In mammals, iron is thought to be transported into cells by transferrin-receptor-mediated endocytosis. However, since the available apo-transferrin in the intestinal lumen (that arising from biliary excretion) is insufficient to account fully for dietary iron absorption6, it is likely that other non-receptor-mediated uptake systems exist in the intestine. Dietary iron comprises two forms, haem and non-haem iron. The bulk of intestinal non-haem iron is absorbed in the first portion of the duodenum, in which the acidic environment promotes solubilization of iron rendered in its 2+ reduced state by ferrireductase7,8 and ascorbate9,10. How Fe2+ is subsequently absorbed however is poorly understood. The divalent cation transporters identified so far include the mono- or oligospecific transporters derived from plants and yeast11. In addition, mammalian divalent cation export systems have been described for zinc12 and copper13, but a mammalian uptake system has not yet been identified.

We report the expression cloning, tissue distribution and initial characterization of a divalent-cation transporter, DCT1. By screening for iron uptake activity in oocytes, we isolated the DCT1 complementary DNA from a cDNA library prepared using duodenal messenger RNA from rats fed on a low-iron diet (Fe(−)D). Xenopus oocytes injected with Fe(−)D mRNA showed a 7-fold increase in 55Fe uptake compared with water-injected (control) oocytes (Fig. 1). Following size-fractionation, maximal 55Fe uptake activity was induced by a 4.0–4.5-kilobase (kb) mRNA fraction, from which we isolated a single cDNA (DCT1) which stimulated 55Fe2+uptake 200-fold (Fig. 1). The 4,409-base-pair (bp) DCT1 cDNA encodes a 561-amino-acid protein with 12 putative membrane-spanning domains (Fig. 2), predicted glycosylation sites in the fourth extracellular loop, and a consensus transport motif in the fourth intracellular loop3,4,5. In bacteria, this motif is believed to participate in the interaction with ATP-coupling subunits; however its function is unknown and it is distinct from the nucleotide-binding fold of ABC transporters.

Figure 1: Uptake of 10 µM 55Fe in Xenopus oocytes injected with poly(A)+RNA from normal or iron-de.
figure 1

ficient (Fe(−)D) rat duodenum or with RNA synthesized from DCT1 cDNA. Data are mean ± s.e.m. from 6–10 oocytes. Inset, relative 55Fe uptake in oocytes injected with size-fractionated poly(A)+Fe(−)D RNA. C, Control water-injected oocytes; 1, 2.0–3.0 kb poly(A)+RNA; 2, 2.5–3.5 kb; 3, 3.3–4.0 kb; 4, 3.8–4.5 kb; 5, 4.4–5.7 kb; 6, 4.5–6.0 kb; 7, 4.8–6.4 kb; 8, 5.5–7.0 kb; and T, unfractionated poly(A)+Fe(−)D RNA.

Figure 2: Sequence alignment of DCT1 and Nramp-related polypeptides.
figure 2

The sequences of rat DCT1, human Nramp1 and Nramp2 (GenBank accession numbers, L32185 and L37347) were aligned using the GCG program, and identical residues are indicated by shading. The 12 putative transmembrane regions are underlined and numbered 1–12. b, Twelve-transmembrane-domain model of the DCT1 protein. Putative transmembrane domains 1 to 12 are indicated. The ‘consensus transport motif’3,4,5 is indicated (CTM) in the fourth intracellular loop, and putative N-linked glycosylation sites are identified in the fourth extracellular loop. c, The potential iron-responsive regulatory-protein-binding site (IRE) consensus sequence (CNNNNNCAGUG)24,25, predicted to form a stem–loop.

