (63,64). Furthermore, the existence of long-range effects on the phosphostatus of KCC2 is supported by the observation that the mutation R1049C causes a significant Ͼ50% decrease in Ser 940 phosphorylation (11). Full understanding of this issue likely requires the 3D structure of KCC2 to modulate the impact of different phosphosites.
In conclusion, a detailed functional analysis of KCC2 amino acid residues revealed a complex phosphoregulatory landscape. Full understanding of phosphoregulation therefore requires a 3D model of KCC2 and the interacting effectors to simulate the competitive (de)phosphorylation events and the associated conformational changes. In addition, mass spectrometric characterization of KCC2 phosphosites under different treatments would be of advantage. These types of analyses will clarify how single and combined (de)phosphorylation events influence the conformational changes of KCC2 and thus its transport activity. Initial progress toward this goal has been made and revealed a flexible multidomain organization of KCC2 (65). Furthermore, future studies should address the role of phosphosites in KCC2 trafficking. Most studies, including our own, analyze phosphosites in nonneuronal cells with a simplified morphology compared with the high degree of compartmentalization in neurons. The N terminus of KCC2, for instance, is required for Table 3 Transport activity in the presence of staurosporine or NEM **, p Ͻ 0.01; ***, p Ͻ 0.001; n.s., not significant.

Regulation of KCC2 by multiple phosphosites
cell-surface delivery in neurons (66). Interestingly, this part contains various phosphorylation sites with unknown function.
KCC protein sequences for a diverse selection of organisms were obtained from a combination of BLAST searches against GenBank TM and data mining of the Ensembl database and the Joint Genome Institute (http://www.jgi.doe.gov/). The protein sequences of human KCC1 (NP_005063.1), KCC2 (NP_065759.1), KCC3 (NP_598408.1), and KCC4 (NP_006589.2) were used as queries. For each protein in each target species, we saved all sequences with an E-value of at least 10 Ϫ2 . These sequences were then reverse blasted (BLASTp or translated BLAST) against the Homo sapiens protein database, and only those protein sequences were retained that showed the same CCC protein sequence of H. sapiens that was used as a query sequence as the best hit (E-value of at least 10 Ϫ2 ). Each obtained sequence was then aligned at the amino acid level using the default settings in MUSCLE (67) as implemented in SeaView v.4.4.2 (68) and manually improved by eye thereafter.

Construction of expression clones
Site-directed mutagenesis of mouse expression clones for KCC2b (GenBank accession NM_020333.2) was performed according to the QuikChange mutagenesis system (Stratagene, Heidelberg, Germany). Forward oligonucleotides for the generation of the mutations are given in Table 4. Only sequenceverified clones were used for this study.

Cell culture
For immunocytochemistry, K ϩ -Cl Ϫ cotransporter activity measurements, and cell-surface biotinylation assays, HEK-293 cells were transiently transfected with the respective construct using polyethylenimine (Sigma-Aldrich). 3 h prior to transfection the medium was replaced. Briefly, 400 l of Dulbecco's modified Eagle's medium (Invitrogen), 18 l of polyethyleni-mine, and ϳ4.5 g of DNA were mixed and incubated for 15 min at room temperature prior to transfection. For K ϩ -Cl Ϫ cotransport activity measurements, HEK-293 cells were plated in a 0.1 mg/ml poly-L-lysinecoated 96-black well culture dish (Greiner Bio-One, Frickenhausen, Germany) at a concentration of 100,000 cells/well 24 h after transfection. The remaining cells were plated on a 0.1 mg/ml poly-L-lysinecoated glass coverslip. After ϳ18 h, coverslips were processed for immunocytochemical analyses to determine transfection rates, which were routinely between 20 and 30% (Fig. S1).

