Cloning and Functional Characterization of a Novel Aquaporin fromXenopus laevis Oocytes*

We have cloned a novel aquaporin (AQP) fromXenopus laevis oocytes, which we have provisionally named AQPxlo. The predicted protein showed highest homology (39–50%) to aquaglyceroporins. Northern blot analysis showed strong hybridization to an ∼1.4-kb transcript in X. laevis fat body and oocytes, whereas a weaker signal was obtained in kidney. We injected in vitro transcribed cRNA encoding AQPxlo into Xenopus oocytes for functional characterization. AQPxlo expression increased osmotic water permeability (P f ), as well as the uptake of glycerol and urea. However, AQPxlo excluded larger polyols and thiourea. An alkaline extracellular pH (pH o ) increasedP f and to a lesser extent urea uptake but not glycerol uptake. Remarkably, low HgCl2 concentrations (0.3–10 μm) reduced P f and urea uptake, whereas high concentrations (300–1000 μm) reversed the inhibition. We propose that AQPxlo is a new AQP paralogue unknown in mammals.

Inasmuch as the water permeability of the native oocyte membrane is low, it was long thought that oocytes possess few endogenous water channels (i.e. aquaporins or AQPs). 1 Recently, Schreiber et al. (2) cloned an AQP3 orthologue (xAQP3) from the A6 X. laevis cell line (derived from kidney). They showed that, in oocytes expressing the cystic fibrosis transmembrane conductance regulator (CFTR), stimulation of CFTR activity with 3-isobutyl-1-methylxanthine (IBMX) increased the water and glycerol permeability of the oocyte membrane. They concluded that CFTR regulates the activity of endogenously expressed xAQP3, because IBMX failed to stimulate glycerol permeability in oocytes injected with antisense oligonucleotides complimentary to xAQP3. However, in the absence of exogenous CFTR expression, xAQP3 activity in the oocytes remained undetectable.
We report here the cloning of a novel member of the AQP family from X. laevis oocytes. In addition, we have studied the permeability properties of this novel clone by overexpressing it in its native host, the oocyte.

Cloning of AQPxlo
By searching the GenBank TM data base for sequences with homology to known AQPs, we identified two novel EST sequences (GenBank TM accession numbers AW200125 and AW199121) from X. laevis oocytes. To obtain the full-length coding region, we performed PCR on an X. laevis oocyte library (X. laevis oocyte 5Ј-STRETCH cDNA TriplEx library, Clontech Laboratories, Inc., Palo Alto, CA) using primers specific to the coding sequence and to the vector sequences flanking the 5Ј and 3Ј ends of the insert. The resulting PCR products were subcloned into a pCR 2.1 TOPO-vector (Invitrogen) and sequenced. We performed end-to-end PCR with primers designed to the 5Ј and 3Ј regions flanking the open reading frame to obtain a continuous cDNA fragment containing the entire coding region. The sense primer was 5Ј-ACTCGAGG-CCGCGTCGACTGG and the antisense primer was 5Ј-TACTAGTGCC-AATCAAGAAGCAGATTCCC, where the underlined sequence corresponds to engineered XhoI (sense) and SpeI (antisense) restriction sites. The resulting PCR product was cloned into pCR 2.1 TOPO-vector and sequenced. All sequencing was performed by the Keck Biotechnology Resource Laboratory (Boyer Center for Molecular Medicine, Yale University). The cDNA sequence was deposited into the GenBank TM (accession number AY120934). We have provisionally named the clone AQPxlo (for Xenopus laevis oocyte).
vector and sequenced. The cDNA (GenBank TM accession number AY120935) contained an 810-bp open reading frame encoding 269 amino acids. The deduced amino acid sequence was 97% identical to the published rat AQP7 sequence (GenBank TM accession number NM_019157).

Northern Blot
Total RNA from oocytes, lung, skin, fat body, muscle, kidney, urinary bladder, stomach, intestine, and liver was extracted from adult female X. laevis using Trizol reagent (Invitrogen), according to manufacturer's directions. Total RNA (10 g) was resolved by formaldehyde-agarose (1%) denaturing gels and blotted to positively charged nylon membrane (Hybond XL, Amersham Biosciences) by capillary elution. Blots were prehybridized by incubation in ExpressHyb hybridization solution (Clontech Laboratories, Inc) at 65°C for 30 min and then hybridized with an ␣-32 P-labeled probe (Random Primer labeling kit, Invitrogen) corresponding to the full-length AQPxlo sequence at 65°C for 90 min. The blots were subjected to a high stringency wash (two 15-min washes at room temperature with 1ϫ SSC, 0.1% SDS and two 30-min washes at 50°C with 0.1ϫ SSC, 0.1% SDS), after which they were mounted and autoradiographed.

