Molecular Imaging of Aquaglycero-Aquaporins: Its Potential for Cancer Characterization

Abstract

Aquaglycero-aquaporins (agAQPs) are one of the water channel proteins located in the cell membrane that transport not only water but also some small solutes such as glycerol. Since agAQPs are involved in cancer proliferation and malignancy, it might be possible to utilize them as new targets for cancer molecular imaging. In this study, we investigated whether agAQPs can be specifically targeted by using [¹⁴C]-labeled glycerol ([¹⁴C]glycerol), which passes through agAQPs. In the in vitro experiments, comparing the cancer cell lines with different expression levels of AQP3 and AQP9, major agAQPs known to be expressed in cancers, and examining the effect of their inhibitors on these cells, the expression of AQP3 and AQP9 in cell lines was shown to be closely related to [¹⁴C]glycerol uptake. When [¹⁴C]glycerol was injected into tumor-bearing mice, Spearman’s rank coefficient analysis revealed that radioactivity levels in tumor and in plasma were mutually correlated only in tumors expressing agAQPs at a high level. These results indicate the possibility of using agAQPs as new targets to characterize cancer using radiolabeled glycerol as a molecular probe.

Aquaporins (AQPs) are a family of membrane transport proteins that assemble in the cell membrane and transport water.^1,2) Thirteen subtypes (AQP0–12) have been identified in mammals. Although AQPs are expressed in almost all tissues, the distribution pattern of the subtypes is tissue-specific. Several studies showed that some AQPs were involved in cancer proliferation, its malignancy and angiogenesis by activation and inactivation of various cell signaling cascades.^1–7)

AQP3, AQP7, AQP9, and AQP10, termed aquaglycero-AQPs (agAQPs), can transport not only water but also small solutes such as some ions and glycerol.⁸⁾ In particular, AQP3 and AQP9 are considered to be closely associated with cancer development because their expression is dramatically changed in various cancers. AQP3 is reported to be overexpressed in cancers of the brain, lung, kidney, colon, skin, and ovary, and to regulate epidermal growth factor (EGF) signaling to promote cell proliferation and cell migration in human skin fibroblasts and human ovarian cancer cells. AQP9 overexpression was observed in brain and liver cancer.^9–12) In addition, AQP9 was reported to regulate filopodia formation at the cell edge, which enhances cell migration and metastasis potential.¹³⁾ Furthermore, some metalloids, such as arsenic and antimony, may also be transported through agAQPs by the same mechanism as glycerol uptake,³⁾ suggesting that the agAQP-overexpressing tumors could be a chemotherapeutic target of metalloid-containing drugs. Based on these previous reports, we considered that agAQPs would make interesting molecular targets for cancer molecular imaging, and that data on their expression in cancer would provide information useful for cancer diagnosis and therapy. Unfortunately, there are few candidates for subtype-specific binding partners or selectively passing molecules for AQPs. With regard to AQP3 and AQP9, glycerol is known as a commonly passing molecule, which does not pass through other major AQPs.

In this study, we investigated whether molecular imaging of agAQPs with radiolabeled glycerol was possible in cancer cells and a mouse xenograft model using [¹⁴C]glycerol.

MATERIALS AND METHODS

Antibodies and Reagents

Rabbit anti-AQP3 antibody was purchased from Millipore (Billerica, MA, U.S.A.) and rabbit anti-AQP9 antibody from Alpha Diagnostics (San Antonio, TX, U.S.A.). Goat anti-actin (C-11) antibody and donkey anti-goat immunoglobulin G (IgG) antibody conjugated with horseradish peroxidase (HRP) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Goat anti-rabbit IgG antibody conjugated with HRP was purchased from GE Healthcare (Uppsala, Sweden). Goat anti-rabbit IgG antibody conjugated with Alexa Fluor 488 was purchased from Invitrogen (Carlsbad, CA, U.S.A.). Curcumin and para-chloromercuriphenyl sulfonate (pCMPS) were obtained from Sigma-Aldrich (St. Louis, MO, U.S.A.).

