Functional characterization of mouse urea transporters UT-A2 and UT-A3 expressed in purified Xenopus laevis oocyte plasma membranes

Bryce MacIver, Craig P. Smith, Warren G. Hill, Mark L. Zeidel


Urea is a small solute synthesized by many terrestrial organisms as part of the catabolism of protein. In mammals it is transported across cellular membranes by specific urea transporter (UT) proteins that are the products of two separate, but closely related genes, referred to as UT-A and UT-B. Three major UT-A isoforms are found in the kidney, namely UT-A1, UT-A2, and UT-A3. UT-A2 is found in the thin, descending limb of the loop of Henle, whereas UT-A1 and UT-A3 are concentrated in the inner medullary collecting duct. UT-A2 and UT-A3 effectively represent two halves of the whole UT-A gene and, when joined together by 73 hydrophilic amino acids, constitute UT-A1. A biophysical characterization of mouse UT-A2 and UT-A3 was undertaken by expression in Xenopus laevis oocytes and subsequent preparation of highly enriched plasma membrane vesicles for use in stopped-flow fluorometry. Both isoforms were found to be highly specific for urea, and did not permeate water, ammonia, or other molecules closely related to urea (formamide, acetamide, methylurea, and dimethylurea). Single transporter flux rates of 46,000 ± 10,000 and 59,000 ± 15,000 (means ± SE) urea molecules/s/channel for UT-A2 and UT-A3, respectively, were obtained. Overall, the UT-A2 and UT-A3 isoforms appear to have identical functional kinetics.

  • urea analog
  • flux
  • water permeability
  • stopped flow
  • functional kinetics

in mammals, urea plays an essential role in the removal of excess nitrogen and maximal concentration of the urine. Compared with electrolytes, urea's relatively high permeability across specific biological membranes was thought to occur without regulation. However, early studies revealed rapid and inhibitable urea transport across erythrocyte (22) and mammalian collecting duct cell membranes (7, 17, 32). In addition, in collecting duct, urea permeability was found to be significantly increased by the antidiuretic hormone vasopressin (33, 44).

These observations ultimately led to the cloning of two separate genes from mammals (27, 48), now termed urea transporters (UT)-A (Slc14a2) and UT-B (Slc14a1), with a number of isoforms identified from a variety of mammalian species (31). UTs are also found in other species, including fish, frogs, and some pathogenic bacteria. Urea transport has also been described in plants (AtDUR3), although the transport mechanism here differs in that it is a H+ symporter (20).

Currently, six protein isoforms of the UT-A gene have been described, with UT-A1, UT-A2, and UT-A3 being the most abundant in rat and mouse kidney (2, 3). In mice, the UT-A gene comprises 24 exons with UT-A1 translating to the full-length protein of 929 amino acid (929aa) and is predicted to have four hydrophobic regions, each with multiple transmembrane domains (10, 37). In comparison, UT-A2 comprises the COOH-terminal half (397aa) of UT-A1 and has two predicted hydrophobic regions, while UT-A3 represents the amino-terminal half of 461aa and also has two predicted hydrophobic regions.

Of the less-abundant isoforms, UT-A4 comprises the amino terminal half of UT-A3 and the COOH-terminal half of UT-A2, producing a predicted protein of 466aa but is barely detectable in rat kidney (14). UT-A5 has so far been found only in mouse testes and is similar to UT-A3 but truncated by 138aa at the amino terminal end. UT-A6 is the smallest isoform described to date with expression in the colon (38). In addition to its role in the erythrocyte, UT-B has a wide tissue distribution, but in the kidney it is found in the descending vasa recta endothelium. It is believed that all UT proteins are built upon a 10-transmembrane segmented structure that is probably based on duplication of a five transmembrane segment early in the evolutionary history of these genes (24). It has been postulated that a duplication event from an ancestral gene gave rise to precursors of the current Slc14a1 (UT-B) and Slc14a2 (UT-A) genes, and that Slc14a2 has since undergone an internal duplication to effectively double the original size to 20 transmembrane segments (39). Comparing the amino acid sequences of UT-A2, UT-A3, and UT-B reveals a high degree of similarity; between 60–65%, with higher homology in the central putative transmembrane segments and more divergence at either end.

Functional studies have been carried out on UT-B by using whole erythrocytes as a convenient membrane system (e.g., 5, 19). However, the lack of such a simple system for UT-A has focused studies on the regulation and localization of several of the isoforms and their role in urine concentration when knocked out in mice. In mice, UT-A1 and UT-A3 are localized on the apical and basolateral membranes, respectively, of the inner medullary collecting duct (IMCD; 25, 40). UT-A2 is found in the thin descending limb of the loop of Henle (9).