Oocyte Expression
The AQPxlo fragment was excised from pCR 2.1 TOPO using XhoI and SpeI and subcloned into a KSM oocyte expression vector, in which the multiple cloning site is surrounded by the 5Ј-and 3Ј-untranslated regions for Xenopus ␤-globin. This expression vector was a kind gift of Dr. William Joiner (Yale University). Rat AQP7 was excised from pCR 2.1 TOPO using XhoI and HindIII and ligated to KSM. Capped cRNA was transcribed in vitro using T3 Message Machine kit (Ambion, Austin, TX).
Stage V-VI oocytes were defolliculated with collagenase type Ia (Sigma) and stored in OR3 medium (Sigma) as described previously (3). On the day after isolation, oocytes were injected with 50 nl of cRNA encoding either AQPxlo (0.4 g/l), rat AQP3 (0.2 g/l), rat AQP7 (0.2 g/l), the rat urea transporter UT-A2 (0.2 g/l), or human AQP1 (0.05 g/l). Control oocytes were injected only with water. The cDNA encoding AQP1 (in the Xenopus expression vector pX␤G) was a gift from Dr. Peter Agre (The Johns Hopkins University, Baltimore, MD); AQP3 (in pSPORT) was a gift from Dr. Lawrence Palmer (Cornell University, New York, NY); and UT-A2 (in pBluescript) was a gift from Dr. Craig Smith (University of Manchester, UK).

Measurement of Oocyte Water Permeability
We used a volumetric assay to measure the osmotic water permeability (P f ) of oocytes injected with various cRNAs or water (control). The oocyte was placed in a perfusion chamber and illuminated from below. We acquired images of the oocyte silhouette every 2 s through a video camera attached to a stereomicroscope. The images were saved to a disk and analyzed using Optimas software version 5.2 (Media Cybernetics, Silver Spring, MD). The oocyte volume was calculated from the cross-sectional area of the oocyte, assuming the oocyte to be a perfect sphere. The volume was calibrated using, as an underwater standard, a brass ball bearing placed near the oocyte.
The oocyte was initially superfused with isotonic ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES, pH 7.5, osmolality 200 mOsm) at a solution flow of 4 ml min Ϫ1 . To induce cell swelling, the perfusate was switched to a hypo-osmotic solution with half the osmolality of ND96, prepared by reducing the concentration of NaCl to 43 mM. Osmolalities were measured using a vapor pressure osmometer (Wescor, Inc. Logan, UT).
To study extracellular pH (pH o ) sensitivity of P f in oocytes expressing AQPxlo, we preincubated the oocyte for 2 min in an iso-osmotic solution with the appropriate pH (5-11.5) before switching to the hypo-osmotic solution. For solutions with pH 5 or 6, we replaced HEPES with MES. For solutions with pH 7.5 or 8.5 we used HEPES. For solutions with pH 9.5, 10.5, or 11.5, we replaced HEPES with CHES. To avoid precipitation, we omitted MgCl 2 and CaCl 2 from the pH 11.5 solutions and increased the amount of NaCl to maintain osmolality.
To study the effect of HgCl 2 on P f of oocytes expressing AQPxlo, we preincubated the oocytes for 5 min in ND96 solution containing the appropriate concentration of HgCl 2 (0.1-1000 M) before switching to hypo-osmotic solution containing the same HgCl 2 concentration. We studied the reversibility of the HgCl 2 treatment by sequentially subjecting the same oocyte to cell swelling three times as follows: (i) without HgCl 2 , (ii) with 1 M HgCl 2 , and (iii) without HgCl 2 but after a 5-min incubation period with ␤-mercaptoethanol (5 mM).
The osmotic water permeability was calculated using Equation 1 (4,5), where (d(V/V o )/dt) is the initial rate of increase in relative oocyte volume following a hypo-osmotic challenge; V o is the initial oocyte volume; S is the actual surface area; ⌬ osm is the osmotic gradient across the oocyte membrane, and V w is the molar volume of water.