High-stringency northern-blot analysis of DCT1 transcripts (Fig. 3a) revealed prominent bands of 4.5 kb from proximal intestine, kidney, thymus and brain, and fainter bands in testis, liver, colon, heart, spleen, skeletal muscle, lung, bone marrow and stomach. By loading an equal amount of each mRNA, we found that DCT1 was expressed at a much higher level in proximal intestine than in kidney, and more so in kidney than in brain. We observed another strong band at 3.5 kb in kidney and thymus, indicating that a second isoform may exist in those tissues. In the intestine, DCT1 mRNA expression was highest in duodenum, and decreased towards the colon (Fig. 3b, top row). Following diet-induced iron deficiency, DCT1 mRNA expression was enhanced dramatically in duodenum (Fig. 3b, bottom row) and also, but to a lesser extent, in kidney, liver, brain, heart, lung and testis (not shown), suggesting that there is marked regulation of DCT1 mRNA by dietary iron and/or tissue iron concentration.

Figure 3: High-stringency northern blot analysis of RNA from rat tissues probed with 32P-labelled DCT1 cDNA.
figure 3

Each well was loaded with 3 µg poly(A)+RNA from a range of a, whole tissues, or b (top row), specific regions of brain, kidney and intestine, including duodenum from rats fed a low-iron diet (Fe(−)D); bottom row, duodenal samples after short exposure.

We investigated the cellular localization of DCT1 mRNA in small intestine, kidney, testis, thymus and brain, in frozen sections using non-radioactive in situ hybridization with a digoxigenin-labelled cRNA probe. High-stringency hybridization yielded specific signals with antisense probe, but no labelling with sense probe (Fig. 4). In small intestine, DCT1 was highly expressed in enterocytes lining the villus, especially in the crypts and lower segments of the villi, but not at villar tips (Fig. 4a). Again, a proximal-to-distal gradient of expression was evident in the small intestine. This pattern of expression is consistent with the primary sites that are responsible for intestinal absorption of most divalent cations9.

Figure 4: Tissue localization of rat DCT1 mRNA detected by in situ hybridization.
figure 4

Bright-field micrographs of cryosections hybridized to digoxigenin-labelled DCT1 antisense (or sense in c, h) cRNA probe. a, Duodenum (M, muscle layer; V, villi; L,lumen). b, Kidney (CO, cortex; OS, outer stripe of the medulla; IS, inner stripe; IM,inner medulla). c, Kidney hybridized with sense probe. d, Thymus (C, cortex; M, medulla). e, Sagittal section through brain; DCT1 mRNA was found in neurons throughout the brain. f, Testis (S, Sertoli cells; L, lumen). g, In hippocampus, positive labelling was prominent in the pyramidal and granule cells. No labelling was obtained with sense probe (h). i, Anterior olfactory nucleus (m, medial; d, dorsal; l, lateral; v, ventral). j, Substantia nigra (pr, pars reticulata; pc,pars compacta). k, Choroid plexus in the fourth ventricle. Scale bars: 100 µm in a, d, f, i, k; 1.5 mm in b, c; 3 mm in e; and 200 µm in g, h, j.

In kidney, DCT1 mRNA labelling was most prominent in S3 proximal tubule segments (Fig. 4b), suggesting that it is involved in the reabsorption of divalent cations. Staining was also significant over the entire length of the collecting ducts, where DCT1 may participate in the final reabsorption of these ions. In testis, DCT1 mRNA was expressed in the Sertoli cells of seminiferous tubules (Fig. 4f), and was more abundant in those tubules containing mature spermatocytes. In thymus, DCT1 labelling was positive in cortical, but not medullary, thymocytes (Fig. 4d).

In each region examined in brain, DCT1 mRNA was found in neurons but not glial or ependymal cells (Fig. 4e). A qualitative examination of sagittal sections indicated that most neurons expressed DCT1 mRNA at low levels. More prominent labelling was present in densely packed cell groups, such as the hippocampal pyramidal and granule cells (Fig. 4g), cerebellar granule cells, the preoptic nucleus and pyramidal cells of the piriform cortex. Moderate amounts of DCT1 mRNA were present in the substantia nigra (Fig. 4j). In Parkinson's disease, iron accumulates in affected neurons of the substantia nigra2; this increased iron deposition may contribute to neuron death by inducing the production of hydroxyl radicals according to the Fenton reaction. One cell group, the ventral portion of the anterior olfactory nucleus (Fig. 4i), contained large amounts of DCT1 mRNA, at least an order of magnitude more than elsewhere in the brain. DCT1 mRNA was localized in epithelial cells of the choroid plexus (Fig. 4k). Although the different functions of DCT1 in the central nervous system still need to be investigated, it is likely that defects in DCT1 contribute to the aetiology of certain neurodegenerative diseases by promoting the generation of reactive oxygen species by divalent cations.

