| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
MEMBRANE TRANSPORTERS, ION CHANNELS, AND PUMPS
Department of Cellular and Molecular Physiology, Yale University School of Medicine, New Haven, Connecticut
Submitted 13 May 2006 ; accepted in final form 17 January 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
40 µM) < NKMIK 
NKMIN KKMIN < KNMIN KNMIK < NNMIK < NNMIN (400 µM) < DDMID < EEMIE (800 µM). Thus the second K is the most important for reversible DIDS blockade. Nevertheless, these mutations had relatively little effect on slope conductance in the absence of DIDS. For KKMIK, RRMIR, NKMIK, KKMIN, KNMIK, and NNMIN, the rates of irreversible inhibition by DIDS roughly parallel the apparent affinities for reversible DIDS binding. The rate was extremely low for DDMID. The fitted maximal inhibitions were 80–91% for the first five constructs, and 66% for NNMIN. Thus DIDS probably reversibly binds before irreversibly reacting with NBCe1-A. Finally, tenidap blocks not only KKMIK, but also NNMIN and EEMIE. bicarbonate; pHi; acid-base
In the proximal tubule, the splice variant NBCe1-A works with a Na+:HCO3– stoichiometry of 1:3, and thus is electrogenic (33). In other tissues, as well as in Xenopus oocytes, NBCe1-A and other splice variants (-B and -C) work with a stoichiometry of 1:2—possibly because of the absence of a kidney-specific protein (32).
Except for the electroneutral Na/HCO3 cotransporter NBCn1 (10) and the borate transporter BTR1 (or NaBC1) (28), a striking characteristic of other members of the SLC4 family is their sensitivity to a class of drugs typified by 4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid (DIDS) (8, 9). This compound, with two negative charges and two isothiocyano groups, can covalently react with amines. SITS, a close relative of DIDS (but with only a single isothiocyano group) interacts with the extracellular surface of the Cl-HCO3 exchanger AE1 in two steps (7), first reversibly (by rapidly interacting with positive charges on AE1) and then irreversibly (by slowly reacting covalently with a lysine residue). Scavenging with albumin can remove reversibly bound, but not covalently reacted stilbene (7). Biochemical studies of AE1 and alignments of deduced amino-acid sequences of the AEs suggest that the lysine (Lys, K) with which H2DIDS reacts covalently is near the extracellular end of the putative transmembrane segment 5 (TM5), with a consensus motif of KLXK, where X may be I or Y (24). Site-directed mutagenesis of the first Lys in the KLIK motif of AE1 to Asn prevents the reversible (23) as well as the covalent reaction (2, 3) of H2DIDS. Others have confirmed, by biochemical analysis of AE1 proteolytic fragments, that H2DIDS covalently reacts with the first K in the KLIK motif (27). The binding kinetics of H2DIDS and DIDS to red blood cells are different (25), and it is not known which AE1 residues are important for the rapid, ionic interaction or slow covalent interactions with SITS or DIDS. As far as the Na+-coupled HCO3– transporters are concerned, no information is available on the residues important for either the ionic or the covalent interaction. At the extracellular end of TM5, human NBCe1-A has the sequence KKMIK, with lysine residues at positions 558, 559, and 562. In fact, most members of the superfamily tend to conserve positively charged residues at the extracellular ends of TM5.
Our goal was to determine whether the KKMIK motif in NBCe1 plays a role in the reversible and irreversible interactions with DIDS. The approach was to make point mutations in the KKMIK motif, express the mutants in Xenopus oocytes, and use a two-electrode voltage clamp to measure the inhibition of NBCe1 currents by DIDS. To verify delivery of wild-type and mutant proteins to the membrane, we fused enhanced green fluorescent protein (EGFP) to the COOH terminus of the NBCs. To study multiple DIDS concentrations in a single oocyte—and thus determine a complete dose-response relationship—we used albumin to scavenge unreacted DIDS. We found that the lysine residues in the KKMIK motif—particularly the second lysine—are extremely important for reversible DIDS binding. We also determined the time course for irreversible DIDS inhibition and found that no single K in the KKMIK motif of TM5 is critical for the irreversible DIDS blockade of NBCe1-A.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
We thank Dr. Leila Virkki for a gift of the NBCe1-EGFP cDNA construct, consisting of human NBCe1-A fused in frame at its 3' end to EGFP (Clontech, Palo Alto, CA), and subcloned into pGH19 Xenopus expression vector (see Ref. 35 for details). Use of this construct was recently reported in Ref. 29. We will abbreviate this clone as wild-type KKMIK.
The missense mutations were introduced into the Xenopus NBCe1-A expression constructs using the QuikChange mutagenesis kit (Stratagene, Cedar Creek, TX). The Keck Oligonucleotide Synthesis Facility at Yale University generated the oligonucleotide primers used to introduce mutations at K558, K559, and K562 within the KKMIK motif within putative TM5 of wild-type KKMIK. The EGFP-tagged constructs with one K, two Ks, or three Ks mutated will be abbreviated as NKMIK, KNMIK, KKMIN, NNMIK, NKMIN, KNMIN, NNMIN, RRMIR, DDMID, and EEMIE. The sequences of the final constructs were confirmed by automated sequencing performed by the Keck Sequencing Center at Yale University.
