Cloning, Heterologous Expression, and Characterization of Three Aquaglyceroporins from Trypanosoma brucei*

Trypanosoma brucei, causative for African sleeping sickness, relies exclusively on glycolysis for ATP production. Under anaerobic conditions, glucose is converted to equimolar amounts of glycerol and pyruvate, which are both secreted from the parasite. As we have shown previously, glycerol transport in T. brucei occurs via specific membrane proteins (Wille, U., Schade, B., and Duszenko, M. (1998) Eur. J. Biochem. 256, 245–250). Here, we describe cloning and biochemical characterization of the three trypanosomal aquaglyceroporins (AQP; TbAQP1–3), which show a 40–45% identity to mammalian AQP3 and -9. AQPs belong to the major intrinsic protein family and represent channels for small non-ionic molecules. Both TbAQP1 and TbAQP3 contain two highly conserved NPA motifs within the pore-forming region, whereas TbAQP2 contains NSA and NPS motifs instead, which are only occasionally found in AQPs. For functional characterization, all three proteins were heterologously expressed in yeast and Xenopus oocytes. In the yeast fps1Δ mutant, TbAQPs suppressed hypoosmosensitivity and rendered cells to a hyper-osmosensitive phenotype, as expected for unregulated glycerol channels. Under iso- and hyperosmotic conditions, these cells constitutively released glycerol, consistent with a glycerol efflux function of TbAQP proteins. TbAQP expression in Xenopus oocytes increased permeability for water, glycerol and, interestingly, dihydroxyacetone. Except for urea, TbAQPs were virtually impermeable for other polyols; only TbAQP3 transported erythritol and ribitol. Thus, TbAQPs represent mainly water/glycerol/dihydroxyacetone channels involved in osmoregulation and glycerol metabolism in T. brucei. This function and especially the so far not investigated transport of dihydroxyacetone may be pivotal for the survival of the parasite survival under non-aerobic or osmotic stress conditions.

The protozoan parasite Trypanosoma brucei undergoes a complex life cycle including two dividing stages, i.e. the long slender form in blood, lymphatic fluid, and cerebrospinal fluid of its mammalian host and the procyclic form in the mid-gut of the tsetse fly. Due to the lack of a functional Krebs cycle and oxidative phosphorylation capabilities, the bloodstream forms depend on glycolysis for ATP production, with glycerol as an alternative substrate (1,2). Glycolysis in trypanosomes differs markedly from the corresponding pathway in higher eukaryotes. Most strikingly, the first seven glycolytic reactions occur in a peroxisome-like organelle called glycosome (3).
Under aerobic conditions, glucose is converted into pyruvate, which is disposed by facilitated diffusion (4). The concurrently formed NADH cannot leave the glycosome and is reoxidized during the formation of glycerol-3-phosphate from dihydroxyacetone phosphate. Glycerol-3-phosphat enters the mitochondrion, in which it is oxidized by glycerophosphate oxidase. This enzyme uses molecular oxygen (alternative respiration) and is inhibited by salicylhydroxamic acid. Finally, the newly formed dihydroxyacetone phosphate re-enters the glycosome and can be glycolytically converted. Obviously, suppression of glycolysis by blocking the release of pyruvate or glycerol is a potential therapeutic approach against sleeping sickness (2, 4 -6).
Under anaerobic conditions or in the presence of salicylhydroxamic acid, glucose degradation results in equimolar amounts of pyruvate and glycerol to maintain the redox balance within the glycosome (7). In this case, glycerol is formed from glycerol-3phosphate by glycerokinase, which in trypanosomes leads to the formation of ATP (8). To drive this reaction to ATP formation, glycerol has to be readily released. Consequently, in the presence of salicylhydroxamic acid, the addition of 5 mM glycerol to the medium reverses the glycerol diffusion gradient across the membrane and results in cell death. A therapeutic exploitation of the glycerol sensitivity of the parasite, however, has not yet been successful in situ (5,9). Hence, more information is required to understand how the parasite copes with non-aerobic conditions, regarding metabolism and glycerol transport.