Cell Culture

Human ovary cancer Caov3 cells, human glioblastoma U87MG cells, and human breast cancer MDA-MB435 cells were obtained from the American Type Culture Collection (Manassas, VA, U.S.A.). Caov3 and U87MG cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich) containing 10% fetal bovine serum (FBS, SAFC Biosciences, Lenexa, KS, U.S.A.), 50 units/mL of penicillin and 50 µg/mL of streptomycin (Invitrogen) in a humidified atmosphere of 5% CO₂ in air at 37°C. MDA-MB435 cells were cultured in DMEM/F-12 (Sigma-Aldrich), containing 10% FBS and antibiotics as above.

Western Blotting

Cells were lysed in lysis buffer [50 mM Tris–HCl (pH 7.5), 1.0 mM ethylenediaminetetraacetic acid (EDTA), 150 mM NaCl, 1.0% NP-40, 0.25% sodium deoxycholate, and protease inhibitor cocktail (Sigma-Aldrich)] and centrifuged at 15000×g, 4°C for 20 min. A portion of the supernatant was used for measurement of protein concentration. Samples mixed with 2× Laemmli buffer (Bio-Rad) were boiled and 15–20 µg of protein was subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore). Nonspecific binding was blocked with Blotto A (Santa Cruz). The membrane was then incubated with primary antibody diluted in Blotto A or 5% bovine serum albumin (BSA) in TBS containing 0.05% Tween 20 (TBST). After washes with TBST, the membrane was incubated with HRP-conjugated secondary antibody. Antibody binding was detected by chemiluminescence using the ECL Plus Western Blotting Detection System (GE Healthcare) following the manufacturer’s instructions. Bands were visualized with a Chemi-Smart 5000 system (Vilber Lourmat, Torcy, France) and Chemi-Capt software (Vilber Lourmat).

Glycerol Uptake

Normal cells or drug-treated cells in 12-well plates were washed with Hanks’ balanced salt solution (HBSS) and incubated in HBSS for 15 min. After aspiration of the buffer in wells, 1 mL of 3.7 kBq/mL [2-¹⁴C]glycerol (3.7 MBq/mL, 2035 MBq/mmol; American Radiolabeled Chemicals, Inc., St. Louis, MO, U.S.A.) dissolved in HBSS was added to the cells for 30 or 60 min. At the indicated times, the buffer was removed from the well and the cells were washed with HBSS three times. Cells were collected in 0.2 N NaOH containing 0.5% SDS. The protein amount was measured using a DC Protein Assay Kit (Bio-Rad, Tokyo, Japan). The samples were then mixed with Hionic-Fluor (PerkinElmer Japan Co., Ltd., Kanagawa, Japan) and radioactivity was measured with a liquid scintillation counter (LS6000, Beckman Coulter, CA, U.S.A.).

Cell Treatment

For AQP3 inhibitor treatment, cells were treated with 0, 25, or 50 µM curcumin for 18 h in serum-free medium. For AQP9 inhibitor treatment, 0, 25, 50, or 100 µM pCMPS in HBSS was applied to cultured cells for 1 h. After these treatments, glycerol uptake was measured as described in the previous section.

Animal Experiments

The animal experiments were approved by the institutional animal care and use committee of our institute. Four-week-old BALB/c AJ nu/nu mice were obtained from CLEA Japan, Inc. (Tokyo, Japan). At 5 weeks of age, the mice were divided into two groups. In one group, mice were subcutaneously injected with 1.0×10⁷ of Caov3 cells on one side of the back and 1.0×10⁷ of MDA-MB435 cells on the other side of the back. In another group, mice were subcutaneously injected with 0.2×10⁷ of U87MG cells on one side and 1.0×10⁷ of MDA-MB435 cells on the other side. Two to three weeks later, the mice were weighed and 37 kBq [¹⁴C]glycerol (2 µg, dissolved in 100 µL of phosphate buffered saline (PBS)) was intravenously injected into tumor-bearing mice with 10 µg of unlabeled glycerol. At 10, 30, and 60 min after the injection, mice were euthanized and approximately 100 mg of tissue of interest was collected. The tissue samples were weighed and dissolved in Soluene 350 (PerkinElmer) and mixed with Hionic-Fluor, and radioactivity was measured with a liquid scintillation counter. The tissue distribution of radioactivity was expressed as the percentage of the injected dose per gram normalized to a mouse body weight of 20 g (%ID/g).