Expression studies in Xenopus oocytes have shown both UT-A1 and -A3, but not UT-A2 to be stimulated by cAMP agonists (10, 36). However, in Madin-Darby canine kidney (MDCK) cells, all three isoforms respond to cAMP (11, 29, 41). Mouse knockout studies of UT-A1/-A3 and -A2 have been carried out with that of UT-A1/-A3 effectively being a double knockout because of the overlapping nature of these isoforms. In the UT-A1/-A3 knockout mouse, urine concentrating ability was reduced to ∼35% of wild-type mouse and urine flow increased (8). Results from the UT-A2 knockout mouse showed little effect under normal dietary conditions, but did show impairment of urine concentrating ability under low dietary protein and dehydration conditions, suggesting that the role of UT-A2 is in maintaining high urea concentrations in the inner medulla (42).

While the localization of the different UT-A isoforms has been defined, the functional significance of the different sequences has not. It is unclear whether the different isoforms have different or similar transport properties. This information will be critical in developing an understanding of how these proteins function and whether the different sequences have functional physiological significance.

Oocytes from Xenopus laevis have proved invaluable for expression studies of numerous transporter proteins because they permit functional studies of transport across the plasma membrane of the intact oocyte. However, in the intact oocyte, there are large unstirred layers and an inability to control precisely the internal chemical composition. These limitations make it difficult to make detailed functional measurements of transporters of water and small nonelectrolytes. In a previous paper (13), we overcame some of these limitations by developing a method for enriching the plasma membrane vesicles derived from whole oocytes. Here we use this methodology to examine the functional properties of two mouse UT-A variants in greater detail.

In this study we have compared the functional characteristics of the mouse UT-A2 and UT-A3 isoforms. We have initially focused on these two isoforms because they effectively split the UT-A gene into two halves, and they are expressed in different regions of the kidney. If the two forms of the protein exhibit similar function, then the differences in sequence between them are of limited significance for transport function. Thus, we are able to investigate the similarities or differences in the functioning of these two isoforms.

Our results show that both UT-A2 and UT-A3 function in a practically identical manner in terms of urea flux and permeability to a number of urea analogs.



Chemicals were purchased from Sigma (St. Louis, MO) and were of reagent grade, unless otherwise noted.

Harvesting oocytes.

All studies with animals were conducted in strict accordance with animal subjects regulations at the University of Pittsburgh and Beth Israel Deaconess Medical Center and were approved by the IACUC of those institutions. Xenopus laevis frogs (Xenopus Express, Plant City, FL) were anesthetized in 1 liter 0.5% (wt/vol) 3-aminobenzoic acid ethyl ester methanesulfonate salt (Tricaine) containing ice for 20 min. Oocytes were removed bilaterally from the abdominal cavity, and the egg mass was cut into small pieces and placed in calcium-free ND96 buffer (in mM; 96 NaCl, 1 KCl, 1 MgCl2 5 HEPES, pH 7.5). Oocytes were then defolliculated in 1.7% (wt/vol) collagenase, 0.17% (wt/vol) trypsin inhibitor in calcium-free ND96 for 55 min with rotation on an Adams Nutator before being washed three times with hypotonic buffer [in mM; 100 K2HPO4, 0.1% (wt/vol) BSA, pH 6.5], and then oocytes were allowed to incubate in hypotonic buffer for 10 min at room temperature. Oocytes were transferred to calcium-free ND96 and then to modified Barth's solution [in mM: 88 NaCl, 1 KCl, 2.4 NaHCO3, 0.82 MgSO4, 0.33 Ca(NO3)2, 0.41 CaCl2, 10 HEPES, pH 7.4] supplemented with 1% vol/vol penicillin/streptomycin (Invitrogen, Carlsbad, CA) and then maintained at 18°C.

Isolation of oocyte plasma membrane.

Plasma membranes were isolated as previously described (13) with additional slight modifications. The full method is described as follows. Between 300 and 600 oocytes were homogenized in oocyte homogenization buffer (OHB; in mM: 250 sucrose, 5 MgCl2, 10 HEPES pH 7.4; 10 μl/oocyte) containing protease inhibitors (Complete Mini; Roche Diagnostics, Mannheim, Germany) using a Maxima homogenizer (Fisher Scientific, Pittsburgh, PA) on setting seven (∼400 rpm) and 15 strokes of a Teflon-coated pestle in a glass dounce. When membranes were intended for permeability measurements, 10 mM 5,6-carboxyfluorescein (CF) was added to the OHB. For pH-sensitive measurements, the concentration of CF was 1 mM. Homogenates were centrifuged at 500 g for 5 min, and the supernatant was recovered. The second homogenization of the resuspended pellet as described in Hill et al. (13) was omitted, but the first supernatant was recentrifuged at 500 g for 5 min. The crude membranes were then overlaid on top of a discontinuous sucrose gradient comprised of 50% sucrose/OHB (6 ml) and 20% sucrose/OHB (6 ml) and centrifuged at 30,000 g for 1 h in a swinging bucket rotor. Two major membrane bands were visible: the lower at the interface between 20% and 50% sucrose (designated heavy membranes) and the other ∼1 cm from the top (designated light membranes). Lipid floating on the top of the gradient was removed with a tap aspirator and Pasteur pipette and then the top band was collected, diluted 20-fold in OHB, and total membranes pelleted at 100,000 g for 30 min. This pellet was resuspended in ∼20 ml OHB, and this wash step was repeated. The final pellet was resuspended in 0.4–0.6 ml OHB for use in stopped-flow experiments. This procedure has been shown to yield a unilamellar, uniformly sized population of vesicles that range between 200 and 400 nm in diameter.