We have shown previously that glycerol uptake is consistent with facilitated diffusion through a transporter protein (10). This investigation gave clear evidence for an additional nonsaturable channel, which remained active after inhibition of the transporter. As we now know, glycerol facilitators are members of the superfamily of major intrinsic protein and fall into the aquaporin branch. Aquaporins facilitate permeation of water (orthodox aquaporins) or of small non-ionic solutes, such as glycerol and urea (aquaglyceroporins). Aquaporins of both types are present throughout all kingdoms of nature and participate in diverse biological processes (11). Here, we report the identification and cloning of three aquaglyceroporins from * This work was funded by the Deutsche Forschungsgemeinschaft (Du-136-11 and Be-2253-2). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ697889, AJ697890, and AJ697891 § Recipient of a personal grant from Consejo de Desarrollo Cientifico y Humanistico (Venezuela). ** Recipient of a personal grant from Deutscher Akademischer Austauschdienst (Germany).

MATERIALS AND METHODS
Strains and Media-T. brucei bloodstream forms (strain 427; clone MITat 1.2) were either propagated in Sprague-Dawley rats and purified from blood using a DEAE-cellulose column (12) or cultured at 37°C in a 5% CO 2 atmosphere using modified minimum essential medium as described previously (13). Procyclic forms, obtained from MITat 1.2 using a standard transformation protocol (14), were also grown in a modified minimum essential medium (15).
Cloning of Aquaglyceroporin Genes from T. brucei and Heterologous Expression in S. cerevisiae-The complete cDNAs of TbAQP1-3 were cloned by reverse transcriptase-PCR using RNA obtained from bloodstream form and primers against the spliced leader sequence (5Ј-CGC TAT TAT TAG AAC AGT TTC TG-3Ј) or against the poly-A tail (5Ј-CCC GGG T (20) VNN-3Ј) along with primers derived from aquaporin-like sequences published by the TIGR data base (www.tigr.org). The open reading frames were cloned into pBSKS ϩ (Stratagene, La Jolla, CA) and sequenced (GATC GmbH, Konstanz, Germany). A list of primers used is available upon request. PCR products were also cloned into SmaI-cut pRS416 (17) to give rise to yeast expression vectors (MET17inducible). EGFP fusions of TbAQP1-3 were constructed by cloning HindIII-cut PCR products obtained with primers carrying HindIII-sites immediately adjacent to sequences flanking the respective open reading frames into HindIII of pRS416 with an egfp gene for which the stop codon was replaced by a HindIII site.
Glycerol Efflux Experiments of TbAQP1, TbAQP2, and TbAQP3 in S. cerevisiae under Iso-osmotic and Hyperosmotic Conditions-Cells were grown at iso-osmotic conditions overnight in synthetic medium until an optical density of 1 OD at ϭ 600 nm was obtained. Part of this culture was harvested by centrifugation (4°C, 3,500 ϫ g, 10 min), resuspended in the same medium containing 5% NaCl (hyperosmotic stress), and further incubated at room temperature and 160 rpm for 2 h. Determination of intracellular glycerol was performed by filtration (18). For the determination of extracellular glycerol, an aliquot of the cellfree supernatant was used. To determine the dry weight of yeast, aliquots of cells were collected onto Whatman GF/C filters and dried at 37°C for 1 week.
Expression of TbAQP1, TbAQP2, and TbAQP3 in X. laevis Oocytes-For expression in Xenopus oocytes, TbAQP1, TbAQP2, and TbAQP3 genes were subcloned into the EcoRV site of the pT7T5 expression vector, which contains 5Ј-and 3Ј-untranslated regions of the Xenopus ␥-globin gene. After linearization with SmaI, cDNAs were transcribed in vitro with 20 units of T7 RNA polymerase at 37°C in the presence of 25 nM rNTPs, 40 units of RNase inhibitor, and 300 M cap analogue m 7 G(5Ј)ppp(5Ј)G. Template plasmids were removed by digestion with 10 units of RNase-free Dnase I for 15 min at 37°C. 10 ng of each cRNA were injected into defolliculated oocytes of stages V-VI. After injection, oocytes were kept at 17°C in ND96 storage solution containing 96 mM NaCl, 2 mM KCl, 1.8 mMCaCl 2 , 1 mM MgCl 2 and 5 mM HEPES, pH 7.4 (19).
