A Mammalian H+ Channel Generated Through Alternative Splicing of the NADPH Oxidase Homolog NOH-1

See allHide authors and affiliations

Science  07 Jan 2000:
Vol. 287, Issue 5450, pp. 138-142
DOI: 10.1126/science.287.5450.138

This article has a correction. Please see:


Voltage-gated proton (H+) channels are found in many human and animal tissues and play an important role in cellular defense against acidic stress. However, a molecular identification of these unique ion conductances has so far not been achieved. A 191–amino acid protein is described that, upon heterologous expression, has properties indistinguishable from those of native H+ channels. This protein is generated through alternative splicing of messenger RNA derived from the gene NOH-1 (NADPH oxidase homolog 1, where NADPH is the reduced form of nicotinamide adenine dinucleotide phosphate).

Voltage-gated H+channels (1) were first described in snail neurons (2) and were further characterized in a variety of mammalian cells (3–5). They are unique among ion channels with respect to their extremely high selectivity (5), marked temperature dependence (6), and unitary conductance, which is three orders of magnitude lower than that of most ion channels (7, 8). An H+ channel protein has not yet been identified, but Arg → His mutations were sufficient to turn the voltage sensor of the Shaker K+ channel into a voltage-gated H+ conductance (9). The critical residues (R/HVIR/HLVR/H) (10) are not found among known proteins, but are reminiscent of a motif (HSAIHTIAH) within the predicted third transmembrane domain of gp91phox, the electron-transporting subunit of the phagocyte NADPH oxidase (11,12).

By analogy with mitochondrial cytochromes, gp91phoxhad been postulated to conduct protons to preserve electroneutrality (13). This prediction was supported by pH measurements in phagocytes and in gp91phoxtransfectants (14–16), and a mutational analysis suggested a role for the histidines of the third transmembrane domain (17). However, normal H+ currents were observed in resting phagocytes from gp91phox-deficient patients (18), and a distinct type of H+ current was activated during assembly of the NADPH oxidase (19). Thus, gp91phox might conduct protons within an active oxidase complex, but a separate protein, possibly sharing the gp91phox histidine motif, mediates the H+ currents of resting phagocytes and other tissues.

A search of expressed sequence tag (EST) databases with the gp91phox third transmembrane domain yielded two cDNA clones coding for a short NH2-terminal gp91phoxhomolog, which ended with 35 unrelated amino acids. A further search with the COOH-terminus of gp91phox yielded another cDNA sequence, which suggested the presence of a longer homolog. A search of genomic DNA databases with the EST clones yielded a common gene (20), localized to the long arm of the human X chromosome (Xq22, Fig. 1), which we termed NOH-1 (NADPH oxidase homolog 1). Exon 1 (homologous to exon 1 of gp91phox) contained a presumable translation initiation site, whereas exon 13 (homologous to exon 13 of gp91phox) contained a stop codon. In 3′ position from exon 13, we identified an additional exon (exon 14) corresponding to the unique COOH-terminus of the short NH2-terminal homolog.

Figure 1

The NOH-1 gene generates NOH-1S and NOH-1L through alternative splicing. The NOH-1 gene, as found in the nonredundant database, localizes to Xq22. The CYBB gene (coding for gp91phox) localizes to Xp21. Exons are shown as horizontal bars; introns are shown as thin vertical lines. We have identified 14 exons of the NOH-1 gene; only 13 exons are known for the CYBB gene. Exons 1 to 13 constitute NOH-1L; exons 1 to 5 and exon 14 constitute NOH-1S. gp91phoxconsists of 13 exons, all with lengths similar to the 13 exons of NOH-1L (boxed numbers, coding nucleotides). Arrow indicates the last exon coding the unique NOH-1S COOH-terminus.

