Specific Inhibition of AQP1 Water Channels in Isolated Rat Intrahepatic Bile Duct Units by Small Interfering RNAs*

Cholangiocytes express water channels ( i.e. aquaporins (AQPs)), proteins that are increasingly recognized as important in water transport by biliary epithelia. How-ever, direct functional studies demonstrating AQP-me-diated water transport in cholangiocytes are limited, in part because of the lack of specific AQP inhibitors. To address this issue, we designed, synthesized, and utilized small interfering RNAs (siRNAs) selective for AQP1 and investigated their effectiveness in altering AQP1-mediated water transport in intrahepatic bile duct units (IBDUs) isolated from rat liver. Twenty-four hours after transfection of IBDUs with siRNAs targeting two different regions of the AQP1 transcript, both AQP1 mRNA and protein expression were inhibited by 76.6– 92.0 and 57.9–79.4%, respectively. siRNAs containing the same percent of base pairs as the AQP1-siRNAs but in random sequence ( i.e. scrambled siRNAs) had no effect. Suppression of AQP1 expression in cholangiocytes re-sulted in a decrease in water transport by IBDUs in response to both an inward osmotic gradient (200 mos M ) or a secretory agonist (forskolin), the osmotic water permeability coefficient ( P f ) decreasing up to 58.8% and net water secretion ( J v ) decreasing up to 87%. A strong cor- relation between AQP1 protein expression and water transport in IBDUs transfected with AQP1-siRNAs was consistent with the decrease in water transport from AQP1 gene silencing by AQP1-siR-NAs. AQPs in direct of the contribution to biliary

Intrahepatic bile duct epithelial cells (i.e. cholangiocytes) play an essential role in bile formation, and by integrated absorptive and secretory processes, they contribute up to 40% of daily bile production in humans (1,2). Because bile is a complex fluid composed of Ͼ98% water, cholangiocytes like other water-transporting epithelial cells are required to rapidly transport large amounts of water in response to osmotic gradients generated by transported ions and solutes, a situation in which specific water channel proteins (i.e. aquaporins (AQPs)) 1 are probably involved (3)(4)(5)(6). Indeed, our initial observation (7) that isolated rat cholangiocytes are capable of rapid mercurysensitive, temperature-independent transmembrane water transport in response to osmotic gradients was consistent with transport via water channels rather than by diffusion through the lipid bilayer. Our more recent molecular studies have demonstrated that rat cholangiocytes express six AQPs (i.e. AQP 0, 1, 4, 5, 8, and 9) from the known 11 AQPs in mammals (5,(7)(8)(9)(10)(11). Moreover, at least two of them (i.e. AQP1 and AQP4) contribute to the water permeability of both the apical and basolateral cholangiocyte membrane domains, AQP1 facilitating mainly the apical transport of water and AQP4 modulating its basolateral movement (7)(8)(9)11). Nevertheless, direct studies of the contribution of AQPs to water transport in intrahepatic bile ducts and other tissues have been severely hampered by the lack of specific AQPs inhibitors.
Recently, several groups have described post-transcriptional gene silencing or RNA interference in a wide variety of organisms using double-stranded RNAs of ϳ200 -1000 nucleotides in length that specifically suppress the expression of a target mRNA (reviewed in Refs. [12][13][14][15][16][17][18]. According to the prevailing model, double-stranded RNA is processed into small interfering double-stranded RNAs (siRNAs) of 19 -25 nucleotides in length, which act as guides for the RNA-induced silencing enzymatic complex required for the cleavage of the target mRNAs (15)(16)(17)(18). Although the physiological significance of post-transcriptional gene silencing and RNA interference is still under study, powerful new technology for selective inhibition of specific gene expression employing siRNAs is rapidly evolving (19 -26).
In this study, we present data demonstrating that AQP1 gene expression in cholangiocytes is specifically suppressed by AQP1-siRNAs, resulting in a significant decrease of water transport by this cell type. These data show the feasibility of utilizing siRNAs to specifically reduce the expression of AQPs in epithelial cells and provide direct evidence of the contribution of AQP1 to water transport in biliary epithelia.

EXPERIMENTAL PROCEDURES
Materials-All of the chemicals were of highest purity commercially available and were purchased from Sigma unless otherwise indicated.
Animals-Male Fisher 344 rats (225-250 g) were obtained from Harlan Sprague-Dawley (Indianapolis, IN), housed in temperaturecontrolled room (22°C) with 12-h light-dark cycles, and maintained on a standard diet with free access to water. All of the experimental procedures were approved by the Animal Use and Care Committee of the Mayo Foundation.
