Cloning and characterization of novel human SLC4A8 gene products encoding Na⁺-driven Cl⁻/HCO₃⁻ exchanger variants NDCBE-A, -C, and -D

Abstract

The reported sequences of the human and mouse Na⁺-driven Cl⁻/HCO₃⁻ exchangers (NDCBEs) differ greatly in their extreme cytosolic COOH termini (Ct). In human NDCBE (NDCBE-B), a 17-amino acid (aa) sequence replaces 66 aa at the equivalent position in mouse NDCBE (NDCBE-A). We performed 5′- and 3′-rapid amplification of cDNA ends (RACE) on human brain cDNA, followed by PCR of full-length cDNAs to determine whether the human SLC4A8 gene was capable of producing the mouselike Ct sequence. Our study confirmed the presence in human cDNA of mouse NDCBE-like transcripts (human NDCBE-A) and also disclosed the existence of three further novel NDCBE transcripts that we have called NDCBE-C, NDCBE-D, and NDCBE-D′. The novel NDCBE-C/D/D′ transcripts initiate at a novel “exon 0” positioned ∼35 kb upstream of the first exon of NDCBE-A/B. NDCBE-C/D/D′ protein products are predicted to be truncated by 54 aa in the cytosolic NH₂ terminus (Nt) compared with NDCBE-A/B. Our data, combined with a new in silico analysis of partial transcripts reported by others in the region of the human SLC4A8 gene, increase the known extent of the SLC4A8 gene by 49 kb, to 124 kb. A functional comparison of NDCBE-A/B/C/D expressed in Xenopus oocytes demonstrates that the Nt variation does not affect the basal functional expression of NDCBE, but those with the shorter Ct have a 25–50% reduced functional expression compared with those with the longer Ct. By comparison with an artificially truncated NDCBE that contains neither 17-aa nor 66-aa Ct cassette, we determined that the functional difference is unrelated to the 66-aa cassette of NDCBE-A/C, but is instead due to an inhibitory effect of the 17-aa cassette of NDCBE-B/D.

electroneutral sodium-driven chloride/bicarbonate exchanger activity was first described in the snail neuron (38, 39) and squid giant axon (6, 7, 10, 33), as well as in giant barnacle muscle fiber (8, 9, 11, 34), crayfish neuron (22), proximal tubules of mudpuppies (17) and rats (2), and a Na⁺/H⁺ exchanger-deficient hamster fibroblast line (20). In these preparations, introducing an extracellular solution containing CO₂/HCO₃⁻ causes an acidification (due to the influx of CO₂) followed by intracellular pH (pH_i) recovery (reflecting acid extrusion). The investigators showed that the pH_i recovery is blocked by disulfonic stilbenes (e.g., DIDS, SITS) and requires extracellular Na⁺ and HCO₃⁻. Furthermore, the uptake of HCO₃⁻ (or a related ion such as CO₃⁼) in these cells has a trans-side Cl⁻ dependence and is accompanied either by an efflux of ³⁶Cl⁻ or a transient fall in intracellular Cl⁻ concentration. Finally, HCO₃⁻ uptake is accompanied by either an influx of ²²Na⁺ or a transient rise in intracellular Na⁺ concentration. These data defined the existence of a Na⁺-driven Cl⁻/HCO₃⁻ exchanger. Later, our laboratory (35) functionally demonstrated also that Na⁺-driven Cl⁻/HCO₃⁻ exchange is the most potent acid extrusion mechanism in freshly dissociated rat hippocampal neurons.

In 1997 Romero et al. (32) cloned the first Na⁺-coupled HCO₃⁻ transporter (NCBT), the electrogenic NaHCO₃⁻ cotransporter NBCe1 from salamander proximal tubule. In the intervening decade, our knowledge of the genes and splice variants that encode NCBTs has greatly increased, such that we now recognize five mammalian NCBTs, each with multiple splice variants (reviewed in Refs. 24 and 31). For example, the human genome has two electrogenic NBCs (NBCe1/SLC4A4 and NBCe2/SLC4A5), two electroneutral NBCs (NBCn1/SLC4A7 and NBCn2/SLC4A10), and a single Na⁺-driven Cl⁻/HCO₃⁻ exchanger (NDCBE/SLC4A8).

The first full-length SLC4A8 gene product (since designated NDCBE-B by GenBank) was cloned in our laboratory from human brain cDNA (16). The second full-length slc4a8 gene product (since designated NDCBE-A by GenBank) was cloned from mouse kidney cDNA (44). The human NDCBE-B transcript encodes a protein of 1,044 amino acids (aa). In mouse NDCBE-A, 66 aa replace the 17 aa at the extreme COOH terminus (Ct) of NDCBE-B [human (dashed) curve vs. mouse (solid) curve in Fig. 1]. Experiments in which NDCBE-B was functionally expressed in Xenopus oocytes provided the definitive characterization of the SLC4A8 gene product as a Na⁺-driven Cl⁻/HCO₃⁻ exchanger (16). First, the pH_i recovery from an acid load required extracellular HCO₃⁻ and Na⁺ and was blocked by DIDS. Second, when the direction of net transport was reversed by the removal of bath Na⁺ (i.e., converting a pH_i recovery/increase to a decrease), subsequently removing bath Cl⁻ blocked the reversed transporter. Finally, oocytes expressing NDCBE-B mediated a ³⁶Cl efflux that required extracellular Na⁺ and CO₂/HCO₃⁻. More recent work employing an extracellular surface-chloride electrode demonstrates a net efflux of Cl⁻ when NDCBE-B operates in the forward direction (26). Our laboratory has also recently demonstrated (14) that NDCBE protein is immunolocalized in the cell bodies and processes of rat hippocampal neurons.

Fig. 1.

Major differences between “human” and “mouse” Na⁺-driven Cl⁻/HCO₃⁻ exchanger (NDCBE). A: scale representation of the putative topology of NDCBE showing the cytosolic NH₂ terminus (Nt) and COOH terminus (Ct). Transmembrane spans are depicted as numbered boxes. Major differences between the originally described human and mouse NDCBEs are confined to the Ct region. The human NDCBE terminates in a 17-amino acid (aa) cassette (H, dashed curve) that is replaced in the mouse Ct (M, solid curve) by a distinct 66-aa cassette. B: alignment of the putative cytoplasmic Ct polypeptide sequences of human and mouse NDCBE. The first 34 aa of human and mouse NDCBE Ct are identical except for the 2 residues shaded in gray.

In the present study, we set out to determine whether NDCBE-A can be found in humans and, given the multiplicity of splice variants reported for other mammalian NCBTs, whether any entirely novel NDCBE splice variants can be detected in human brain cDNA. Employing a combination of 5′- and 3′-rapid amplification of cDNA ends (RACE)—together with PCR of full-length transcripts from human brain, kidney, and heart cDNA libraries—we were able to detect human NDCBE-A and NDCBE-B. In addition, we detected three entirely novel cDNA variants (NDCBE-C, -D, and -D′) that encode two transporters (D and D′ have identical deduced amino acid sequences). Compared with NDCBE-A and -B, NDCBE-C and -D are truncated at the NH₂ terminus (Nt), probably because they are transcribed from an alternative promoter. NDCBE-C and -D differ from each other by having alternate Ct that correspond to NDCBE-A and -B, respectively. In human heart cDNA, we detected a second NDCBE-D transcript (NDCBE-D′), with a variant 5′-untranslated region (UTR). Our data, together with an in silico analysis of expressed sequence tags (ESTs) and partial clones reported by others from chromosome 12q13.13 (where the SLC4A8 gene is located), extend the length of the known human SLC4A8 gene by ∼65%.

