I. INTRODUCTION

Top Previous Next References

Probably the earliest suggestion of a coupled Na⁺-Cl cotransport process was made by Shanes (318). Evidence for a tightly coupled Na⁺-K⁺-Cl cotransport (NKCC) mechanism, blocked by so-called loop diuretics and unaffected by ouabain, was first presented by Geck et al. (90). At the time Geck et al. (90) presented their findings, there had already been numerous reports of cotransport of all the possible combinations of Na⁺, K⁺, and Cl cotransport processes. For a while it was hoped that all of the reported cation-coupled Cl cotransport (CCC) processes merely represented different modes of a single transport entity. However, as a result of tremendous progress in this field, in the last 5 years we now know that K⁺-Cl (KCC) and Na⁺-Cl (NCC) cotransporters are separate gene products from one another as well as from the NKCC (see sects. II and III). This review focuses almost exclusively on the NKCC.

A by-product of the widespread interest in the NKCC has been the appearance of a substantial number of review articles (e.g., Refs. 43, 50, 86, 109, 110, 114, 163, 171, 172, 239, 255, 266, 277, 281, 297, 311, 320). So why one more? Previous reviews approach the subject from the standpoint of a somewhat special interest within the field, e.g., the properties of the transporter in some particular cell type (red blood cells, Ehrlich ascites tumor cells, and epithelial tissue), some functional aspect (cell volume regulation), and, most recently, the molecular biology and molecular structure of the NKCC.¹ This review draws on data from all these sources in an attempt to distill a focused picture of our current understanding (and ignorance) of this important ion transport mechanism. Perhaps the best reason for a comprehensive review at this time is that the field is poised on the brink of an exciting new era that will undoubtedly yield interesting and unexpected insights into the workings of these CCC processes.

The past 5 years have seen the major energy in this field being directed into molecular characterization of the cotransporter. These efforts have been rewarded in that two isoforms of the NKCC have been cloned as have two isoforms of the KCC and one isoform of the NCC. We now enter a period in which we can extend our understanding of how these cotransporters work using the newly available molecular biological tools. This review will be a success if it serves to help the reader distinguish between what is well understood from what is poorly understood about the NKCC, its transport mechanism, its regulation, and its function. There are several excellent recent reviews that focus on the molecular biological information that has just become available about this cotransporter (110, 114, 163, 277, 281). Therefore, this review does not attempt an exhaustive coverage of that information; rather, a molecular biological framework is presented.

	II. OVERVIEW OF THE CATION-COUPLED CHLORIDE COTRANSPORT FAMILY

Top Previous Next References

The cotransport of Cl along with Na⁺ and/or K⁺ has been reported for a variety of cells since the 1970s (see sect. III) and was surmised even earlier than that (78). Not long ago, it was difficult for even those working in the field to know with any certainty whether a given K⁺-dependent Cl transport process was a KCC or NKCC or whether a given Na⁺-dependent Cl transport process was NCC or NKCC. In fact, there were reports in the literature which suggested that, in some situations, a single transport moiety could switch "modes" from one to the other given certain stimuli, such as increased osmolarity of bathing solutions (77) or application of vasopressin (334). In addition to an uncertainty about the absolute ion requirements for each of the putative ion cotransporters, there was confusion regarding their pharmacology. Much of this uncertainty resulted from the simultaneous presence of more than one coupled cotransporter in the tissues/cells being studied. We can now be certain that all three of these cotransporters are separate proteins encoded by the same gene family. The gene products of the NCC (80) and the KCC (92, 139, 282) share ~45-50% and 25% identity, respectively, with the gene products of the NKCC (277, 281).

Ion transport studies show all three of these cotransporters share an absolute requirement for Cl as well as at least one cation (either Na⁺ and/or K⁺) and that all three cotransport processes are electrically silent. Haas (110) has proposed that this gene family be termed the cation chloride cotransporter, or CCC, gene family. Pharmacologically, they are somewhat distinguishable. Only the NCC is blocked by thiazide diuretics, whereas only the KCC is blocked by disulfonic acid stilbenes, such as DIDS. The loop diuretics (bumetanide, furosemide) inhibit both the NKCC and the KCC but are much more potent in their action against the NKCC. In addition, there may be an NCC that is inhibited by loop diuretics (e.g., Refs. 196, 241, 334), although there is some disagreement about whether the NCC blocked by bumetanide is the same as that blocked by thiazides (134).

On the basis of a combination of functional and genetic results, both the NKCC and the KCC are found in a wide variety of tissue types. In general, the NCC seems to be largely confined to epithelial tissue such as kidney (especially in the distal nephron), where it participates in the net movement of Na⁺ and Cl across an epithelial barrier. Functionally, the KCC has been best characterized in red blood cells but has also been described in other tissues. The KCC has been associated with regulatory volume decrease in response to cell swelling. In nervous tissue, where it has been recently identified (282), it may participate in maintaining an intracellular Cl concentration ([Cl]_i) at lower than electrochemical equilibrium levels. What follows is a brief description of the properties of the KCC and the NCC. Table 1 summarizes the key similarities and differences among these three CCC.

Of the three members of the CCC family, this is the one for which general recognition and acceptance was perhaps the slowest. This probably relates to the fact that in mammals its most definitive location is the early distal convoluted tubule in the kidney, a region where the NKCC is also prominent. In addition, it was often mistaken for parallel Na⁺/H⁺ and Cl/HCO₃ exchangers. The functional properties of the NCC were first definitively characterized by Stokes et al. (332) using the urinary bladder of the winter flounder, a preparation that apparently lacks the NKCC. They demonstrated 1) a clear interdependency between Na⁺ and Cl effects on net absorption of Cl and Na⁺, respectively; 2) that the transport process did not require K⁺; and 3) that the NCC was inhibited by thiazide diuretics but not by the so-called loop diuretics (which, as we shall see, are the most specific inhibitors of the NKCC) such as furosemide, nor by stilbenedisulfonic acid derivatives (which inhibit the KCC and Cl/HCO₃ exchange, among other anion transport processes). However, there is functional evidence for bumetanide-sensitive K⁺-independent NaCl cotransport in trachea (207, 241) and Necturus gallbladder (196). The later confusion about whether the NCC could be converted into a NKCC probably arose because in some epithelial tissues, as mentioned above, the two transporters spatially coexist (e.g., Ref. 143).

There is some functional evidence for this cotransporter in nonepithelial tissue. In the rabbit heart, treatment with chlorothiazide caused a reduction of cell volume (57). This cell shrinkage effect of thiazide treatment was tentatively attributed to the inhibition of a NCC-mediated uptake process. However, in vascular smooth muscle, thiazides activate a Ca²⁺-activated K⁺ channel (37). It is quite possible the cell shrinkage observed in the heart muscle after thiazide treatment was due to K⁺ (Cl) loss and not to inhibition of an NCC. Also, high-stringency Northern blots failed to reveal any NCC mRNA in rat heart (80). There were early reports of apparent NCC activity in Ehrlich ascites tumor cells (136), but subsequent functional evidence showed that this cell has only the NKCC (89, 155, 202) and the KCC (182). Finally, Northern blot analysis of mammalian tissues (including human tissues) localize it mainly to kidney, but also possibly to small intestine, placenta, prostate, colon, and spleen (in humans, Ref. 45). In rat tissues, high-stringency assays revealed the NCC to be only in the kidney (79). In fact, in situ hybridization probes (79) and mRNA localization (359) reveal that the NCC is exclusively localized to the distal convoluted tubules of the kidney. This result fits very well with what is known about thiazide-sensitive NaCl absorption by the kidney.

The winter flounder urinary bladder behaves functionally much like the mammalian kidney distal tubule and, as we have seen, was the first site of functional characterization of the NCC. The flounder urinary bladder NCC was the first of the CCC to be cloned (80). The human NCC has also been cloned (40) and expressed (225).

Gitelman's syndrome is a human genetic disease associated with a mutation of this thiazide-sensitive cotransporter in the renal distal convoluted tubule (326). The syndrome is characterized by hypokalemic metabolic alkalosis, hypocalcuria, hypomagnesemia, and natriuresis. The mutation, located on chromosome 16, is believed to lead to a loss of function of the NCC.

During the same series of studies that began the linkage between cell volume regulation and NKCC (see sect. IVC), Kregenow (185), also made several critical observations for our understanding of the KCC. The KCC mediates the coupled cotransport of K⁺ and Cl across plasma membranes. Although reversible, it is thermodynamically poised to effect net efflux. Because of this net loss of K⁺ and Cl, it can promote regulatory volume decrease. Thus it is activated by a cell volume increase (135). It may also be involved in "pumping" [Cl]_i to lower than equilibrium levels in neurons (279, 282). At present, there are two known isoforms of the KCC, the "housekeeping" form (KCC1) that is found in a variety of tissues (e.g., Ref. 92) and a neuron-specific isoform (KCC2, Ref. 282). Structurally, both of these isoforms differ more from the NKCC than does the NCC (92, 282). They have an estimated molecular mass of 120-125 kDa (92, 282).

Functional studies have confirmed that the KCC is found in several cell/tissue types, including ascites tumor cells (182) and mammalian kidney epithelial cells (65, 294). However, by far the most detailed functional studies have been performed in the red blood cell (RBC). These properties have been well covered in recent reviews (RBC, Ref. 199; all cells, Ref. 134). I highlight only the key properties that differentiate the KCC from the NKCC. As already mentioned, the KCC is activated by cell swelling, and as we shall see (sects. IIIC and IXC), the NKCC is activated by cell shrinkage. Activation of the KCC by swelling leads to the net loss of K⁺ and Cl along with an osmotic equivalent of water. This results in a reduction of cell volume ("regulatory volume decrease"). Another intriguing functional difference between the KCC and the NKCC is that dephosphorylation activates KCC (e.g., Ref. 154), whereas it inactivates the NKCC (see sect. IXA). The KCC differs from the NKCC in its response to an increase of [Cl]_i as well. Whereas the NKCC is inhibited by such an increase (e.g., Refs. 28, 93; see sect. IXB1), the KCC is stimulated (51, 197, 198). Unlike either the NKCC or the NCC, the KCC can be inhibited by the disulfonic acid stilbenes such as DIDS (52). Because it is unlikely that DIDS crosses the plasmalemma, this property suggests that the DIDS-sensitive site of the KCC is externally accessible.

	III. MOLECULAR CHARACTERIZATION OF THE SODIUM-POTASSIUM-CHLORIDE COTRANSPORTER

Top Previous Next References

Since 1994, all three members of the CCC family have been cloned from a variety of species (see Refs. 50, 163, 277, 281). A quantitative comparison of the derived amino acid sequences of members of this family with other known proteins suggests that it is a unique family, perhaps distantly related to the amino acid-polyamine-choline family of transporters (277). Full-length clones of two isoforms (see below) of the putative NKCC have been reported from rat kidney (79), shark rectal gland (358), inner medullary collecting duct of the mouse kidney (53), the loop of Henle from rabbit kidney (280), human colon (283), mouse kidney (143), human kidney (325), and bovine aortic endothelium (360). The deduced amino acid sequences of these clones have revealed a protein whose molecular mass varies from ~120 to ~130 kDa. A very similar protein has been reported as encoded by the gene YBR235w of chromosome II of the yeast Saccharomyces cerevisiae (18). Hydropathy profiles of the deduced proteins (using the Kyte-Doolittle algorithm) have shown there are three general regions. A central hydrophobic region of ~50 kDa flanked by an amino- (~20-30 kDa) and a carboxy-terminal (~50 kDa) region. These latter two regions are more hydrophilic than the central region of the molecule. Based on such analyses, Xu et al. (358) suggested that there are 12 -helical membrane-spanning regions. These putative transmembrane regions are highly conserved between isoforms, with 75-90% identity between the NKCC1 and NKCC2 isoforms. It has been suggested that alternative topological models (e.g., 14 membrane-spanning regions) might better conform to other predictors of membrane protein topology (277). Both isoforms have consensus N-linked glycosylation sites on a putative extracellular loop that fits with biochemical evidence that the cotransport protein is significantly glycosylated (293).

As mentioned above, there are two isoforms of the NKCC. The isoform initially found on the basolateral membrane of the shark rectal gland is known as NKCC1 (Ref. 358; also known as SLC12A2). The same isoform was also identified in mouse kidney by Delpire et al. (53) but named BSC2 (for bumetanide-sensitive cotransporter 2). The NKCC nomenclature is used in this review. The other isoform is referred to as NKCC2 (or BSC1; also known as SLC12A1). The NKCC1 isoform is the larger of the two with ~1,200 amino acid residues and a transcript size of ~7.4 kb. It has an overall 58% amino acid identity with NKCC2. The NKCC2 isoform is somewhat smaller than NKCC1, containing ~1,100 amino acid residues with a transcript size of ~5 kb. The difference in molecular size is almost entirely accounted for by an additional 80 amino acids at the amino terminus of the NKCC1. In contrast, the carboxy end of the molecule is relatively well conserved, exhibiting >65% identity among the two isoforms and between the same isoforms from different sources (277). Given the difference in transcript size and the relatively low overall amino acid identity between the two isoforms, it is not surprising that they are products of two different genes. Delpire et al. (53) showed in the mouse that the NKCC1 is localized to chromosome 18, whereas the NKCC2 is localized to chromosme 2 (289). The human NKCC1 is localized to chromosome 5 (283).

The NKCC1 isoform is by far the most widely distributed of the two currently identified isoforms. In addition to being found on the basolateral membrane of secretory epithelia, Northern probe studies have indicated this is the isoform found in the plasmalemma of a wide variety of cell types, including most nonepithelial cells (283, 358). There is some evidence for tissue-specific variants of the NKCC1. For example, skeletal muscle presents a somewhat smaller mRNA transcript than is found in other tissues (6.7 vs. 7-7.5 kb, Ref. 283). Because it is found on the basolateral membrane of epithelial cells, this isoform is often referred to as the "secretory" isoform. However, when it is found in the membrane of nonpolar cells, it may be referred to as the housekeeping isoform.

Figure 1 is a general model based on the hydropathy profile for human colonic NKCC1. As already mentioned, the molecule has three regions. The amino-terminal region has the lowest amino acid identity of the three regions among NKCC1 from different species. There is at least one phosphorylation site in this region. This site has the greatest degree of amino acid identity found in the amino terminus across species (213, 358). On the other hand, the carboxy terminal region is relatively highly conserved among species extending from Cyanobacterium to human (281). The carboxy-terminal end has several consensus phosphorylation sites that have a relatively high degree of identity across species (see sect. IXB2). In addition, there are several hydrophobic regions in the carboxy terminus that might also be embedded in the membrane. It is hypothesized that the central ~500-amino acid residue region is arranged into 12 transmembrane domains with connecting loops. The hydrophobic membrane-spanning region is not only highly conserved between isoforms, but also between the same isoform from different species (e.g., Ref. 281). There are two consensus N-linked glycosylation sites on the extracellular loop between transmembrane (TM) segments 7 and 8. It is currently unknown whether a functional NKCC is a monomer or an oligomer.

View larger version (34K): [in this window] [in a new window]   Fig. 1. Model of Na+-K+-Cl cotransport (NKCC) isoform 1 protein based on its hydropathy profile. NKCC1 protein is ~1,200 amino acids with 12 transmembrane (TM)-spanning domains, numbered TM1-TM12. Transmembrane domains 1, 3, 6, 8, and 10 are rather highly conserved between NKCC1 and NKCC2. In addition, intracellular loop linking transmembrane domains 2 and 3 is highly conserved. Virtually entire difference in molecular mass between NKCC1 (~195 kDa) and NKCC2 (~121 kDa) can be accounted for by an additional 80 amino acids on NH2 terminus of NKCC1. Site-directed mutagenesis studies have shown that changes in cation affinity result from changes in TM2, whereas Cl affinity was affected by changes in TM4 to TM7. Bumetanide binding appears to be associated with transmembrane TM2 to TM7 and TM11 to TM12. [Adapted from Payne and Forbush (281).]

Structure-function studies have just begun. In a recent series of studies, Forbush's group made chimeras from shark and human NKCC1. They also made some point mutations. These studies show that ion binding and transport as well as bumetanide binding depend on the ~500 amino acids believed to comprise the conserved hydrophobic transmembrane regions (147-150). These studies have provided strong evidence that the second transmembrane region (TM2; a region highly conserved across the entire CCC family) is an important site in determining cation and bumetanide binding, but not Cl binding. Chloride binding determinants appear to reside in TM4 and TM7. Bumetanide binding determinants appear to be somewhat more diffuse than those for the ions, being in TM2, -7, -11, and -12.

The NKCC2 (BSC1) isoform has thus far been identified only in the medullary regions of the kidney (rat, Ref. 79; rabbit, Ref. 281). This anatomic location of the NKCC2 corresponds with the location of the loop of Henle and the juxtaglomerular apparatus. Thus NKCC2 is believed to play a critical role in two of the crucial homeostatic functions of the kidneys, namely, regulation of extracellular fluid volume and osmolarity.

In studies on the rabbit kidney NKCC2, Payne and Forbush (280) identified three distinct variants that differed only by an alternately spliced 96-bp exon. With the use of such a hydropathy plot model very similar to that for NKCC1 (refer to Fig. 1), it was shown that these exons coded for the amino acids in TM2 as well as 13 amino acids that are contiguous with TM2, but which extend into the intracellular compartment (280). A particularly interesting addendum to this observation was that high-stringency Northern probes indicated these three variants had very distinct regional localization within the rabbit nephron; one was found only in the renal cortex, one only in the renal medulla, and the third in both the cortex and medulla. Igarashi et al. (143) have made a similar observation for the mouse kidney. This finding is consistent with a proposal made by Knepper and Burg (177) that there are some differences between NKCC function in different regions of the thick ascending limb of the loop of Henle. It is possible that the proposed regional differences in Na⁺ reabsorption along the thick ascending limb might be mediated by the different isoforms of the NKCC2. This interpretation fits well with the finding that TM2 is a critical component of Na⁺ and K⁺ affinities for the NKCC (148, 149).

The NKCC2 has also been identified in the renal juxtaglomerular apparatus (143, 163). Igasrashi et al. (143) showed that one of the alternately spliced isoforms (NKCC2B) is found in macula densa cells. Macula densa cells are involved with two important renal homeostatic processes: tubuloglomerular feedback and secretion of renin. LaPointe's group (194, 195) has presented evidence that strongly suggests a key role for the NKCC in the tubuloglomerular feedback process. LaPointe et al. (195) suggest that the NKCC (NKCC2B?) in the macula densa cells is very near its thermodynamic equilibrium point such that small changes in the NaCl concentration in the nephron tubule lumen could affect the net direction of NKCC transport and thereby change the equilibrium potential of Cl (E_Cl). This in turn is hypothesized to affect the resting membrane potential (V_m) of the basolateral membrane of the macula densa cells which somehow transmit this membrane voltage information to the granule cells of the afferent artery. Evidence for this function of the NKCC2 comes from reports that luminal furosemide inhibits tubuloglomerular feedback (152) and renin secretion (209).

Mutations of NKCC2 have been linked to some forms of Bartter's syndrome, a severe human genetic disease involving hypokalemic alkalosis with hypercalcuria largely attributable to dysfunction of the NKCC2 in the thick ascending limb of the loop of Henle (325). The authors reported a variety of mutations, all in the putative transmembrane region of the molecule. The interested reader is directed to a thoughtful discussion of the roles of the NKCC and NCC in Bartter's and Gitelman's syndromes by Hebert and Gullans (129).

	IV. HISTORY OF THE DISCOVERY OF THE SODIUM-POTASSIUM-CHLORIDE COTRANSPORTER

Top Previous Next References

The NKCC is distinguished by the fact that it transports Na⁺, K⁺, and Cl stoichiometrically by means of a tightly coupled mechanism that can be blocked by loop diuretics. Although this distinguishing "fingerprint" can be stated briefly, it took nearly 15 years for workers to put the pieces together (90). Three key elements were required: 1) a conceptual breakthrough; 2) a laborious, empirical characterization of Na⁺ and K⁺ fluxes; and 3) a reason to consider that Cl transport might be directly linked to cation transport. The conceptual breakthrough occurred in the early 1960s with Crane's seminal formulation of the sodium gradient hypothesis and the concept of secondary active transport (see Refs. 46, 47) whereby the transmembrane electrochemical gradient (maintained by a primary active transport process) of one solute (usually Na⁺) provides the energy for the thermodynamically uphill movement of a second solute. It is difficult to overemphasize the importance of this concept to the study of cell solute homeostasis. The labor-intensive characterization of Na⁺ and K⁺ fluxes was performed by a variety of workers whose original intent was to characterize further the operation of the Na⁺ pump. The reasons to consider Cl linkage to Na⁺ and K⁺ movements were almost serendipitous. As we shall see, the Cl linkage idea grew out of studies designed to characterize amino acid transport mechanisms.

Hoffman and Kregenow (137) demonstrated in ouabain-treated human RBC that there was a fraction of Na⁺ efflux that was stimulated by intracellular Na⁺, blocked by ethacrynic acid, and which might not use ATP as an energy source. To explain this finding, they hypothesized that in addition to the ouabain-sensitive sodium pump, there was a second, energy-dependent mechanism capable of moving Na⁺ against its electrochemical gradient. They called this mechanism pump II to distinguish it from the ouabain-sensitive Na⁺/K⁺ exchange pump (pump I). Over the next 8 years, the pump II hypothesis was ultimately disproven. However, it served well to stimulate much-needed research into the properties of coupled Na⁺-K⁺ cotransport processes. In the interim, some workers concluded that the ouabain-insensitive Na⁺ efflux was mainly a Na⁺/Na⁺ exchange process (60, 210), whereas others concluded that these fluxes were simply manifestations of different properties of pump I, depending on the particular experimental conditions (309).

By the late 1960s, Na⁺ pump workers began to notice that K⁺ uptake, in the presence of ouabain, was saturable (e.g., Refs. 84, 310). This was a perplexing finding since it was generally believed at that time that K⁺ movements consisted of only two pathways: 1) via the Na⁺ pump and 2) via electrodiffusive leaks (the pump-leak hypothesis). Thus, in the presence of ouabain, K⁺ influx was not expected to be saturable as a function of extracellular K⁺ concentration ([K⁺]_o), unless there was a hitherto unknown mechanism for K⁺ uptake.

Evidence for such an unknown mechanism was provided by Sachs (309). Using human RBC, he performed a critical series of studies that linked K⁺ uptake with Na⁺ uptake via a ouabain-insensitive mechanism. He showed that (in the presence of ouabain) there was a Na⁺ uptake dependent on external K⁺ that could be blocked by 1 mM furosemide. Under the appropriate set of conditions, he could demonstrate a small net efflux of Na⁺ (in the presence of ouabain) that was also blocked by furosemide. This is the first published report of linking K⁺-dependent Na⁺ fluxes to inhibition by a loop diuretic. However, because he and others had demonstrated that furosemide could inhibit pump I (Na⁺-K⁺-ATPase) fluxes, Sachs (309) concluded that the effects seen in the presence of ouabain were probably mediated by pump I, which had different properties under the experimental conditions being used.