Solutions-The composition of isotonic (290 mosM) Ringer-HCO 3 buffer was (in mM): 120.0 NaCl, 5. AQP1-siRNAs Design, Synthesis, and Labeling-Sequence information regarding mature rat AQP1 mRNA was extracted from the NCBI Entrez nucleotide data base. Two target sites within AQP1 gene were chosen from the rat AQP1 mRNA sequence (GenBank TM accession NM_012778). Following selection, each target site was searched with NCBI BlastN to confirm specificity only to AQP1. Two different siRNAs designated AQP1-siRNA(a) and AQP1-siRNA(b), which target nucleotides 71-91 and 673-693 of the rat AQP1 mRNA sequence, respectively, and two nonspecific siRNA duplexes containing the same nucleotides but in irregular sequence (i.e. scrambled AQP1-siRNA(a) and AQP1-siRNA(b)) were prepared by a transcription-based method using the Silencer siRNA construction kit (Ambion, Austin, TX) according to manufacturer's instructions. The 29-mer sense and antisense DNA oligonucleotide templates (21 nucleotides specific to AQP1 and 8 nucleotides specific to T7 promoter primer sequence 5Ј-CCTGTCTC-3Ј) were synthesized by the Mayo Molecular Core facility. One of the constructed siRNAs, AQP1-siRNA(b), was labeled with Cy3 following manufacturer's instructions (Ambion). The efficacy of AQP1-siRNA labeling with Cy3 was estimated by acrylamide gel analysis of the Silencer siRNAlabeling positive control experiment (Ambion) and found to be ϳ20%.
IBDUs Isolation and Transfection with siRNAs-IBDUs, which are portions of intrahepatic bile ducts ranging in luminal diameter from 100 to 125 m and in length from 0.6 to 1.2 mm, were isolated from normal rat liver as we described previously (27). IBDUs were cultured from 0 to 24 h in normal rat cholangiocyte medium containing 10 nM AQP1-siRNAs or corresponding scrambled AQP1-siRNAs. Exogenous delivery of siRNAs to cholangiocytes was carried out with or without a lipid carrier (i.e. TransMesenger TM transfection reagent (Qiagen, Valencia, CA)).
Fluorescence Analysis of the siRNAs Uptake by IBDUs-IBDUs were incubated in normal rat cholangiocyte medium with 0, 0.5, 1, 5, 10, and 20 nM Cy3-AQP1-siRNA(b) with or without a lipid carrier for 24 h at 37°C. AQP1-siRNA(b)-Cy3 then was visualized in IBDUs by fluores-cent microscopy, and Cy3 fluorescence was measured by a method proposed for analysis of fluorescent antisense oligonucleotides (28). After the incubation, IBDUs were washed three times with PBS and then lysed in 200 l of a lysis buffer. Total cellular Cy3-AQP1-siRNA(b) fluorescent emission ( ex ϭ 552; em ϭ 568 nm) was measured with a PerkinElmer LS 55 luminescence spectrophotometer. An aliquot of the cell lysate was taken to measure the amount of total protein using the fluorescence assay, and the Cy3 fluorescence was normalized to 100 g of total protein. For visualization of AQP1 suppression in IBDUs by AQP1-siRNA, IBDUs were incubated with 10 nM Cy3-AQP1-siRNA(b) for 0, 12, and 24 h at 37°C on poly-L-lysine-treated chamber slides. Following the incubation, the IBDUs were fixed with cold 100% methanol for 5 min and air-dried. The slides were then washed three times with 1ϫ PBS and permeabilized in 0.2% Triton-PBS for 2 min at room temperature. IBDUs were blocked for 20 min in blocking buffer (10% normal sheep serum, 0.05% Tween 20 in PBS) at room temperature and incubated with affinity-purified AQP1 antibody (Alpha Diagnostics, San Antonio, TX) at a 1:50 dilution in blocking buffer overnight at 4°C. Following the primary antibody incubation, the IBDUs were washed with 1ϫ PBS three times and incubated at a 1:100 anti-goat IgG fluorescein isothiocyanate conjugate (Sigma) secondary antibody for 1 h at room temperature. AQP1-siRNA(b)-Cy3 and AQP1 fluorescence in IBDUs was then determined by using scanning laser confocal microscopy keeping the pinhole and detector gain setting identical while analyzing the different IBDUs and quantified using LSM 510 Image Examiner software (Carl Zeiss, Thornwood, NY). (Ambion) added as a co-precipitant and stored at room temperature for 5 min. After the addition of 0.1 ml of 1-bromo-3-chloro-propane, the samples were vigorously shaken, incubated for 15 min at room temperature, and centrifuged at 12,000 ϫ g for 15 min at 4°C. The aqueous phase was transferred to a new tube, and 0.5 ml of isopropyl alcohol was added. The samples were mixed, stored for 10 min, and centrifuged at 12,000 ϫ g for 15 min at 4°C. After removing the supernatant, the RNA pellet was washed with 75% EtOH and repelleted by centrifugation at 12,000 ϫ g for 15 min at 4°C. RNA was resuspended in RNA Secure solution (Ambion), and the concentration and purity were determined by spectroscopy. Quantitation of AQP1 message was accomplished by real-time PCR. A standard curve to AQP1 was generated by amplifying AQP1 from freshly isolated rat cholangiocyte cDNA using rat AQP1specific primers (sense, 5Ј-AGTTGAGCACCAGGCATCC-3Ј, and antisense, 5Ј-CACTGATGTGACCCACACTTTG-3Ј). The 259-bp amplicon was electrophoresed on a 1.5% agarose gel, visualized with ethidium bromide, and gel-extracted. The amplicon was subsequently diluted and used as template for the PCR reaction. One-step reverse transcriptase-PCR for AQP1 and 18 S ribosome (Ambion) was performed using the LightCycler RNA Master SYBR Green I kit according to manufacturer's instructions.