Kinetic Glycerol Uptake of TbAQPs in Oocytes-For transport assays, 8 -10 oocytes were washed and placed in 0.5 ml of fresh ND96 buffer. Glycerol uptake was started by replacement of the ND96 solution by ND96 containing 1 mM unlabeled glycerol and 1 Ci/ml of [U-14 C] glycerol. Uptake of radiolabeled glycerol was stopped at defined time points by washing the cells three times in ND96 solution (at about Ϫ3°C) containing 100 mM unlabeled glycerol (20,21). Individual oocytes were dissolved in 1 ml of 10% SDS and subject to scintillation counting.
Standard Oocytes Swelling Assay-Xenopus leavis oocytes were injected with 50 nl of water (control), 50 nl of water containing 10 ng of cRNA (TbAQP1, TbAQP2, or TbAQP3), or 50 nl of water containing a combination of 10 ng (TbAQP1, TbAQP2, and TbAQP3) and 2.5 ng of AQP1 as described previously (22). To measure water permeability, oocytes were transferred into 1:3 diluted ND96 medium. Solute permeability was assayed in iso-osmotic ND96 in which 65 mM NaCl were replaced by 130 mM non-ionic test solute or 96 mM NaCl were replaced by the salt of an ionic test solute. Swelling assays were carried out at room temperature and were video-monitored. The relative oocyte volume was calculated from the covered area. Osmotic water permeability (P f , M/s) was calculated from the oocyte surface area (S ϭ 0.045 cm 2 ), the initial slope of the relative volume increase (d(V/V 0 )/dt in s Ϫ1 ), the molecular water volume (V w ϭ 18 cm 3 /mol), and the osmotic gradient (osm in -osm out ) by the following equation: . The initial swelling rates (d(V/V 0 )/dt in s Ϫ1 ) were used to compare solute permeabilities.
Northern Blot Analysis-Northern blot analysis was performed according to Jungwirth et al. (23). Total RNA from procyclic and bloodstream form trypanosomes (the latter from exponentially growing or stationary phase parasites) was prepared as described in the Qiagen RNeasy handbook. RNA samples (15 g of each) were denatured and separated on a 1.2% agarose gel containing 1.2 M formaldehyde. rRNA was used to monitor RNA loading. The gels were transferred to a Hybond-N filter and hybridized with a probe containing about 500 bp from the coding region of the TbAQP genes or of the ␤-tubulin gene as control. Northern blots were visualized by autoradiography.
Glycerol and Protein Determination-The glycerol concentration was determined enzymatically (24). Protein concentrations were estimated using the Bradford method (25) with serum albumin as a standard.

Cloning of TbAQPs
We have cloned three genes from T. brucei with similarity to major intrinsic proteins using a reverse transcriptase-PCR approach. The sequences have been submitted to the EMBL data base as TbAQP1-3 (accession numbers: AJ697889, AJ697890, and AJ697891). The open reading frames code for proteins of 321 (TbAQP1), 312 (TbAQP2), and 304 (TbAQP3) amino acids, which are very similar to each other (77% identity, Fig. 1A). TbAQP1 and -3 possess two canonical NPA motifs, whereas TbAQP2 contains unusual NSA/NPS motifs as part of the pore region and a leucine residue instead of the highly conserved arginine residue following the second NP(A/S) motif (Fig. 1A, asterisk). All three TbAQPs contain characteristically elongated N termini of up to 80 amino acids. Sequence alignments and analysis of the phylogenetic tree suggest that these proteins are more closely related to aquaglyceroporins, e.g. human AQP3, -7, and -9, than to orthodox aquaporins (Fig. 1B).

