Neuronal ClC-3 Splice Variants Differ in Subcellular Localizations, but Mediate Identical Transport Functions*

Background: Alternative splicing can result in proteins with distinct subcellular distributions and functions. Results: Three ClC-3 splice variants are expressed in the mammalian brain with different subcellular localizations, but identical transport properties. Conclusion: Differences in the subcellular localization of ClC-3 splice variants suggest diverse cellular functions. Significance: The existence of multiple splice variants needs to be considered when studying cellular functions of ClC-3. ClC-3 is a member of the CLC family of anion channels and transporters, for which multiple functional properties and subcellular localizations have been reported. Since alternative splicing often results in proteins with diverse properties, we investigated to what extent alternative splicing might influence subcellular targeting and function of ClC-3. We identified three alternatively spliced ClC-3 isoforms, ClC-3a, ClC-3b, and ClC-3c, in mouse brain, with ClC-3c being the predominant splice variant. Whereas ClC-3a and ClC-3b are present in late endosomes/lysosomes, ClC-3c is targeted to recycling endosomes via a novel N-terminal isoleucine-proline (IP) motif. Surface membrane insertion of a fraction of ClC-3c transporters permitted electrophysiological characterization of this splice variant through whole-cell patch clamping on transfected mammalian cells. In contrast, neutralization of the N-terminal dileucine-like motifs was required for functional analysis of ClC-3a and ClC-3b. Heterologous expression of ClC-3a or ClC-3b carrying mutations in N-terminal dileucine motifs as well as WTClC-3c in HEK293T cells resulted in outwardly rectifying Cl− currents with significant capacitive current components. We conclude that alternative splicing of Clcn3 results in proteins with different subcellular localizations, but leaves the transport function of the proteins unaffected.

ClC-3 belongs to the sub-branch of the CLC family of anion channels and transporters that resides primarily in intracellular organelles. Its functional relevance in the central nervous system is illustrated by Clcn3 Ϫ/Ϫ knock-out animal models (1-3) that exhibit pronounced hippocampal and retinal degeneration. Changes in synaptic transmission in these animals suggest that ClC-3 is present in synaptic vesicles and contributes to the regulation of neurotransmitter accumulation and release from the presynaptic nerve terminal (2,4,5).
However, besides experimental data that supports localization of ClC-3 in synaptic vesicles or lysosomes (2)(3)(4)(5)(6)(7)(8), there are also results that argue in favor of surface membrane localization of this protein (9,10). Moreover, multiple functional properties have been reported for ClC-3. Our group expressed mutant ClC-3 after removal of an N-terminal dileucine motif and observed outwardly rectifying anion-proton exchange current that resemble currents mediated by ClC-4 and ClC-5 (11)(12)(13)(14)(15). A characteristic property of ClC-3 was the occurrence of prominent capacitive currents, which indicate a large percentage of transporters mediating incomplete transport cycles (12,16). Other groups assigned a postsynaptic Ca/CaMK-regulated anion channel in hippocampal neurons to ClC-3 and hypothesized that ClC-3 might regulate neuronal excitability as anion channels by modifying the postsynaptic membrane potential and/or length constant (9,10,17).
A potential reason for such functional differences between native and heterologously expressed proteins might be the existence of alternatively spliced ClC-3 variants with distinct subcellular localizations and transport functions. So far, five splice variants of Clcn-3 have been identified; ClC-3a, ClC-3b, ClC-3c, ClC-3d, and ClC-3e, and partially characterized (18 -20). We decided to clone all ClC-3 splice variants from mouse brain and to compare their functions and subcellular distributions. We found three splice variants that differ in the N-terminal domain and exhibit identical transport function, but different subcellular distributions.

Experimental Procedures
Cloning and Expression Profile of ClC-3a, ClC-3b, and ClC-3c-To clone the complete coding regions of ClC-3a, ClC-3b, and ClC-3c, cDNAs were amplified from mouse brain using the SuperScript TM one step RT-PCR system with platinum Taq (Invitrogen, Carlsbad, CA). We used primers that were specific to the different 5Ј coding region together with a common reverse primer hybridizing to the 3Ј-end. After assembly of amplified bands into the pRSETB vector (Invitrogen) variants were identified by sequencing.