Protein Analysis by Western Blotting-Total lysates were obtained by lysing the IBDUs with M-PER TM mammalian protein extraction reagent (Pierce). The lysates were heated to 60°C for 10 min in sample buffer containing 0.8 M dithiothreitol and 10% SDS for protein denaturation and solubilization. The samples were then subjected to electrophoresis through a 12% SDS-polyacrylamide gel and transferred overnight to a nitrocellulose membrane. The blots were blocked with 5% (w/v) nonfat dry milk and 0.2% (v/v) Tween 20. After blocking, the blots were incubated with affinity-purified rabbit anti-rat antibodies to AQP1 (Alpha Diagnostics) at a dilution of 1:1000 overnight at 4°C. The blots were washed and incubated for 1 h with a horseradish peroxidaseconjugated goat anti-rabbit immunoglobulin (1:2000 dilution) at room temperature. Protein bands were detected using an enhanced chemiluminescence detection system (ECL Plus, Amersham Biosciences). After exposing the nitrocellulose membranes to Kodak X-Omat AR film, the autoradiographs were scanned and quantified by densitometry using Molecular Analyst software (Bio-Rad).
Measurement of Water Movement across Intrahepatic Biliary Epithelia-IBDUs were perfused with 1 mM of the impermeable volume marker, fluorescein sulfonate (fluorescein-5(6)-sulfonic acid trisodium salt (Molecular Probes, Eugene, OR)) at a rate of 20 -80 nl/min as described previously in detail (3). Net water movement (J v ) and osmotic water permeability coefficient (P f ) in response to established osmotic gradient (200 mosM) or stimulated by forskolin were measured as described previously (3).
Statistical Analysis-All of the values are expressed as the mean Ϯ S.E. Statistical analysis was performed by the Student's t test, and results were considered statistically different at p Ͻ 0.05.

RESULTS
Design of AQP1-siRNAs-We selected two target regions of rat AQP1 mRNA (i.e. 71-91 and 673-693 sequences) by scanning the length of the AQP1 gene for AA-dinucleotide sequences and downstream 19 nucleotides without significant homology to other genes by using an appropriate genome data base. The antisense strands of synthesized AQP1-siRNAs are the reverse complement of the target sequences (Fig. 1). The sense strands of the AQP1-siRNAs have the same sequences as the target mRNA sequences with the exception that they lack the 5Ј-AA sequence (Fig. 1). A uridine dimer was incorporated at the 3Ј end of the sense strands siRNAs (Fig. 1). Thus, the end products are two double-stranded 21-mer siRNAs (i.e. AQP1-siRNA(a) and AQP1-siRNA(b)) that theoretically should reduce the expression of AQP1 mRNA and protein and two siRNAs (i.e. scrambled AQP1-siRNA(a) and AQP1-siRNA(b)) that theoretically should not be effective in AQP1 gene silencing.
Uptake of AQP1-siRNA by IBDUs-IBDUs incubated for 24 h in normal rat cholangiocyte culture medium containing various amounts (i.e. 0 -20 nM) of Cy3-labeled AQP1-siRNA(b) in the absence or presence of the lipid carrier took up AQP1-siRNA in a dose-dependent manner (Fig. 2). However, no lipid carrier-dependent uptake of siRNA by IBDUs was observed. Given that only 20% AQP1-siRNA(b) transfected into cholangiocytes was labeled with Cy3 (see "Materials and Methods" for details), we conclude that the amount of AQP1-siRNA taken up by cholangiocytes was 5 times greater, suggesting that IBDUs could be effectively transfected with siRNAs in the absence of the lipid carrier. Based on this observation and given that a transfection reagent could potentially affect cholangiocytes function, studies of AQP1 gene suppression in IBDUs were performed by using naked AQP1-siRNAs.
