Identification of Key Functional Domains in the C Terminus of the K+-Cl– Cotransporters*

The K+-Cl– cotransporter (KCC) isoforms constitute a functionally heterogeneous group of ion carriers. Emerging evidence suggests that the C terminus (Ct) of these proteins is important in conveying isoform-specific traits and that it may harbor interacting sites for 4β-phorbol 12-myristate 13-acetate (PMA)-induced effectors. In this study, we have generated KCC2-KCC4 chimeras to identify key functional domains in the Ct of these carriers and single point mutations to determine whether canonical protein kinase C sites underlie KCC2-specific behaviors. Functional characterization of wild-type (wt) and mutant carriers in Xenopus laevis oocytes showed for the first time that the KCCs do not exhibit similar sensitivities to changes in osmolality and that this distinguishing feature as well as differences in transport activity under both hypotonic and isotonic conditions are in part determined by the residue composition of the distal Ct. At the same time, several mutations in this domain and in the proximal Ct of the KCCs were found to generate allosteric-like effects, suggesting that the regions analyzed are important in defining conformational ensembles and that isoform-specific structural configurations could thus account for variant functional traits as well. Characterization of the other mutants in this work showed that KCC2 is not inhibited by PMA through phosphorylation of its canonical protein kinase C sites. Intriguingly, however, the substitutions N728S and S940A were seen to alter the PMA effect paradoxically, suggesting again that allosteric changes in the Ct are important determinants of transport activity and, furthermore, that the structural configuration of this domain can convey specific functional traits by defining the accessibility of cotransporter sites to regulatory intermediates such as PMA-induced effectors.

In their active state, the KCCs usually generate net outward movement of K ϩ -Cl Ϫ ions because of large intra-to-extracellular K ϩ gradients across the surface of most animal cells and because transport stoichiometries are probably 1K ϩ :1Cl Ϫ (13,14). For this reason, and given that K ϩ -Cl Ϫ cotransport is enhanced under conditions that lead to cell swelling, the KCCs are believed to play a key role in regulatory volume decrease responses (15)(16)(17). However, K ϩ -Cl Ϫ cotransport is not completely abolished under isotonic conditions (18 -20) 4 and is sensitive to changes in intracellular ionic strength (21)(22)(23), suggesting that the KCCs are also of central importance in intracellular Cl Ϫ (Cl i Ϫ ) regulation (22,23) and can thus affect a variety of physiological processes (24 -26). To this effect, it has been claimed that KCC4-mediated Cl Ϫ efflux in renal intercalated cells promotes secondary HCO 3 Ϫ basolateral efflux via an anion exchanger (24) and KCC2-mediated Cl Ϫ efflux in neuronal cells, secondary GABA responses (25,26). The importance of the KCCs in these processes is suggested more specifically by the behavior of a KCC4 Ϫ/Ϫ mouse, which was found to exhibit renal tubular acidosis (24), and a KCC2 Ϫ/Ϫ mouse, which was found to exhibit generalized epilepsy (26). The mechanisms that lead to KCC deactivation or activation during a change in cell volume and Cl i Ϫ are still poorly understood despite a large number of studies on the subject of K ϩ -Cl Ϫ cotransport regulation. Based on the effect of various pharmacological agents, a popular model that has emerged over the years is one in which K ϩ -Cl Ϫ cotransport is reduced through carrier phosphorylation and induced through carrier dephosphorylation, and in which such modifications result from the concerted action of protein kinases and protein phosphatases at relevant regulatory sites (14 -17, 27-36). It should be mentioned, however, that most of these studies have addressed the issue of cell volume-dependent, rather than Cl i Ϫ -dependent regulation.
Although different types of protein phosphatase and kinases that may play a role in KCC regulation have been implicated and a few possibly identified (14 -17, 27-38), the specific isoenzymes involved have remained elusive. Recently, we have found that KCC2-or KCC4-mediated activity is inhibited by 4␤-phorbol 12-myristate 13-acetate (PMA), suggesting that PKC is one of the candidate enzymes. 4 Carrier loci with which such enzymes interact have also remained unidentified, but a few studies suggest a central role for the C terminus (Ct) in this regard. For example, Strange et al. (39) have found that a Y1087D mutation in KCC2, a neuronal-specific isoform that displays robust constitutive activity under isotonic conditions, leads to a large decrease in KCC activity. Along the same line, Mercado et al. (40) have concluded from functional characterizations of KCC2-KCC4 chimeras that constitutive activity for KCC2 is conveyed through a unique 15-residue stretch (amino acids 1021-1035) in the distal Ct. For this reason, they termed the stretch in question the "ISO segment." In this work, the role of the Ct was explored further by analyzing distinctive sets of KCC2-KCC4 chimeras as well as selected mutants in which putative PKC sites were altered. Our studies led to the identification of protein segments that play unique functional roles but not that of the underlying the PMA effect through canonical PKC phosphorylation sites. Our studies have also revealed that the Ct may impart isoformspecific functional traits by adopting defined conformational ensembles and that, for KCC2, such ensembles are probably more important than the ISO segment per se in determining transport activity under isotonic conditions.