Expression in Xenopus Oocytes
We synthesized capped mRNAs in vitro for both the wild-type KKMIK and the mutants with the T7 mMessage mMachine kit (Ambion, Austin, TX). Stage V and VI oocytes from Xenopus laevis were isolated as described previously (31). One day after isolation, we injected the oocytes with 50 nl of a solution containing 0.5 ng/nl of mRNA encoding wild-type KKMIK or one of the mutants. Control oocytes were injected with 50 nl of sterile water. The oocytes were used in experiments 4–6 days after injection. All experiments were performed at room temperature (22°C).
We routinely used a plate-reader assay (34) to screen oocytes for expression of EGFP-tagged NBCe1-A.
Solutions
Nominally CO2/HCO3–-free ND96 solution contained (in mM) 96 NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2, and 5 HEPES at a pH of 7.50. HCO3–-containing solution was prepared by replacing 33 mM NaCl in ND96 with 33 mM NaHCO3, and equilibrating the solution with 5% CO2/balance O2. DIDS (Sigma-Aldrich, St. Louis, MO) solutions were made freshly by adding the powder to the 5% CO2/33 mM HCO3– solution. Sigma-Aldrich reports the DIDS purity as "minimum 80%". The results of negative ion electrospray mass spectroscopy (W. M. Keck Foundation Biotechnology Resource Laboratory, Yale University) are consistent with such a level of purity. The actual DIDS concentrations listed in this paper were adjusted by a factor of 0.8 to account for the purity of the compound. The single major contaminant had a molecular mass of 410.98 Da, and is likely to be 4-amino,4'-isothiocyanatostilbene-2,2'-disulfonic acid. 0.2% Albumin (Sigma-Aldrich) solution was made freshly by the addition of 0.2 g/100 ml into 5% CO2/33 mM HCO3– solution. [5-chloro-2,3-dihydro-3- (hydroxy-2-thienylmethylene)-2-oxo-1H-indole-1-carboxamide] (Tenidap) was a gift of Pfizer, Groton, CT. We made a fresh 1 M stock solution of tenidap powder in DMSO, and then diluted this 1:1,000 into the CO2/HCO3– solution. DMSO at 1:1,000 does not affect NBCe1-A currents (not shown). The osmolalities of all solutions were adjusted to 200 mosmol/kgH2O by the addition of water or mannitol.
Two-Electrode Voltage Clamp
In all experiments, the oocyte was initially superfused in a plastic perfusion chamber with the ND96 solution, which is nominally CO2/HCO3– free. Solutions were contained in 140-ml plastic syringes (Sherwood Medical, St. Louis, MO), carried via Tygon tubing (Formulation R3603–3; OD 4.8 mm/ID 1.6 mm; Ryan Herco Products, Burbank, CA), and delivered at a rate of 4 ml/min to the chamber using syringe pumps (Harvard Apparatus, South Natick, MA). Solutions were switched with pneumatically operated valves (Clippard Instrument Laboratory, Cincinnati, OH).
We used two-electrode voltage clamp to measure whole cell ionic currents of oocytes expressing wild-type KKMIK or mutants. Currents and voltages were recorded with an oocyte clamp (model OC-725C; Warner Instruments, Hamden, CT), under the control of the Clampex module of pCLAMP software (version 8; Axon Instruments, Foster City, CA). The current and voltage electrodes were fabricated from thin-walled borosilicate glass (part no. 300077, Harvard Apparatus, Holliston, MA); resistances were 0.3–1.0 M
when filled with 3 M KCl. A third, 3 M KCl electrode with negligible tip resistance was used as the reference in the bath (ISense connection of the OC-725C).
We began all experiments with the oocyte in the ND96 solution. After the spontaneous membrane potential (Vm) stabilized, we turned on the clamp at a command potential close to the spontaneous Vm until initiating the voltage-clamp protocol. Before switching the extracellular solution to one containing CO2/HCO3–, we turned off the clamp, made the solution change, and finally reapplied the clamp later at the new spontaneous Vm. Current-voltage (I-V) relationships were generated by stepping the holding potential from –160 to +20 mV in 20-mV increments, each step lasting 100 ms.
Data Analysis
Data were analyzed using Clampfit 8.0 and Microsoft Excel 97. Because of technical limitations of the OC-725C oocyte clamp, the raw I-V relationships were not acquired precisely at even 20-mV increments. We interpolated/extrapolated values to even multiples of 20 mV in Excel 97. Values are given as means ± SE, with the number of replicate experiments (n) in the data set from which they were calculated. We also wrote an in-house, nonlinear least-squares curve-fitting program (see APPENDIX) to compute simultaneously—from the currents at various DIDS concentrations at a particular holding potential—the apparent inhibitory constant (Ki) for reversible DIDS binding and the non-NBC background current (IBack).
