The Molecular Basis of Chloride Transport in Shark Rectal Gland

With respect to the movement of NaCl, the tubular epithelial cell of the shark rectal gland can be considered ‘un porteur par excellence’. This is because it employs a secondary active transport system involving the coupled translocations of at least three different ions at several transporters and channels to accomplish the secretion of salt from the blood to sea water. Although the specific porters participating in NaCl secretion or absorption in different epithelial tissues vary, the basic paradigm remains the same (Nellans et al. 1973). The shark rectal gland has served as an extremely useful model for the characterization of such transcellular porter systems primarily because of the high degree of specialization of the tissue (Burger and Hess, 1960; Epstein and Silva, 1985); its utility has been analogous to that of the electric tissues of rays and eels as overexpressors of sodium channels and acetylcholine receptors. Its ready availability to seaside laboratories may also have contributed to the attention this organ has received. As will be discussed explicitly at the end of this article, tissues in other organisms have apparently evolved similar mechanisms of transepithelial NaCl movement (Haas, 1989).

The experimental work describing the salt transport process at the level of the whole organism, the tissue, the isolated tubules, subcellular fractions in the form of membrane vesicles and more recently cultured cells has been extensively documented during the past two decades (Siegel et al. 1976; Stoff et al. 1979; Epstein et al. 1983; Forbush et al. 1992; La et al. 1991; Greger and Schlatter, 1984a,b; Greger et al 1984). This brief treatise will not attempt to be comprehensive in outlining all the important contributions to our present understanding of the system. Rather, initial glimpses at the underlying molecular events will be emphasized. This has become possible because of the molecular cloning of three of the four main transporters and channels that combine to accomplish NaCl secretion. Because much of the regulation of the overall process appears to occur at the levels of the basolateral symporter, which moves Cl⁻ (as well as Na+ and K+) from the blood into the cells, and the apical Cl⁻ channel, which allows the anion to move out of the cell and into the duct lumen, these molecules will receive most attention.

The secretory tubules of the gland all empty into a single collecting duct, just as the arterial and venous capillaries with which they interdigitate connect to a single artery and vein. The tubular walls are composed of a highly uniform population of epithelial cells, one of which is illustrated in Fig. 1. The steps in the NaCl secretion process as they are presently conceived and the involvement of each of the transport and permeation pathways will be outlined, followed by the evidence supporting this interpretation. The stimulus for secretion is mediated by cyclic AMP. Elevation in the intracellular level of this cyclic nucleotide activates the apical CFTR chloride channel. This activation allows Cl⁻ to exit the luminal side of the cell down its electrochemical gradient. The resulting transient reduction in the intracellular Cl⁻ concentration can be sensed by the basolateral Na+/K+/Cl⁻ cotransporter (NKCC). This activation of the symporter results in the inward flux of all three ions from the blood side on the driving force provided by Na+ movements down its large electrochemical gradient. The immediate energy supply for the overall process derives from the active transport of Na+ back out across the basolateral membrane by the Na+/K+-ATPase.

The other cation which has entered via NKCC also exits on the basolateral side through a K+ channel. The Cl⁻ which has been transported in by NKCC restores its level to above electrochemical equilibrium with respect to the luminal cell exterior so that exit through the CFTR channel can continue if its activation persists. This transcellular chloride movement contributes substantially to the lumen-negative transepithelial potential that provides the driving force to pull Na+ through a paracellular permeation pathway from blood to lumen (Silva et al. 1983b). The net effect is Cl⁻ transport through the cell via the integrated action of transporters and channels and passive Na+ diffusion around the cell.