Wiley and Cooper (355) were the first to demonstrate the obligatory coupled cotransport of Na⁺ and K⁺. They extended Sach's finding by demonstrating furosemide-inhibitable, mutually dependent Na⁺ and K⁺ fluxes occurring in the presence of ouabain. Figure 2 illustrates their finding that the furosemide-inhibited Na⁺ influx in human RBC required the presence of extracellular K⁺. Figure 3 shows that a portion of the ouabain-insensitive K⁺ uptake required the presence of extracellular Na⁺ and could be inhibited by furosemide. This was the first demonstration that the furosemide-sensitive components of the unidirectional influxes of both Na⁺ and K⁺ required the presence of the other cation. They also noted, as has had Glynn et al. (95), that there was a saturable component of the ouabain-insensitive Na⁺ uptake. This saturable component of Na⁺ uptake was inhibited by furosemide and had a Michaelis constant (K_m) for external Na⁺ of 24 mM. Furthermore, and this was very important at the time, they demonstrated net, furosemide-sensitive fluxes. This observation, coupled with the observation that furosemide inhibited ouabain-insensitive effluxes of K⁺ and Na⁺ in the absence of extracellular Na⁺ and K⁺, showed that the transport process they were studying was not simply isotopic exchange. The conclusion of these workers was that the RBC possessed a cotransport mechanism that moves both Na⁺ and K⁺ and that there was a codependency in their transport. This represented a very critical stage in the development of the idea of the NKCC. It is important to remember that at that time the idea of secondary active transport in general, and cotransporters in particular, was relatively new and by no means universally accepted.

View larger version (17K): [in this window] [in a new window]   Fig. 2. Inhibition by 1 mM furosemide of Na+ influx into human red blood cells bathed either in K+-containing or K+-free media. Each isotonic medium contained 10 mM NaCl plus requisite concentration of indicated cation as chloride salt. Ouabain (0.1 mM) was present in all media. Values are means ± SD. [From Wiley and Cooper (355).]

View larger version (24K): [in this window] [in a new window]   Fig. 3. Inhibition by 1 mM furosemide of ouabain-insensitive K+ influx into human red blood cells bathed in either Na+-rich or Na+-free media. Each isotonic medium contained 6 mM KCl plus requisite concentration of indicated cation as chloride salt. Ouabain (0.1 mM) was present in all media. Values are means ± SD. [From Wiley and Cooper (355).]

The long and continuing association between what has come to be recognized as the NKCC and cell volume regulation that occurs in response to cell shrinkage began with several critical observations by Kregenow (185, 186). He showed that duck RBC, shrunken by exposure to a hypertonic medium, could (in the continued presence of the hypertonic medium) recover toward their original volume. This recovery depended on [K⁺]_o being elevated somewhat above normal levels and the presence of extracellular Na⁺. In the absence of ouabain, cell volume recovery was accompanied by a net uptake mainly of K⁺ and Cl plus the osmotically obligated water. However, Kregenow (186) showed that the volume regulation did not require a functioning Na⁺ pump (Fig. 4). With the Na⁺ pump inhibited, the main increase was seen in the Na⁺ and Cl contents, with a somewhat smaller increase of K⁺. This important observation implied that the primary solute transport mechanism responsible for the volume increase moved Na⁺ (+Cl) in addition to K⁺ (-Cl). Kregenow (186) supported this interpretation by showing that cell shrinkage was accompanied by an increase in the ouabain-insensitive unidirectional isotopic influx and efflux of both Na⁺ and K⁺. However, in keeping with the cationocentric point of view of those times, the increase of Cl content was believed to be "passive," in response to changes in membrane potential caused by the cation movements. Thus Kregenow's findings showed that cell shrinkage activated a ouabain-insensitive Na⁺ and K⁺ transport pathway or pathways that moved these two cations into the cell (K⁺ against its electrochemical gradient) to effect a net uptake of osmotically active particles. For the first time, Cl was also implicated in the process (albeit in an apparently passive role).

View larger version (22K): [in this window] [in a new window]   Fig. 4. Lack of effect of ouabain on response of duck red blood cells to exposure to a hypertonic medium. Duck erythrocytes were incubated in isotonic or moderately hypertonic fluids containing either 2.5 or 19 mM K+. Ouabain (104 M) was added just before beginning experiment. [K+]o, extracellular K+ concentration. [From Kregenow (186).]

D.  Linking Cl to the Coupled Na⁺ and K⁺ Movements

1.  Evidence in RBC

Since the early 1970s, red cell workers were aware that the coupled Na⁺-K⁺ movements they observed were accompanied by Cl movements in the same direction (see above). However, it was not until the late 1970s that investigators directly addressed a critical question: Are the Cl movements obligatorily linked to the cation movements or are they functionally linked via changes in the electrochemical driving forces? Although the question seems straightforward now, at the time there were conceptual and technical roadblocks both to formulating as well as to addressing this question. The conceptual roadblock was the then-current view (based on results from RBC and frog skeletal muscle) that Cl crossed membranes only by electrodiffusive pathways and was distributed at electrochemical equilibrium across all cell membranes. According to this view, Cl simply traversed membranes through very high permeability pathways in response to primary changes in the electrochemical driving forces as set by the movements of cations. Moving beyond this concept required overcoming two technical roadblocks. One was the difficulty of measuring Cl movements. The principal radioisotope of chlorine (³⁶Cl) has a half-life of over 300,000 years, meaning that it can only be produced with a relatively low specific activity, and it is expensive. The measurement of net Cl movements, while possible, was laborious. The second roadblock had to do with a property of the membrane of RBC: it has both a high Cl conductance and a high Cl exchange flux via the band 3 anion exchanger (AE1).

Both the problems outlined above came together to influence the interpretation of the results of an elegant study by Schmidt and McManus (316). Using duck RBC, they examined the role of Cl in the shrinkage-induced volume increase that had first been demonstrated by Kregenow (185, 186). The duck RBC is an excellent subject to study cell-volume regulation and the associated Na⁺ and K⁺ fluxes. However, like other RBC, it is ill-suited for the study of Cl fluxes associated with volume regulation, because the Cl flux via the Cl/HCO₃ exchanger (band 3 or AE1) is orders of magnitude greater than the Cl flux via the NKCC. As a result of this problem, Schmidt and McManus (316) were unable to measure any Cl fluxes directly. Instead, they looked at effects of varying Cl concentrations and concluded that the Cl movements during cell reswelling were in response to the change in the electrochemical driving force on this anion, caused by the net cotransport of Na⁺ and K⁺. There is an additional circumstance that contributed importantly to the conclusion that Cl movements were passive and not obligatorily linked to those of Na⁺ and K⁺. That is the fact that Cl is distributed at thermodynamic equilibrium across the RBC membrane. This means the E_Cl is equal to the V_m. Thus, as long as the process(es) responsible for this passive distribution were operating, measurements of net Cl fluxes could never reveal Cl flux via the NKCC.

In a later publication from the same laboratory (121), the question of whether Cl was directly linked to cation transport was readdressed using duck RBC whose membrane potential was "voltage-clamped" (using valinomycin) and whose band 3 fluxes were blocked by DIDS. Under these conditions, they completely replaced Cl with permeant anions across the duck red cell membrane. Now, with V_m unable to change and band 3 unable to exchange anions, they showed that only Cl and Br could support furosemide-sensitive net Na⁺ transport. In their earlier report (316), they had reported that reducing both [Cl]_i and [Cl]_o to 20 mM (using acetate as the replacement) still permitted a norepinephrine-stimulated uptake of Na⁺ and K⁺ without any change of [Cl]_i. In the earlier work, this observation had been interpreted to mean there was no absolute Cl requirement by the cotransporter. When this experiment was repeated in the presence of DIDS, it was found that only Cl or Br would support the net uptake of the cations (121). Furthermore, in DIDS-treated cells, they were able to show a furosemide-sensitive net uptake of Cl that equaled the furosemide-sensitive net uptake of Na⁺ plus K⁺. In a particularly convincing series of studies summarized in Figure 5, they also showed that the direction of the transmembrane Cl chemical gradient dictated the direction of the furosemide-sensitive net flux of Na⁺ when the membrane potential was "clamped" with valinomycin (142). Thus this study not only demonstrated that Cl was transported by the cotransporter, but also that the chemical gradient for Cl could "energize" net cotransport-mediated fluxes of the cations.

View larger version (15K): [in this window] [in a new window]   Fig. 5. Demonstration in duck red blood cells that Cl chemical gradient can drive furosemide-sensitive Na+ fluxes at a constant membrane potential (calculated to be 12.4 mV over entire range of [Cl]o, where subscript o refers to extracellular). Cells pretreated with 105 M DIDS (to block band 3 anion exchanger) were incubated with 2 × 106 M valinomycin (to "voltage-clamp" membrane potential), ouabain (to block Na+ pump), and norepinephrine (to stimulate NKCC). In addition, media contained 30 mM Na+ and 100 mM K+ ± 1 mM furosemide. Initial intracellular concentrations were as follows (in mmol/l cell water): Na+ = 7.8; K+ = 174.9; Cl = 95.6. Methylsulfate was used to substitute for external Cl. Subscript c refers to cytosolic. [From Haas et al. (121).]

The original misinterpretation of their results was reinforced by an analysis of the thermodynamic equilibrium of the putative Na⁺-K⁺ cotransporter that agreed with their actual results. That one could arrive at the correct final thermodynamic expression starting with incorrect initial assumptions was the result of two pieces of bad luck: 1) Cl is distributed at thermodynamic equilibrium and 2) Cl is a transported species. In the first report (316), they interpreted their results as being due to an electrogenic Na⁺-K⁺ cotransport with Cl following via an electrodiffusive pathway. Starting with this assumption, they derived an equilibrium equation for such a model as follows.

At thermodynamic equilibrium, the net electrochemical potential (_net) is

(1)

where _Na and _K are the electrochemical potentials for Na⁺ and K⁺, respectively. Knowing that Cl is distributed at electrochemical equilibrium (i.e., V_m = E_Cl) allows one to write the following expression for the net electrochemical potential

(2)

Unfortunately, the exact same equation can be derived starting with the assumption that the cotransporter is electroneutral, that is, there are one Na⁺, one K⁺, and two Cl transported per cycle, i.e.

(3)

Thus we see that although the initial assumptions were quite different (Eqs. 1 and 3), the final equation (Eq. 2) is the same. This identity contributed to the initial misinterpretation by Schmidt and McManus (316). It took the development of the voltage-clamped RBC treated with DIDS (to block the band 3 anion exchanger) to distinguish between the two very different models.

Three other groups studying Na⁺ and K⁺ fluxes in human RBC reported a close relationship between ouabain-insensitive cation influxes and the presence of extracellular Cl. In a preliminary note, Kregenow and Caryk (189) reported that Cl was cotransported with Na⁺ and K⁺ during volume regulation in duck RBC. Dunham et al. (59) showed that ~75% of the ouabain-insensitive influx of K⁺ was dependent on [Cl]_o in a concentration-dependent manner. They further showed that 1 mM furosemide abolished this [Cl]_o-dependent K⁺ influx. Similarly, they showed that a portion of the ouabain-insensitive Na⁺ influx required the presence of external Cl and that this fraction of the Na⁺ influx was eliminated by treatment with furosemide. The fact that the magnitude of the [Cl]_o-dependent Na⁺ influx was significantly smaller than that of the same K⁺ influx as well as the fact that Wiley and Cooper (355) had shown that furosemide blocked a substantial portion of K⁺ influx into cells bathed in Na⁺-free media caused them to conclude that they were studying a KCC. At about the same time, Chipperfield (43) also reported a [Cl]_o-dependent K⁺ influx, blocked by furosemide, but noted that "direct evidence for Cl transport by the same system is lacking." Clearly, the inability to measure non-band 3-mediated Cl fluxes in the RBC was a critical impediment to correctly assessing the role of Cl in the RBC cotransporter.

Thus the first definitive experiments showing the direct linkage between the furosemide-sensitive Na⁺ and K⁺ fluxes with Cl fluxes were performed in a different preparation (Ehrlich ascites tumor cells) and for a completely different reason than the series of studies we have just outlined. In the mid 1970s, Heinz et al. (131) were attempting to determine the energy sources for amino acid uptake. In the course of those studies, his group noticed an apparently paradoxical movement of Cl (131). They had demonstrated a strong electrogenic Na⁺ pump component to the V_m of the Ehrlich cells. Thus, as [K⁺]_o was increased from 0 to 15 mM, the V_m became increasingly negative, an effect largely blocked by ouabain. According to the prevailing dogma of those times, Cl is distributed passively and can rapidly move to maintain Cl electrochemical equilibrium (V_m = E_Cl). It was expected that Cl content would decrease in response to the intracellular negativity. In fact, they reported that the Cl content of the Ehrlich cells actually increased as a function of [K⁺]_o, and this response was exaggerated in the presence of ouabain. Furthermore, they showed that this apparently paradoxical Cl uptake could be blocked by furosemide. These findings were followed (87, 88) by a demonstration that furosemide blocked the unidirectional fluxes of Cl and K⁺.

Finally, in a landmark paper, Geck et al. (90) demonstrated virtually all the basic properties of the NKCC (see sect. V). As seen in Figure 6, they first showed that when K⁺-depleted cells were reexposed to external K⁺, they took up K⁺, Na⁺, and Cl in a ratio of 1:1:2. Using a membrane potential-sensitive probe, tetraphenylphosphonium, they showed that these ouabain-insensitive fluxes had no effect on the membrane potential of the cells. Conversely, they also showed that changing the membrane potential had no effect on the apparent cotransport fluxes. Thus the cotransporter was electroneutral. Using a coupling analysis derived from irreversible thermodynamics, they showed that the net movements of all three ions were tightly coupled to one another and to an isosmotic water flow. From these results, they explicitly proposed that there existed a strongly coupled NKCC in the membrane of the Erhlich ascites tumor cell. They further speculated that reports from experiments on RBC and several epithelia could be explained by the existence of a coupled triple cotransporter, the NKCC, rather than either Na⁺-K⁺, K⁺-Cl, or Na⁺-Cl cotransporters as originally postulated.

View larger version (12K): [in this window] [in a new window]   Fig. 6. Furosemide-sensitive net ion movements. Ehrlich ascites tumor cells, previously K+ depleted and Na+ enriched, were incubated for periods up to 5 min in Krebs-Ringer phosphate buffer with 38 mM K+ and 93 mM Na+. Furosemide-inhibitable net ion movement is defined as difference in intracellular ion content (mmol/g dry wt) between cells incubated in presence of 1 mM ouabain and those incubated in 1 mM ouabain and 2 mM furosemide. This furosemide-inhibitable net movement is plotted against incubation times on left panel. Right panel shows ratio of Cl to K+ net movements and ratio of Na+ to K+ movements. [From Geck et al. (90).]

Although for many years the widely accepted view of Cl distribution was that it was at equilibrium and crossed membranes easily and quickly, results from a number of tissues had never been able to fit into that model (e.g., heart muscle, Ref. 191; nerve cells, Ref. 165; muscle, Ref. 238). For technical reasons, data from the squid giant axon were the most convincing in this regard. As early as 1939 chemical analysis of extruded axoplasm had shown [Cl]_i of the giant axon to be very high (~120-150 mM) (23). (It must be remembered that the ionic composition of the squid's extracellular fluid is similar to seawater, i.e., [Cl]_o is ~500 mM.) Keynes (166) made a careful study of the [Cl]_i in the axon to rule out several potential technical problems and very clearly showed that at 120-130 mM, [Cl]_i was at least three times greater than expected if it were distributed at electrochemical equilibrium (i.e., V_m = 70 mV; E_Cl = 35 mV). Furthermore, he made the intriguing observation that 2,3-dinitrophenol, an inhibitor of oxidative phosphorylation, significantly reduced ³⁶Cl influx into axons. In addition, the conductive (electrodiffusive) pathways for Cl transmembrane movement across the squid axolemma are very small. Adelman and Taylor (1) showed that the Cl conductance was so small that it could account for no more than 10% of the "leakage" conductance across the voltage-clamped squid axolemma. Thus it seemed clear that some sort of active transport process must be responsible for this high, nonequilibrium distribution of Cl in squid axoplasm. The search for the mechanistic basis of this nonequilibrium Cl distribution resulted in a series of studies that used the internally dialyzed squid axon preparation (see sect. VA). These studies showed that the mechanism responsible for the high [Cl]_i 1) required intracellular ATP (5, 302), 2) was inhibited at high [Cl]_i (28, 302, 303), 3) involved the cotransport of Na⁺ and was blocked by furosemide (303), and 4) involved the cotransport of K⁺ and was blocked by bumetanide (306).

In the early 1980s, several labs provided evidence that the NKCC process was found in the thick ascending limb of the loop of Henle (101, 103), in the distal tubule of the kidney (245, 246), as well as in the intestine (242). It was demonstrated that these epithelial transport processes required all three co-ions (101, 242, 246) and were blocked by the sulfamoyl benzoic acid diuretics (see sect. VB, Ref. 245).

Thus, by the early 1980s, investigators had reported functional evidence of the NKCC in a wide variety of cell types. This was a period during which much of the functional characterization of this coupled cotransporter was obtained.

	V. FUNDAMENTAL CHARACTERISTICS

Top Previous Next References

There are three functionally defining and unique characteristics of the NKCC. 1) Ion translocation by the NKCC requires that all three ions (Na⁺, K⁺, and Cl) be simultaneously present on the same side of the membrane. 2) Bumetanide and its congeners (5-sulfamoyl benzoic acid loop diuretics) bind to the cotransporter protein and inhibit ion translocation of all three ions. 3) Under normal ionic conditions, all three ions are translocated, and the translocation process has an electrically silent stoichiometry: 1 Na⁺:1 K⁺:2 Cl for most cells or 2 Na⁺:1 K⁺:3 Cl for at least one cell (the squid giant axon). As a minimum, these three criteria should be met before a given process can be functionally identified as being mediated by the NKCC. In the following section, we consider the evidence for these widely accepted properties and what these properties tell us about the NKCC.

Wherever it has been possible to rigorously examine the individual fluxes of Na⁺, K⁺, and Cl, it has been shown that the NKCC-mediated fluxes of all three ions require the presence of the other two ions on the side of the membrane from which the flux originates (termed the cis-side; e.g., Refs. 8, 90, 121, 229, 303, 306).

Although it would seem to be a simple, straightforward matter to test for this property, using the sensitivity of each flux to bumetanide, it is now clear that complications can arise. Care must be exercised to remove, as much as possible, contributions from other transport pathways for the ion whose flux is being measured. For example, McManus' group initially believed Cl was not directly involved because the band 3 anion exchanger obscured the net Cl fluxes via the NKCC (see sect. IVD1). Prevention of this kind of error requires a preknowledge and understanding of the ion transporters that exist in the particular cell type one is investigating.

Unidirectional flux studies (using radioactive tracers) can be subject to a subtle kind of complicating error: isotopic exchange. In some RBC, the NKCC can mediate isotopic exchange fluxes under the appropriate set of ionic conditions (e.g., Refs. 38, 58, 69, 118, 218; but see Refs. 160, 161, 181; see sect. VIIA for further details). This property can lead to errors when determining the apparent stoichiometry of the cotransport process. It can also lead to misidentification of NKCC as KCC or as NCC. The important thing to remember is that because of this possibility one cannot rule out the presence of NKCC just because the unidirectional flux is not reduced by removal of one of the three cotransported ions. One must strive to arrange conditions that will not permit exchange fluxes of the ion whose flux one is measuring.

Both of these potential problems have been overcome using the internally dialyzed squid giant axon. Because of its large size (~500 µm in diameter and 6-7 cm in length), it has been possible to develop a technique to control the intracellular as well as the extracellular solute composition while measuring unidirectional fluxes. The technique of intracellular dialysis was developed by Brinley and Mullins (31). Briefly, it involves threading a miniature dialysis capillary longitudinally down the axis of the squid axon. The dialysis capillary used has a molecular mass cutoff of ~1,000 kDa, quite satisfactory for the control of small inorganic ions, organic solutes, and ATP. This gives the investigator two powerful advantages for studies on ion transporters. First, it is possible to control the internal milieu, and thereby maintain a steady-state condition in terms of solute concentrations that cannot be maintained in nondialyzed cells. Second, it permits one to directly measure either the unidirectional influx, or the efflux, simply by placing the relevant isotope in the extracellular fluid or the intracellular (dialysis) fluid, respectively, and collecting the fluid from the opposite side of the axolemma.

This technique has been used to examine a number of functional properties of the NKCC, including the requirement for cis-side ions. For the latter studies, the trans-side fluid was designed so that at least one of the necessary co-ions was absent, thereby reducing the possibility of isotopic exchange fluxes (see sect. VIIB). In a series of papers, Russell and co-workers have demonstrated the absolute requirement for the cis-side presence of all three ions both for unidirectional influx (303, 306) as well as for efflux (8). Each of the three co-ions was systematically removed while measuring the bumetanide-sensitive influx of the other two, using a double-label flux approach. The results of three representative experiments are shown in Figure 7. Figure 7A shows the effects on ³⁶Cl and ²⁴Na influx when an axon was sequentially superfused with external fluids that were, in turn, K⁺ free, K⁺ containing, and finally K⁺, 10 µM bumetanide containing. In this axon, it can be seen that providing extracellular K⁺ increased the influx of both Cl (by ~21 pmol·cm²·s¹) and Na⁺ (by ~14 pmol·cm²·s¹) and that both these increases were reversed by the application of extracellular bumetanide. Figure 7, B and C, shows the same general protocol applied to examine the external Na⁺ (Fig. 7B) and Cl requirement. In each case, the increase in the influx of the two co-ions caused by supplying the third one is reversed by bumetanide treatment. These results show that the cis-side absence of any one of the three ions is the same as treating with bumetanide. Thus it is clear that the NKCC requires all three ions be present for transport to occur. In addition, this particular demonstration also shows that after "delivering" the three ions to the inside of the membrane, the cotransporter can return to an outward-facing conformation even though only K⁺ is present on the inside. Because there is no evidence of bumetanide-sensitive K⁺/K⁺ exchange in the squid axon (i.e., K⁺ influx in Fig. 7, B and C, is the same in either the absence of the third co-ion or in the presence of bumetanide), this suggests that the reorientation of the binding sites to the outward-facing conformation must not require intracellular binding, translocation, and extracellular release of ions.