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
The protocol. In 1972, Cabantchik et al. (7) established that the rapid, ionic interaction between SITS and AE1 could be reversed by scavenging the SITS with albumin. By using albumin to scavenge reversibly bound DIDS from NBCe1, an approach previously used by Kietz et al. (23) to scavenge H2DIDS from AE1, we could perform multiple rounds of DIDS exposures and wash offs—and therefore obtain a full DIDS dose-response curve—on a single oocyte. Figure 1 summarizes an experiment on an oocyte expressing wild-type KKMIK.
|
In a similar fashion, we performed additional cycles for [DIDS] values of 10 µM (Fig. 1B), 20 µM (Fig. 1C), and so on. The Albumin/post-DIDS I-V curve of the preceding cycle is the same as the "Albumin/pre-DIDS" of the current cycle. In these other panels, for the sake of simplicity, we plot only the mean albumin currents (averaging the Albumin/pre-DIDS and Albumin/post-DIDS). Note that the mean albumin current (IAlbumin) gradually fell from Fig. 1A to Fig. 1H, presumably reflecting a gradual increase in the fraction of NBCe1-A molecules that have become irreversibly blocked by DIDS. We will consider this irreversible inhibition in greater detail below.
The analysis.
At a given Vm, the NBCe1-A current in DIDS (IDIDS) falls in a regular pattern as [DIDS] rises, and would presumably fall to the background current (IBack) at a [DIDS] of 
. Of course, one must know IBack to compute the fractional inhibition of the current produced by DIDS, and this fractional inhibition is required to compute Ki. As described in the APPENDIX, our analysis software uses a nonlinear, least-squares approach to simultaneously compute IBack and Ki at each value of Vm. Figure 1I shows the voltage dependence of fitted IBack for one oocyte. This procedure works very well for Vm values of +20 to a range of –40 to –80 mV, where IAlbumin and IBack are sufficiently different that we can accurately compute the difference (IAlbumin – IBack). At Vm values of –100 and –120 mV—near the point where the IAlbumin-vs.-Vm curve, the IBack-vs.-Vm curve, and all IDIDS-vs.-Vm curves cross one another–any analysis (computerized or otherwise) always fails because of the similarities of the various currents. At Vm values of –140 and –160 mV—where the I-V curves sometimes have diverged sufficiently—the computerized analysis works reasonably well. As described below, we have used another inhibitor—tenidap, which acts at a site different from that of DIDS—to confirm our analysis.
The mutants.
We performed protocols—similar to that in Fig. 1—on a total of 10 NBCe1-A mutants. The first two columns of Table 1 list the mutants and the number of experiments. In the absence of DIDS, these mutations had relatively little impact on the slope conductance of NBCe1-A (Table 1, column 3). As far as the action of DIDS is concerned, at a Vm of 0 mV, the apparent Ki for reversible DIDS binding increased in the sequence RRMIR < KKMIK (wt, 40 µM) < NKMIK NKMIN KKMIN (60 µM) < KNMIN KNMIK (130 µM) < NNMIK (300 µM) < NNMIN (400 µM) < DDMID (500 µM) < EEMIE (800 µM). Thus mutants in which we replaced positively charged Lys with the neutral Asn at one, two, or all three residues in the motif at the end of TM5 have lower apparent affinities for DIDS, indicating that the KKMIK motif is important for reversible DIDS binding. A positive charge in the motif therefore appears to be important for DIDS binding. What is more, the Ki of the single mutant KNMIK is twice that of the two other single mutants (NKMIK and KKMIN), or even the double mutant NKMIN. Similarly, the Ki of the double mutants lacking the second Lys (NNMIK and KNMIN) are higher than that of the other double mutant (NKMIN). Thus K559 in the KKMIK motif of TM5 (analogous to the first K in the SKLIK motif of AE1) is the most important of the three lysines for reversible DIDS blockade of NBCe1-A. Finally, the triple mutant with Asn residues replacing all three Lys residues has the lowest Ki of all clones with K
N substitutions.
|
Concentration dependence of DIDS inhibition. As noted in MATERIALS AND METHODS, the Sigma DIDS preparation consists of 80% DIDS and 20% of a related contaminant. Fig. 2 shows the DIDS dose-response relationship for two representative data sets: the wild-type KKMIK and the NNMIN mutant, both at 0 mV. Each symbol represents a mean fractional inhibition computed for individual oocytes using the nonlinear, least-squares approach outlined in the APPENDIX. The curves, which represent best fits of the Michaelis-Menten equation to the data, show that the DIDS preparation behaves as if only one agent was responsible for the reversible blockade of NBCe1-A activity. As noted in the DISCUSSION, this behavior could reflect the simultaneous actions of DIDS and the contaminant.
|
Figure 3A shows I-V curves for wild-type KKMIK exposed to albumin vs. 40 µM DIDS (similar to that shown in Fig. 1D), as well as a third I-V curve that represents the voltage dependence of the fitted IBack. Finally, it shows a fourth I-V curve for the initial condition in the CO2/HCO3–-free ND96 solution. Figure 3B shows comparable I-V curves for the NNMIN mutant exposed to 400 µM DIDS. We chose the two [DIDS] levels to be similar to the Ki values of the respective constructs. The most striking feature of both Fig. 3, A and B, especially panel B, is the increasing inward rectification of the IDIDS curve over a Vm range corresponding to increasing outward currents (i.e., inward movements of HCO3– or CO
).
|
|
20%, which is somewhat below the extrapolation of the fractional-inhibition vs. Vm line that we obtained between +20 and –40 mV. Thus, for wild-type KKMIK, the DIDS blockade was somewhat voltage dependent. The squares in Fig. 4A summarize mean data for the computer analyses of the blockade of the NNMIN mutant at a [DIDS] of 400 µM (i.e., near the Ki at 0 mV). From +20 to –80 mV (i.e., for outward currents/inward movements of HCO3–/CO3=), fractional inhibition fell more steeply with Vm than was the case for KKMIK. At very negative voltages (i.e., –140 to –160 mV, where the currents are inward), we were unable to compute fractional inhibition because the I-V curves nearly merged.