The apparent paradox is that NaCl is actively moved in a vectorial fashion across the epithelium without primary active transport of either Na+ or Cl⁻ in that direction. In fact, the active Na+ transport step is in the opposite direction across the basolateral membrane and Cl⁻ moves by secondary active transport across the basolateral membrane, coupled to the Na+ gradient by the cotransporter (NKCC) and by passive transport at the apical membrane Cl⁻ channel (CFTR). The rate-limiting and primary regulatory step resides at the CFTR Cl⁻ channel, which is optimally designed to respond to hormones employing cyclic AMP as second messenger. However, the means by which the regulation of each of the translocation steps is coordinated are of equal importance and provide an interesting prototype for the dissection of multicomponent transport systems. The hormonal control involves distinct up-regulation and down-regulation arms. The most physiologically relevant secretory agonist may be vasointestinal peptide (VIP), a neurotransmitter released by nerves supplying the rectal gland, which activates adenylate cyclase (Stoff et al. 1979). Adenosine is an especially interesting regulator which also activates through a stimulatory A₂ receptor in the micromolar concentration range but inhibits through an A₁ receptor in the low nanomolar range (Kelley et al. 1990, 1991). The former effect is mediated through cyclic AMP and probably reflects CFTR activation; however, the down-regulation, which may be a form of feedback control of the hormonal activation, occurs mainly downstream of cyclic AMP although the specific site has not yet been identified.

The ubiquitous Na+/K+-ATPase maintains the basic monovalent cation homeostasis in virtually all vertebrate cells and is basolaterally located in epithelia. It has been extensively studied over nearly the past 40 years and hence is one of the best understood P-type transport ATPases (Shull et al. 1985), although detailed structure–function relationships remain to be established. In contrast, the K+ channel, which is also essential to rectal gland NaCl secretion, has not been characterized beyond its sensitivity to inhibition by Ba2+, which also blocks secretion (Silva et al. 1981). The key molecules that are directly responsible for the passage of Cl⁻ across the basolateral and apical membranes are NKCC (Xu et al. 1994) and sCFTR (Hanrahan et al. 1993a,b), respectively. The latter was discovered 5 years ago and, because it is the site of the defect in cystic fibrosis, has been extensively studied since then (Riordan et al. 1989; Riordan, 1993). NKCC, however, has been described at the molecular level only this year (Xu et al. 1994) and as yet is less well understood.

Shortly after cloning of the human CFTR gene (Riordan et al. 1989), cDNAs with very high sequence similarity were isolated from libraries constructed using RNA isolated from rectal glands of Squalus acanthias (Grzelczak et al. 1990; Marshall et al. 1991). Overall amino acid sequence identity approached 80% and was greater than 90% for the putative transmembrane and cytoplasmic domains (excepting the R domain). At that time, evidence that CFTR was itself a Cl⁻ channel had not yet been obtained and its presence in such a highly conserved form in a tissue devoted to Cl⁻ secretion supported the idea that the molecule was directly involved in this process (Riordan et al. 1991). Extensive heterologous expression and in vitro mutagenesis studies (Welsh and Smith, 1993), in addition to the purification and reconstitution of human CFTR (Bear et al. 1992), subsequently provided convincing evidence that the protein constitutes a finely regulated low-conductance Cl⁻ channel. Hence, in many epithelial tissues that perform cyclic-AMP-stimulated Cl⁻ secretion, such as mammalian intestine, CFTR is the major apical conduit as well as the main regulatory element. In the case of sCFTR, it has more recently been possible to assemble a stable full-length cDNA and to express it in heterologous systems including mammalian and insect somatic cells and Xenopus oocytes (Hanrahan et al. 1993a). As illustrated in Fig. 2, sCFTR forms a channel with very similar properties to human CFTR. The macroscopic currents produced in Xenopus oocytes expressing cloned sCFTR had properties that were identical to those obtained by Sullivan et al. (1991) after injection of poly(A+) mRNA from shark rectal glands. Thus, conservation of function over this rather large evolutionary span is apparently as great as that of the primary structure. These data, however, do not prove that CFTR is the secretory Cl⁻ channel. In fact, earlier studies of chloride channels in cells of rectal gland tubules had shown the presence of two different channels, a smaller one with a conductance of 11 pS, which is close to that of human CFTR, and a larger one with a conductance of approximately 40 pS (Greger et al. 1985, 1987). Detection of the latter channel was enhanced by secretory stimuli (i.e. cyclic AMP), suggesting that it might carry the secretory current. However, in a more recent study carried out in conjunction with measurement of Cl⁻ channel activity associated with heterologously expressed sCFTR, it was found that a still smaller channel of approximately 4 pS appeared more frequently in stimulated tubules (Hanrahan et al. 1993a). The similarities between the endogenous channel and the recombinant one are shown in Fig. 3. We are unsure of the basis for the apparent discrepancy in the two studies of endogenous chloride channels. The role of the larger channel, if indeed it is not involved in stimulated Cl⁻ secretion, remains to be determined. However, present evidence favours sCFTR as the secretory channel and its properties seemed to be aptly suited to this role.

