Cytosolic malate dehydrogenase confers selectivity of the nucleic acid-conducting channel

Methods

Results

We have shown previously that nucleic acid-affinity chromatography resulted in purification of a highly specific nucleic acid-conducting channel (1). Recent efforts to increase purity and yield have led to the development of a four-step FPLC protocol. Detergent solublized rat-kidney brush border membranes (Fig. 1A, lane 1) were applied to two sequential ion-exchange columns. Fractions with NACh activity were pooled, analyzed by SDS/PAGE (Fig. 1A, Lane 2), and purified further by hydrophobic interaction chromatography. The active peak from this separation contained two protein bands; a 45-kDa band (p45) and a second band at 36-kDa (p36; Fig. 1A, lane 3). The bands were separated by addition of a reducing agent (DTT, 10 mM) and separation by gel filtration chromatography. This separation resulted in two protein peaks as detected by UV spectroscopy (data not shown) that corresponded to p45 (Fig. 1A, lane 4) and p36. Nucleic acid-dependent channel activity was lost by this step of purification, suggesting the nucleic acid channel is a heteromultimer of p45 and p36. Yield of p36 was low in this step of purification and consequently the amount of protein recovered was at the lower limit of detection on silver-stained gels.

To test the hypothesis that the nucleic acid channel is a heteromultimer of p45 and p36, fractions containing either p36 or p45 were used to form proteoliposomes and channel activity was assessed in planar lipid bilayers. These experiments are summarized in Fig. 1 B–E. Fig. 1B depicts current traces from a single experiment (Upper) and the mean current-voltage plot (IV-plot) from six experiments (Lower) when p45 was reconstituted alone and ODN was not present. Under these conditions, significant current was observed in the absence of ODN indicating that p45 forms an ion-conducting channel, which conducts ion(s) other than ODN. Furthermore, transitions in current level were rare. This is exemplified in Fig. 1B Upper where transitions were observed only in the trace collected at a holding potential of +100 mV. It is unclear whether these intermediate levels of current represent subconductive states of a single channel or whether multiple channels are present in the bilayer. In analyzing the data from each experiment, we have assumed they result from the presence of multiple channels in the bilayer, and therefore, the transitions represent single channel gating, and unitary conductance can be determined from these transitions. Consequently, if transitions were observed in any of the traces in an experiment, these transitions were used to define single channel current. All of the traces from that experiment were then factored by this single channel current, and unitary conductance was determined. The data in Fig. 1B (Lower), which have been adjusted in this way, shows that the current-voltage relationship of p45 was linear from −50 to +50 and reached a plateau at holding potentials above +50 mV and below −50 mV, suggesting saturation of the channel. Channel conductance (slope conductance) between −50 mV and +50 mV was 13.7 ± 2.5 pS (n = 6).

In several experiments, p45 was reconstituted, channel activity was recorded, ODN was then added to both solution chambers, and channel activity was again recorded. In these experiments, addition of ODN resulted in small fluctuations in current (data not shown) indicating some form of interaction between ODN and p45.

Because we have reported previously that NACh activity is blocked by heparan sulfate (1), we examined possible interactions between p45 and heparan sulfate (Fig. 1C). Under control conditions, channel open probability of p45 in the absence of ODN was 1.0 (Fig. 1C, top trace). Addition of 10-μg/ml heparan sulfate decreased open probability to 0.88 (Fig. 1C, middle trace) and 20 μg/ml blocked channel activity completely (Fig. 1C, bottom trace). This dose-response is identical to that previously reported (1) and indicates that p45 is a component of NACh. Furthermore, these data suggest that heparan sulfate blocks nucleic acid channel activity in this system by binding to p45.

Ion gradient and ion substitution studies were performed to identify the ion(s) conducted by p45. In symmetrical solution of 200 mM KCl, 1 mM CaCl₂, and 10 mM Hepes, pH 7.4 (Fig. 1D, ●), channel conductance was 11 pS with a reversal potential (E_rev) of ≈0 mV. When a 10:1 gradient for KCl was created (cis to trans), E_rev was +32 mV (Fig. 1D, ▪). The Goldman-Hodgkin-Katz equation predicts the channel is five times more permeable to anions than cations, indicating that an anion (i.e., Cl⁻) is preferentially conducted through the channel. When the buffered solution in the trans chamber was changed to 200 mM CsCl, thus eliminating the Cl⁻ gradient, E_rev shifted back to zero. Together, these data indicate that although p45 conducts several ions, it is somewhat more selective for anions than cations.

When a single fraction containing both p45 and p36 was reconstituted (Fig. 1E), channel activity was seen only when ODN was present. The channel characteristics (conductance and kinetics) were identical to the previously reported characteristics of NACh (1, 7).

