Involvement of aquaporins in colorectal carcinogenesis

Abstract

Aquaporins (AQPs) are important in controlling water permeability. As AQP1 is known as a serum-responsive gene, we hypothesized that AQP expression may be involved in the development of human cancer. By reverse transcriptase–polymerase chain reaction analysis, expression of AQPs 1, 3, and 5 was found in seven colon and colorectal cancer cell lines. Western blot analysis confirmed their expression in four of these cell lines. In situ hybridization demonstrated that during colorectal carcinogenesis, the expression of AQPs 1 and 5 was induced in early-stage disease (early dysplasia) and maintained through the late stages of colon cancer development. Expression of AQPs 1 and 5 was maintained even in metastatic lesions in the liver. These findings demonstrate that the expression of several AQPs is found in tumor cells and is associated with an early stage of colorectal cancer development. These novel observations suggest that multiple AQP expression may be advantageous to tumorigenesis, which may lead to a better understanding of colorectal carcinogenesis.

Introduction

The molecular basis of membrane water permeability remained elusive until the recent discovery of aquaporins (AQPs), water channel proteins (Preston et al., 1992; Agre et al., 1995). The fundamental importance of these proteins is suggested by their conservation from bacteria through plants to mammals (King and Agre, 1996; Maurel, 1997; Calamita et al., 1998). Thus far, 10 mammalian AQPs have been identified, each with a distinct tissue distribution (King and Agre, 1996). In the kidney, lung, eye, and brain, multiple water-channel homologues are expressed, providing a network for water transport in those locations, and it was assumed that alterations in AQP expression or function can be rate-limiting for water transport across certain membranes (King and Agre, 1996; King et al., 1997; Agre et al., 1998). Therefore, it was believed that AQPs have some valuable role in human pathophysiology and that changes in such specific pathways have the potential to cause disease or to be the basis of treatment (King et al., 2000). For example, mutations in AQP2 were identified as a cause of nephrogenic diabetes insipidus (Deen et al., 1994). Yet, the significance of AQPs in mammalian systems is frequently doubted, mainly because a human AQP1-knockout and several AQP-knockout mouse models did not show expected clinical phenotypes (Preston et al., 1994; Bai et al., 1999).

Previous reports have shown that AQP1 is one of the protein families involved in cell cycle control known as delayed early-response genes (Moon et al., 1995; Ma and Verkman, 1999). Therefore, we believed that this gene may have a potential role in uncontrolled cell replication, as in cancer (Saadoun et al., 2002). To test our hypothesis, we initially used reverse transcriptase–polymerase chain reaction (RT–PCR) to screen the expression profiles of human AQPs in seven colon and colorectal cancer cell lines, and confirmed our results by Western blot analysis. Based upon these observations, by performing in situ hybridization on 16 colorectal cancer tissue samples, we further studied the expression of AQPs in vivo by using the well-established multistep colorectal carcinogenesis model (Fearon and Vogelstein, 1991). Our results provide a clear and novel example of multiple AQP expression during human colorectal carcinogenesis.

Results

Sequence-specific antisense and sense primers were designed for AQPs 1, 2, 3, 5, and 8. Seven colon and colorectal cancer cell lines were screened with RT–PCR. Of the five AQPs examined, AQPs 1, 3, and 5 were expressed in all the seven cell lines (Figure 1a). AQPs 2 and 8 were not expressed in any of these cell lines (data not shown). These data indicate that, although AQP3 is the only known AQP to be expressed in normal colonic surface epithelium, different AQPs seem to be expressed simultaneously in tumor cells. Western blot analysis demonstrated that AQPs 1, 3, and 5 were expressed concurrently in all four of the tested cancer cell lines (Figure 1b), confirming our RT–PCR results. NIH3T3 cells transfected with AQP1 and AQP3 expression constructs were used as positive controls for the Western blots of AQPs 1 and 3, respectively. 293, the kidney cell line, was used as a positive control for AQP5. In addition, there was a relatively lower expression of AQP3 in the HCT116 p53 −/− clone compared to the HCT116 p53 +/+ clone, consistent with a previous report that p53 promoted the expression of AQP3 (Zheng and Chen, 2001).