Although sCFTR has not yet been extensively studied, human CFTR has been well characterized and present indications are that the molecules are extremely similar. Fig. 4 shows a schematic representation of the molecule. This two-dimensional depiction was originally based primarily on hydropathy analyses and sequence similarities to members of a superfamily of ATP-binding integral membrane proteins (Riordan et al. 1989). The unit structure, which is the hallmark of this class of molecules, is a polytopic membrane-associated segment followed by a nucleotide binding fold (NBF). Two such units separated by a large hydrophilic central domain (R domain) and moderately sized N-and C-terminal extensions make up the molecule. This topological disposition with respect to the lipid bilayer has recently been verified experimentally (Chang et al. 1994); there is increasing evidence that several of the twelve transmembrane sequences contribute to the ion pore (Anderson et al. 1991; Tabcharani et al. 1993). Regulation of the channel is extremely complex (Tabcharani et al. 1991), involving the action of ATP as substrate for several protein kinases that phosphorylate CFTR as well as binding to the NBFs (Baukrowitz et al. 1994). From the point of view of cyclic-AMP-mediated hormonal activation of Cl⁻ secretion, such as that occurring in the rectal gland, phosphorylation of the R domain by PKA and dephosphorylation are most relevant (Cheng et al. 1991; Chang et al. 1993). Some insight is beginning to be gained into the mechanism whereby phosphorylation of multiple R domain sites activates the channel (Dulhanty and Riordan, 1994a,b). The additional layer of regulation stemming from NBF interactions with ATP becomes most obvious in CFTR variants created by in vitro mutagenesis in which the R domain is either removed (Rich et al. 1991) or its PKA phosphorylation sites converted to residues which are already negatively charged (aspartates or glutamates; Riordan et al. 1993) without addition of phosphoryl groups. However, the physiological basis of regulation by ATP action at the NBFs is not yet clear, although it has been suggested that it is simply a means of turning off secretion in cells with a suboptimal energy charge so that remaining ATP would not be further dissipated (Quinton and Reddy, 1992). Nevertheless, even without this capability, phosphorylation of the R domain of sCFTR by PKA on hormonal stimulation would suffice to initiate the series of steps outlined above, including activation of NKCC.

The non-electrogenic coupled transport of Na+, K+ and Cl⁻ has been extensively characterized in many different cell types, in both ordinary symmetrical cells and bipolar epithelial cells and in plasma membrane vesicles derived from them (Kinne et al. 1985). Hallmarks of the process include the absolute interdependency of the movement of four ions (1 Na+, 1 K+ and 2 Cl⁻) in the inward direction across the plasma membrane and sensitivity to inhibition by drugs such as furosemide and bumetamide (Haas, 1989). Activation of the cotransporter is brought about by treatments which elevate cyclic AMP levels, such as VIP, and by manipulations which decrease cellular Cl⁻ concentration. The binding of some members of this class of compounds, such as benzmetamide or 4-benzoyl-5-sulphamoyl-3-(3-thenyloxy)benzoic acid, to cells or membranes provided a means of identifying and measuring NKCC (Forbush et al. 1992). The latter photoactive compound bound specifically to a protein in rectal gland membranes with an apparent M_r of approximately 195 (Lytle et al. 1992a). This binding occurred only when the transporter was in an active state. This specific labelling provided a means of following the protein during purification from detergent-solubilized rectal gland membranes. Use of this highly specialized tissue was especially advantageous for this purpose because of its elevated level of expression of the transporter protein (approximately 1% of total protein). This enabled the acquisition of sufficient protein for the immunization of mice for monoclonal antibody production and the generation of proteolytic fragments for amino acid sequencing (Lytle et al. 1992b). It also ensured abundant representation in cDNA libraries constructed using the RNA isolated from the tissue (Xu et al. 1994). This is in contrast to CFTR, which is also dedicated to the specialized function of the tissue but, being a channel, need not be present in very high copy number to provide adequate Cl⁻ conductance.

