Functional characterization of human SLC41A1, a Mg2+ transporter with similarity to prokaryotic MgtE Mg2+ transporters

Angela Goytain, Gary A. Quamme

Abstract

We have begun to identify and characterize genes that are differentially expressed with low magnesium. One of these sequences conformed to the solute carrier SLC41A1. Real-time RT-PCR of RNA isolated from renal distal tubule epithelial [mouse distal convoluted tubule (MDCT)] cells cultured in low-magnesium media relative to normal media and in the kidney cortex of mice maintained on low-magnesium diets compared with those animals consuming normal diets confirmed that the SLC41A1 transcript is responsive to magnesium. Mouse SLC41A1 was cloned from MDCT cells, expressed in Xenopus laevis oocytes, and studied with two-electrode voltage-clamp studies. When expressed in oocytes, SLC41A1 mediates saturable Mg2+ uptake with a Michaelis constant of 0.67 mM. Transport of Mg2+ by SLC41A1 is rheogenic, voltage dependent, and not coupled to Na+or Cl. Expressed SLC41A1 transports a range of other divalent cations: Mg2+, Sr2+, Zn2+, Cu2+, Fe2+, Co2+, Ba2+, and Cd2+. The divalent cations Ca2+, Mn2+, and Ni2+ and the trivalent ion Gd3+ did not induce currents nor did they inhibit Mg2+ transport. The nonselective cation La3+ abolished Mg2+ uptake. The SLC41A1 transcript is present in many tissues, notably renal epithelial cells, and is upregulated in some tissues with magnesium deficiency. These studies suggest that SLC41A1 is a regulated Mg2+ transporter that might be involved in magnesium homeostasis in epithelial cells.

  • magnesium
  • differentially regulated
  • expression
  • Xenopus oocytes

magnesium plays a fundamental role in cellular metabolism so that its control within the body is critical. Magnesium homeostasis is principally a balance between the intestinal absorption of dietary magnesium and renal excretion of urinary magnesium (11). We have shown that the kidney, mainly the distal convoluted tubule, controls magnesium reabsorption (3). Although renal reabsorption is under the influence of many hormones, selective regulation of magnesium transport is due to intrinsic control involving transcriptional processes and synthesis of transport proteins (4). Using microarray analysis, we have begun to identify the genetic elements involved with this transcriptional control.

Recently, Wabakken and co-workers (19) reported that human SLC41A1 has a homology to MgtE Mg2+ transporters found in certain bacteria. SLC41A (GenBank accession no. AJ514402) is a member of the human solute carrier family designated by the human genome nomenclature committee (www.gene.ucl.ac.uk/nomenclature/), specifically, solute carrier family 41 member 1. Computer analysis of the SLC41A1 protein structure reveals that it comprises 10 putative transmembrane domains, 2 of which are highly homologous to the integral membrane part of the prokaryote MgtE protein family. MgtE was initially identified in the bacteria Providencia stuartii and Bacillus firmus by Smith, Maguire, and their colleagues in their search for members of the CorA (cobalt resistance) family of Mg2+ uptake transporters (1517). On the basis of the similarity of SLC41A1 to MgtE, Wabakken et al. (19) suggested that the human solute carrier SLC41A1 may be a eukaryotic Mg2+ transporter.

With this knowledge, we set out to determine whether SLC41A1 is differentially expressed in renal epithelial cells and if it encodes a functional Mg2+ transporter. Our initial experiments showed that the SLC41A1 transcript is upregulated in response to low magnesium, satisfying our premise that transcriptional changes are involved with magnesium homeostasis (3). Subsequently, we cloned mouse SLC41A1 cDNA from mouse distal convoluted tubule (MDCT) cells, sequenced it, prepared capped cRNA, injected it into Xenopus oocytes, and performed two-electrode voltage-clamp studies. Our data show that SLC41A1 translocates Mg2+ in oocytes expressing the protein. We conclude that the SLC41A1 is a Mg2+ transporter that is responsive to magnesium balance.

EXPERIMENTAL PROCEDURES

Animal preparation and cell culture.

