Alternative Splicing of NHE-1 Mediates Na-Li Countertransport and Associates With Activity Rate

Human erythrocytes and reticulocytes were separated by double centrifugation through Ficoll-Hypaque (24). Total RNA was extracted from human erythrocytes and reticulocytes and from transfected and nontransfected PS200 cells according to the guanidine thiocyanate method (25), and cDNA was retrotranscribed using SuperScript II Reverse Transcriptase (Gibco, Gaithersburg, MD).

RT-PCR primers were designed along the sequence of NHE-1 (GenBank AC S68616) as follows: NHE-1.1, position 534–556 F (forward), 5′-ACTCTCATTTGGGTAAGGTCGA-3′; NHE-1.2, position 1098–1118 R (reverse), 5′-CCGCAGTGGCAGGAAGTAGC-3′; NHE-1.3, position 1066–1084 F, 5′-CTTCCTGCTGCCGCCCAT-3′; NHE-1.4, position 1656–1676 R, 5′-ATAGGGGCGCATCACCACTC-3′; NHE-1.5, position 1610–1630 F, 5′-GCCGAGCTCTTCCACCTGTC-3′; NHE-1.6, position 2106–2126 R, 5′-GCGCTTCGTCTCTTGCTTTT-3′; NHE-1.7, position 2020–2042 F, 5′-CACTGCCATCATCACTGTCATC-3′; NHE-1.8, position 2653–2672 R, 5′-CCGAGACATGGTGGGTGAG-3′; NHE-1.9, position 2508–2527 F, 5′-AGCGGCTGCGGTCCTACAA-3′; NHE-1.10, position 3201–3221 R, 5′-TCTGGTGGGGAGGATGCTTC-3′.

Full length NHE-1 (kind gift of J. Pouyssegur and L. Counillon, Universitè de Nice, Nice, France) was cloned into pcDNA3 expression plasmid (Invitrogen, Carlsbad, CA) to obtain pcNHE-1. pcNHE-1 was then mutagenized by replacing the BstZI–Csp45I fragment (spanning nucleotides 715-1281) of NHE-1 DNA with the corresponding fragment of the alternative spliced form, lacking nucleotides 809–1105 and therefore obtaining the expression construct pcDNHE-1.

An NHE-deficient PS200 hamster fibroblast cell line [a twin clone of the PS120 hamster fibroblast cell line (26), gift of J. Pouyssegur and L. Counillon] was cultured in minimal essential medium supplemented with 10% fetal bovine serum. Before transfection, PS200 cells were trypsinized, seeded at a density of 5 × 10⁵ cells/dish in 60-mm dishes, and incubated overnight. Transfection was performed according to the calcium phosphate–DNA coprecipitation method (27).

Three days after transfection, cells transfected with NHE-1 or DNHE-1 were subject to the proton killing test (28), an acid-loading selection test that allows elimination of those cells that do not express functional NHE. Cells were allowed to load with NH₄ during 60-min incubation at 37°C in an NH₄-saline medium. External NH₄ was then rapidly removed by two washes with an Na-rich medium and the cells were finally incubated in this same medium for 60 min at 37°C. At the end of incubation, the Na-rich medium was replaced by culture medium. This treatment was repeated five times over a period of 2 weeks, and the successfully transfected cells were used for experiments.

Three days after transfection, cells transfected with DNHE-1 were treated with 400 μg/ml geneticin (Gibco). This treatment was repeated 10 times over a period of 3 weeks, and finally the transfected cells were used for experiments.

Na-driven Li efflux was evaluated as previously described for human skin fibroblasts (29). In brief, Li efflux was measured in PS200 cells, in PS200 cells transfected with NHE-1, and in PS200 cells transfected with DNHE-1 in the presence of ouabain, bumetanide, and 4,4′-diisothiocyanato-stilbene-2,2′-disulfonic acid to inhibit the Na pump, Na-K cotransport, and the anion exchanger, respectively. Dimethyl-amiloride (DMA, the specific inhibitor of NHE) and/or phloretin (the strongest inhibitor of SLC) were added as appropriate to probe the NHE and SLC, respectively (Fig. 3). Na-driven Li efflux in human erythrocytes was evaluated as previously described (22).

A rabbit polyclonal antibody against amino acids 56–70 of human NHE-1 (common to both NHE-1 and DNHE-1) (NH₂-SIGDVTTAPPEVTPE-COOH) was generated and purified by absorption in protein A-Sepharose.