A panel of monoclonal antibodies (mAbs) was obtained and successfully employed as probes for the isolation of cDNAs for NKCC by conventional bacterial expression cloning techniques (Xu et al. 1994) as well as in immunoprecipitations of the protein from detergent lysates of activated tubules (Lytle and Forbush, 1992). The latter methodology was used to detect phosphorylation of the polypeptide, which correlated with activation by either cyclic AMP agonists or osmotic manipulations (Lytle and Forbush, 1992). The data obtained supported the idea that phosphorylation is directly involved in cotransporter activation. This interpretation is based primarily on the observations that broad protein kinase inhibitors blocked the stimulated binding of benzmetamide, which parallels cotransporter activity, and the fact that the kinetics of phosphorylation and dephosphorylation were synchronized with NKCC activation and inactivation. Both phosphoserine and phosphothreonine were identified by phosphoamino acid analysis. Proteolysis and fractionation of phosphopeptides that could be sequenced accounted for much of the phosphorylation at two sites near the N and C termini of the protein (see below). As yet, the protein kinases responsible for phosphorylation at these sites have not been identified. Lack of correspondence of the polypeptide sequences around these two sites with the consensus sequences for phosphorylation by PKA (Xu et al. 1994) suggests that activation does not result from the direct action of PKA (Lytle and Forbush, 1992). Because the rectal gland cotransport process is known not to be strongly influenced by either phorbol esters or calcium ionophores, it is unlikely that PKC or calcium/calmodulin-dependent protein kinases are responsible. Hence, in contrast to CFTR, in which at least a major part of the cyclic AMP activation can be accounted for by the direct action of PKA on the R domain, there must be a more indirect route from both cyclic AMP and osmotic stimuli to the phosphorylation and activation of NKCC. It appears that a decrease in intracellular [Cl⁻] is an important part of this pathway (Lytle et al. 1992a). Elucidation of these pathways should now be possible with the availability of both the purified protein and expressible cDNAs.

Proof that the cDNA isolated with the mAb probes contained authentic NKCC sequence came from the correspondence between the amino acid sequences of several peptides generated from the purified 195 kDa polypeptide and regions of the translated cDNA sequence as well as the functional effects of the heterologous expression of the full-length cDNA in mammalian cells (Xu et al. 1994). This resulted in the production of a polypeptide which, after removal of oligosaccharide chains, had the same apparent molecular size as the deglycosylated protein endogenous to the rectal gland. Reduction of intracellular [Cl⁻] was necessary for cotransporter activity of the recombinant protein, which is consistent with evidence that reduction in [Cl⁻]_i may provide the signal for NKCC activation when the CFTR Cl⁻ channel is initially activated by secretory stimuli. When function was assessed by the uptake of 86Rb+ (as a K+ surrogate) into the heterologous cells, bumetamide-sensitive flux was elevated 10-fold and was absolutely dependent on the presence of extracellular Na+ and Cl⁻.

Properties of the NKCC protein discerned largely from its primary structure are illustrated schematically in Fig. 5. The predicted 1191-residue polypeptide has features consistent with its identity as an integral membrane glycoprotein. Analysis of the sequence using hydropathy algorithms suggests that there may be 12 membrane-spanning sequences as in CFTR. These are bracketed by long hydrophilic sequences at the N and C termini which, on the basis of the presence of verified phosphorylation sites and intracellular epitopes for antibodies (Xu et al. 1994), should be cytoplasmically oriented. This topological disposition would provide five cytoplasmic and six extracytoplasmic loops separating the transmembrane segments. Of the total of nine consensus sequences for N-glycosylation, one is present in the first and three in the fourth extracytoplasmic loop. The diagram suggests that the latter three are employed, although this has not yet been demonstrated experimentally. If it is confirmed, it would mean that the same relative location is glycosylated in the CFTR and NKCC molecules. The two positions which are phosphorylated in the secretory tubules in situ at residues T189 and T1114 in the N-and C-terminal hydrophilic domains, respectively, are indicated in Fig. 5.