Study protocols were approved by the University of British Columbia Animal Care Committee. Male mice were maintained for 5 days on a low-magnesium diet (ICN diet no. 902205, Nutritional Biochemicals; Cleveland, OH) or on this diet supplemented with 0.05% MgSO4, which is comparable to normal mouse chow. The plasma concentrations were 0.13 ± 0.01 and 0.75 ± 0.09 mM, respectively, confirming that mice consuming a magnesium-restricted diet were relatively magnesium deficient (11).

MDCT cells were originally derived by Pizzonia et al. (8). The MDCT cell line has been extensively used by us to study renal magnesium transport (3). Cells were grown in basal DMEM-Ham’s F-12 (1:1) media (GIBCO) supplemented with 10% fetal calf serum (Flow Laboratories; McLean, VA), 1 mM glucose, 5 mM l-glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin in a humidified environment of 5% CO2-95% air at 37°C. Where indicated, subconfluent MDCT cells were cultured in Mg2+-free media (Stem Cell Technologies; Vancouver, British Columbia, Canada) for 16 h. Other constituents of the Mg2+-free culture media were similar to the complete media.

Quantitative analysis of SLC41A1 transcripts by real-time RT-PCR.

Total RNA of cells was extracted by TRIzol (Invitrogen). Genomic DNA contamination was removed by a DNA-free kit (Ambion) before the first-strand cDNA was made. Standard curves were constructed by serial dilution of a linear pGEM-T vector (Promega) containing the SLC41A1 gene. The primer set of mouse SLC41A1 was forward, 5′-CAT CCA TCG CAG CCG TCG TCT TT-3′, and reverse, 5′-AGC CTG AAT ACA ACA CCT CCC T-3′. PCR products were quantified continuously with AB7000 (Applied Biosystems) using SYBR green fluorescence according to the manufacturer’s instructions. The relative amounts of SLC41A1 RNA were normalized to mouse β-actin transcripts.

Plasmid construction and generation of expression constructs.

A RT-PCR strategy was used to isolate mouse SLC41A1 cDNA from the MDCT cell line. cDNA comprising the open reading frame (ORF) of SLC41A1 was amplified from the MDCT cDNA using the cloning primers (sense: 5′-TGC CTG TCA TGT CCT CTA AG-3′; antisense: 5′-TGA ATG GCT AAG CTA GTC CC-3′) and subcloned into the XbaI and SacII restriction sites of the pcms-EGFP expression vector. Isolated clones were subcloned and sequenced to confirm sequence integrity. Standard protocols were used for Escherichia coli cloning procedures and DNA sequencing (12). To synthesize cRNA, the cDNA constructs were linearized and then transcribed with T7 polymerase in the presence of m7GpppG cap using the mMESSAGE MACHINE T7 KIT (Ambion) transcription system.

Protein motifs were identified using BLASTP and the SWISSPROT database. Membrane topology was predicted by SOSUI and Kyte-Doolittle hydrophobicity analysis.

Expression of SLC41A1 in Xenopus oocytes and current measurements.

Xenopus oocytes were prepared and injected with cRNA, and electrophysiological recordings were preformed according to techniques previously described (10). Briefly, defolliculated stage V-VI oocytes were typically injected with 50 ng cRNA in 50 nl H2O. Oocytes were incubated at 18°C for 3–5 days in multiwell tissue culture plates containing Barth’s solution [88 mM NaCl, 1.0 mM KCl, 2.4 mM NaHCO3, 1.0 mM MgSO4, 1.0 mM CaCl2, 0.3 mM Ca(NO3)2, 10 mM HEPES-NaOH (pH 7.6), 2.5 mM Na-pyruvate, 0.1% BSA, 10,000 U/l penicillin, and 100 mg/l streptomycin]. Oocytes were injected with SLC41A1 cRNA or, for control observations, H2O or kidney total poly(A)+ RNA; no Mg2+-induced currents were detected in the latter. To record expressed membrane currents, the oocytes were placed in a recording chamber (0.3 ml) and perfused with modified Barth’s solution (96 mM NaCl and 10 mM HEPES-NaOH) containing various concentrations of MgCl2, as indicated, in substitution for osmotically equivalent amounts of NaCl . All experiments were performed at room temperature (21°C).