Permeability assays.

Permeabilities of membrane vesicles were measured with stopped-flow fluorometry as described in (12, 30). Water permeability was measured from the rate of shrinkage after exposing membrane vesicles to a hyperosmotic solution with double the osmolality. Urea and urea analog transport was measured in membrane vesicles preequilibrated for 20 to 30 min in OHB containing 1 M urea or analog. Vesicles were then rapidly mixed at a 1:1 ratio with an osmotically balanced solution of OHB plus additional sucrose, exposing the vesicles to 500 mM gradient of urea or analog. Urea or analog efflux in response to the chemical gradient leads to vesicle shrinkage. In some experiments, the UT was inhibited by preincubation with 0.5 mM phloretin for 30 min. Permeability coefficients were calculated using Mathcad (Mathsoft, MA) software as previously described (30).

Ammonia flux assays.

Permeabilities of ammonia was measured as previously described (30). Briefly, plasma membrane preparations were made from oocytes as above, but each set was split in two with one set prepared at 10 mM CF concentration to allow for assaying of urea permeability and the second at 1 mM CF for ammonia. CF at 1 mM is used to allow for measurements based on the pH sensitivity of the dye. For ammonia permeability measurements plasma membrane vesicles were equilibrated in OHB buffer at pH 6.8 for >30 min. Stopped-flow experiments were performed by abruptly diluting the vesicles into OHB titrated to pH 6.8 and supplemented with 40 mM NH4Cl. Ammonia flux across the membrane results in an increase in pH within the vesicle, which is reported by increased CF fluorescence.

Preparation of cDNA and cRNA.

To enable relative and absolute quantitation a c-Myc tag was incorporated by PCR at the 3′ end of both UT-A2 and UT-A3. Antisense primers were designed to incorporate the c-Myc tag in frame with the UT protein followed by a stop codon and then a SpeI restriction site. Sense primers were designed to anneal to a section of nucleotide sequence 5′ to a unique restriction site (UT-A2, SphI; UT-A3, NheI) within the coding region of each gene, giving 0.6 kb fragments. These fragments were initially cloned into the pCR4 TOPO cloning vector (Invitrogen), verified by sequencing, and then subcloned into the respective Xenopus expression vector pT7TS that carried untagged versions of the UTs using the restriction sites SphI-SpeI (UT-A2) and NheI-SpeI (UT-A3). Sequencing of both tagged UTs in the pT7TS vector was carried out to confirm integrity. Constructs were linearized with XbaI for cRNA synthesis. Capped RNA was prepared with a T7 mMessage mMachine kit (Ambion, Austin, TX). The primer sequences used were: UT-A2 sense; 5′-CACAATCTCTTCTTCCCC-3′ UT-A2 antisense (with c-myc tag); 5′-ACTAGTTTACAGATCCTCTTCTGAGATGAGTTTTTGTTCGGAGACGTCGTAGGC-3′, UT-A3 sense; 5′-CTACAACCTTTTCTTCCCC, UT-A3 antisense (with c-myc tag); 5′-ACTAGTCTACAGATCCTCTTCTGAGATGAGTTTTTGTTCGTCACCATTGGGGGA-3′.


Membrane samples were denatured at 65°C for 10 min in deglycosylating enzyme PNGaseF (New England Biolabs, Danvers, MA), treated with 125 units PNGaseF by following the manufacturer's instructions. Samples were denatured at 65°C for 15 min, 4× sample buffer and run on SDS-PAGE using 8–16% precast Long-Life gels (NuSep, Austell, GA) and then electrotransferred to Immun-Blot polyvinylidene difluoride membrane (Bio-Rad Laboratories, Hercules, CA) at 40 volts for 90 min. The membrane was blocked overnight in a Tris-buffered saline + 0.1% vol/vol Tween 20 solution containing 5% skim milk powder. Monoclonal antibodies to the c-Myc epitope 9E10 (Santa Cruz Biotechnology, Santa Cruz, CA) were incubated at 1:500 for a minimum of 3 h at room temperature, washed, followed by 1:5,000 horseradish peroxidase-coupled anti-mouse secondary antibody (GE Biosciences, Piscataway, NJ), and detected by enhanced chemiluminescence (GE Biosciences). Blots were exposed to standard X-ray film for various times. The film was scanned at 600 dpi resolution with the resulting image saved in TIFF format. Bands were quantitated using ImageQuant software (GE Biosciences). Total protein concentration was measured using the Coomassie Plus assay (Pierce, Rockford, IL) with a standard curve derived from bovine serum albumin and spectrophotometric measurements made on a Biophotometer (Eppendorf, Westbury, NY). For mouse UT-A2 antibody ML194 and mouse UT-A3 antibody ML446 an anti-rabbit horseradish peroxidase-coupled secondary antibody was used at 1:5,000 (GE Biosciences).