When the NKCC sequence was compared with others in protein databases, homology was found with several genes coding for proteins of unknown function as well as with a Na+/Cl⁻ cotransporter from flounder urinary bladder (Gamba et al. 1993). Overall sequence identify with the latter was 47%, with regions of much greater identity indicated by the thickened lines along the polypeptide in Fig. 5. They include the third, sixth, eight and tenth putative transmembrane sequences, the first cytoplasmic loop and a short peptide about 180 residues from the C terminus. Because these two molecules share the ability to cotransport Na+ and Cl⁻, it can be speculated that these regions may participate in the handling of these two ions. These and other structure–function relationships are now amenable to investigation by in vitro mutagenesis. The construction of hybrid molecules in which portions of the two can be interchanged, as well as with other related molecules that have been detected by cross-hybridization, should provide important clues to domain functions and guide the design of site-specific mutagenesis. The ability to co-express shark NKCC and CFTR should also facilitate the elucidation of the regulatory interrelationsips between these two molecules, which are the primary determinants of the Cl⁻ concentrative and Cl⁻ dissipative arms of this transepithelial transport mechanism.

The complete description of this overall cellular symport system that accomplishes NaCl secretion requires not only a characterization of each of the transporters and channels involved, but also of how their activities are coordinately regulated. Although there were some earlier suggestions that the Na+/K+-ATPase might be directly activated during rectal gland stimulation via cyclic AMP (Silva et al. 1983a), these effects were small. A more recent study indicated an inhibition of Na+/K+-ATPase by kinase-mediated phosphorylation (Bertorello et al. 1991). However, the pump probably works primarily to maintain cellular Na+ and K+ homeostasis as it does in most cells. What is clear is that activity of the sodium pump is essential to the secretory process, since ouabain blocks rectal gland secretion completely (Silva et al. 1977). Control of the K+ exit channel has not been extensively examined and its molecular make-up is entirely unknown. Greger et al. (1984) showed that changes in K+ conductance on stimulation of secretion were modest compared with the large increase in Cl⁻ conductance. Rapid advances in understanding of CFTR as a highly apical Cl⁻ channel that is activated and inactivated by PKA-mediated phosphorylation and dephosphorylation, respectively, has provided insight into the initial regulatory event in Cl⁻ secretion. Even more recent molecular studies of NKCC clearly indicate that it is not an analogous PKA substrate. However, phosphorylation by other kinase(s) probably does contribute to activation, possibly as a consequence of the reduction in [Cl⁻]_i resulting from CFTR channel activity (Lytle et al. 1992a).

This multimolecular mechanism of Cl⁻ secretion that is so magnified in shark rectal gland is widely employed in other secretory epithelia. Avian salt gland is another very specialized example which functions very similarly (Simon, 1982). However, not all of the constituents are identical in each case. For example, additional or alternative Cl⁻ channels may be employed in some tissues, such as tracheal epithelium (reviewed by Hanrahan et al. 1993b; see also Boucher, 1993).

Some reabsorptive epithelia use most of the same transporters and channels to move chloride in the opposite direction (Epstein and Silva, 1985). The thick ascending limb of the loop of Henle in the kidney is among the best characterized examples (Haas, 1989). The sidedness, with respect to the apical and basolateral surfaces, of the Cl⁻ channels, the cotransporter and the K+ channel is reversed from that in the secretory tissues so that salt is moved from the luminal to the blood side. The cotransporter from reabsorptive kidney tubules in particular has many of the same properties as the rectal gland NKCC, and isolation of the homologous cDNA has recently confirmed this relatedness (Payne and Forbush, 1994).