To characterize the functional role of p36 in NACh activity, p36 was reconstituted alone or in combination with p45. Current was not seen when p36 was reconstituted alone in either the absence or presence of ODN (data not shown), suggesting that p36 alone does not form a channel. When p45-containing proteoliposomes were reconstituted, the expected ODN-independent activity was observed (Fig. 2A). The channel closed on addition of p36-containing proteoliposomes (Fig. 2B), and current was not observed until ODN was added (Fig. 2C). Channel activity observed after reconstitution of p36-proteoliposomes with p45-proteoliposomes was comparable to activity observed after reconstitution of a single FPLC fraction, which contained both proteins (Fig. 1E), and to the activity of the channel complex as previously described (1). These data support the hypothesis that the nucleic acid channel is a heteromultimeric protein complex consisting of at least two proteins, p45 and p36. P45 forms a nonselective transmembrane pore that is more permeable to anions than cations. The 36-kDa subunit, which does not have channel activity when reconstituted alone, is a regulatory subunit that interacts with p45 to form the nucleic acid specific channel.

To identify p45 and p36, the proteins were purified by two-dimensional gel electrophoresis and sequenced by MS. These efforts resulted in several peptide sequences each from p45 and p36. The sequences from p45 do not match any sequence in searchable databases; whereas, the sequences from p36 are identical to rat cytosolic malate dehydrogenase (cMDH). Alignment of p36 sequences with rat cMDH is shown in Fig. 3A. We used hydropathy analysis (11) and tmpred software (12), which is an algorithm that predicts membrane-spanning domains, to assess the potential that cMDH is an integral membrane protein. Hydropathy analysis demonstrates highly hydrophobic domains from amino acids 8–21 and 37–53 (Fig. 3B) and tmpred predicts putative transmembrane-spanning domains from amino acids 7–25 and 35–53 (Fig. 3C). These computational analyses suggest that cMDH may be an integral membrane protein and as such we would hypothesize that cMDH interacts with p45 to form the nucleic acid channel.

Experiments were performed to test this hypothesis and confirm that cMDH is a subunit of the nucleic acid channel. First, we explored whether l-malate alters channel function (Fig. 4A). Reconstitution of a single fraction containing both p45 and p36 resulted in ODN-dependent channel activity (Fig. 4A Top). At least four channels are present in the trace shown. Addition of 20 μM l-malate to both solution chambers resulted in a decrease in channel open probability from 0.71 to 0.60 (Fig. 4A Middle). Increasing l-malate concentration to 25 μM resulted in a further decrease in open probability to 0.43 (Fig. 4A Bottom). Open probability increased when l-malate was washed out (data not shown). In four experiments, 20 μM l-malate resulted in a mean decrease in open probability of 79 ± 11% (P < 0.05 vs. control). To validate the l-malate blockade, these open probability data were determined from 3-min current traces. Solution pH did not change after the addition of l-malate. To ensure the effect of l-malate was specific to NACh activity (i.e., specific to the activity of the p45-p36 channel complex), l-malate effects on p45 were examined (n = 4). In the representative experiment shown in Fig. 4B, symmetrical l-malate at concentrations as high as 30 μM had no effect on open probability of p45-alone or its conductance (Fig. 4B Bottom) when compared to zero malate control (Fig. 4B Top). These data demonstrate that NACh activity is inhibited by l-malate and that this inhibition requires the presence of the 36-kDa subunit.

We generated a polyclonal antiserum against pig heart cMDH (Sigma) in a mouse at the Mount Sinai School of Medicine Hybridoma Core Facility by using previously publish methodology (13). As seen in Fig. 5A, the animal mounted a strong immune response against cMDH as indicated by a titer of >1:10,000 by ELISA. In Western blots of pig heart cMDH, this serum primarily recognized the 36-kDa pig heart cMDH as well as two larger molecular weight products present in the commercial cMDH preparation (Fig. 5B, lane 1). The serum was immunoreactive with only a single protein that comigrated with a cMDH control when blots of whole kidney lysate were probed (Fig. 5B, lane 2). We then tested the ability of the antiserum to alter NACh activity. Antiserum at a dilution of 1:200 caused a mean decrease in channel open probability of 80 ± 13% (P < 0.05 vs. control, n = 6). In these, open probability data also were determined from an entire 3-min current trace. Normal serum, at the same dilutions, had no effect on channel open probability (data not shown). The results from a single experiment are depicted in Fig. 5C.

These data suggest that cMDH is the regulatory subunit of the nucleic acid channel. To provide direct evidence for this hypothesis, proteoliposomes were formed with either p45 or pig heart cMDH (Sigma), fused with lipid bilayers either individually or together, and channel activity was characterized. To form cMDH proteoliposomes, ≈20 μg of cMDH (Sigma) was sonicated with a 1:1 mixture of phosphatidylethanolamine and phosphatidylserine in a volume of 25 μl. Five to ten microliters was then added to the trans solution cup and allowed to fuse with the preformed lipid bilayer. Reconstitution with p45 alone resulted in the expected nucleic acid-independent channel activity (Fig. 6A) previously described in Fig. 2. Channel activity was not observed after reconstitution of pig heart cMDH alone either in the presence (Fig. 6B) or absence (data not shown) of ODN. Channel activity was not seen when p45 and commercial cMDH were reconstituted together in the absence of ODN (Fig. 6C). When 10 μM ODN was added, however, channel gating was observed (Fig. 6D). Channel conductance of reconstituted p45 and pig heart cMDH was 12.4 ± 3.9 pS (n = 11), which is comparable to the conductance of the ODN-gated channel published previously (1). In control experiments, proteoliposomes containing mitochondrial MDH did not restore nucleic acid selectivity to p45 (data not shown). These data provide direct evidence that cMDH is the regulatory subunit of the nucleic acid channel.