Determination of membrane vesicle diameters.

Size distributions were determined by quasielastic light scattering using a DynaPro LSR particle sizer and Dynamics data collection and analysis software (Protein Solutions, Bucks, UK).

Isotopic uptake assays.

UT activity in oocytes was assayed by [14C]urea (56 mCi/mmol; American Radiolabeled Chemicals, St. Louis, MO) uptake using 10 mM cold urea and 0.1 mM [14C]urea added to modified Barth's solution buffer. Oocytes (from 4–12) were incubated in 1 ml of solution in 1.5 ml microcentrifuge tubes for the required time period at room temperature (23°C). At the end of the incubation period, the isotope solution was removed and the oocytes were washed with five quick exchanges of ice-cold modified Barth's solution + 10 mM urea. Oocytes were then individually distributed to scintillation vials and dissolved in 10% wt/vol SDS, and then 3 ml ScintiSafe (Fisher, Pittsburgh, PA) was added. Counting was performed on a Packard model 1,500 liquid scintillation counter (Perkin Elmer, CA).

Calculation of single-channel flux rates.

The urea flux of a single channel was calculated based on Eq. 1 Math(1) where Jurea is the rate of uptake in mol/s, SA is the surface area of a vesicle in cm2, and [urea] is the concentration in mol/cm3. The urea permeability coefficient (Purea) was obtained from stopped-flow experiments corrected by subtracting control values.

To calculate the surface area of vesicles in a preparation two parameters were required: 1) the surface area (and volume) of a single vesicle, which was readily determined from the diameter measured by quasielastic light scattering (see Determination of membrane vesicle diameters), and 2) the total number of vesicles in a given volume of a preparation. The latter parameter was measured by taking advantage of the entrapped CF and assuming the vesicles captured the full initial concentration of CF. A standard curve of CF fluorescence vs. CF concentration was measured using an Aminco Bowman Series 2 luminescence spectrophotometer, using excitation and emission wavelengths of 490 nm and 520 nm, respectively. The fluorescence of vesicles was measured after lysis with Triton X-100 (final concentration 0.5% vol/vol). From these measurements, the concentration of CF entrapped in the vesicles could be calculated, and, knowing the original CF concentration (10 mM), the volume of vesicles present could be calculated using the formula Math(2) where V1 is the volume of the vesicles, V2 is the volume occupied by the released CF (usually 2.05 ml), C2 is the concentration of CF from the standard curve after lysis, and C1 is the starting concentration of CF in the vesicles (8.33 mM to allow for dilution when the CF/OHB solution is added to a volume of oocytes at the homogenization step). The total number of vesicles was multiplied by the surface area of a single vesicle to provide the SA value for Eq. 1. The concentration of urea was 0.5 mol/1,000 cm3. The amount of UT present in a plasma membrane vesicle preparation was estimated from Western blot analysis (see above) using recombinant c-Myc protein (Actif Motif, Carlsbad, CA) as a standard. The single-channel flux was calculated by dividing Jurea from Eq. 1 with moles UT calculated from the Western blot quantitation experiments (Mr UT-A2-myc = 44,344, UT-A3-myc = 51,846).

Statistical tests.

Standard two-tailed student's t-tests were performed on permeability data using Microsoft Excel (Microsoft, Redmond, WA).


Expression of UTs results in higher urea fluxes.

Unless otherwise stated, all experiments were conducted with the c-Myc epitope tagged version of UT-A2 and UT-A3. Both UT-A2 and UT-A3 were shown to transport urea (Fig. 1A). Initial experiments were performed to ascertain optimal expression conditions for these transporters in Xenopus oocytes. We tested differing amounts of cRNA injected per oocyte and the time of expression from 2–4 days. There was a modest difference between 2 and 10 ng of cRNA and no further improvement in expression using 30 ng of cRNA (results not shown). While we observed higher levels of expression with increasing time, experiments were performed at 3 days postinjection as a compromise between maximal expression and loss of oocytes (plasma membrane preparations require a minimum of 300 surviving oocytes in good condition to obtain sufficient material). We next confirmed that the c-Myc tag added to UT-A3 does not change the transport activity of the protein using [14C]urea uptake assays (Fig. 1B), while Potter et al. (29) report a similar result for c-Myc-tagged UT-A2 in MDCK cells. Additionally, Stewart et al. (41) recently have shown the same result for UT-A3 in MDCK cells.

Fig. 1.

Isotopic urea transport measurements in control oocytes and oocytes expressing urea transporter (UT)-A2 and UT-A3. A: time course [14C]urea uptake experiment in whole oocytes (8–12 per experiment) with c-Myc-tagged UT-A2 and UT-A3. B: similar experiment comparing UT-A3 with and without c-Myc tag. Data from a single experiment are shown, although each experiment was repeated 3 times. The results have not been averaged because the time points varied between experiments. Error bars represent means ± SE.