We investigated the functional characteristics of DCT1 by using oocytes injected with DCT1 mRNA. Two-microelectrode voltage-clamp analysis indicated that divalent-cation transport mediated by DCT1 was rheogenic, with currents of up to −1,000 nA (Fig. 5). Whereas Fe2+ evoked no appreciable current (<5 nA) in control oocytes clamped at −50 mV (Fig. 5a (1)), in oocytes expressing DCT1 (Fig. 5a (2)) we observed large inward currents (IFe) when 50 µM Fe2+ was superfused at pH 5.5. In DCT1-injected oocytes in the absence of substrate, switching extracellular pH (pHo) from 7.5 to 5.5 caused a larger shift in the baseline (an inward current) than in control oocytes. In two oocytes expressing DCT1, this pH-dependent current was −24.0 nA and −23.5 nA (each the average of several replicates), compared with −5.7 ± 2.7 nA (mean ± s.d.) in control oocytes from the same batch. Ca2+ did not mimic the effect of Fe2+; instead, 10 mM Ca2+ produced a small outward current at pHo 5.5 and partially inhibited the current evoked by 50 µM Fe2+(Fig. 5a (2)). Large inward currents were obtained when Zn2+, Mn2+, Cu2+, Co2+ or Cd2+ were applied (each at 50 µM); smaller currents were evoked by Ni2+ and Pb2+(Fig. 5b).

Figure 5: Currents associated with the divalent cation transporter DCT1 expressed in oocytes.
figure 5

a, Current was continuously monitored in (1) a control oocyte and (2) a single oocyte expressing DCT1, each clamped at −50 mV andsuperfused for periods indicated by the boxes in the top panel, first at pH7.5(blank) then at pH 5.5 (diagonal hatch). Iron or additional calcium was applied for the periods shown by the solid bars, then washed out with substrate-free medium, pH 7.5. The dotted horizontal lines indicate the approximate baseline obtained in pH 5.5 medium before the addition of test substrates. b, Substrate specificity of DCT1, applying test substrates at 50 µM (pHo 5.5) in a single oocyte (Vh = −50 mV); evoked currents were normalized to the Fe2+-evoked current (−98.5 nA). c, An oocyte expressing DCT1 (superfused at pHo5.5) was held at Vh of −50 mV, and Vm was stepped to between +90 mV and −110 mV (in 20-mV increments) for 200 ms. Currents are displayed from 3 ms after the voltage step (on and off); for clarity, only the currents obtained at Vm of +70, +50, +30, −10, −50 and −90 mV are shown. d, Concentration-dependence of the Fe2+-evoked currents (pHo 5.5, Vh = −50 mV). The currents evoked by 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 25 and 50 µM FeCl2 were fitted to equation (1): ImaxFe= −105 ± 1 nA, K0.5Fe= 2.2 ± 0.1 µM, and nH = 1.2 ± 0.1 for iron. e, Voltage-dependence of ImaxFe (neither K0.5Fe nor nH varied with Vm). f, Simultaneous recordings of intracellular pH (pHi) changes and current in an oocyte expressing DCT1. The oocyte was clamped at Vh = −90 mV and superfused for the periods indicated by the boxes in the top panel, with pH 7.5 medium (blank), then with pH 5.5 medium (diagonal hatch). 50 µM FeCl2 was added for the period shown by the solid bar. g, Currents evoked by 10 µM Fe2+ as a function of extracellular pH (pHo). h, Proton activation of the currents evoked by 10 µM Fe2+ at Vh = −50 mV. Data were fitted to equation (1): ImaxH= −121 ± 21 nA (compare with ImaxFe), K0.5H= 1.3 ± 0.5 µM, nH = 1.1 ± 0.2 for H+. (ImaxH varied with Vm in a similar manner to ImaxFe.) All kinetic data presented (d, e, g, h) were obtained from the same oocyte; errors represent the error in the estimates of kinetic parameters (equation (1), see Methods).