Comparing the activities of human NDCBE-A through -D by functional expression in Xenopus oocytes, we determined that all four transporter variants mediate electroneutral, Na⁺-dependent, and DIDS-sensitive HCO₃⁻ transport. Significant variations in the rates of pH_i recovery indicate that the choice of Ct (NDCBE-A/C vs. -B/D)—but not the choice of Nt (NDCBE-A/B vs. -C/D)—influences the functional expression of NDCBE. Our data indicate that the unique 17-aa Ct of NDCBE-B is inhibitory, whereas the unique 66-aa Ct of NDCBE-A does not influence functional expression.

Portions of this work have been presented in abstract form (12).

MATERIALS AND METHODS

Performing PCR on Human cDNA Libraries

5′- and 3′-RACE.

We performed nested RACE to screen a Marathon-Ready whole brain human cDNA library (Clontech, Palo Alto, CA) for Nt and Ct variants of NDCBE. PCR was performed in a FGEN02TP Genius Thermocycler (Techne, Burlington, NJ). Cycling parameters were 1 × 300 s at 95°C, 35 × (60 s at 94°C, 30 s at 65°C, and 120 s at 72°C), 1 × 600 s at 72°C, followed by a 12°C hold. In PCRs, we used the adaptor-specific primers AP1 (unnested) and AP2 (nested) supplied with the Marathon cDNA amplification kit (Clontech), according to the manufacturer's recommendations. For 5′-RACE we used the SLC4A8-specific antisense (“reverse” or R) unnested primer 5′-GGCTGTTTCCTTCCACTCAGCATCTTCTCCCTC-3′ (EXON4R; complement of nt 557-589 of human NDCBE-B, GenBank accession no. NM_004858) and the nested (NEST) primer 5′-GTGGGCCAGGGCTTCCAGGCCTTCCTCCCCCTG-3′ (EXON4RNEST; complement of nt 422-454 of NM_004858).

For amplifying cDNAs containing transcribed sequence from predicted exon 23 onwards (i.e., downstream) of human SLC4A8—homologous to the cDNA that encodes the Ct of mouse NDCBE-A—we used the gene-specific sense (“forward” or F) primer 5′-GAACAGAGAGTGACAGGC-3′ (EXON19F; nt 2807-2824 of NM_004858) and the antisense primer 5′-CTCTTTTCCTTTGCATT-3′ [EXON24R; complement of nt 14042999-14043015 of contiguous sequence (contig) NT_029419.11].

For 3′-RACE of cDNAs containing sequence from predicted exon 23 onwards, we used the gene-specific sense primer 5′-TCTGATGAAATGCCTAAA-3′ (EXON23F; nt 14042950-14042968 of contig NT_029419.11).

For amplifying cDNAs containing transcribed sequence from predicted exon 25 onwards, we used the gene-specific sense primer 5′-GAGCCAAGAAGGGACATTTGAGTTGTGTCGC-3′ (EXON25F; nt 14045219-14045249 of contig NT_029419.11) and the antisense primer 5′-GAGTAGGGAATGGTTCCTACCACAGAG-3′ (AK124576R; complement of nt 14049548-14049574 of contig NT_029419.11). A second round of PCR was performed, using the same primers to generate sufficient product for direct DNA sequencing.

Oligonucleotides and peptides were synthesized by the Keck Biotechnology Resource Laboratory at Yale University and by Operon Biotechnologies (Huntsville, AL). 5′- and 3′-RACE products were subcloned into the pCR 2.1 TOPO vector (Invitrogen, Carlsbad, CA), following the manufacturer's protocol. DNA sequencing of clones was performed by the Keck Biotechnology Resource Laboratory and by the Genomics Core Facility at Case Western Reserve University.

PCR of full-length NDCBE cDNA.

Full-length NDCBE cDNAs were created by PCR using as a template Marathon-Ready cDNA libraries from human brain, heart, and kidney (Clontech).

To obtain full-length NDCBE-A cDNA, we performed nested PCR. In the first round of PCR, we used the unnested sense primer 5′-TCGCCGTTCGAGTGATCTGCTCAGACCCG-3′ (EXON1F; nt 13961983-13962011 of contig NT_029419.11) and the unnested antisense primer 5′-TGGCAGGCACTGCCAATGCCCAGATC-3′ (EXON25R; complement of nt 14045127-14045102 of contig NT_029419.11). In the second round of PCR, we used the nested sense primer 5′-ATGCCGGCCGCCGGGAGTAACGAGCCG-3′ (EXON1FNEST; nt 179-205 of NM_004858) and a nested antisense primer “A,” 5′-GGGACCTCACTGAAGGACTCCTCAGTGGCAG-3′ (EXON25RNESTA; complement of nt 14044920-14044890 of contig NT_029419.11).

To obtain full-length NDCBE-C cDNA we performed semi-nested PCR. In the first round of PCR, we used the sense primer 5′-GTGAGGAGGAAGAGGAAGAGGCGAAGGCTGGCGGAGGAGGAGAGACCA-3′ (EXON0F; nt 13928537-13928578 of contig NT_029419.11 plus the next 6 contiguous nt identified in our 5′-RACE product) and the unnested antisense primer 5′-TGGCAGGCACTGCCAATGCCCAGATC-3′ (EXON25R; complement of nt 14045102-14045127 of contig NT_029419.11). In the second round of PCR, we again used the sense primer EXON0F and the nested antisense primer “B,” 5′-GCCCAAATCTTCCATCCCAGCAAAGACTC-3′ (EXON25RNESTB; complement of nt 14044557-14044529 of contig NT_029419.11).

To obtain full-length NDCBE-D cDNA, we performed semi-nested PCR. In the first round of PCR, we used the sense primer EXON0F and the unnested antisense primer 5′-TCCTCCAGCTTCAACTCCCAAAAGCCCTG-3′ (EXON22R; complement of nt 14034895-14034867 of contig NT_029419.11). In the second round of PCR, we again used the sense primer EXON0F and the nested antisense primer 5′-TCCATTCTTAGCTGAGAACATCT-3′ (EXON22RNEST; complement of nt 14034365-14034343 of contig NT_029419.11).

The cycling parameters were 1 × 300 s at 94°C, 35 × (30 s at 94°C, 30 s at 55°C, and 360 s at 68°C), 1 × 600 s at 68°C, followed by a 12°C hold. PCR products were either sequenced directly or, in the case of NDCBE-A, subcloned into the pCR 2.1 TOPO vector before sequencing.

Cloning Full-Length NDCBE into pGH19

NDCBE-A.pGH19.