Immunohistochemistry

Tumors were removed from the mice and immediately fixed in 10% formaldehyde overnight. Tumor samples were embedded in paraffin and 4-µm sections were cut and mounted on glass slides. After deparaffinization and rehydration, slides were incubated in Target Retrieval Solution Citrate (pH 6.0) (Dako Cytomation, Carpinteria, CA, U.S.A.) at 95°C for 20 min. After one wash with Tris-buffered saline [TBS, 25 mM Tris (pH 7.4), 150 mM NaCl, and 2 mM KCl], the sections were covered with 3% H₂O₂ at room temperature (RT) for 5 min if detected with HRP and DAB, followed by an additional wash. Slides were then placed in a humid chamber and incubated with the primary antibody at 4°C overnight. After washes with TBS, the slides were incubated with HRP-conjugated secondary antibody and visualized with a DAB substrate chromogen solution (Dako) according to the manufacturer’s instructions and counterstained with hematoxylin. The slides were examined with an optical microscope. In some cases, after incubation with primary antibody the slides were incubated with Alexa Fluor 488-conjugated secondary antibody for 2–3 h at RT. The slides were examined with a confocal fluorescence microscope (FV1000, Olympus, Tokyo, Japan).

Statistical Analysis

Quantitative data are presented as mean (n≥2 for in vitro studies, and n≥3 for in vivo studies). Aspin–Welch’s t-test or Student’s t-test was used, depending on the results of F-test. p values <0.05 were considered to indicate a significant difference. In the in vivo experiment, we compared [¹⁴C] activity in tumor and some tissues with that in plasma in individual animals using Spearman’s rank correlation coefficient. p values <0.05 were considered to indicate a significant difference.

RESULTS

We first analyzed the presence of AQP3 and AQP9 proteins in various human cancer cell lines. The representative Western blotting images in Fig. 1A show that AQP3 was expressed strongly in Caov3 cells and U87MG cells, whereas no expression was observed in MDA-MB435 cells. AQP9 was expressed strongly in U87MG cells, slightly in MDA-MB435 cells, and negligibly in Caov3 cells. To investigate whether glycerol uptake was consistent with the level of AQP3 and/or AQP9 (AQP3/AQP9) expression, we next tested [¹⁴C]glycerol uptake. The amount of glycerol uptake was increased in AQP3/AQP9 overexpressing cells (Fig. 1B). To further study the relationship between AQP3/AQP9 expression and glycerol uptake, cells were treated with their inhibitors: curcumin for AQP3, and pCMPS for AQP9. When Caov3 cells were treated with curcumin, glycerol uptake was markedly decreased in Caov3 cells (Fig. 2A). A decrease in glycerol uptake with curcumin treatment was also observed in U87MG cells (Fig. 2B), in which AQP3 protein is moderately expressed (Fig. 1A). Treatment of U87MG cells with pCMPS dramatically decreased glycerol uptake in a concentration-dependent manner (Fig. 2C).

Fig. 1. AQP3 and AQP9 Expression and Glycerol Uptake in Cancer Cells

(A) Western blotting of proteins from MDA-MB231, MDA-MB435, Caov3 and U87MG cells. (B) Cellular uptake of glycerol after 60 min-incubation with [¹⁴C]glycerol. Data are shown as mean with S.D. (n=4–5, * p<0.05).