The tissue distribution of the different ClC-3 mRNAs was determined by RT-PCR. After isolation of total RNA from brain, heart, pancreas, kidney, liver, lung, retina, olfactory bulb, and spinal cord from 2-month-old mice and from hippocampi from 2, 13, 30, 60, or 120 days old mice RT-PCR was performed with the following primers: for ClC-3a and ClC-3b 5Ј-CGCC-CAGCTTGCTATGCCTCTGAG-3Ј (forward), ClC-3c 5Ј-ATG-GATGCTTCTTCTGATCC-3Ј (forward) and a common antisense primer 5Ј-AGCTAGTGCCCCTGATGCCAGTC-3Ј (reverse). Three PCR products with the predicted size of 324 bp/ClC-3a, 500 bp/ClC-3b, and 379 bp/ClC-3c were obtained. To identify ClC-3e (ClC-3d or ClC-3f), 5Ј-TGCCCTCAGAA-GAGACCTGACTATTGC-3Ј (forward) and 5Ј-AACGAACT-TCCTCTTCTGTCTCCTCTCTG-3Ј (reverse) primers were applied. These primers recognize sequences in the 3Ј-coding region of Clcn3 and generates RT-PCR products with expected sizes of 485 bp and 409 bp corresponding to the ClC-3 with the long and short C termini, respectively. PCR products were separated by gel electrophoresis and quantified using ImageJ 1.44p software (National Institutes of Health, Bethesda, MD) (21). To account for age-dependent changes in cell number or size these values were normalized to mRNA levels of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). GAPDH mRNA levels were determined using 5Ј-CAGTATGACTCCACTCACGGCAAATTC-3Ј as forward primer and 5Ј-CACAGTCTTCTGGGTGGCAGTGATG-3Ј as reverse primer, generating a PCR product with an expected size of 423 bp.
Heterologous Expression-cDNAs encoding full-length mouse ClC-3a, ClC-3b, or ClC-3c (GenBank TM Accession Number NM_007711.3, NM_173873.1, NM_173876.3) were fused in-frame to the 5Ј-end of the coding sequences of enhanced green or monomeric red fluorescence protein (eGFP or mRFP) and cloned into FsY1.1 G.W. or p156rrL vectors (were kindly provided by Dr. M. Filippov, Nizhny Novgorod, Russia, and Dr. D. Bruns, Homburg, Germany). For each construct, two independent recombinants from the same transformation were examined and shown to exhibit indistinguishable functional properties.
Electrophysiology-Standard whole-cell patch clamp recordings were performed using an EPC-10 amplifier, software controlled by PatchMaster (HEKA) (11). Borosilicate pipettes (Harvard Apparatus) were pulled with resistances of 0.9 -2 M⍀. We only recorded from cells with series resistances below 4.5 M⍀. More than 80% of the series resistance was routinely compensated, resulting in a voltage error of less than 5 mV. P/4 leak subtraction with a baseline potential of Ϫ30 mV was used to cancel linear capacitances (26). Currents were low-pass filtered at 2.9 kHz and digitalized with a sampling rate of 100 kHz. The standard external and internal recording solutions contained (in mM) 160 NaCl, 15 HEPES, 4 K-gluconate, 2 CaCl 2 , 1 MgCl 2 , pH 7.4 (bath solution), or 105 NaCl, 15 HEPES, 5 MgCl 2 , 5 EGTA; pH 7.4 (pipette solution).