Functional Expression of TbAQPs in S. cerevisiae
To characterize channel activity of TbAQPs, the corresponding proteins were N-terminally fused with GFP and functionally expressed in an S. cerevisiae fps1⌬ mutant, which lacks a functional glycerol channel. Expression of GFP alone served as a control. As shown by fluorescence microscopy, GFP-TbAQPs were clearly located within the cellular membrane of the yeast (Fig. 2). For a functional characterization, TbAQP-expressing cells were grown overnight in isotonic medium. Subsequently, intra-and extracellular glycerol concentrations were determined. As shown in Fig. 3, control cells produced very little glycerol, which was equally distributed between the cells and the surrounding medium, whereas TbAQP1-and TbAQP3transformed fps1⌬ mutant cells showed a dramatic increase of glycerol production, which was almost exclusively secreted into the medium. Notably, yeast TbAQP2-transformants produced twice as much glycerol as compared with TbAQP1 and -3, which was also mainly secreted, although in this case, some glycerol was retained intracellularly (note the different scales between TbAQP1/3 and TbAQP2 in Fig. 3).
To further characterize this phenomenon, the experiment was repeated under hypertonic conditions (5% NaCl for 2 h), which was described to produce a very high accumulation of glycerol in S. cerevisiae (18). Due to the absence of a functional glycerol channel, control cells showed indeed a high retention of glycerol, although some glycerol appeared also in the medium, indicating that glycerol leaked out through the membrane. TbAQP1-and TbAQP3-transformed fps1⌬ mutant cells, however, showed similar results as those obtained under iso-osmotic conditions, i.e. these cells were unable to accumulate glycerol and released it constitutively into the medium. This behavior is similar to that described for fps1⌬ yeast cells expressing the glycerol facilitator from Escherichia coli (GlpF) and indicates a non-regulated high glycerol channel activity (26,27).
TbAQP2-transformed fps1⌬ mutant cells showed a similar phenotype as fps1⌬ mutant cells expressing TbAQP1 and TbAQP3. However, a significant amount of glycerol was retained within the cytosol, regardless of the ionic strength of the medium (Fig. 3C). In summary, the results are consistent with a role of TbAQPs in glycerol efflux. In addition, TbAQP2 appears to affect glycerol production and its intra-and extracellular distribution in S. cerevisiae in a different way as compared with TbAQP1 and -3.

Phenotypes of TbAQPs
To analyze the phenotype of TbAQP1, -2, and -3, respective transformants were grown and then spotted in serial dilutions onto agar plates with different osmolytes (Fig. 4). Fps1 is involved in glycerol transport in S. cerevisiae. This regulated glycerol channel is active under hypotonic conditions but inactive under hypertonic conditions (28). Thus, growth of fps1⌬ mutant cells was limited under hypoosmotic conditions but did not show any phenotype under hyperosmotic conditions (27). TbAQPs expressed in fps1⌬ mutant cells suppressed hypoosmosensitivity. In addition, these cells grew unaffected under hypertonic conditions if the respective osmolyte was 1 M glycerol. Growth was drastically reduced, however, if 1 M sorbitol or 5% NaCl was applied instead of glycerol. This phenotype of hyper-osmosensitivity for other osmolytes than glycerol was also described for yeast mutants lacking either the N-or the C-terminal regulatory domains of Fps1, indicating that TbAQPs are non-regulated glycerol channels (26,27).

Functional Expression of TbAQPs in Xenopus Oocytes
For further functional characterization, TbAQPs were heterologously expressed in Xenopus oocytes.
Glycerol Uptake-Glycerol uptake was measured by incuba-  Fig. 5, control oocytes mediated only little glycerol uptake (57.9 pmol/oocyte within 40 min), consistent with simple membrane diffusion. In contrast, oocytes injected with 10 ng of any of the TbAQPs accumulated glycerol (about 240 pmol in 40 min) rapidly, with very similar transport kinetics. To confirm these results and for comparisons with other aquaglyceroporins, glycerol permeability was tested using the oocyte swelling assay.