View larger version (19K): [in this window] [in a new window]   Fig. 7. Dependence of cotransport influxes of Na+, K+, and Cl on simultaneous presence of all 3 ions in external fluid. Experiments were carried out on internally dialyzed giant squid axons. Internal fluid in all cases contained the following (in mM): 400 K+, 0 Na+, 0 Cl, 8 Mg2+, and 4 ATP (pH 7.3, osmolality = 970 mosmol/kgH2O). Ouabain (105 M) and tetrodotoxin (107 M) were present at all times. A: effect of external K+ on 24Na and 36Cl influxes. After a control period in 0 mM [K+]o [N-methyl-D-glucamine (NMDG) replacement], 10 mM K+ was applied and both influxes increased; 36Cl influx increased by ~21 pmol·cm2·s1 and 24Na influx by ~14 pmol·cm2·s1. Finally, 10 µM bumetanide was applied causing both fluxes to decrease to levels very near that seen under 0 [K+]o conditions. [Na+]o = 425 mM; [Cl]o = 561 mM. B: effect of external Na+ on 42K and 36 Cl influxes. After a control period in 0 mM [Na+]o (NMDG substitution), 425 mM Na+ was applied leading to increases of both 42K (~8 pmol·cm2·s1) and 36 Cl (~20 pmol·cm2·s1) influxes. Subsequent application of 10 µM bumetanide caused influxes to decrease to levels near those observed in absence of extracellular Na+. [K+]o = 10 mM; [Cl]o = 561 mM. C: effect of external Cl on 42K (~7.5 pmol·cm2·s1) and 22Na (~15 pmol·cm2·s1) influxes. After a control period in 0 mM [Cl]o (gluconate substituted for Cl), 561 mM Cl was applied leading to increases of both 42K and 22Na influxes. Subsequent application of 10 µM bumetanide caused influxes to decrease to levels near those observed in absence of extracellular Cl. [Na+]o = 425 mM; [K+]o = 10 mM. (From A. A. Altamirano, G. E. Breitwieser, and J. M. Russell, unpublished observations.)

Bumetanide is the prototype for the loop diuretics. They are called loop diuretics because the diuresis (increase in urine production) they produce is the result of their action in the thick ascending limb of the loop of Henle. This region of the renal nephron reabsorbs Na⁺ and Cl from the tubular fluid and is responsible for the establishment of the hypertonic interstitium of the renal medulla. Thus, when this isoform of the NKCC (NKCC2) is inhibited, a large increase in urine flow is observed, and the urine is nearly isosmotic with plasma, regardless of the hydration state of the individual, i.e., the individual loses the ability to excrete either a concentrated or a dilute urine. Other representatives of this group (5-sulfamoylbenzoic acid derivatives) are furosemide, an agent used in early studies on the NKCC, but only rarely used experimentally nowadays, and piretanide, a congener that has been used mainly in Europe. Both are chemically closely related to bumetanide but less potent and less specific for the NKCC (e.g., Ref. 313) than bumetanide.

It is difficult to overstate the importance of the loop diuretics to the development of our present level of understanding of the NKCC. They are routinely used to identify fluxes mediated by the NKCC. This latter use has contributed in a major way to our learning that the NKCC is found in a very wide variety of cells (e.g., Ref. 269). They have played an important role in the identification of the protein and the subsequent cloning of the cotransporter, and they have provided information that has been used to interpret ion binding to the NKCC.

Given this importance, it is necessary to emphasize that these agents are not the NKCC's equivalent of the Na⁺ pump's ouabain. In fact, they are far from it. Although these agents inhibit the NKCC with a reasonably high affinity (see below), they can also inhibit other anion transport processes when used in higher concentrations, e.g., Cl/HCO₃ exchange (108, 153), Cl channels (63), and KCC (199). Thus, used alone and/or in high concentrations, not even bumetanide can provide positive proof that a given function is mediated by the NKCC. For example, Kracke et al. (180) showed in human RBC that bumetanide inhibits components of Na⁺ and K⁺ efflux that are neither dependent on cell Cl nor on the mutual presence of the other cation. So one must know the properties of one's experimental subject well before using these "specific" agents, especially for binding studies designed to identify the NKCC. A good general rule of thumb would be when used at concentrations at or below 10 µM, bumetanide is specific for the NKCC.

Bumetanide reversibly inhibits the NKCC in a variety of preparations by a concentration-dependent mechanism, with a half-inhibitory constant of ~1 × 10⁷ M (see Table 2). From the early days of work in this field, there have been attempts to relate the potency of these agents to the concentration of the three cotransported ions. Some workers have observed that the apparent inhibitory potency is inversely related to the [Cl] of the bathing media (117, 264) and positively correlated with [K⁺] and [Na⁺] (116, 264; see sect. VIIIA for further discussion of this important observation).

Lytle and Forbush (213-215) have suggested that there might be significant differences in bumetanide sensitivity between NKCC found in secretory versus those found in absorptive epithelia. Thus they point out that absorptive epithelia such as renal thick ascending limb of the loop of Henle and flounder intestine have apparent dissociation constants (K_D) that are significantly <1 µM, whereas the apparent affinities for some secretory epithelia such as rectal gland and parotid gland are >1 µM. We now know that these absorptive epithelia express NKCC2 on their apical membranes, whereas the secretory epithelia express NKCC1 on their basolateral membranes. Thus this suggestion implies a difference in bumetanide binding affinity between the two NKCC isoforms. This interpretation is borne out by recent studies in which NKCC1 and NKCC2 were expressed in HEK-293 cells. Isenring et al. (150) showed an about threefold difference in the inhibition constant (K_i) for bumetanide-sensitive ⁸⁶Rb flux via the NKCC2 isoform (0.08 µM) compared with the NKCC1 isoform (0.28 µM). Table 2 is a collation of a published K_i values measured in situ in a variety of cells. Most of these cells would be expected to express NKCC1. For such NKCC1 cells, the apparent K_i values span a range of almost two orders of magnitude, from 8.7 µM (Ehrlich ascites cells) to 0.06-0.07 µM (duck erythrocytes and Xenopus oocytes). By comparison, dog kidney outer medulla vesicles, which are very likely to to be enriched with NKCC2, had a K_i value of 0.045 (Table 2). Thus, although the pattern of a somewhat higher bumetanide affinity for the NKCC2 isoform appears to hold for in situ measurements, there is more variability within the NKCC1 cohort than between the NKCC1 and NKCC2 isoforms. Clearly, one could not identify the isoform based on its bumetanide sensitivity except in the more extreme cases.

These compounds have a very high lipid solubility (ether/water partition coefficient = 0.25 at pH 7 and 0.04 at pH 8; Ref. 73). Thus they would be expected to partition easily into and cross biological membranes. This raises the question of their site of inhibitory action. Nitschke et al. (244) addressed this question by constructing two high-molecular-weight congeners of piretanide. One was dextran-sulfonylurea piretanide (Pir-Dex) and the other was polyethylene glycol piretanide (Pir-PEG). Both had molecular weights of ~5,300. Thus they were presumably too large to cross the membrane and thereby gain access to the cytoplasmic-facing face of the membrane. These agents as well as the parent compound isothiocyanopiretanide (Iso-Pir) were tested for their effectiveness in inhibiting NKCC activity in the isolated, perfused thick ascending limbs of rabbit kidneys. In this preparation, the NKCC is located on the apical membrane that faces the tubule lumen. There was no effect on NKCC activity (measured as short-circuit current) when these compounds were applied to the basolateral membrane or the bath side of the preparation. However, when applied to the luminal side, all three agents inhibited the NKCC with the following half-inhibitory constants: Iso-Pir = 3 × 10⁶ M; Pir-Dex = 6 × 10⁶ M; Pir-PEG = 2 × 10⁶ M. In an earlier work, these investigators used piretanide itself and found the half-inhibitory constant to be 1 × 10⁶ M (313). Since there is a great deal of evidence that these agents bind directly to the NKCC protein itself (see sect. VIIB), these results are consistent with the site of inhibitory action being on a portion of the NKCC molecule accessible from the external face of the plasmalemma.

In keeping with the above interpretation, Haas (110) has tentatively suggested that bumetanide binds exclusively to the outwardly facing cotransporter. Unfortunately, this suggestion is now being cited as proof (e.g., Ref. 183). Nevertheless, at present, there is absolutely no evidence excluding the possibility that bumetanide may also bind to the inward-facing cotransporter. Indeed, if bumetanide acts by competing with Cl for the second anion transport site as has been suggested (116, 117; but see Ref. 149), then one might reasonably expect inhibition from an internal site as well.

Given that the NKCC mediates the transmembrane movement of three ions, Na⁺, K⁺, and Cl, there is the possibility that the overall transport process generates a transmembrane current and/or is affected by the membrane potential. This would be most obvious if the ion transport stoichiometry was such that a net charge was translocated with each turnover of the cotransporter. The term electrically silent as used here means that the overall transport process of Na⁺, K⁺, and Cl via the cotransporter is not driven by the cellular transmembrane voltage, nor does the transport process directly generate a membrane current and thereby affect the transmembrane voltage. (This does not exclude the possibility that the "carrier" is charged and hence affected by membrane voltage in some, as yet, unrecognized way.) In fact, it was the movement of Cl against its electrochemical gradient ("paradoxical movement of Cl") that first led Geck et al. (90) to undertake the studies that resulted in the initial published description of NKCC (see sect. IVD2). The electrical silence of the NKCC is a fundamental characteristic of the NKCC that is now well accepted by workers in the field, yet the direct evidence for the assertion is surprisingly sparse. Electrical silence is often inferred from the apparent stoichiometry (see sect. VI) of the cotransport process, but arriving at this conclusion can potentially involve circular reasoning.

Conclusive proof of electrical silence requires direct measurements of membrane potential combined (ideally) with simultaneous measures of cotransporter-mediated ion transport. For obvious technical reasons, such studies are rare. For example, Geck et al. (90) varied extracellular Na⁺, K⁺, and Cl concentrations while measuring furosemide-sensitive water movements. The membrane potential was estimated using the transmembrane distribution ratio of the lipid-soluble cation [³H]tetraphenylphosphonium. They observed that wide variations of the extracellular ion concentrations caused large changes of the furosemide-sensitive water fluxes but had no effect on the distribution ratio of the tetraphenylphosphonium, i.e., on the membrane potential. Although this result was interpreted to mean that cotransport was voltage insensitive, it also implies that the V_m was quite insensitive to very large variations in the electrochemical gradients of the three major extracellular ions, a surprising result. Thus either the cells were very "leaky" or the tetraphenylphosphonium distribution was not accurately reflecting the membrane potential. Either way, this finding casts a serious doubt on the overall interpretation of the result.

Haas et al. (121) addressed the question of whether the effects of varying [Cl]_o on the norepinephrine-stimulated, furosemide-inhibitable Na⁺ and K⁺ net fluxes in the duck RBC were direct (via a coupled cotransporter) or indirect, via a change of V_m. By comparing the effects of variations of [Cl]_o on changes of cell Na⁺ content (i.e., measuring net Na⁺ fluxes) under conditions in which V_m was expected to vary widely, these authors concluded that changes of V_m played no role in the Cl-dependent net fluxes of Na⁺ (Fig. 8). Thus, in the presence of valinomycin and DIDS (to block Cl conductance through the band 3 anion transporter), V_m is expected to be "clamped" to the K⁺ equilibrium potential because, (as the authors showed) under these conditions, the membrane permeability to K⁺ is about seven to eight times larger than the membrane permeability to Cl. In a similar experiment, they also showed that the K⁺ dependence of the norepinephrine-stimulated net Na⁺ fluxes was unchanged by treatment with valinomycin plus DIDS. In this case, [K⁺]_o was varied such that the membrane potential of valinomycin-treated cells was expected to vary between 8 and 129 mV, whereas for the untreated cells, the membrane potential would remain essentially equal to the E_Cl, 29 mV. Although these studies contained no direct measurements of membrane potential, the RBC literature contains considerable support for the view that treatment with valinomycin will make the RBC membrane a reasonably good "K⁺ electrode" (121). Thus the fact that treatment with valinomycin had no effect on the furosemide-sensitive net flux of Na⁺ caused by variations of either [Cl]_o or [K⁺]_o strongly argues that NKCC in the duck RBC is insensitive to changes of V_m.

View larger version (18K): [in this window] [in a new window]   Fig. 8. Lack of effect of membrane potential on Cl-dependent net Na+ fluxes. Initial intracellular concentrations of K+, Na+, and Cl are shown on bottom right. Net Na+ uptake was measured at 41°C from a solution containing 50 mM Na+, 100 mM K+, 105 M DIDS, 104 M ouabain, 106 M norepinephrine, ±2 × 106 M valinomycin, and ±1 mM furosemide. Changes in [Cl]o were made by substituting methylsulfate to maintain a constant extracellular osmolality. In presence of valinomycin, authors estimated PK/PCl = 7.5. With the use of this permeability ratio and with the assumption of a constant field, calculated membrane potential for valinomycin-treated cells was 7.9 mV over entire range of [Cl]o used. In absence of valinomycin, it was assumed that PCl >>> PK, PNa and that PMeSo4 = 0.5PCl. Again, with the assumption of a constant field, membrane potential was calculated to vary between 0.4 and 18.3 mV. [From Haas et al. (121).]

McRoberts et al. (229) used a very similar approach in Madin-Darby canine kidney (MDCK) cells. They varied [K⁺]_o in the presence and absence of 20 µM valinomycin while measuring ²²Na uptake. They found that valinomycin had no effect on the [K⁺]_o-stimulated ²²Na uptake in Na⁺-depleted cells. Hannafin and Kinne (125) found a similar lack of valinomycin effect on bumetanide-sensitive ³⁶Cl uptake by membrane vesicles prepared from the thick ascending limb of rabbit loop of Henle. It should be noted that neither of these studies provided evidence that varying [K⁺]_o in the presence of valinomycin actually affected V_m.

The question of electrical silence could be more directly addressed in the internally dialyzed squid giant axon because membrane potential and influx could be directly measured in the same axon (307). Membrane potential was measured with a microelectrode and was varied using veratridine, an agent which opens Na⁺ channels in the squid axon, resulting in the depolarization of the V_m. The action of veratridine was reversed by the blocker of Na⁺ channels tetrodotoxin. As seen in Figure 9, depolarization of the V_m (in this case, by almost 30 mV) caused Cl influx to increase slightly, but this increase was insensitive to bumetanide and therefore probably occurred via voltage-sensitive Cl channels (146). The bumetanide-sensitive, steady-state influx in the presence of veratridine (V_m depolarized by an average of 28 mV) averaged 19.8 ± 2 pmol·cm²·s¹ (n = 4). This value of bumetanide-sensitive Cl influx is the same as that reported for the squid NKCC under similar conditions of external ion concentrations and of normal membrane potential (306). It is important to recognize that at 17°C, a 26-mV change in membrane potential should result in an e-fold (2.7×) change in the influx if the transporter carried one net charge. Obviously, no such change of bumetanide-sensitive influx was observed. Therefore, these data support the view that the steady-state, unidirectional influx via the NKCC in the squid giant axon is insensitive to changes of the membrane potential.

View larger version (16K): [in this window] [in a new window]   Fig. 9. Effect of membrane potential depolarization on bumetanide-sensitive and -insensitive Cl influx into internally dialyzed squid giant axon. Cl influx was measured using 36Cl. Axon was dialyzed with a Cl-free fluid to prevent contamination of unidirectional Cl flux data with possible Cl/Cl exchange fluxes. Transmembrane resting potential was measured with a 0.5 M KCl-filled micropipette inserted longitudinally alongside intracellular dialysis capillary. When membrane potential of axon was depolarized ~30 mV by application of veratridine, Cl influx increased by ~7 pmol·cm2·s1 (a). Repolarization of axon membrane potential (Vm) was accomplished by treating axon with tetrodotoxin (TTX). Upon repolarization, Cl influx decreased by ~6 pmol·cm2·s1 (b) even though axon was treated with 1 µM bumetanide, a concentration which inhibits >90% of Na+-K+-Cl cotransport flux in this preparation. Thus voltage-sensitive Cl influx was insensitive to bumetanide and therefore presumably not occurring through Na+-K+-Cl cotransporter. [From Russell (307).]

Kracke and Dunham (181) further examined the question of voltage sensitivity of the NKCC by measuring ²²Na uptake (unidirectional influx) into human RBC. Like Haas et al. (121), they also used valinomycin and DIDS to voltage-clamp the V_m to the K⁺ equilibrium potential. Over a membrane potential range calculated to vary between 42 and 118 mV, these workers demonstrated that there was no significant change in furosemide-sensitive ²²Na influx. If the NKCC mediated the movement of a single charge per transport cycle, the constant field equation would predict a 37-fold change in ²²Na efflux over such a membrane potential range. The authors controlled for the possibility that much of the ²²Na influx might be via an Na⁺/Na⁺ exchange mode of the cotransporter by measuring the furosemide-inhibitable ²²Na influx in cells at three different [Na⁺]_i values. Here they found that, if anything, furosemide-inhibitable ²²Na influx was reduced by increased [Na⁺]_i (see sect. IXB2A). Thus, under the conditions of their experiments, there was no evidence of Na⁺/Na⁺ exchange occurring via the NKCC. The overall conclusion once again is that the cotransport mechanism is unaffected by membrane potential changes, in this case, rather large changes.

Alvarez-Leefmans et al. (11-13) have shown that frog dorsal root ganglion cells maintain a higher than equilibrium [Cl]_i by means of an external Na⁺- and K⁺-dependent process. Removal of external Cl resulted in a decrease of [Cl]_i from a value of 25-30 to <2-4 mM (as measured with Cl-selective microelectrodes). When the external Cl was returned, the [Cl]_i recovered by a mechanism that required the presence of extracellular Na⁺ and extracellular K⁺, and which could be greatly inhibited by either furosemide (1 mM) or bumetanide (10 µM). The recovery of [Cl]_i was not accompanied by any significant change of the resting membrane potential. These results of direct measurements of [Cl]_i and resting membrane potential offer further strong proof of the electrically silent nature of the operation of the NKCC.

Thus the finding that the overall transport process is electrically silent provides strong evidence that the cotransport stoichiometry is such that the sum of the cations transported during one cotransport cycle equals the number of Cl cotransported during that same cycle. However, it should be pointed out that nothing in the preceding analysis precludes the possibility that ion binding and/or transport might be influenced by the membrane potential or the membrane electrical field.

	VI. STOICHIOMETRY/THERMODYNAMICS

Top Previous Next References

Because the overall transport of the ions is an electroneutral process (see sect. VC), it follows that during a single turnover of the cotransporter the sum of the cations (Na⁺ + K⁺) crossing the membrane will be equal to the sum of the Cl crossing the membrane. Therefore, the simplest possible transport stoichiometry is 1 Na⁺:1 K⁺:2 Cl. Indeed, this was the stoichiometry first reported by Geck et al. (90). Using K⁺-depleted (and probably Cl-depleted, as well), Na⁺-enriched, Ehrlich ascites tumor cells, these workers measured the net, furosemide-sensitive uptake of each of the three ions as a function of each ion's extracellular concentration. For example, when the net uptakes of Na⁺ and Cl were measured as a function of [K⁺]_o, the ratio of Cl uptake to K⁺ uptake was 2.0, and the ratio of Na⁺ uptake to K⁺ uptake was 0.8 (see Fig. 6, right). Identical studies varying [Na⁺]_o or [Cl]_o yielded the same relative flux ratios. These results led to their conclusion that the NKCC transported ions into the cell with a stoichiometry of 1 Na⁺:1 K⁺:2 Cl and was the first report of the cotransporter's stoichiometry of ion transport.

Haas et al. (121) studied the simultaneous net effluxes of K⁺ and Na⁺ from duck RBC. The net effluxes were measured into an external fluid that contained no Na⁺ or K⁺. This design served two purposes: 1) to maximize the gradient down which these fluxes occur, thereby enhancing their magnitude, and 2) to prevent any possibility of a backflux of either Na⁺ or K⁺. Figure 10 shows that the ratio of the net losses of Na⁺ to K⁺ (virtually all of which were furosemide sensitive) was unaffected by changes in intracellular concentrations of either K⁺ or Na⁺. In addition, they reported (but did not show) that the net loss of Cl in these two studies was equal to the sum of the Na⁺ plus K⁺ losses. This result supports both the concept of electroneutral cotransport of the three ions as well as a stoichiometry of 1 Na⁺:1 K⁺:2 Cl. However, unless some other ion is moving (highly unlikely in this case given the design of the study and the size of the fluxes), the sum of the net K⁺ plus Na⁺ movements has to equal the sum of the net Cl movement to maintain macroscopic electroneutrality. Because it is possible that the net loss of cations created an outwardly directed electrochemical driving force on Cl, which exited the cells by an alternate pathway (e.g., a channel), these results alone cannot prove the stoichiometry of the actual coupled cotransport event. However, the combination of these results with those discussed previously (see Fig. 8), in which these same workers demonstrated that the Cl movements were not voltage driven, makes a powerful argument for the 1:1:2 stoichiometry. Notice that the net losses (effluxes) depended entirely on the simultaneous presence of all three ions intracellularly; removal of either intracellular Na⁺ or K⁺ completely prevented the net loss.

View larger version (23K): [in this window] [in a new window]   Fig. 10. Effect of internal K+ (left) and internal Na+ (right) on net Na+ + K+ efflux from duck red blood cells. Cotransport was stimulated by adding 106 M norepinephrine; furosemide concentration was 1 mM. Co-ion gradients were maximized by incubating cells in media free of Na+ and K+ (choline chloride replaced cations); this also prevented any possibility of back-fluxes of the two cations (but not of Cl). All external media contained 0.1 mM ouabain. Left: with the use of nystatin technique, cells were preloaded to contain [Na+]i of 60.5 mM. [K+] was varied in nystatin loading solutions to yield different levels of cytosolic K+ ([K+]c) noted on abscissa. Right: cells were preloaded with nystatin technique to contain [K+]i of 50.2 mM. [Na+] was varied in nystatin-loading solutions to yield different levels of [Na+]c noted on abscissa. [From Haas et al. (121).]

Among the earliest use of the kinetic analysis approach to the determination of the stoichiometry of cotransport by the NKCC was a study by McRoberts et al. (230). Working on confluent monolayers of cultured MDCK cells, they measured bumetanide-sensitive ⁸⁶Rb and ²²Na uptake as a function of the extracellular concentrations of Na⁺, K⁺, and Cl. They found that the ⁸⁶Rb or ²²Na uptakes were hyperbolic functions of either [K⁺]_o or [Na⁺]_o, suggesting only a single K⁺ and Na⁺ were required to activate the cotransport process. In contrast, the plot of bumetanide-sensitive ²²Na uptake as a function of [Cl]_o was fit by a sigmoidal curve.

The data could be linearized by plotting the bumetanide-sensitive flux (V) against V/[Cl]_n^o, where n was between 1.4 and 2.1. This result suggested a degree of cooperativity between cotransport flux and [Cl]_o. Because the n was greater than 1 for both Na⁺-stimulated ⁸⁶Rb uptake and K⁺-stimulated ²²Na uptake, at least two Cl would be required for each Na⁺ plus K⁺ that was transported. This means that, theoretically, if the cooperativity between Cl binding and ⁸⁶Rb and ²²Na uptake is very high, then a minimum of two Cl must bind to effect cotransport. To the degree that the cooperativity is less, the stoichiometry obtained from this analysis will be an underestimate. This general approach has been used by a large number of workers in the field (e.g., Refs. 169, 260, 346), and all derived the same stoichiometry, 1 K⁺:1 Na⁺:2 Cl. Unfortunately, none has measured the fluxes of all three ions under the same circumstances. Most have measured only one flux, usually ⁸⁶Rb, so the possibility exists that the actual ion transport stoichiometry might differ from the stoichiometry of ion binding to the cotransporter.