Determination of K ؉ -Cl ؊ cotransport
KCC2 transport activity was determined by Cl Ϫ -dependent uptake of Tl ϩ in HEK-293 cells as described previously (27,55,56). To initiate the flux measurement, the medium in the 96-well culture dish was replaced by 80 l of preincubation buffer (100 mM N-methyl-D-glucamine chloride, 5 mM HEPES, 5 mM KCl, 2 mM CaCl 2 , 0.8 mM MgSO 4 , 5 mM glucose, pH 7.4) with 2 M FlouZin-2 AM dye (Invitrogen) plus 0.2% (w/v) Pluronic F-127 (Invitrogen) and incubated for 48 min at room temperature. Cells were then washed three times with 80 l of preincubation buffer and incubated for 15 min with 80 l of preincubation buffer plus 0.1 mM ouabain to block Na ϩ /K ϩ -ATPase activity. This was done for three technical replicates for each construct. Afterward, the 96-well plate was placed into a fluorometer (Fluoroskan Accent, Thermo Scientific), and each well was injected with 40 l of 5ϫ Tl ϩ stimulation buffer (12 mM Tl 2 SO 4 , 100 mM N-methyl-D-glucamine chloride, 5 mM HEPES, 2 mM KCl, 2 mM CaCl 2 , 0.8 mM MgSO 4 , 5 mM glucose, pH 7.4). Fluorescence was measured in a kinetic-dependent manner (excitation, 485 nm; emission ,538 nm; one frame in 6 s in a 200-s time span) across the entire cell population in a single well. By using linear regression of the initial values of the slope of Tl ϩ -stimulated fluorescence increase, transport activity was calculated (55,56).
The effects of the kinase inhibitor staurosporine or NEM were determined by adding 8 M staurosporine or 1 mM NEM to the preincubation buffer 15 min prior to flux measurement. This was done for two technical replicates for each construct, and at least four independent measurements were performed 3 Please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party-hosted site.

Regulation of KCC2 by multiple phosphosites
for each construct. To specifically block transport activity of cation-chloride cotransporters, the loop diuretic furosemide (2 mM) was added to the preincubation buffer. Again, this was done for two technical replicates for each construct, and at least four independent measurements were performed for each construct.

Statistical analyses
Transport activity of each phosphomutant was tested against the control sample (WT), both under control treatment (untreated), using two-sample unequal-variances t test after Welch test (69). Because three technical replicates were measured from each independent preparation, we deflated the number of degrees of freedom according to the actual sample size (number of independent preparations) to avoid pseudoreplication. The resulting p values were corrected using the Benjamini-Hochberg method (70), which controls the falsediscovery rate in multiple comparisons.
For each phosphomutant, we compared the flux measured under four different treatment conditions (untreated, furosemide, staurosporine, and NEM) in a nested analysis of variance where we set the treatment condition as the fixed effect and the replicates as nested random effects. The mixed effects were modeled with the lme function from the R package nlme, and Tukey post hoc analysis was performed with the glht function from the R package multcomp. The p values from the post hoc comparisons were adjusted using Holm's sequential Bonferroni correction (71). Note that only p values Ͻ0.01 were considered to reduce the chances of false positives (type I errors).

Cell-surface biotinylation
KCC2 cell-surface levels were assessed by surface biotinylation as described previously (55,56). For this purpose, 90 -95% confluent 10-cm culture dishes of transfected HEK-293 cells were treated with the membrane-impermeant Sulfo-NHS-SS-Biotin (Thermo Fisher Scientific) according to the provided protocol. After several washes and cell lysis, biotinylated proteins were recovered by a NeutrAvidin-agarose column. After three rounds of washes, biotinylated proteins were eluted in sample buffer. Aliquots of the cell homogenate and of the eluate were collected and analyzed by immunoblot analysis.
To quantify the amount of KCCs expressed at the cell surface, dilution series of each sample were loaded onto a 10% SDS-polyacrylamide gel system. After separation and electrotransfer onto polyvinylidene difluoride membranes, membranes were incubated with N1-12 antibody (1:1,000). After incubation for 2 h at room temperature, membranes were washed four times with TBS-T (20 mM Tris, 150 mM NaCl, 1% Tween, pH 7.5), and the secondary antibody donkey antimouse IgG-horseradish peroxidase (Santa Cruz Biotechnology, Heidelberg, Germany) was applied for 2 h. After washing, bound antibodies were detected using an enhanced chemiluminescence assay (GE Healthcare) and a LAS-3000 documentation system (Fujifilm, Düsseldorf, Germany). Quantification of the bands was performed using MultiGauge software V3.1 (Fujifilm). The cell lysate corresponds to the total protein amount and was set to 100%. Only data with recovery values of 100 Ϯ 20% were included in our analysis. Four to six biological replicas were performed for each construct. Data are given as mean Ϯ S.D. Significant differences between the groups were analyzed by Student's t test.