NDCBE-A.pCR2.1 was used as the template in a PCR with the sense primer 5′-CGAAGCCCGGGATGCCGGCCGCCGGGAGTAACGAGCC-3′ (i.e., a spacer and an XmaI site followed by the first 26 coding nt of NDCBE-A) and the antisense primer 5′-CGAAGTCTAGACCAGAACTTCCCTGAGGCACCACACGGCC-3′ (i.e., a spacer and an XbaI site, followed by the complement of nt 27-55 of the NDCBE-A 3′-UTR).1 The resulting PCR product included the entire open reading frame of NDCBE-A plus 55 nt of 3′-UTR positioned between XmaI and XbaI restriction sites. The full-length clone was ligated into a XmaI/XbaI-digested pGH19 vector. The final construct, NDCBE-A.pGH19, contained the NDCBE-A open reading frame under the control of a T7 promoter and flanked by the 5′- and 3′-UTRs of Xenopus β-globin (41). To optimize translation of NDCBE-A cRNAs, we added an optimal Kozak sequence (19) immediately upstream of the initiator codon (Met₁) by performing QuikChange mutagenesis (Stratagene, La Jolla, CA) according to the manufacturer's recommendations with the sense primer 5′-CTTTGGCAGATCAATTGCCACCATGCCGGCCGCCGGG-3′ and its complement (underscoring indicates optimized Kozak nt; italics indicates the start codon).

NDCBE-B.pGH19.

The construction of human NDCBE-B.pGH19 (aka NDCBE1.pGH19) has been described previously (16). An optimized Kozak sequence was positioned upstream of Met₁ by QuikChange mutagenesis as described above.

NDCBE-C.pGH19 and NDCBE-D.pGH19.

NDCBE-A.pGH19 and NDCBE-B.pGH19 were used as templates for QuikChange mutagenesis to remove Met₁ of each clone (using the sense primer 5′-GTTACTCCCGGCGGCCGGCAATTTGGCAATTGATCTGC-3′ and its complement) and place Met₅₅ of each construct under the control of an optimal Kozak sequence [using the sense primer 5′-CTCTGTATGTGGGAGCCACCATGCCGCTTGGCCGG-3′ and its complement (underscoring indicates optimized Kozak nt; italics indicates the start codon)]. In this way, NDCBE-A.pGH19 was converted into NDCBE-C.pGH19 and NDCBE-B.pGH19 was converted into NDCBE-D.pGH19.

NDCBE-X.pGH19.

An artificial construct truncated at the point of NDCBE-A/B Ct sequence divergence was created by using NDCBE-B.pGH19 as a template for QuikChange mutagenesis to replace Val₁₀₂₈ with a termination codon, using the sense primer 5′-GGCCAAGGAGGAAGAGTAGATAGTCCTTGCACCAACTG-3′ and its complement (italics indicates the termination codon).

cRNA Synthesis and Injection into Xenopus Oocytes

cRNA synthesis.

cDNA constructs in pGH19 were linearized with NotI. Linearized cDNA was purified with the QIAquick PCR purification kit (Qiagen). Capped cRNA was transcribed from the linearized cDNA constructs with the T7 mMessage mMachine kit (Ambion, Austin, TX) according to the manufacturer's recommendations. cRNA was purified and concentrated with the RNeasy MinElute RNA Cleanup Kit (Qiagen).

Xenopus oocyte isolation.

The isolation, separation, and maintenance of oocytes were recently described (40). Ovaries were surgically removed from anesthetized frogs in house or purchased, predissected, from NASCO (Fort Atkinson, WI). Protocols were approved by the Institutional Animal Care and Use Committee through the Animal Resource Center of Yale University. Oocytes were separated by collagenase treatment. Stage V–VI oocytes were manually sorted from the total population of oocytes and stored in OR3 medium (∼0.5× Leibovitz's L-15 medium supplemented with 5 mM HEPES, pH 7.50 with NaOH, adjusted to ∼195 mosmol/kgH₂O by dilution with distilled H₂O), containing 500 U each of penicillin and streptomycin, at 18°C until use.

Microinjection of cRNAs.

One day after isolation, oocytes were injected with either 25 ng of cRNA (25 nl of a 1.0 ng/nl cRNA solution) or 25 nl of sterile water. Oocytes were stored in OR3 medium at 18°C for 5–7 days before assay.

Electrophysiological Measurements

Oocyte solutions.

Our nominally CO₂/HCO₃⁻-free ND96 solution contained (in mM) 93.5 NaCl, 2 KCl, 1 MgCl₂, 1.8 CaCl₂, and 5 HEPES at a pH of 7.50. HCO₃⁻-containing solutions were prepared by replacing 33 mM NaCl in ND96 with 33 mM NaHCO₃ and equilibrating the solution with 5% CO₂-balance O₂. Our Na⁺-free solution contained 93.5 mM NMDG⁺/Cl⁻ in place of NaCl. The osmolalities of all solutions were adjusted to ∼195 mosmol/kgH₂O by adding water or mannitol. DIDS concentrations quoted are corrected for the 80% purity quoted by the manufacturer (Sigma, St. Louis, MO).

Measurement of intracellular pH.

We assayed HCO₃⁻ transport of oocytes expressing NDCBE variants by measuring pH_i with pH-sensitive microelectrodes. The electrodes were fabricated and used as described previously (30, 32, 37). Briefly, the oocyte was impaled with two microelectrodes, one for measuring the membrane potential (V_m) and the other for measuring pH_i. Technical aspects of this protocol were recently described in detail elsewhere (40).

Data Analysis

Data were analyzed with Microsoft Excel 97. Values are given as means ± SE and the number of replicate experiments (n) in the data set from which they were calculated. Slopes of pH_i trajectories were calculated with customized software. Statistical analyses were performed on data with Kaleidagraph 4.0 [1-way and 2-way ANOVA with Student-Newman-Keuls (SNK) multiple-comparison post hoc analysis].

RESULTS

Detection of Two 5′ Variant Transcripts of Human NDCBE by 5′-RACE

The presence of a number of unannotated, partial cDNA clones corresponding to transcriptional activity upstream of NDCBE exon 1 (e.g., GenBank accession no. AB018282) indicated that the human SLC4A8 gene locus might extend further upstream than previously thought. We performed nested 5′-RACE on a human brain cDNA library, using a combination of the gene-specific primers EXON4R/EXON4RNEST (see materials and methods) and the adaptor primers AP1/AP2. We separated and visualized the 5′-RACE product as a single diffuse band of ∼0.5 kb by agarose gel electrophoresis (see Supplemental Data)2 and cloned the product into the pCR2.1 TOPO vector for sequencing.

DNA sequence analysis indicated that the visualized band was composed of at least three distinct products. Figure 2A summarizes the gene structure of SLC4A8, and elements 1–3 in Fig. 2B show how these products align against the gene. 1) The first and shortest of the 5′-RACE products (Fig. 2B, element 1) comprises the adaptor sequence joined directly (i.e., without a 5′-UTR) to exons 3–4 of SLC4A8 and potentially represents a transcript that was degraded before the creation of the Marathon cDNA library. 2) The second 5′-RACE product (Fig. 2B, element 2) comprises the adaptor sequence joined to exons 1–4 of SLC4A8. This product contains the start codon originally described by Grichtchenko et al. (16) for NDCBE but reveals that the 5′-UTR within exon 1 is at least 167 nt in length rather than the originally described 14 nt. 3) The third 5′-RACE product (Fig. 2B, element 3) comprises the adaptor primer joined via a novel 42-bp intervening sequence to exon 2–4 of SLC4A8—that is, the product entirely omits exon 1, which encodes the initiator methionine (Met₁) of NDCBE-B. This novel intervening sequence corresponds to nt 13928537-13928578 of human genome contig NT_029419.11 and represents a novel “exon 0” located ∼33 kb upstream of the previously described exon 1.