This unusual stoichiometry has been reported for two preparations, the squid axon and the ferret RBC. Evidence from the squid giant axon rests on the results of two kinds of experiments. The initial indication that the stoichiometry in the squid axon might be different from that reported in other preparations came from collating the results of experiments that were originally designed to determine 1) the dependence of the influx of each ion on the presence of the other two ions in the extracellular fluid or 2) the effect of raising the [Cl]_i on NKCC-mediated influxes of all three cotransported ions, or 3) the effect of furosemide on the influxes of all three ions (303-306). In all these experiments, the steady-state, unidirectional influxes of either ⁴²K, ²²Na, or ³⁶Cl had been measured in axons whose intracellular Na⁺ and Cl were set at 0 mM by intracellular dialysis. This condition prevents isotopic self-exchange fluxes of these two ions. [Because increasing [Cl]_i and [Na⁺]_i both inhibit NKCC influx in the squid axon (see sect. IXB1), there is no evidence that either Cl/Cl or Na⁺/Na⁺ exchange occurs in the squid axon.] Collating the results of experiments carried out over several years (303, 305, 306) and which examined the effects of several different perturbations on the individual influxes of each of the cotransported ions strongly suggested that a stoichiometry of 1:1:2 might not apply to the squid axon. These results suggested that the ratio might be closer to 2 Na⁺:1 K⁺:3 Cl, but because the experiments had not been designed specifically to address the stoichiometry issue, no firm conclusions could be reached.

To directly address the issue of stoichiometry of transport, steady-state, unidirectional influxes of ion pairs were measured simultaneously. For each pair, one member was always K⁺ so that all stoichiometric ratios were determined relative to the K⁺ influx. Thus either ⁴²K and ²²Na or ⁴²K and ³⁶Cl influxes were measured simultaneously (306). Again, the control internal ionic conditions were set to 0 mM [Cl]_i and 0 mM [Na⁺]_i to prevent the possibility of isotopic self-exchange for Na⁺ and Cl. The effects of two perturbations on the ratio of the flux pairs were examined. One perturbation was to increase [Cl]_i from 0 to 150 mM. This is a treatment known to inhibit flux through the NKCC (see sect. IXB1). This protocol yielded an apparent stoichiometry of 1.8 Na⁺:1 K⁺:3.3 Cl. The second perturbation was to measure the ion-pair influxes in the absence and presence of furosemide. The furosemide-sensitive influxes had an apparent stoichiometry of 2.2 Na⁺:1 K⁺:3.1 Cl. Thus both of these experiments yielded a stoichiometry of ~2 Na⁺:1 K⁺:3 Cl which suggests an "extra" Na⁺ and Cl are transported per cotransporter turnover. It is noteworthy that the results of Figure 7 (which show the external ion dependence of the three ions) also support the notion that, in the squid axon, the transport stoichiometry is 2 Na⁺:1 K⁺:3 Cl.

A similar conclusion regarding the cotransport stoichiometry was reached by Hall and Ellory (122). They measured isotopic uptake of ⁸⁶Rb (for K⁺), ²²Na, and ³⁶Cl into ferret RBC. The ferret RBC contains very high concentrations of Na⁺ (e.g., Refs. 69, 122) and is able to engage in Na⁺/Na⁺ and Cl/Cl self-exchange fluxes via a partial reaction of its NKCC (215, 219). This raised the possibility that some of the isotopic uptake fluxes of ²²Na and ³⁶Cl measured by Hall and Ellory (122) did not represent a full turnover of the NKCC. Flatman (69) tested for this possibility by measuring bumetanide-sensitive net effluxes of Na⁺ and K⁺ into Na⁺- and K⁺-free solution. With this experimental design, which prevented self-exchange fluxes, Flatman (69) showed that the ratio of bumetanide-sensitive Na⁺-to-K⁺ efflux was about 1:1 in the ferret RBC. With the assumption that the NKCC is electroneutral in ferret RBC as has been shown for other cells (see sect. VC), this implies an overall transport stoichiometry of 1:1:2 (Na⁺:K⁺:Cl).

Thus, at the present time, every cell examined except the squid giant axon appears to transport Na⁺, K⁺, and Cl with a stoichiometry of 1 Na⁺:1 K⁺:2 Cl. It is possible that the squid possesses an isoform of the NKCC that is different from those characterized to this point.

Unlike the situation for [Na⁺]_i and [K⁺]_i, [Cl]_i can vary over a fairly wide range depending on the particular cell and the activity of other Cl transporting mechanisms (e.g., Cl channels; Cl/HCO₃ exchanger; Na⁺-dependent Cl/HCO₃ exchanger). In mammalian cells, [Cl]_i can vary from as low as a few millimolar to as high as 50-60 mM. Table 3 collates data from eight representative cells and shows the range of ion concentrations and the calculated net free energies available to the NKCC using both stoichiometries. It can be seen from these data that the net energy available to the NKCC of different cells can vary from strongly favoring net uptake (i.e., G being quite negative) to being at, or very near, thermodynamic equilibrium. In a "typical" mammalian cell (e.g., [K⁺]_i = 140 mM, [K⁺]_o = 4.5 mM, [Na⁺]_i = 15 mM, [Na⁺]_o = 140 mM, [Cl]_o = 110 mM), the thermodynamic equilbrium for the NKCC (with a stoichiometry of 1 K⁺:1 Na⁺:2 Cl) would be with a [Cl]_i of ~60 mM. The 2:1:3 stoichiometry derives much more energy from the same ionic gradients to pump Cl into the cell. In the case of the squid giant axon, this stoichiometry yields an equilibrium [Cl]_i of nearly 750 mM.

View this table: [in this window] [in a new window]   Table 3. Net free energy in the combined Na+ + K+ + Cl chemical gradients in several representative cells

Consideration of the information in Table 3 raises the question of why all cells that express the NKCC do not have a [Cl]_i at or very near 60 mM. What prevents the NKCC from achieving thermodynamic equilbrium? What serves as the physiological "brake" on the NKCC in these cells? Considerable evidence points to the phosphorylation state of the cotransporter protein and/or the level of [Cl]_i (see sect. IX).

	VII. TRANSPORT MODEL OF THE SODIUM-POTASSIUM-CHLORIDE COTRANSPORTER

Top Previous Next References

Normal operation of the NKCC requires that 1 (or 2) Na⁺,1 K⁺, and 2 (or 3) Cl must all bind to the NKCC on the same side of the membrane before cotransport across the cell membrane will occur (see sect. VA). Is the order of binding of these cotransported ions random? Or is there a preferred order of binding? A preferred order of binding would occur from an allosteric effect when the binding of one ion increases the apparent affinity of a binding site for the next ion. This is a special form of cooperativity. Experimental evidence for ordered binding can be obtained by performing activation studies for any one of the cotransported ions, e.g., Cl, at various concentrations of the other two co-ions, e.g., Na⁺ and K⁺. Several such studies have been reported for cotransport influx. In all cases, the apparent external affinity for each of the cotransported ions increases with increased extracellular concentrations of the co-ions (e.g., Refs. 33, 58, 125, 234, 295).

If, as seems very likely, the cotransported ions bind in an ordered fashion, what is that order? Although the question is simple, getting an answer has proven difficult. Three groups have addressed this issue, and the result is three somewhat different models.

The best-tested model to date resulted from an insightful use of the ion dependencies of self-exchange fluxes mediated by the cotransporter in duck RBC (200, 217-219) and in human RBC (58). Briefly, these authors made two key observations regarding the ionic requirements necessary to get either K⁺/K⁺ self-exchange or Na⁺/Na⁺ self-exchange via the NKCC. They showed that the NKCC of the duck RBC mediated K⁺/K⁺ exchange when the extracellular medium contained only K⁺ and Cl and the cytoplasm contained K⁺ at a high concentration as well as lesser, but finite concentrations of Na⁺ and Cl. On the other hand, Na⁺/Na⁺ exchange was favored by a different set of internal and external ion compositions: the extracellular fluid needed to contain all three co-ions while the internal fluid needed to contain only Na⁺.

From this pattern of ionic requirements for NKCC-mediated self-exchange fluxes, the binding order model seen in Figure 11 was postulated by McManus and co-workers (217-219). The model assumes that only the fully loaded and the completely unloaded form can "cross" the membrane and that fully loaded "carriers" oscillate between inwardly and outwardly facing conformations faster than do completely unloaded carriers. It assumes that the NKCC can mediate either net influx or efflux of all three co-ions, depending on the thermodynamic gradient (see sect. VIB). It further assumes glide symmetry, which means the first ion to bind on one side will be the first ion to be released on the other. Net cotransport influx would require that the ions bind to the cotransporter in the order shown as steps 1-4. Step 5 represents the reorientation of the NKCC molecule such that the binding sites are now accessible from the intracellular compartment, and steps 6-9 represent the ordered release of the ions. Step 10 is the reorientation of the completely unloaded cotransporter to an outwardly facing conformation. According to this model, Na⁺/Na⁺ exchange involves only steps 1-6, followed by steps 6-1. The high [Na⁺]_i serves to prevent net Na⁺ unloading at the internal site (only Na⁺/Na⁺ self-exchange can occur). On the other hand, K⁺/K⁺ exchange involves steps 3-8 followed by steps 8-3. The high [K⁺]_i serves to prevent the release of the second Cl. Duhm (58) also reported data obtained from human RBC, most of which could be well-fit by this clever model.

View larger version (49K): [in this window] [in a new window]   Fig. 11. Hypothetical model of ordered ion binding with glide ("first on/first off") symmetry. Model assumes that cotransport is completely reversible and that only fully loaded and completely unloaded forms of cotransporter protein are capable of changing orientation from inward-to-outward (or vice versa) facing. Numbers refer to order of binding, reorientation, and release for one complete influx cycle. Because both binding and release are ordered, a high concentration of substrate on release side can hinder release of subsequent ions. Thus a rise in [Na+]i would be expected to prevent subsequent release of Cl and K+ by preventing step 7. This could be manifested as a reduction of net influx or as an increase in Na+/Na+ exchange depending on exact ionic conditions. [Modified from Lytle et al. (219).]

Recently, this model has been quantitatively tested using simulations (24). Flux data from duck RBC, HeLa cells, and epithelial cells were fitted with differential equations developed based on the Lytle-Haas-McManus model. The experimental data were well fit by these equations.

However, there are nagging problems that suggest that this model may not be the absolute final word on the mechanism of ion binding by the NKCC. The model was initially derived from, and explicitly predicts, exchange fluxes for all the co-ions. Although these appear to be present and prominent in duck RBC and in ferret RBC (69), they are not universally described. For instance, Kracke et al. (180) were unable to demonstrate K⁺/K⁺ exchange in human RBC. Kaji (160) also failed to demonstrate K⁺/K⁺ exchange in cultured mouse medullary thick ascending limb cells. As already mentioned, there is no evidence for Cl/Cl or Na⁺/Na⁺ exchange in squid axon (see sect. VIB). Other qualitative predictions of this model have also not been verified. Milanick (232) pointed out that the Lytle-McManus-Haas model predicts that high [K⁺]_o or [Cl]_o would have inhibitory effects on net efflux via the NKCC, as well as on Na⁺/Na⁺ exchange fluxes. No such effects of extracellular Na⁺, K⁺, or Cl have been reported. Although increases of [Cl]_i have profound inhibitory effects on NKCC-mediated influxes (and effluxes; see sect. IXB), the complete removal of extracellular Cl has only a modest stimulatory effect on NKCC-mediated effluxes in the squid giant axon (8, 29). Of course, it is very possible that different isoforms of the NKCC may have somewhat different ion binding mechanisms or patterns.

If the glide symmetry model is correct, then the simplest structural correlate would be that the Na⁺ binding site would be found deeper in the membrane than the K⁺ site. Isenring et al. (149) have recently shown by means of mutagenesis studies that the transmembrane region apparently conferring Na⁺ binding properties is located more superficially within the plane of the bilayer than is the putative K⁺ binding site. Of course, this interpretation depends on the correctness of the topological model (Fig. 1). It is of interest that this structural information suggests that the first ion to bind ought to be the K⁺, and a recent squid axon flux study arrived at that same conclusion (7).

Miyamoto et al. (234) have developed a different ion-binding model using flux results obtained from studies on HeLa cells. In agreement with Lytle and McManus (218), they postulate that Na⁺ binds before K⁺. However, their model suggests that one Cl binds simultaneously with the Na⁺ at a "low Cl affinity" site, and the second Cl binds simultaneously with K⁺ at a "high Cl affinity" site. In addition, they postulate mirror symmetry, i.e., that debinding at the intracellular face occurs in the opposite order as binding on the extracellular face. This model does not account for Na⁺/Na⁺ or K⁺/K⁺ exchange (but see above). Finally, Benjamin and Johnson (24) show that the data of Miyamoto et al. (234) could be fit rather well with the Lytle-McManus-Haas model.

McRoberts et al. (229) were able to describe much of the kinetic properties of the NKCC found in the MDCK cells (a canine kidney cell line) beginning with the assumption of random but cooperative binding of the ions to the NKCC. This model predicts mutual dependence of each ion's apparent affinity on the concentrations of the other two co-ions. At the same time, it predicts that the maximum velocity (V_max) values obtained would be unaffected by the co-ion concentrations. These predictions were verified by the results of Rindler et al. (295) for MDCK cells as well as by Miyamoto et al (234) for HeLa cells.

Clearly, the issue of whether the NKCC binds ions in a preferred order or simply displays cooperativity of binding is still far from resolved. Whether the differences described above are the results of unsuspected experimental complications or are the result of different isoforms of the NKCC occurring in the different cells being examined remains to be resolved.

One such anion is NO₃. Numerous studies have shown that NO₃, when present as the only anion, will not support NKCC ion transport (e.g., Refs. 86, 130,174, 311). However, several studies have demonstrated that, in the presence of low, nonsaturating [Cl], NO₃ will support some NKCC-mediated ion transport (33, 174, 234, 346). Brown and Murer (33) used vesicles from apical membranes of LLC-PK₁ cells and showed that when external Cl was completely replaced by NO₃, there was no bumetanide-sensitive ⁸⁶Rb uptake. However, in the presence of small concentrations of extracellular Cl, NO₃ was quite effective at supporting NKCC cotransport. This manifested itself as a hyperbolic relation between [Cl]_o and NKCC-mediated ⁸⁶Rb uptake when NO₃ was the Cl substitute. In contrast, when gluconate was used as the Cl substitute, the relation between [Cl]_o and NKCC-mediated ⁸⁶Rb uptake was sigmoidal. Using vesicles from rabbit parotid glands, Turner and George (345) showed that NO₃ alone could support NKCC-mediated ²²Na influx ~10-25% as well as Cl. However, in the presence of a subsaturating concentration of Cl, addition of 90-100 mM NO₃ increased the ²²Na influx 40-60% as much as adding 90-100 mM Cl. Similarly, Kinne et al. (174) using vesicles from the rabbit kidney outer medulla showed that NO₃ and SCN, by themselves, could only support ~26% as much ²²Na influx as an equal [Cl] when they were the only anions, but in the presence of 10 mM Cl, 90 mM SCN was 73% as effective as 90 mM Cl and NO₃ ~55% as effective. These results have been interpreted to mean that, in the presence of small concentrations of Cl, the NKCC can bind and transport other anions. This suggests that the sequence of binding of the cotransported ions is Na⁺:Cl:K⁺:A, where A is NO₃ or SCN. If this interpretation is correct, then there may be two distinct anion binding sites whose binding characteristics differ (see sect. VIIIB1 for more evidence of two distinct sites).

It is important to point out that not all workers have been able to demonstrate such effects for NO₃ and SCN. Hegde and Palfrey (130), working with intact duck RBC, found that neither of these anions could support NKCC-mediated ⁸⁶Rb influx in the presence of 10 mM Cl. Indeed, these workers showed both anions further inhibited the bumetanide-sensitive ⁸⁶Rb influx. Miyamoto et al. (234) reported that, in the complete absence of extracellular Cl, NO₃ supported a small fraction of furosemide-sensitive Rb⁺ uptake, but in the presence of Cl, NO₃ inhibited NKCC-mediated Rb⁺ uptake. These latter workers reported that SCN, even in the complete absence of Cl, did not support NKCC-mediated Rb⁺ uptake, and in the presence of Cl, SCN was inhibitory. Given what we are beginning to learn about structure-function properties of the two isoforms, and the existence of splice variants, it seems highly likely that at least some of these functional differences will turn out to reflect some structural differences.

There is additional evidence for the two Cl binding/transport sites that may have quite different apparent affinities. Brown and Murer (33) plotted their [Cl]_o (gluconate replaced Cl) versus ⁸⁶Rb uptake data with an Eadie-Scatchard plot (used to evaluate kinetic parameters of two enzymes using the same substrate, Ref. 317). This plot suggested there were two Cl binding sites, one with an apparent K_m of 5.1 mM, and the other with an apparent K_m of 55.3 mM. They suggest the lower affinity site is the one that NO₃ binds to in the presence of Cl. Miyamoto et al. (234), using intact HeLa cells, have presented data that could be fit with a two-site model. The two sites had significantly different affinities (24 and 103 mM).

Taken together, these results strongly suggest that at least some of the isoforms of the NKCC have two distinct Cl binding/transport sites, and they have rather different binding properties. This implies a rather different structure for each of these two Cl sites. Verification of this hypothesis awaits future structural studies.

It has been shown in several preparations that Rb⁺ is as effective or nearly as effective as K⁺ in supporting fluxes via the NKCC (e.g., Refs. 33, 130). It has also been reported that Cs⁺ can partially substitute for K⁺ (e.g., Refs. 86, 130) and that Tl⁺ may actually be more effective than K⁺ at supporting the cotransport fluxes (86). Among physiologically relevant cations, NH₄⁺ may substitute at the K⁺ site (duck RBC, Ref. 130) particularly for the NKCC2 isoform in the thick ascending limb of the kidney (15, 96, 174). This may play an important role in the nephron for NH₄⁺ recycling, which is important for excretion of excess acid. Sodium is completely ineffective at replacing K⁺ on the cotransporter. The Na⁺ site may accept Li⁺ to a certain degree (e.g., Refs. 117, 130, 311). No other cation has been reported to substitute for Na⁺ on the NKCC.

	VIII. BUMETANIDE BINDING STUDIES

Top Previous Next References

As discussed in section VB, inhibition of ion transport by bumetanide or its congeners is a critical piece of evidence that the responsible mechanism is the NKCC. Another important use of the loop diuretics was as a probe for the NKCC protein. Forbush's group (358) was ultimately able to clone the NKCC from shark rectal gland as a result of having first labeled a membrane protein with [³H]bumetanide. They developed antibodies to this protein and then used the antibodies to screen an expression library for the bumetanide-binding protein. Their strategy started with the assumption that [³H]bumetanide bound uniquely to the NKCC protein. This assumption is based on findings that the binding of bumetanide (and its congeners) to membranes have, in many (but not all) cases, three properties. These properties are consistent with the view that this binding is specifically to one of the Cl binding sites of the NKCC involved in Cl transport across the membrane (e.g., Refs. 117, 173). They are as follows: 1) bound molecules are displaced by other members of the loop diuretic family with apparent affinities that match their potency as inhibitors of cotransport fluxes; 2) Na⁺, K⁺, and Cl all must be present in the binding medium in order for specific binding to occur; and 3) raising [Cl] in the binding medium above the optimal level necessary for binding results in a reduction of the specific binding.

A.  Functional Evidence for Effects of External Ions on Bumetanide Binding

1.  Effects of external Cl

Even before the recognition of a coupled NKCC mechanism, some early workers examining the effects of the loop diuretics in isolated perfused nephrons had suggested that the loop diuretic furosemide, which carries a net negative charge at physiological pH, might act by competing with Cl for a transport site on the apical membrane of kidney thick ascending limb cells. The results of the initial experiments designed to test this idea were inconclusive (35). However, a little later, Ludens (211) using the toad cornea as a model for renal thick ascending limb cells showed that the ability of furosemide to block the short-circuit current (which depends on an apical membrane NKCC) was reduced as the [Cl]_o was increased. When these results were plotted as a double-reciprocal plot, the characteristic relationship of a competitive inhibitor was revealed. The fitted K_m for Cl in the absence of furosemide was 53 mM; in the presence of 22 µM furosemide, the apparent K_m for external Cl rose to 145 mM.

Haas and McManus (117) reexamined this issue by measuring norepinephrine-stimulated net uptake of Rb⁺ (as a K⁺ surrogate) and Li⁺ (as a Na⁺ surrogate) into duck RBC as a function of [Cl]_o. Just as Ludens (211) had shown, they found that increasing [Cl]_o had the effect of reducing the apparent affinity for bumetanide (see Fig. 12). Similar results showing apparent competitive interaction between loop diuretics and Cl have been reported for several other preparations (e.g., shark rectal gland plasma membranes, Ref. 173; chick cardiac cells, Ref. 77). It should be pointed out that these data cannot be fit by a simple Michaelis-Menten relationship. However, if these data are plotted as percent inhibition versus bumetanide concentration, they can be fitted with a Hill equation. In this case, the apparent Hill coefficients are considerably <1 (for [Cl] = 20 mM, the Hill n = 0.46; for [Cl] = 100 mM, n = 0.55). This implies a more complex (and less direct) interaction between bumetanide and the NKCC protein than a one-to-one binding competition. Thus there may be allosteric interactions that account for the observed effects. From a standpoint of the practical use of these agents, it should be noted that at normal [Cl]_o values, concentrations of bumetanide above 1-10 µM are likely to be fully inhibitory. Interestingly, this concentration of bumetanide is also fully inhibitory for the squid NKCC, where [Cl]_o is 560 mM.

View larger version (18K): [in this window] [in a new window]   Fig. 12. Effect of varying external [Cl] ([Cl]o) on bumetanide inhibition of Cl-dependent uptake of Rb+ and Li+ in duck red blood cells stimulated with norepinephrine. Lines were drawn by a curve-fitting program that also calculated IC50 values shown. Difference in IC50 for 2 curves was statistically significant (P < 0.001). [From Haas and McManus (117).]

Palfrey et al. (264) examined the effects of reducing the extracellular concentrations of Na⁺ or K⁺ on cAMP-stimulated ⁸⁶Rb influx into turkey RBC. They reported that such reductions actually shifted the bumetanide dose-response curves to the right, i.e., reduced the apparent affinity for the inhibitor. Kracke et al. (180) also demonstrated a clear dependence of the degree of furosemide inhibition of the human RBC NKCC on the concentration of extracellular K⁺. These results suggest that the cotransported cations are necessary for loop diuretic inhibition and, presumably, for binding of the diuretics to the NKCC.

The fact that high [Cl] seemed to reduce the apparent affinity for bumetanide, whereas Na⁺ and K⁺ seemed to increase it, has been interpreted in light of the ordered ion-binding hypothesis (see sect. VIIA). This led to the view that bumetanide competes for the second of two Cl transport binding sites. This site is also believed to have a lower Cl affinity and to be less selective among anions than the "first" site (see sect. VIIB; but also see Ref. 149). Thus the thought is that bumetanide and Cl compete for the second Cl binding site following the binding of Na⁺, the first Cl and then a K⁺. When bumetanide binds to the "second" Cl site, cotransport is blocked. Alternatively, these results are consistent with a mechanism in which the loop diuretics bind to some other anion binding site on the NKCC whose affinity is affected (allosterically) by the presence of Na⁺, K⁺, and Cl.