RT-PCR was performed for 40 cycles on 2 μl cDNA (2–10 ng/ml) from six human tissues (brain, leukocyte, heart, liver, kidney, spleen) (OriGene Technologies, Rockville, MD) using primers NHE-1.1 and NHE-1.4, as described above (Fig. 6).

Total RNA was extracted from whole blood of 10 individuals with low SLC activity (142.5 ± 8.7 μmol · liter⁻¹ cells · hour⁻¹, mean ± SE, range 90–174) and from 10 individuals with high SLC activity (939.0 ± 58.6 μmol · liter⁻¹ cells · hour⁻¹, range 768–1,369), and cDNA was retrotranscribed. PCR conditions were optimized for Mg²⁺ concentration, primer concentration, primer annealing temperature, and number of cycles for NHE-1, DNHE-1, and β-actin cDNA amplification. Oligonucleotide positions were, respectively, 721–740 F 5′-CTCAGAGCCACCACGAGAAC-3′ and 1120–1139 R 5′-TGCCCAGGTTTTCTGTGAAC −3′. In the case of β-actin, oligonucleotides 5′-TGACGGGGTCACCCACACTGTGCCCATCTA-3′ and 5′-CTAGAAGCATTGCGGTGGACGATGGAGGG-3′, upstream and downstream primers, respectively, were used for amplification of a 646-bp fragment of human β-actin cDNA located between nucleotides 2198 and 3065 of the reported human gene sequence (GenBank AC E00829). Semiquantitative RT-PCR experiments were performed by incorporating α-³²P-dCTP in two systems: primers 1 and 2 on NHE-1 and primers 3 and 4 on β-actin. Expression of NHE-1 + DNHE-1 was normalized by β-actin values.

Data are shown as arithmetical mean with standard error. Comparisons between observations were addressed by Student’s t test. Correlations were sought by simple linear regression. The null hypothesis was rejected for two-tailed P values <5% (JMP software for the Apple Macintosh, SAS Institute, NC, USA).

To search for possible amiloride-insensitive alternative splicing(s) of NHE-1, cDNA prepared with total RNA obtained from human erythrocytes and reticulocytes was amplified using 10 specific primers designed along the entire sequence of NHE-1 (NHE-1.1–NHE-1.10, see research design and methods for details). Amplification of cDNA using NHE-1.1 and NHE-1.4 primers revealed two independent PCR fragments of 1,142 and 845 bp respectively (Fig. 1A). Sequence analysis of cloned RT-PCR products confirmed the upper band as NHE-1 and showed an in-frame deletion of 297 nucleotides (nucleotides 809-1105 of NHE-1; Fig. 1B, C, and D and Fig. 2A) in the lower band (DNHE-1; Fig. 2B). The possibility that this cDNA variant originates from a second gene homologous to NHE-1 or a pseudo-gene was proven to be highly unlikely after database searching using both the canonical NHE-1 or the new DNHE-1 variant sequence as probes. No sequence other than the expected NHE-1 gene was identified in the human genome (data not shown).

DNHE-1 might have originated either directly through the splicing of the GT-TC splice donor-acceptor pair or through a first splicing involving the canonical 5′ GT (of NHE-1) splice site and the noncanonical 3′ TC (of DNHE-1) splice site followed by a second splicing of a smaller intron involving the canonical 5′ GT (of DNHE-1) splice site and the canonical AG (of NHE-1), now as a 3′ splice site (Fig. 1D).

The open reading frame of the DNHE-1 isoform leads to a predicted 99 amino acid–smaller protein (missing transmembrane domains M2, M3, and M4), which lacks the putative amiloride-binding region (30). The putative topology of DNHE-1 obtained from TMAP, a web-based tool for prediction of transmembrane segments in proteins (31), is shown in Fig. 2B. According to this model, DNHE-1 would contain eight transmembrane domains with NH₂- and COOH-terminals on the internal side of the membrane. To evaluate the functional activity of this isoform, we inserted full-length (NHE-1) and DNHE-1 transcripts into a pcDNA3 expression plasmid and then transfected the obtained expression constructs pcNHE-1 and pcDNHE-1 into an NHE-deficient PS200 hamster fibroblast cell line (PS200).

NHE transfectants were selected according to the proton killing test by their ability to survive 1-h of acidification induced by NH₄ prepulse acid loading (28). As expected, cells transfected with NHE-1 survived the acid-loading selection test. On the contrary, cells transfected with DNHE-1 did not survive this procedure, suggesting that DNHE-1–mediated exchange of Na for H, if any, was insufficient to allow cell survival.

To clarify whether DNHE-1 can mediate SLC activity even in the absence of a significant exchange of Na for H, DNHE-1–transfected cells were selected by their acquired resistance to geneticin (pcDNA3 expression plasmid contains the sequence conferring resistance to geneticin).