Confocal Imaging-Images were acquired 24 -36 h after transfection with a Leica TCS SP5 II inverted microscope (Manheim, Germany) using a 63ϫ oil immersion objective from living cells in PBS containing Ca 2ϩ and Mg 2ϩ (GIBCO) at room temperature (22-24°C). EGPF and YFP (enhance green and yellow fluorescence proteins) fluorophores were excited with a 488-nm Argon laser and mRFP (monomeric red fluorescence protein) with a 594-nm He-Ne laser. Emission signals were detected after filtering with at 500 -550 nm, 520 -560 nm, or 600 -650 nm bandpass filters. To determine the fraction of ClC-3b mutants inserted into the plasma membrane (Fig. 4H) we co-expressed farnesylated eGFP as surface membrane marker together with ClC-3b S3/S2 ClC-3b S3/S1 orClC-3b S3/S2/S1 as mRFP fusion proteins. Surface membrane insertion was then quantified in confocal images as mRFP fluorescence intensity overlapping with eGFP fluorescence. For all mutants we used similar microscope settings in these experiments. Images were analyzed and assembled for publications in ImageJ 1.44p software (National Institutes of Health) (21).
Data Analysis-Data analysis was performed using a combination of FitMaster (HEKA), Origin (OriginLab), SigmaPlot (Systat Software), and Excel (Microsoft) software. All data are presented as mean Ϯ S.E.
Hippocampal degeneration in Clcn3 Ϫ/Ϫ mice starts about 2 weeks after birth (1)(2)(3). We reasoned that developmental changes in splice variant expression might contribute to this age dependence. Since there are no splice variant-specific antibodies available that distinguish between ClC-3a, ClC-3b, and ClC-3c, quantification of protein expression levels by Western blot analysis is not possible. We therefore examined mRNA profiles in hippocampal tissue from 2, 13, 30, 60, or 120 days old . We did not observe significant age-dependent changes in the mRNA levels for any ClC-3 splice variants relative to the amount of GAPDH mRNA (Fig. 2, A and B).
Comparison of mRNA levels demonstrated relatively low levels of ClC-3a mRNA and much stronger transcription of ClC-3b and ClC-3c mRNA at all tested ages. These data show that mRNA levels of ClC-3a, ClC-3b and ClC-3c remain unchanged at juvenile, early adult and adult ages and that ClC-3b and ClC-3c are the predominant ClC-3 splice variant in hippocampal neurons.

ClC-3 Splice Variants Exhibit Different Subcellular
Localizations-Differences in primary structure might result in altered transport functions and/or subcellular distribution of ClC-3 splice variants. We therefore studied biophysical properties and subcellular localization of ClC-3b and ClC-3c and compared them with the well characterized short isoform ClC-3a (16). Whole-cell recordings of HEK293T cells heterologously expressing WT ClC-3a or WT ClC-3b yielded ionic currents undistinguishable from non-transfected cells (Fig. 3A). In contrast, we were able to record ClC-3-specific currents from cells expressing WT ClC-3c. At positive potentials these cells display outwardly rectifying Cl Ϫ currents with amplitudes up to 1.5 nA at ϩ175 mV, whereas no measurable currents could be observed at negative potentials. Upon depolarizing voltage steps, there are large peaks at the beginning of the applied voltage steps that resemble the gating charge movements of ClC-5 (29) and ClC-3 13-19A (a ClC-3a mutant in which an N-terminal dileucine motif had been mutated (8, 16)) (Fig. 3A).
The differences in functional expression are due to separate subcellular targeting of the distinct splice variants (Fig. 3B). Upon expression of ClC-3a or of ClC-3b transfected cells exhibit large vesicular structures that co-localize with the lysosomal marker LAMP1 and therefore likely originate from lysosomal compartments. ClC-3c exhibited a different intracellular localization, which results in staining of the surface membrane and of intracellular vesicular compartments that do not contain LAMP1 (Fig. 3B). Complementary experiments revealed identical subcellular distribution of ClC-3 splice variants in MDCK cells as in HEK293T cells (data not shown).