We next used the reverse transcription polymerase chain reaction (RT-PCR) to amplify the expected products of the NOH-1 gene (21). An exon 1–exon 13 primer pair yielded two PCR products (∼1.75 and ∼1.6 kb) in the colon carcinoma cell line CaCo-2 (Fig. 2A). An exon 1–exon 14 primer pair yielded a ∼0.5-kb band in CaCo-2 cells, HL-60 cells, and leukocytes, but not in HEK-293 cells (Fig. 2A), heart, and skeletal muscle myotubes (22). The amino acid sequences of the threeNOH-1 products, NOH-1S (short), NOH-1L [long (23)], and NOH-1Lv (long variant, corresponding to the 1.6-kb PCR product), are shown in Fig. 3A (24). With the use of a ribonuclease (RNase) protection assay (25), NOH-1L was detected in colon, uterus, prostate, and CaCo-2 cells, whereas NOH-1S was detected only in colon and CaCo-2 cells (Fig. 2, A and B). The positive NOH-1S PCR results were confirmed by nested PCR and by sequencing (22), which suggested that NOH-1S was indeed present in HL-60 cells and leukocytes but in amounts below the detection threshold of the RNase protection assay.

Figure 2

(A) Left panel: RT-PCR detection of mRNA coding for NOH-1L (lanes 2, 4, 6, and 8) and NOH-1S (lanes 3, 5, 7, and 9). Lane 1, DNA length markers. Right panel: RNase protection assay using a probe protecting 355 nucleotides of NOH-1S but only 240 nucleotides of NOH-1L. The undigested NOH-1 and GAPDH probes are labeled with asterisks (25). HL-60 cells were differentiated with DMSO. Human peripheral blood leukocytes (Leuk.) were purified as described (30) and included both lymphocytes and granulocytes. (B) Tissue distribution of NOH-1L and NOH-1S assessed by RNase protection assay using the probes described in (A).

Figure 3

Structure of NOH-1 and sequence comparison. (A) The deduced amino acid sequence of NOH-1S (GenBank accession number AF166326) and NOH-1L (GenBank accession number AF166327) aligned with the sequence of gp91phox. Shaded boxes indicate identical residues. Presumed membrane-spanning regions are overlined. The fourth transmembrane region of NOH-1S is predicted in positions 161 to 181, whereas it is predicted in positions 173 to 193 for NOH-1L. The four boxed histidines in NOH-1L and gp91phox are the conserved heme-spanning residues of heme cytochromes (31). The underlined sequences show the FAD-binding region (residues 338 to 344 of gp91phox) and the NADP pyrophosphate, NADP ribose, and NADP 2′-phosphate-binding regions and the nicotinamide C-4 atom approaching site (32) (residues 406 to 416, 442 to 446, 504 to 513, and 535 to 538 of gp91phox). The amino acids absent in NOH-1Lv (GenBank accession number AF166328) are in italics (residues 433 to 481). The NOH-1S EST clone (GenBank accession number AI821410) contained the complete coding sequence shown here. (B) Model of the transmembrane topology and of putative functional domains of NOH-1S and NOH-1L. The transmembrane helices, NH2- and COOH-termini, conserved histidines, and FAD and NADPH binding sites are indicated.

On the basis of the sequence information, we identified the exons ofNOH-1 gene that generate NOH-1S and NOH-1L (Fig. 1). Exons 1 to 5 code for the common NH2-terminal 158 amino acids of NOH-1L and NOH-1S. Exons 6 to 13 code for the COOH-terminal 406 amino acids of NOH-1L, with exon 11 missing in NOH-1Lv. Finally, exon 14 codes for the COOH-terminus of NOH-1S. Note that the splice donor site for the generation of NOH-1S is not located at the end of exon 5, but within the exon. The gp91phox geneCYBB localizes to the short arm of the X chromosome (Xp21). The size of the corresponding exons is conserved betweenNOH-1 and CYBB, with the exception of exon 14, for which no corresponding sequence in CYBB is known. However, the lengths of the introns are markedly different. Thus, the presence of the two homologous genes is most likely due to a relatively ancient gene duplication.