As described in Fig. 3A, in the presence of specific inhibitors of Na pump, Na-K cotransport, and anion exchanger, exchange of Na for Li in PS200 cells was virtually absent (0.13 ± 0.1 nmol Li⁺ · mg⁻¹ protein · min⁻¹, mean ± SE, n = 3). NHE-1–tranfected cells were instead characterized by an active exchange of Na for Li (10.8 ± 1.6), which was blunted by the addition of DMA, as expected (0.41 ± 0.2, P < 0.05, n = 3) (19). Finally, DNHE-1–transfected cells expressed a small but significant exchange of Na for Li (1.21 ± 0.18), which was consistent with SLC-mediated transport because it was resistant to DMA (1.64 ± 0.08, NS) but was largely reduced by the addition of phloretin (0.10 ± 0.09, P < 0.05, n = 3) (11,20). The inhibitory effect of phloretin on Na-driven Li efflux was similar in DNHE-1–transfected cells (Fig. 3B) and in human erythrocytes (Fig. 3C). Only partial (43%) inhibition of Na-driven Li efflux was exerted by phloretin in NHE-1–transfected cells (Fig. 3D), in line with our previous report in human erythrocytes (22). As shown in Fig. 3E, K_m for external Na was similar in NHE-1–and DNHE-1–transfected cells. Finally, no inhibitory effect of DMA could be demonstrated on either DNHE-1–transfected cells or erythrocytes (Fig. 3F, G). The evidence that Na-driven Li efflux in human erythrocytes is completely insensitive to DMA, while confirming previous reports (20), could conflict with our hypothesis that human erythrocytes express both amiloride-sensitive NHE-1 and amiloride-insensitive DNHE-1. However, we have previously demonstrated that in the human erythrocyte, the amiloride-sensitive (NHE-1–mediated) component of Na-driven Li efflux is virtually inactive in resting conditions (as in this case) and can be activated only after osmotic shrinkage (21).

Because DNHE-1 should be transduced into a smaller protein than NHE-1, we examined protein extracts from the erythrocyte membrane and from PS200 cells transfected with either NHE-1 or DNHE-1 using a polyclonal antibody raised against a peptide shared by NHE-1 and its spliced variant. After BLAST analysis on all nonredundant protein databases, the peptide described above showed only nine 100% homologue matches, and all of them were specific for NHE-1. As expected (Fig. 4), two independent bands were identified in human erythrocyte membranes by immunoblotting and were consistent with full-length NHE-1 (upper band) and with the smaller variant due to the predicted alternative splicing (lower band).

As shown in Fig. 5A, RT-PCR amplification of cDNA extracted from PS200 cells transfected with NHE-1 displayed a single band consistent with full-length NHE-1, while amplification of cDNA extracted from PS200 cells transfected with DNHE-1 originated a lower single band consistent with the alternative splicing.

In parallel, as shown in Fig. 5B, immunoblotting of protein extracted from PS200 cells transfected with NHE-1 originated a single band consistent with full-length NHE-1, while immunoblotting of protein extracted from PS200 cells transfected with DNHE-1 originated a lower single band consistent with the alternative splicing. No bands could be detected in nontransfected PS200, thus confirming the specificity of the antibody. Because the methodology we used to extract membrane proteins was proven to disrupt glycosylation (data not shown), neither erythrocytes nor transfected cells showed glycosylated NHE-1.

To clarify whether the spliced variant of NHE-1 is expressed in tissues other than erythroid cells, we performed a PCR using cDNAs from different human tissues with the same primers (NHE-1.1 and NHE-1.4) employed to identify NHE-1 and DNHE-1 in human erythrocytes and reticulocytes. As shown in Fig. 6, the DNHE-1 isoform was detected in kidney and spleen, suggesting that its presence is not restricted to the hematopoietic system.

Finally, as shown in Fig. 7A, NHE-1 and DNHE-1 expression levels were correlated to each other, as individuals characterized by low NHE-1 expression showed low DNHE-1 expression and vice versa. For this reason, we performed a semiquantitative RT-PCR aimed at evaluating the sum of the expression levels of NHE-1 and DNHE-1. As described in Fig. 7B, individuals with higher SLC activity were also characterized by an increased cumulative expression level of NHE-1 + DNHE-1 normalized for β-actin (1.88 ± 0.2 vs. 0.71 ± 0.04 arbitrary units, mean ± SE, P = 0.0002) when compared with individuals with low SLC activity.