S of the oocyte was obtained by multiplying the geometric surface area (calculated from the measured volume of each oocyte) by 8, as it has been shown that the actual surface area, accounting for the numerous infoldings and microvilli of the plasmalemma, is 8 -9 times larger than the area of a sphere of equivalent size (5).

Measurement of Osmolyte Permeation in Oocyte
Radioisotope Uptake-We measured unidirectional influx of urea and glycerol in oocytes using 14 C-labeled urea ([ 14 C]urea) and glycerol ([U-14 C]glycerol; Moravek Biochemicals, Brea, CA). Four oocytes were placed in a 1.5-ml microcentrifuge tube in ND96 solution. Measurement of isotope uptake was started by replacing the ND96 solution with 700 l of uptake solution, which consisted of ND96 containing 1 mM unlabeled analyte and 1 Ci/ml (37 kBq/ml) of the radioisotopically labeled analyte. Oocytes were incubated on a horizontal shaker for 2 min at room temperature. In preliminary experiments, we monitored the time course of 14 C uptake to ensure that the uptake was linear during the first 2 min. Radioisotope uptake was stopped by washing oocytes three times in ice-cold ND96 containing 10 mM of the unlabeled analyte. Individual oocytes were transferred to scintillation vials and lysed in 400 l of 10% SDS with continuous shaking. The 14 C activity of individual oocytes was assessed by liquid scintillation counting (LKB-Wallac Rackbeta, Turku, Finland).
We calculated an apparent rate constant k using Equation 2, where t is the incubation time; A OOCYTE is the radioactivity accumulated by an oocyte (assuming an oocyte volume of 1 l) over the time of incubation (t), and A U is the radioactivity of 1 l of uptake solution. When inhibitors were used, the oocytes were pretreated for 5 min in ND96 solution containing the inhibitor before adding the uptake solution (which also contained the inhibitor). Both HgCl 2 (1 and 300 M) and phloretin (0.5 mM, final concentrations) were prepared fresh daily.
Reflection Coefficient -The reflection coefficient is an index of the effectiveness of a solute in generating an osmotic driving force across a semipermeable membrane. For ϭ 1, the membrane is impermeant to the solute, and for ϭ 0, the permeability of the solute is indistinguishable from that of the solvent. For a semipermeable solute, 0 Յ Յ 1. To determine , we used a variation of the volumetric method (see above) that we used to measure P f . For each oocyte, we obtained an initial P f value by subjecting the oocyte to swelling by exposing it to a hypoosmotic (100 mOsm) ND96 solution. After allowing the oocyte to recover in iso-osmotic ND96 solution for at least 5 min, we superfused the oocyte with a solution in which we iso-osmotically substituted 100 mM of the test solute for NaCl. When Յ 1, the solute will enter the oocyte together with osmotically obliged water, leading to cell swelling. The oocyte was again allowed to recover before superfusing with another solute (see below). For AQPxlo-expressing oocytes we measured for six different solutes as follows: the polyols glycerol (C-3), xylitol (C-5), ribitol (C-5), and mannitol (C-6), as well as urea and thiourea. Ribitol and xylitol were chosen because in ribitol, each of the -OH groups is on the same side of the carbon backbone, whereas in xylitol the -OH groups have a mixed stereospecific arrangement. In the bacterial glycerol facilitator GlpF, such stereoselectivity plays an important role in determining the permeability properties of the channel (6). As controls, we measured for glycerol and urea in water-injected oocytes, as well as in oocytes expressing human AQP1 and rat AQP3.
We calculated according to Equation 3,

Measurement of Intracellular pH
An oocyte, placed in a perfusion chamber, was superfused at a rate of 4 ml min Ϫ1 . pH-sensitive electrodes were fabricated and used as described previously (7,8). Briefly, the oocyte was impaled with two microelectrodes, one for measuring the membrane potential (V m ) and the other one for measuring intracellular pH (pH i ). The tip of the pH electrode contained a pH-sensitive liquid membrane (Hydrogen Ionophore I, mixture B, Fluka Chemical Corp., Ronkonkoma, NY). Voltages were measured using an FD 223 Electrometer (World Precision Instruments, Inc., Sarasota, FL), and the voltage due to pH i was obtained by an analogue subtraction of the signals from the V m and pH electrodes. The pH calibration was performed as described previously (9).