AQP1 Gene Suppression in IBDUs by siRNAs-The levels of AQP1 mRNA and protein in IBDUs transfected with 10 nM each of four different naked siRNAs (i.e. two siRNAs to different regions of the AQP1 and two corresponding scrambled AQP1-siRNAs) are shown in Fig. 3. Both siRNAs to different sequences within the AQP1 gene effectively inhibited AQP1 mRNA (Fig. 3A) and protein expression (Fig. 3B). AQP1 mRNA and protein levels were inhibited by AQP1-siRNA(a) by 76.6 and 57.9%, respectively. AQP1-siRNA(b) was more effective, suppressing AQP1 mRNA and protein levels by 92.0 and 79.4%, respectively. In contrast, two scrambled AQP1-siRNAs had no effect on AQP1 gene expression. Moreover, AQP1-siRNAs had no effect on mRNA expression for AQP4 (data not shown), an AQP, which is also expressed in rat cholangiocytes (11), providing evidence that both AQP1-siRNA(a) and AQP1-siRNA(b) were specific for AQP1. Because there is a strong correlation between AQP1 mRNA and protein suppression by siRNAs (Fig.  3C), these data suggest that AQP1 silencing in cholangiocytes results from a reduction in the amount of AQP1 mRNA available for translation. These data also suggest that AQP1-siR-NAs were highly specific and efficient in AQP1 gene silencing in rat IBDUs.
Visualizing of AQP1 Gene Suppression in IBDUs by AQP1-siRNA-IBDUs incubated with 10 nM Cy3-labeled AQP1-siRNA(b) for 12-24 h were analyzed for AQP1 expression using immunofluorescence as described under "Materials and Methods." The data in Fig. 4 show that fluorescence intensity of AQP1 (green) decreased in transfected IBDUs as the fluorescence intensity of AQP1-siRNA(b) (red) increased. 12 and 24 h after IBDUs transfection with AQP1-siRNA, AQP1 immunofluorescence in cholangiocytes was reduced by 63 and 72%, respectively, suggesting that AQP1 could be silenced in cholangiocytes as quickly as 12 h post-transfection.
Inhibition of AQP1-mediated Water Transport by siRNAs-Water transport characteristics (i.e. P f and J v ) in AQP1-siR-NAs-nontransfected IBDUs and in IBDUs transfected with 10 nM AQP1-siRNAs were determined from the volume of water transported into the lumen of IBDUs either in response to a 200 mOsM transepithelial osmotic gradient (lumen osmolality Ͼ bath osmolality) (Fig. 5) or in response to 100 M forskolin (Fig. 6), an agent known to stimulate biliary bicarbonate and water secretion (2-4). The P f in AQP1-siRNA-nontransfected IBDUs and IBDUs transfected with scrambled AQP1-siRNAs ranged from 50 ϫ 10 Ϫ3 to 72 ϫ 10 Ϫ3 cm/sec (Figs. 5 and 6). When a transepithelial osmotic gradient was established in IBDUs transfected with AQP1-siRNA(a) and AQP1-siRNA(b), P f was decreased by 29.4 and 58.8%, respectively (Fig. 5A). In IBDUs transfected with AQP1-siRNA(b), P f was also decreased by 53.1% in response to forskolin (Fig. 6A). In IBDUs transfected with AQP1-siRNA(b), J v across biliary epithelia in response to an osmotic gradient or forskolin was decreased by 42.8 and 87.6%, respectively (Figs. 5B and 6B). A strong correlation between AQP1protein expression and water transport in IBDUs transfected with AQP1-siRNAs was shown (Fig. 5C), suggesting that a decrease of cholangiocyte osmotic water permeability is a result of specific AQP1 gene silencing by AQP1-siRNAs. fected with AQP1-siRNA(b) compared with nontransfected IBDUs. C, a correlation between AQP1 protein expression and P f was seen (y ϭ 0.014192 ϩ 0.00039212 ϫ x; r ϭ 0.9254). Values are the mean Ϯ S.E. of 4 -6 microperfused IBDUs in each group (*, p Ͻ 0.05).