Fig. 2.

PCR products aligned with the exon structure of the SLC4A8 gene. A: the human SLC4A8 gene locus on chromosome 12q13.13. Solid vertical bars represent coding exons, and open vertical bars represent 3′-untranslated regions (UTRs). Positions of initiator codons (M) and termination signals (*) as well as exon numbers (1–22) are marked. The gray shading indicates the originally appreciated extent of the SLC4A8 gene locus. TMD, transmembrane domain. B: structure of PCR products obtained by 5′-rapid amplification of cDNA ends (RACE). C: structure of a gene-specific PCR product confirming the presence of novel 3′ exons. D: structure of a PCR product obtained by 3′-RACE. E: structure of a gene-specific PCR product confirming the presence of an extended 3′-UTR that extends beyond the confines of our original 3′-RACE product.

We predict that translation of SLC4A8 transcripts that contain “exon 0,” but lack exon 1, will commence at an in-frame initiator Met encoded within exon 3. This Met corresponds to an internal methionine (Met₅₅) of NDCBE-B. The resulting protein product would be effectively truncated in its Nt by 54 aa.3

Detection of a 3′ Variant Transcript of Human NDCBE by PCR and 3′-RACE

Human and mouse SLC4A8 gene products differ greatly in the sequence of their cytosolic Ct. We performed a Basic Local Alignment Search Tool (BLAST; www/ncbi.nlm.nih.gov/blast) search of the human genome, using the protein sequence of mouse NDCBE (GenBank Protein accession no. NP_067505). In this way we identified two regions 6–8 kb downstream of exon 22 (previously regarded as the last exon) of the human SLC4A8 gene (nt 14041119-14041211 and nt 14042920-14043018 of contig NT_029419.11). If transcribed, spliced, and translated, these novel exons would produce a protein product with ∼90% identity to the mouse Ct sequence.

We amplified cDNA including sequence encoded by the above putative exons, using the gene-specific sense primer pair EXON19F and EXON24R (see materials and methods). We separated and visualized the PCR product as a single band of ∼0.65 kb by agarose gel electrophoresis (see Supplemental Data) and cloned the product into the pCR2.1 TOPO vector for DNA sequencing. The PCR product (Fig. 2C) was composed of exons 19–21 of SLC4A8 joined to a 3′-truncated exon 22 (which we call exon 22a; see Fig. 3; nt 14034077-14034214 of contig NT_029419.11), followed by sequence transcribed from the two novel exons predicted by our BLAST search: the entirety of novel exon 23 (see Fig. 3; nt 14041119-14041210 of contig NT_029419.11) and the initial portion of novel exon 24 (nt 14042925-14043015 of contig NT_029419.11). This PCR product did not include an in-frame termination codon. It should be noted that it was impossible in this PCR to define the 3′ extent of exon 24 because primer EXON24R corresponds to a sequence within this exon.

Fig. 3.

Conceptual translation of novel SLC4A8 exons. The DNA sequence of exons encoding the novel human NDCBE Ct is shown in black. The novel Ct is encoded by transcripts in which exons 22a, 23, 24, and 25 are spliced together. The predicted amino acid sequence of the NDCBE Ct encoded by such transcripts is provided in the line above the DNA sequence. Residues encoded by across exon/exon boundaries are shown in parentheses. For transcripts encoding NDCBE variants with a short Ct (i.e., NDCBE-B/D), the cryptic splice in exon 22 site (marked with an arrow) is ignored and processed transcripts include the exon 22 sequence in gray (including a ∼0.7-kb 3′-UTR), but not sequence from exons 23, 24, or 25. Note that exon 25 is much longer than depicted here and may include ∼8.3 kb of 3′-UTR sequence (see discussion).

To determine the full extent of the novel exon 24, and to detect any downstream sequence including the in-frame termination codon, we performed 3′-RACE on a human brain cDNA library, using the gene-specific primer EXON23F (see materials and methods) and the adaptor primer AP1. We visualized the 3′-RACE product as a single band of ∼1 kb by agarose gel electrophoresis (see Supplemental Data) and cloned the product into the pCR2.1 TOPO vector for DNA sequencing.

Sequence analysis of the 3′-RACE product (Fig. 2D) revealed the full extent of exon 24 (see Fig. 3; nt 14042925-14043020 of contig NT_029419.11) and a further downstream novel SLC4A8 exon 25 (see Fig. 3; nt 14044517-14045425 of contig NT_029419.11) joined directly to the adaptor sequence. Exon 25 includes 10 bp of coding sequence, an in-frame “TAA” termination codon, and 895 bp of 3′-UTR. The 3′-RACE product concludes at a point in the genome ∼10 kb downstream of the 3′ end of exon 22—the last exon included in NDCBE-B transcripts. We predict that translation from SLC4A8 RNA that includes sequence transcribed from exons 22a–25 will not terminate in exon 22, but instead will continue through exons 23–24 and terminate at the aforementioned stop codon in exon 25. Thus the resulting transporter would have a Ct that contains 66 aa not found in NDCBE-B. This alternative human Ct is 89% identical to the unique 66-aa Ct of mouse NDCBE (see protein sequence alignment in Fig. 11).

Noting that the 3′-UTR included in our new 3′-RACE product lacked a consensus polyadenylation site, we hypothesized that the true extent of the novel 3′-UTR is longer than can be detected by 3′-RACE using the cDNA library and PCR conditions employed in our study. Indeed, the 3′-UTR of mouse slc4a8 (GenBank accession no. NM_021530) extends to 8.3 kb. Studying the human genome database, we note several EST sequences representing transcriptional activity in the 10 kb downstream of the SLC4A8 gene. Most relevant is a ∼5-kb transcribed stretch of sequence (GenBank accession no. AK124576 from cerebellum cDNA, nt 14047871-14052853 of contig NT_029419.11) that begins ∼2.5 kb downstream of exon 25 and includes two consensus polyadenylation signals “AATAAA” positioned 196 nt and 16 nt from its 3′ end. To investigate whether this transcribed sequence could be a continuation of exon 25, we performed a PCR on human brain cDNA using a sense primer within exon 25 (EXON25F) and an antisense primer complementary to a region within AK124576 (AK124576R). We visualized the PCR product as a single band of ∼4.5 kb by agarose gel electrophoresis (see Supplemental Data). Sequence analysis demonstrated that the PCR product (Fig. 2E) is a cDNA representing an unspliced transcript spanning 4.3 kb (nt 14045192-14049574 of contig NT_029419.11), that is, 0.2 kb of our 3′-RACE product, 2.4 kb of intervening sequence, and 1.7 kb of AK124576. Thus the full extent of exon 25 spans nt 14044517-14052853 of contig NT_024919.11 and includes 8.3 kb of 3′-UTR. The sequence of this novel portion of SLC4A8 3′-UTR has been deposited in GenBank with the accession number EU637925.