Steady-state membrane currents were recorded with the two-microelectrode voltage-clamp technique using a Geneclamp 500 amplifier (Axon Instruments). Electrophysiology consisted of a voltage-clamp step profile consisting of a holding potential of −15 mV, followed by eight episode series of +25-mV steps of 2-s duration from −150 to +25 mV within an episode duration of 6.14 s. Each episode recorded 1,536 data points collected at 4-ms intervals. The data were filtered at the appropriate frequency before digitization. To assess the permeability of different divalent cations, we used the shift in the reversal potentials of the respective cation from the reversal potentials of Mg2+ currents (Erev) and calculated the permeability ratio (Px/PMg) by Math where R is the gas constant, T is temperature, and F is Faraday’s constant. Voltage-clamp episodes in the presence of extracellular test cations were corrected against episodes in the absence of external test cations.

All experimental conditions were performed on oocytes harvested from a minimum of three different animals.

RESULTS

Identification of SLC41A1 as a magnesium-responsive gene.

With the knowledge that differential gene expression is involved with selective control of epithelial cell magnesium conservation, our strategy was to identify candidates that were upregulated with low magnesium. Using real-time RT-PCR, we first showed that SLC41A1 mRNA is increased 4.1-fold in the kidney cortex of hypomagnesemic mice (n = 3 separate animals) and 1.6-fold in MDCT cells (n = 3 independent preparations) cultured in low magnesium compared with normal animals and cells, respectively (data not shown).

SLC41A1 elicits Mg2+ currents in expressing Xenopus oocytes.

To determine if SLC41A1 encodes a functional Mg2+ transporter, we cloned mouse SLC41A1 cDNA from MDCT cells, prepared cRNA, injected it into Xenopus oocytes, and characterized Mg2+-evoked currents using two-microelectrode voltage-clamp analysis. The electrophysiological data gave evidence for a rheogenic process with large inward currents, in the order of 1.0 μA, in SLC41A1 cRNA-injected oocytes, whereas there were no appreciable currents in control H2O- or total poly(A)+ RNA-injected cells from the same batch of oocytes (Fig. 1A). Mouse SLC41A1-mediated Mg2+-evoked uptake was linear for at least 20 min and did not display any time-dependent decay during repetitive stimulation with voltage steps (data not shown). Erev was significantly shifted to the right, as would be expected of a magnesium transporter (Fig. 1A). Steady-state Mg2+-evoked currents were saturable (Fig. 1B). The Michaelis-Menten constant (Km) was 0.67 ± 0.09 mM (n = 29) when measured at a holding potential of −125 mV (Fig. 1B, inset) and 0.86 ± 0.13 mM (n = 29) at −75 mV (Fig. 1C, inset). Km was independent of the membrane potential used to determine the saturation kinetics.

Fig. 1.

SLC41A mediates Mg2+ currents in Xenopus oocytes. A: current-voltage (I-V) relationships obtained from linear voltage steps from −150 to +25 mV in the presence of Mg2+-free solutions or those containing the indicated concentrations of MgCl2. Oocytes were clamped at a holding potential of −15 mV and stepped from −150 to +25 mV in 25-mV increments for 2 s at each of the concentrations indicated. Shown are average I-V curves obtained from control H2O-injected (n = 3) or SLC41A1-expressing (n ≥ 3) oocytes. Note the positive shift in reversal potential, indicated by arrows, with increments in Mg2+ concentration. Values are means ± SE of observations measured at the end of each voltage sweep for the respective Mg2+ concentration. Vm, membrane potential. B: summary of concentration-dependent Mg2+-evoked currents in SLC41A1-expressing oocytes using a holding potential of −125 mV. Mean ± SE values are those given in A. Inset: Eadie-Hofstee plot of concentration-dependent Mg2+-evoked currents demonstrating a Michaelis-Menten constant (Km) of 0.67 ± 0.09 mM. C: summary of concentration-dependent Mg2+-evoked currents in SLC41A1 oocytes using a holding potential of −75 mV. Mean ± SE values are those given in A. Inset, Eadie-Hofstee plot with a Km of 0.86 ± 0.13 mM. Km was independent of the respective holding potential.

Mg2+-evoked currents were not altered with deletion of external NaCl by substitution with choline Cl or replacement of NaCl and MgCl2 with appropriate amounts of Na gluconate and Mg2SO4, indicating that transport does not depend on extracellular Na+ or Cl or is influenced by these electrolytes (data not shown).