Hydropathy plots showed a similar profile for NOH-1L and gp91phox (22). Predicted transmembrane topology models of NOH-1L and NOH-1S are shown in Fig. 3B. NOH-1L appears similar to gp91phox, with a short cytosolic NH2-terminus, six transmembrane domains, and a long cytosolic COOH-terminus. NOH-1S shares the cytosolic NH2-terminus and the first three transmembrane domains with NOH-1L, but terminates with a distinct fourth transmembrane domain followed by a short, intracellular COOH-terminus. NOH-1L contains a heme-spanning histidine motif as well as flavin adenine dinucleotide (FAD) and NADPH binding regions, consistent with its oxidoreductase function (23). In contrast, NOH-1S only retains the histidine-rich transmembrane motif postulated to generate voltage-dependent H+ currents (Fig. 3B).

To investigate whether NOH-1S functions as an H+ channel, we stably expressed the protein in HEK-293 cells and chose experimental conditions to isolate H+ currents (4, 5,26). In the absence of a cell line completely devoid of H+ currents, HEK-293 cells provided a good expression system, as they had only very small H+ currents (Fig. 4, B and D) and did not express theNOH-1 isoforms (Fig. 2A). Cells were acidified through the patch pipette to maximize H+ currents, and depolarizing voltage pulses were applied. No currents above background levels were observed in HEK-293 cells stably transfected with the empty vector (mock-transfectants) or with an unrelated channel protein, hTRP4 (Fig. 4). In contrast, voltage-dependent outward currents were observed in two independent NOH-1S–expressing clones. The currents activated slowly upon depolarization, and several seconds were required to elicit maximal amplitude, as expected for voltage-dependent H+currents of epithelial cells and phagocytes (27). The NOH-1S currents were reversibly blocked by Zn2+ (100 μM), a known H+ channel inhibitor (5) (Fig. 4C). The experiments are summarized in Fig. 4D. Nontransfected, mock- and hTRP4-transfected cells had only minor currents, whereas cells stably transfected with NOH-1S had large currents that were blocked by Zn2+.

Figure 4

Cells expressing NOH-1S have voltage-activated, pH-dependent, Zn2+-sensitive currents. (A and B) Currents in NOH-1S–expressing cells and in mock-transfected cells, elicited by 3-s depolarizing voltage steps ranging from −40 to +80 mV (inset); pHi 5.7, pHo 7.5. (C) The currents in NOH-1S–expressing cells were reversibly inhibited by 100 μM Zn2+. (D) Current densities in wild-type cells (WT) and clones transfected with the empty vector (B4), a calcium channel (hTRP4), or NOH-1S (D3, E9). Zn2+ (100 μM) reduced the currents to background levels in NOH-1S–expressing cells. Data are means ± SE of leak-subtracted currents measured at +60 mV, pHi/pHo 5.7/7.5 (NS: not significant versus WT; *P < 0.00004 versus WT, unpairedt test). (E) Families of currents recorded at the indicated pipette pH values in NOH-1S–expressing cells (pHo 7.5). (F) Current-voltage relation of theNOH-1–dependent currents recorded at the indicated pipette pH values (means ± SE from 5 to 19 individual cells).

Low intracellular pH generally activates H+ channels and shifts their threshold of voltage activation toward more negative values. We therefore performed experiments using pipette solutions of differing pH. The amplitudes of the currents were larger and the thresholds of voltage activation were lower as the cytosol was more acidified (Fig. 4, E and F).

The NOH-1S currents share several characteristics of previously described H+ currents: activation by depolarization and cytosolic acidification, slow kinetics of activation, and reversible block by Zn2+. To directly demonstrate that the observed currents were carried by H+ ions, we investigated the selectivity of the conductance by tail current analysis. The reversal potential of the current, E rev, depended on the pH gradient and, within the physiological pH range, changed by 46 mV per pH unit (Fig. 5, A and B). At very low pipette pH, E rev deviated from the proton equilibrium potential (E H) but was not affected by substitution of K+ or Na+ in the extracellular solution (Fig. 5B). The 35-mV deviation ofE rev from E H at pHi 5.7 most likely reflects an imperfect submembranous pH clamp in the presence of large outward currents. However, even assuming that the deviation was due to permeation of other ions, the relative H+ permeability is still very large [p H/p Cs > 106 (28)] given the low concentration of H+ ions. This degree of selectivity is three orders of magnitude higher than Ca2+ channels [p Ca/p Cs = 1:4200 (29)] and in the same range as native H+ channels (27).