Detection and Distribution of Four Novel, Full-Length Human NDCBE Transcripts

To determine whether the novel 3′-RACE product is incorporated into a full-length human SLC4A8 gene product corresponding to human NDCBE-A, we performed nested full-length PCR, using a human brain cDNA library as a template. In the forward direction, we used sense primers complementary to sequence in exon 1 (EXON1F and EXON1FNEST; see materials and methods). In the reverse direction, we used antisense primers complementary to sequence in exon 25 (EXON25R and EXON25RNESTA; see materials and methods), ∼0.5 kb downstream from the termination codon.

Using agarose gel electrophoresis of the PCR product, we visualized a band of ∼3.5 kb (Fig. 4A). The cDNA was cloned into pCR2.1, and DNA sequence analysis confirmed that the product corresponded to the expected full-length cDNA encoding the human ortholog of NDCBE-A. The structure of NDCBE-A vs. NDCBE-B transcripts is represented in Fig. 5A, and the NDCBE-A protein product is represented in Fig. 6.

Fig. 4.

Full-length PCR products from NDCBE-A, -C, and -D/D′. Agarose gels show the products of PCR reactions designed to amplify entire open reading frames from human brain (lane B), heart (lane H), and kidney (lane K) cDNA libraries. A: NDCBE-A. B: NDCBE-C. C: NDCBE-D/D′.

Fig. 5.

Exon structure of SLC4A8 transcript variants. Alignment is in the style of Fig. 2, depicting the exon structure of novel SLC4A8 splice variants. A: NDCBE-A and NDCBE-B. B: NDCBE-C, NDCBE-D and NDCBE-D′. Positions of initiator codons (M) and termination signals (*) as well as exon numbers (0–25) are marked. The position of the extended exon 3 in NDBCE-D′ is marked (†). GenBank accession nos. are shown in parentheses. C: the new extent of the SLC4A8 gene, based on the present data, depicting exons 0–25. Exon 25 has the potential to include 8.3 kb of 3′-UTR (see results), but this is not represented here because we cannot attribute this extended 3′-UTR to a specific full-length transcript.

Fig. 6.

NDCBE protein variants. Diagrammatic representation (to scale) of mouse NDCBE (mNDCBE; NP_067505) aligned against the 4 human NDCBE protein variants (NDCBE-A, NP_001035049; NDCBE-B, NP_004849; NDCBE-C, ABJ09587; and NDCBE-D, ABJ91576). Horizontal gray bars represent the protein sequence aligned from Nt to Ct, with the overall length of the protein in amino acid residues noted on right. The lengths of protein cassettes are marked within or adjacent to their corresponding horizontal gray bars. The limits of the cytosolic Nt, TMD, and cytosolic Ct are marked at top. Positions of putative transmembrane spans are denoted by numbered vertical boxes. A detailed alignment of the 66-aa and 17-aa Ct protein cassettes is given in Fig. 11.

To determine whether the novel 5′-RACE product is incorporated into full-length human SLC4A8 gene products, we performed semi-nested full-length PCR using human brain, heart, and kidney cDNA libraries as template. In the forward direction, we used the sense primer EXON0F. In the reverse direction, we used either an antisense primer pair designed to detect transcripts terminating in exon 22 (EXON22R and EXON22RNEST; see materials and methods) or an antisense primer pair designed to detect transcripts terminating in exon 25 (EXON25R and EXON25RNESTB; see materials and methods).

Using agarose gel electrophoresis of the PCR products expected to represent full-length cDNA transcribed from exons 0 through 22a and then on through 25, we visualized bands of ∼3.0 kb from brain, heart, and kidney (Fig. 4B, lanes B, H, and K). DNA sequence analysis of the PCR products confirmed that they corresponded to the expected full-length cDNA, which we named NDCBE-C. The structure of the NDCBE-C transcript is represented in Fig. 5B. We predict that the protein product of NDCBE-C is identical to NDCBE-A, but is truncated by 54 aa at its Nt (Fig. 6).

Using agarose gel electrophoresis of the PCR products expected to represent full-length cDNA transcribed from exons 0–22, we visualized bands of ∼3.0 kb from brain and kidney (Fig. 4C, lanes B and K). DNA sequence analysis of the PCR products confirmed that they corresponded to the expected full-length cDNA, which we named NDCBE-D. The structure of the NDCBE-D transcript is represented in Fig. 5B. We predict that the protein product of NDCBE-D is identical to NDCBE-B, but is truncated by 54 aa at its Nt (Fig. 6). An unexpected observation was that the NDCBE-D PCR product amplified from human heart cDNA (Fig. 4C, lane H) was noticeably larger than the NDCBE-D PCR product amplified from human brain or kidney cDNA. DNA sequence analysis of this PCR product demonstrated that the extended heart variant (which we call NDCBE-D′) differs from the shorter brain/kidney variant (i.e., NDCBE-D) in that exon 3 is extended in the 5′ direction (such that exon 3 of NDCBE-D′ encompasses nt 13987732-13988112 of contig NT_029419.11). We predict that this extension in NDCBE-D′ should not affect the protein sequence of the encoded transporter, but should extend the 5′-UTR by 234 nt. The structure of the NDCBE-D′ transcript is represented in Fig. 5B.

We deposited the sequence of the four novel full-length human NDCBE cDNAs in GenBank, and they were assigned the following accession numbers: DQ063579 (NDCBE-A), DQ975204 (NDCBE-C), DQ996537 (NDCBE-D), and DQ996398 (NDCBE-D′). The correspondingly extended gene structure of SLC4A8 is represented in Fig. 5C.

Functional Detection of NDCBE-A, -B, -C, and -D at the Plasma Membrane of Xenopus Oocytes

We subcloned NDCBE-A, -C, and -D into the Xenopus expression vector pGH19. We injected NDCBE-A, -B, -C, and -D cRNAs into Xenopus oocytes and, after 5 days, monitored V_m and the pH_i recovery from a CO₂-induced intracellular acid load. Like the previously characterized NDCBE-B (16), all three novel variant transporters expressed DIDS-sensitive NCBT activity. That is, in all four cases, NDCBE variant-expressing oocytes exhibited a pH_i recovery from a CO₂/HCO₃⁻-induced acid load that was 1) not evident in H₂O-injected oocytes (representative experiments are shown in Fig. 7, A–J), 2) Na⁺ dependent (representative experiments are shown in Fig. 7, A–E), and 3) blocked by 160 μM DIDS (representative experiments are shown in Fig. 7, F–J).

Fig. 7.