Figure 4B summarizes mean data for the computer analyses of the voltage-dependence of Ki for wild-type KKMIK. The diamond symbols show that Ki values rise monotonically from +20 to –180 mV. The squares in Fig. 4C summarize the mean data for the computer analyses of the apparent Ki of the NNMIN mutant at voltages from +20 to –180 mV. At more negative voltages, the computed Ki values rise substantially.
The fifth column in Table 1 summarizes the voltage dependence of Ki values (i.e., 
Ki/Vm) between –40 and +20 mV for all 11 clones. This Vm range corresponds to the linear portion of Fig. 4, B and C, and also to the Vm range where the I-V curves are the most widely separated, and thus the calculations most reliable. The negative Ki/Vm values indicate that, for all clones, the apparent Ki always increased as Vm became more negative. The Ki/Vm for KKMIK was –0.34 µM/mV, and the Ki was 42 µM at 0 mV. Thus the Ki changed 0.8%/mV (Table 1, column 6). The Ki for NNMIN changed 1.4%/mV. As expected, the ratio 0.8/1.4 roughly corresponds to the ratio of the slopes of the dotted lines (between –40 and +20 mV) in Fig. 4A. For all clones, the (Ki/Vm)/Ki ratios were within a factor of 2, indicating that all of the clones exhibit similar degrees of voltage-dependent reversible inhibition by DIDS.
The use of tenidap to block NBCe1—and thus directly reveal IBack—confirms the importance of the KKMIK motif and voltage-dependent blockade by DIDS.
As noted above, the computer analysis works well at voltages where the family of I-V curves is sufficiently divergent. However, when this is not true, the calculation of IBack and Ki is problematic or impossible. However, we reasoned that if we had an independent method of obtaining IBack, we could improve the analysis in problematic Vm regions. Ducoudret et al. showed that tenidap inhibits NBCe1-A (14). Using an approach similar to that described above for DIDS, in a single experiment we found that the apparent Ki of tenidap for inhibiting wild-type KKMIK was 26 ± 1 µM at 0 mV. In a single experiment with NNMIN, the value was 26 ± 2 µM at 0 mV.
Figure 5, A–C, is analogous to Fig. 3, A and B, but, aside from including EEMIE, differs in two respects. First, in this series of experiments, we used a single (or a few levels of) DIDS in the vicinity of the Ki value for the constructs. Thus, this data set is not amenable to the computer analysis described earlier (i.e., we have no fitted estimate of IBack). Second, we concluded each experiment with an exposure to 1,000 µM tenidap. We regard the current in tenidap (Itenidap) as our best estimate of IBack. In Fig. 5A, the I-V relationship for wild-type KKMIK in the presence of tenidap is virtually linear (gray diamonds), as is the corresponding I-V relationship in the absence of drugs (squares). Thus the I-V relationship for NBCe1-A must be virtually linear in the voltage range –160 to +20 mV.
|
Figure 6 is analogous to Fig. 4, but differs in three respects. First, Fig. 6 includes EEMIE. Second, we used Itenidap (assumed to be equal to IBack) and Eq. A2 to compute the fractional inhibition of DIDS at a single [DIDS]. Third, in Fig. 6B we used this fractional inhibition, single value of [DIDS], and Eq. A1 to compute the apparent Ki at each Vm.
|
Figure 6B shows—similar to what we saw in Fig. 4B and C—that the Ki values tend to rise monotonically as Vm becomes more negative. For the two mutants, which have lower apparent affinities for DIDS, these Ki values rise substantially at extremely negative values of Vm (i.e., inward currents/outward movements of HCO3–/CO3=). For the wild-type KKMIK, the apparent Ki values ranged from 38 ± 6 µM at +20 mV to 133 ± 41 µM at –160 mV.
No Single K in the KKMIK Motif of TM5 is Critical for the Irreversible DIDS Blockade of NBCe1-A
As noted above, when we exposed oocytes to DIDS for 30 s—and then removed DIDS and scavenged with albumin—we observed that the albumin/pre-DIDS current at 0 mV was always somewhat greater than the albumin/post-DIDS current. This difference was small in Fig. 1A (4.8 µM DIDS), but increased with increasing [DIDS] (not shown). We hypothesized that this irreversible blockade represents the covalent reaction of DIDS with NBCe1-A.