Following step-changes in membrane potential (Vm) of oocytes expressing DCT1, we observed presteady-state and steady-state currents at pHo 5.5 (Fig. 5c). These comprised a fast capacitive transient (decaying with a time constant of 1 ms), which was also observed in control oocytes, and a transporter-mediated presteady-state current (half-time, τ= 30–70 ms), decaying to the steady-state value (Fig. 5c). DCT1-mediated presteady-state currents were abolished in the presence of divalent cations serving as substrates (Fig. 5b) or 10 mM Ca2+, and were H+-dependent, suggesting that H+ may bind to the transporter—our first indication that DCT1-mediated metal-ion transport might be H+-coupled. The presteady-state currents were similar to the charge movements reported for other H+-coupled transporters14,15,16, attributed to reorientation of a charged carrier or to H+ binding or dissociation in the plane of the membrane. DCT1-mediated presteady-state currents decayed more slowly than those of other cloned ion-coupled transporters (see Table 1 in ref. 15) except for the Na+/Cl/GABA transporter17; the Vm midpoint of charge transfer (V0.5) at pHo 5.5 was >+70 mV.

Steady-state Fe2+-evoked currents were saturable (Fig. 5d): the Fe2+ concentration at which current was half-maximal (K0.5Fe) was 2 µM and did not vary with Vm. The maximal Fe2+-evoked current (ImaxFe) showed a curvilinear dependence on Vm (Fig. 5e), approaching a zero-current asymptote at positive Vm. ImaxFe was approximately linear at hyperpolarized Vm and did not saturate with hyperpolarization within the Vm range tested. The kinetic characteristics of DCT1-mediated Fe2+ transport were confirmed in radiotracer studies (not shown). At pHo 6.2, 55 Fe uptake in NaCl medium was 88 ± 12 pmol over 1.5 h per oocyte (n = 8 oocytes), compared with 100 ± 10, 118 ± 12 and 118 ± 16 pmol per 1.5 h per oocyte respectively in NaNO3 (n = 9), NaSCN (n = 6), and choline chloride (n = 6). Therefore, DCT1-mediated transport does not additionally depend on extracellular Na+ or Cl.

We further investigated the possibility of H+-coupling of metal-ion transport by simultaneously monitoring intracellular pH (pHi) changes and Fe2+-evoked currents (Fig. 5f). In an oocyte superfused with pHo 7.5 medium at Vh = −90 mV, pHi was 7.35 and remained stable. The inward shift in baseline current after switching to pHo 5.5 (in the absence of metal ion substrate) was associated with a significant decrease in pHi that was not seen in control oocytes. Applying 50 µM Fe2+ at pHo 5.5 induced a much faster and larger intracellular acidification, together with an inward current that peaked at about −200 nA. This manoeuver had no effect in control oocytes. Thus DCT1-mediated Fe2+ transport is H+-coupled. Removal of Fe2+ halted the decline in pHi and IFe returned to baseline.

At any given Vm, the current evoked by 10 µM Fe2+ was smaller at higher pHo, so that the IFe/Vm relationship appeared to be shifted to the left with increasing pHo (Fig. 5g): that is, Fe2+ transport was driven at higher rates at hyperpolarized potentials and/or low pHo.IFe increased in magnitude as a hyperbolic function of [H+]o (Fig. 5h); current was half-maximal at 1.3 µM H+(K0.5H). The Hillcoefficient for H+(nH) was about 1 (as for iron; Fig. 5d) and did not vary appreciably from unity at each Vm tested (not shown).Hill analysis suggested a transport stoichiometry for DCT1 of 1 H+ : 1 Fe2+. At physiological membrane potentials (−90 to −30 mV), the apparent affinity constant for H+(K0.5H) was 1–2 µM. H+ concentrations of at least 1 µM (pH 6.0) are expected in the brush-border extracellular microenvironment18 in the small intestine, even after the bulk luminal pH has risen substantially.