KCC Constructs-A total of sixteen different cDNAs was used. They were subcloned in the vector pGEM-HE or Pol1, which are designed to generate cRNA off of cDNA inserts and to increase the stability and translatability of transcription products in Xenopus laevis oocytes. Both these vectors contain a T7 promoter, a cloning site flanked by the X. laevis ␤-globin-untranslated regions, a poly(A) tract, and an NheI linearizing site. In all experiments, cDNA amplification and mutagenesis were carried out using the pGEM-HE-or Pol1-based plasmids.
c-myc-tagged Wild-type (wt) Rat (rt) KCC2 (KCC2 wt ) and wt Mouse (ms) KCC4 (KCC4 wt )-These constructs, provided to us by Dr. Eric Delpire and coworkers (Vanderbilt University, Nashville, TN), were generated from wt rtKCC2/pBF and wt msKCC4/pGEM-HE. The insert of the first construct consists of a cDNA that contains 3351 bp of open reading frame and more than 373 bp of untranslated regions, whereas that of the other construct consists of a cDNA that contains 3252 bp of open reading frame and Ͼ250 bp of untranslated regions. Before adding the c-myc tag to wt rtKCC2, the insert was moved from pBF to Pol1 as an XbaI-HindIII fragment.
The tag per se (MEQKLISEEDL) was introduced in front of the coding sequences of wt rtKCC2/Pol1 and wt msKCC4/pGEM-HE. This task was achieved by 1) cutting the wt constructs at restriction sites that are very close to the first ATG (XbaI-NarI for wt rtKCC2/Pol1 and XmaI-DrdI for wt msKCC4/pGEM-HE) and 2) ligating the larger fragments with prehybridized complementary oligonucleotides (Table 1) designed to encode the

TABLE 1 Oligonucleotides
Synthetic oligonucleotides were used for sequencing or to generate c-myc epitope tags and to introduce single point substitutions in KCC2 wt (A) or KCC4 wt (B). They are all written 5Ј to 3Ј. Capital letters designate codons that were substituted or inserted. Note that for KCC4 wt , the oligonucleotides of "c-myc tag step 1" were used to create the tag per se, and those of "c-myc tag step 2", to generate an in-frame tag-KCC4 coding sequence. Also note that several of the oligonucleotides that were used to generate substitutions served as sequencing primers as well to verify whether substitutions at other sites were correctly introduced.  4-2-4 , and KCC4 4-4-2 -These mutants were engineered by fragment exchange after creating silent restriction sites in the Ct of KCC2 wt and KCC4 wt using the QuikChange mutagenesis kit and pairs of oligonucleotides ( Table 1). The three sites used to generate junction points were Bst1107I, AvrII, and EcorV at bp 2046 -2051, 2401-2406, and 2804 -2809 in the KCC2 wt open reading frame, and 2063-2068, 2418 -2423, and 2821-2826 in the KCC4 wt open reading frame. In KCC2 wt , they correspond to residues Gly 644 /Ile 645 , Leu 763 /Gly 764 , and Asp 897 / Ile 898 , and in KCC4 wt , to residues Gly 664 /Ile 665 , Leu 783 /Gly 784 , and Asp 917 /Ile 918 (see arrows in Fig. 1). In this work, the chimeras are termed KCC2 x-x-x or KCC4 x-x-x , where the first, second, and third "x" in subscripted font correspond, respectively, to the proximal, middle, and distal third of the Ct, and where "x" is assigned the number 2 or 4 to designate the isoform from which each third is derived.

Oligonucleotide sequences
Single Point Substitutions: KCC2 (T34A) , KCC2 (S728N) , KCC2 (T787A) , KCC2 (S940A) , KCC2 (S1034A) , KCC2 (T34A-S728N-T787A-S940A-S1034A) or KCC2 (0PKC) , KCC2 4-2-2(N728S) , and KCC2 4-2-2(N728D) -The first six mutants consist of wt KCC2s in which putative PKC phosphorylation sites that are conserved among species and occur in cytosolic domains were altered either one by one or in combination by generating (S/T) 3 (A/N) mutations. In KCC2 wt and KCC4 wt , one of these sites is in the predicted N terminus (Nt), whereas the other sites are in the predicted Ct ( Fig. 1). Other putative PKC sites are present in KCC2 wt and KCC4 wt , but they are not conserved among species or they are predicted to occur in the central domain. The other mutants consist of KCC2 4-2-2 chimeras in which additional changes were made in the proximal Ct by creating a PKC site or a Asn 3 Asp substitution.
Except for KCC2 0PKC , all of the substitutions were generated from KCC2 wt or KCC2 4-2-2 using pairs of mutagenic oligonucleotides (Table 1B) and the QuikChange mutagenesis kit. They were called KCC2 (y) s, where "y" in parenthesis is used to designate the substitution that was generated based on the amino acid number in the template carrier. As for KCC2 0PKC , it was generated in four steps: 1) by recreating the S1034A substitution as above but from the template KCC2 (S940A) , 2) by recreating the T787A substitution also as above but from KCC2 (S940A-S1034A) , 3) by ligating the XbaI-HincII fragment of KCC2 T34A to that of KCC2 (S728N) , and 4) by ligating the SfiI-HindIII fragment of KCC2 (T34A-S728N) to that of KCC2 (T787A-S940A-S1034A) .