RESULTS
Sequence Data-The cDNA encoding AQPxlo contains an open reading frame of 894 bp, which encodes 297 amino acids. Fig. 1A shows a sequence alignment with AQPxlo and human AQP3, AQP10, and the bacterial glycerol facilitator GlpF. Fig.  1B shows a dendrogram containing members of all vertebrate AQPs identified to date (AQPs 0 -10). We also included xAQP3 and GlpF for comparison. An amino acid sequence-alignment analysis of these AQPs shows that AQPxlo is ϳ49% identical to mammalian and X. laevis AQP3; ϳ39 -45% identical to mammalian AQPs 7, 9, and 10; and ϳ30% identical to the bacterial GlpF. Sequence identity to mammalian AQP1 is ϳ18%. Thus, it appears that AQPxlo is a member of the aquaglyceroporin subgroup of the AQP family. Fig. 1C shows a hydropathy plot for the deduced amino acid sequence of AQPxlo. The plot is similar to that of other members of the AQP family and is consistent with six transmembrane segments and five connecting loops. Loops B and E contain the aquaporin family signature motif, the amino acid sequence Asn-Pro-Ala. Loop B of AQPxlo contains a consensus protein kinase A/G phosphorylation motif (KKLS) at positions 94 -97. Ser 97 is also part of a consensus protein kinase C phosphorylation motif (SWK) at residues 97-99. Loop C contains a consensus N-glycosylation site at position 134.
Northern Analysis- Fig. 2 shows a Northern blot analysis of total RNA from multiple tissues from Xenopus. A strong ϳ1.4-kb signal is present in oocytes and fat body, and a somewhat weaker signal is seen in kidney. We detected no signal in lung, skin, skeletal muscle, urinary bladder, stomach, intestine, or liver.
Water Permeability- Fig. 3 shows the osmotic water permeabilities of control oocytes and of oocytes expressing human AQP1, rat AQP3, rat AQP7, and AQPxlo. For each clone, we injected the minimum amount of cRNA/oocyte required to produce a near-maximum P f . In oocytes expressing the archetypal water channel AQP1, P f was 30-fold higher than in waterinjected oocytes, whereas in oocytes expressing the aquaglyceroporins AQP3 and AQP7, P f was 15-and 4-fold higher, respectively. The P f of AQPxlo-expressing oocytes increased 6-fold, compared with controls. Thus, AQPxlo appears to function as a water channel when overexpressed in the oocyte. Compared with oocytes expressing AQP1, the water permeability of oocytes expressing AQPxlo is low. However, because we have no information about the actual number of functional channels at the membrane, our data do not permit a direct comparison of the water permeabilities of the different AQPs.
Effect of pH o Changes on Water Permeability-Zeuthen and Klaerke (10) have shown that the P f and the glycerol permeability of AQP3 are virtually 0 at extracellular pH (pH o ) values below 6 and rise to maximum values at pH o 6.5-7.0. Fig. 4 shows how changes in pH o affect P f in control oocytes and in oocytes expressing AQPxlo. As pH o increases from 5.0, the P f of AQPxlo-expressing oocytes increases markedly. Maximal P f is observed at pH o 9.5, which is 4-fold higher than the value at pH 7.5 and 8-fold higher than at 5.0. On the other hand, increasing pH o from 9.5 to 10.5 or 11.5 caused small decreases in P f . In water-injected oocytes, increasing pH o from 5.0 to 11.5 did not have a statistically significant effect on P f .
In order to determine whether shifts in pH i may have played a role in the P f changes shown in Fig. 4A, we measured pH i and V m of water-injected and AQPxlo-expressing oocytes exposed to ND96 at different pH o values. Changing pH o from 7.5 to 5.0, 9.5, or 10.5 resulted in changes in pH i of less than 0.05 pH units ( Fig. 4B). Lowering pH o from 7.5 to 5.0 caused the oocyte to depolarize reversibly by ϳ15 mV, whereas raising pH o from 7.5 to either 9.5 or 10.5 caused the oocyte to hyperpolarize reversibly by 10 -15 mV. In contrast to raising pH o to only 9.5 or 10.5, raising pH o to 11.5 (Ca 2ϩ -and Mg 2ϩ -free) caused a large and rapid intracellular alkalinization and an irreversible depolarization. Our observation that oocyte pH i is remarkably resistant to pH o changes over the range from 5.0 to 10.5 implies that, over this range, the effects on the P f of AQPxlo-expressing oocytes are mediated by extracellular protons.