Forbush and Palfrey (73) synthesized and studied in detail [³H]bumetanide and [³H]benzmetanide binding to membranes isolated from the outer medulla of the dog kidney. Given the region of the kidney from which the membranes were prepared and the properties of the membranes to which the loop diuretics preferentially bound, there is good reason to believe they were studying the binding properties of the apical membrane of thick ascending limb cells. Northern and in situ hybridization experiments have subsequently shown that this region of the kidney contains the absorptive or NKCC2 isoform of the cotransporter (79, 143, 280).

They demonstrated a saturable component of reversible [³H]bumetanide binding to these membranes. A Scatchard analysis of the binding data was consistent with one bumetanide molecule per binding site. They showed that, at 22°C, binding had a half-time of ~5 min and a dissociation half-time of ~10 min. (At 0°C, dissociation had a half-time of >120 min.) From the association and dissociation rates they calculated an equilibrium binding constant for bumetanide of 4.9 × 10⁸ M. These determinations were performed in a medium that contained 30 mM each of Na⁺, K⁺, and Cl. Further evidence that they were studying binding to the site that effects transport inhibition came from results of displacement studies using other members of the sulfamoylbenzoic acid loop diuretic family (e.g., furosemide, benzmetanide). These agents displaced [³H]bumetanide binding with apparent affinities that were in the same rank order as those reported to effect inhibition of transport in other preparations such as avian RBC (probably the NKCC1 isoform) or rabbit kidney thick ascending limb of the loop of Henle (probably the NKCC2 isoform). The rank order of displacement was benzmetanide > bumetanide > furosemide > p-methoxybenzmetanide > 3-amino-4-phenoxy-5-sulfamoylbenzoic acid.

In view of the functional evidence cited previously for interaction among the three cotransported ions and inhibition of the NKCC by the loop diuretics, Forbush and Palfrey (73) performed a series of studies that examined the effect of independently varying each transported ion on saturable [³H]bumetanide binding. Figure 13 shows the results. Several important points emerged from this study. First, in the absence of any one of the three cotransported ions, [³H]bumetanide binding was reduced to no more than 10-20% of maximal binding levels (i.e., binding measured in the presence of all 3 ions). Second, both Na⁺ and K⁺ activate [³H]bumetanide binding, with the relationship between [³H]bumetanide binding and both [K⁺] and [Na⁺] being monotonic, saturating functions. This result has been cited as further support for a 1:1 Na⁺:K⁺ stoichiometry of NKCC cotransport (see sect. VIA1). Third, the effects of Cl on [³H]bumetanide binding were biphasic. Very low concentrations of Cl were necessary for maximal binding, but increasing [Cl] above 4 mM resulted in a significant reduction of diuretic binding. Physiological levels of [Cl] reduced the binding by >50% when the membranes were exposed to a subsaturating concentration (2.3 × 10⁷ M) of bumetanide. This result was generally interpreted to mean that there are two Cl sites; the first site must bind a Cl, and the second site is where Cl and bumetanide compete for binding.

View larger version (14K): [in this window] [in a new window]   Fig. 13. [3H]bumetanide binding is a function of [Na+], [K+], and [Cl]. Membranes prepared from dark red outer medulla of dog kidneys were incubated with [3H]bumetanide (0.25 mM) for 20 min in presence of various salts. Unless otherwise indicated, ion concentrations were 128 mM Na+, 64 mM K+, 128 mM Cl, and 32 mM SO42. In each experiment, concentration of only one ion was varied. , K+ substituted with choline; , Na+ substituted with choline; , Cl substituted with SO42. Each curve is a separate experiment and has been normalized to maximal binding in that experiment. In the cases of both Na+ and Cl studies, ion concentration was increased to 128 mM, as shown by 2 data points at extreme right of figure. , Cl substitution data from a separate experiment in which membranes were isolated in a Cl-free medium (i.e., at 0 on abscissa). , Na+ replacement in an experiment with 10 mM K+ and 15 mM Cl instead of 64 mM K+ and 128 mM Cl (i.e., at 0 on abscissa). [From Forbush and Palfrey (73).]

The general pattern of overall ion effects on [³H]bumetanide binding to the putative NKCC protein has been reported by several investigators for a variety of cells (e.g., Refs. 106, 111, 126, 130, 160, 255, 256, 287, 345). Only one significant deviation from this overall pattern of ion-dependent [³H]bumetanide binding to putative NKCC proteins has been reported. Altamirano et al. (9) studied binding to squid optic lobe microsomes and were unable to demonstrate any Na⁺ requirement for binding, although the effects of K⁺ and Cl were qualitatively as reported for all other preparations. Whether this constitutes additional evidence that the squid NKCC is a different isoform (squid NKCC has a different transport stoichiometry also; see sect. VIA2) remains to be determined.

The forgoing results established ion-dependent [³H]bumetanide labeling as a potentially important tool for studying the NKCC. There have been a large number of studies using labeled loop diuretics (mainly [³H]bumetanide) in a variety of tissues examining what we now know to be both of the currently identified isoforms of the NKCC. Virtually all the results agree that specific [³H]bumetanide binding is enhanced by the presence of all three cotransported ions and can be competed off by other members of the loop diuretic family with a characteristic relative potency sequence (e.g., benzmetanide > bumetanide > piretanide > furosemide, Ref. 269). However, there are numerous differences in other details regarding the properties of diuretic binding. Some of these differences undoubtedly relate to true differences in properties of different isoforms; others may relate to differences in tissue or membrane purity or unidentified differences in experimental conditions.

Another possible reason for some of the aforementioned differences may have to do with the assumed near-absolute specificity of the binding. As a result of the forgoing studies, it has been widely accepted that the three key fingerprints for the specificity of diuretic binding to the NKCC protein are 1) demonstration that the putative site binds the diuretics with relative binding affinities of benzmetanide > bumetanide > piretanide > furosemide; 2) binding to the putative NKCC site must depend on Na⁺, K⁺, and Cl as described above; and 3) raising [Cl] above a critical level must result in a reduction of specific [³H]bumetanide binding.

However, the literature of this area provides at least one clear example that those three criteria alone are inadequate to guarantee the identity of the diuretic binding protein. The Tamm-Horsfall protein (338) is found in the urine of many mammals (312). Immunohistochemical studies showed that it is expressed in the distal regions of the nephron, especially in the luminal membrane of the thick ascending limb of the loop of Henle, which, not coincidentally, is the site of the NKCC2 isoform. It is a highly glycosylated protein with a molecular weight of ~100,000. Soon after Forbush and Palfrey (73) reported the properties of [³H]bumetanide binding to dog kidney membranes prepared from predominantly thick ascending limb of the loop of Henle cells, a report appeared on a study of labeled furosemide binding to purified Tamm-Horsfall protein (105). There was a remarkable similarity between the key properties of the [³H]bumetanide binding reported by Forbush and Palfrey (73) and those of furosemide binding to the Tamm-Horsfall protein. Thus [¹⁴C]furosemide binding was greatly enhanced by the presence of Na⁺, K⁺, and Cl to the binding media. Raising the [Cl] in the binding medium above a critical level led to reduced furosemide binding. Also, it was reported (but not shown) that other furosemide congeners displaced the [¹⁴C]furosemide. Thus, qualitatively, this binding met the general criteria for binding to the NKCC. Quantitatively, there were some differences from those reported by Forbush and Palfrey (73). For instance, the apparent mean affinity constant (K_0.5) for stimulation of diuretic binding is lower for Na⁺ than for K⁺, an order opposite to that usually observed, and the inhibitory effect of Cl was not observed until [Cl] was greater than 100 mM, a much higher value than is usually reported. Finally, Santoso et al. (312) using immunohistochemical techniques showed a lack of correlation between the location of the Tamm-Horsfall protein and known locations of transport via the NKCC. The recent success of cloning and expressing the NKCC has clearly shown that the Tamm-Horsfall protein is not the NKCC, yet the uncanny similarity of the diuretic binding properties of this protein to those of the NKCC should serve as another warning against the use of reversible ligands for the unequivocal identification of proteins.

Nevertheless, the use of labeled loop diuretic congeners has led to important insights about three aspects of NKCC structure/function: 1) to identify the NKCC protein, which, in turn, led to one of the successful strategies for cloning the NKCC; 2) to identify the activated state of the NKCC protein; and 3) to study the site of loop diuretic binding.

The first convincing demonstration that near-ultraviolet (UV) irradiation ( 300-400 nm) would result in an irreversible inhibition of NKCC-mediated flux was made by Amsler and Kinne (17). They exposed LLC-PK₁ cells (a cell line derived from pig kidney) bathed in bumetanide-containing solution to near-UV light. This treatment caused a selective inhibition of the NKCC-mediated ⁸⁶Rb influx with little or no effect on sodium pump-mediated influx or on the ouabain- and bumetanide-insensitive influx as long as the bumetanide concentration at the time of UV exposure was <5 µM. Exposure of the cells to the same near-UV irradiation in the absence of bumetanide was without effect. Thus this result shows that photoactivation of bumetanide will result in an irreversible inhibition of the cotransporter functionally without obvious effects on other transport processes.

The definitive study using irreversible loop diuretic binding was that of Haas and Forbush (112). In that study, the authors made use of the fact that one congener of bumetanide, 4-benzoyl-5-sulfamoyl-3-(3-phenyloxy)benzoic acid (BSTBA), contains a photoactivatable group, benzophenone. Their study used membranes prepared from the cortex of dog kidneys (a site of NKCC1), either a crude membrane preparation or fractions of membrane prepared by centrifugation through a sucrose gradient. They demonstrated that BSTBA could competitively displace [³H]bumetanide from these membranes with an apparent affinity nearly identical to that of bumetanide itself. They exposed these membranes bathed in a solution containing 0.27 µM [³H]BSTBA plus Na⁺, K⁺, and Cl to UV light ( = 366 nm), in the absence and presence of 10 µM unlabeled bumetanide. They then separated the proteins in these membranes using polyacrylamide gel electrophoresis as seen in Figure 14. Of the three peaks of [³H]BSTBA labeling observed, only one (the peak centering around gel slice number 15) is blocked by the unlabeled bumetanide. The molecular mass of the protein centered in this region was ~150 kDa. They did observe another peak of labeling in the 50- to 60-kDa region of the gel (gel slice number 30). Several observations argue against this peak being the NKCC protein. First, cold bumetanide did not compete it off. Second, binding to this peak was linear up to 1.2 µM, whereas the binding to the 150-kDa region saturated with an apparent half-saturation constant of 0.094 µM. Third, binding to the 50- to 60-kDa protein was largely unaffected by omission of K⁺ and Cl (but not Na⁺) from the media. Thus not all three requirements were met in the case of the 50- to 60-kDa protein. Could it be a proteolytic fragment of the NKCC? That also seems unlikely since the authors showed that the 50- to 60-kDa protein binding came predominantly from much lighter membrane fragments than did the 150-kDa protein. Interestingly enough, the authors mention that when they photolabeled membranes using [³H]bumetanide, they, like Jørgensen et al. (159), found labeling only in the low-molecular-weight region. No functional evidence of irreversible inhibition caused by UV irradiation was given for either the BSTBA or the bumetanide.

View larger version (18K): [in this window] [in a new window]   Fig. 14. Evidence of photoactivated, irreversible binding of 3H-labeled 4-benzoyl-5-sulfamoyl-3-(3-phenyloxy)benzoic acid (BSTBA) binding to membranes prepared from dog kidney cortex. Top: a densitometric scan of SDS-polyacrylamide gel of [3H]BSTBA-photolabeled membranes just before cutting gel into slices for scintillation counting. Arrows indicate position of molecular mass standards (from left to right, in kDa) of 200, 116, 92, and 66, and last arrow is tracking dye. Bottom: distribution of [3H]BSTBA in gels of membranes incubated in 0.27 mM [3H]BSTBA and exposed to UV light. Solid symbols and heavy lines refer to membranes in absence of 10 mM bumetanide, and open symbols and thin lines represent data from membranes prepared in presence of bumetanide. Gels were cut into 4-mm slices, digested, and counted. [From Haas and Forbush (112).]

Forbush and co-workers have used photoactivated, irreversible binding of [³H]BSTBA to study the NKCC of two other preparations, duck RBC (113), and the shark rectal gland (72, 213-215). The results of these later studies agreed well with the original study in terms of providing several lines of evidence that the irreversible binding was to a protein that represented at least a part of the NKCC protein. In addition, two new pieces of information were obtained. First, Haas and Forbush (113) demonstrated that BSTBA would irreversibly block NKCC-mediated fluxes in parallel with its binding to the 150-kDa protein. Second, in the shark rectal gland, they identified the binding protein as a 195-kDa protein that was present in great abundance. This permitted Forbush and co-workers (213, 221) to partially purify the protein and use it to produce several monoclonal antibodies. These antibodies were, in turn, used to screen a shark rectal gland cDNA library and isolate the shark rectal gland NKCC cDNA (358).

Nevertheless, there is considerable qualitative evidence showing that stimuli that increase cotransport flux often increase the level of [³H]bumetanide binding (e.g., Refs. 76, 111, 118, 162, 257), although not always proportionately. There are also notable exceptions to this generalization. For example, two groups (175, 249, 250) have shown that although bradykinin significantly enhances cotransport fluxes in vascular endothelial cells, it has no effect on [³H]bumetanide binding. Slotki et al. (327) have shown that an 8-h incubation of HT-29 cells with forskolin increases the bumetanide-sensitive ⁸⁶Rb influx by two- to fourfold without changing the B_max of [³H]bumetanide binding. In T84 cells, Mathews et al. (228) showed that cAMP-stimulated NKCC flux activity could be attenuated by treatment with phalloidin without any effect on the level of [³H]bumetanide binding. Hegde and Palfrey (130) showed that both [³H]bumetanide binding and ⁸⁶Rb influx into duck RBC were stimulated by raising the pH of the external medium from 6 to 8. However, the degree of stimulation of the two properties was quite different. To further illustrate this, the results of Haas and McManus (118) and Haas and Forbush (111) have been combined in Figure 15. These two papers reported the effects of shrinking on duck RBC in the absence or presence of 10⁶ M norepinephrine. Figure 15 shows that although norepinephrine increases both ion fluxes and [³H]bumetanide binding, the changes are not proportional.

View larger version (22K): [in this window] [in a new window]   Fig. 15. Comparison of [3H]bumetanide binding and Na+ flux in duck red blood cells exposed to a range of external osmolalities either in presence or absence of 106 M norepinephrine (N-E). Exposure to different osmolalities caused cell volume (cell water) to vary on either side of normal value of 1.49 kgH2O/kg cell solids. All data were normalized to value obtained at normal cell volume. In absence of N-E, relative changes of [3H]bumetanide binding () and Na+ flux () increase caused by cell shrinkage were parallel. However, in presence of N-E, [3H]bumetanide binding () is enhanced much more than Na+ flux (). [Flux data from Haas and McManus (118); binding data from Haas and Forbush (111).]

These observations may suggest some error in the hypothesis (e.g., bumetanide does not bind to a Cl cotransport site, but to some other anion binding site on the NKCC) or faulty experimental design. In many cases, the measurements of flux and bumetanide binding were made in separate studies and were not designed to test the hypothesis per se.

Thus, as pointed out by Haas and McBrayer (115), the use of [³H]bumetanide binding to monitor NKCC activity is not fully proven. Nevertheless, it has been a useful tool to demonstrate that a number of agents that either stimulate or inhibit cotransport activity also similarly affect diuretic binding.

As discussed in section VIIIA1, it is postulated that bumetanide binding requires the first Cl to bind to a transport site and then bumetanide competes with Cl for the second Cl transport site (e.g., Refs. 110, 117, 173). Thus it has been shown that Na⁺ and K⁺ and low concentrations of Cl increase the affinity of the NKCC for both Cl binding and bumetanide binding. However, Griffiths and Simmons (106), working with rabbit renal cortical membranes, showed that although inclusion of all three ions increases the total specific [³H]bumetanide binding, it does not increase the apparent affinity for such binding.

Turner's group (237, 345) has specifically addressed the issue of whether bumetanide competes with Cl for a transport site or whether bumetanide binds to another site (or sites) for which Cl may have an affinity. In the first of these studies, Turner and George (345) directly compared the effects of a group of anions on both NKCC-mediated ion fluxes and [³H]bumetanide binding. Using basolateral membrane vesicles prepared from rabbit parotid gland, they characterized the ionic requirements for [³H]bumetanide binding and found, as others have (see above), that Na⁺, K⁺, and Cl must all be present for specific bumetanide binding to occur. Furthermore, they showed that binding was maximal when [Cl] is 5 mM; an additional increase of [Cl] resulted in a reduction of specific binding. They next examined the effect on [³H]bumetanide binding of having 5 mM [Cl] present plus 95 mM of the test anions. In a parallel study, they examined bumetanide-sensitive ²²Na uptake from a media containing 50 mM Cl plus 100 mM of the test anions. Table 4 compares the relative effectiveness of this series of anions at either supporting cotransport in the presence of a subsaturating [Cl] for transport or at inhibiting [³H]bumetanide binding in the presence of an optimal [Cl] for binding. For four of the eight anions, Br, Cl, NO₃, and formate, the relative order of effectiveness for the two functions is the same, although there are quantitative differences. However, for the other four anions, the differences are extreme. Sulfate, acetate, and gluconate do not support cotransport either alone (not shown) or in the presence of 50 mM Cl, yet they displace [³H]bumetanide. In particular, sulfate is nearly as effective as Cl at displacing [³H]bumetanide from its binding sites. Conversely, fluoride with a small, but finite ability to support cotransport fluxes has no effect on diuretic binding. This suggests that the second anion binding site for cotransport may have very different properties from the [³H]bumetanide binding site, a finding which casts doubt on the view that they are one and the same site.

Turner and co-workers (237) followed up this study by showing that whereas formate alone could monotonically stimulate [³H]bumetanide binding, SO₄² could only inhibit such binding. Furthermore, when 50 mM SO₄² was added to 5 mM Cl, [³H]bumetanide binding was considerably reduced. On the other hand, 50 mM formate in the presence of 5 mM Cl greatly stimulated [³H]bumetanide binding. Neither of these anions had any effect on [³H]bumetanide dissociation. They interpret their findings as supporting a model in which there are two intracellular anion sites (separate from the cotransport sites) that influence bumetanide binding. One is a relatively high-affinity site that stimulates bumetanide binding, hence the need for low concentrations of Cl for [³H]bumetanide binding. The second is a relatively low-affinity site that inhibits [³H]bumetanide binding. The authors postulate that this anion site that inhibits [³H]bumetanide binding may be the same as that which inhibits NKCC flux. In this regard, it should be noted that intracellular SO₄² had only modest inhibitory potency against NKCC influx. Regardless of whether the bumetanide-binding inhibitory site is the same as the NKCC flux inhibitory site, it seems clear that the effect of Cl to inhibit [³H]bumetanide binding is unlikely to be mediated through competition with Cl for a cotransport site.

Further evidence suggesting that the bumetanide binding site is likely to be different from the Cl binding/transport sites comes from work of Hegde and Palfrey (130). They studied [³H]bumetanide binding to duck RBC membranes as a function of [Cl]. They found, as many others have found, that increasing [Cl] above a certain level (15 mM in this case) results in a reduction of the diuretic binding. However, when they plotted these data several ways, they found that raising [Cl] caused a reduction in the B_max with no change of K_D. This is indicative of noncompetitive inhibition, which also suggests that bumetanide does not bind to a Cl binding/transport site, but rather to another site (cf. Refs. 148, 149). In addition, they point out that the second Cl binding/transport site has a halide selectivity sequence (I > Br > Cl) that is similar to anion selectivity sequence 1 (357). Because this is a common sequence for many anion binding proteins in biological systems, it could not account for the relatively high selectivity bumetanide has for the NKCC.

Recently, Isenring and co-workers (148, 149) have presented structural evidence showing that different regions of the NKCC1 molecule are responsible for binding of bumetanide and Cl. They studied effects on ion and bumetanide binding caused by making chimeras and point mutations of the putative transmembrane regions of shark and human NKCC. They showed that TM2 is critical for both Na⁺ and K⁺ binding and cotransport, TM4 is critical for K⁺ and Cl binding and cotransport, whereas TM7 is critical for all three ions. On the other hand, bumetanide binding is critically dependent on TM2, TM11, and TM12. These results are consistent with the view that the bumetanide binding site(s) is not the same as those for Cl transport.

	IX. REGULATION/MODULATION OF COTRANSPORTER ACTIVITY

Top Previous Next References

In their seminal paper, Geck et al. (90) struggled with the possible role of ATP. They had reported earlier (131) that metabolic inhibition (treatment with antimycin A) prevented the kind of ion movements that they were now showing to be via the cotransporter. In their 1980 study, they attempted to distinguish between primary and secondary active transport by plotting furosemide-sensitive water flow into the Ehrlich cells as a function of the calculated free energy in the chemical gradients of Na⁺, K⁺, and Cl, using a stoichiometry of 1 Na⁺:1 K⁺:2 Cl. This analysis appeared to show that insufficient energy resided within the chemical gradients of the three cotransported ions to account for the observed net transport. They pointed out that their analysis rested on two unverified assumptions: 1) that the ionic activity coefficients in the cytoplasm and the incubation medium were the same and 2) that the ions were not compartmentalized within the cell. The actual raw ion concentration data used by Geck et al. (90) to calculate the relationship were not given in the paper. It therefore becomes a moot point that there is some evidence from studies with ion-selective microelectrodes that the intracellular activity coefficients for Na⁺ and K⁺ are lower than their extracellular activity coefficients (49). The data given simply do not permit one to correct for low activity coefficients nor test whether another Na⁺, K⁺, Cl stoichiometry would better fit the data.

Geck et al. (90) tested for the stoichiometric utilization of ATP by NKCC by asking whether furosemide-sensitive net fluxes caused ATP hydrolysis or stimulated cellular glycolysis. First, they inhibited ATP production from mitochondrial oxidative phosphorylation using antimycin A and measured the rate of decline of cellular ATP content. Treatment with ouabain (an inhibitor of the Na⁺-K⁺ pump) slowed down the rate of ATP depletion by a factor of ~2. However, furosemide treatment had no measurable ATP-sparing effect either with or without ouabain. Because the magnitudes of the NKCC fluxes and the Na⁺-K⁺ pump fluxes are comparable, the authors concluded that ATP hydrolysis did not provide the "missing" energy for NKCC. In addition, they were unable to demonstrate any furosemide-sensitive changes in glycolytic lactate production. Thus the role of ATP in the NKCC mechanism of Ehrlich cells was unclear, although Dr. Heinz stated during a discussion at a meeting that the energy source for NKCC cotransport must be something other than ATP (85).