The N Terminus of ClC-3b Contains Three Potential Dileucine Motifs-Alternative splicing in the N-terminal region might not only modify the subcellular distribution, but also the function of ClC-3, as reported for many other proteins (30 -33). We therefore searched for the signals that are responsible for the intracellular localization of ClC-3b and whose deletion might allow membrane surface insertion and electrophysiological characterization. For ClC-3a removal of a dileucine motif sequence (LLDLLDE (S1) Fig. 4A) allows surface membrane insertion and functional analysis of the protein (8,16,34). ClC-3b contains the same sequence motif, however, its removal did not result in surface membrane insertion (data not shown). We therefore screened the N-terminal region of ClC-3b for additional dileucine motifs (Fig. 4A). We found two such sequences, 42 EDDNLL 47 (S2) and 26 EELL 29 (S3), and generated mutant constructs in which either two of the three motifs (ClC-3b S3/S2 and ClC-3 S3/S1 ) or all dileucine motifs (ClC-3b S3/S2/S1 ) were substituted by alanine. Removal of only two dileucine motifs (ClC-3b S3/S2 and ClC-3 S3/S1 ) resulted in surface membrane localization of a fraction of the expressed proteins. However, there was still some fluorescence staining of intracellular compartments and large LAMP1-positive vesicular structures. ClC-3b S3/S2/S1 , in which all three dileucine motifs were removed, inserted predominantly into the surface membrane so that the large vesicular structures induced by ClC-3b S3/S2 and ClC-3 S3/S1 were absent in cell expressing this mutant protein (Fig. 4B).
To investigate interactions of the dileucine motifs with components of the endocytotic machinery using a pull-down strategy, we generated recombinant GST fusion proteins of N-terminal regions of ClC-3b wild type and ClC-3b S3/S2/S1 . After purification N-terminal fusion proteins were incubated with equal amount of mice brain lysate, and potential binding partners were analyzed by immunoblotting with antibodies to clathrin. Whereas GST-NT ClC-3b exhibits strong binding to clathrin (Fig. 4C), this interaction was markedly reduced for mutant GST-NT ClC-3b S3/S2/S1 (Fig. 4C). These results suggest that the removal of ClC-3b dileucine motifs results in reduced internalization of the mutant protein (8). Alternatively, these mutations might enhance ClC-3b insertion into the plasma membrane via impaired recognition of mutant sorting motifs by adaptor proteins in the trans-Golgi network or in endosomal compartments (35).
The altered localization of mutant ClC-3b permits the electrophysiological characterization of this splice variant. The existence of various ClC-3b mutants with different dileucine motifs also provides the possibility to test whether mutations within the internalization motifs change functional properties. Mutant ClC-3b proteins with or without one dileucine motif expressed at sufficient amounts in the surface membrane to account for measurable outwardly rectifying Cl Ϫ currents (Fig.  4, D and E). In all cases, we observed time and voltage-dependent currents that resemble ClC-3a S1 (16). Expression of ClC-3b S3/S2, ClC-3 S3/S1, or ClC-3b S3/S2/S1 resulted in voltage-dependent outwardly rectifying currents at potentials positive to ϩ35 mV, without inward currents at negative voltages (Fig. 4, D and  E). Depolarizing voltage steps elicited a capacitive current followed by ionic current that slightly increased with time.
Stepping back to the holding potential resulted in a capacitive current with identical amplitude as upon membrane depolarization. For CLC exchangers, a plot of the time integral of these capacitive currents, the "gating charge movement," versus the preceding voltage step provides the voltage dependence of activation (12,16,29,36). Such analysis did not reveal any marked differences between the three mutants (Fig. 4F). For ClC-3, ClC-4, and ClC-5, such capacitive currents have been shown to originate from transporters that only perform incomplete transport cycles (12,16), and the charge movement upon voltage steps thus provides a measure of transport-incompetent transporters. On the other hand, ionic currents are propor-tional to Cl Ϫ -H ϩ exchange rates. Plotting gating charges versus ionic currents at the same voltage provides a value proportional to the transport competence of the different constructs (Fig.  4G). We observed identical slopes for ClC-3b S3/S2 , ClC-3b S3/S1 , and ClC-3b S3/S2/S1 . The different macroscopic current amplitudes of cells expressing ClC-3b S3/S2 , ClC-3b S3/S1 , and ClC-3b S3/S2/S1 are likely due to separate protein densities in the surface membrane (Fig. 4, A and B), but could be also affected by variation in individual transport rates. To distinguish between these two explanations we co-expressed mutant ClC-3b fusion proteins with farnesylated eGFP as surface membrane marker and calculated surface insertion probabilities as ratio of the mRFP fluorescence intensity in regions overlapping with farnesylated eGFP by whole-cell fluorescence in confocal images. A plot of mean macroscopic current amplitudes from cells expressing ClC-3b S3/S2 , ClC-3b S3/S1 , or ClC-3b S3/S2/S1 against these values revealed a linear relationship (Fig. 4H), as expected for sole differences in trafficking and identical transport rates of the mutant transporters. We conclude that dileucine motifs in the N terminus exclusively affect trafficking, but not the transport activity of ClC-3b.