Expression of KCCs-Defolliculated stage V-VI oocytes were generally each injected with ϳ10 ng of KCC2 wt -, KCC4 wt -, or mutant KCCderived cRNAs, and maintained for ϳ3 days at 18°C in Barth's (B) medium (Table 2) plus 125 M furosemide. In each of the conditions tested, groups of oocytes were also injected with H 2 O alone to determine the component of fluxes that is due to endogenous KCC-mediated cotransport.
Functional Studies-As in previous studies by us and other groups (4,21,39,40), KCC activity was assessed through influx assays. The isotope 86 Rb ϩ was used as tracer, and all experiments were carried out at ϳ22°C with several types of media (Table 2) adjusted to a pH of 7.4. One of these mediums, called low Cl Ϫ /hypotonic (LH), is hypoosmolar relative to the X. laevis plasma, whereas three of these media, called low Cl Ϫ (or L), normal Cl Ϫ (R) and wash (W), are approximately isoosmolar. In one experiment, two other groups of media, termed low Cl Ϫ /osmolality variable (LO ⌬ ) and near normal Cl Ϫ /osmolality variable (O ⌬ ), were used to determine the sensitivity of certain carriers to hypoosmotic swelling at 5 mM Cl Ϫ (KCC2 wt , KCC4 wt , KCC2 2-2-4 , and KCC2 4-2-2 ) and at 55 mM Cl Ϫ (KCC2 wt and KCC4 wt ). For all of these media, sucrose was used to adjust osmolality, whereas gluconate and/or SO 4 2Ϫ were used as replacement of Cl Ϫ .
Before the flux assay, furosemide was first removed through rinses in large volumes of medium B. Next, oocytes were subjected to two consecutive incubation steps, one of variable duration in medium R, L, LH, O ⌬ , or LO ⌬ (with or without 250 M furosemide, with or without 0.1-1.0 M PMA or 4␣-PMA, with or without 0.1 M OA) and another of 45 min in medium R (plus 1-2 Ci/ml 86 Rb ϩ , plus 10 M ouabain, with or without 5 M bumetanide, with or without 250 M furosemide). It should be noted that, in most studies, bumetanide was added at low concentrations during the second incubation to inhibit Na ϩ -K ϩ -Cl Ϫ cotransport while leaving K ϩ -Cl Ϫ cotransport intact (4,21). Fluxes were ended with several rinses in medium W (plus 10 M ouabain, plus 250 M furosemide, plus 250 M bumetanide), and oocytes were transferred to 96-well plates prefilled with 2% SDS and scintillation fluid. After a FIGURE 1. Models of rtKCC2, msKCC4, and 6 KCC2-KCC4 chimeras, showing the localization of putative PKC sites in the wt carriers and the chimeras. Carriers are represented schematically by horizontal bars that are aligned to scale on an "x" axis. Black lines correspond to gaps in the amino acid sequence, red or blue rectangles are used to designate groups of residues that belong to rtKCC2 or msKCC4, respectively, and yellow corresponds to residues that belong to putative PKC sites ((S/ T)X(K/ R)) identified as being conserved among several species for a given KCC. Note that the positions of these sites within the horizontal bars are also to scale. In the central domain, wide rectangles represent transmembrane segments, and narrow rectangles represent connecting segments. In the section including the Ct domains, arrows indicate the positions of the junction points that were used to generate the chimeras, and the asterisk indicates the site that was subjected to Asn 3 Ser or Asn 3 Asp substitutions. The nomenclature used to identify the chimeras is explained under "Experimental Procedures"; based on this nomenclature, KCC2 2-2-2 ϭ KCC2 wt and KCC4 4-4-4 ϭ KCC4 wt . Nt, N terminus; Ct, C terminus.
brief stabilization period, 86 Rb ϩ was detected with the TopCountNXT counter (Packard Instrument Co.).
For each condition tested, counts among 3-12 oocytes (typically ϳ8 oocytes) were averaged and transposed into flux rates (FRs) based on the equation: FRs ϭ (counts/oocyte ϫ [ 85 Rb ϩ ] in flux medium) Ϭ ([counts] ϫ incubation time in flux medium), assuming that membrane surface areas among stage V-VI oocytes are the same. In all experiments, FRs measured in KCC-expressing oocytes were also converted into cotransporter-specific FRs (FRs KCC ), where FRs KCC ϭ (FRs without furosemide for KCC-expressing oocytes Ϫ FRs with furosemide for KCC-expressing oocytes) Ϫ (FRs without furosemide for H 2 O-injected oocytes Ϫ FRs with furosemide for H 2 O-injected oocytes), or they were converted into background-subtracted FRs, where FRs ϭ FRs without furosemide for KCC-expressing oocytes Ϫ FRs without furosemide for H 2 O-injected oocytes. Note that when calculated FRs were from experiments in which the effect of phorbol esters or a change in osmolality was assessed, they were converted a third time by normalizing them to those measured with the lowest concentration of these agents or lowest osmolality.
From the above calculations, absolute or normalized FRs among several replicate experiments (n ϭ 2-9) were re-averaged and expressed as mean FRs, relative units, or % changes Ϯ S.E. Normalized FRs were also used to calculate K i (PMA) values based on KCC2 wt activity versus [PMA] or [4␣-PMA] relationships. These constants were generated with the program PLOT (Biff Forbush) using a three-parameter Hill equation that includes the variables V max , K i , and the Hill number. They are expressed in this work as averaged fits Ϯ S.E. among six replicate experiments.