To measure quantitatively the permeabilities of urea and its analogs with a view to deriving single-channel urea fluxes, we utilized enriched plasma membrane vesicles from UT-expressing oocytes and measured fluxes using stopped-flow fluorometry. To measure fluxes of small nonelectrolytes, we preloaded vesicles with the solute of interest and then monitored the time course of volume reduction (using CF self-quenching) following abrupt dilution of the vesicles into an iso-osmotic solution in which 50% of the permeant solute was replaced with an impermeant solute. As the permeant solute inside the vesicles effluxes down its concentration gradient, the efflux causes an outwardly directed osmotic gradient. The resulting efflux of water causes volume reduction, which is reflected as a simple, unidirectional exponential curve. Representative traces typical of all experiments are shown in Fig. 2, A and B, where it can be seen that vesicles with UTs shrink at a much faster rate, indicating more rapid efflux of preloaded urea. In all experiments where a UT was present, we observe a curve that requires a double exponential to create a good fit to the data. The majority of control curves (either water injected or uninjected oocytes) also require fitting double exponentials, which is unexpected, and we comment on this in discussion. Purea coefficients were calculated for these experiments based on the first and fastest rate. We obtained permeability coefficients of 3.73 ± 0.30 × 10−6 cm/s (mean ± SE, n = 9) for UT-A2 and 3.69 ± 0.54 × 10−6 cm/s (mean ± SE, n = 9) for UT-A3, which are threefold higher than the coefficient obtained for controls; 1.21 ± 0.12 × 10−6 cm/s (mean ± SE, n = 18) (Fig. 2C). The permeability coefficients between UT-A2 and UT-A3 are not statistically different.

Fig. 2.

Urea transport in plasma membrane vesicles isolated from Xenopus oocytes: A and B: representative stopped-flow traces from a single experiment for mouse UT-A2 (A) and UT-A3 (B). The stopped-flow apparatus was loaded with 50-μl vesicles (protein per microliter vesicle ranged from 0.05–0.8 μg across all experiments) that had been diluted into 1 ml/1 M urea in oocyte homogenization buffer (OHB) and allowed to equilibrate for 30 min. Note that these traces represent the shrinkage rate of the vesicles present and are independent of the amount of material loaded, which only affects the voltage level required to obtain a measurable signal. Data are normalized to represent a 50% reduction in volume. Each panel shows both raw data points (averaged from 6–8 runs) and fitted curves. Double exponential curves were fit to both control and UT traces. C: summarizes all stopped-flow-based urea permeability experiments (mean ± SE, n = 9 each for UTs and 15 for controls). The results show a significant difference when an UT is expressed over controls (P < 0.0001 for both UT-A2 and UT-A3). The difference between UT-A2 and UT-A3 is not significantly different (P = 0.195). Purea, urea permeability coefficients.

In several experiments, we used phloretin, a known inhibitor of UTs, and observed reductions in Purea ranging from 25% to 64% (Fig. 3). The average reduction was 42% for UT-A3 (n = 4) and 50% for UT-A2 (n = 4). Other studies have generally shown a higher degree of phloretin inhibition in intact oocytes. The lower levels of inhibition observed in this study suggest that not all UT molecules are accessible to phloretin, possibly because the vesicles may be a mixture of inside-out and outside-out.

Fig. 3.

Inhibition of urea transport by phloretin (Phl). A representative stopped-flow tracing shows reduction of urea efflux in response to phloretin treatment. The tracing is truncated to the initial 20% of the total time course to clearly reveal the difference. Each tracing has been normalized to a 50% relative volume change. Inset shows averaged Purea coefficients value for several experiments where phloretin was tested. Reductions in Purea coefficients averaged 50% for UT-A2 (n = 4) and 42% for UT-A3 (n = 3).

Permeability to water.

We have tested whether UT-A2 or UT-A3 transports water as has been shown for UT-B (46), although there is some controversy over this result (1, 21). This was done by exposing the UT and control vesicles to an osmotic gradient of 550 mosmol/kgH2O in stopped-flow experiments. We did not observe any significant difference between control and UT vesicles in water permeability (Table 1). Permeability values were, respectively, 2.85 ± 0.26 × 10−4 cm/s for UT-A2 and 4.38 ± 0.41 × 10−4 cm/s for UT-A3, compared with 4.34 ± 0.40 × 10−4 cm/s (all means ± SE) for all controls, slightly lower than values previously reported using this methodology (8.07 × 10−4 cm/s) (13), but reaffirming that the Xenopus oocyte plasma membrane has barrier membrane characteristics. The UT-A2 water permeability coefficient results appear almost significantly lower (P = 0.068) than control (or UT-A3) results. We attribute this to variation in oocytes used from experiment to experiment. The value for controls presented is an average of all experiments, whereas those controls paired with the respective UT-A2 experiments were more comparable (3.23 ± 0.56 × 10−4 cm/s, mean ± SE) to the UT-A2 water permeability coefficient result (2.85 ± 0.26 × 10−4 cm/s, mean ± SE).

View this table:
Table 1.