Because the irreversible DIDS blockade after a 30-sec DIDS exposure was relatively small, especially when [DIDS] was small, we performed a final series of experiments in which we exposed oocytes for up to 60 min to 800 µM DIDS. We gathered data for DIDS exposure of 1, 2, 4, and 8 min using the same albumin wash-off protocol as depicted in Fig. 1. In this protocol, we continuously impaled a single oocyte for as long as 8 min, turning off the clamp between I-V curves. Because oocytes did not generally survive this kind of "continuous" recording for more than 8 min, for longer DIDS incubations we used a somewhat different "discontinuous" approach. We 1) recorded each oocyte's initial I-V relationship in albumin, and 2) removed the oocyte from the recording chamber to a petri dish containing 800 µM DIDS for an incubation of either 7.5, 15, 30, or 60 min. After the incubation, we 3) returned the oocyte to the chamber through which we were now flowing 800 µM DIDS. Finally, we 4) then switched the solution to albumin for 1 min and (e) obtained a second I-V curve. The difference in the two I-V curves obtained in albumin from the same oocytes—separated by a certain time of DIDS incubation—represents the irreversible DIDS blockade. Because this protocol allowed some time for the oocyte to reseal and recover, the oocyte survived the 60-min DIDS incubation. Thus, for each cDNA construct, we merged the data from the two DIDS protocols.
Figure 7 summarizes the time dependence of the irreversible DIDS blockade for all the NBCe1-A constructs. For example, the brown diamonds represent the mean time course of irreversible blockade produced by 800 µM DIDS for the wild-type KKMIK. The brown line represents a single exponential fit through all these data points. The best-fit maximal irreversible DIDS blockade for wild-type KKMIK was 88%, with a time constant of 8 min. As summarized in Table 2, the constructs fell into three groups. 1) The wild-type KKMIK, the three single mutants, and RRMIR all had high maximal inhibitions and time constants between 8 and 18 min. 2) NNMIN had a maximal inhibition of only 66% and a time constant of 33 min. And 3) DDMID exhibited such a slow increase in irreversible inhibition that it was impossible to obtain a meaningful fit.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
3. The present study establishes, for the first time for any other member of the SLC4 family, that the comparable residue (i.e., the second K in KKMIK of NBCe1-A) is also critical for the reversible binding of DIDS. In fact, we observed virtually the same threefold decrease in affinity. On the other hand, we found that no single Lys residues in the KKMIK motif of NBCe1-A is critical for irreversible blockade. Purity of DIDS
Mass spectroscopy revealed that the single major contaminant in the commercial DIDS preparation is likely to be 4-amino,4'-isothiocyanatostilbene-2,2'-disulfonic acid. One reacts 4,4'-diaminostilbene-2,2'-disulfonic acid (DADS) with thiophosgene to replace the two amino groups of DADS with two isothiocyanato groups, yielding DIDS (7). Thus, the contaminant is likely to be a partial reaction product in the synthesis of DIDS, in which only one of the two amino groups of DADS is replaced. As a reversible inhibitor of anion transport in red blood cells (7), DADS has an extremely low apparent affinity (Ki 10–3 M), whereas DIDS has a much higher apparent affinity (Ki 2 x 10–7 M). Thus, it is likely that the contaminant (a chimera of DIDS and DADS) has a much lower affinity for NBCe1-A than DIDS does, and thus is likely to behave as an inert substance in our assays.
Figure 2 shows that the inhibition produced by the DIDS preparation at 0 mV follows Michaelis-Menten kinetics both for KKMIK and NNMIN. Rapid-equilibrium kinetics predict the following relationship for one inhibitor
 | (1) |
 | (2) |
 | (3) |
may be 0.8 and 
may be 0.2. The terms in parentheses in Eq. 1 and Eq. 3 must be equal, so that
 | (4) |
), and we multiplied all "apparent DIDS concentrations" by 0.8. If these assumptions are valid, our reported Ki values would correspond to Ka in the above equations. If the /Kb term in Eq. 4 should prove to be appreciable, then our reported Ki values (which would correspond to 80% of the Ki value in Eq. 4) would represent an amalgamated, apparent affinity of the two inhibitors for NBCe1-A. Regardless of the value of Kb, our data provide valuable insight into the properties of the disulfonic-stilbene binding site of NBCe1-A. If the contaminant is indeed 4-amino,4'-isothiocyanatostilbene-2,2'-disulfonic acid, it could covalently react with Lys residue via its lone isothiocyano group, and presumably with a lower irreversible reaction rate.
Voltage Dependence of Reversible Blockade by DIDS
An unexpected finding was that, although the I-V relationship for NBCe1-A itself is virtually linear (see tenidap data in Fig. 5A), the reversible inhibition of NBCe1-A by DIDS is voltage dependent. Voltage dependence can arise by three nonexclusive mechanisms: 1) conformational change (20), 2) Boltzmann effect (19, 20), and 3) knockoff (19).
Conformational change. A voltage-sensitive step in the reaction sequence (13, 18) of NBCe1-A could produce changes in DIDS affinity that could produce voltage-dependent inhibition.
Boltzmann effect.
If the DIDS binding site on NBCe1-A experienced a portion of the membrane's electrical field, then the local concentration of DIDS (which has two negative charges) near this binding site would be governed by a Boltzmann-type relationship
 | (5) |
is the DIDS concentration near the DIDS-binding site, which is at a point that corresponds to the fraction through the electrical field ( = 0 in the extracellular bulk solution and = 1 in the intracellular bulk solution). [DIDS]0 is the DIDS concentration in the extracellular bulk solution; z is the valence of DIDS (i.e., –2); F, R, and T have their usual meanings.