Heterologous expressed SLC41A1 mediates the transport of a number of other divalent cations in addition to Mg2+ (Fig. 2A). Relative to Mg2+, the divalent cations Sr2+, Fe2+, Ba2+, Cu2+, Zn2+, and Co2+ evoked considerable currents, whereas Cd2+ and Mn3+ were poorly permeable (Fig. 2B). Ni2+ and Ca2+ and the trivalent cation Gd3+ did not evoke appreciable currents. These data indicate that SLC41A1 is a relatively nonselective divalent cation transporter.

Fig. 2.

Characterization of SLC41A1-mediated currents in Xenopus oocytes. A: substrate specificity of SLC41A1 after application of test cations (2.0 mM) in the absence of external Mg2+. For clarity, only Mg2+, Sr2+, Zn2+, Cd2+, and Ca2+ are shown. Oocytes were clamped at a holding potential of −15 mV and stepped from −150 to +25 mV in 25-mV increments for 2 s for each of the cations. The reversal potential for each of the cations is indicated by an arrow. Values are means ± SE of currents measured at the end of each voltage sweep for the respective divalent cation. B: summary of permeability sequence of the tested divalent cations. Shown are average permeability ratios given in A. C: inhibition of Mg2+-evoked currents with 0.2 mM test cation except for Ca2+, which was tested at 5.0 mM, in the presence of external 2.0 mM Mg2+. For clarity, only Mg2+, Ni2+, Mn3+, Ca2+, Gd3+, and La are shown. The reversal potentials for each of the cations are indicated by arrows. Values are means ± SE of currents measured at the end of each voltage sweep for the respective cation. D: summary of potency sequence of inhibition by multivalent cations based on the change in reversal potential given in C. The inhibitor was added with MgCl2, and voltage clamp was performed about 5 min later.

Next, we determined if the cations that were not transported by SLC41A1 would inhibit Mg2+-evoked currents (Fig. 2C). Relatively large concentrations, 0.2 mM, of the respective inhibitors Mn2+, Ni2+, and Gd3+ and 5.0 mM Ca2+ were tested in the presence of 2.0 mM MgCl2 (Fig. 2D). These concentrations were large relative to what would be physiological values. Erev for Mg2+ was not altered in the presence of 5.0 mM Ca2+ or 0.2 mM of either Ni2+ or Mn2+; however, 0.2 mM Gd3+ significantly shifted Erev to the left. The nonselective blocker La3+ markedly inhibited expressed Mg2+ currents at 0.2 mM (Fig. 2C). On balance, these data indicate that SCL42A1 cRNA-induced transport in oocytes supports a number of divalent cation substrates, but Ca2+ is neither transported nor does it inhibit Mg2+ transport.

Sequence analysis and chromosomal location.

Isolated mouse SLC41A1 cDNA contains an ORF of 1,539 bp, which predicts a protein of 512 amino acids with a calculated molecular mass of 54.9 kDa (Fig. 3A). Mouse SLC41A1 cDNA is similar (98%) to the human sequence (19). Hydrophobicity analysis using the Susui program predicts an integral membrane protein containing 10 hydrophobic transmembrane-spanning regions (Fig. 3B). Analysis of the amino acid sequence by the NetNGlyc 1.0 and NetPhos 2.0 server programs reveals several consensus sites for posttranslational modification. There is a possible glycosylation site at amino acid residue N-475 on the last putative extracellular loop. SLC41A1 contains four putative cAMP-dependent protein kinase phosphorylation sites at residues S-157, S-308, S-393, and T-508 and four possible protein kinase C phosphorylation sites at residues T-80, T-167, T-387, and S-407. There is also one predicted casein kinase II motif at residue S-317 and a myristoylation sequence at G-143. The presence of these consensus sequences might indicate posttranslational modification and regulation of localization and function.

Fig. 3.

Molecular characterization of SLC41A1. A: amino acid sequence of mouse SLC41A1. Transmembrane segments are underlined, and predicted NH2-linked glycosylation sites (×), protein kinase A (▪), protein kinase C phosphorylation (⧫), casein kinase II (•), and myristoylation (+) sites are indicated.

Query of GenBank revealed genomic sequences that were identical to SLC41A1 (GenBank accession number NM-173865), and alignment with SLC41A1 cDNA predicts the gene structure. The coding sequence possesses 11 exons, each of which is about 200 bp in length. The mouse SLC41A1 gene is localized to chromosome 1E4. The human genomic sequence of SLC41A1 (GenBank accession no. NM.173854) was assigned to chromosome band 1q31–32.

