Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The movement of fluid between the airspace and vascular compartments in lung plays an important physiologic role in a number of processes, including the maintenance of proper airway hydration, the reabsorption of alveolar fluid in the neonatal period in preparation for alveolar respiration, and the resolution of pulmonary edema (1). The general paradigm for the transport of fluid across epithelia in lung is that active salt transport drives osmotic water movement. Recent data indicate that alveolar and airway epithelia in lung are highly water-permeable (4). Type I alveolar epithelial cells, which compose the majority of alveolar surface area, appear to have the highest plasma membrane water permeability of any mammalian cell membrane studied to date (7).

Several molecular water channels (aquaporins, AQP) are expressed in lung (reviewed in [8] and [9]): AQP1 (channel-forming integral protein of 28 kD) in alveolar endothelia and some pneumocytes (6, 10), AQP4 (mercurial insensitive water channel) at the basolateral membrane of airway epithelia (13, 14), and AQP5 at the apical membrane of type I alveolar epithelial cells (15). The aquaporins are structurally related membrane-spanning proteins of ~ 30 kD. Expression of aquaporins 1, 4, and 5 is strongly increased in the perinatal period (16, 17) in parallel to increased water permeability of the airspace-capillary barrier (18). Water channels have not been identified at the basolateral membrane of alveolar epithelia or the apical membrane of airway epithelia, each of which probably has high water permeability. Outside of the lung, there are many examples of cell membranes that are likely to be highly water-permeable but do not express any of the known water channels, such as the basal membranes of choroid plexus and ciliary body. Whether aquaporin-type water channels provide the primary route for water movement across most water-permeable cell membranes is unknown. It is noted that water can also move to a limited extent through membrane lipids, and across various membrane transport proteins that serve other functions, including the GLUT1 glucose transporter (19, 20), the GLUT5 Na⁺-coupled glucose transporter (21), and the cystic fibrosis transmembrane conductance regulator (CFTR) cystic fibrosis Cl-channel (22).

The purpose of this study was to test the hypothesis that T1 is a water transporter or a regulator of aquaporin-type water channels. T1 is a protein of unknown function that was cloned from a rat-lung complementary DNA (cDNA) library using a monoclonal antibody specific for type I cells (23). Sequence analysis of rat T1 indicated that a mouse homolog had been cloned several years earlier from an osteoblast cell line (OTS-8 [24]) and a medullary thymic epithelial cell line (gp-38 [25]). Hydropathy analysis suggested that T1 is an integral membrane protein in which each monomeric unit contains a single hydrophobic membrane-spanning domain. Immunocytochemistry indicated that T1 is expressed strongly at the apical membrane of type I alveolar epithelial cells, in choroid plexus, and in ciliary body epithelium (26). Because each of these membranes is highly water permeable, it was proposed that T1 may function as a water channel. Because of significant sequence differences in cloned rat and mouse T1 protein (see DISCUSSION), we cloned two human T1 isoforms and then expressed each of the T1 isoforms in Xenopus oocytes for measurement of osmotic water permeability (P_f). Effects of protein kinase A and C activation were examined because some T1 protein sequences contain consensus sites for phosphorylation, and phosphorylation has been proposed to regulate several water channels (27, 28). In addition, we examined the role of T1 as a regulator of AQP5, a water channel expressed with T1 in type I cells, and AQP1, a water channel expressed in ciliary body and choroid plexus.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Cloning of Human T1 cDNAs

A human lung gt11 cDNA library (HL1041b; Clontech, Palo Alto, CA) was screened with a 501-base pair (bp) rat T1 coding region probe labeled with [-³²P]dCTP (Rediprime DNA labeling kit; Amersham, Arlington Heights, IL). One strongly positive clone was obtained and purified from 10⁶ screened plaques. Phage DNA was prepared using a lambda Mini kit (QIAGEN, Valencia, CA). A 1.2-kb insert (hT1-1) was released with EcoRI and subcloned into plasmid pGEM3Zf(+) (Promega). A second isoform was cloned using the hT1-1 coding sequence to search the dbEST database. Fifteen overlapping sequences from various human tissues were recovered. Two sequences (AA026967 and AA468871) were purchased from Research Genetics (Huntsville, AL) and cDNAs were sequenced from both orientations by the dideoxy chain termination method (U.S. Biochemical sequenase kit, version 2.0; Cleveland, OH).