Neuronal ClC-3 Splice Variants
OCTOBER 23, 2015 • VOLUME 290 • NUMBER 43 Biophysical Properties of ClC-3 Splice Variants- Fig. 5 summarizes the electrophysiological analysis of the three variants, ClC-3a S1 , ClC-3b S3/S2/S1 , and ClC-3c. Each of the three ClC-3 proteins mediates outwardly rectifying currents (Fig. 5, A and  B) with identical properties. In all cases, we observed large capacitive currents upon depolarization and subsequent repolarization to the holding potential. We quantified the voltage dependence of ClC-3a S1 , ClC-3b S3/S2/S1 , and ClC-3c by measuring the area under the off-gating (Q off ) currents and plotting these "gating" charges versus the preceding voltage steps (12,16,37). This analysis revealed identical voltage dependences with a half-maximal activation voltage of ϳϩ65mV for all ClC-3s proteins (Fig. 5C). A plot of gating charge versus ionic current at the same voltage revealed identical transport competences for all ClC-3 splice variants expressed in the central nervous system (Fig. 5D). We conclude that alternative splicing leaves functional properties of ClC-3 unaffected.
with the limited overlap with LAMP1 or RAB7 (Figs. 3B and 6C), indicates localization of ClC-3c in the recycling endosome.
Among recycling endosomes two functionally distinct populations can be distinguished: endosomes that express RAB11 (38) and endosomes that contain the transferrin receptor TfR (40). To further study the localization of ClC-3c we co-expressed ClC-3c-eGFP with the transferrin receptor TfR. We observed substantial co-localization ClC-3c with TfR (Fig. 6C) indicating that ClC-3c localizes to both, RAB11-and TfR-positive compartments.
ClC-3c Targets to Recycling Endosomes via an Isoleucine-Proline (IP) Motif-ClC-3a, ClC-3b, and ClC-3c share dileucine motifs in the N terminus, and the distinct subcellular localization of ClC-3c must therefore be caused by additional targeting sequences. The ClC-3c N terminus contains a sequence motif ( 8 YLPY 11 ), which is reminiscent of a consensus binding motif YXX[FYL] for AP1, AP2, AP3, and AP4 mu subunits (41,42). This motif contains the PY residues that were suggested to result in the internalization of ClC-5 and barttin (43,44) (Fig.  7A). To determine whether 8 YLPY 11 is involved in ClC-3c targeting, we substituted all amino acids by alanine and evaluated whether removal of this motif redirects ClC-3c from recycling endosomes to late endosomes/lysosomes. Such a change in localization would be visible as co-localization of mutant ClC-3c with the late endosomal/lysosomal markers RAB7/ LAMP1 and characteristic enlargement of endosomal/lysosomal vesicles in cells expressing mutant ClC-3c. However, mutation of all amino acids in 8 YLPY 11 to alanine neither resulted in obvious changes in the subcellular distribution nor in the morphology of intracellular compartments (data not shown).