Figure 1
figure1

Expression of AQPs 1, 3, and 5 in seven colon and colorectal cancer cell lines. RT–PCR was performed using sense and antisense primers for AQPs 1, 2, 3, 5, and 8 for seven cancer cell lines (W489, HT20, H455, HCT116 p53 +/+, HCT116 p53 −/−, DLD1, SW480). The name of each cell line is marked on the top of each lane, and the size of each PCR product is marked with an arrow. PCR mixtures without templates were used negative controls, and GAPDH was the loading control. Each PCR product was cloned and its sequence confirmed. All seven cell lines demonstrated expression of AQPs 1, 3, and 5(a). Western blot analysis was performed using antibodies for AQPs 1, 3, and 5 for the cell lines HCT116 p53 +/+, HCT116 p53 −/−, DLD1, and SW480. The name of each cell line is marked on the top of each lane, and the size of the protein is marked with an arrow. NIH3T3 cells transfected with AQP1 and AQP3 expression constructs were used as positive controls for the Western blots of AQPs 1 and 3, respectively. 293, a kidney cell line, was used as a positive control for the Western blot of AQP5. NIH3T3 cells were used as a negative control, and β-actin was the loading control. All four cell lines demonstrated AQPs 1, 3, and 5 expression(b)

To confirm these findings in vivo and to evaluate the pattern of AQP induction during carcinogenesis, we performed in situ hybridization with AQP1, AQP3, and AQP5 riboprobes generated by cloning the RT–PCR products. We then tested the expression of these AQPs in colon cancer and in different stages of preneoplastic lesions (Fearon and Vogelstein, 1991). From five colon cancer patients 16 tissue samples were studied, including one patient with metastasis to the liver. Of these samples, 12 consisted of adenocarcinomas and the surrounding normal colonic tissue, while four were metastatic lesions from the liver. Of the five patients, staged surgically resected series were available for three patients. All 12 colon cancer samples showed strong expression of AQPs 1, 3, and 5. AQP1 was strongly expressed in colonic adenoma with almost no staining in surrounding normal colonic mucosa (Figure 2a). In addition, it was consistently expressed in primary colon cancer (Figure 2b) and in metastatic lesion in the liver (Figure 2c). AQP5 was also expressed in early adenoma, late adenoma (Figure 2d, e), and in adenocarcinoma (Figure 2f) with almost no staining in surrounding normal colonic mucosa (Figures 2d, e). The expression of AQP3 in colonic surface epithelium (reviewed by Ma and Verkman, 1999) and expression of AQP1 in the vascular endothelium were previously described (Nielsen et al., 1997; King et al., 2000). The germinal centers of the tonsils have also been shown to express AQPs 1 and 5 (Moon et al., 2003, manuscript in review). As internal controls for in situ hybridization, slides of the germinal centers of the tonsils were probed with antisense riboprobes of AQP1 (data not shown) and AQP5 (Figure 2g) and sense riboprobes of AQP1 (data not shown) and AQP5 (Figure 2h). The vascular endothelium was stained with antisense riboprobes of AQP1 (Figure 2i). The expression of AQP3 was maintained in colon cancer and in different stages of preneoplastic lesions (data not shown). These findings strongly suggest that at least two different AQPs are induced in the early stage of colorectal carcinogenesis.

Figure 2
figure2

In situ hybridization with AQP1 and AQP5 probes in the multistep carcinogenesis model of colon cancer development. The AQP1 antisense riboprobe clearly stains not only colonic adenoma (a, arrow) but also the primary colon cancer (b, arrow) and metastatic lesions in the liver (c, arrow). Likewise, the AQP5 antisense riboprobe clearly stains early adenoma with moderately dysplastic cells (d, arrow), late adenoma with severe dysplastic cells (e, arrow), and adenocarcinoma (f, arrow). There is almost no staining in the surrounding normal colonic mucosa when probed with AQP1 and AQP5 antisense riboprobes (a and d, star). As positive controls, the germinal centers of the tonsils were stained with antisense riboprobes of AQP1 (data not shown) and AQP5 (g). Sense riboprobes of AQP1 (data not shown) and AQP5 (h) were used as negative controls. As an internal control, vascular endothelium was stained with the AQP1 antisense riboprobe (i)

Discussion

This report provides a novel example of the involvement of water channel proteins in human colon cancer. Thus far, AQP3 is the only AQP known to be expressed in normal colonic surface epithelium (reviewed by Ma and Verkman, 1999). Therefore, finding expression of multiple AQPs in both colorectal cancer cell lines and tissue samples was unexpected. More surprisingly, AQPs 1 and 5 were found to be expressed in the early stage of colon cancer development (mild dysplasia); based upon RT–PCR data from cell lines, we had expected AQP expression to occur exclusively in the carcinoma cell itself.