Northern Blot Analysis of Human T1

A human multiple-tissue Northern blot membrane (Origene, Rockville, MD) was probed with ³²P-labeled coding sequences of T1-1 and T1-2 at 68°C in rapid hybridization buffer (Amersham) for 1 h, then washed twice in 0.2× standard saline citrate, 0.1% sodium dodecyl sulfate (SDS) at 68°C, each for 30 min. The membrane was autoradiographed with Hyperfilm-MP (Amersham) with intensifying screen for 16 h at 80°C.

Constructs for Translation and Xenopus Oocyte Expression

The longest open reading frame of hT1-1 cDNA (492 bp) was amplified by reverse transcription-polymerase chain reaction (RT-PCR) using human lung cDNA as template and primers: sense, 5'-GAAGATCTATGTTACATATTTTATCACCGATC-3'; antisense, 5'-GACTAGTTCAAATTCCTGGGCACAAGTGAACC-3' with engineered BglII and SpeI restriction sites (underlined). After restriction digestion, the hT1-1 coding-region fragment was subcloned at BglII/XbaI sites into plasmid pSP64T, which contains a 60-bp Xenopus -globin translation enhancer sequence downstream from the SP6 promoter. The coding sequence of hT1-2 was subcloned similarly using primers: sense, 5'-GAAGATCTATGTGGAAGGTGTCAGCTCTG-3'; antisense 5'-GCTCTAGATTAGGGCGAGTACCTTCCCGAC-3' with engineered BglII and XbaI sites underlined. The rat T1 (rT1) cDNA coding sequence (23) was PCR-amplified using rat-lung cDNA as template and primers: sense, 5'-GAAGATCTATGTGGACCGCGCCAGTGTTGCTC-3'; antisense, 5'-GACTAGTTTAGGGCGAGAACCTTCCAGAAATC-3'; the mouse T1 (mT1) cDNA coding sequence (24) was PCR-amplified using mouse-lung cDNA as template and primers: sense, 5'-GAAGATCTATGTGGACCGTGCCAGTGTTGTTC-3'; antisense, 5'-GCTCTAGATTATCTTCCTCCACAGGAAGAGGA-3'. The rT1 and mT1 cDNA coding sequences were subcloned into pSP64T at BglII/XbaI sites. The PCR reactions were performed using a High Fidelity PCR kit (Boehringer, Indianapolis, IN). The entire insert region of each construct was found by sequence analysis to be identical to that reported previously for the rat and mouse T1 isoforms. For epitope tagging, a c-myc epitope was fused with the N-terminus of T1. A 33-bp DNA sequence (5'-ATG GAA CAA AAG CTG ATT TCT GAA GAA GAC CTG) encoding a 10 amino-acid c-myc epitope tag (amino acids EQKLISEEDL) and a translation initiation codon was introduced into the pSP64T vector immediately downstream from the globin enhancer sequence. Each of the T1 coding sequences was ligated downstream and in-frame with the c-myc coding sequence.

RNA Transcription

Complementary RNA (cRNA) was transcribed in vitro using SP6 polymerase (BRL Life Technologies, Gaithersburg, MD) and 2 µg of plasmid DNA (linearized at EcoRI site downstream of each T1 insert) in a 100-µl volume at 37°C for 1 h in the presence of diguanosine triphosphate (1 A₂₅₀ unit; Pharmacia, Piscataway, NJ). Plasmid DNA was digested with RNase-free DNase (Promega), extracted with phenol-chloroform, and precipitated twice in ethanol. The cRNA was suspended in diethylpyrocarbonate-treated water for cell-free translation and oocyte injection. For studies of T1-aquaporin interactions, cRNAs encoding rat AQP1 and AQP5 were prepared as described previously (29).

Cell-free Translation

In vitro transcribed cRNA was added to a rabbit reticulocyte lysate mixture containing [³⁵S]methionine for 1 h at 23°C. Microsomal membranes prepared from dog pancreas were added in some experiments at the start of translation to a final concentration of 8 OD₂₈₀. Samples were resolved on a 12% polyacrylamide gel, dried, and autoradiographed as described elsewhere (30).