Water permeability coefficients for plasma membrane vesicles expressing UT-A2 and UT-A3

Permeability of urea analogs.

To explore the selectivity of the UT-A isoforms to urea we identified a number of molecules structurally related to urea to test in the plasma membrane vesicle preparations. The urea analogs were tested using stopped-flow fluorometry in a similar manner to urea transport; that is, the vesicles were preloaded by incubation in 1 M solution of the analog of interest in OHB buffer and then rapidly mixed with an iso-osmotic solution balanced with sucrose. Analogs that could be tested were formamide, acetamide, methylurea, and 1,3-dimethylurea. Additionally, methylamine and ethylamine (pKa: 10.62 and 10.67, respectively) were also selected, but found to be mostly charged at the pH used for these experiments (pH 7.4) and thus would not load into vesicles to necessary levels. The results are summarized in Table 2 where we show that there is no increase in permeability of these analogs when a UT is present in the plasma membrane. These data indicate that the UTs are highly specific for urea to the exclusion of molecules that are equally small and with some similarity in structure (formamide).

View this table:
Table 2.

Permeability coefficients for formamide, acetamide, methylurea, dimethylurea, and urea for plasma membrane vesicles expressing UT-A2 and UT-A3 compared with control (water injected or uninjected)

In considering the passive permeability of urea and the analogs through oocyte plasma membrane (for example, compare controls in Table 2), it should be recognized that many factors, such as molecular volume, dipole moment, and partition coefficient, contribute to the permeability coefficient of a molecular species and that a direct correlation between permeability coefficient and molecular weight is not seen. This phenomenon has been observed in liposomes of varying compositions where acetamide (Mr 59) permeability always exceeded that of urea (Mr 60) (18). We also tested thiourea, which is widely reported to be an inhibitor of UT activity, but it would not preload into vesicles. This observation was also made for vesicles with expressed UT from Actinobacillus pleuropneumoniae (G. Godara and J. Mathai, personal communication).

Permeability to ammonia.

Ammonium is a vital part of acid-base regulation in mammals, and its conjugate base ammonia can be toxic to many tissues at low concentrations. Transport of ammonium is facilitated by Rb proteins and potassium channels in the kidney (15), but aquaporin 8 (AQP8) has also shown ability to transport ammonia (34). Hence, we investigated the possibility of ammonia transport through these UT isoforms with results shown in Fig. 4. The batch of injected oocytes was split into two samples, with one being homogenized in 1 mM CF for the ammonia measurements and the second in 10 mM CF for urea permeability measurements. Although both UTs markedly increased urea fluxes, they did not increase the rate of ammonia flux.

Fig. 4.

Ammonia fluxes across plasma membrane vesicles from control oocytes and UT-expressing oocytes. Plasma membrane vesicles were equilibrated at pH 6.8 and then rapidly mixed with the same solution containing 40 mM NH4Cl. Ammonia traversing the membrane (or transporter) into the vesicle protonates to NH4+ resulted in an increase in pH that is measured as a rise in pH sensitive CF fluorescence. AC: typical stopped-flow traces for UT-A2, UT-A3, and uninjected control oocytes, respectively. D: averaged results from a series of stopped-flow experiments. There is no significant difference between control values and those for either UT (two-tailed t-test: P > 0.6, n = 4–7). Results are presented as means ± SE.

Calculation of flux rates across single UTs.

The addition of a c-Myc epitope tag to the COOH-terminal ends of each mouse UT allows us to quantify the amount of protein present in plasma membrane preparations by quantitative Western blot analysis. Recombinant c-Myc protein (Actif Motif) was used to generate a standard curve on each blot, thus allowing the amount of c-Myc-tagged UT in membrane vesicles to be calculated (Fig. 5C). We conducted several trial experiments with differing denaturing temperatures (room temperature, 42°C, 65°C, and boiling) to optimize signal intensities on Western blots and found 65°C optimal. Additionally, we treated the plasma membrane samples with an N-glycosidase F enzyme (PNGaseF; New England Biolabs) to reduce the multiple and variable banding pattern observed, presumably due to glycosylation, to a single major band (Fig. 5A). Each sample was run several times to ensure reproducibility. X-ray films were exposed for a variety of times to prevent saturation and were scanned to create a TIFF digital image file. This file was then imported into the ImageQuant (GE Biosciences) image analysis software package, and bands were quantified using autotracing and manual tracing methods. All signals detected are due to the c-Myc tag incorporated into the UT protein as blotting of several different control plasma membrane preparations, resulted in no detectable signal. (Fig. 5D).

Fig. 5.