We next progressively deleted the N-terminal region of ClC-3c by removing stretches of 5, 6, or 8 amino acids (Fig. 7A). Neither deletion of the first five amino acids (ClC-3c ⌬1-5 , data not shown) nor of the following six amino acids (ClC-3c ⌬6 -11 ) (Fig. 7B) changed the localization of the protein or the morphology of intra-vesicular compartments. In contrast, the subsequent deletion of the amino acids stretch 12 DGGGDSIP 19 caused insertion of ClC-3c ⌬12-19 into lysosomes and enlargement of endosomal vesicles (Fig. 7C). We observed substantial co-localization of ClC-3c ⌬12-19 with LAMP1, but not with Rab11. Further deletion ClC-3c ⌬20 -25 did not alter the subcellular distribution (data not shown). Fusing DGGGDSIP directly to the N terminus of ClC-3a (Fig. 8A) resulted in localization of ClC-3a DGGGDSIP in the recycling endosomes (Fig. 8B). This result was confirmed by different co-localization pattern of RAB11/LAMP1 with ClC-3a or ClC-3a DGGGDSIP and by the absence of large vesicles formation in cells expressing ClC-3a DGGGDSIP (Fig. 8B). Taken together, our findings indicate that the amino acids stretch 12 DGGGDSIP 19 contains a potential sorting motif to the recycling endosome.
To delineate the minimum sequence necessary for the specific sorting of ClC-3c, we mutated groups of two amino acids within this stretch jointly to alanine. Substitution of Asp 12 and Asp 16 to alanine (ClC-3c D12/A D16/A ) left targeting of ClC-3c unaltered (Fig. 8, A and C). In contrast, alanine insertion at 18 I and 19 P (ClC-3c IP/AA ) was sufficient to target mutant ClC-3c to late endosomes/lysosomes (Fig. 8, A and D), resulting in prom-

Neuronal ClC-3 Splice Variants
OCTOBER 23, 2015 • VOLUME 290 • NUMBER 43 inent vesicular enlargement of LAMP1 positive compartments in cells expressing mutant ClC-3c. We conclude that an N-terminal isoleucine-proline (IP) motif is responsible for targeting of ClC-3c to the recycling endosomes.

Discussion
Alternative splicing permits translation of diverse proteins from a single gene by including or excluding certain exons from the processed messenger RNA. We here studied alternative splicing of Clcn3 and the consequences of this process on protein function and subcellular distribution. The exon-intron arrangement of Clcn3 suggests translation of six alternatively spliced gene products, referred to as ClC-3a to ClC-3f. We amplified ClC-3 splice variant from different mouse tissues by RT-PCR (Fig. 1A) and demonstrated that only three splice variants are expressed in the brain, the olfactory bulb and the spinal cord, ClC-3a, ClC-3b, and ClC-3c, with ClC-3b and ClC-3c being the predominant ClC-3 splice variants ( Fig. 1C and Fig. 2).

Upon heterologous expression in mammalian cells ClC-3a
and ClC-3b exclusively localize to the late endosomal/lysosomal system, whereas ClC-3c can be found in recycling endosomes and also in the surface plasma membrane. ClC-3b is targeted to the late endosomal/lysosomal system via multiple dileucine retention signals (Fig. 4, A and B), similar to the signals that control localization of ClC-3a (8,16). For ClC-3c we identified an isoleucine-proline (IP) signal that is responsible for recycling endosome localization. Removal of this signal hinders targeting to recycling endosomes and surface membrane expression of ClC-3c (Fig. 8). Moreover, insertion of the isoleucine-proline (IP) signals reroutes ClC-3a from the late endosomal/lysosomal system to the recycling endosomes (Fig. 8).
We studied localization of ClC-3 splice variants exclusively in cultured mammalian cells of epithelial origin and not in cultured neurons or even native neuronal tissue. Cultured cells are well established for studying trafficking and function of membrane transport proteins, and a large body of evidence supports the notion that similar motifs might direct targeting in epithelia  Recently, the ClC-3 splice variant ClC-3d was cloned from mouse liver and functionally analyzed by heterologous expression in HEK293T cells (20). The authors demonstrated that ClC-3d differed from ClC-3a and ClC-3b in surface membrane expression, but exhibit similar transport properties. These results demonstrate that alternative splicing within the C terminus also affects only trafficking and not function of ClC-3.