Oocyte Expression and Water Transport Assay

Stage V and VI oocytes from Xenopus laevis were isolated and defolliculated with collagenase (type 1A; Sigma, St. Louis, MO; 1 mg/ml for 2-4 h at 20°C) in Barth's buffer (200 mOsm). Oocytes were microinjected with 50-nl samples of cRNA (0-5 µg/µl) encoding human, rat, or mouse T1, or rat AQP1 or AQP5, or combinations (see RESULTS), and incubated at 18°C for 24-27 h. Osmotic water permeability (P_f) was measured from the time course of oocyte swelling at 10°C in response to a 5-fold dilution of the extracellular Barth's buffer with distilled water (31). In some experiments, cyclic adenosine monophosphate-dependent protein kinase was simulated in oocytes by a 15- 30-min incubation in 10 µM forskolin at 23°C, or protein kinase C was stimulated by a 15-30-min incubation with 100 nM phorbol myristate acetate (PMA) at 23°C. Oocyte P_f was calculated from the initial rate of swelling, d(V/V₀)/ dt, by the relation

where V/V₀ is relative oocyte volume, S/V₀ is surface-to-volume-ratio (50 cm¹), V_w is the partial volume of water (18 cm³/mol), and the Osm_out  Osm_in is the osmotic gradient difference.

Immunoprecipitation

After microinjection of cRNA encoding c-myc tagged T1 or aquaporin constructs, groups of ten oocytes were incubated for 24 h at 18°C in 100 µl Barth's buffer containing 50 µCi [³⁵S]methionine. Oocytes were washed in ice-cold Barth's buffer and plasma membranes were isolated by microdissection (29). Membrane pellets were solubilized in 1 ml phosphate-buffered saline (PBS) (pH 7.4) containing 100 mM -octylglucoside for 1 h and incubated with protein A-Sepharose CL-4B beads (Pharmacia Biotech) for 1 h at 4°C. The beads were pelleted at 10,000 × g. The supernatant was incubated with primary c-myc antibody (1:300) for 1 h and then a rabbit antimouse secondary antibody (BRL) for 1 h. The c-myc antibody (19E10) was generated as ascites in mice using hybridoma cells purchased from American Type Culture Collection (Rockville, MD). Protein A-Sepharose CL-4B beads were added, pelleted, and washed 4 times with PBS (pH 7.4) containing 0.1% SDS, 1% deoxycholate, and 1% Triton-X100. The proteins were released from the beads in 30 µl of SDS loading buffer and resolved on 12% SDS-polyacrylamide gel electrophoresis (PAGE). Gels were soaked in 15% (wt/vol) 2,5-diphenyloxazole in dimethylsulfoxide for 30 min, dried, and exposed to hyperfilm (Amersham) at 70°C for 3-10 d.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Figure 1 presents the cDNA and deduced amino-acid sequences of the two human isoforms of T1 (hT1-1 and hT1-2). The longest open reading frame of hT1-1 encoded a 164 amino-acid protein (Figure 1A). There were two in-frame ATG start codons (seven residues apart) with a stop codon at 18 bp upstream from the first ATG. The predicted amino-acid sequence of hT1-1 contained one consensus sequence for phosphorylation by protein kinase A and no consensus sites for N-linked glycosylation. The longest open reading frame of hT1-2 encoded a 162 amino-acid protein. The predicted amino-acid sequence of hT1-2 contained one consensus site for phosphorylation by protein kinase A and one site for protein kinase C (Figure 1B).

View larger version (72K): [in this window] [in a new window]   Figure 1.   cDNA and deduced amino-acid sequence of two isoforms of human T1. (A) hT1-1. (B) hT1-2. Consensus sites for phosphorylation by protein kinase A are circled. For hT1-1, an upstream in-frame stop codon (at nucleotide-18) is underlined. For hT1-2, the consensus site for phosphorylation by protein kinase C is indicated by #.