Immunolocalization of KCCs Expressed in X. laevis Oocytes-These studies were performed as described in previous publications by our group (41)(42)(43)(44). Briefly, oocytes cryosections (10 m) were postfixed for 30 min in paraformaldehyde and incubated sequentially with a primary and secondary antibody (for 1 h each at room temperature). Micrographs were taken with a TE-2000 Nikon Eclipse epifluorescence microscope.
Western Analyses of KCC2 wt , KCC4 wt , KCC2 2-2-4 , and KCC4 4-4-2 Purified from the Cell Surface by Biotinylation-The aim of these experiments was to provide evidence that immunofluorescence studies in oocytes are reliable in assessing levels of cell-surface protein expression and to determine the effect of a change in cell volume on protein distribution. Intact oocytes were first subjected to a flux assay as described above using medium LH, L, or R for preincubation, medium R (plus 10 M ouabain, with or without 5 M bumetanide, plus 1 mM Sulfo-NHS Biotin) for the second incubation during which KCC activity is generally measured during the assay, and medium W for wash. Subsequent to these procedures, cells were transferred to a lysis solution called medium X (20 mM Tris, pH 8.0, 1 mM EDTA, 4 mM MgCl 2 , 80 mM sucrose, 10% glycerol, 1% Triton-X, 1 mM phenylmethylsulfonyl fluoride) and homogenized mechanically through pipette tips. Oocyte extracts were then incubated in this medium for another 30 min at 4°C after which they were cleared by centrifugation and brought to 1 ml in medium X. Cell-surface biotinylated proteins were extracted from this medium after adding 50 l of agarose-coupled streptavidin and pelletting beads by centrifugation 2 h later, and they were released from the pellets in heated protein sample buffer. For the Western analyses, biotinylated proteins were migrated on 7.5% SDS-polyacrylamide Tricine gels, transferred onto Immobilon-P membrane blots (Millipore), and revealed by chemiluminescence using the Amersham Biosciences ECL solutions after sequential incubations of the blots with a primary and secondary antibody.
Software and Statistical Analyses-DNA characterizations were performed by automated sequencing, using plasmid-or KCC-derived primers (Table 1), and by restriction analyses. For BLAST searches, sequence alignments, and structure predictions, we used a combination of programs, including PLOT (B. Forbush) and DNAStar (Lasergene). When appropriate, differences between groups of variables were analyzed by Student two-tail t-tests, rejecting the null hypothesis for p Ͼ 0.05.

RESULTS
Preamble-Cell swelling was induced in the current study with a 5 mM Cl Ϫ , 125 mosM solution (medium LH), and its effect on KCC activation was determined by comparing flux rates (FRs) measured after this maneuver to those measured with a 5 mM Cl Ϫ , 200 mosM solution (medium L). The reason for reducing external Cl Ϫ (Cl o Ϫ ) to low levels was to minimize differences in Cl i Ϫ between cells incubated in the hypotonic versus control solution and, accordingly, Cl i Ϫ -dependent differences in KCC activity, as could occur based on the positive correlation that exists between the latter two variables (21)(22)(23). Given, however, that medium LH could still lead to slightly lower Cl i Ϫ than medium L by diluting cytosolic ions through cell swelling (21,45,46), additional studies were performed to verify whether changes in ion transport, if they arose, resulted from changes in cell volume or Cl i Ϫ , assuming that medium L acts by decreasing anion concentration. We suspect that this is the case based on previous work (47) 5   incubations in a 86 mM Cl Ϫ , 200 mosM solution (medium R) and compared with measurements obtained after incubations in medium L.

Synthesis of wt and Mutant KCCs in X. laevis
Oocytes-Carrier expression levels and distributions were analyzed through immunofluorescence studies using oocytes that had been maintained in isotonic media before membrane permeabilization and through Western analyses of cell-surface-biotinylated KCCs using oocytes that were subjected to different preincubations (in medium LH, L, or R). Immunofluorescence data are depicted in Fig. 2A using representative fields among Ն2 oocytes for each of the KCCs expressed (panels 1-16) and Western analyses in Fig. 2B through duplicate experiments using four different KCCs (panels 1 and 2).
In the upper group of images ( Fig. 2A), it is seen that the c-Myc antibody labels all of the heterologous proteins tested strongly (panels [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] and that it does so specifically, as revealed by the absence of signal in the H 2 O controls (panels 17 and 18). It appears, moreover, that the signal is localized primarily at the cell surface and is of similar intensity among the KCC-expressing oocytes. In the lower group of images (Fig. 2B), it is seen there that differences in expression at the cell surface among the KCCs tested (KCC2 wt , KCC4 wt , KCC2 2-2-4 , and KCC4 4-4-2 ) are indeed not important. For KCC2 wt and KCC4 4-4-2 , the bands are in fact of slightly fainter density but as will be shown below, transport rates by these carriers are not lower than those by KCC4 wt and KCC2 2-2-4 . The Western analyses presented in Fig. 2B Fig. 3 using panel A for condition "LO ⌬ " and B for "O ⌬ ". As can be seen in both these panels, interestingly, KCC2 wt is more sensitive to a decrease in osmolality than is KCC4 wt ; at 150 mosM, e.g. KCC2 wt is nearly fully stimulated while KCC4 wt is less than half fully stimulated. Compared with KCC4 wt , conversely, KCC2 wt is constitutively more active at higher osmolalities when Cl o Ϫ is 55 mM (panel B) instead of 5

mM (panel A).