An ion channel is a pore with a single gate. When the gate is open, the flow of electrical current through the resistive pore causes the electrical potential to fall (an I·R drop) from near one end of the pore ( = 0) to the other ( = 1). The electrical field—along the axis of the pore—created by this current flow is, in general, not constant (i.e., voltage need not change linearly with physical distance). Imagine that NBCe1-A is a pore with two gates that can create an occluded state, in which the transported substrates (e.g., Na+ and CO3= or HCO3–) have access to neither the extra- nor the intracellular fluid. Because these gates are almost never both open at the same time, NBCe1-A would hardly ever conduct current from one bulk aqueous solution to the other through an open pore. Thus, no continuous I·R drop exists along the axis of the pore, and thus no sustained electrical field due to such an I·R drop.
Islas and Sigworth (21) have developed electrostatic models of the electrical field in the neighborhood of a membrane into which a cavity intrudes from one or both bulk aqueous solutions. Their simplest system was a membrane 30 Å thick with a right cylindrical cavity 6 Å in diameter and 25 Å deep (i.e., 83% of the way through the membrane).1 They calculate that would be 0.1 at the deepest point in the cavity. Extending the cavity to 90% of the way through the membrane would increase to only 0.13, as would reduction of the cavity diameter to nearly zero (assuming that the cavity is always filled with H2O). Given the size and hydrophilicity of DIDS, it is reasonable to expect that the cavity into which DIDS penetrates into NBCe1-A would be filled with water molecules, as required by the Islas-Sigworth model.
From the fractional-inhibition data between +20 and –40 mV—and assuming no competition between DIDS and other solutes—we used Eq. 5 to calculate values for our 11 constructs. As summarized in the last column of Table 1, these range from 0.08 for wild-type KKMIK to 0.20 for NKMIN and NNMIN. Given the uncertainties in the calculations, we regard this range of values as indistinguishable. It is interesting to note that Jennings et al. (22) also estimated a of 0.10–0.15 for the site at which extracellular SO
and Cl– compete on AE1 in red blood cells. According to the model of Islas and Sigworth—and assuming that the DIDS cavity in NBCe1-A is filled with water—it is possible that the computed of 0.1 could reflect the effect of the electrical field on the local [DIDS] near a binding site deep (i.e., 80–90%) into the NBCe1-A molecule as measured from the extracellular membrane surface. A site at such a depth would almost certainly have to correspond to a region of the pore that becomes occluded during transport, and the DIDS would have access to this site only with the extracellular gate open.
However, two lines of reasoning suggest that such a location of the DIDS—binding site-very close to the intracellular edge of the protein—is unrealistic. First, the KKMIK motif is generally regarded to be at the extracellular end of membrane spanning segment 5 (38). Indeed, the last K of the KKMIK motif is only 30 amino acids upstream from an Asn residue of NBCe1-A that is glycosylated when the other two consensus N-glycosylation sites are removed, and only 35 amino acids upstream from another Asn residue that is normally glycosylated (11).
Second, our calculation of values for NBCe1-A has assumed no competition between DIDS (a divalent anion) and HCO3–, CO3=, or another HCO3–-related species (e.g., the NaCO3– ion pair). That is, we assumed that, when we change the holding potential, alterations in the NBCe1-A current would be due entirely to a Boltzmann-like change in local [DIDS]. However, in AE1 of erythrocytes (15) as well as the Na+-driven Cl-HCO3 exchanger of squid axons (6), an HCO3–-related species appears to compete with DIDS. Let us assume that a HCO3–-related species also competes with DIDS for binding to NBCe1-A. When we change the holding potential, we would not only be altering the local [DIDS], but would be altering—in the same direction—the local concentration of the DIDS competitor, and thus be tempering the effect of the alteration in local [DIDS]. If the competitor were CO3=, as suggested by preliminary data (16), changes in holding potential would cause similar fractional changes in [DIDS] vs. [CO3=] and we would have observed little voltage dependence. If the competitor were HCO3– or NaCO3–, then changes in holding potential would have a greater effect on [DIDS] than on the monovalent competitor. To account for the observed voltage-dependent change in fractional inhibition, we would need a larger change in local [DIDS]. Thus the computed value would increase substantially, beyond the maximal value predicted by Islas and Sigworth, assuming that their model is applicable to our data.
In their SO4=/AE1 study, Jennings et al. (22) took into consideration the effects of voltage on the local concentrations of both [Cl–] and the competing [SO4=], and still estimated a value as high as 0.10–0.15.
Knockoff. The third possibility is knockoff or punchthrough, namely, that at very negative voltages (i.e., inward currents) exiting HCO3– or CO3= would displace DIDS from extracellular binding sites. This model remains a viable possibility for the effect of DIDS at very negative voltages, such as those shown in Figs. 4 and 6.
Inhibition by Tenidap
Adding CO2/HCO3– to a cell could potentially activate many currents: 1) currents through NBCe1-A, 2) HCO3– currents via other routes, and 3) currents activated by CO2 or the resulting decrease in pHi. Thus, the current in ND96 may not be a valid estimate of IBack, that is, the non-NBC current in CO2/HCO3–. At relatively positive voltages (e.g., +20 to –40 mV), where the I-V curves are rather divergent, the curve-fitting and the tenidap approaches yielded similar results for fractional inhibition and apparent Ki. This agreement confirms the usefulness of the curve-fitting approach, at least for relatively positive voltages. However, at more negative voltages, tenidap proved to be a useful tool for estimating IBack.