The first direct demonstration that a cation-Cl cotransporter required cellular ATP was made in the internally dialyzed squid giant axon (303). In this preparation, ATP concentration can be varied while maintaining normal ion gradients, intracellular pH (pH_i), and cell volume, something not possible in other preparations. In a series of studies it was shown that reduction of ATP concentration to <10 µM resulted in 1) complete inhibition of furosemide/bumetanide-sensitive, external Na⁺-, external K⁺-dependent Cl influx (5, 303, 306); 2) complete inhibition of furosemide/bumetanide-sensitive, external K⁺- and external Cl-dependent, Na⁺ influx (303, 304, 306); 3) complete inhibition of furosemide/bumetanide-sensitive, external Na⁺- and external Cl-dependent K⁺ influx (304, 306); 4) complete inhibition of bumetanide-sensitive, internal K⁺- and internal Na⁺-sensitive Cl efflux (8); 5) complete inhibition of bumetanide-sensitive, internal K⁺- and internal Cl-dependent, Na⁺ efflux (8). The foregoing results clearly demonstrate the obligatory nature of the ATP requirement for unidirectional fluxes via the reversible NKCC found in the internally dialyzed squid giant axon. The relationship between ATP concentration and bumetanide-sensitive ³⁶Cl influx was hyperbolic with an apparent K_0.5 for ATP of ~90 µM (5). It is important to note that the inhibitory effect of ATP deprivation develops as rapidly as intracellular ATP is washed out. Thus either a "very long-lived" phosphorylated species is not involved or a protein phosphatase rapidly dephosphorylates the active species. The latter would imply a phosphorylated system with a relatively rapid turnover. Finally, the inhibition by ATP depletion is readily and fully reversible.

Experiments with the, ,-methylene analog of ATP (a "nonhydrolyzable" analog) showed that this analog would not support NKCC but could apparently compete with normal ATP for the activation of the cotransporter (304). In the presence of cellular ATP, intracellular vanadate (0.3 mM) had no effect on furosemide-sensitive K⁺ or Cl influx (306). Thus, although the cotransporter requires ATP, it seems highly unlikely that ATP activates NKCC via an E₁-E₂-type ATPase.

Further early evidence for a role for ATP in the operation of the cotransporter was provided by Saier's group. They demonstrated that MDCK cultured cells possess an NKCC and that reducing cellular ATP concentration to ~3% of control (probably ~30-60 µM) was sufficient to abolish the K⁺-dependent ²²Na and the Na⁺-dependent ⁸⁶Rb uptake (229, 295). Subsequently, ATP requirements have been demonstrated for NKCC in human RBC (3, 48), ferret RBC (70, 122), turkey erythrocytes (268, 347), and human cultured colonic cells (169). In the case of the human RBC, inhibition of furosemide-sensitive (1 mM) Na⁺ and K⁺ fluxes was incomplete. The reported ATP levels were <100 and 30-50 µM, respectively.

There is now widespread agreement that ATP is necessary for cotransport. It is therefore somewhat surprising that there are several reports of NKCC-mediated ion fluxes using membrane vesicle preparations in which ATP was not added to the intravesicular fluid (33, 80, 173, 174, 345, 353). It is possible that these vesicles contained compartmentalized ATP (230). Alternatively, the process of vesicle preparation may have "frozen" the NKCC in a phosphorylated state, e.g., perhaps the protein phosphatase(s) responsible for dephosphorylation (see sect. IXA2C) are cytoplasmic and lost in the process of vesicle preparation.

The results of the experiments utilizing turkey erythrocytes are particularly interesting in view of the fact that two means of stimulating NKCC were tested for their ATP sensitivity. One way of stimulating the avian cotransporter was to increase cellular levels of cAMP. Obviously, one would expect ATP to be required for a cAMP-mediated response, both to generate cAMP in response to the agonist (e.g., epinephrine) and to provide the necessary phosphate group for the cAMP-activated protein kinase to attach to protein(s). Satisfyingly enough, two groups (268, 347) found the cAMP-stimulated NKCC activity was reduced as ATP concentration was reduced. Another means of activating NKCC activity in avian RBC was to cause the cell to shrink. Apparently, this mode of activation is independent of the cAMP-stimulated mode. Again, both groups showed that the enhanced cotransport activity was inhibited by ATP depletion. However, a puzzling aspect of these findings arises from quantitative considerations. Both groups found that the cAMP-elicited NKCC activity was more sensitive to ATP depletion than the cell shrinkage-elicited activity. As mentioned above, the two mechanisms are apparently mediated via separate pathways so a difference in the ATP requirement might not be surprising. However, the relatively high K_0.5 for ATP of the cAMP-elicited activity is puzzling (1.7 mM, Ref. 347; 0.6 mM, Ref. 268). Membrane-bound protein kinase A (PKA) from human RBC has been shown to have, like most protein kinases, a high affinity for ATP (K_0.5 = 10 µM; Refs. 97, 301). Whether these apparent discrepancies reflect technical problems (e.g., ATP compartmentalization in intact cells that is not reflected in whole cell ATP determinations) or fundamentally different experimental conditions that might make hazardous the comparison of biochemical measurements of enzyme activity in membrane fragments with rates of ion transport, is unknown at the present time. Alternatively, cAMP-mediated stimulation may require higher ATP concentrations at an activation step that occurs after cAMP production and the presumed protein phosphorylation.

Other transport processes believed to be secondary active or gradient driven have been shown to be affected by ATP depletion. Sodium/calcium exchange (19, 26, 56) as well as Na⁺-glutamate cotransport (20) in squid giant axon are examples of such processes. Perhaps even more surprising is the finding that sugar uptake in RBC (a facilitated diffusion process, e.g., no movement against a gradient) is also modulated by cellular ATP (132). For the Na⁺/Ca²⁺ exchange and Na⁺-glutamate uptake systems in squid axon and the glucose uptake mechanism in RBC, the evidence available supports the concept that ATP affects the affinity properties of the carriers such as to permit them to be closer to saturation at physiological substrate levels. In general, no effect on V_max was noted in these kinetic studies of the ATP effect. It is important to note that in these instances ATP had a modulating effect, that is, the transport processes were not abolished in the absence of ATP. In contrast, ATP depletion completely abolishes NKCC cotransport activity (see above).

Only one attempt has been made to examine the kinetic effects of ATP depletion on the transport behavior of the NKCC. Ikehara et al. (145) treated HeLa cells with various concentrations of glucose in the presence of the inhibitor of oxidative phosphorylation, carbonyl cyanide m-chlorophenylhydrazone (CCCP). This treatment achieved stable intracellular ATP concentrations within 90-120 min. These cells were then exposed to various concentrations of extracellular Rb⁺, and the furosemide-sensitive net uptake of this K⁺ replacement was measured. They showed that the apparent K_0.5 for the ATP was ~0.95 mM, and the Hill coefficient was between 1.5 and 2. This K_0.5 was ~10 times higher than was reported for the squid giant axon (5). Whether this difference in ATP affinity is a real difference between the mammalian and the squid NKCC is unclear. The means used to reduce ATP concentration in the HeLa cells would necessarily result in an increase of ADP concentration, and it is possible that this nucleotide might engage in competitive inhibition with ATP for the ATP-binding site and thus increase the apparent K_0.5. Ikehara et al. (145) further showed that as cellular ATP concentration was decreased, the apparent affinity of the NKCC for extracellular Rb⁺ also decreased with no effect on the V_max of the cotransporter. This result suggests that phosphorylation of the cotransporter may activate cotransport by a multistep process rather than a simple "all-or-none" mechanism. Although this seems consistent with Lytle's (212) result showing increasing levels of phosphorylation of the NKCC protein with increasing levels of stimulation, it does not seem to fit with the observation that ATP depletion will completely block NKCC transport activity (see above).

Although the majority of available evidence supports the present view that the effects of ATP on the NKCC are mediated via a protein phosphorylation/dephosphorylation mechanism, other possibilities exist. We have already discussed the evidence that the ability of loop diuretics such as bumetanide to bind to the NKCC may be in some cases a measure of the activation state of the NKCC protein. To the extent this is true, it is of interest that the maximal [³H]bumetanide binding (B_max) to squid optic lobe microsomes could be increased as much as 50% simply by including 100 µM ATP in the binding media (9). No effect on the apparent binding affinity of the bumetanide was noted. This effect by ATP had an apparent half-saturation constant of 0.9 µM, well under the half-saturation value of ~90 µM reported to activate NKCC ion fluxes in the internally dialyzed squid axon (5). Because this effect was observed in the nominal absence of Mg²⁺ (0 mM Mg²⁺ and 5 mM EDTA in the binding medium), it seems very unlikely to be the result of additional NKCC phosphorylation. Further evidence against this effect being via protein phosphorylation comes from the fact that a range of adenine nucleotides was also effective at increasing the saturable [³H]bumetanide binding, including ADP, cAMP, ,-methylene ATP, ,-methylene ATP, and adenosine 5'-O-(3-thiotriphosphate). Even adenine (at 1 mM) caused a slight enhancement. However, thymine and cytosine were completely without effect at 1 mM. It is important to note that in the nominal absence of ATP, specific [³H]bumetanide binding to the microsomes was about one-half to two-thirds the value seen in the presence of ATP. Therefore, this effect of adenine nucleotides may be in addition to the phosphorylation mechanism.

A) EVIDENCE FOR NKCC PHOSPHORYLATION AND COTRANSPORT ACTIVATION.  If ATP is not stoichiometrically consumed during NKCC activation, then why is its intracellular presence an absolute requirement for cotransport? Given the wide variety of protein-mediated biological activities that are regulated by protein phosphorylation and dephosphorylation (e.g., Ref. 184), it has always seemed likely that the NKCC protein must be phosphorylated to mediate cotransport.

There are several reports showing that stimuli that result in an increase of cotransport activity also cause an increase in the degree of phosphorylation of the NKCC protein (e.g., Refs. 119, 175, 212, 213, 254, 335, 341, 342). In at least two studies a temporal correlation has been demonstrated between transport activation and NKCC protein phosphorylation (212, 213). (In all these studies, the NKCC protein was identified using NKCC-specific antibodies.) These reports provide strong circumstantial evidence for the widely held view that the NKCC is activated by being phosphorylated. If the NKCC protein is only able to engage in ion transport when it is phosphorylated, then there ought to be a quantitative relationship between the degree of transport activation and the degree of phosphorylation caused by any given stimulus. At least three studies have measured both cotransporter activity and ³²P incorporation under the same conditions and reported that cotransport activity and NKCC protein phosphorylation increase in parallel. Haas et al. (119) measured the relative change in transepithelial ³⁶Cl flux across dog tracheal epithelial cells (see Table 5) and the relative change of ³²P incorporation into the NKCC protein (identified using the T4 monoclonal antibody, Ref. 221). They subjected the cells to a variety of stimuli including isoproterenol, UTP, hypertonic fluid, and a nystatin-induced change of [Cl]_i (see sect. IXB1A). Each of these treatments gave very nearly identical relative changes of transport and ³²P labeling. Similarly, O'Donnell et al. (254) studied bumetanide-sensitive ⁸⁶Rb uptake into endothelial cells as the measure of NKCC cotransport activity and ³²P incorporation into protein identified using the T4 monoclonal antibody. They showed that a 10-min exposure to hypertonic fluid (~1.3× normal osmolality) increased the bumetanide-sensitive ⁸⁶Rb uptake by 1.6-fold, whereas phosphoimage analysis of Western blots showed a 1.9-fold increase of ³²P incorporation. Lytle (212) used duck RBC to elegantly show that as cell volume is decreased by exposure to hyperosmotic fluids, the degree of NKCC transporter activity (measured as bumetanide-sensitive ⁸⁶Rb uptake) and the degree of phosphorylation of the NKCC protein (isolated using NKCC-specific antibody) increased in parallel (Fig. 16).

View this table: [in this window] [in a new window]   Table 5. Correlation of NKCC cotransport in tracheal epithelial cells with "intracellular" [Cl]

View larger version (24K): [in this window] [in a new window]   Fig. 16. Cell volume changes affect both cotransport activity and phosphorylation of NKCC protein in parallel. Duck red blood cells were labeled with 32P and exposed to a range of osmolalites ranging from 220 to 420 mosmol/kgH2O. NKCC protein was isolated by immunoprecipitation (using T14 antibody) and SDS-PAGE. Its 32P content was analyzed by Cerenkiv counting. Cell volume equals cell water content, and cotransport activity equals bumetanide-sensitive 86Rb influx. Normal cell volume is denoted by shaded region. [From Lytle (212).]

However, not all workers have been able to demonstrate a near one-for-one relationship between degree of ATP phosphorylation and activation of the NKCC. For example, Lytle and Forbush (213) used the shark rectal gland and measured NKCC transport activity indirectly, using specific labeling by [³H]benzmetanide. By varying the extracellular osmolality, they were able to change cell volume. They measured the effects of a range of cell volumes on both the specific [³H]benzmetanide binding and the content of ³²P in the 200-kDa band on an electrophoresis gel. Figure 17 shows that there is good qualitative agreement between these two variables. Quantitatively, however, there was a much greater relative increase of [³H]benzmetanide binding (~15-fold) than there was of total ³²P content (~4-fold). These workers also determined that the only phosphoamino acids they could detect were phosphoserine and phosphothreonine; no phosphotyrosine was found. For stimulation both by 20 µM forskolin or by exposure to an external fluid that was about 1.5 times normal osmolality, the phosphoserine and phosphothreonine content increased by ~3.5- and 5.7-fold, respectively. Given that [³H]benzmetanide binding is, at best, an indirect measure of cotransport activity, it may be that the degree of such binding and transport activity of the NKCC are not quantitatively related. Alternatively, it may be that measuring the phosphorylation of the 200-kDa protein included other, non-NKCC proteins. However, it should be noted that Lytle (212), using immunoblotting techniques, found that only the NKCC protein exhibited an increase of phosphorylation after cell shrinkage of duck RBC.

View larger version (22K): [in this window] [in a new window]   Fig. 17. Effects of osmolality on both [3H]benzmetanide binding and 32P labeling of NKCC protein. Osmolality of fluids bathing shark rectal gland cells was varied to cause a gradation of stimulation of NKCC (solid symbols). NKCC was maximally stimulated by applying 20 mM forskolin (). Normal cell volume was achieved at 915 mosM (see arrow on abscissa). Cell shrinkage occurred when osmolality was increased by adding sucrose. Cell swelling occurred when osmolality was reduced by NaCl removal. An inverse relationship was observed between external osmolality and cell water content [kg cell/kg cell solids = (3.237/mosM)  0.752]. Top: effect of osmotic changes on specific [3H]benzmetanide binding. These data represent means ± SE of 4 experiments. [Data adapted from Lytle and Forbush (214).] Bottom: effect of osmotic changes on 32P content of 195-kDa NKCC protein. Degree of phosphorylation was determined in 2 ways: scintillation counting of 195-kDa gel band or by autoradiography and densitometry. These data are representative of 3 experiments. [From Lytle and Forbush (213).]

Finally, it is important to note that not all workers have been able to show that all stimuli that increase NKCC-mediated fluxes cause an increase in phosphorylation. Tanimura et al. (339), working with rat parotid acini, were unable to demonstrate an increase of phosphorylation of the NKCC protein identified using an antibody when they stimulated fluxes using calyculin A or hypertonic shrinkage of the cells. Interestingly, when they stimulated fluxes with agents that increase cAMP concentration, they were able to detect increased protein phosphorylation. These authors suggested that there may be multiple pathways by which the NKCC could be stimulated.

B) WHAT PROTEIN KINASE(S) MIGHT BE INVOLVED?  Thus we have seen that protein phosphorylation is crucial to the activation of the NKCC1, and there is considerable evidence that the NKCC1 protein itself is phosphorylated when active (e.g., Refs. 212, 213, 288). Similar proof that the NKCC2 isoform is phosphorylated is lacking; thus the remainder of this discussion focuses on the NKCC1 isoform.

What protein kinase or kinases are involved in mediating this phosphorylation? As already discussed, the only residues on the NKCC1 that Lytle and Forbush (213) could identify as being phosphorylated were serine and threonine. Recent successes in cloning the NKCC1 (e.g., Refs. 53, 79, 283, 358) have permitted workers to deduce consensus phosphorylation sites from the predicted amino acid sequence of the NKCC1 protein. There are as many as 20 consensus sequences for kinases. Protein kinase C (PKC) and the protein kinase casein kinase II (CK2) consensus phosphorylation sites are found in all the transcripts so far isolated (53, 79, 283), whereas PKA consensus sites have been identified in all isoforms except shark (358). Of these three candidates, functional evidence has been presented to support claims for both PKA and PKC as mediators of NKCC function. Almost no functional evidence currently exists that addresses a possible role for protein kinase CK2.

Recently, Lytle (212) has argued that no matter what the stimulus, a single kinase is responsible for activating the NKCC via phosphorylation. Using duck RBC, he stimulated bumetanide-sensitive ⁸⁶Rb fluxes and measured phosphorylation in immunoprecipitated protein using a specific monoclonal antibody against NKCC1. Four different stimuli (osmotic cell shrinkage, norepinephrine, fluoride, and calyculin A) were all capable of increasing ⁸⁶Rb fluxes by 10-fold. All four stimuli exclusively phosphorylated serine and threonine sites in agreement with the results of Lytle and Forbush (213) mentioned above. As seen in Figure 16, when the NKCC-mediated flux was zero, there was still measurable phosphorylation. Phosphorylation increased with increasing stimulus in parallel with the increase of NKCC-mediated ⁸⁶Rb influx. Stoichiometrically, the baseline phosphorylation corresponded to 1 ATP/NKCC copy, and maximal stimulation corresponded to ~6 ATP/NKCC copy. Thus NKCC-mediated influx increased as a function of the degree of phosphorylation of each individual cotransporter rather than as a result of an increased number of cotransporters that were phosphorylated. There was no additivity to the effects of the four stimuli, and each of the cotransport stimuli caused the same increase in phosphorylation of the immunoprecipitated protein (Fig. 18). No matter which stimulus was applied, the same eight major tryptic phosphopeptides were generated. Finally, staurosporine, a broadly selective kinase inhibitor, inhibited each of the stimuli with equal potency, implying a common site of inhibition for all modes of stimulation (although leaving unanswered the question of the site of inhibition by this agent). Thus these three observations strongly point to a common kinase mechanism causing the NKCC protein phosphorylation that in turn causes the functional activation of the NKCC.

View larger version (38K): [in this window] [in a new window]   Fig. 18. Four different stimuli all increase cotransporter activity and NKCC protein phosphorylation, but their effects are not additive. Duck red blood cells were exposed to the following stimuli (all at their maximally effective doses): norepinephrine (NE), 10 mM; sodium fluoride (F), 10 mM; sucrose (hyper), 100 mM; calyculin A (cal), 0.2 mM. Cotransport activity (open bars) was measured as bumetanide-sensitive 86Rb uptake. NKCC protein phosphorylation (hatched bars) was measured as amount of 32P content of immunoprecipitated NKCC protein. All values were normalized to those of calyculin A. [From Lytle (212).]

At this time, workers have been studying regulation of the NKCC for at least 15 years. What we know for sure is that NKCC phosphorylation results in the functional activation of the cotransporter. We do not know the identity of the final, common kinase despite the fact that consensus sites for three different protein kinases have been identified in the putative cytoplasmic domain of the protein. We discuss below the conflicting, or nonexistent evidence for these three protein kinase candidates. It seems likely, therefore, that regulation of the functional activity by protein phosphorylation is going to prove to be complex and multifactorial. This should not be surprising. The postulated functions of this cotransporter are quite varied (see sect. X) requiring that the cotransporter receive input from a variety of stimuli. Only further experiments will tell us whether a final common kinase is responsible for NKCC phosphorylation.

I) PKA? As previously mentioned, the mammalian NKCC1 has consensus PKA sites (e.g., Refs. 53, 281), whereas the shark NKCC1 does not (358). Nevertheless, cAMP stimulates NKCC1 transport in the shark rectal gland. It is now believed that cAMP stimulates the shark NKCC1 indirectly, by increasing the electrodiffusive permeability of the apical membrane to Cl (104), thereby causing a net Cl secretory efflux and reducing [Cl]_i. As discussed in section IXB1, there is considerable evidence that reducing [Cl]_i increases NKCC cotransport activity and the degree of phosphorylation of the NKCC protein. The activating effect of reduced [Cl]_i is almost certainly not mediated by PKA as shown by Tanimura et al. (339). They immunoprecipitated a 175-kDa protein from rat parotid acini that binds bumetanide and is phosphorylated after treatment with isoproterenol, dibutyryl cAMP, and forskolin. However, treatment designed to reduce the [Cl]_i did not lead to such phosphorylation, leading the authors to argue the effect of cAMP must be direct (and not via a change of [Cl]_i).

Although the mammalian NKCC1 presumably does have PKA phosphorylation sites, the functional evidence for a direct role of PKA-mediated activation of cotransport function is decidedly mixed. For example, treatments designed to elevate cellular cAMP levels stimulate NKCC activity in avian erythrocytes (e.g., Refs. 121, 265, 347). However, this stimulation is not via PKA (212).

In contrast to the stimulation reported above, there are numerous reports of cAMP-induced inhibition of NKCC function. For example, cAMP has been reported to inhibit the NKCC function in several RBC types including human (82) and ferret RBC (263). Other tissues exhibiting NKCC inhibition after cAMP treatment include PC12 cells (201), flounder intestine (292), vascular smooth muscle (328), rat vascular smooth muscle (261), and human fibroblasts (260). In the case of the epithelial preparations, these inhibitory effects may be mediated, at least partially, by decreasing the membrane conductance to Cl (e.g., Ref. 292).

In still other cells, with well-defined NKCC mechanisms, cAMP has no discernible effect whatsoever (squid giant axon, Ref. 304; MDCK cells, Ref. 311; Ehrlich ascites tumor cells, Refs. 86, 145, 179; chick heart cells, Ref. 77; ferret RBC, Ref. 223).

The pattern (or lack thereof) of the forgoing results argues against a direct role for PKA in the activation of the NKCC1. However, it must be kept in mind that the precise NKCC isoform has not been identified in many of the preparations on which the functional studies outlined above were conducted. Thus it is possible that there are some isoforms that are directly phosphorylated by PKA and others that are not.

II) PKC? As already mentioned, there is structural evidence for consensus PKC binding sites on the NKCC1 protein in both mammalian and shark isoforms (53, 358). From a functional point of view, the situation for PKC playing a direct role in the phosphorylation-induced activation of the NKCC is very similar to that just described for PKA. There are reports of PKC stimulating NKCC activity (e.g., rat mesangial cells, Ref. 140; avian salt gland, Ref. 342), inhibiting NKCC activity (Balb/c 3T3 preadipose cells, Ref. 247; aortic endothelial cells, Ref. 251; HT-29 cells, Ref. 76), and having no effects (chick heart cells, Ref. 77; shark rectal gland, Ref. 323; avian RBC, Ref. 287; flounder intestine, Ref. 337; ferret erythrocytes, Ref. 275; eccrine clear cells, Ref. 343). In none of these studies can it be determined whether the functional effect being measured is due directly to PKC-mediated phosphorylation of the NKCC protein. The possibility of PKC exerting its effect on NKCC function distally via a multistep regulation seems to offer the best ad hoc explanation for the gamut of functional effects caused by PKC stimulators.