We also examined P f at pH o values of 7.5 and 9.5 in oocytes expressing rAQP3 and rAQP7. As summarized in Fig. 5A, P f is significantly increased at alkaline pH o in AQPxlo-expressing oocytes but not in rAQP3-or rAQP7-expressing oocytes. In this batch of oocytes, the P f of water-injected oocytes is significantly increased at alkaline pH o , suggesting that they may have had a higher level of endogenous AQPxlo expression than those in Fig. 4.
Effect of pH o Changes on Glycerol and Urea Uptake-Because our sequence analysis placed AQPxlo among the aquaglyceroporins, we measured the uptake of glycerol and urea into water-injected oocytes, as well as oocytes expressing rAQP3, rAQP7, or AQPxlo. As shown in Fig. 5B, all of the AQPs tested substantially increased glycerol uptake (6 -7-fold compared with water-injected oocytes). By contrast, Fig. 5C shows that only AQPxlo produced a substantial increase in urea uptake (11-fold). This compares to a 2.6-fold increase in urea uptake seen in rAQP3-expressing oocytes. In contrast to previous reports (11), we did not observe an increase in the rate of urea uptake in rAQP7-expressing oocytes.
Because raising pH o from 7.5 to 9.5 substantially increased P f in AQPxlo-expressing oocytes, we examined the effect of this pH o on glycerol and urea uptake. As shown in Fig. 5B, increasing pH o from 7.5 to 9.5 did not affect the rate of glycerol uptake in any of the oocytes expressing an AQP. Curiously, only the water-injected oocytes showed a significant increase in glycerol uptake at alkaline pH o , perhaps reflecting the presence of an endogenous pH-sensitive glycerol transport pathway other than AQPxlo.
In oocytes expressing AQPxlo, the effect on urea uptake of raising pH o seemed to be intermediate between that on P f (2.7-fold; see Fig. 5A) and on glycerol uptake (no effect; see Fig.  5B). As shown in Fig. 5C, raising pH o from 7.5 to 9.5 caused a significant increase in urea uptake (1.5-fold) in oocytes expressing AQPxlo. Raising pH o failed to increase urea uptake in any of the other groups of oocytes tested.
Effect of HgCl 2 on Water Permeability- Fig. 6 shows how HgCl 2 affects the P f of water-injected oocytes and oocytes ex-pressing hAQP1 or AQPxlo. For AQPxlo-expressing oocytes the interaction with HgCl 2 appears biphasic. At low concentrations of HgCl 2 (0.3-10 M) the P f is low, whereas at high HgCl 2 concentrations (300 -1000 M), P f is similar to the value obtained in the absence of HgCl 2 . In water-injected oocytes, 1 M HgCl 2 produced a small but statistically significant inhibition of P f . In contrast, the effects of HgCl 2 on the water permeability of AQP1-expressing oocytes are monophasic. Correcting for the background P f in water-injected oocytes, 0.3 M HgCl 2 reduced the AQPxlo-dependent P f by 78%, and 300 M HgCl 2 reduced the AQP1-dependent P f by 82%.
To investigate whether HgCl 2 has time-dependent effects on P f , we added 1 or 300 M HgCl 2 and then measured P f after 2, 5, and 20 min. The time of preincubation did not affect P f at either concentration of HgCl 2 (not shown).
Because alkaline pH o increases the P f of AQPxlo-expressing oocytes (Fig. 4), we investigated whether HgCl 2 at low concentration (1 M) still blocks P f when pH o is increased from 7.5 to 9.5. Because HgCl 2 may act as a Cl Ϫ /OH Ϫ ionophore (12), we decided not to examine P f at different pH o values in the presence of HgCl 2 . Instead, we first pretreated AQPxlo-expressing oocytes with 1 M HgCl 2 for 5 min in iso-osmotic ND96 at pH o 7.5. We then exposed the oocytes to either (i) a hypo-osmotic solution at pH o 7.5 with HgCl 2 , (ii) a 2-min wash with an iso-osmotic HgCl 2 -free solution at pH o 7.5, followed by the comparable hypo-osmotic solution, or (iii) a 2-min wash with an iso-osmotic HgCl 2 -free solution at pH o 9.5, followed by the comparable hypo-osmotic solution. Fig. 7A shows that, at pH o 7.5, HgCl 2 reduced P f by ϳ70% whether it was present or absent from the hypotonic swelling solution. In contrast, when P f was measured at pH o 9.5, HgCl 2 pretreatment reduced P f by only 43%. Fig. 7A shows that, in AQPxlo-expressing oocytes, inhibition of P f by HgCl 2 is not reversed by washing. We therefore determined whether inhibition by HgCl 2 could be reversed using a reducing agent by superfusing an oocyte for 5 min with 5 mM ␤-mercaptoethanol (after first measuring P f before and after treatment with 1 M HgCl 2 ). Fig. 7B shows that ␤-mercaptoethanol treatment completely reversed the inhibition of AQPxlo by HgCl 2 .