What is the phsyiological significance of coupling of divalent-cation transport with H+? First, coupling to the movement of H+ down its electrochemical gradient will always ensure a concentrative uptake of divalent cations, even at trace amounts. Second, it is possible that cotransported H+ could result in a slight lowering of pH at the intracellular surface of the membrane, sufficient to maintain the solubilization of iron before its binding to intracellular proteins. In addition to operating as a H+/divalent cation cotransporter (symporter), DCT1 can apparently operate as a H+ uniporter: that is, a H+‘leak’ pathway can proceed uncoupled from divalent cation transport. This was demonstrated by the inward currents and intracellular acidification observed after switching pHo from 7.5 to 5.5 and is analogous to the Na+/glucose cotransporter SGLT1 in which the phlorizin-inhibitable leak current in the absence of sugar is carried by Na+(ref. 19). Similar observations have been made for other ion-coupled transporters, such as the plant H+/amino acid20 and H+/sucrose16 cotransporters, and a low-affinity homologue of SGLT1 (ref. 21). It is unlikely that these phenomena were due to H+-coupled transport of Ca2+ or Mg2+ (present at low concentrations in the medium) because, instead of increasing the inward currents, addition of excess (5 mM) Mg2+ had no effect, and excess Ca2+ (which partially inhibited the Fe2+ -evoked current) resulted in a small outward current, presumably as a result of blocking the H+ leak pathway. The effects of Ca2+ (which, like Fe2+, also abolished the presteady-state currents) were reminiscent of the effects of phlorizin on the Na+/glucose cotransporters19,21. Our data therefore suggest that Ca2+ is a blocker (or very weak agonist), reacting with DCT1 at very low affinity (Ki ≈ 10 mM).

Further kinetic measurements will distinguish between co-transporter models22, by showing for example which ligand (H+ or divalent cation) binds first and whether they are transported simultaneously or consecutively. The biophysical properties of the DCT1 transporter are strikingly similar to those of several H+- or Na+-coupled transporters (see Table 1 in ref. 15)14 that serve a wide range of substrates (including sugars, oligopeptides, amino acids), despite a lack of significant sequence homology.

Taken together, our data show that there is an active, cellular uptake mechanism for divalent cations (including Fe2+) in mammalian cells. DCT1 accepts a broad range of metal ions, favouring the divalent cations Fe2+, Cd2+, Co2+, Cu2+, Ni2+, Mn2+, Pb2+ and Zn2+ among those tested. The existence of a common intestinal absorptive mechanism for a range of metals has important nutritional implications. Absorption of these divalent cations will proceed even at physiological Ca2+ levels. However our results suggest that excessive luminal Ca2+ could interfere with the normal absorption of trace metals (see also ref. 23). DCT1 also transports the toxic heavy metals Cd2+ and Pb2+. We cannot at present exclude the involvement of DCT1 in receptor-mediated endocytic absorption of iron (endosomal pH of about 6.0), a possibility that could be tested using immunocytochemistry.

DCT1 was upregulated in response to dietary iron deficiency in most, if not all, tissues. DCT1 cDNA contains in its 3′ untranslated region a putative iron-responsive element which forms a stem–loop containing the consensus sequence (Fig. 1c) found in the 3′ and 5′ iron-responsive elements of the transferrin receptor and the ferritin mRNAs, respectively24,25. By analogy with transferrin receptor mRNA, the iron-responsive element may regulate DCT1 mRNA levels by RNA degradation mediated by a protein that binds to this element.