III) Other protein kinases?.  There are numerous consensus protein kinase CK2 phosphorylation sites on the NKCC protein (53, 79, 281, 358). This suggests a possible role for protein kinase CK2 as a modulator of NKCC activity. To date, little functional evidence exists regarding this potential pathway, although O'Donnell et al. (254) demonstrated that two inhibitors of protein kinase CK2 (A3 and CKI-7) both reduce the hypertonically induced stimulation of the NKCC of bovine vascular endothelial cells. Because these agents also inhibit myosin light chain kinase, the interpretation of this result is uncertain (see Refs. 175, 254).

Two groups have reported that ML-7, an inhibitor of myosin light chain kinase, will prevent the shrinkage-induced stimulation of the NKCC found in bovine aortic endothelial cells (175, 254). Klein and O'Neill (175) further demonstrated that cell shrinkage resulted in increased phosphorylation of both the myosin light chain as well as the NKCC (identified using the T4 antibody). As expected, ML-7 inhibited the shrinkage-induced phosphorylation of the myosin light chain but, surprisingly, had no effect on the phosphorylation of the cotransporter. Thus the inhibitory effect of ML-7 on NKCC function does not seem to be the result of a reduction of the phosphorylation of the NKCC but may be mediated through a reduction of the phosphorylation of the myosin light chain or another as-yet-unidentified protein.

At present, it is safe to say that if a single protein kinase is responsible for the phosphorylation and subsequent functional activation of the NKCC protein, it either must be different in different cells or it has not yet been identified.

C) EVIDENCE FOR PROTEIN PHOSPHATASE INVOLVEMENT.  A protein phosphorylation/dephosphorylation scheme for NKCC activation/deactivation implies an important role for protein phosphatases in addition to protein kinases. Reports from studies on a variety of cells have provided evidence that protein dephosphorylation reduces NKCC functional activity (e.g., Refs. 5, 6, 176, 201, 212, 254, 267, 287, 335). If the NKCC is constitutively fully activated when the phosphatase inhibitor is added, then no effect is observed (e.g., squid axon, Ref. 7). To demonstrate the action of protein phosphatase, intracellular ATP was washed out (using intracellular dialysis). The rate at which the bumetanide-sensitive unidirectional ³⁶Cl influx was inactivated in the presence and absence of phosphatase inhibitors was measured (5). In this way, the effects of intracellular fluoride (5 mM) and vanadate (40 µM) were studied. Neither agent affected the rate of ³⁶Cl influx in the presence of intracellular ATP (continuously supplied via internal dialysis). Therefore, it was concluded that neither agent had direct effects on the NKCC itself nor did they appear to interfere with the process by which the NKCC is activated by ATP. However, if either agent was present during the washout of cellular ATP, the inactivation of bumetanide-sensitive ³⁶Cl influx was slowed by a factor of two- to threefold. Another means of demonstrating protein phosphatase activity in the squid axon was to reduce the [ATP]_i to levels near the apparent half-saturation constant for the nucleotide (~75 µM; Ref. 5) and then apply an inhibitor of protein phosphatases. Using a more specific protein phosphatase inhibitor, okadaic acid (151), Altamirano et al. (6) provided further evidence that the NKCC in the squid axon is inactivated by protein phosphatase. ATP "washout" studies using okadaic acid as the protein phosphatase inhibitor confirmed the results originally reported using vanadate and fluoride (Russell, unpublished observations). If, indeed, the action of the protein phosphatase inhibitors was to prevent the dephosphorylation of the "activated" NKCC, one might reasonably ask why, in the presence of high concentrations of okadaic acid, was there inactivation at all? To address this question, we must consider the following. First is the fact that these phosphatase inhibitors do not block all classes of phosphatases (21, 151). This, coupled with the fact that phosphoprotein phosphatases are not particularly substrate specific, means that dephosphorylation of the activated NKCC could continue to occur by other phosphatases not blocked by the particular agent being used. In addition, because phosphorylated proteins are relatively "energy rich" (e.g., Ref. 184), spontaneous dephosphorylation will eventually inactivate the cotransporter. Therefore, it is not surprising that agents that are believed to be acting to block phosphoprotein phosphatases might not totally prevent the inactivation of the NKCC. A similar lack of complete inactivation of an ATP-dependent Ca²⁺ channel in heart muscle has been reported after treatment with protein phosphatase inhibitor 2 (133).

Most other cell types that have been studied for protein phosphatase activity appear to have the NKCC inactivated under "resting" conditions. Pewitt et al. (287) studied the effects of okadaic acid on ⁸⁶Rb influx into duck RBC. They reported that in otherwise unstimulated erythrocytes, okadaic acid treatment would increase the bumetanide-sensitive ⁸⁶Rb uptake rapidly (within 4-5 min) with an ED₅₀ of ~5 × 10⁷ M. Similar effects have been reported for the NKCC of PC12 cells (201), bovine aortic endothelial cells (254), and brain microvessel endothelial cells (335) using okadaic acid and/or calyculin A.

Two studies have reported a large difference in the sensitivity of the NKCC-mediated flux stimulation to these two protein phosphatase inhibitors. In both PC12 cells (201) and duck erythrocytes (267), it takes about a 100 times higher concentration of okadaic acid than of calyculin A to cause maximal stimulation of the bumetanide-sensitive flux. Given the relative selectivities of these two agents for protein phosphatase 1 and 2A (I151), this difference in sensitivities of the NKCC has been taken to mean that protein phosphatase 1 is more important than protein phosphatase 2A in dephosphorylating the NKCC. An interesting and, from an operational point of view, important observation has been that both these inhibitors exhibit a biphasic dose response (201, 335). Thus, if too high a concentration of either agent is used, the stimulation of the cotransported ion flux is reduced.

In addition to the functional evidence that protein phosphatases may be involved in controlling the level of activity of the NKCC, there has recently been some structural evidence as well. O'Donnell's group (254) has demonstrated that the NKCC protein (identified with a monoclonal antibody) from endothelial cells has a much higher level of phosphorylation when otherwise unstimulated cells are exposed to either okadaic acid or calyculin A. Using duck RBC, Lytle (212) has recently shown that treatment with calyculin A stoichiometrically increased both cotransport flux and phosphorylation of the NKCC.

Recently, Flatman and Creanor (71) have postulated that the protein phosphatase responsible for dephosphorylating (and thereby inactivating) the NKCC protein itself must be dephosphorylated to be active. They then suggest that the effects of a variety of protein kinase inhibitors are exerted not on the final common kinase but on other protein kinases responsible for phosphorylating (and inactivating) the protein phosphatase. Thus future work may well reveal that regulation of NKCC function is accomplished by a variety of cascades of protein kinases and phosphatases.

In the internally dialyzed squid giant axon, elevation of [Cl]_i inhibits the unidirectional fluxes of all three cotransported ions in both the influx and efflux directions (28, 29, 302, 303, 306, 307). These findings grew out of an early observation that removal of intracellular Cl had the surprising (at the time) effect of increasing ³⁶Cl influx (302). This was surprising because it was expected that a substantial portion of the unidirectional ³⁶Cl influx would occur via isotopic exchange flux pathways, in which case reducing [Cl]_i would reduce the unidirectional ³⁶Cl influx. Later, as the pathway for ³⁶Cl influx was being identified as the NKCC, it was shown that in addition to inhibiting the influx of ³⁶Cl, high levels of [Cl]_i also inhibited the influxes of ²²Na (303) and ⁴²K (306). Figure 19 illustrates the profound and reversible effect of changing the [Cl]_i on simultaneously measured ³⁶Cl and ⁴²K influxes into an internally dialyzed squid axon. The inhibitory effect of elevated intracellular Cl was not observed when the axons were pretreated with furosemide or bumetanide or when ATP was removed by internal dialysis, showing it was a specific effect on the NKCC.

View larger version (16K): [in this window] [in a new window]   Fig. 19. Effect of removing intracellular Cl (replaced isosmotically with glutamate) on unidirectional influxes of both Cl () and K+ () measured simultaneously into an internally dialyzed squid giant axon. Throughout experiment, intracellular fluid was Na+ free. Although not shown here, stimulation of both influxes seen upon removal of intracellular Cl is not observed in presence of 10 mM bumetanide. SSW, squid seawater. Ouabain concentration, 105 M; tetrodotoxin concentration, 107 M. [From Russell (307).]

Intracellular Cl is not the only, or even the most, effective anion at inhibiting the NKCC. Thiocyanate, I, and Br are all more potent inhibitors than Cl, whereas NO₃ and SO₄² exhibit significant but much less inhibitory potentcy (29). Because SCN, NO₃, and SO₄² alone are not suitable substrates for the NKCC (e.g., Refs. 234, 260), this suggests that the site(s) being affected by the intracellular anions may not be transport sites. Further evidence that the intracellular site(s) may not be the same sites that Cl binds to for cotransport is the fact that extracellular Cl has very little inhibitory effect (29). Nevertheless, the exact site of inhibition by intracellular Cl is still unknown.

Because the binding and release of substrate ions is believed to be highly ordered (see sect. VIIA), one possibility is that raising the trans-side concentration of a substrate could inhibit the reaction by an end-product inhibition mechanism. Although this mechanism may partially account for the observed inhibition by intracellular Cl, there are at least three lines of evidence that argue against this as the sole explanation. First, as mentioned above, there is the fact that some nontransported anions are even more effective at inhibition when presented intracellularly than is Cl. However, it is possible that nontransported ions bind at the transport sites but do not promote or support cotransport. Second, if the mechanism is simple trans-side inhibition, then removal of extracellular Cl ought to stimulate NKCC-mediated ion efflux. This was not what actually happened when Cl (normally 561 mM) was completely replaced by gluconate in the external fluid bathing the squid axon. In this case, there was only a small increase of cotransport efflux (29). Third, and perhaps the strongest argument, is the fact that raising [Cl]_i not only blocks bumetanide-sensitive influxes in a concentration-dependent manner but also the bumetanide-sensitive effluxes (29). Figure 20 shows the effects of increasing [Cl]_i on both NKCC-mediated influx and efflux in the internally dialyzed squid axon. In the case of the bumetanide-sensitive Cl influx into the squid axon, the relationship between [Cl]_i and influx is rather steep, with a Hill coefficient of 4.5. This suggests multiple sites of action by intracellular Cl. The relationship between [Cl]_i and bumetanide-sensitive efflux is complex, since for efflux, intracellular Cl is both a substrate and an inhibitor. The efflux versus [Cl]_i data cannot be fitted by simply assuming that there are cooperative Cl binding sites for transport and another set of cooperative binding sites for inhibition. Note that the steepest region of the intracellular Cl inhibition is in the [Cl]_i range of normal intracellular Cl levels (normal [Cl]_i for squid axons is ~125 mM; reviewed in Ref. 307). This suggests that [Cl]_i plays an important negative-feedback role in controlling the activity of the NKCC. Because the intracellular ionic conditions for these influx and efflux experiments were different (e.g., efflux experiments required intracellular Na⁺, which was omitted from the intracellular fluids for the influx studies), quantitative comparisons of the two unidirectional fluxes are not useful.

View larger version (12K): [in this window] [in a new window]   Fig. 20. Effect of varying [Cl]i on bumetanide-sensitive unidirectional influx () and unidirectional efflux () of Cl measured using 36Cl in internally dialyzed squid axons. [Cl]i was adjusted by dialyzing with fluids identical in all respects except a varying amount of Cl was replaced with glutamate. A nonlinear least-squares analysis of Hill equation was fitted to influx data. Fitting parameters were as follows: Vmax = 56.6 ± 1.4 pmol·cm2·s1; Hill n = 4.5 ± 0.3; K0.5 = 101 ± 2.0 mM. Efflux data are fitted by eye. [K+]o was 100 mM (for influx and efflux studies), which promotes a larger influx without affecting efflux. For influx studies, [K+]i was 400 mM and [Na+]i was 0 mM. For efflux studies, [K+]i was 200 mM and [Na+]i was 200 mM. All other ionic conditions both inside and outside were identical for two sets of data. External fluid contains ouabain (105 M) and tetrodotoxin (107 M). Each point represents data averaged from no less than 3 axons and in most cases 6 axons. [Influx data from Breitwieser et al. (28); efflux data from Breitwieser et al. (29).]

As early as 1982, Ussing (348) reported the results of studies on the recovery of cell volume after exposure of the blood side of the frog skin to reduced [K⁺] or [Cl] solutions. He reasoned that exposure to KCl-depleted solutions caused the cells to lose K⁺ and Cl and an osmotic equivalent of water. He showed that the recovery process was furosemide sensitive and proposed that it was via a coupled NCC process. Based on his assumption that the cell shrinkage resulted from a loss of (K⁺) Cl, Ussing (348) suggested that the cotransporter might be activated by a fall of [Cl]_i, although no definitive proof was offered.

The remarkable effect of intracellular Cl on the NKCC has recently attracted a good deal of attention as the implications have become better understood. For example, there are a number of cell types which, when shrunken by direct exposure to hyperosmotic fluids, do not perform regulatory volume increase (RVI; see Table 6). However, several will perform a pseudo-RVI if they are first swollen by exposure to hypotonic fluids, then permitted to undergo regulatory volume decrease (RVD), which they do by losing K⁺ and Cl. This has been termed a "post RVD RVI" (134). Levinson (202-204) pursued this observation using Ehrlich ascites tumor cells and showed that the subsequent RVI was the result of a bumetanide-sensitive net uptake of K⁺, Na⁺, and Cl that always results in an increase of intracellular Cl content. Thus, when [Cl]_i recovered, cell volume recovered, suggesting that somehow [Cl]_i was playing an important role in regulating the activity of the NKCC. An insightful observation made in the discussion of the 1990 paper (203) was that, although the gradient of the chemical potential (_net) for the NKCC favored net uptake by the NKCC, such net uptake did not occur until the [Cl]_i was reduced by the prior RVI-driven loss of Cl. This led Levinson (203) to speculate that the activity of the NKCC might be "somehow regulated by the intracellular [Cl] and not by the gradient of chemical potential." Similarly, studies on shark rectal gland (104) and vascular endothelial cells (257) revealed that there are situations in which when the net free energy in the chemical gradients of Na⁺, K⁺, and Cl favors net uptake by the NKCC, it does not occur. O'Neill and Klein (257) suggested, based on the report of trans-inhibition in squid axon (see above), that intracellular Cl may play a regulatory role in the operation of the vascular endothelial cell NKCC.

Homma and Harris (141) showed that soaking cultured rat glomerular mesangial cells in Cl-free or Na⁺-free media for 60-90 min resulted in enhanced cotransport flux and [³H]bumetanide binding. Bumetanide-sensitive ⁸⁶Rb uptake was increased between 55 and 70% by such treatment. [³H]bumetanide binding was increased 50-60%. Similarly, in a study using the osteosarcoma cell line UMR-106-01, Whisenant et al. (351) showed that presoaking the cells in a Cl-free medium for 30 min resulted in a 2.5-fold increase in bumetanide-sensitive ⁸⁶Rb uptake. These workers had previously shown that such a pretreatment depleted the cells of Cl. The preceding results support the view that reduced [Cl]_i stimulates transport by the NKCC. Finally, it has been noted that to study the functional expression of cloned NKCC, it is necessary to soak HEK-293 cells in Cl-free medium, presumably to reduce [Cl]_i (e.g., Refs. 283, 358).

Recently, Gillen and Forbush (93) restudied the issue of [Cl]_i and NKCC activity using an epithelial cell line (HEK-293 cells) transfected with human NKCC1. They measured both bumetanide-sensitive ⁸⁶Rb uptake as well as [Cl]_i. As was the case for squid axon, they found a very steep relationship between bumetanide-sensitive ion flux and [Cl]_i. They likewise showed that at the resting [Cl]_i level, the NKCC-mediated uptake was very low. Both these observations nicely confirm in another cell type earlier reports for the key importance of [Cl]_i to the functional activity of the NKCC.

A) [CL]_i AND REGULATION OF TRANSEPITHELIAL SECRETION: THE [CL]_i-COUPLING HYPOTHESIS.  Secretory epithelial organs (e.g., salivary glands, tracheal epithelium, shark rectal glands) secrete a Cl-rich aqueous solution in response to a variety of stimuli. Secretion requires the transepithelial movement of Cl (plus an attendant cation, usually Na⁺, and osmotically obliged water) from the blood to the lumen of the gland. This means that the Cl must enter the cell across the blood-side membrane (the basolateral membrane) via the NKCC, diffuse through the epithelial cytoplasm, then cross the apical membrane, via Cl channels, into the lumen of the gland. The attendant cation follows either via a channel (transcellular) or across the tight junction (paracellular). Thus secretion requires the coordination of not less than four ion transport pathways: at least two Cl transport pathways located on two different membranes as well as a basolateral K⁺ channel and an apical Na⁺ (or K⁺) channel (5 if you count the Na⁺/K⁺ pump). In some cases, the direct secretagogue action is apparently only at the apical membrane. Here the effect is to cause Cl channels to open, thereby permitting Cl to run down its electrochemical gradient, out of the cell and into the secretory organ lumen (e.g., cAMP-mediated agonists in shark rectal gland, Refs. 100, 104; purinergic agonists in the tracheal epithelium, Refs. 120, 224).

In 1984, two studies appeared that, in retrospect, provided the first direct evidence that [Cl]_i might be the link between the Cl channel on the apical membrane and the NKCC on the basolateral membrane. Both Shorofsky et al. (322) and Greger et al. (104) used Cl-sensitive liquid ion exchanger microelectrodes to monitor [Cl]_i. (Actually, these microelectrodes measure ion activity a_i^Cl, but no serious error is made if we use the more familiar term [Cl]_i.) They used two different secretory epithelia, but both clearly demonstrated that resting [Cl]_i was much higher than expected if Cl were distributed across the membrane according to electrochemical equilibrium considerations. Furthermore, they both showed that when the cells were stimulated to secrete, [Cl]_i declined significantly. Furthermore, Greger et al. (104), using the shark rectal gland, showed that the increase in the Cl conductance of the apical membrane appeared to precede the fall of [Cl]_i.

Opening apical Cl channels would be expected to reduce [Cl]_i. Could a reduction of [Cl]_i work by simply increasing the net chemical driving force on the NKCC? As discussed in section VIB, the net driving force on the cotransporter is a function of the transmembrane chemical gradients of Na⁺, K⁺, and Cl. Greger et al. (104) monitored the intracellular ion activities of Na⁺, K⁺, and Cl before and during cAMP stimulation using ion-selective microelectrodes. They showed that the cAMP treatment that increased the transepithelial Cl flux by 10- to 30-fold increased the [Na⁺]_i from 11 to 29 mM while it decreased the [Cl]_i from 49 to ~40 mM and left [K⁺]_i unchanged. Using these ion activity measurements and the net driving force equation (see Table 3), it is clear that the 10- to 30-fold increase in furosemide-sensitive Cl flux cannot be explained by the change of net driving force on the NKCC represented by the changes of ion concentrations that these workers measured. In the discussion of their paper, Greger et al. (104) considered at least six possible links between the increase of the apical membrane Cl conductance and the increased NKCC rate. One of the possibilities was that reducing the normally high [Cl]_i itself might act to stimulate the cotransporter, an effect which had already been clearly demonstrated in the squid axon (see sect. IXB1).

Forbush's group (213, 214) has termed this interaction between [Cl]_i and epithelial secretion, the "[Cl]_i-coupling hypothesis." Thus Lytle and Forbush (213; see also Ref. 72) explicitly postulated that a reduction of [Cl]_i, resulting from Cl loss through the activated apical Cl channels, was the key to the apparent "cross-talk" between different transport mechanisms located on the two separate membranes. Furthermore, they have suggested that upregulation of transport by the NKCC results from an enhanced phosphorylation of the NKCC protein that in some way is caused by the reduced [Cl]_i (213). Figure 21 illustrates the relevant anatomical considerations and the Cl transport pathways involved in this hypothesis. Results obtained from several epithelial preparations directly (115, 119) and indirectly (120, 216, 296) support this [Cl]_i coupling hypothesis.

View larger version (19K): [in this window] [in a new window]   Fig. 21. Possible role for [Cl]i in regulation of secretion by a secretory epithelium. The "[Cl]i coupling hypothesis" is as follows: 1) agonist causes opening of a Cl channel on apical membrane of epithelial cell. 2) [Cl]i, which is normally maintained above its electrochemical equilibrium concentration (see sect. IXA) by NKCC, now falls as a result of increased apical membrane conductance. 3) Fall of [Cl]i reduces [Cl]i-mediated inhibition of NKCC (see Fig. 20), possibly by permitting an increase in phosphorylation of NKCC. 4) K+ recycles via a K+ channel in same membrane in which NKCC is located. 5) Na+ follows Cl via a paracellular pathway. See text for details.

Haas and co-workers (115, 119, 120) have performed a series of studies on airway epithelial cells that provide the best direct evidence in support of the [Cl]_i coupling hypothesis in epithelia. First, they showed that secretagogue-stimulated NKCC activity (measured 2 ways) could be reduced by treatments expected to prevent or reduce the loss of Cl across the apical membrane. In one study, they measured saturable [³H]bumetanide binding to dog tracheal epithelial cells as an index of NKCC activity (120; see sect. VIII for a discussion of this approach). As seen in Figure 22, the apical Cl channel inhibitor IAA-94 (94) reduced or abolished the secretagogue-stimulated saturable [³H]bumetanide binding. When ³⁶Cl efflux was used as a measure of NKCC activation, IAA-94 blocked 75% of the cAMP-stimulated ³⁶Cl flux and 69% of the ATP-stimulated flux. These results support the [Cl]_i coupling hypothesis by demonstrating that two different measures of NKCC activity are reduced when the apical Cl channels are inhibited. Presumably, such inhibition of the apical Cl channels would prevent a fall in [Cl]_i and thereby prevent the activation of the NKCC.

View larger version (28K): [in this window] [in a new window]   Fig. 22. Effects of IAA-94 (an inhibitor of apical membrane-located Cl channels) on saturable binding of [3H]bumetanide to confluent cultures of dog tracheal epithelial cells (grown on permeable supports) caused by various stimulators of transepithelial Cl secretion. The following concentrations of secretagogues were used: isoproterenol, 5 × 106 M (basolateral medium only); ATP and UTP, 10 µM (apical membrane only). aSignificantly different from appropriate control (with or without IAA-94) at P = 0.05 level by Student's t-test. bSignificantly different from identical condition without IAA-94 (P < 0.05). [Modified from Haas et al. (120).]