Fig. 2. Changes in Glycerol Uptake by Inhibitor Treatment

(A, B) Cellular uptake of glycerol after curcumin treatment (A: Caov3 cells, B: U87MG cells). The cells were incubated with [¹⁴C]glycerol for 60 min. (C) Cellular uptake of glycerol after pCMPS treatment. After U87MG cells were treated with pCMPS as described in Materials and Methods, the cells were incubated with [¹⁴C]glycerol for 30 min. All data are shown as mean (n=2).

Since we observed a positive correlation between AQP3/AQP9 expression level and the amount of glycerol uptake in in vitro assays, we studied the [¹⁴C]glycerol distribution in tumor-bearing mice (Fig. 3A). Mice were subcutaneously implanted with MDA-MB435 cells and either Caov3 cells or U87MG cells. The [¹⁴C] activities were not significantly different between tumors with different levels of AQP3/AQP9 expression. However, the temporal change in [¹⁴C] activity in tumors was found to be different. Radioactivity in tumors expressing AQP3/AQP9 at a high level (Caov3 and U87MG) decayed in a time-dependent manner similar to that in the plasma, whereas radioactivity in tumors expressing AQP3/AQP9 at a low level (MDA-MB435) remained constant. Spearman’s rank correlation coefficients showed a good correlation between [¹⁴C] activity in plasma and tumor for Caov3 and U87MG tumors (Figs. 3B, C and Table 1). The coefficient was 0.636 for U87MG and 0.222 for MDA-MB435 tumors in mice with the two tumors, and it was 0.741 for Caov3 tumor and 0.399 for MDA-MB435 tumor in mice bearing the two tumors. When we compared the radioactivity in normal tissue (skin, adipose or muscle) and plasma, radioactivity in skin, where AQP3 is expressed at high level,¹⁴⁾ and adipose, where AQP7 (another agAQP) is expressed at high level,^15,16) decayed in a time-dependent manner similar to that in the plasma whereas radioactivity in muscle, where only AQP1 and AQP4 are expressed,¹⁷⁾ statistically did not change (Fig. 3D). When performing the Spearman’s rank correlation test, Spearman’s rank correlation coefficients showed also a good correlation between [¹⁴C] activity in plasma and skin or adipose, not in plasma and muscle (Fig. 3E and Table 2).

Fig. 3. Biodistribution of [¹⁴C]Glycerol in Tumor-Bearing Mice

(A) Radioactivities in plasma and tissues of interest at the indicated times after [¹⁴C]glycerol injection are expressed as %ID/g (n=3–5). (B, C) The radioactivity in the plasma and the tumor in each mouse were plotted in MDA-MB435 (filled circle) and U87MG (open circle) tumor bearing mice (B) or in MDA-MB435 (filled circle) and Caov3 (open circle) tumor bearing mice (C). (D) Radioactivities in skin, adipose, muscle and plasma at the indicated times after [¹⁴C]glycerol injection are expressed as %ID/g (n=3–5, * p<0.05). (E) The radioactivity in the plasma and the tissues in each mouse were plotted (skin: gray circle, adipose: open circle, muscle: filled circle).

Table 1. Comparison of Radioactivity in the Plasma and in the Tumors with Different Levels of AQP Expression: Spearman’s Rank Correlation Coefficients (ρ) between Plasma and Xenografts

	MDA-MB435	U87MG	MDA-MB435	Caov3
ρ	0.222	0.636	0.399	0.741
p	0.138	<0.01	0.1	<0.01

Table 2. Comparison of Radioactivity in the Plasma and in the Skin or Adipose or Muscle: Spearman’s Rank Correlation Coefficients (ρ) between Plasma and Each Tissue

	Skin	Adipose	Muscle
ρ	0.937	0.748	0.455
p	<0.01	<0.05	0.132

AQP3 and AQP9 expression in the tumors was confirmed by immunohistochemistry (Fig. 4). Histological analysis showed negative staining for AQP9 in MDA-MB435 tumors (Fig. 4A) and positive staining in U87MG tumors (Fig. 4B). We observed AQP3 expression in Caov3 tumors (Figs. 4D, F, H). Although the expression of AQP3 and AQP9 in MDA-MB435 cells was negligible in the in vitro experiment (Fig. 1A), the tumor was slightly stained with anti-AQP3 antibody (Figs. 4C, E, G).