The swelling rate in a 130 mM glycerol gradient was similar among all three TbAQPs (Fig. 6) and about 10-fold higher as compared with water-injected control oocytes. This was in good agreement with the result of [ 14 C]glycerol uptake in the same time interval of 1 min (Fig. 5). These results were comparable with those obtained using aquaglyceroporins from other protozoa, e.g. Plasmodium falciparum and Toxoplasma gondii (22,29).
Water Permeability-To analyze water transport, oocytes expressing TbAQPs were exposed to hypotonic shock using a 1:3 diluted medium, which generates an osmotic gradient of 140 mosM/kg across the oocyte membrane (22). Expression of TbAQPs in Xenopus oocytes equally led to a 6 -7-fold increase of the swelling rate as compared with control oocytes (Fig. 7).
Selectivity Profile for Other Solutes-Beside water, aquaglyceroporins usually facilitate permeation of other small uncharged solutes. We thus measured the swelling rates of TbAQP-expressing oocytes in the presence of a variety of poly-ols such as urea, dihydroxyacetone, alanine, and pyruvate. Although permeability for polyols was restricted, erythritol and ribitol led to about half the swelling rate of glycerol in TbAQP3expressing oocytes (Fig. 6). Urea passed the pores of TbAQP2 and -3 but was hardly transported by TbAQP1 (Fig. 6). Surprisingly, dihydroxyacetone (DHA) was excellently transported by all TbAQPs; when compared with glycerol, the transport rate was about 2-fold higher for TbAQP2, 1.5-fold higher for TbAQP1, and about equal for TbAQP3.
Interestingly, a concentration of 0.2 M DHA is toxic for yeast cells (30). In glucose-containing medium, 0.2 M DHA led to a slightly decreased growth rate of the fps1⌬ deletion strain, whereas growth of TbAQP-transformed fps1⌬ was drastically reduced (Fig. 7). These results are consistent with a role of TbAQPs for DHA permeation and stress the importance of aquaglyceroporins for secretion of toxic metabolites.

Transcription of the TbAQP Genes from T. brucei
Northern blot analysis using total RNA isolated from different life cycle stages and TbAQP genes as probe showed a stage-specific regulation of their transcripts. TbAQP1 was slightly expressed in log phase, highly expressed in stationary phase, and virtually the only one expressed in procyclic trypanosomes. TbAQP2 was scarcely expressed throughout the life cycle, whereas transcripts of TbAQP3 were only detectable in bloodstream form parasites (Fig. 8). DISCUSSION The importance of aquaporins can be inferred from their wide distribution throughout nature from bacteria to human. Single cell organisms may possess only one type, like P. falciparum (22), or several of them, e.g. S. cerevisiae. Metazoa usually express numerous isoforms, e.g. 11 AQP isoforms in mammalians (30).
Due to the presence of glycosomes in T. brucei, the energy metabolism of this parasite is unique. Especially important is the sensitivity for glycerol under anaerobic conditions, which can be mimicked by salicylhydroxamic acid. We have cloned TbAQP1, TbAQP2, and TbAQP3. Experimental evidence for the function of TbAQPs as glycerol facilitators was obtained by complementation studies of functionally expressed TbAQPs in an S. cerevisiae fps1⌬ mutant, which suppressed the hypoosmotic sensitivity of this strain. In addition, these transformants showed a hyperosmotic phenotype similar to a yeast fps1⌬ mutant expressing either GlpF or Fps1 with N or C termini deletions (26,27). TbAQP transformants further showed incapacities to accumulate glycerol under isotonic and hypertonic conditions, indicating a glycerol efflux function of TbAQPs. The existence of channels involved in glycerol efflux stresses the role for glycerol release depending on oxygen supply. This may not only be important under anaerobic conditions, but also under a diminution of oxygen, because glycerol production was observed even under aerobic conditions (6).