These findings raise many possibilities as to the role of AQPs during colorectal carcinogenesis. It is possible that induction of AQPs is required in the early stage of colorectal cancer development to function as one of the essential driving forces for initiating carcinogenesis. Considering prior promoter study results, expression of AQP1 in the early dysplasia of colon adenoma could have been induced by the binding of immediate early-response gene products to AQP1 binding sites of the human AQP1 promoter as a result of the expression of an oncogene like Fos/Jun (Moon et al., 1993, 1997).

In addition, the expression of at least three AQPs in colon and colorectal cancer cell lines suggests the possibility that tumor cells may need multiple AQPs to gain advantages for replication or survival. During the cell cycle, as cell volume needs to increase rapidly by absorbing water from the outside with a minimal amount of energy, expression of several types of AQPs in the tumor cells is likely to be advantageous compared with normal cells having a single type of AQP expression. Also, tumor cells may require several AQPs for high metabolic turnover or tumor-specific metabolic pathways needed for survival. In fact, there is evidence suggesting that AQPs function as gas channels (Nakhoul et al., 1998), and it is possible that by accumulating mutations inside the critical motifs in their channel structures, some of these AQPs can gain novel functions other than of a water channel (King and Agre, 1996; Heymann et al., 1998).

These questions of whether the expression of multiple AQPs is a driving force or only an associated phenomenon in colorectal carcinogenesis can be at least partially answered by transfecting overexpression or antisense expression constructs (Splinter et al., 2002). In fact, through a recent functional study, we have identified preliminary evidence that AQP1 has oncogenic properties. Ectopic expression of full-length cDNA of AQP1 induced many phenotypic changes characteristic of transformation, including cell proliferation enhancing activity and anchorage-independent growth (Moon and Hoque, 2003, data not yet published).

Based upon our observations, we propose that changes in expression levels of several AQPs may play a role in tumorigenesis, and we provide evidence for the potential role of AQPs in human pathophysiology (Marples, 2000). We speculate that studying the expression status or modifying the expression of these AQPs may lead to the design of novel ways of diagnosing or controlling human cancers.

Materials And Methods

Cell lines

Three colorectal cancer cell lines (HCT116 p53 +/+, HCT116 p53 −/−, DLD1) were maintained in McCoy's 5A modified medium with 10% fetal bovine serum (FBS), and the colon cancer cell line SW480 was maintained in Leibovitz's L-15 medium with 10% FBS. Other colon cancer cell lines (W489, HT20, H455) were maintained in Dulbecco's modified Eagle's medium with 5% FBS.

Tissues

In all, 16 premalignant and malignant specimens of colon epithelium were archived tissue samples of surgically resected lesions from five patients treated at Hospital Tenon, Paris, France. In one patient, metastatic lesions in the liver were also available. Of these, 12 samples were comprised of adenocarcinomas and the normal surrounding colonic tissue, while four were metastatic lesions from the liver. Of the five patients studied, a complete staged series consisting of normal colon epithelium, mild and moderate dysplasia (early adenoma), severe dysplasia (late adenoma), and invasive adenocarcinoma tissues were available for three patients.