Effect of Inhibitors on Glycerol and Urea Uptake-We investigated whether HgCl 2 has a similar biphasic inhibition profile on glycerol and urea uptake as it has on P f (Fig. 6). Fig. 8, A and B, shows that a low (1 M) concentration of HgCl 2 reduces both glycerol and urea uptakes of AQPxlo-expressing oocytes. In contrast, a high (300 M) concentration of HgCl 2 inhibits glycerol uptake but has no effect on urea uptake. Thus, HgCl 2 acts with a different concentration profile on P f and urea uptake than on glycerol uptake. In water-injected oocytes or rAQP3expressing oocytes, 300 M HgCl 2 had no effect on glycerol uptake. In contrast, in water-injected oocytes, urea uptake was reduced by the low (1 M) concentration of HgCl 2 but was increased by the high (300 M) HgCl 2 concentration.
Of the 11 mammalian AQPs cloned to date, only AQP9 (13) and AQP10 (14) show substantial glycerol and urea permeabilities, raising the possibility that AQPxlo is the amphibian orthologue of one of these. Because phloretin blocks the urea permeabilities of both AQP9 and AQP10, we investigated the effect of phloretin (500 M) on the urea uptake of AQPxlo. Fig.  8C shows that phloretin strongly inhibits urea uptake in both water-injected oocytes and oocytes expressing the rat urea transporter rUT-A2 but has no effect on AQPxlo-expressing oocytes.
Measurement of Reflection Coefficient - Fig. 9A shows the reflection coefficient of AQPxlo-expressing oocytes for glycerol, xylitol, ribitol, mannitol, urea, and thiourea. Of these, glycerol is the only solute with a (0.57 Ϯ 0.09) substantially smaller than 1, indicating that glycerol permeates AQPxlo-expressing oocytes. This result is in good agreement with the isotopic flux data in Fig. 5B. The obtained for urea in AQPxlo-expressing oocytes (0.95 Ϯ 0.02) is only slightly less than 1, but the difference is statistically significant (p Ͻ 0.0089). Neither xy-litol, ribitol, mannitol, nor thiourea had values significantly lower than 1, indicating that they are not able to permeate AQPxlo-expressing oocytes.  7. HgCl 2 inhibition of P f , effect of pH o changes, and reversibility. A, effect of pH o . AQPxlo-expressing oocytes were either untreated or pretreated with 1 M HgCl 2 at pH o 7.5 for 5 min. Subsequently, P f was measured (i) in untreated oocytes at pH o 7.5 or 9.5 (black bars); (ii) in pretreated oocytes in the continued presence of HgCl 2 at pH o 7.5 (light gray bar); or (iii) in pretreated oocytes in the absence of HgCl 2 either at pH o 7.5 or 9.5 (dark gray bars). B, reversibility of HgCl 2 inhibition. P f was measured in the same oocyte before HgCl 2 treatment, after treatment with HgCl 2 (1 M), and after treatment with ␤-mercaptoethanol (5 mM). Error bars, S.E. Asterisk indicates that the difference between HgCl 2 -treated and -untreated oocytes is statistically significant (p Ͻ 0.05) in a two-tailed t test. Fig. 9B shows measurements for water-injected oocytes as well as oocytes expressing hAQP1 and rAQP3. rAQP3-expressing oocytes had a of 0.37 Ϯ 0.01, in agreement with the isotopic flux data in Fig. 5B. Interestingly, water-injected oocytes had a (0.68 Ϯ 0.18) for glycerol that was similar to that of AQPxlo-expressing oocytes, consistent with the presence of endogenous AQPxlo. DISCUSSION We have cloned from Xenopus oocytes a novel AQP that is predominantly expressed in fat body, oocytes, and kidney. AQPxlo is permeable to water, glycerol, and urea but excludes the larger polyols xylitol, ribitol, and mannitol, as well as thiourea. Judging from its strong expression in fat body, AQPxlo may play a role in the triglyceride metabolism in this tissue by promoting glycerol transport. The fat body, an organ associated with the anterior testis or ovary, serves as a storage organ for fat and undergoes marked changes in size in many anurans, depending on the season and the nutritional status of the animal.