Fig. 4. Expression of agAQPs in Tumors

Tumor sections were stained with anti-AQP9 antibody (A; MDA-MB435, B; U87MG) and anti-AQP3 antibody (C, E, G; MDA-MB435, D, F, H; Caov3). Scale bar indicates 100 µm in A–F and 25 µm in G and H.

DISCUSSION

We selected glycerol as a molecular probe for agAQPs, which are known to be related to carcinogenesis and cancer malignancy,¹⁾ and performed basic studies using [¹⁴C]glycerol. Molecular imaging of cancer is based upon cancer characteristics such as accelerated metabolism, activated molecular function, and molecular overexpression. Accordingly, a molecular probe to detect a target is usually designed by modification and labeling of its specific ligands, inhibitors, antagonists, agonists, substrates, or enzymes. With regard to AQPs, one candidate for an imaging probe would be a radiolabeled water molecule. However, water has no specificity with respect to AQP subtypes. Another candidate for an AQP molecular probe would be one based on specific inhibitors.¹⁸⁾ Although development of inhibitors for AQP4 is in progress, no appropriate candidates are known for agAQPs. At present, glycerol appears to be a possible molecular probe for agAQP imaging.

In vitro experiments showed that expression levels of AQP3 and AQP9 in the cancer cells tested in this study were positively reflected to [¹⁴C]glycerol uptake. Furthermore, glycerol uptake was decreased on treatment with curcumin or pCMPS, which were known to inhibit the expression or function of AQP3 or AQP9, respectively. These findings indicated that glycerol uptake was mainly regulated by AQP3 and AQP9 in the cell membrane, which is consistent with the results of an experiment with AQP9 knockout mice that showed the importance of AQP9 in glycerol metabolism,¹⁹⁾ although we could not exclude the possibility that other agAQPs, AQP7 and AQP10, and/or diffusion were involved.

Contrary to the in vitro results, in an in vivo biodistribution study performed 10 min after administration of [¹⁴C]glycerol, uptake was only slightly higher in tumors with high AQP3/AQP9 expression than in those with low expression. After 30–60 min, the [¹⁴C] radioactivity found in tumors was similar regardless of AQP3/AQP9 expression level. At least two explanations are possible. One is that the time points used in the in vivo experiment were too long after the time of [¹⁴C]glycerol administration, since glycerol can be metabolized quickly after administration.²⁰⁾ Within 2 h after [¹⁴C]glycerol administration, some [¹⁴C]glycerol may exist as lipid constituents and some [¹⁴C]glycerol may be metabolized to amino acids and other compounds,²¹⁾ or may even have left the body after conversion to CO₂ gas.²²⁾ At earlier time points, however, we were concerned that a high level of radioactivity in the circulation might obscure the specific distribution of the compound. A second explanation might be the low but detectable expression of AQP3 in MDA-MB435 tumors, as seen in the immunohistochemistry study. The MDA-MB435 tumor mass includes stromal cells, in which AQP3 is known to be expressed,²³⁾ and this might have affected [¹⁴C] radioactivity in the tumor in the biodistribution study.