Intracellular pH (pH_i) and membrane potential (V_m) traces of representative oocytes expressing NDCBE variants. Xenopus oocytes heterologously expressed NDBCE-A (A and F), NDCBE-B (B and G), NDCBE-C (C and H), or NDCBE-D (D and I). Control oocytes (E and J) were injected with H₂O. Oocytes were initially superfused with ND96 solution before acidification with our CO₂/HCO₃⁻ solution (e.g., segment a–b in A). Oocytes were then exposed to either a CO₂/HCO₃⁻ solution that contained NMDG⁺ in place of Na⁺ (0 Na in A–E) or a CO₂/HCO₃⁻ solution that was supplemented with 160 μM DIDS (DIDS in F–J). pH_i recovery (e.g., segment b–c in A) was detected only in cells expressing NDCBE variants (A–D and F–I, but not E or J). The Na⁺ dependence of the transport process was assessed by the reversal of the pH_i recovery (Na⁺-coupled HCO₃⁻ efflux from the cell) on application of a Na⁺-free bath solution (e.g., segment c–d in A). The stilbene sensitivity of the process was assessed by the stoppage of pH_i recovery on the application of a DIDS-containing bath solution (e.g., segment e–f in F). The slow acidification of control cells elicited by Na⁺ removal or DIDS application are phenomena observed in other recent work (see Fig. 6B and Fig. 8B of Ref. 26). These findings may reflect the presence of a low endogenous Na⁺-dependent acid extrusion in control cells or—in the case of the DIDS effect—a relaxation from a transient, DIDS-induced alkalinization (see segment e′–e in G). The average resting pH_i and V_m of oocytes did not differ between cells injected with NDCBE variant cRNA (see Supplemental Data for statistics). Averaged data for these experiments are shown in Fig. 8.

Figure 8 summarizes averaged data from experiments represented in Fig. 7. Among the NDCBE variants, we observed small, but significant, differences in the rates of pH_i recovery (open bars in Fig. 8). We used a two-way ANOVA to determine whether a difference in Ct or Nt length influenced the rate of pH_i recovery mediated by the variant transporters. The results of this analysis (see Supplemental Data) disclosed that the rate of pH_i recovery was not significantly influenced by the presence vs. absence of the first 54 aa of the Nt (NDCBE-A/B vs. NDCBE-C/D; P = 0.13), whereas the choice of the 66-aa NDCBE-A/C Ct compared with the 17-aa NDCBE-B/D Ct significantly enhanced the rate of pH_i recovery (P = 0.002). The mean pH_i recovery rate for NDCBE-B/D was ∼25% less than that for NDCBE-A/C. The two-way ANOVA further indicated that the effect of the choice of Ct is not significantly influenced by the choice of Nt (P = 0.59). That is, we do not detect a functional consequence of interactions between the Nt and the Ct.

Fig. 8.

Effects of Na⁺ removal or DIDS application on NDCBE-mediated HCO₃⁻ transport rates. Bars represent average ± SE data from experiments such as those presented in Fig. 7, including the average rate of pH_i recovery mediated by each variant under control conditions (open bars), in the absence of extracellular Na⁺ (gray bars), or after the application of DIDS (black bars). *P < 0.001 compared with H₂O-injected oocytes, †P ≤ 0.01 compared with NDCBE-D in a 1-way ANOVA with post hoc Student-Newman-Keuls test (SNK). Average resting pH_i and V_m of oocytes did not differ between cells injected with NDCBE variant cRNA. See Supplemental Data for statistics. Numbers above bars are the number of oocytes in the data set. Transporters with the long Ct (NDCBE-A/C) mediate a pH_i recovery faster than transporters with the short Ct (NDCBE-B/D). The rates of acidification elicited by Na⁺ removal or DIDS application are not substantially different between NDCBE-expressing and control cells. Thus both treatments block NDCBE-mediated NCBT activity.

Functional Detection of a COOH-Terminally Truncated NDCBE at the Plasma Membrane of Xenopus Oocytes

To determine whether the functional differences between Ct variants of NDCBE could be attributed to a stimulatory effect of the 66-aa Ct cassette or an inhibitory role of the 17-aa Ct cassette, we created an artificial NDCBE variant, NDCBE-X, that is truncated in its Ct at the point where the NDCBE-A and NDCBE-B sequences diverge. That is, the Ct of NDCBE-X contains neither the 17-aa Ct cassette of NDCBE-B nor the 66-aa Ct cassette of NDCBE-A. We expressed NDCBE-A, NDCBE-B, and NDCBE-X in Xenopus oocytes and compared the rate of pH_i recovery mediated by each transporter from a CO₂/HCO₃⁻-induced acid load (Fig. 9). In these experiments, NDCBE-X mediated an average rate of pH_i recovery from a CO₂/HCO₃⁻-induced acid load that was not significantly different from the pH_i recovery mediated by NDCBE-A (P = 0.964; result of a 1-way ANOVA with SNK post hoc analysis, see Supplemental Data). However, the pH_i recovery rates for both NDCBE-A and NDCBE-X were significantly higher than that for NDCBE-B (P ≤ 0.01 in both cases; see Supplemental Data).

Fig. 9.

Comparison of pH_i recovery rates mediated by NDCBE-A or -B with those mediated by a COOH-terminally truncated NDCBE. Bars represent average ± SE data from a series of experiments in which we measured the pH_i recovery rate from a CO₂/HCO₃⁻-induced acid load mediated by NDCBE-A, NDCBE-B, and NDCBE-X as expressed in oocytes. NDCBE-X is an artificial construct that is truncated in its Ct at the point where the NDCBE-A and NDCBE-B sequences diverge. Numbers above bars are the number of oocytes in the data set. *P < 0.001 compared with H₂O-injected oocytes; †P < 0.005 compared with NDCBE-X in a 1-way ANOVA with post hoc SNK test. See Supplemental Data for statistics. Thus the differences in functional expression detected between NDCBE variants in Fig. 8 are likely due to an inhibitory effect of the short Ct and not a stimulatory effect of the long Ct.

Solubility of the Unique 17-aa Cassette of NDCBE-B/D

A synthetic peptide corresponding to this 17-aa cassette, _NH2-CVIVLAPTVYLGASNYRT-_COOH, failed to dissolve in an aqueous conjugation buffer (Pierce, Rockford, IL).

DISCUSSION

Transcription from alternative promoters and the use of alternative splice sites are common mechanisms employed at the pre- and posttranscriptional levels to tailor the structural, functional, and temporal properties of each transporter to the cell type in which it is being expressed. The electrogenic Na⁺/HCO₃⁻ cotransporter NBCe1 is a well-characterized example of an NCBT under such regulation, there being three distinct variants (NBCe1-A, -B, and -C) transcribed from the SLC4A4 gene. NBCe1-A and -B/C mRNA are transcribed from alternative promoters (1), with the result that NBCe1-A and -B/C proteins (which have distinct Nt) differ in their distribution and activity (21). NBCe1-B and -C mRNA are alternatively spliced and differ in the presence or absence of a single exon (exon 24 of NBCe1-A/B being spliced out to produce NBCe1-C), with the result that NBCe1-A/B and -C proteins have distinct Ct (5). At the commencement of our study, NDCBE was noticeably different from other NCBTs in that human and mouse SLC4A8 each had only one known transcript/protein. Moreover, the differences between the human and mouse versions of NDCBE are sufficiently minor that one could have hypothesized that they represented a species difference. Our study reveals that 1) the human SLC4A8 gene is in fact much larger than previously thought; 2) with alternative 5′ and 3′ ends, the extended SLC4A8 gene is capable of producing at least five unique full-length mRNAs, encoding four distinct protein products, two with relatively low and two with relatively high NCBT activity; and 3) the human gene can be spliced to produce the ortholog of the previously known mouse variant.