Sensitivity to Changes in pH o -The P f of AQPxlo falls sharply as one lowers pH o below 9.5 and more gradually as one raises pH o above 9.5 (Fig. 4A). Regarding other AQPs, both the P f and glycerol permeability of AQP3 fall sharply at acidic pH o values, with a pK of 6.4 for P f and 6.1 for glycerol (10). According to one report, the P f of bovine major intrinsic protein (AQP0) is maximal at pH o 6.5 and falls at both more acidic and more alkaline pH o values (15).
The biphasic pH o dependence of P f in AQPxlo-expressing oocytes is consistent with the idea that AQPxlo has more than one titratable site that influences the permeability pathway for water, and that the sum of effects is positive in its influence from pH 5.0 to 9.5 and negative above pH 9.5. Because oocyte pH i is remarkably resistant to pH o changes between pH o 5.0 and 10.5 (Fig. 4B), it is likely that these sites are located where they are accessible to extracellular protons.
In stark contrast to P f , which increased 4-fold as we raised pH o from 7.5 to 9.5, glycerol uptake in AQPxlo-expressing oocytes was unaffected (Fig. 5B). The effect of pH o on urea uptake in AQPxlo-expressing oocytes falls between that of P f and glycerol uptake in that increasing pH o from 7.5 to 9.5 produced a small but significant increase in the rate of urea uptake (Fig. 5C). Our observation that the movement of the small H 2 O is very pH o -sensitive, whereas the uptake of the much larger glycerol is not pH o -sensitive at all, suggests that pH o sensitivity does not depend in any simple way on molecular size. Rather, the complex pattern of pH o sensitivities may reflect interactions between the transported species and specific titratable residues lining the channel.
Sensitivity to HgCl 2 -The effect of HgCl 2 on the P f (Fig. 6) and urea uptake (Fig. 8B) of AQPxlo-expressing oocytes is novel. A low (1 M) HgCl 2 concentration reduces both P f and urea uptake, whereas a higher (300 M) concentration reverses the inhibition. In contrast, HgCl 2 inhibited glycerol uptake equally well at both low (1 M) and high (300 M) HgCl 2 concentrations (Fig. 8A).
The simplest way to interpret the data is to assume that the action of HgCl 2 involves at least two reactive groups. One (a high affinity site) reacts at low concentrations of Hg 2ϩ , resulting in a marked decrease in the permeability of AQPxlo to H 2 O, glycerol, and urea. The other (a low affinity site) reacts only when the HgCl 2 concentration is elevated, and its effect is to reverse only the inhibition of P f and urea uptake. Presumably the reaction of the first Hg 2ϩ partially obstructs the pore of AQPxlo, reducing the transport of all substances. The second Hg 2ϩ might then cause a conformational change that relieves the obstruction just enough to let H 2 O and urea (but not glycerol) pass. We do not know the nature of the Hg 2ϩ -reaction sites, but they are likely to be thiol groups of cysteine residues, as is the case for several other AQPs. AQPxlo has six cysteines, at positions 24, 40, 74, 105, 174, and 210. Preston et al. (16) showed that of the four cysteines present in AQP1, Cys 189 mediates the blocking effect of mercurials. AQPxlo has a tyro- Asterisk indicates that the difference between that value and water-injected control is statistically significant (p Ͻ 0.05) in a two-tailed t test. Dagger indicates that the difference between that value and no-drug control (black bar) is statistically significant (p Ͻ 0.05) in a two-tailed t test. sine (Tyr 212 ) at the position equivalent to Cys 189 in AQP1 but has a cysteine nearby (Cys 210 ).
Interestingly, Yasui et al. (17) showed that HgCl 2 causes an ϳ10-fold increase in the P f of oocytes expressing human AQP6, and also induces an ion conductance. The authors showed that the double mutant Cys 155 -Cys 190 is unresponsive to stimulation by HgCl 2 . AQPxlo has a cysteine (Cys 174 ) at the position equivalent to Cys 155 in AQP6 (Cys 152 in AQP1), but has the aforementioned Tyr 212 at the position equivalent to Cys 190 in AQP6 (Cys 189 in AQP1). Thus, of the six cysteine residues in AQPxlo, Cys 174 and Cys 210 are good candidates for HgCl 2 reaction sites.