Although it is difficult to estimate agAQP expression from glycerol uptake in vivo at a single time point, comparison of the temporal changes in radioactivity in the plasma and tumor might be a useful method of estimation, as the radioactivity in tumors with high AQP3/AQP9 expression decreased together with the radioactivity in the plasma, while it remained constant in tumors with low expression. A similar correlation was also found between plasma and skin where AQP3 is highly expressed,^23–25) and between plasma and adipose tissue where AQP9 and AQP7, another agAQP, are expressed.^2,16) On the other hand, radioactivity did not change statistically with time in muscle where only AQP1 and AQP4 are expressed.¹⁷⁾ As water exchange through AQP was reported to be quite fast,²⁶⁾ glycerol exchange between extra- and intra-cellular space by the cells expressing high level of AQP3/AQP9 can be expected to be much faster compared with the time resolution of in vivo measurement in this study, which could explain the nearly equilibrated tissue [¹⁴C]glycerol concentration with that in plasma. Further analysis using Spearman’s rank correlation test revealed that [¹⁴C] radioactivity in the plasma was significantly correlated with that in the tumors with high AQP3/AQP9 expression but not with that in the tumors with low expression in the same mice. These data suggest the potential of radiolabeled glycerol to distinguish tumors expressing high levels of AQP3/AQP9. Further refinement of the system might enable us to estimate the expression level of agAQPs, and hence, the effectiveness of an anti-cancer drug targeting agAQPs. Under physiological conditions, agAQPs also allow the passage of trivalent metalloid by the same intake mechanism as glycerol.³⁾ The detection of agAQP expression would be beneficial in cancer chemotherapy using these metalloid compounds.

Since water in and surrounding cells forms an important “scaffold field” for chemical reactions, such as metabolism, and physical conditions, such as pH, one can expect some difference in the status of water between cancer cells and normal cells.²⁷⁾ Water crosses the cell membrane directly and indirectly through AQPs. While direct movement is very slow, AQP-mediated movement is rapid.²⁸⁾ An increase in AQP expression in tumor cells would lead to a dramatic change in water movement into or out of the cell, which could determine some characteristics of the tumor. In this study, we showed that glycerol could be used for molecular imaging of agAQP, which would mean that glycerol could be used for indirect water imaging. There are already some imaging methods for monitoring water or blood, such as magnetic resonance imaging (MRI) and positron emission tomography (PET). MRI can visualize micromovements of water using the technology of diffusion-weighted imaging (DWI).²⁹⁾ Cell membrane water permeability affects contrast in DWI, although it is not a major signal source for DWI under the usual clinical settings. It is possible to obtain absolute quantification by PET. In PET studies, [¹⁵O]water is used as an authentic molecular probe.^30,31) However, since the short half-life of [¹⁵O] (2 min) makes it difficult to use [¹⁵O]water, there are few validation studies on [¹⁵O]water PET.^32–35) Our study of glycerol uptake in vivo and analysis with Spearman’s rank correlation coefficient indicates a new approach to the delineation of the different behaviors of water and the demand for it in tumors and normal tissues.

Although we used [¹⁴C]glycerol as a probe for agAQPs in this study, [¹⁴C]-labeled molecules cannot be serially traced in the same animal in the living state and cannot be used in humans. The positron emitter [¹¹C] could replace [¹⁴C] in sequential PET scans in animals and in humans; this would make it possible to perform more elaborate studies to elucidate the relationship between [¹¹C]glycerol and agAQP expression in the region of interest.

In conclusion, based on the present study we propose that radiolabeled glycerol could be used as a molecular probe targeting agAQPs. This might enable the estimation of agAQP expression by comparing radioactivity in the tissue of interest and plasma in the same individual, thus providing valuable information about the characteristics of the tissue.

Acknowledgments

We would like to thank the members of the Diagnostic Imaging Program, Molecular Imaging Center, NIRS, for helpful discussions and valuable suggestions.