Consequences of Extended 5′-UTRs

Our study extends the known 5′ limit of the 5′-UTR of NDCBE-A/B, demonstrating that transcription of SLC4A8 to produce NDCBE-B begins 153 bp further upstream in the gene locus than previously thought. This assignment is an important first step for determining the boundaries of SLC4A8 regulatory elements. That is, one could examine the genomic sequence upstream of this new boundary for promoter elements that could affect transcription of SLC4A8, and one could examine downstream elements that could influence the stability and/or translational efficiency of SLC4A8 mRNA.

Although presently nothing is known about the identity of the SLC4A8 promoter region(s), we might speculate that the confirmation and assignment of exon 0 as a novel alternative site for transcriptional initiation of SLC4A8 mRNA suggests that the gene locus contains two distinct promoters. 1) One promoter—upstream of exon 0 but perhaps downstream of the neighboring GALNT6 gene (encoding GalNAc transferase 6) (Fig. 10A; i.e., within a range of ∼12 kb; nt 13916871-13928537 of contig NT_029419.11)4 —could drive transcription of NDCBE-C/D. 2) A second promoter—perhaps located between exons 0 and 1 (i.e., within a range of ∼33 kb; nt 13928578-13961900 of contig NT_029419.11)—could drive transcription of NDCBE-A/B. The most obvious implication of NDCBE-A/B and -C/D being controlled by separate promoters is that their transcription could be differentially influenced by factors such as cell type, developmental stage, and/or responsiveness to stimuli such as metabolic acidosis (3).

Fig. 10.

Reassignment of the SLC4A8 gene locus on chromosome 12q13.13. Scaled representations of the SLC4A8 gene locus are shown. A: genomic view before the assignment of novel SLC4A8 from the present study. Here, the SLC4A8 gene locus is flanked by the GALNT6 gene (encoding GalNAc transferase 6), LOC728476 (labeled as LOC), and the SCN8A genes (encoding a voltage-gated Na⁺ channel), each represented by a gray box. Arrows indicate the direction of transcription. LOC728476 is likely an errantly predicted open reading frame composed of exons that comprise the extended 5′-UTRs of both the GALNT6 and SLC4A8 genes. B: newly reassigned SLC4A8 gene locus.

In NDCBE-A/B transcripts, exons 2 and 3 of SLC4A8 encode protein sequence. However, in NDCBE-C/D transcripts—where the initiator Met is located within exon 3—all of exon 2 and part of exon 3 are now 5′-UTR sequence. It is possible that contextual requirements for the translational initiation signal at Met₁ of NDCBE-C/D (corresponding to Met₅₅ of NDCBE-A/B) has informed the choice of codons for residue 54 (i.e., “CGG” encoding Arg) in NDCBE-A/B. Neither of the closely related transporters NBCn1 and NBCn2 has an internal Met at the position equivalent to NDCBE-A/B residue 55 or an Arg at the position equivalent to NDCBE-A/B residue 54. Instead, in both NBCn1 and NBCn2, His-Val replaces the Arg₅₄-Met₅₅ of NDCBE-A/B. However, the Arg-encoding CGG at codon position 54 of NDCBE-A/B does not contribute to a theoretically better Kozak sequence than either the His-coding “CAT” or “CAC” triplets at the equivalent positions of NBCn1/n2 (18).

A final novel feature is the presence of a unique cassette in the 5′-UTR of NDCBE-D′, cloned from human heart cDNA. This cassette derives from the choice of an alternative 5′ splice site for exon 3 that extends this exon by 234 nt in the 5′ direction. What role this unique 5′-UTR sequence plays in the heart-specific regulation of NDCBE-D′ mRNA is presently unknown. Others report partial clones/ESTs of SLC4A8 that indicate further possible variations in the 5′-UTR of NDCBE-C/D transcripts in other human cDNA preparations by incorporating sequence from cryptic exons located between exons 0 and 1 (e.g., partial SLC4A8 from thymus/AK128321, testis/DB090766, and embryonic stem cells/CN286464). We are now able to designate an unannotated, and apparently partial,5 human brain cDNA clone reported by others (GenBank accession no. AB018282) as a further full-length NDCBE-D variant. This variant includes a novel exon (nt 13940248-13940379 of contig NT_029419.11)—which in the genome would appear between exons 0 and 1—that contributes an extra 131 nt to the 5′-UTR between exons 06 and 2. However, we did not isolate this particular variant in the present study.

Consequences of a Novel 3′-UTR

Figure 10B illustrates the extended SLC4A8 gene locus on chromosome 12q13.13. Our finding that NDCBE-A/C transcripts have a 3′-UTR distinct from those of NDCBE-B/D transcripts provides another potential mechanism for the differential regulation of transcription and translation of SLC4A8 variants. Although the ∼8.3-kb 3′-UTR of human NDCBE-A/C transcripts (encoded by the single exon shown in Fig. 10B) are similar in length to the 3′-UTR of mouse slc4a8 transcripts (GenBank accession no. NM_021530), only the last ∼250 bp of this ∼8.3-kb extended human SLC4A8 3′-UTR shares sequence identity with its mouse equivalent. NDCBE-A transcripts, which include this ∼8.3 kb of 3′-UTR, would account for the unexpectedly large (∼12 kb) band seen in Northern blots of NDCBE from human brain RNA using an NDCBE-A/B-specific probe (16). Intriguingly, a 1.5-kb portion (which includes a polyadenylation signal) near the end of this extended 3′-UTR of NDCBE-A/C is 95% identical to a region 25 kb downstream of the presently recognized 3′ boundary of the human serotonin receptor 2C (HTR2C) gene on chromosome Xq24.

With regard to the 3′-UTR specific to NDCBE-B/D, we were unable to obtain by PCR any product longer than the ∼0.7-kb stretch previously described in Ref. 16, despite our using an entirely different source of human brain cDNA (new 3′-RACE data using Marathon human brain cDNA; not shown).

Consequences of NDCBE Nt Truncation

We predict that NDCBE-C/D translation initiates at an “ATG” codon that encodes an internal Met₅₅ residue in the Nt of NDCBE-A/B. Thus it is possible that full-length NDCBE-A/B transcripts can be translated to produce both NDCBE-A/B and NDCBE-C/D proteins. It is also possible that other internal in-frame “ATG” codons could serve as alternative foci for translational initiation. Notably, an in-frame Met₁₂₅ in NDCBE-A/B (in exon 4) is a feasible initiation site for a functional NDCBE. In the present study, we did not detect a clone in which the codon for Met₁₂₅ was the first possible initiation site, possibly because our reverse nested 5′-RACE primer was complementary to sequence within exon 3 (i.e., we would not have detected such variants if their 5′-UTR did not include exon 3 sequence). Interestingly, Met₁₂₅ of NDCBE-A/B is located at a position analogous in eAE1 to the internal Met₆₆, which is the first available Met in the truncated kidney variant, kAE1 (13). Three other Na⁺/HCO₃⁻ cotransporters—NBCe1, NBCe2, and NBCn1—also have internal, in-frame “ATG” codons within 6 residues of Met₆₆ in eAE1, and thus could potentially serve as alternative initiation points. It is not clear whether initiation at “ATG” codons even further downstream into the Nt sequence—that is, where the Nt is predicted to become highly structured (15, 46)—could produce functional NCBTs.