These results suggest that Cl i Ϫ is an important determinant of constitutive activity for KCC2 wt under isotonic conditions. Functional Characterizations of KCC2-KCC4 Chimeras-These mutants were used to determine whether certain regions within the Ct, such as those that were described just above and those identified in previous studies, confer isoform-specific functional behaviors (18,31,40). Preincubations were carried out in three types of solutions, media LH (5 mM Cl Ϫ , 125 mosM), L (5 mM Cl Ϫ , 200 mosM), or R (86 mM Cl Ϫ , 200 mosM), and results are expressed (in Fig. 4) as background-subtracted furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs.
With medium LH (Fig. 4, panels A and B), both the wt KCCs (KCC2 2-2-2 and KCC4 4-4-4 in Fig. 4) are found to exhibit above background transport activity with FRs that are significantly different statistically from "0" but much higher for KCC2 wt . As for the reciprocal chimeras, differences in FRs are also observed, but the direction and amplitude of changes among each group are quite variable compared with the wt carriers. In panel A, e.g. the activity of a KCC2 that has the proximal third of its Ct replaced by that of KCC4 wt (KCC2 4-2-2 ) is much higher compared with KCC2 wt , whereas the activity of a KCC2 that has the middle third or distal third of its Ct replaced (KCC2 2-4-2 and KCC2 2-2-4 ) is much lower. In panel B, FRs among the carriers tend to be less variable except for KCC4 4-4-2 , which is much more active that the others and more active than KCC2 2-2-4 as well. For unknown reasons, the behavior of KCC4 2-4-4 in this group is not reciprocal to that of KCC2 4-2-2 .
With medium L (Fig. 4, panels C and D), the pattern of changes in transport activity among the carriers is almost completely identical to that observed with medium LH, suggesting that the mutations have not acted solely by altering transport activity to changes in osmolality. It can be noted here that FRs are much lower than those measured with medium LH; for three of these carriers, in fact, transport activity is not different from "0" statistically. These results suggest that some of the mutations may have generated inactive carriers or that differences in FRs between cRNA-injected and H 2 O-injected oocytes were too small to be detected in this series of assays. We favor the second hypothesis, because KCC4 wt has been shown on several occasions by our group to exhibit definite transport activity (generally between 0.1 to 0.2 nM/oocyte/h) under the "L" condition, 4 and because all of the chimeras tested in this study were found to be active under at least one condition.
In contrast to medium L, medium R (Fig. 4, E and F) is found to induce above background furosemide-sensitive, bumetanide-insensitive FRs for all of the carriers tested except KCC4 4-2-4 . Here again, and even if FRs measured under this condition tend to be intermediate between those measured with medium LH and those measured with medium L, the pattern of changes among the KCCs is very similar to that observed with medium LH. An obvious exception in these regards is KCC2 2-4-2 , which is seen to display a KCC2 wt -like phenotype in panel E, when it was clearly less active than the native carrier in panels A or C.
The behavior of the chimeras suggests thus far that the protein segments analyzed through various substitutions accomplish very different roles. It is noticeable that KCC2 2-2-4 and KCC4 4-4-2 are not only functionally reciprocal under all of the conditions tested but that they also adopt the functional properties of the wt carrier from which the distal Ct  Table 2. originates. The chimera KCC2 2-4-2 , on the other hand, exhibits a hybrid behavior between the wt carriers, and the chimera KCC2 4-2-2 , a rogue behavior with FRs that are outside the range of those measured for the wt carriers. The implication of these various findings will be discussed below.
Additional Studies for KCC2 4-2-2 and KCC2 2-2-4 -Such studies were carried out to determine the phenotype of certain chimeras in greater detail and whether the loss of a putative PKC site in the proximal Ct of KCC2 4-2-2 , such a site is present in KCC2 wt but not in KCC4 wt , accounts for heightened K ϩ -Cl Ϫ cotransport by this chimera. In one of the mutants, an Asn 3 Ser substitution was created in the proximal Ct to reconstitute the PKC site, and in the other, an Asn 3 Asp substitution to mimic phosphorylation. Results are summarized in Fig. 5.
In Fig. 5A, the chimera KCC2 2-2-4 is seen to behave exactly as KCC4 wt even if it is almost entirely composed of the KCC2 wt sequence, suggest-ing that the distal Ct conveys increased sensitivity to changes in osmolality. In panel B, on the other hand, KCC2 4-2-2 is seen to display once more peculiar characteristics in that its sensitivity to changes in external osmolality is much higher than that of KCC2 wt . Among various possibilities, these results suggest that mutations in the proximal Ct have induced allosteric effects or that they have led to the loss or gain of key regulatory sites.
The latter hypotheses were tested more specifically by analyzing the functional features of the two additional mutants KCC2 4-2-2(N728S) and KCC2 4-2-2(N728D) (Fig. 5C). They were also tested by analyzing the effect of PMA and OA on KCC2 wt and KCC2 4-2-2 (shown in panel D as PMAor OA-induced % changes in furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs), given that these agents were shown by us and others (see below and Refs. 14 -17 and 27-36) to decrease ion transport by certain isoforms. Note that in certain studies oocytes were preincubated  Table 2. for only 15 min in medium LH, L, or R to avoid PMA-or OA-induced toxicity; as will be explained in Fig. 6, fortunately, KCC activity after this incubation time is still higher with medium LH compared with medium L.