Together, these findings demonstrate that SLC is mediated by an alternative splicing of NHE-1 and that its activity directly depends upon NHE-1 expression.

The results of the present study demonstrate that SLC is mediated by an amiloride-insensitive splicing of NHE-1 and that its activity correlates with the expression of both NHE-1 and its spliced transcript. As shown above, a very large (∼40-kb) intron with canonical splice sites in the relevant gene segment generates NHE-1 mRNA. DNHE-1 is instead the likely result of a splicing event using splice junction sequences in the coding region of NHE-1 involving a GT-TC splice donor-acceptor pair.

Although the identification of a GT-TG splice pair of the human adenylyl cyclase stimulatory G-protein Gα_s has recently been described (32), ours may nonetheless be the first described example of a GT-TC noncanonical splice donor-acceptor pair. The most simple mechanism to explain the origin of DNHE-1 would be that the canonical intron yielding NHE-1 is spliced first, followed by splicing of the noncanonical intron, to yield DNHE-1. If so the NHE-1 expression construct would also generate DNHE-1; however, this is not the case, as described in results and in Fig. 5, panel A. This result supports the alternative possibility that the splicesome recognizes both the canonical AG (of NHE-1) and, less frequently, the noncanonical TC donor splice site. Such splicing and resplicing mechanisms have been described previously, especially when very large introns are involved (33). The latter mechanism might also be consistent with the evidence (32) that atypical noncanonical sites are often near canonical sites. In our case, canonical NHE-1 and noncanonical DNHE-1 3′ splice sites are not that close (175 bp), but relative to the large size of the intron they become close, which could account for the use of the atypical site.

The hypothesis, based on functional evidence, that SLC could represent an operational mode of NHE has actually long been suggested (13,21,22). Nonetheless, this is the first time that this assumption has been definitively confirmed by a molecular approach.

The significance of the identification of the gene responsible for SLC activity rests in two main arguments: 1) This result might be of help in perspectively finding a genetic marker useful for the early detection of patients predisposed to the development of essential hypertension and diabetic nephropathy. The molecular dysfunction that underlies the deranged SLC activity should in fact be, at least in principle, a more sensitive marker of disease than the SLC overactivity by itself. 2) This finding, by translating to NHE-1 the characteristics of inheritance and predictivity previously attributed to SLC, once more designates NHE-1 as a candidate to explain the pathogenesis of essential hypertension and, consequently, diabetic nephropathy.

A few years ago, after genetic linkage analysis, NHE-1 was excluded as the gene responsible for elevated SLC activity and therefore as a candidate gene for essential hypertension (34). Our finding that SLC activity is mediated by DNHE-1 and correlates with the expression of NHE-1 transcripts should nonetheless renew interest in the potential contribution of this gene to the pathogenesis of essential hypertension. At present, indeed, it is not possible to exclude the possibility that recently identified NHE-1 gene single-nucleotide polymorphisms (National Center for Biotechnology Information’s Entrez system) might correlate with SLC activity, NHE-1 transcript levels, and hypertension.

As shown by our experiments, DNHE-1, although virtually unable to mediate a significant exchange of Na for H when transfected into NHE-deficient cells, is nonetheless capable of mediating a “remnant” exchange of Na for Li. This finding, although surprising at first sight, is justified by a few pieces of evidence: 1) previous studies indicate that mutations in the amiloride binding region induce a reduced sensitivity to amiloride paralleled by slower Na transport (35); 2) functional studies have suggested that SLC is unable, possibly because of an excessive affinity for external H, to exchange Na for H (36,37); and 3) SLC activity in human erythrocytes has been demonstrated to be 10–50 times less active than NHE (38). Finally, recent evidence (39) suggests that mutations in the fourth transmembrane domain of NHE-1 (a segment that is lost in DNHE-1, as shown in Fig. 2) reduce the affinity of the exchanger for external Na. Interestingly, the same report also indicates that none of the mutations introduced affect the affinity of the exchanger for Li.

In conclusion, we have identified in human erythrocytes and reticulocytes a spliced variant of NHE-1 lacking the nucleotide sequence that specifies for the amiloride-binding region. Once transfected in NHE-lacking cells, this smaller variant restores an amiloride-insensitive, phloretin-sensitive exchange of Na for Li compatible with SLC activity. Finally, expression of both regular and spliced transcripts of NHE-1 is increased in subjects with high SLC activity. Analysis of NHE-1 polymorphisms should clarify the molecular basis of NHE and SLC hyperactivity found in essential hypertension and diabetic nephropathy.

The financial support of Telethon, Italy (Grant no. E.816) is gratefully acknowledged.