Others have concluded that 250 µM tenidap does not cause the appearance of other currents in oocytes (14), and that the inhibition of a Ca2+-activated anion conductance is also minimal (26). We found that, at very negative voltages, 1,000 µM tenidap did produce a slight change in the inward current. However, this effect is extremely small compared with the currents carried by NBCe1-A.
Although the interaction of tenidap with NBCe1-A was not the focus of the present study, we find it interesting that the apparent Ki of tenidap is not affected in a major way by mutating all three Lys residues in KKMIK to NNMIN. Thus it is likely that tenidap interacts with NBCe1-A at a site that is distinct from that at which DIDS interacts.
Reversible DIDS Binding Site
As anticipated, removing positive charges in the KKMIK motif of NBCe1-A reduced the apparent affinity of the transporter for the reversible action of DIDS. The largest effect that we observed was a 46-fold shift in Ki between the RRMIR (Ki = 18 µM) and EEMIE (Ki = 834 µM) mutants. Note that the Ki values reported in this paper are apparent values that pertain to a specific set of conditions, including a [HCO3–]o of 33 mM, a [Na+]o of 96 mM, a pHo of 7.50, and a temperature of 22°C in a Xenopus oocyte expression system.
Within the KKMIK motif, the second Lys residue appears to be the single most important of the three Lys residues for reversible DIDS interactions. Yet converting all three lysines to electroneutral residues—or even to negatively charged residues—produced a mutant transporter that still retained reversible DIDS sensitivity. Thus KKMIK cannot be the only motif important for reversible DIDS binding. As pointed out previously (36), certain SLC4 family members have conserved KXXK-like motifs—sometimes with Arg instead of Lys residues—at the putative extracellular ends of TM3 and TM13. These motifs and perhaps still others could contribute to the remnant DIDS binding in the NNMIN, DDMID, and EEMIE mutants of NBCe1-A. Perhaps by mutating all of the aforementioned conserved lysines or arginines would eliminate reversible DIDS sensitivity.
A striking observation was that even the most radical mutations of the KKMIK motif—such as converting all Lys residues to neutral Asn or anionic Glu—did not have substantial effect on the slope conductance of the transporter in the absence of DIDS (Table 1, column 3). For example, Ki increased by a factor of nearly 50 between RRMIR and EEMIE, but the slope conductance decreased by <20% (Table 1, columns 3 and 4). If the KKMIK motif were important for binding HCO3– or CO3= for subsequent transport into the cell, then the mutations that markedly reduced the apparent DIDS affinity should also have reduced the apparent HCO3– or CO3= affinity. Assuming that the apparent Km of NBCe1-A for extracellular HCO3– is 7 mM (17), then, at a [HCO3–]o of 33 mM, NBCe1-A should have been at 33/(33+7) or 83% of maximal velocity. Even a 10-fold reduction in HCO3– affinity should have reduced NBCe1-A to 33/(33+70) or 32% of its maximal velocity. A reduction from 83% to 32% of maximal velocity would correspond to a 60% reduction in slope conductance, a shift we would easily have detected. Moreover, our routine plate-reader assays of EGFP indicate that all of the clones expressed to similar extents.
Thus, the KKMIK motif is probably not essential for the electrogenic transport of HCO3– or CO3=. Then, why is a similar motif so highly conserved among members of the SLC4 family, not only in mammals, but also in species as distant as squid and flies? Because these organisms live at rather divergent temperatures, it is unlikely that the difference between room temperature (in our experiments) and 37°C (the physiological temperature of hNBCe1-A) materially affected the response to DIDS. Perhaps KKMIK participates in the binding of HCO3– or a similar anion for an as-yet unknown modulatory role.
Irreversible DIDS Inhibition
Figure 7 and Table 2 show that for KKMIK, RRMIR, NKMIK, KKMIN, KNMIK, and NNMIN, the time constants for irreversible DIDS blockade roughly parallel the apparent Ki for reversible DIDS binding (Table 1). Thus the irreversible interaction of DIDS with NBCe1-A could occur by the following two-step model:
 | (6) |
It is interesting that the triple mutants RRMIR, NNMIN, and DDMID all retain the ability to irreversibly block NBCe1-A, albeit at a rather low rate in the case of NNMIN and at an extremely low rate in the case of DDMID (Table 2). These data suggest that DIDS can covalently react with a Lys residue other than those in the KKMIK motif. One possibility is K924, which is part of a KSTV motif that is predicted to be at the extracellular end of TM12 (30)—based on the 13 TM model of AE1 proposed by Zhu et al. (38). Working with AE1 in RBCs, Okubo et al. (27) showed that H2DIDS at high pH can cross-link K851 (which corresponds to K924 in NBCe1-A) and K539 (which corresponds to the second K of KKMIK in NBCe1-A).