Figure 2A shows an alignment of deduced amino-acid sequences of the hT1 isoforms with reported sequences for the mouse and rat T1 isoforms. The sequences for mouse and rat T1 were confirmed here (see MATERIALS AND METHODS). A highly conserved 29-residue span of nonpolar region is seen, with 79-90% amino-acid sequence identity. Sequence comparison of the rat and mouse isoforms showed 78% amino-acid identity up to an early stop codon in the rat sequence (23). The hT1-1 sequence differed remarkably from the mouse and rat isoforms outside of the conserved region. The amino-acid sequence of hT1-2 is 50% identical to rat T1 and 39% identical to mouse T1. The two human T1 isoforms share an identical 38-amino-acid sequence around a putative transmembrane region. There were four consensus sequences for phosphorylation by protein kinase C in the N-terminus polar segments in both the mouse and rat isoforms; hT1-2 has only one in this region. A consensus sequence for phosphorylation by protein kinase A was conserved at an identical site in the C-terminus polar region of the four isoforms; the mouse isoform contained an additional protein kinase A consensus sequence distally. Only the mouse isoform had consensus sequences for N-linked glycosylation. Figure 2B shows hydropathy profiles for human, mouse, and rat T1 isoforms. The highly conserved nonpolar region found in each isoform suggests a membrane-spanning domain. The nonconserved flanking regions are considerably more polar and probably represent extramembrane domains.

View larger version (42K): [in this window] [in a new window]   Figure 2.   (A) Alignment of deduced amino-acid sequences of human, rat, and mouse T1 isoforms. Sequences for rat and mouse isoforms were taken from Rishi and coworkers (23) and Nose and associates (24), respectively. Conserved residues are shown in bold. Consensus sites are indicated for phosphorylation by protein kinase C (underlined), protein kinase A (double underlined), and N-linked glycosylation (asterisk). The highly conserved nonpolar region is boxed. (B) Kyte-Doolittle hydropathy plots (window 11) of human, rat, and mouse T1 isoforms.

Cell-free translation using rabbit reticulocyte lysate and endoplasmic reticulum-derived microsomes was done to determine protein size and glycosylation. Figure 3A shows that translation of rat T1 gave a single nonglycosylated protein band at ~ 18 kD in the absence of microsomes, consistent with predicted protein size. Translation of mouse T1 gave a single protein of molecular size ~ 25 kD in the absence of microsomes, somewhat greater than its predicted size of 21.5 kD, which is probably related to anomalous mobility on SDS-PAGE. An additional band at ~ 27 kD in the presence of microsomes indicated N-linked glycosylation as confirmed by its absence when translation was done in the presence of AcAsn-Tyr-Thr, a tripeptide inhibitor of N-linked glycosylation (not shown). Translation of hT1-1 gave a single band at ~ 13 kD without glycosylation, which also appeared to migrate anomalously.

View larger version (30K): [in this window] [in a new window]   Figure 3.   (A) Cell-free translation of T1 isoforms. Translation was carried out using rabbit reticulocyte lysate in the absence and presence of endoplasmic reticulum-derived microsomes as described in MATERIALS AND METHODS. (B) Northern blot of human-tissue mRNAs (2 µg/lane) probed at high stringency with 237 bp of the coding regions of hT1-1 (top) and hT1-2 (bottom).

Northern blot analysis showed expression of hT1-1 transcript in human lung >> skeletal muscle > heart (Figure 3B, top). hT1-1 transcripts of size 2.6 and 1.4 kb were observed. RT-PCR amplification of the hT1-1 coding sequence using lung and skeletal muscle human cDNA templates and primers flanking the coding sequence gave single bands of ~ 0.5 kb as predicted (not shown). A single hT1-2 transcript of size 1.2 kb was detected in human lung (Figure 3B, bottom).

Functional measurements of water permeability were made in Xenopus oocytes from the rate of oocyte swelling in response to an osmotic gradient. Figure 4A shows representative oocyte swelling curves after a 5-fold dilution of the extracellular buffer with distilled water. Water permeability in oocytes expressing aquaporin water channels AQP1 and AQP5 was strongly increased over that in control (water-injected) oocytes, whereas water permeability in oocytes expressing hT1 was not increased above control.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Assay of T1 water permeability. (A) Representative data showing the time course of swelling in oocytes microinjected with water or 5 ng of cRNA encoding hT1 isoforms AQP1 or AQP5. The constructs did not contain the c-myc epitope tag. Measurements were made at 10°C after a 24- to 36-h incubation at 18°C. (B) Summary of osmotic water permeability coefficients (Pf, mean ± SE, n = 8-10, 10°C) for oocytes microinjected with the indicated cRNAs. As indicated, oocytes were incubated with 10 µM forskolin or 100 nM PMA at 23°C for 15-30 min prior to water permeability assay.