Finally, cMDH has previously been described as a globular cytosolic protein and has not been considered to be associated with cell membranes (14). To function as the subunit of a plasma membrane channel complex, cMDH must interact with p45 and be in close proximity to the cell membrane. To demonstrate proximity of cMDH with cell membranes, anti-cMDH antiserum was used to immunolocalize cMDH in LLC-PK cells, a cell line derived from pig kidney proximal tubule. We have recently purified NACh from this cell line (data not shown). LLC-PK cells were grown on acid-washed, collagen-coated coverslip, fixed, permeabilized, immunostained, and viewed by confocal microscopy. As seen in Fig. 5D, the pattern of immunoreactivity is reticular, suggesting intracellular membrane association. In addition, staining on the margin of cells (arrowheads) is also detected, which is consistent with localization of cMDH to the plasma membrane. Preimmune sera were not immunoreactive with LLC-PK cells (data not shown). These data, together with the computational analyses shown in Fig. 3 B and C, support the hypothesis that cMDH associates with p45 at the cell membrane to regulate channel activity.

Discussion

The data presented herein demonstrate that the nucleic acid conducting channel is a complex of at least two proteins. One of these proteins has been identified as cMDH, whereas the second protein, p45, has yet to be identified. These data indicate that p45 and cMDH must both be present to reconstitute nucleic acid channel function. When we first identified and characterized the nucleic acid channel (1), p36 (i.e., cMDH) was not detected and therefore NACh activity had been attributed to a single protein, p45. Subsequent to a significant scale-up of the purification methodology, p36 was seen to copurify with p45. These experiments were initiated to determine whether p36 was an artifact of purification or whether it participates in nucleic acid channel function. As detailed above, the data support the hypothesis that p36 is a critical component of the channel. We would contend, based on the striking difference between the channel activity of p45 alone and that of the p45-cMDH channel complex, that p36 was present, although not detected on gels, in the original affinity-purified samples.

MDH is a highly conserved, ubiquitous enzyme present in bacteria, plants, and animals. In mammalian cells there are two isoforms; a mitochondrial isoform (mMDH) and a “cytosolic isoform” (cMDH). Both isoforms are NAD-dependent enzymes that catalyze l-malate to oxaloacetate. Despite similar enzymatic function, the isoforms share little homology at the amino acid level (≈20% sequence identity and <35% homology (15). Sequence analysis of mdh genes from several species indicate that the cytosolic and mitochondrial isoforms resulted from a gene duplication before the evolution of the eukaryotic cell (14, 16). Although the biochemical role of mitochondrial MDH is well documented, the functional significance of cMDH remains unclear. The genetic divergence of the two genes, however, would suggest that they might have distinct biochemical or physiologic functions. The studies detailed herein provide compelling evidence for an additional, novel function of cMDH.

It is also interesting to note that the enzymatic reaction catalyzed by MDH is thought to be an obligate-ordered reaction, which requires NAD interaction with the enzyme before malate can bind. In our experiments, l-malate blocks channel activity in the absence of NAD suggesting that p45 results in a conformational change in cMDH, which opens the malate binding site. Conversely, it is possible that NAD purifies with cMDH as has been suggested for the β-subunit of the voltage-dependent K⁺ channel (17).

Because p45, when reconstituted alone, is nonselective and essentially constitutively open, its presence in the plasma membrane would lead to a dissipation of the charge separation across the membrane, and as such would be incompatible with cell viability. We would hypothesize, therefore, that p45 cannot exist in the plasma membrane alone and must be in complex with a regulatory subunit. Our data suggest that cMDH plays this regulatory role in the nucleic acid channel complex.

In summary, we have provided evidence for an important function of cMDH. We have shown that the nucleic acid channel is a heteromultimer consisting of a 45-kDa subunit and a 36-kDa subunit; p45 forms a nonselective transmembrane pore which is more permeable to anions than cations whereas p36, which has no channel activity when reconstituted alone, is a regulatory subunit that regulates channel selectivity and gating kinetics. Protein sequence data and functional reconstitution studies have shown that cMDH is the regulatory subunit of NACh. Finally, we have localized cMDH to cell membranes by immunofluorescence. The cellular function of cMDH is not completely understood, and its involvement as a component of a plasma membrane transport complex has not, until now, been described. Although such a role for cMDH has not been described, other enzymes have been shown to participate in channel function. Two notable examples are a recent report that the beta subunit of voltage-dependent potassium channels is a putative oxido-reductase enzyme (17) and the observation that a splice variant of the NADPH oxidase homolog-1 gene (NOH-1) forms an H⁺ channel when expressed in HEK-293 cells (18). The current observations together with the findings of these other studies suggest an emerging duality of function for some metabolic enzymes and it is intriguing that these newly described functions are related to regulation of cell membrane permeability.