Western blot analysis for estimation of single UT flux rates. Examples of c-Myc-tagged UTs and recombinant c-Myc protein used for quantitation from Western blots. A: treatment of plasma membrane vesicle samples with the deglycosylating enzyme PNGaseF (UT-A2, lane 1, and UT-A3, lane 2) results in a single major band compared with the same samples untreated, respectively lanes 3–4 (UT-A2; 6.7 μg total protein loaded, UT-A3; 15.5 μg total protein loaded). B: two independent plasma membrane vesicle samples of UT-A2 (5 and 12 μg total protein) and UT-A3 (5 μg and 7 μg total protein) shows that the antibody to c-Myc identifies the same band on the Western blot as do anti-mouse UT-A2 antibody ML194 and anti-mouse UT-A3 antibody ML446. C: representative blot with recombinant c-Myc (lanes 13, 50, 20, and 10 ng, respectively), and PNGaseF-treated independent samples of UT-A3-myc (lanes 46, respectively: 7, 7, and 17.4 μg total protein loaded) or UT-A2-myc (lanes 79, respectively: 12.3, 12.8, and 20.2 μg total protein loaded). D: control vesicles produce no signal (lane 2) compared with a UT-A3 vesicle sample (lane 1, not PNGaseF treated). We note that the apparent molecular weight of both UT proteins on the blot are lower than predicted with UT-A2-myc ∼34 kDa (predicted 44.3 kDa) and UT-A3-myc ∼39 kDa (predicted 51.8 kDa) probably because of the denaturing conditions.

The flux of urea in moles per second was calculated using Eq. 1 (materials and methods), and from this information we calculated a single-channel urea flux of 46,000 ± 10,000 (mean ± SE) urea molecules·s−1·channel−1 for mUT-A2 (n = 4 independent samples) and 59,000 ± 15,000 (mean ± SE) for mUT-A3 (n = 4 independent samples). These values are not statistically different based on a two-tailed t-test (P = 0.66), indicating that both UTs have similar transport capabilities.


The UT-A transporter has been reported to have saturability >200 mM urea (35). We performed stopped-flow urea permeability experiments with final urea concentrations between 50 mM and 1 M on plasma membrane preparations from UT-A2, but could not generate a clear saturation curve (results not shown).


We purified plasma membrane vesicles from Xenopus oocytes expressing UT-A2 and UT-A3 to define the functional characteristics of these transporter isoforms. We chose these two UT-A isoforms because they represent the two halves of the full-length protein and are expressed at different sites within the kidney. We found that both proteins function in a very similar manner in their ability to transport urea and in their ability to exclude water and several urea analogs. We also found the neither UT isoform transports ammonia.

Under ideal conditions (e.g., a transporter protein reconstituted in liposomes), we would expect to see single exponential curve fits for both control and transporter-containing vesicles with the latter giving a significantly greater rate constant over the former when specific transport occurs. Indeed, stopped-flow traces for water and urea analog experiments were always readily fit with a single exponential curve; however, we found it necessary to fit double exponential curves to all UT-derived curves and the majority of control curves. There are two possibilities to explain these observations. First, our procedure is an enrichment for plasma membrane vesicles from a complex biological source [see Hill et al. (13) for a detailed characterization of the methodology], and, while the vesicles obtained fall within a uniform size distribution, a portion of these may not be plasma membrane derived and thus contain no functional UT. This class of vesicles would contribute to a slower rate of urea efflux. Second, we speculate that there may be a low-level endogenous urea transport activity in Xenopus oocytes. An aquaporin-like molecule, AQPxlo, with urea transporting capabilities has been identified in the Xenopus oocyte (43), although we must stress that the endogenous levels of this protein are unknown, and the characterization was accomplished by overexpression in the oocyte. A third possibility is the high urea concentrations used in these studies, which were designed to mimic conditions in the renal medulla, the natural location of these transporters, might have enhanced the relative activity of nonspecific transport processes by oocyte proteins.

The difference in urea permeabilities between control and UT-expressing oocyte plasma membranes is somewhat lower than has been observed in studies measuring isotopic uptake in intact oocytes, where differences of 10- to 20-fold have been typically reported. However, our results represent an overall average from many experiments, and we have observed quite a degree of variation from experiment to experiment. The largest difference between control and UT expressing vesicles within a particular experiment is ninefold for UT-A2 and 6.9-fold for UT-A3. The highest permeability calculated in an individual experiment was 4.8 × 10−6 cm/s for UT-A2 and 7.9 × 10−6 cm/s for UT- A3; the lowest were 1.81 × 10−6 cm/s and 0.78 × 10−6 cm/s, respectively.

We have used a number of urea analogs that, while not physiological, give insight into the selectivity of the UTs. Perhaps the most notable is formamide, which can be viewed structurally as urea lacking an amino group and is not transported by either UT. This result would suggest that the selectivity of the UTs for urea requires the presence of both amino groups. The urea analogs chosen here represent those that are small and structurally most resemble urea and that could passively diffuse through the plasma membranes to allow analysis by stopped-flow fluorometry. Final concentrations of analogs in stopped-flow experiments are 500 mM, which may, in fact, be inhibitory to the transporter. Dimethylurea has been reported as inhibitory to UT function with a Ki of 4.9 mM (11). Additionally, acetamide has been reported to reduce urea flux in perfused IMCD by 35% (6); however, in another study using oocytes expressing UT-A2, it did not inhibit urea uptake at a concentration of 150 mM (48). Although we have not fully addressed the inhibitory nature of the urea analogs, our data clearly demonstrate that none of the analogs passes through the transporters.