Figure 5

The NOH-1S currents are H+selective and alter cellular pH. (A) The reversal potentials of tail currents were measured upon depolarization to various test potentials after a 2-s depolarization to +60 mV (pHi 5.7, pHo 7.5). (B) Reversal potentials measured in solutions containing Cs+, Na+, or K+, plotted against pipette pH; pHo 7.5. Values are means ± SE of 3 to 12 experiments. The dotted line shows the H+ equilibrium potential. (C and D) Combined recordings of whole-cell currents and of cytosolic pH, measured with carboxy–SNARF-1 in NOH-1S and mock-transfected HEK-293 cells. Long-lasting depolarizing steps to +60 mV were applied (middle), and the currents (bottom) and cytosolic pH changes (top) were measured concomitantly (pHi6.5, pHo 7.5). Cytosolic pH changes were measured with carboxy-SNARF-1 as described (4). HEK-293 cells were incubated with 5 μM carboxy–SNARF-1 acetoxymethyl ester for 20 min at room temperature just before recordings in the whole-cell patch-clamp configuration. To compensate for the diffusion of the dye into the patch pipette, we included 100 μM carboxy–SNARF-1 (free acid) in the pipette solution. Data are expressed as the ratio of carboxy-SNARF-1 emission at 640 nm to emission at 580 nm.

Finally, we investigated whether the NOH-1S–associated H+channel participates in cytosolic pH homeostasis by measuring intracellular pH during activation of the currents. Cells were patch-clamped in the whole-cell configuration, and cytosolic pH changes were measured with the fluorescent pH indicator carboxy–SNARF-1. As shown in Fig. 5, C and D, the activation of voltage-dependent H+ currents by long-lasting depolarization from −60 mV to +60 mV led to an increase of cytosolic pH in NOH-1S–expressing cells. In contrast, no pH changes and no H+ currents above background levels were detected in mock-transfectants. Thus, NOH-1S is able to participate in the regulation of cellular pH and can extrude H+ ions under intracellular acidic stress.

NOH-1S–expressing HEK-293 cells display electrophysiological properties similar to previously described voltage-gated H+channels. On the basis of current kinetics, DeCoursey has suggested the presence of at least four different types of H+ channels (27); the relatively slow time constant of NOH-1S current activation resembles the epithelial type and the phagocyte type. At this point we cannot exclude the possibility that NOH-1S is an H+ channel regulator, rather than the channel itself. However, H+ conduction appears to be a general property of several proteins containing an intramembranous histidine motif, as demonstrated by (i) pH measurements in cells expressing full-length or truncated gp91phox (14–16), (ii) studies with the mutated Shaker voltage sensor (9), and (iii) studies demonstrating H+ currents in cells expressing NOH-1L and gp91phox (22). The fact that NOH-1S does not contain an electron transport chain suggests that H+ conductance is its main physiological function, whereas flavocytochromes such as NOH-1L or gp91phox might conduct H+ ions as part of their electron transport mechanism.

Given its limited tissue expression, it is clear that NOH-1S should only be considered as a prototype H+ channel. We recently identified two additional gp91phox homologs [NOH-2 and NOH-3 (22)] that have a high degree of conservation within the histidine-rich third transmembrane domain. Thus, new candidate H+ channel proteins are already emerging and should provide new insights into H+ channel heterogeneity.

Note added in proof: A recent publication demonstrates H+ currents in cells transfected with gp91phox(33).

  • * To whom correspondence should be addressed. E-mail: Nicolas.Demaurex{at}


View Abstract

Navigate This Article