Figure 4B summarizes water-permeability measurements from a large group of oocytes expressing one of the T1 proteins or the aquaporins, or T1 together with an aquaporin. The P_f in oocytes microinjected with cRNAs encoding each of the T1 isoforms was not increased significantly over that in control oocytes. Oocyte P_f was not stimulated by activation of protein kinase A or C. As a positive control, P_f was increased strongly in oocytes microinjected with cRNAs encoding AQP1 or AQP5; the magnitude of the increase in P_f was related to the amount of microinjected cRNA. The P_f in oocytes coinjected with cRNAs encoding one of the T1 isoforms together with AQP1 or AQP5 was not different from P_f in oocytes injected with equivalent amounts of AQP1 or AQP5 alone. No significant effect of T1 expression on aquaporin function was found even when a considerable excess of T1 cRNA was coinjected with the aquaporin cRNA. These results suggest that T1 is involved in neither the formation nor the regulation of a water-transporting pathway.

Immunoprecipitation of oocyte plasma membranes was done to confirm that T1 protein is expressed at the oocyte plasma membrane and to establish an upper limit to the T1 intrinsic water permeability. To immunoprecipitate the T1 and aquaporin proteins with similar efficiencies, each of the proteins was epitope-tagged with c-myc at its N-terminus. Immunostaining (with c-myc antibody) of oocytes microinjected with cRNA encoding c-myc-hT1 indicated a plasma membrane expression pattern, as expected (not shown). Figure 5 shows an autoradiogram of [³⁵S]methionine-labeled proteins from oocyte plasma membranes that were immunoprecipitated with c-myc antibody. Immunoprecipitated proteins of the sizes found by cell-free translation (Figure 3A) were seen for oocytes expressing mouse and human T1, as well as the expected sizes for AQP1 and AQP5. No proteins were immunoprecipitated in control oocytes. To estimate the amount of expressed T1 relative to AQP1, autoradiograms were analyzed by quantitative densitometry to compute background-subtracted integrated band intensities. Relative band intensities were: 1.0 (AQP1), 0.96 (AQP5), 0.3 (hT1-1), and 2.4 (mouse T1). Since the single-channel water permeability of AQP1 has been measured (32), the ratio of oocyte P_f to band intensities was used to establish an upper limit to T1 single-channel water permeability (see DISCUSSION).

View larger version (69K): [in this window] [in a new window]   Figure 5.   Immunoprecipitation of proteins from microdissected plasma membranes. Oocytes were microinjected with 5 ng of cRNAs encoding each of the constructs (with N-terminus c-myc epitope tag). An autoradiogram of [35S]methionine-labeled proteins is shown (see MATERIALS AND METHODS).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The purpose of this study was to test the hypothesis that T1 functions as a water transporter or as a regulator of aquaporin-type water channels. This hypothesis was motivated by reports that T1 is an integral membrane protein of unknown function that is expressed strongly at the apical plasma membrane of type I alveolar epithelial cells and several other highly water-permeable membranes (23, 26). In addition, we found recently that type I cells have extremely high osmotic water permeability (7), suggesting the involvement of water-transporting protein(s) in addition to AQP5. Because of substantial differences of reported sequences for the mouse and rat forms of T1, two human T1 isoforms were cloned here and analyzed. The deduced amino-acid sequence of T1-1 had high homology to mouse and rat T1 in a putative membrane-spanning region. The two hT1 isoforms share an identical 38-amino-acid sequence around the putative membrane-spanning region, suggesting that they are derived from a single gene by alternative messenger RNA (mRNA) splicing. Heterologous expression studies in Xenopus oocytes indicated that T1 is not a water-transporting protein, a regulated water-transporting protein, or a regulator of aquaporin-type water channels. Although the cDNA cloning and functional experiments here do not define the biological role(s) of T1, they provide direct evidence against its proposed role in membrane water transport. In addition, the cloning of human T1 isoforms should permit the assay of T1 in human airspace fluid as a marker of type I cell injury.