Thiourea has been used extensively in erythrocyte UT investigations as a competitor and inhibitor of urea transport mechanisms (e.g., see Refs. 4, 23), so it was of interest that we were unable to preload the plasma membrane vesicles with this compound and thus make measurements using the stopped-flow system. The inability of thiourea to load either by passive diffusion through the plasma membrane or through the UTs themselves is suggestive of differing properties between UT-A and UT-B transporter function. It is also possible that thiourea enters the erythrocyte through a different mechanism. Moreover, thiourea cannot load into liposomes prepared from pure phospholipids (J. Mathai, personal communication). A possible explanation for the failure of vesicles to load is that the molecule is charged at the pH of our experiments [pKa of thiourea of 1.19 (28)].

We tested the ability of both UTs to transport ammonia, a compound that is toxic to mammals at low concentrations, but which is important for renal pH regulation in the form of the ammonium ion. Transport of these compounds is facilitated by Rh proteins and potassium channels (15). However, other molecules, such as AQP8, have been reported to have ammonia-conducting properties with recent data showing AQP8 to have about twofold higher ammonia flux over water (34), although the physiological relevance of this has yet to be determined (47). In this study, we found that neither UT shows ammonia transport ability.

A value of 0.3–1 × 105 urea molecules·s−1·transporter−1 has been reported in the literature for UT-1 in perfused rat IMCD using an ELISA-based method to quantify the transporter (16). UT-1, now called UT-A1, is the full-length protein and effectively represents both UT-A2 and UT-A3, and thus, the single-channel urea flux of UT-A1 could be expected to be twice that of UT-A2 or UT-A3. It should also be noted that this study did not take into account the presence of UT-A3 in the IMCD, as this isoform had not been recognized as being present in this tissue at the time of the study (16). Our values of 46,000 ± 10,000 and 59,000 ± 15,000 (means ± SE) urea molecules·s−1·channel−1 for UT-A2 and UT-A3, respectively, agree well with the data of Kishore et al. (16). Our single-channel flux calculation assumes that all c-myc-tagged UT protein detected is functional and present in plasma membrane-derived vesicles. This assumption is likely optimistic, but our technique cannot determine the relative levels of functional UT to nonfunctional UT, and therefore, the single-channel flux calculations are underestimated. Single-channel urea fluxes have been calculated for UT-B (39) at 5 × 106 urea molecules·s−1·UT-B molecule−1, which is about 100-fold higher than our UT-A data. As noted by others, these values are higher by several orders of magnitude than the flux of the glucose transporters GLUT1 and GLUT4 at 102-103 glucose molecules·s−1·transporter−1 (26).

UT-B has also been shown to contribute to water transport in the erythrocyte (46), although there is some debate regarding this observation (1; for a review, see Ref. 21). The results of Yang and Verkman (46) using AQP1 and UT-B single and double knockouts show that, when the differing levels of protein expression between AQP1 and UT-B are taken into account, the single-channel water flux of UT-B is significant. Our results clearly show that neither UT-A2 or UT-A3 transport water.

Our data shows that the two UT-A isoforms have very similar, if not identical, functional properties. There is considerable sequence similarity (60–65%) between both UT-A isoforms and UT-B. However, the differences between the two UT-A isoforms do not appear to confer functional significance, whereas those between UT-A and UT-B isoforms do, based on higher single transporter urea flux and water permeability data for UT-B. Compared with the GLUT transporters we obtained a significantly higher single transporter flux. Furthermore, we were unable to generate saturation kinetics from our plasma membrane preparations. These data firmly support a channel-like mode of transport in these proteins. These differences will be better understood when the structures of these molecules are elucidated and compared.

This study examined mouse UT-A2 and UT-A3 transporters. Mice concentrate their urine to much higher levels than humans or rats: ∼2,650 mosmol/kgH2O for mice compared with ∼650 mosmol/kgH2O for humans and ∼1,500 mosmol/kgH2O for rats (45). This higher concentrating ability in mice is in part due to anatomical differences in the kidney. At present it is unknown whether the mouse UT-A isoforms have greater functional capability than those in mammals that do not concentrate their urine as much. Our method should enable a quantitative comparison of UTs from different species to address this issue.

In summary, we have utilized membrane vesicles from Xenopus oocytes to functionally characterize two mouse UT-A gene isoforms. These two isoforms effectively represent each half of the total gene product, and we cannot discern any significant differences in the biophysical properties of these isoforms. From our results, it appears that sequence differences in the splice variants of UT-As in the kidney do not result in differences in function as UTs. However, it is possible that these sequence differences relate to successful trafficking of the transporters to specific sites in specific cell types. We plan to utilize and improve the membrane vesicle preparation method to study the biophysical properties of other UT-A isoforms and the mouse UT-B.


This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-43955 and DK-48217 (to M. L. Zeidel).


We thank Nicole Southern and Jia Yin for technical assistance with these experiments.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract