Background
Nucleoside analogs are currently employed in cancer treatment. These compounds exert cytotoxic effects by interfering with the uptake and metabolism of their natural counterparts. They trigger transcriptomic responses preferentially encompassing up-regulation of a set of genes implicated in cell cycle regulation and apoptosis along with other genes of undefined function in cancer chemotherapy [
1‐
4]. Among these “non-anticipated” genes, we identified aquaporin 3 (AQP3) [
4]. AQP3-related mRNA levels dramatically increased (8-fold) after treatment of MCF7 breast cancer cells with the capecitabine catabolite, 5
′-deoxy-5-fluorouridine (5
′-DFUR), a direct precursor of 5-fluorouracil (5-FU). Treatment of these cells with the human Equilibrative Nucleoside Transporter-1 (hENT1) inhibitor, NBTI, led to significant resistance to 5
′-DFUR, which was associated with a marked decrease in AQP3 up-regulation. Thus, it appears that changes in AQP3-related mRNA levels parallel the cytotoxic effects of nucleoside derivatives on breast cancer cells.
Aquaporins (AQPs) are integral membrane proteins implicated in the selective transport of water across the plasma membrane. A subset of the AQP family that includes AQP3 also mediates glycerol uptake. Accordingly, these proteins are designated aquaglyceroporins [
5‐
7]. When AQP3 was initially identified as putative drug target, limited information was available on the role of this protein family in cancer. Recent evidence suggests that selective AQP participate in angiogenesis, cell migration and metastasis (reviewed by [
8]). AQP1-null mice display reduced tumor growth after subcutaneous implantation of melanoma cells, which is associated with reduced endothelial cell migration and angiogenesis [
9]. Moreover, AQP1 expression promotes tumor cell extravasation and metastasis [
10]. AQP3 has been implicated in skin tumorigenesis. AQP3-null mice are resistant to the development of skin tumors, while skin squamous cell carcinomas overexpress this protein [
11]. Clinical data from a number of studies provide evidence for the heterogeneous expression of different AQP family members in solid tumors, and in most cases, AQP overexpression [
12‐
15].
The possibility that a particular AQP gene member is implicated in the chemotherapeutic response to antitumor agents has not been addressed. Moreover, previous studies reporting acute AQP3 up-regulation following nucleoside-derived drug treatment in cultured cancer cells do not provide insights into whether changes in the AQP3-related mRNA level represent a collateral effect of treatment or, on the contrary, it participates in drug response, either by promoting it or by acting as a resistance gene. In this study, we address whether AQP3 is implicated in drug responses by monitoring the effects of gene silencing on expression patterns of nucleoside analogs-induced target genes, cell cycle progression, and cell growth in the breast cancer cell line MCF7 and the colon adenocarcinoma cell line HT29.
Methods
Reagents
5′-DFUR, 5-fluorouracil, cisplatin (cis-diaminedichloroplatinum or cis-DDP) and propidium iodide were purchased from Sigma-Aldrich (Saint Louis, MO, USA). Gemcitabine (2′,2′-difluorodeoxycytidine, dFdC, Gemzar®) was obtained from Eli Lilly and Company (Indianapolis, IN, USA).
Cell culture and treatments
The human colorectal carcinoma cell line HT29 (HTB-38, ATCC-LGC Promochem Partnership, USA) and two human breast carcinomas cell lines, MCF7 (HTB-22, ATCC-LGC Promochem Partnership, USA) and MDA-MB-468 (HTB-132, ATCC-LGC Promochem Partnership, USA) were purchased from the American Type Culture Collection with the indicated references. MCF7 and MDA-MB-468 cell lines are characterized by the fact that the former expresses the estrogen and progesterone receptors whereas the latter is negative for both. NP-29 cells were derived from human pancreatic adenocarcinomas, which had been perpetuated as xenografts in nude mice and further characterized for different oncogene and tumor suppressor profiles [
16]. MCF7 and HT29 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (GIBCO-BRL, Grand Island, NY, USA), 2 mM glutamine, and a mixture of antibiotics (100 U penicillin, 0.1 mg/ml streptomycin and 0.25 μg/ml fungizone). The MDA-MB-468 cell line was maintained in DMEM and F12 mixture (1:1) supplemented with 10% fetal bovine serum, 2 mM glutamine and 100U penicillin, 0.1 mg/ml streptomycin. NP-29 cells were maintained in DMEM and F12 mixture (1:1) supplemented with 5% fetal bovine serum, 2 mM glutamine and 100U penicillin, 0.1 mg/ml streptomycin. Cells were maintained as monolayer cultures at 37°C in an atmosphere containing 5% CO
2, and subcultured by trypsinization every 4–5 days.
Mycoplasma test assays, verification of morphology and growth curve analysis were performed as a routine protocol for all of them. Cells were treated 24 h after seeding at 20 000 cells/cm
2. Cultures were exposed to drugs for 90 min (5
′-DFUR: 250 μM; 5-FU: 250 μM; gemcitabine: 100 nM for MCF7, 250 nM for MDA-MB-468 and NP-29 and 50 μM for HT29; cisplatin: 50 μM), and measurements performed at 24 or 48 h after drug addition. Drug concentrations were chosen based upon the EC
75 values calculated from MTT cell viability assays, as previously described [
4,
17]. The choice of 90 min was based upon the need to highlight the role transport processes play in drug action but, more importantly, to better mimic the in vivo exposure time to the drug, which is far less shorter than the “classical” cytotoxicity assays in which cells are exposed to drugs for 24, 48, and even 72 hours.
RNA isolation and quantitative RT-PCR
Isolation of mRNA was performed after treatment using the SV Total RNA Isolation System (Promega Biotech, Madison, WI, USA), following the manufacturer’s protocol. Total DNase-treated RNA (1 μg) was used to generate cDNA using M-MLV Reverse Transcriptase (Promega Biotech) and random hexamers (Amersham Pharmacia, Buckinghamshire, UK) for reverse transcription. Quantitative real-time PCR was performed with the ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) using the manufacturer’s recommendations. Assays-on-Demand Taqman probes (Applied Biosystems) for AQP3, CDKN1A/p21, TNFRSF6/FAS and GAPDH were employed (Hs00185020_m1, Hs00355782_m1, Hs00163653_m1 and 4310884E, respectively). Relative quantification of gene expression was performed as described in the TaqMan user manual with GAPDH as an internal control.
Measurement of cell volume and cell counting
Cells were plated in 24-well culture plates. After 24 h, cells were treated for 90 min with different genotoxic agents. Cultures were allowed to proceed for 48 h. The cell culture was washed and the remaining cells were trypsinized and collected in culture medium. Cell volume and number were measured using a cell counter Coulter Multisizer (Beckman Coulter, Inc., Fullerton, CA) or Quanta SC flow cytometer (Beckman Coulter). The population of viable cells was discriminated by size and the number of cells was calculated as a percentage by comparing the cell number from treated cultures with that from cultures not exposed to cytotoxic drugs.
Transfection with small interfering RNA (siRNA) for AQP3
AQP3 siRNA (ID: 147362) was purchased from Ambion (Austin, TX, USA). Silencer® Negative Control siRNA #1 (Ambion) was employed as the negative control to ensure silencing specificity in all the experiments.
Transfection of cells with 20–25 nM (MCF7) or 200 nM (HT29) of siRNA was performed using Lipofectamine 2000® (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s recommendations. Transfection efficiency was measured using AQP3 siRNA (ID: 147362) labeled with FAM (6-carboxy-fluorescein) and a Beckman Coulter flow cytometer (Fullerton, CA). Depletion of AQP3 expression following siRNA transfection was confirmed by real-time RT-PCR, as described above.
Cell cycle analysis
At 48 h after treatment, cells were collected by centrifugation at 1200 g for 4 min and fixed in cold 70% ethanol. After 24 h, cells were washed and resuspended in 0.5 ml of PBS containing RNase (10 μg/ml). Flow cytometry analysis was performed within 1 h after the addition of propidium iodide (0.1 mg/ml) at room temperature using a Coulter XL (Beckman Coulter).
Western blot analysis
Cells were lysed in a RIPA buffer containing 1% Complete Mini protease inhibitors (Roche, Mannheim Germany). Protein concentration was determined by the Bradford assay (Bio-Rad, Hercules, CA) and 30 μg of total protein were resolved by electrophoresis on 12% SDS-PAGE gels and transferred to PVDF membranes by standard methods. Membranes were immunoblotted with anti-p21 (Santa Cruz, Santa Cruz, CA), anti-Fas (Roche, Mannheim, Germany) and anti-tubulin (Sigma, St Louis, Mo) and the corresponding secondary antibodies, horseradish peroxidase (HRP)-conjugated antibodies (Bio-Rad, Hercules, CA). Antibody labeling was detected using the chemiluminiscence detection kit (Biological Industries, Israel).
Apoptosis detection
Apoptosis was measured using the Annexin V-FITC Apoptosis Detection Kit I (BD Biosciences, San Diego, CA). Cells were harvested by centrifugation (including detached cells) 48 h after treatment with increasing doses of 5-fluorouracil (5–500 μM), washed twice in PBS, and pelleted again. They were resuspended at 106 cells/ml in binding buffer, 100 μl of cells were stained with 5 μl Annexin-V and 5 μl propidium iodide, and incubated in the dark for 15 min at room temperature, as recommended by the manufacturer. Following the addition of 400 μl binding buffer, cells were processed within 1 h using the FACScan flow cytometer Coulter XL (Beckman Coulter).
Statistical analysis
The paired or unpaired Student’s t-test was used to compare experimental data. Analysis was performed using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA).
Discussion
High-throughput transcriptomic analysis of anticancer drug activity is a suitable tool to identify novel target genes. However, confirmation that a particular drug-modulated gene specifically contributes to drug response requires detailed analysis similar to that performed for AQP3, a gene up-regulated by the 5-FU precursor and capecitabine catabolite, 5
′-DFUR, in the breast cancer cell line MCF7 [
4].
AQP3 is a broadly expressed aquaglyceroporin found in most epithelia, where it localizes to the basolateral membrane, as well as in several types of nonepithelial cells [
21]. The extensive distribution pattern suggests that this water channel protein is a major player in barrier hydration and water and osmolyte homeostasis. AQP3 is a target of aldosterone in the collecting duct [
22] and under osmotic control in renal and keratocarcinoma cells, thus presumably contributing to cell volume adaptive regulatory processes [
23,
24]. While previous studies suggest that changes in cell size associated with cell division are facilitated by increased AQP1 abundance at the plasma membrane [
25], our results support a putative role of AQP3 in maintaining or promoting cell swelling induced by nucleoside-derived drugs. Interestingly, AQP3-related mRNA levels were not modified during cell cycle progression, suggesting that the role of the water channel in the increased cell volume is related to drug response. The nucleoside analogs 5
′-DFUR and gemcitabine triggered G
1/S cell cycle arrest, but not cisplatin. This DNA alkylating agent appeared to induce S/G
2 arrest, which did not result in increased cell volume, in contrast to the effects of nucleoside-derived drugs.