The marked responsiveness of DCT1 expression (at least at the mRNA level) to manipulations of dietary iron has implications for hereditary haemochromatosis (HH). This common autosomal recessive disease occurs primarily in individuals of northern European origin. The primary pathophysiological defect is in the small intestine where the absorption of iron in HH patients is increased, leading to progressive iron deposition in the liver, heart and other organs. Interestingly, HH patients also accumulate other metals, such as Co2+ and Mn2+(ref. 26). As this is consistent with the broad substrate range of DCT1, we propose that DCT1 is constitutively upregulated as an indirect result of a defect in the recently identified haemochromatosis gene HFE. A single point mutation in HFE (resulting in the amino-acid substitution Cys 282 → Tyr) is found in the majority of HH patients1. As the HFE gene product corresponds to a protein related to the major histocompatibility complex, with only a single membrane-spanning region and a short cytoplasmic tail, it is unlikely that HFE is a transporter. Instead, HFE may be involved in ‘sensing’ iron levels and regulating the expression of other gene products, including DCT1, thereby explaining how a defective HFE gene product might result in iron overload.

DCT1 is the most recently identified member of an emerging family (Nramp) of mammalian proteins (Fig. 2a). Among these, the mouse macrophage protein Nramp1 confers resistance to mycobacterial infection in mice, but the host defence mechanism and biochemical function of Nramp1 are unknown4. The amino-acid sequence of DCT1 is 73% identical to human Nramp1 and probably corresponds to the rat isoform of the ubiquitously expressed human Nramp2 (with which it shares 92% identity). Based on our functional characterization of DCT1 expressed in oocytes, we propose that the Nramp family of proteins are divalent-cation transporters. We found that 55Fe uptake was stimulated in oocytes expressing mouse Nramp1 (not shown), although the apparent affinity for Fe2+ was much lower than for DCT1. Functioning as a divalent-cation transporter, Nramp1 could confer resistance to infection by either of the following mechanisms. (1) H+-coupled uptake of Fe2+ in macrophages by Nramp1 may provide these cells with Fe2+ to produce toxic hydroxyl radicals via the Fenton reaction, killing pathogens in phagosomes as part of the defence mechanism. (2) Fe2+, Mn2+ or other divalent cations may be depleted from the phagosome by Nramp1 in the phagosomal membrane. This could result in a lack of essential metal ions which are required by the pathogen for the activity of enzymes such as superoxide dismutase and catalase in order to detoxify superoxide anions or hydroxyl radicals, thereby preventing propagation of the bacteria. A similar proposal27 has been made for the SMF1 transporter, a yeast homologue of the Nramp family.

Collectively, the data presented constitute a major advance in our understanding of a critical biological question: how divalent cations cross cell membranes, particularly those of epithelia. Moreover, the existence of a single transport mechanism serving a range of divalent cations and driven by the proton electrochemical gradient has nutritional, clinical and toxicological implications.


Expression cloning. DCT1 cDNA was cloned from rats fed on a low-iron diet (ICN) with milli-Q water ad libitum (3–4 weeks). Briefly, RNA was extracted from duodenal mucosal scrapes from iron-deficient (Fe(−)D) male rats and size-fractionated using preparative gel electrophoresis. A positive (4.5 kb) poly(A)+ RNA fraction, which stimulated 55Fe uptake in oocytes, was used to construct a directional cDNA library using a SuperScript cDNA synthesis system (GibcoBRL). cDNA was ligated into the pSPORT1 vector and ElectroMax DH10B cells were transformed as described28. RNA synthesized in vitro from pools of 200 clones was injected into oocytes. A positive pool was sequentially subdivided and analysed until we identified a single clone (DCT1) that was sequenced at the Yale University Sequencing Facility.

Radiotracer uptake. Collagenase-treated Xenopus oocytes were injected with 50 ng poly(A)+RNA or 25 mg DCT1 RNA synthesized in vitro. Radiotracer uptake was determined (3 d after RNA injection) by incubating 6–10 oocytes for 1 or 1.5 h in 750 µl standard uptake medium (100 mM NaCl, 10 mM HEPES, 1 mM ascorbic acid, pH 6.2) together with 10 µM 55FeCl2. Oocytes were solubilized using 10% SDS and the 55Fe content was measured by liquid scintillation counting.