Looking at Fig. 5C, one can observe that the Asn 3 Ser substitution in KCC2 4-2-2 results in decreased K ϩ -Cl Ϫ cotransport under all conditions used, and that the Asn 3 Asp substitution almost completely abolishes this process. Although such results could indicate that Ser 728 in KCC2 wt is in fact a phosphoregulatory site, one can also observe by looking at panel D that PMA decreases KCC2 4-2-2 -mediated activity much more than KCC2 wt -mediated activity (25-30% inhibition versus 65-70%) and that the effect OA is quantitatively similar between KCC2 4-2-2 and KCC2 wt . Taken together, hence, the results of these studies suggest that the loss of a phospho-acceptor site (Ser 728 ) in the proximal Ct of KCC2 does not account for the behavior of KCC2 4-2-2 and are consistent with the idea of mutationinduced allosteric effects. As will be shown below in Fig. 8, the consequence of certain single point substitutions in KCC2 wt is consistent with this conclusion as well.
Initial Characterizations of wt and Mutant KCC2s in Which Putative PKC Sites Were Altered-Theses characterizations were conducted through immunofluorescence studies, which have already been presented, and functional studies, which are presented in Fig. 6, using once more averaged background-subtracted furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs from several replicate experiments. Here, oocytes were all preincubated for 15 min in various media, the purpose of these studies being to study the effect of PMA on carrier function. These initial characterizations ( Fig. 2A, panels 1 and 9-16, and Fig. 6) show that expression levels and transport activity under each of the conditions tested are similar among the various carriers with 2-fold increases in FRs between medium LH and medium L.
Sensitivity of KCC2 wt to PMA under Low Cl Ϫ /Isotonic versus Low Cl Ϫ /Hypotonic Conditions-As mentioned and presented earlier, studies by our group have shown that PMA can alter the activity of certain KCCs in X. laevis oocytes. Additional experiments were carried out here to determine whether the effect of PMA differs between the isotonic and hypotonic conditions by preincubating oocytes for 15 min in medium L (Fig. 7A) or LH (Fig. 7B) with increasing amounts of PMA or 4␣-PMA (from 0.0 to 1.0 M). Results are shown in Fig. 7 expressed as background-subtracted furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs normalized to those measured with "0" M PMA or "0" 4 ␣-PMA.
In both panels, a [PMA]-dependent decrease in carrier activity is observed, but the kinetics of this effect appears to differ between the conditions. Indeed, a plateau begins to appear at ϳ0.3 M in panel A but at ϳ0.5 M in panel B, and apparent K i (PMA)s (based on a three-parameter Hill equation) were 0.11 M and 0.32 M, respectively. Under either condition, importantly, 86 Rb ϩ FRs are seen to be nearly unaffected by an inactive analogue of PMA, 4␣-PMA. These findings indicate that the mechanisms by which PMA exerts its effect may differ between the two conditions but that the observed inhibition is probably accounted for by changes in kinase activity at the cell membrane.
Effect of PMA on wt and Mutant KCC2s under Low Cl Ϫ /Isotonic versus Low Cl Ϫ /Hypotonic Conditions-To determine whether PMA leads to changes in KCC2 wt activity by promoting direct PKC-carrier interactions or by inducing other regulatory events, and whether it exerts its effect through the same mechanisms with medium LH versus L, we repeated part of the experiments described above using six different mutants in which putative cytosolic PKC sites were removed, either a single site per mutant or all of the sites together. Results are shown in Fig. 8 for a [PMA] of 0.5 M and are expressed as phorbol ester-induced % decreases in furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs. . Initial functional characterizations of KCC2 wt and KCC2s in which putative PKC sites were altered. Oocytes expressing wt KCC2 or mutant KCC2 T34A , KCC2 S728N , KCC2 T787A , KCC2 S940A , KCC2 S940A , KCC2 S1034A , or KCC2 0PKC were assayed for 86 Rb ϩ influx measurements (alongside with H 2 O-injected controls) in medium R with or without 250 M furosemide, plus 1-2 Ci/ml 86 Rb ϩ , plus 10 M ouabain, plus 5 M bumetanide after 15-min incubation also in medium LH (5 mM Cl, 125 mosM) or medium L (5 mM Cl, 194 mosM). Data correspond to background-subtracted furosemide-sensitive, bumetanide-insensitive 86 Rb ϩ FRs and are shown as averages Ϯ S.E. among six to nine experiments (six to twelve oocytes/experiments). All of the FRs measured in medium LH were significantly different statistically (p Ͻ 0.01) than those measured in medium L, and all of the FRs measured in medium L were significantly different statistically (P also Ͻ 0.01) from "0". The composition of the media is as in Table 2.  Table 2.