All three ClC-3 splice variants in the mammalian central nervous systems exhibit closely similar transport properties. We recently performed a detailed electrophysiological analysis of ClC-3a and demonstrated that this splice variant functions as Cl Ϫ -H ϩ exchanger with low transport efficiency (16). ClC-3a, ClC-3b, and ClC-3c exhibit identical ratios of the moved charges by the transport current (providing values proportional to the number of complete transport cycles (Fig. 5D)) and identical voltage dependences of these capacitive currents (Fig. 5C). The importance of these specific functional features of ClC-3 is not clear (16). The extreme outward rectification results in maximum transport rates at voltages far away from physiological values. The large percentage of incomplete transport cycles result in transport effectivities that are much lower than those of ClC-4 and ClC-5 (16). To account for the multiple pro-nounced effects of ClC-3 ablation we recently proposed that the main function of ClC-3 might be enlarging the capacitance of their resident compartments (16). Such a function nicely accounts for the effects of ClC-3 ablation for synaptic function, but makes it difficult to assign a cellular role for ClC-3 splice variants in early or late endosomes/lysosomes.
Because of its predominant intracellular localization, the functional characterization of ClC-3 has been difficult and multiple transport functions have been assigned to ClC-3 since its identification. Initially, a large conductance, slightly outwardly rectifying anion channel, which was blocked by intracellular calcium, was assigned to ClC-3 (47,48). Later, ClC-3 was postulated to represent a volume-activated anion channel (49 -52). Another ClC-3 candidate channel is a Ca 2ϩ /calmodulin-dependent chloride channel at postsynaptic localizations (10,17). Work with Clcn3 Ϫ/Ϫ mice (2) and our functional data on all existing ClC-3 splice variants strongly suggests that these anion channels are not identical with ClC-3 and demonstrate that neuronal ClC-3 splice variants rather function as Cl Ϫ -H ϩ exchangers with strong voltage dependence and low transport efficiency.
Whereas ClC-3a and ClC-3b can only be found in intracellular compartments, ClC-3c is part of the recycling endosome with a considerable percentage of transporters present in the surface membrane. ClC-3c co-localizes with endosomes that express RAB11 as well as with endosomes that contain the transferrin receptor TfR (31). RAB11 is present in mature synaptic vesicles of the mammalian brain, and it has been speculated that it might contribute in determining the secretory fate of a transport vesicle (58). Upon expression in cultured neurons, RAB11 localizes to synaptic boutons and moderately copurifies with synaptic vesicle markers (59). So far, we have not determined the localization of the different splice variants in neurons, but these data suggest that ClC-3c might account for altered synaptic transmission in Clcn3 Ϫ/Ϫ (2, 4, 5). Alternative ClC-3aDGGGDSIP merge merge FIGURE 8. ClC-3c is targeted to recycling endosomes via an IP motif. A, schematic representation of the approach used to dissect the sorting signal of ClC-3c. Amino acid substitutions and insertions are highlighted in red. B, C, D, confocal images of HEK293T cells co-expressing ClC-3a DGGGDSIP (B), ClC-3c D12/A D16/A (C) or ClC-3c IP/AA (D) with RAB11 or LAMP1. The scale bar represents 10 m. Insets show changes in cell morphology upon expression of ClC-3c IP/AA , but neither with ClC-3a DGGGDSIP nor with ClC-3c D12/A D16/A . splicing of ClC-3 permits targeting intracellular CLC transporters to multiple distinct cellular compartments. ClC-3 is known to hetero-multimerize with ClC-4 and ClC-5 (60), and alternative splicing of ClC-3 will thus also affect subcellular localization of ClC-3-ClC-4 oligomers in the central nervous system. Moreover, hetero-dimers between different splice variants are likely to assemble. At present, it is not clear into which compartment these different hetero-oligomers will insert.
In summary, we demonstrate that alternative splicing leads to the occurrence of three ClC-3 splice variant with differences in the N terminus in the mammalian system. All three variants exhibit identical transport properties, but distinct localization in late endosomes/lysosomes or recycling endosomes. Alternative splicing enables ClC-3 to fulfill diverse cellular functions, and our work provides an important step toward understanding the role of ClC-3 in diverse cellular compartments.