Tissue distribution of SLC41A1 mRNA expression and the response to magnesium.

We used quantitative real-time RT-PCR analysis to show that the SLC41A1 transcript is present in many tissues of normal mice (Fig. 4). The heart possesses the most transcript, but mRNA is abundant in the brain, kidney, and liver with lesser amounts in the colon and small intestine, in that order. The variability, particularly in the heart and brain, shown here may be due to the different cells comprising these tissues as we did not attempt to isolate specific cell types.

Fig. 4.

Tissue distribution of mouse SLC41A1 expression and responsiveness of the SLC41A1 transcripts to Mg2+. Quantitative RT-PCR analysis was performed with SLC41A1 mRNA purified from the indicated tissues derived from mice maintained on a normal (0.5% by weight) or low-Mg2+ (<0.01%) diet for 5 days. Mouse SLC41A1 and murine β-actin RNA were measured with real-time RT-PCR (AB7000, Applied Biosystems) using primers designed to mouse SLC41A1 and β-actin. Standard curves for each were generated by serial dilution of each plasmid DNA. The expression level of the mouse SLC41A1 transcript was normalized to that of the mouse β-actin transcript, which was measured in the same cDNA sample. Relative tissue distribution of SLC41A1 mRNA was normalized to mRNA found in the small intestine, the tissue with the lowest transcript. Also shown is the response of SLC41A1 mRNA to low-Mg2+ balance for each of the tissues again normalized to the lowest transcript found in the small intestine. Data were obtained from triplicate experiments and are indicated as means ± SE.

Next, we determined the change in expression levels of the SCL41A1 transcripts in selected tissues with a decrease in magnesium balance. Mice were placed on low-magnesium- or normal-magnesium-containing diets for 5 days before death. The mean plasma concentration was 0.75 ± 0.09 mM for normal animals and 0.13 ± 0.01 mM for mice maintained on low-magnesium diets. The respective urinary magnesium concentrations were 13.2 ± 1.2 and 1.1 ± 0.3 mM. SCL42A1 mRNA significantly increased in the kidney (4.0-fold), colon (2.1-fold), and heart (2.0-fold) (Fig. 4). The transcript decreased in the liver and small intestine. Taken together, SLC41A1 is widely distributed among tissues and is responsive in some organs to magnesium balance.

DISCUSSION

In an effort to identify additional members of the CorA family, the major Mg2+ transport system of Eubacteria and Archaea, Maguire and colleagues (6) isolated the MgtE family and found them represented in a number of bacteria. MgtE comprises a protein of about 40 kDa and possess five to six transmembrane segments. Bacterial MgtE translocates Mg2+ and Co2+ inward but does not transport Ni2+ (16, 17). Moreover, Ca2+, Mn2+, or Zn2+ did not inhibit Mg2+ influx. Although a multivalent transporter, the primary function of MgtE family in bacteria is not fully understood (3). Recently, Wabakken and colleagues (19) identified and structurally characterized a human gene and protein product with homology to bacterial MgtE. The human protein was designated solute carrier family 41 subtype 1, SLC41A1, by the human genome nomenclature committee. It consists of 513 amino acids and has a molecular size of 56 kDa. Hydropathy analysis revealed that the protein consists of 10 putative transmembrane domains, two [NH2-terminal PX6GN and COOH-terminal P(D/A)X4P X6D] of which are very similar (52% and 46%, respectively) to the integral membrane domains of the MgtE protein with an overall similarity of 15% between human SCL41A1 and Bacillus firmus (SWISSPROT accession no. I40201) amino acid sequences (19). On the basis of this structural similarity, Wabakken et al. (19) suggested that SLC41A1 may have similar functions in eukaryotic cells as those in prokaryotes.

With the knowledge that epithelial Mg2+ absorption is regulated to a major degree by differential expression of genes encoding magnesium transport proteins, we speculated that if SLC41A1 functions to translocate Mg2+, it would be responsive to magnesium. Real-time RT-PCR showed that the transcript was upregulated with low magnesium. This led us to express SLC41A1 in Xenopus oocytes to determine whether it mediates Mg2+ transport as speculated by Wabakken and colleagues (19). In keeping with our initial premise, we found that SLC41A1 expression is responsive to magnesium and functions as a Mg2+ transporter.