When the preincubation is carried out in medium LH (Fig. 8A), both the wt and mutant KCCs are seen to behave analogously that is, their transport activity decreases by Ͼ30% with PMA. When, on the other hand, preincubation is carried out in medium L (Fig. 8B), two of the mutants are found to become paradoxically more sensitive to the inhibitory effect of PMA, that is, their transport activity decreases by Ͼ50%. These results suggest that the PMA effector involved in carrier inhibition does not produce its effect via canonical PKC sites. They also suggest that residue 940 (or the region in which it occurs) could play an important role under isotonic conditions by defining the accessibility of other sites to PMA-dependent effectors.

DISCUSSION
In the current study, we have used a mutagenic approach to identify important functional domains and residues in the putative Ct of the Na ϩ -independent group of CCCs. More specifically, we have determined the role of three protein segments in this domain by analyzing a series of KCC2-KCC4 chimeras in which short residue stretches were interchanged between the carriers, and we have also determined whether canonical PKC sites in KCC2 wt modulate or underlie the functional purpose of the protein segments identified. As a result of such studies, we have found that the three segments in question each play a definite role in carrier function but that transport activity by KCC2 wt is not influenced through PKC phosphorylation of canonical PKC sites in the cytosolic domains of this carrier.
One of the functional domains analyzed through this study corresponds to the proximal Ct of the KCCs (residues 645-763 in KCC2 wt and 665-783 in KCC4 wt ). Surprisingly, it was shown to harbor residues that are probably crucial in defining conformational ensembles rather than conveying isoform-specific traits. This assertion is sup-ported by the peculiar behaviors of KCC2 4-2-2 and KCC2 4-2-2(N728S) that are consistent with the occurrence of allosteric effects due to substitutions in the proximal Ct of KCC wt . Indeed, sensitivity to changes in osmolality and transport activity under all conditions tested were higher for KCC2 4-2-2 compared with KCC2 wt , and although transport activity for KCC2 4-2-2 was decreased following the N728S substitution, the PMA effect was paradoxically more pronounced for KCC2 4-2-2 , and it was not prevented by producing the same S728N substitution in KCC2 wt . These results also imply that the KCC2 3 KCC4 mutations did not alter the phenotype of KCC2 wt , because a PKC site was lost in the proximal Ct.
Some of the functional traits exhibited by KCC2 4-2-2 raise the possibility that this chimera was expressed at higher levels and not that it was simply more active at the cell surface. Based on our immunofluorescence data, on the other hand, and because KCC2 4-2-2 exhibited increased sensitivity to changes in osmolality, we view this possibility as unlikely. In this study and in a recent study by Mercado et al. (40), moreover, Western analyses of various KCC2-KCC4 chimeras purified by cell-surface biotinylation showed that changes in activity for these types of mutants were apparently not associated with changes in cell-surface protein expression.
As a complement to the studies that were conducted for this analysis, we have not attempted to narrow the list of residues that may play a role in defining conformational ensembles throughout certain portions of the carrier. It is unlikely, however, that such residues would be found exclusively in the proximal Ct or, if this were the case, that they would interact with residues that only belong to this domain. These assumptions are based on the following data: 1) Mercado et al. (40) have observed that a chimera made of the Nt and central core of KCC4 and of the Ct of KCC2 also displayed increased transport activity compared with KCC2 wt . Because the junction point of this chimera was similar to the one that was used for KCC2 4-2-2 , residues proximal to amino acid 644 could thus play a structural role as well (2). We have recently found that the proximal and distal Ct of the NKCCs, which are closely related to the KCC isoforms, are capable of self-interactions (43,49), suggesting that residues between these distant protein segments can associate with one another in the context of intramolecular and intermolecular interactions.
The second domain that was examined in the current study corresponds to the middle Ct. In this protein segment of KCC2 wt , interestingly, the substitutions were found to induce subtler effects, altering transport activity only under the "LH" and "L" conditions. In fact, KCC2 2-4-2 behaved exactly as KCC4 wt when Cl o Ϫ was 5 mM and as KCC2 wt when it was 86 mM, suggesting that the middle Ct encloses residues that confer differences in transport activity at lower ranges of Cl i Ϫ . It should be mentioned, however, that the functional characteristics of KCC4 4-2-4 did not mirror those of KCC2 2-4-2 under the "L" condition, indicating that differences in carrier activity at lower ranges of Cl i Ϫ are probably specified through additional domains. Although the substitutions that were generated in the middle Ct could have also acted by disrupting conformation, we are reassured with the idea that changes in functional traits reflect changes in behavior-defining residues by finding that the substitutions led to hybrid properties as well as FRs that were all within the range of those observed for KCC2 wt or KCC4 wt . The third domain that was analyzed through this study corresponds to the distal Ct. Surprisingly, the effects of interchanging the latter protein segment between KCC2 wt and KCC4 wt were very different from those of interchanging the proximal or middle Ct. Indeed, both KCC2 2-2-4 and KCC4 4-4-2 assumed the phenotype of the carrier from which the distal Ct originated. For each of these mutants, in fact, the adopted transport  Table 2.
activities (at 5 mM Cl Ϫ or 86 mM Cl Ϫ ) as well as the adopted sensitivity profiles to changes in osmolality were almost identical to those of the wt carrier. These results are consistent with our previous suggestion that differences in ion transport rates at lower ranges of Cl i Ϫ cannot be solely accounted for through variant compositions of the middle Ct. They also point toward the possibility that the distal Ct encloses key residues that convey isoform-specific behaviors.