REFERENCES

• 1)  Verkman AS, Hara-Chikuma M, Papadopoulos MC. Aquaporins–new players in cancer biology. J. Mol. Med. (Berl.), 86, 523–529 (2008).
• 2)  Verkman AS. More than just water channels: unexpected cellular roles of aquaporins. J. Cell Sci., 118, 3225–3232 (2005).
• 3)  Bhattacharjee H, Rosen BP, Mukhopadhyay R. Aquaglyceroporins and metalloid transport: implications in human diseases. Handbook Exp. Pharmacol., 190, 309–325 (2009).
• 4)  Saadoun S, Papadopoulos MC, Hara-Chikuma M, Verkman AS. Impairment of angiogenesis and cell migration by targeted aquaporin-1 gene disruption. Nature, 434, 786–792 (2005).
• 5)  McCoy E, Sontheimer H. Expression and function of water channels (aquaporins) in migrating malignant astrocytes. Glia, 55, 1034–1043 (2007).
• 6)  Hu J, Verkman AS. Increased migration and metastatic potential of tumor cells expressing aquaporin water channels. FASEB J., 20, 1892–1894 (2006).
• 7)  Papadopoulos MC, Saadoun S, Verkman AS. Aquaporins and cell migration. Pflugers Arch., 456, 693–700 (2008).
• 8)  Hibuse T, Maeda N, Nagasawa A, Funahashi T. Aquaporins and glycerol metabolism. Biochim. Biophys. Acta, 1758, 1004–1011 (2006).
• 9)  Zheng X, Chen X. Aquaporin 3, a glycerol and water transporter, is regulated by p73 of the p53 family. FEBS Lett., 489, 4–7 (2001).
• 10)  Ji C, Cao C, Lu S, Kivlin R, Amaral A, Kouttab N, Yang H, Chu W, Bi Z, Di W, Wan Y. Curcumin attenuates EGF-induced AQP3 up-regulation and cell migration in human ovarian cancer cells. Cancer Chemother. Pharmacol., 62, 857–865 (2008).
• 11)  Levin MH, Verkman AS. Aquaporin-3-dependent cell migration and proliferation during corneal re-epithelialization. Invest. Ophthalmol. Vis. Sci., 47, 4365–4372 (2006).
• 12)  Cao C, Sun Y, Healey S, Bi Z, Hu G, Wan S, Kouttab N, Chu W, Wan Y. EGFR-mediated expression of aquaporin-3 is involved in human skin fibroblast migration. Biochem. J., 400, 225–234 (2006).
• 13)  Loitto VM, Huang C, Sigal YJ, Jacobson K. Filopodia are induced by aquaporin-9 expression. Exp. Cell Res., 313, 1295–1306 (2007).
• 14)  Boury-Jamot M, Sougrat R, Tailhardat M, Le Varlet B, Bonte F, Dumas M, Verbavatz JM. Expression and function of aquaporins in human skin: Is aquaporin-3 just a glycerol transporter? Biochim. Biophys. Acta, 1758, 1034–1042 (2006).
• 15)  Funahashi T, Nagasawa A, Hibuse T, Maeda N. Impact of glycerol gateway molecule in adipocytes. Cell. Mol. Biol. (Noisy-le-grand), 52, 40–45 (2006).
• 16)  Maeda N, Funahashi T, Shimomura I. Metabolic impact of adipose and hepatic glycerol channels aquaporin 7 and aquaporin 9. Nat. Clin. Pract. Endocrinol. Metab., 4, 627–634 (2008).
• 17)  Frigeri A, Nicchia GP, Balena R, Nico B, Svelto M. Aquaporins in skeletal muscle: reassessment of the functional role of aquaporin-4. FASEB J., 18, 905–907 (2004).
• 18)  Huber VJ, Tsujita M, Nakada T. Identification of aquaporin 4 inhibitors using in vitro and in silico methods. Bioorg. Med. Chem., 17, 411–417 (2009).
• 19)  Rojek AM, Skowronski MT, Fuchtbauer EM, Fuchtbauer AC, Fenton RA, Agre P, Frokiaer J, Nielsen S. Defective glycerol metabolism in aquaporin 9 (AQP9) knockout mice. Proc. Natl. Acad. Sci. U.S.A., 104, 3609–3614 (2007).