DISCUSSION
The major objectives of this study were to develop and utilize siRNAs to directly investigate the importance of AQP1 in water transport by biliary epithelia. Although several lines of evidence indicate that AQPs are involved in water transport by cholangiocytes (4 -11), the absence of specific AQP inhibitors has impaired direct testing of hypotheses related to this issue. Indeed, until this study, no specific pharmacological inhibitors of AQPs had been reported.
Our results show that AQP1-siRNAs effectively (i.e. by 57.9 -79.4%), specifically (i.e. AQP1-siRNAs but not scrambled AQP1-siRNAs), and rapidly (i.e. during 12 h after transfection of IBDUs) down-regulate endogenously expressed AQP1 but not AQP4 in cholangiocytes. Moreover, direct quantitation of water transport in IBDUs transfected with siRNAs to AQP1 demonstrates that P f and J v are significantly decreased compared with IBDUs transfected with scrambled siRNAs. The observed decrease of net water secretion in IBDUs transfected with AQP1-siRNA(b) under different experimental conditions (i.e. an inward osmotic gradient or a secretory agonist) suggests that as much as 65% of water transported across biliary epithelia may be AQP1-mediated. Taken together, these findings demonstrate for the first time that inhibition of AQP1 expression in mammalian cells by siRNAs results in inhibition of water transport.
We found that cholangiocytes effectively take up siRNA duplexes if IBDUs are incubated with AQP1-siRNAs either in the absence or presence of transfection reagent. This observation is in contrast to published data (29) that siRNAs are taken up by cultured HeLa, COS-1, and 293 cells only in the presence of transfection agent. However, our data are consistent with recently published results that hepatocytes in vivo can be effectively transfected with naked siRNAs (30,31). Although a precise explanation for these differences is not apparent, they emphasize the importance of assessing both forms of siRNAs when employing this gene-silencing approach.
In our study, we utilized two different siRNAs constructed against different regions of AQP1 mRNA. Whereas both constructs inhibited AQP1 expression and function in cholangiocytes, siRNA-designated AQP1-siRNA(b) was more effective. We have no explanation as to why these two siRNAs differed in their inhibitory activity, but this phenomenon has been observed for other genes. For example, several siRNAs synthesized against different sites on the same target mRNA demonstrated striking differences in silencing efficiency (29).
To our knowledge, this study is the first to successfully utilize siRNA gene-silencing technology to achieve AQP gene suppression in epithelial cells. AQP gene silencing has previously been achieved primarily by developing transgenic null mice lacking AQPs and by using dominant negative mutants or antisense oligonucleotides (ODNs). The phenotype analysis of transgenic mice deficient in AQP1, AQP3, AQP4, and AQP5 has provided new insights into their critical role in water transport in the kidney, brain, eye, ear, salivary glands, skin, and gastrointestinal organs (32)(33)(34)(35)(36)(37)(38)(39)(40). However, given that cholangiocytes express numerous AQPs, deletion of a single AQP might not significantly affect biliary water transport in vivo because other cholangiocyte AQPs could undergo compensatory up-regulation. Dominant negative mutants, which exert their effects in the presence of the wild-type gene product, have also been employed in water-transporting studies, demonstrating dominant negative effects of AQP2 and AQP4 mutants on AQP-mediated water transport in Xenopus oocytes (41-43), LLC-PK 1 cells (44), and mouse cholangiocytes (45). This experimental approach also has its limitations being applicable principally to studies utilizing cultured cell systems. Antisense ODNs have been used to a limited degree to suppress AQP function. For example, the P f of Xenopus laevis oocytes injected with poly(A) ϩ RNA from cultured bovine corneal epithelial cells was inhibited by coinjection with AQP5 ODNs (46). Also, the expression of AQP1 in human trabecular meshwork cells was suppressed by corresponding ODNs (47). Although useful, the specificity of this approach is somewhat limited because nonspecific hybridization of ODNs with intracellular protein rather than mRNAs has been reported previously (48). Moreover, the efficiency of gene silencing by ODNs is relatively low (e.g. only one of eight antisense oligonucleotides is expected to provide a specific suppression of a targeted gene) (49). Our own experience has been that AQP1 ODNs are not effective in inhibiting AQP1 expression and function in isolated rat IBDUs (data not shown).
In summary, we designed and utilized siRNAs to AQP1 that markedly diminished the expression of this protein in IBDUs. As a result, cholangiocyte osmotic water permeability and net water secretion were significantly reduced, suggesting that at least 65% of water transported across biliary epithelia is AQP1mediated. These data are the first to demonstrate the feasibility of utilizing siRNAs to specifically reduce the expression of AQPs in epithelial cells. They also provide further evidence of the importance of AQP1 in water transport by biliary epithelia.