Clearly, SLC41A1 does not have the same cationic specificity as MgtE (16, 17). SLC41A1 does not transport Co2+ to any degree, and Ni2+ is not a selective inhibitor of Mg2+ transport. This is not surprising as SLC41A1 shares only 40% similarity with 2 of 12 membrane segments of the bacterial transporter with an overall similarity of about 15%. SLC41A1 prefers Mg2+ as a substrate compared with other divalent cations; nevertheless, it is evident that it could perform a physiological function within cells to transport Fe2+, Zn2+, Cu2+, Co2+, and Cd2+. Because SLC41A1 is upregulated with low magnesium, it is interesting to speculate that the transport of these multivalent cations may be increased in magnesium deficiency. In support of this notion, we reported that cadmium uptake into the renal epithelial distal tubule cell line, Madin-Darby canine kidney cells, was greater after cellular magnesium depletion than that in control cells (9). These studies used fluorescence changes of mag-fura-2 to sensitively measure cellular free Cd2+ activity to determine Cd2+ transport. This observation lead to the notion that heavy metal intoxication may be more severe in individuals with low magnesium (9).

Although there is ample evidence for unique Mg2+ transporters in a variety of animals and tissues, only a few have been identified at the molecular level (11). Of the identified mammalian magnesium transporters, Nadler et al. (7) first identified TRPM7, a widely expressed member of the transient receptor potential melastatin (TRPM) ion channel family that produces a Mg2+ current in a wide variety of cells. TRPM7 is regulated by intracellular Mg·ATP levels and is similarly permeable to both major divalent cations Ca2+ and Mg2+ but also many of the trace elements, such as Zn2+, Mn2+, and Co2+ (14). Using a positional cloning approach, Schlingmann et al. (13) and Walder et al. (20) found that hypomagnesemia with secondary hypocalcemia (HSH) was caused by mutations in TRPM6, a new member of the TRPM family. HSH is an inherited disease affecting both intestinal and renal Mg2+ absorption (4). The functional characteristics of the TRPM6 transporter have not been fully elucidated (2, 18). Schweyen and colleagues (1, 5) have demonstrated that the mitochondrial RNA splicing2 (Mrs2) gene encodes a protein that is present in yeast and mammalian inner mitochondrial membranes. Mrs2 mediates high-capacity Mg2+ influx in isolated yeast mitochondria driven by the inner membrane potential but also transports a range of divalent cations such as Ni2+, Co2+, and Cu2+ (1). Overexpression of Mrs2 increases influx, whereas deletion of the gene abolishes uptake, suggesting that it is the major mitochondrial system. These data suggest that the Mrs2 protein may mediate Mg2+ transport in mammalian mitochondria. It is disparaging that, despite the significance of cellular Mg2+, only three specific magnesium transporters have been described in mammalian cells to date. Thus it is important that SLC41A1 was found to mediate Mg2+ transport. The role of SLC41A1 in the cellular magnesium balance is unclear, but it is regulated as would be expected of a physiologically relevant transporter (11).

In summary, we have shown that the eukaryote solute carrier SLC41A1 functions as a multivalent transporter with preference to physiological magnesium concentrations. Although it has only a small similarity to the bacterial Mg2+ transporter family, MgtE, it might have evolved from a common ancestor. The cationic permeability of mouse SLC41A1 is certainly different from bacterial MgtE. We also demonstrate that SLC41A1 is regulated by magnesium in that the transcript is increased with low magnesium. This is consonant with our premise that magnesium transporters that are involved with renal magnesium balance are regulated by magnesium-dependent transcriptional mechanisms. We infer from these data that SLC41A1 may play a role in epithelial magnesium homeostasis.

GRANTS

This study was supported by Canadian Institutes of Health Research Research Grant MOP-53288 and the Kidney Foundation of Canada.

Footnotes

  • Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

    Address for reprint requests and other correspondence: G. A. Quamme, Dept. of Medicine, Vancouver Hospital, Koerner Pavilion, 2211 Wesbrook Mall, Vancouver, British Columbia, Canada V6T 1Z3 (E-mail: quamme{at}interchange.ubc.ca).

    doi:10.1152/physiolgenomics.00261.2004.

REFERENCES

View Abstract