Participation of TbAQPs in glycerol catabolism was inferred from glycerol influx experiments. As judged from both uptake of radiolabeled glycerol and standard oocyte swelling assays, TbAQPs expressed in Xenopus oocytes transport glycerol efficiently, comparable with AQP3 and AQP9 (31).
Under aerobic conditions, glycerol serves as an alternative substrate (6) and inhibits glucose uptake efficiently (50% at 0.8 mM glycerol (2)); it also competes for ATP (32). Therefore, the glycerol blood concentration of about 50 M could be relevant as an energy source for trypanosomes.
TbAQPs are aquaglyceroporins, which are able to transport water and other small solutes. The P f values of oocytes expressing TbAQPs are in the range of aquaporins with intermediate water permeability, such as T. gondii (29). It is generally assumed that T. brucei live under constant conditions in blood and are thus not exposed to osmotic stress. However, it is well known that erythrocytes possess a high density of AQP1 in their plasma membrane, which shows a high capacity for water transport. This aquaporin seems to be fundamental to respond to abrupt changes in extracellular osmolarity, e.g. when blood cells travel through the renal medulla (33). This may also be important for trypanosomes within their mammalian host. However, osmoregulation may become pivotal for survival of trypanosomes also during the course of transmission and during their life within the insect vector.
Transport of other solutes by TbAQPs was restricted (Fig. 6). Surprisingly, uptake of DHA in oocytes expressing TbAQPs was similar to or even better than glycerol. This result was confirmed by experiments with TbAQP-transformed fps1⌬ mutant yeast. So far, the physiological relevance of the DHA transport is unknown, but a simple explanation would be that DHA may be used as energy source. However, this seems not to be the case since detailed searches of the trypanosome gene data bank (TIGR and Sanger) for enzymes related to DHA consumption (i.e. glycerol dehydrogenase and dihydroxyacetone kinase) gave negative results. Moreover, trypanosomes FIG. 8. Northern blot analysis of bloodstream form (BSF) and procyclic form trypanosomes. Total RNA was isolated from the logarithmically growing bloodstream form, the stationary phase bloodstream form, and logarithmically growing procyclic forms and then separated on an agarose gel (15 g/lane of each). After blotting, membranes were first hybridized with probes corresponding to the different TbAQPs, stripped thereafter, and reprobed with a ␤-tubulin sequence as a control for loading. incubated in buffer containing different DHA concentrations as the sole energy source did not survive (data not shown). In literature, DHA has been related to anaerobic glycerol metabolism and osmoregulation, but cell toxicity has also been described (30,34,35).
The expression profile of TbAQP transcripts suggests a distinct importance of the respective proteins throughout the life cycle. We interpret these data as follows. Although TbAQP3 seems to be the main AQP in the logarithmically growing slender bloodstream form, the procyclic form relies on the expression of TbAQP1. Stationary phase trypanosomes, equivalent to stumpy bloodstream forms, still express some TbAQP3 but also a huge amount of TbAQP1. This may reflect either the onset of the differentiation process or, more likely, the specific need for aquaglyceroporins during the transmission process. TbAQP2 is scarcely expressed in all three life stages examined. Since this AQP shows significant changes in the NPA and the charged amino acid motifs and thus the formation of the pore, it may be a candidate for an organelle localization. Alternatively, it could be expressed only under defined stress conditions.
In conclusion, we report the characterization of three aquaglyceroporins from T. brucei. These proteins constitute channels with a high glycerol and an intermediate water permeability, which may be mainly involved in glycerol uptake and release and in osmoregulation. In addition, their transcripts seem to be regulated in a stage-specific fashion. We also studied permeability of TbAQPs for DHA, which was in the same range (TbAQP3) or even more prominent as the glycerol transport (TbAQP1 and -2). Since the metabolic function of DHA in T. brucei has not been experimentally addressed yet, experiments are underway to improve our understanding of the relevance of this metabolite in the biology of trypanosomes.