As for the substitutions created in the middle Ct of KCC2 wt and KCC4 wt , it is reasonable to assume that those created in the distal Ct did not act simply by disrupting conformation. For example, KCC2 2-2-4 and KCC4 4-4-2 exhibited transport rates that were once more within the ranges of KCC2 wt and KCC4 wt , and, in addition, their phenotypes mirrored one another near perfectly. At the same time, it may be of concern that a number of functional parameters were altered as a result of these mutations, indicating that the distal Ct could also be prone to conformational transitions when its composition is altered. In fact, substitution-induced allosteric effects could explain why KCC2 (S940A) exhibited a paradoxical increase in PMA sensitivity and why several chimeras in the Mercado et al. (40) study, mostly KCC4 structures in which short regions along the distal Ct were replaced by KCC2 wt counterparts, deployed "off range" transport activities. In this regard, however, it should be mentioned that, in contrast to KCC4 4-4-2 , the distal end of these functionally peculiar structures all enclosed a number of KCC4 wt residues.
In the same work by Mercado et al. (40), it was concluded that KCC2 wt displays much higher transport activity than KCC4 wt under isotonic conditions because of a unique 15-residue stretch (amino acids 1021-1035) in its distal Ct. For this reason, the stretch in question was called the "ISO segment" and ascribed the role of conveying constitutive activity to the neuronal isoform. Based on our own data, on the other hand, we do not concur with such conclusions. Indeed, the absence of an ISO segment in KCC4 wt and KCC2 2-2-4 did not prevent carriers from displaying obvious transport activity under isotonic conditions, and its presence in KCC2 wt and KCC4 4-4-2 did not prevent carriers from displaying higher transport activity than KCC4 wt under hypotonic conditions. As mentioned earlier, moreover, several mutants in the Mercado et al. study behaved as if their normal structure had been altered. Hence, it would appear that the ISO segment is not a prerequisite for K ϩ -Cl Ϫ cotransport to occur under isotonic conditions and that the region in which it occurs also specifies variant levels of transport activities following changes in cell volume.
In this study, the mechanisms by which variant residues in the Ct convey isoform-specific functional traits have not been elucidated. The differential behavior of closely related carriers such as KCC2 wt versus KCC2 2-2-4 or KCC4 wt versus KCC4 4-4-2 suggests that the Ct plays an independent role in this regard, perhaps by influencing the membrane segment (that mediates ion movement) through long range conformational influences or intra-molecular interactions between the ion binding sites and the Ct (50 -52). As such, and because the Ct of all KCCs is in contact with the cytosol, variant residues in this domain could impart unique traits by allowing each isoform to interact with a different subset of regulatory enzymes or, alternatively, by defining different levels of accessibility to the same regulatory enzymes. Two lines of evidence indicate that the latter possibility may correspond to at least one of these mechanisms: 1) The KCCs share very high levels of homology in their Ct and very similar behaviors in response to given environmental cues, and 2) KCC2 (S940A) , which encloses a mutation in a 41-add-on-residue stretch that is unique to KCC2 wt , behaves as if the accessibility of a site "x" to a PMA effector had been altered.
Based on preliminary data that were confirmed in the current analysis, various enzymes that could have interacted with conserved or iso-form-specific regulatory sites included the PKCs. The conventional isoenzyme PKC␦ appeared as a particularly interesting candidate given that is endowed with Cl i Ϫ -sensitive activity and was shown in previous work to regulate NKCC1, a CCC that shares relatively high levels of homology with the KCCs (53). Somewhat unexpectedly, however, mutagenic studies showed that the effect of PMA on KCC2 wt was not mediated through such enzymes, at least not through novel or conventional PKCs that interact with the carrier at canonical phosphorylation sites for these enzymes. Other regulatory intermediates that can be recruited directly or indirectly through PMA and could interact with the KCC isoforms at various sites include WNK (with no lysine kinase), PASK (proline arginine stress kinase), p38, and JNK, as suggested by a number of recent studies (37, 38, 48, 54 -58).
We have stated earlier that hypoosmotic cell swelling, which is often used to activate various KCCs, affects at least two determinants of transport activity for these carriers: the cell volume and Cl i Ϫ (21)(22)(23). With this maneuver, hence, it is not possible to determine the contribution of each determinant to a change in ion transport. Our study has provided insight in this regard, as we have used isoosmolar preincubating solutions to determine the effect of a change in Cl o Ϫ , and solutions with very low [Cl Ϫ ] to determine the effect of a change in osmolality. To our surprise, we have found that KCC2 wt was more sensitive than KCC4 wt to a change in osmolality and that increased activity for KCC2 wt under isotonic conditions was largely Cl i Ϫ -dependent. Given that KCC2 wt was found to be more active at low Cl i Ϫ as well, one might postulate that the carrier has evolved to optimize Cl Ϫ transport under a variety of environmental cues, a feature that would be particularly advantageous for maintaining neuronal cells hyperpolarized through ␥-aminobutyric acid responses (25,26). In summary, we have shown that the distal Ct of the KCCs plays an important role in defining functional traits for this group of carriers. We have also shown that the basal structural arrangement that is assumed by a large portion of the Ct is probably important in specifying isoformspecific behaviors. One such behavior, higher levels of activity at lower levels of Cl i Ϫ , may correspond to a unique and indispensable characteristic of the neuronal-specific isoform.