• 20)  Krause RF, Beamer KC, Plow JH. Phospholipid metabolism in vitamin A-deficient rats. J. Nutr., 101, 161–168 (1971).
• 21)  Graham OL, Kit S. Glycerol metabolism of normal and malignant lymphatic tissue; the preferential labeling of tumor serine and glycine. Cancer Res., 16, 937–942 (1956).
• 22)  Holub BJ, Kuksis A. Differential distribution of orthophosphate-32 P and glycerol-14 C among molecular species of phosphatidylinositols of rat liver in vivo. J. Lipid Res., 12, 699–705 (1971).
• 23)  Hara-Chikuma M, Verkman AS. Prevention of skin tumorigenesis and impairment of epidermal cell proliferation by targeted aquaporin-3 gene disruption. Mol. Cell. Biol., 28, 326–332 (2008).
• 24)  Zheng X, Bollinger Bollag W. Aquaporin 3 colocates with phospholipase d2 in caveolin-rich membrane microdomains and is downregulated upon keratinocyte differentiation. J. Invest. Dermatol., 121, 1487–1495 (2003).
• 25)  Nakahigashi K, Kabashima K, Ikoma A, Verkman AS, Miyachi Y, Hara-Chikuma M. Upregulation of aquaporin-3 is involved in keratinocyte proliferation and epidermal hyperplasia. J. Invest. Dermatol., 131, 865–873 (2011).
• 26)  Ibata K, Takimoto S, Morisaku T, Miyawaki A, Yasui M. Analysis of aquaporin-mediated diffusional water permeability by coherent anti-stokes Raman scattering microscopy. Biophys. J., 101, 2277–2283 (2011).
• 27)  Damadian R, Zaner K, Hor D, DiMaio T. Human tumors detected by nuclear magnetic resonance. Proc. Natl. Acad. Sci. U.S.A., 71, 1471–1473 (1974).
• 28)  Jablonski EM, Mattocks MA, Sokolov E, Koniaris LG, Hughes FM Jr, Fausto N, Pierce RH, McKillop IH. Decreased aquaporin expression leads to increased resistance to apoptosis in hepatocellular carcinoma. Cancer Lett., 250, 36–46 (2007).
• 29)  Anderson AW, Xie J, Pizzonia J, Bronen RA, Spencer DD, Gore JC. Effects of cell volume fraction changes on apparent diffusion in human cells. Magn. Reson. Imaging, 18, 689–695 (2000).
• 30)  Kotz B, West C, Saleem A, Jones T, Price P. Blood flow and Vd (water): both biomarkers required for interpreting the effects of vascular targeting agents on tumor and normal tissue. Mol. Cancer Ther., 8, 303–309 (2009).
• 31)  de Langen AJ, van den Boogaart VE, Marcus JT, Lubberink M. Use of H2(15)O-PET and DCE-MRI to measure tumor blood flow. Oncologist, 13, 631–644 (2008).
• 32)  van der Veldt AA, Hendrikse NH, Harms HJ, Comans EF, Postmus PE, Smit EF, Lammertsma AA, Lubberink M. Quantitative parametric perfusion images using 15O-labeled water and a clinical PET/CT scanner: test–retest variability in lung cancer. J. Nucl. Med., 51, 1684–1690 (2010).
• 33)  de Langen AJ, Lubberink M, Boellaard R, Spreeuwenberg MD, Smit EF, Hoekstra OS, Lammertsma AA. Reproducibility of tumor perfusion measurements using 15O-labeled water and PET. J. Nucl. Med., 49, 1763–1768 (2008).
• 34)  Lodge MA, Jacene HA, Pili R, Wahl RL. Reproducibility of tumor blood flow quantification with 15O-water PET. J. Nucl. Med., 49, 1620–1627 (2008).
• 35)  Lodge MA, Carson RE, Carrasquillo JA, Whatley M, Libutti SK, Bacharach SL. Parametric images of blood flow in oncology PET studies using [15O]water. J. Nucl. Med., 41, 1784–1792 (2000).