The constructs employed in the present study to express NDCBE-A/B in Xenopus oocytes used an optimized Kozak sequence to direct translational initiation strongly to the desired ATG, minimizing the expression of a mixed product. Our functional studies show that Nt choice (Met₁ vs. Met₅₅) does not detectably influence the transporter activity (see NDCBE-A/B vs. NDCBE-C/D in Fig. 8). However, our preliminary work on another electroneutral NCBT, NBCn2 (25, 27), indicates that a region within residues 44–92 of the Nt—and, by homology, an analogous region in all three electroneutral NCBTs—is required for interaction with the inositol 1,4,5-trisphosphate (IP₃) receptor binding protein IRBIT (4). This interaction with IRBIT increases the rate of HCO₃⁻ transport mediated by NCBTs (27, 36). Thus, if our preliminary results transfer to NDCBE, it is possible for NDCBE-A/B, but not NDCBE-C/D, to be under the stimulatory control of IRBIT. However, the physiological significance of IRBIT-mediated signaling pathways in respect to HCO₃⁻ transport has yet to be elucidated.

Consequences of Alternative NDCBE Ct Choice

The longer Ct of NDCBE-A/C and the shorter Ct of NDCBE-B/D share 34 aa at the beginning of the putative cytosolic Ct (see Fig. 1B). This common region contains two motifs that satisfy the requirements (42, 43) for binding carbonic anhydrase II (CAII): “LDDLM” and “LDDAK.” However, Piermarini et al. (28) found that the Ct per se of three SLC4 family members—AE1, NBCe1-A/B, and NDCBE-B/D—did not bind CAII.

The longer Ct of NDCBE-A/C.

The unique 66-aa Ct cassette of mouse NDCBE-A (Fig. 11, top row) is ∼90% identical to the equivalent Ct portion of human NDCBE-A/C (Fig. 11, 2nd row). Within this human 66-aa Ct cassette, a 35-aa section shares 75% identity with splice cassette III of human NBCn1 (Fig. 11, 4th row, boxed region) and 56% identity with the equivalent, although invariant, region in the Ct of NBCn2 (Fig. 11, bottom row). One notable feature of the long Ct of NDCBE-A/C—compared with the short Ct of NDCBE-B/D (and in fact compared with the Ct of NBCn1 and NBCn2)—is the presence of a “KLLSSP” sequence (underlined in Fig. 11) that matches a consensus binding site for 14-3-3 proteins (29). 14-3-3 Proteins are known to influence the trafficking (reviewed in Ref. 23) and activity (e.g., Ref. 47) of a variety of transporters and channels. The NDCBE-A/C Ct also contains two consensus protein kinase C phosphorylation sites (45), “SRK” and “SCR” (circled in Fig. 11), not present in the Ct of NDCBE-B/D or the Ct of NBCn1 or NBCn2.

Fig. 11.

Features of the NDCBE Ct. Protein sequence alignment of the presumed cytosolic Ct domains of mouse NDCBE (mNDCBE-A; GenBank accession no. NP_067505), the novel human NDCBE-A reported in the present paper (NP_001035049), the original human NDCBE (hNDCBE-B; NP_004849), human NBCn1 containing splice cassette III (boxed; ACB47400), and human NBCn2 with the shorter of the 2 known Ct (e.g., NP_071341). The position of a putative 14-3-3 protein binding site is underlined. Two novel protein kinase C phosphorylation sites are circled. Sequences conserved between human NDBCE-A and other human Ct are shaded in gray.

The shorter Ct of NDCBE-B/D.

It is not reported whether the shorter Ct of human NDCBE-B/D (Fig. 11, 3rd row) has an ortholog in the mouse. Human NDCBE-B/D Ct arises as a consequence of ignoring a cryptic 3′ splice site within exon 22 (Fig. 3), such that an extended exon 22 (including ∼0.7 kb of 3′-UTR sequence) is incorporated into the final mRNA. Ignoring the same splice site during the processing of mouse slc4a8 pre-mRNA would result in a hypothetical mouse NDCBE-B/D ortholog terminating with the protein sequence “VESGLAPAL” (underlined residues are conserved between the 2 species). This sequence is 8 residues shorter than the human NDCBE-B/D Ct. However, we note that the Val residue is conserved merely out of the necessity for having a “GTX” codon to preserve the 3′ splice site, and it is possible that the “Leu-Ala-Pro” sequence is conserved for similar reasons.

The results of our functional comparisons of the alternative Ct of human NDCBE variants in Fig. 8 and Fig. 9 suggest that those with the shorter Ct (NDCBE-B and -D) have a 25–50% reduced HCO₃⁻ transport rate compared with those with the longer Ct (NDCBE-A and -C). By comparing the transport rates of NDCBE-A and NDCBE-B with NDCBE-X—that is, an NDCBE construct truncated at the point of NDCBE-A/B Ct divergence—we determined that the difference in transport rates between NDCBE-A and NDCBE-B is unrelated to the 66-aa Ct sequence of NDCBE-A but is due to an inhibitory effect of the 17-aa Ct sequence of NDCBE-B. Because the fractional difference (as low as 25%) in functional expression is not sufficiently large to detect reliably by tagging, at present we are unable to distinguish whether the differences are due to an increased amount of NDCBE expressed at the oocyte plasma membrane or whether individual transporter molecules are intrinsically less active.

Notably, a synthetic peptide corresponding to the above 17-aa cassette is water insoluble. Thus this domain may have a tendency to bury itself in a hydrophobic environment at the cell membrane, perhaps even in the Nt or transmembrane domain of the transporter itself.7

Implications

The identification of four novel variants of human NDCBE has important implications for the design of specific antibodies, oligonucleotide probes, and genetically modified mice. It is now evident that mature mRNA for NDCBE can be produced without transcription of mRNA from exon 1, and that these variant mRNAs have the potential to encode functional transporters that lack the first 54 aa of NDCBE-A/B and/or that replace the last 17 aa of NDCBE-B/D with the last 66 aa of NDCBE-A/C. The reassignment of ∼49 kb of 12q13.13—previously considered to be SLC4A8 upstream and downstream flanking sequence—as part of the SLC4A8 gene locus greatly extends the region over which investigators might look for disease-associated polymorphisms. We presently consider the fullest extent of the SLC4A8 gene locus to be ∼124 kb, encompassing nt 13928438-14052853 of contig NT_024919.11. The observation that the short Ct of NDCBE-B/D has an inhibitory effect on functional expression of the transporter now raises the possibility that the functional expression of NDCBE-B/D could be stimulated by an as yet unidentified mechanism.

GRANTS

This work was supported by National Institutes of Health Grants R01-NS-18400, P01-HD-32573, and R37-DK-30344.