In conclusion, our data show for the first time that KKMIK motif is important for both reversible and irreversible DIDS blockade of NBCe1-A. The second of the three Lys residues is most important of the three for reversible DIDS blockade, which is voltage dependent. However, none of the Lys residues is essential for irreversible blockade, which is voltage independent. Elucidating the basis for DIDS sensitivity would not only help provide a structural underpinning for understanding why some SLC4 family members have low DIDS sensitivities, but also provide important topological information.
|
APPENDIX |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
 | (A1) |
We can also create an operational—and totally independent—definition of fractional inhibition from an experiment in the style of Fig. 3, A and B, or Fig. 5, A–C:
 | (A2) |
IAlbumin is the total current—the current produced by NBCe1-A plus the background current—with no inhibitor present (albumin scavenges free inhibitor). IBack is the true background current (i.e., all currents other than the NBCe1-A current). Thus (IAlbumin – IBack) is the current due to NBCe1-A in the absence of any inhibition. IDIDS is the total current—the remaining current produced by NBCe1-A plus the background current—in the presence of some known concentration of the inhibitor DIDS. Thus (IDIDS – IBack) is the remaining current due to NBCe1-A in the presence of DIDS.
Clearly, the two expressions for fractional inhibition must be equivalent. Therefore, by combining Eqs. A1 and A2 we have
 | (A3) |
This equation has two unknowns, Ki and IBack. However, if we have a series of data records, each consisting of one value of [DIDS] and one value of IDIDS, we can use a nonlinear, least-squares curve-fitting approach to simultaneously obtain best-fit values for both Ki and IBack. A complication is that IAlbumin will gradually fall during an experiment on a single oocyte as the irreversible inhibition by DIDS gradually inactivates a larger and larger fraction of NBCe1-A molecules. Therefore, in practice the algorithm must employ a separate value of IAlbumin for each data record. That is, each record consists of one [DIDS] value and one IDIDS value as well as the matched IAlbumin, which we assumed to be the average of IAlbumin immediately before applying DIDS (i.e., Albumin/pre-DIDS in Fig. 1A) and IAlbumin after removing DIDS and scavenging with albumin (i.e., Albumin/post-DIDS in Fig. 1A).
We wrote a computer program (available on request) in Microsoft Visual Basic 6.0 to execute the above algorithm for a series of ([DIDS], IDIDS, IAlbumin) records from a single oocyte, separately for each value of Vm. As noted in RESULTS, this algorithm is rather robust as long as the values of IAlbumin, IDIDS, and the true IBack are sufficiently different from one another, given the precision and accuracy of the measuring system. Indeed, when we manually computed fractional inhibition using the current remaining in the presence of the independent inhibitor tenidap (Itenidap) as an estimate of the true IBack, the results agreed well with those obtained by the curve-fitting procedure over a voltage range where the I-V curves were reasonably divergent (e.g., +20 to –80 mV in the present study).
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 They assumed a dielectric constant of 80 for the bulk aqueous solution and cavity, and 2 for the membrane. The assumed ionic strength was 318 mM. 
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXGRANTSREFERENCES |
|---|
2. Bartel D, Hans H, Passow H. Identification by site directed mutagenesis of Lys-558 as the covalent attachment site of dihydro DIDS in the mouse erythroid band 3 protein. Biochim Biophys Acta 985: 355–358, 1989.[Medline]
3. Bartel D, Lepke S, Layh-Schmitt G, Legrum B, Passow H. Anion transport in oocytes of Xenopus laevis induced by expression of mouse erythroid band 3 protein-encoding cRNA and of a cRNA derivative obtained by site-directed mutagenesis at the stilbene disulfonate binding site. EMBO J 8: 3601–3609, 1989.[ISI][Medline]
4. Bevensee MO, Schmitt BM, Choi I, Romero MF, Boron WF. An electrogenic Na/HCO3 cotransporter (NBC) with a novel C terminus, cloned from rat brain. Am J Physiol Cell Physiol 278: C1200–C1211, 2000.
5. Boron WF, Boulpaep EL. Intracellular pH regulation in the renal proximal tubule of the salamander: basolateral HCO3– transport. J Gen Physiol 81: 53–94, 1983.
6. Boron WF, Knakal RC. Intracellular pH-regulating mechanism of the squid axon: interaction between DNDS and extracellular Na+ and HCO3–. J Gen Physiol 93: 123–150, 1989.
7. Cabantchik ZI, Rothstein A. The nature of the membrane sites controlling anion permeability of human red blood cells as determined by studies with disulfonic stilbene derivatives. J Membrane Biol 10: 311–328, 1972.[CrossRef][ISI][Medline]
8. Cabantchik ZI, Rothstein A. Membrane proteins related to anion permeability of human red blood cells. I. Localization of disulfonic stilbene binding sites in proteins involved in permeation. J Membrane Biol 15: 207–226, 1974.[CrossRef][ISI][Medline]
9. Cabantchik ZI, Rothstein A. Membrane proteins related to anion permeability of human red blood cells. II. Effects of proteolytic enzymes on disulfonic stilbene sites of surface proteins. J Membrane Biol 15: 227–248, 1974.[CrossRef][ISI][Medline]
10. Choi I, Aalkjær C, Boulpaep EL, Boron WF. An electroneutral sodium/bicarbonate cotransporter NBCn1 and associated sodium channel. Nature 405: 571–575, 2000.[CrossRef]