Comparison of the deduced amino-acid sequences of T1 isoforms from mouse, rat, and human indicate a highly conserved nonpolar span of 29 amino acids flanked by nonconserved, more polar sequences. An interesting finding was that the predicted sizes of the T1 proteins from mouse, rat, and human differed remarkably. The variability of the flanking sequences and the cloning of the two isoforms of human T1 suggest that the T1 isoforms cloned to date are members of a superfamily containing multiple isoforms resulting from alternative splicing and/ or gene duplication. The conserved nonpolar sequence suggests that T1 is a membrane protein, as confirmed by immunocytochemistry (26), and that each T1 monomer spans the bilayer once. If T1 were to function as a water channel containing an aqueous pathway, assembly in higher- order oligomers would probably be necessary, as was the case for a new type of chloride channel in which each monomer spans the membrane only once (33). T1 oligomer formation would not be required if it were to function as a regulator of aquaporin-type water channels.

Water permeability in Xenopus oocytes expressing each of the T1 isoforms was not significantly increased above the baseline water permeability measured in control (water-injected) oocytes. The water permeability assay utilizes quantitative imaging and has the sensitivity to detect an incremental water permeability coefficient (P_f) of ~ 2 × 10⁴ cm/s (31). T1 expression at the oocyte plasma membrane was confirmed by immunoprecipitation of c-myc-tagged T1 fusion constructs with a monoclonal c-myc antibody. By comparing T1 expression with that of c-myc-tagged AQP1, which has a measured single-channel water permeability of 6 × 10¹⁴ cm³/s (32), an upper limit to the T1 single channel (per molecule) water permeability of 2 × 10¹⁶ cm³/s was computed. This value is too low for T1 to be a physiologically important water channel because its density would need to be unreasonably high (> 50,000 T1 molecules per square micron) to increase membrane water permeability appreciably (P_f > 0.01 cm/s).

Phosphorylation has been proposed to be a modulator of the function of several water channels. The best-documented system is plant water channel -TIP, where protein kinase A-mediated phosphorylation is associated with a 2-fold increase in water permeability (34). Water permeability of the vasopressin-regulated water channel AQP2 in oocytes was reported to increase by 20-30% after protein kinase A-mediated phosphorylation (28); however, similar oocyte expression studies by our group (35), and measurements of water permeability in control and in vitro phosphorylated endosomes (36), did not confirm these findings. Recent experiments suggest that a more likely role for AQP2 phosphorylation is in the regulation of AQP2 trafficking in mammalian cells (37). Recently, it was reported that protein kinase A-mediated phosphorylation in oocytes expressing AQP1 results in increased water permeability and the appearance of a de novo ion conductance (27). However, these findings could not be repeated by several laboratories (38, and others) and remain controversial. Because the deduced T1 amino-acid sequences contained phosphorylation consensus sites in polar amino-acid regions, we tested whether phosphorylation of T1 by protein kinase A or C might increase its intrinsic water permeability. Under conditions in which substantial kinase effects were observed in oocytes, such as activation of CFTR Cl-channels (22), no effect on T1 water permeability was found.

After excluding the proposed role as a water-transporting protein, the hypothesis was tested that T1 functions as a regulator of aquaporin-type water channels. AQP1 and AQP5 were studied because T1 is coexpressed with AQP5 at the apical membrane of type I alveolar epithelial cells (14), and with AQP1 in choroid plexus and ciliary body epithelium (11, 12). Experiments were done in Xenopus oocytes expressing comparable amounts of each of the T1 isoforms, and AQP1 or AQP5, and in oocytes expressing each aquaporin together with an excess of T1 protein. In each case, the increased oocyte water permeability conferred by aquaporin expression was not affected by the coexpression of T1. Therefore, assuming that functional measurements of water permeability made in Xenopus oocytes apply to mammalian cells, our results do not support a significant role of T1 in the formation or regulation of a water-transporting pathway.