Northern analysis. Poly(A)+RNA (3 µg each well) was separated on a 18% formaldehyde/1% agarose gel and blotted onto a nitrocellulose filter (Schleicher and Schuell). The DCT1 insert was excised from pSPORT1 and labelled with 32P using a T7 QuickPrime kit (Pharmacia). Hybridization was for 16 h at 42 °C in 50% formamide, and filters were washed in 5× SSC/0.1% SDS at 50 °C for 2× 30 min, 0.1 × SSC/0.1% at 65 °C for 3× 20 min.

In situ hybridization. Digoxigenin-labelled antisense and sense run-off transcripts were synthesized using a Genius Kit (Boehringer-Mannheim) from a PCR fragment, which contained about 1.7 kb of DCT1 sequence (bases 105–1,788) and which was flanked by promoter sites for SP6 and T7 RNA polymerase. Transcripts were alkali-hydrolysed to an average length of 200–400 base pairs. In situ hybridization was performed on cryosections (12 µm) of fresh-frozen tissue as described29. The hybridization buffer consisted of 50% formamide, 5× SSC, 2% blocking reagent (Boehringer-Mannheim), 0.02% SDS, 0.1% N-laurylsarcosine; probe concentrations were 200 ng ml−1. Sections in slide mailers were immersed in hybridization solution and hybridized at 70 °C for 18 h. Sections were then washed 3 times in 2× SSC and for 2× 30 min in 0.1 × SSC at 70 °C. The hybridized labelled probes were visualized using anti-digoxigenin Fab fragments (Boehringer-Mannheim) and BCIP/NBT substrate29. Sections were developed in substrate solution for 42 h, then rinsed in 10 mM Tris, 1 mM EDTA, pH 8.0, and coverslipped with 50% PBS/glycerol.

Electrophysiology. A two-microelectrode voltage clamp was used to measure steady-state and presteady-state currents in control oocytes and oocytes injected with 25 ng of in vitro transcribed DCT1 RNA, 4–7 days after injection. Microelectrodes (resistance 0.5–5 MΩ) were filled with 3 M KCl. Ooctyes were superfused at 23–24 °C in a low-calcium medium with ascorbic acid (96 mM NaCl, 2 mM KCl, 0.6 mM CaCl2, 1 mM MgCl2, 100 µM ascorbic acid, 2.5 mM HEPES, 2.5 mM MES, buffered to between 5.5 and 7.5 with Tris) and clamped at a holding potential (Vh) of −50 mV. The current was low-pass filtered at 5 kHz and digitized at 5 kHz. Step-changes in membrane potential (Vm) were applied (from +50 mV to −150 mV for steady-state studies, and +90 mV to −110 mV for presteady-state analysis, in 20-mV increments), each for a duration of 200 ms, at pHo 5.5–7.5 before and after the addition of the test substrate. Additionally (Fig. 5a and b), current was monitored continuously (sampling at 1 Hz) in oocytes clamped at −50 mV, but without step changes in Vm. Test solutions were washed out by superfusing the oocyte with substrate-free medium at pH 7.5 for several minutes. Steady-state data (obtained by averaging the points over the final 11 ms at each Vm) were fitted to equation (1), for which I is the evoked current (that is, the difference in steady-state current measured in the presence and absence of substrate), Imax the derived current maximum, S the substrate (divalent cation or H+) concentration, K0.5S the substrate concentration at which current was half-maximal, and nH the Hill coefficient

Intracellular pH measurement. Changes in intracellular pH (pHi) associated with DCT1-mediated transport in oocytes were measured using pH microelectrodes. We used silanized borosilicate pipettes back-filled with phosphate buffer at pH 7.0, with the tips filled with the hydrogen ionophore I-cocktail B (Fluka) as described30. pHi and current (filtered at 25 Hz) were measured simultaneously and digitized at 0.5 Hz in voltage-clamped oocytes, according to a method to be described elsewhere (M.F.R. and W.F.B., manuscript in preparation).