Increasing [K⁺] on the basolateral side of the membrane ([K⁺]_b) is an alternative means to reduce or inhibit Cl efflux across the apical membrane. This reduces the recycling of K⁺ via basolateral K⁺ channels, thereby reducing the cell-negative intracellular voltage, which is the driving force for Cl to leave the cell through the apical channel. The effects of raising [K⁺]_b on [³H]bumetanide binding were tested in dog and human tracheal epithelial cells treated with isoproterenol or with UTP. The pattern of effects was similar to that observed with IAA-94 treatment. Elevated [K⁺]_b prevented the extra [³H]bumetanide binding caused by UTP, but not that caused by isoproterenol. Thus these results suggested that NKCC activation caused by ATP/UTP depended on the movement of Cl out of the cell across the apical membrane, whereas the stimulation by isoproterenol/cAMP was only partially accounted for by this mechanism. Because Cl loss via the apical channels might result in a reduction of [Cl]_i and because the stimulatory effects of reduced [Cl]_i in the squid axon were known (e.g., Ref. 28), Haas and co-workers suggested that the link between apical Cl channel activation and NKCC activation might be the reduced [Cl]_i.

The [Cl]_i-coupling hypothesis specifies that it is the reduction of [Cl]_i that leads to the activation or upregulation of the NKCC. Thus a direct test requires experiments in which [Cl]_i is either controlled or measured and changed independently of other factors known to stimulate Cl secretion. Haas and McBrayer (115) achieved control of [Cl]_i by applying nystatin to the apical membrane of a confluent monolayer of dog tracheal epithelial cells. Nystatin increases the membrane permeability to small, univalent ions, especially cations (39, 206, 308), thereby giving some control over [Cl]_i (but causing changes in the intracellular concentrations of Na⁺ and K⁺ at the same time). The secretagogue effects of isoproterenol and UTP on transepithelial fluxes were measured under two conditions. In one case, the [Cl]_i was uncontrolled (without nystatin), and in the other case, [Cl]_i was nystatin-clamped to the level of the [Cl] in the apical media (~125-130 mM). Figure 23 compares the effects of isoproterenol and UTP in the absence or presence of nystatin treatment on monolayers. When [Cl]_i changes were prevented by "nystatin clamping," UTP no longer stimulated ³⁶Cl transepithelial fluxes. Thus qualitatively, if not quantitatively, these results agree with those reported earlier using IAA-94 or high [K⁺]_b and measuring [³H]bumetanide binding and ³⁶Cl fluxes. In the most direct test to date of the [Cl]_i-coupling hypothesis, Haas and McBrayer (115) measured the bumetanide sensitivity of the transepithelial ³⁶Cl flux across confluent epithelia whose "intracellular" [Cl] was varied over the range of 124, 66, and 32 mM by applying nystatin to the apical membrane in media containing the test [Cl]. Figure 24A shows that when [Cl]_i was reduced, the total ³⁶Cl flux was increased, whereas the bumetanide-insensitive flux was essentially unaffected. When the difference between these two fluxes, which is the bumetanide-sensitive ³⁶Cl flux, is plotted against the nominal [Cl]_i in Figure 24B, a relatively steep relationship is revealed. This relationship is well-fitted by a Hill function when the Hill coefficient is set to 2. 

View larger version (34K): [in this window] [in a new window]   Fig. 23. Effect of [Cl]i on response of apical-to-basolateral 36Cl fluxes across confluent cultures of dog tracheal epithelial cells to isoproterenol and UTP. Control cells had normal [Cl]i (66 mM, assuming an activity coefficient of 0.72; see Ref. 212). Exposure of apical membrane to nystatin has effect of "clamping" [Cl]i to [Cl] of fluid bathing apical face of epithelium, in this case, 132 mM. Isoproterenol (5 mM) and UTP (10 µM) were presented to apical membrane. Each control bar represents 8 determinations; each isoproterenol bar and each UTP bar represents 4 determinations each. aSignificantly different from appropriate control (with or without nystatin) by Student's t-test. bSignificantly different from same condition without nystatin. [From Haas and McBrayer (115).]

View larger version (22K): [in this window] [in a new window]   Fig. 24. Apparent effect of [Cl]i on bumetanide-sensitive 36Cl flux across confluent monolayers of dog epithelial cells grown on permeable supports. A: primary data from Reference 115 showing effects of varying [Cl] of nystatin-containing media bathing apical membrane on bumetanide-sensitive 36Cl flux. Circles, 124 mM [Cl]; triangles, 66 mM [Cl]; squares, 32 mM [Cl]; open symbols, bumetanide treated. B: bumetanide-sensitive transepithelial 36Cl flux, taken from primary data in A plotted against [Cl] in apical membrane bathing media. Line is drawn using Hill equation with Hill coefficient set to 2.0. [A adapted from Haas and McBrayer (115).]

Could the UTP effects on transepithelial Cl flux illustrated in Figure 24 be explained entirely on the basis of a channel-mediated reduction of [Cl]_i? Examination of Figure 24 shows that the total ³⁶Cl transepithelial flux (in the absence of nystatin) is increased about 2.3-fold; this is close to the 2.64-fold stimulation reported by Haas et al. (119) for the same treatment. Because these latter measures of UTP stimulation used total ³⁶Cl flux (including the bumetanide-insensitive flux), the stimulation of the NKCC-mediated flux (the bumetanide-sensitive flux) is likely to be somewhat larger. According to the relationship illustrated in Figure 24 to achieve a 2.5-fold stimulation of bumetanide-sensitive ³⁶Cl flux by reducing [Cl]_i from its resting level of [Cl]_i equals 60 mM, [Cl]_i must fall to ~11 mM. Shorofsky et al. (322), using Cl-sensitive microelectrodes, measured the decrease of intracellular Cl ion activity after exposure to epinephrine and found it decreased from 47.2 to 32.2 mM. Using an activity coefficient of 0.78, this converts to a [Cl]_i decrease from 60.5 to 41.3 mM. Thus, as we saw earlier in this section for cAMP-stimulated fluxes in shark rectal gland, the actual decrease of [Cl]_i is much less than apparently required according to the relationship in Figure 24. There are at least two major reasons for this apparent discrepancy between the data and the prediction of the [Cl]_i-coupling hypothesis. First, Haas and co-workers have shown that the stimulatory effects of -adrenergic agonists such as isoproterenol and epinephrine do not depend entirely on a fall of [Cl]_i to stimulate transepithelial fluxes. Second, it should be pointed out that the conditions of the experiments whose results were used to generate the relationship illustrated in Figure 24 were such that as [Cl]_i was being varied, there was a concomitant and substantial increase of [Na⁺]_i as well as a depletion of [K⁺]_i occurring. Because it has been shown that an increase of [Na⁺]_i will inhibit the NKCC (29, 351; see sect. IXB2), it is likely that this figure underestimates the actual [Cl]_i sensitivity of the NKCC of the dog tracheal epithelial cell.

The possibility that [Cl]_i might serve as a proximate signal for the activation of the NKCC was also addressed by Foskett and co-workers (74, 75, 296) using another secretory epithelial preparation, the rat salivary acinar cell. The results of these studies by Foskett and co-workers taken all together are consistent with a fall of [Cl]_i being necessary (but not sufficient) for the coordinated activation of at least three Na⁺ uptake pathways (74, 75, 296).

In a study that tied together several seemingly disparate lines of evidence regarding muscarinic stimulation of salivary secretion, Robertson and Foskett (296) measured net Na⁺ uptake into single salivary cells as a measure of the combined activities of the NKCC, the Na⁺/H⁺ exchanger (NHE), and a nonselective cation channel. In this preparation, carbachol elevates [Ca²⁺]_i, which in turn activates apical Cl channels and basolateral K⁺ channels (cf. Fig. 21). It was known that secretion by the salivary glands can be inhibited by bumetanide and by amiloride analogs. It was further known that the activities of the NKCC and the NHE were stimulated by carbachol (243). The authors proposed that a fall of [Cl]_i is critical to the activation of all three of these Na⁺ transporting pathways. They showed that raising [K⁺]_o, which would be expected to reduce the loss of K⁺ and Cl and hence prevent a fall of [Cl]_i, prevented the increase of [Na⁺]_i which ordinarily is observed upon application of carbachol. This result is consistent with their hypothesis, but a complication is that carbachol also causes the salivary cells to shrink, and cell shrinkage is a well-known stimulant for both the NKCC and the NHE. Thus it was necessary to show that the shrinkage was not the proximate signal to stimulate these Na⁺ transporters. To address this, the authors performed a clever study in which they prevented the carbachol-induced cell shrinkage by simultaneously exposing cells to carbachol and a gradually reducing extracellular osmolarity such that the tendency to shrink caused by K⁺-Cl loss was just offset by the extracellular hypotonicity. Figure 25 shows the results of one of these experiments. It shows that even when cell volume is unchanged, the presumed fall of [Cl]_i caused by the loss of KCl is sufficient to stimulate Na⁺ uptake. Although these results support the [Cl]_i coupling hypothesis, they add two new pieces of information. First, in this cell type, the fall of [Cl]_i stimulates not only NKCC activity, but also stimulates NHE-mediated transport. In contrast, others have reported that increases of [Cl]_i stimulate NHE-mediated transport (e.g., Refs. 138, 278). It is possible the different results reflect different isoforms of the NHE. In fact, the contribution of the NKCC to the measured variable ([Na⁺]_i) is relatively small, being only ~15% of the total measured Na⁺ uptake. The second important point is that Foskett's group (356) has shown that without a rise in [Ca²⁺]_i, treating the cell in ways that would be expected to reduce [Cl]_i does not stimulate these Na⁺ transport pathways. This suggests an as yet unknown key role for Ca²⁺.

View larger version (14K): [in this window] [in a new window]   Fig. 25. Demonstration that carbachol stimulation of Na+ uptake by rat salivary acinar cells (, bottom line) does not require cell shrinkage (, top line). Cell was surperfused with a solution that contained 10 µM carbachol and whose osmolarity was reduced by reduction of NaCl concentration to 50 mM. Reduction in tonicity was achieved gradually by continual perfusion of various gradations of hypotonicity to maintain a steady-state volume. Thus osmotic water loss (osmotically tied to net KCl loss) that normally accompanies carbachol treatment is just offset with hypotonically induced water gain. Ouabain (1 mM) was present to ensure that change of [Na+]i was reflecting changes of Na+ uptake. Other experiments showed that this Na+ uptake occurred through three pathways: NKCC (~30%), Na+/H+ exchanger (~55%), and a nonselective cation channel (~15%). Thus cell volume per se is not a signal to stimulate Na+ uptake pathways. These data are from a single cell but are representative of 20 similar cells. V/Vo = cell volume (V) normalized to initial cell volume (Vo). [From Robertson and Foskett (296).]

B) [CL]_i AND PHOSPHORYLATION OF THE NKCC.  This brings us to a consideration of how a reduction of [Cl]_i might exert a regulatory effect on the NKCC. Lytle and Forbush (213) showed that activation of the NKCC of the shark rectal gland by either cAMP-dependent secretagogues (such as vasoactive intestinal polypeptide, adenosine, or forskolin) or by cell shrinkage promoted the phosphorylation of the cotransport protein at serine and threonine residues. They also demonstrated that treatment with some protein kinase inhibitors could prevent activation of the NKCC (measured as [³H]benzmetanide binding). They suggested that [Cl]_i affected the degree of phosphorylation (and hence, activity) of the cotransporter.

Haas et al. (119) directly tested this part of the [Cl]_i coupling hypothesis using nystatin-treated tracheal epithelial cells. As discussed in section XB1A, the application of nystatin to the apical membrane permits control over [Cl]_i in the absence of changes in cell volume. In a series of parallel studies examining either transepithelial ³⁶Cl flux or ³²P incorporation into the NKCC protein (identified via immunoprecipitation), these workers demonstrated a parallel increase of transport activity and NKCC protein phosphorylation as the [Cl]_i was reduced.

Lytle and Forbush (216) examined the effects of reduced [Cl]_i in isolated shark rectal gland tubules on either [³H]benzmetanide binding (as a measure of NKCC activity) or ³²P incorporation into the NKCC protein (identified by immunoprecipitation). Two treatments were used to reduce [Cl]_i, which according to the hypothesis should upregulate [³H]benzmetanide binding and stimulate phosphorylation of the NKCC cotransporter. They showed that reducing the Cl concentration of the fluid bathing the tubules resulted in increased [³H]benzmetanide binding and increased the degree of phosphorylation of the NKCC. Because external Cl replacement by an impermeant anion (gluconate) would be expected to result in a net loss of Cl plus a cation (to maintain macroscopic electroneutrality), these cells may well have been shrunken. Cell shrinkage per se is known to stimulate both [³H]benzmetanide binding and NKCC phosphorylation (120, 160, 214). Another means of reducing [Cl]_i was to reduce [Na⁺]_o (e.g., Ref. 102). Again, they noted that a treatment designed to lower [Cl]_i resulted in a parallel increase of [³H]benzmetanide binding and NKCC phosphorylation. But again there is the possibility that these cells might shrink as a result of NaCl loss.

A rise of [Cl]_i has been shown to inhibit fluxes through the NKCC (e.g., squid axon, Refs. 28, 29; HEK-293 cells, Ref. 93). Lytle and Forbush (216) raised [Cl]_i by raising the K⁺ concentration of the media bathing tubules from the shark rectal gland. This treatment would be expected to raise [Cl]_i as a result of two actions: by stimulating the NKCC on the basolateral membrane to promote Cl uptake and by reducing Cl loss across the apical membrane through a reduction of the V_m and hence the electrochemical driving force on Cl (i.e., V_m  E_Cl). They observed that this treatment reduced [³H]benzmetanide binding in cells treated with forskolin. The inhibitory effect required the presence of extracellular Cl, consistent with the idea that it is caused by an increase of [Cl]_i. However, a net increase of [Cl]_i, along with a presumed increase of [K⁺]_i, might also be expected to increase cell volume, and the NKCC has been shown to be inhibited by cell volume increases (see sect. XC). In fact, the authors provide data which show that the high [K⁺]_o-treated cells did, in fact, swell. However, it should be pointed out that exposure to hypotonic media, which also causes the cells to swell, was shown to stimulate [³H]benzmetanide binding (214). This result might be explained by a decrease of [Cl]_i that must inevitably occur when a cell swells (but see sect. XC). However, Pewitt et al. (286) and Kaji (160) found hypotonicity to reduce [³H]bumetanide binding. Still others (e.g., Ref. 111) have shown that cell swelling has no effect on [³H]bumetanide binding. A set of results such as these raises additional questions about the quantitative relationship between diuretic binding and NKCC functional activation state (see sect. VIIIB3).

Despite these uncertainties, there is no question that raising [Cl]_i promotes the inhibition of the NKCC. Plus, there is strong evidence to support the view that such inhibition is the result of a reduction in the degree of phosphorylation of the cotransporter protein. In principle, decreased phosphorylation could result from either inhibition of phosphorylation (via a [Cl]_i-dependent inhibition of a protein kinase) or stimulation of dephosphorylation (via a [Cl]_i-dependent stimulation of a protein phosphatase). At this time, there is insufficient direct evidence to distinguish between the two possibilities, but circumstantial evidence favors an inhibitory effect of internal Cl on a protein kinase.

A possible role for intracellular Cl as an activator of protein phosphatase was tested by Altamirano et al. (6). Inhibition of bumetanide-sensitive ³⁶Cl influx caused by raising [Cl]_i could not be overcome or even offset by the protein phosphatase inhibitor okadaic acid in the internally dialyzed squid giant axon. The fact that okadaic acid could stimulate NKCC flux under conditions of limiting ATP concentration showed the agent to be capable of preventing NKCC dephosphorylation. Taken together, these results suggest that the inhibitory effect of raising [Cl]_i was not mediated by an enhancement of the okadaic acid-sensitive protein phosphatases (serine/threonine protein phosphatases 1 and 2A). In addition to the above experimental evidence against [Cl]_i-triggered regulatory role for protein phosphatases, there is the general observation that protein phosphorylation is rarely, if ever, found to be regulated at the protein phosphatase level. This is presumably because of the relative lack of specificity of such enzymes.

In view of the possibility that the intracellular Cl effect is mediated via a protein kinase, a recent report of an effect by Cl on a protein kinase in a nasal epithelium is quite interesting (344). These workers demonstrated that protein phosphorylation was a biphasic function of [Cl]_i. This is highly reminiscent of the effect of [Cl]_i on NKCC cotransport in the squid giant axon (28,29), tracheal epithelial cells (115), and HEK-293 cells (93). A particularly interesting part of this observation is that the protein kinase being affected by the change of [Cl]_i was one selective for GTP. (Protein kinase CK2 is known to be able to utilize GTP almost as easily as it uses ATP, e.g., Refs. 94, 128.) As discussed previously, there is only one report of an effect of a CK2 inhibitor on NKCC activity (254).

Finally, a recent report by Isenring and Forbush (147) may provide some further insight into the relationship between [Cl]_i and activation of the NKCC. Xu et al. (358) had reported that functional activation of expressed NKCC in HEK-293 cells requires preincubation in Cl-free medium. Isenring and Forbush (147) examined the kinetics of activation of bumetanide-sensitive ⁸⁶Rb uptake in HEK-293 cells exposed for various times to Cl-free medium. This treatment would reasonably be expected to lower [Cl]_i in a time-dependent manner. They observed that the longer the preincubation time, the greater is the stimulation of NKCC-mediated flux. The stimulation was accomplished by an increase of V_max without a change in the apparent K_m for Cl. This result seems to indicate that reducing [Cl]_i activates the NKCC by "turning on" a population of cotransporters that were previously inoperative. Whether this activation occurs as a result of phosphorylation or translocation from a subcellular site is presently unknown.

C) A DIRECT EFFECT OF INTRACELLULAR CL?  Alternatively, it is possible that intracellular Cl directly affects the NKCC protein itself, rather than indirectly through an effect on a protein kinase or phosphatase.

Intracellular Cl might exert its inhibitory effect by an allosteric effect on the NKCC protein directly to hinder or prevent its phosphorylation. Chloride has been shown to have an allosteric effect on the binding of oxygen to hemoglobin (285, 321). An increased [Cl] leads to a reduced oxygen affinity by neutralizing the electrostatic repulsion of positive charges found within the oxygen binding cavity of hemoglobin. At present, there is no evidence either for or against this possibility.

A) EFFECTS OF INTRACELLULAR NA⁺ AND K⁺.  In 1993, two groups reported findings which suggested that NKCC-mediated ion influxes were inversely related to [Na⁺]_i. Whisenant et al. (351) reported that a 20-min incubation of osteosarcoma cells (UMR-106-01 cells) in a Na⁺-free media (N-methyl-D-glucammonium⁺ replaced Na⁺) resulted in a tripling of the bumetanide-sensitive ⁸⁶Rb uptake. Because no measurements were provided of intracellular ion concentrations or cell volume, one cannot be certain whether the observed effects were the result of a parallel depletion of Cl, a potential result of the inhibition of the NKCC by the 0-[Na⁺]_o condition, and/or by a possible cell shrinkage caused by a net loss of Na⁺ plus Cl.

Ikehara et al. (144) varied the duration of ouabain treatment to examine the effect of changing [Na⁺]_i on ouabain-insensitive ⁸⁶Rb influx into HeLa cells. The fundamental observation was that as [Na⁺]_i increased (and [K⁺]_i decreased), the ouabain-insensitive ⁸⁶Rb unidirectional influx decreased. They interpreted this result to mean that intracellular Na⁺ inhibits influx via the NKCC, whereas intracellular K⁺ is necessary to interact with a "regulatory cation binding site." Although this is a very intriguing hypothesis, the design of the study makes it difficult to know the basis of the observed changes. In addition, there are important uncertainties about the effects of their treatment on pH_i, Ca²⁺ concentration, and membrane potential and how these variables might affect ouabain-insensitive ⁸⁶Rb influx.

Breitwieser et al. (29) replaced intracellular Na⁺ with N-methyl-D-glucammonium⁺ while maintaining [K⁺]_i, [Cl]_i, [Ca²⁺]_i, pH_i, and cell volume constant in the internally dialyzed squid giant axon. They reported that increasing [Na⁺]_i to 200 mM could partially (~65-80%) inhibit the bumetanide-sensitive influxes of both Cl and Na⁺. The apparent affinity of Na⁺ for the inhibitory "site" was increased when the [Cl]_i was increased. This is the expected effect for an increase of intracellular product concentration (e.g., [Na⁺]) if the cotransporter follows an ordered binding and release kinetic model. Importantly, raising [Na⁺]_i does not inhibit bumetanide-sensitive ²²Na efflux; the relationship between [Na⁺]_i and bumetanide-sensitive ²²Na efflux is a simple monotonic activation curve. This latter result is completely different from the relationship between [Cl]_i and bumetanide-sensitive effluxes of Cl, Na⁺, and K⁺ where a high [Cl]_i can completely block the NKCC-mediated cotransport effluxes of all three ions. Thus intracellular Na⁺ appears to inhibit by a kind of "end-product inhibition" mechanism (unlike the effect of intracellular Cl, which is more consistent with action at a non-cotransporting ion binding site).

B) EFFECTS OF INTRACELLULAR MG²⁺.  Given that ATP is essential for the operation of the NKCC, presumably by phosphorylating the cotransporter protein, it is not surprising that changes in intracellular Mg²⁺ levels have significant effects on cotransport function. Although studies on the role of Mg²⁺ in cotransport function have been relatively few (e.g., Refs. 62, 68), they have yielded clear results. Flatman (68) measured the bumetanide-sensitive ⁸⁶Rb uptake into ferret RBC as a function of the [Mg²⁺]_i using the ionophore A-23187 to adjust the [Mg²⁺]_i. Figure 26 shows that cotransporter-mediated ⁸⁶Rb uptake is increased as the [Mg²⁺]_i is increased. Flatman (68) also showed that the effects of varying the [Mg²⁺]_i cannot be attributed to changes of total ATP content. Two interesting features can be seen in this relationship. First, the bumetanide-sensitive ⁸⁶Rb influx rate is dependent on intracellular Mg²⁺ in a concentration-dependent manner. The sigmoid shape of the relationship suggests that the cooperative interaction of more than one Mg²⁺ is involved in the activating effect. This is consistent with a role for Mg²⁺ in the phosphorylation of the NKCC. Phosphoryltransferase reactions require Mg²⁺ to bind ATP in a stoichiometry of 2 Mg²⁺:1 ATP (350). Second, even at very low [Mg²⁺]_i (nominally 0.16 µM), there is still a substantial bumetanide-sensitive ⁸⁶Rb uptake (Fig. 26). How is this possible if NKCC activity depends on the protein being phosphorylated and such phosphorylation would be expected to be severely compromised at such low [Mg²⁺]_i? There are at least two possibilities. One is that not only do protein kinases require Mg²⁺ for their activation, but so do protein phosphatases. Hence, in the nominal absence of ionized magnesium, dephosphorylation may also be severely compromised, leaving the NKCC protein in a partially phosphorylated state (but see Ref. 71). Another possibility is that by reducing the [Mg²⁺]_i, KCC was activated. Potasssium-chloride cotransport is known to be activated by treatments that reduce the phosphorylation state (e.g., Ref. 154). However, given the relatively high bumetanide concentration used in this study (0.1 mM), KCC-mediated fluxes, if present, would be largely inhibited. The effects reported are generally consistent with [Mg²⁺]_i being a cofactor essential in the phosphoryltransferase reactions believed to be involved in the activation/deactivation of the NKCC.