Does AQPxlo Contribute to Transport in Native Oocytes?-Because the water permeability of native Xenopus oocytes is low, it is perhaps surprising that they should contain significant amounts of mRNA encoding an endogenous water channel, as indicated by our Northern blot analysis (Fig. 2). On the other hand, the AQPxlo protein may not be expressed at the cell membrane or at all. Indeed, oocytes contain large amounts of stored mRNA that are not translated into protein until the egg starts developing. Our data indicate that native AQPxlo expression in the oocyte is, at most, low. For example, the P f of water-injected oocytes showed little (Fig. 5A) or no (Fig. 4) change in response to changes in pH o . Also, glycerol uptake in water-injected oocytes was not blocked by HgCl 2 . On the other hand, HgCl 2 at 1 M reduces both P f and urea uptake in water-injected oocytes, consistent with native AQPxlo expression. It is possible that if oocytes express the AQPxlo protein at the cell membrane, then the level of expression might vary among batches of oocytes.
However, native AQPxlo might contribute significantly to transport in oocytes heterologously expressing other membrane proteins. Several groups (18 -20) have reported examples in which the heterologous expression of a membrane protein in Xenopus oocytes has led to the up-regulation of native Xenopus membrane proteins. For example, the heterologous overexpression of any of several membrane proteins leads to the appearance of a native nonspecific cation channel in Xenopus oocytes (18). In addition, expression of rBAT or 4F2hc in Xenopus oocytes induces amino acid transport, although the proteins themselves do not possess intrinsic transport activity (19). Instead, they associate with native Xenopus membrane proteins to form a heteromeric amino acid transporter (20).
The up-regulation of endogenous AQPxlo could account, at least in part, for novel transport activity observed in oocytes heterologously expressing two other membrane proteins. First, expressing major intrinsic protein (AQP0) in oocytes leads to an increase in glycerol permeability (21)(22)(23), even though major intrinsic protein in its native environment in the lens does not contribute to significant glycerol permeability (24). Second, Schreiber et al. (25) have shown that IBMX triggers an increase in water and glycerol permeability in oocytes expressing CFTR. In principle, the up-regulation of AQPxlo could account for this observation. On the other hand, Schreiber et al. (2) eliminated the IBMX-stimulated water and glycerol permeabilities by injecting antisense oligonucleotides directed against a Xenopus AQP3 and concluded that CFTR up-regulates endogenous xAQP3. We were unable to obtain from oocytes a PCR product corresponding to xAQP3 2 ; however, it is possible that the Xenopus used by Schreiber et al. (2) contained a significant amount of mRNA encoding xAQP3, whereas those used by us did not.
Is AQPxlo a New Aquaporin?-Our sequence and functional data indicate that AQPxlo is an aquaglyceroporin. Four paralogous aquaglyceroporins are known in mammals, AQPs 3, 7, 9, and 10. The dendrogram in Fig. 1 suggests that AQPxlo is most closely related to AQP3 and AQP10, sharing ϳ49% amino acid sequence identity with mammalian and Xenopus AQP3 and ϳ45% identity with AQP10. This level of sequence similarity is less than seen between mammalian AQP2 and AQP5 (ϳ60%), which are recognized as distinct AQP paralogues. Thus, a comparison of deduced amino acid sequences suggests that AQPxlo represents a new paralogous AQP. However, searching the human and mouse genome data bases has thus far failed to reveal a mammalian orthologue of AQPxlo.
The permeability properties of AQPxlo are unique. Unlike AQP3 and AQP7, AQPxlo is permeable to urea. Unlike AQP9 (26), AQPxlo does not serve as a pathway for mannitol or thiourea and is not sensitive to inhibition by phloretin. Unlike the recently cloned AQP10 (14), which takes up glycerol 7-fold faster than urea, AQPxlo takes up glycerol and urea equally fast (as measured by 14 C uptake). Also unlike AQP10, whose urea uptake is blocked by phloretin, AQPxlo-mediated urea uptake is not inhibited by phloretin. Thus, the conclusion that AQPxlo is a new AQP paralogue is supported by our observation that the permeability properties of AQPxlo are unlike those of any of the known mammalian aquaglyceroporins.