Knockdown of AQP3 expression produced a partial but significant reversion of increased cell swelling associated with nucleoside-derived drug treatment, further supporting a role of AQP3 in this process. Nevertheless, the magnitude of cell volume reversion in MCF7 and HT29 (about 25%), even assuming that AQP3 expression is only partially blocked in siRNA-transfected cells, suggests that this water channel protein is not the only contributor to cell swelling associated with drug treatment. Interestingly, under similar conditions, suppression of AQP3 preserved cell growth inhibition to a better extent, and the magnitude of reversion of G
1/S cell cycle arrest was significantly higher than reversion of cell swelling for 5
′-DFUR and gemcitabine in MCF7 cells. Furthermore, in spite of achieving only a 20% of AQP3 mRNA knockdown in HT29, AQP3 suppression partially reverted cell cycle arrest and preserved cell growth inhibition in 5
′-DFUR treated cells. Thus, it is possible that AQP3 plays roles other than those derived from its ability to mediate water transport. In fact, AQP3 plays a variety of roles in cell physiology associated with its ability to take up glycerol. AQP3-deficient mice show defective skin hydration and elasticity, which can be corrected by glycerol replacement [
26]. Moreover, wound healing is significantly impaired in these animals, with low keratinocyte proliferation, a feature that can also be reversed
in vivo by feeding mice with glycerol [
27]. Interestingly, inhibition of AQP3 in keratinocyte cell cultures results in reduced water and glycerol permeability and impaired cell migration. The protein facilitates migration by functioning as a water channel, but is also implicated in epidermal cell proliferation as a glycerol transporter [
27]. Consistent with this finding, mice lacking AQP3 expression not only display impaired epidermal cell proliferation but are also resistant to skin tumorigenesis [
11]. This appears to be related to the ability of AQP3 to take up glycerol, a suitable energy substrate that supports cell growth. Nucleoside-derived drugs, particularly those used in antiviral therapy, may induce severe mitochondrial toxicity [
28,
29]. While this is not evident for nucleosides used in the treatment of solid tumors, recent evidence suggests that gemcitabine triggers moderate mitochondrial toxicity [
30] and blocks the activity of human mitochondrial DNA polymerase [
31]. Nucleoside derivatives additionally compete with intracellular nucleotides and inhibit key enzymes of the nucleoside salvage pathways [
32,
33], consequently impairing the cellular energy metabolism. In this context, it is feasible to assume that AQP3 induced after exposure to these drugs plays a compensatory role as a provider of energy substrates.
AQP3 silencing also reversed the up-regulation of selective p53-dependent transcriptional targets, such as the death receptor, FAS, implicated in apoptosis, and the inhibitor of the cyclin-CDK2 and -CDK4 complexes, p21, implicated in the modulation of cell cycle progression at G
1. It is not clear from these observations whether AQP3 contributes to apoptosis in addition to its reported effect on cell cycle arrest, which is significantly reversed upon silencing of the gene. Interestingly, AQP3 itself is transcriptionally regulated by p73, a member of the p53 family, which exhibits similar biochemical properties but is rarely mutated in cancer cells [
34]. p73 interacts with the transcriptional coactivator, Yes-associated protein (YAP), leading to enhanced p73-dependent apoptosis in response to DNA damage. YAP is stabilized by the product of the p73/YAP target gene, PML, under negative control by the proto-oncogenic AKT/PKB kinase [
35]. Interestingly, the anticancer drug, curcumin, down-regulates AQP3 expression in cancer ovarian cells via a mechanism that involves, at least partially, inhibition of the EGFR pathway and downstream AKT [
19]. While AQP3 is a p73 target, its association with pro-apoptotic processes does not appear relevant, at least under the conditions used here. This hypothesis is based on evidence that AQP3 up-regulation is observed only at 5-FU concentrations triggering cell cycle arrest, but not at higher doses in which cells are committed to programmed cell death. Moreover, the decrease in cell growth associated with short treatment with low doses of 5-FU is significantly reversed by knockdown of AQP3 expression. These observations collectively support the view that induction of this aquaglyceroporin is related to cell cycle arrest rather than apoptosis.
Aquaporins, including AQP3, are overexpressed in different tumors [
12,
13,
15] and an oncogenic role was suggested for AQP5, although this protein is not an aquaglyceroporin [
13,
36]. To our knowledge, no correlation of basal or drug-induced AQP3 expression with drug chemoresistance has been reported to date. In view of the above findings, this issue deserves further investigation.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
LTM performed the experiments described in Figures
4,
5 and
6, contributed to the experiments shown in Figures
1,
2 and
3, and drafted and submitted the manuscript. SPT contributed to the experiments described in Figures
1,
2 and
3, and helped with the writing and revision of the manuscript. FJC helped with the study design and critically reviewed the manuscript. MMA and MPA are responsible for the study design, provided guidance, supervised the work and helped with the writing and revision of the manuscript. All authors read and approved the final manuscript.