RT–PCR analysis of AQPS expression in human tumor cell lines

For the cell lines W489, HT20, and H455, total RNA extracted using an RNeasy total RNA kit (Qiagen) from cell lines was reverse transcribed by random primer and Superscript II reverse transcriptase (Gibco/BRL). For the cell lines HCT116 p53 +/+, HCT116 p53 −/−, DLD1, and SW480, total RNA was extracted using Trizol reagent (Invitrogen) and reverse transcribed using random primers and M-MLV reverse transcriptase (Invitrogen). The resulting cDNA was subjected to PCR with exon spanning primers of AQP1, AQP2, AQP3, AQP5, and AQP8. The AQP1, AQP2, and AQP5 products were amplified using sequences spanning part of exons 1 and 2, AQP3 product was from part of exons 2 and 3, and AQP8 product was from part of exon 2, all of exon 3 and part of exon 4. The following are the primer sequences and the sizes of each PCR: AQP1, 5′-IndexTermCGCAGAGTGTGGGCCACATCA and 5′-IndexTermCCCGAGTTCACACCATCAGCC, amplifying a 217-bp product; AQP2, 5′-IndexTermTGGCACGGTGGTACAGGCTCT and 5′-IndexTermGCCGTCGTGCTGTTGCTGAG, amplifying a 227-bp product; AQP3, 5′-IndexTermGCTGTCACTCTGGGCATCCTG and 5′-IndexTermGCGTCTGTGCCAGGGTGTAG, amplifying a 131-bp product; AQP5, 5′-IndexTermCGTTTGGCCTGGCCATAGGCA and 5′-IndexTermTGGCCCTGCGTTGTGTTGTTG, amplifying a 247-bp product; AQP8, 5′-IndexTermCGTGATTGCCACGCTGGGGAA and 5′-IndexTermCACCTCAGGTCCAAAAGCACG, amplifying a 423-bp product; and GAPDH, 5′-IndexTermGAAATCCCATCACCATCTTCCAGG and 5′-IndexTermCATGTGGGCCATGAGGTCCACCAC, amplifying a 782-bp product. For HCT116 p53+/+, HCT116 p53 −/−, DLD1, and SW480, different primers were used for AQP3. The primer sequences are as follows: 5′-IndexTermGCCGGATCCGCCGCCGCCATGGGTCGACAGAAG and 5′-IndexTermGCCTCTAGACTATCAGATCTGCTCCTTGTGC, amplifying an 885-bp product. PCR was performed in a DNA thermal cycler (Perkin-Elmer Co.). For W489, HT20, and H455, we performed 40 cycles using the following conditions: 10 s at 94°C, 50 s at 63°C, and 50 s at 72°C. β-Actin mRNA amplification was performed on the cDNA as a positive control for reaction efficiency and was performed for 25 cycles. For HCT116 p53+/+, HCT116 p53−/−, DLD1, and SW480, the following conditions were used: AQP1, 35 cycles for 1 min at 95°C, 1 min at 64°C, and 2 min at 72°C; AQP3, 33 cycles for 4 5 s at 95°C, 45 s at 60° C, and 1 min and 35 s at 72°C; for AQP5, 35 cycles for 1 min at 95°C, 1 min at 61°C, and 1 min at 72°C; and for GAPDH, 24 cycles for 1 min at 95°C, 1 min at 60°C, and 1 min and 30 s at 72°C. All of the PCR products were introduced into the pCR®II-TOPO plasmid vector (Invitrogen) using the TOPO TA Cloning System and each AQP gene-specific sequence was confirmed. Water was used as a negative control.

Western blot analysis

Phosphate-buffered saline (PBS)-washed cells were lysed by sonication in ice-cold RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM PMSF, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, and 0.1% SDS). Protein concentrations were determined by the Bio-Rad DC protein assay (Bio-Rad). Equal amounts of protein (50 μg) were subjected to 14% SDS–polyacrylamide gel electrophoresis, and the proteins were transferred onto nitrocellulose membranes. After 1 h blocking with 5% nonfat dry milk in 1 × PBS at room temperature, the membranes were probed with rabbit anti-AQP1 antibody (Chemicon International), goat anti-AQP3 antibody, or goat anti-AQP5 antibody (Santa Cruz Biotechnology), all at 1 : 100 dilutions in 1 × PBS for 1 h at room temperature. After washing three times for 5 min each in 1 × PBS, the membranes were hybridized for 1 h at room temperature with 1 : 5000 dilutions of the secondary antibody (HRP-conjugated anti-rabbit antibody (Amersham) and HRP-conjugated anti-goat antibody (Santa Cruz Biotechnology)). To ensure equal loading and transfer of proteins, the filters were probed with β-actin (Sigma-Aldrich) at a 1 : 5000 dilution. Detection was performed using the enhanced chemiluminescence system (Amersham). NIH3T3 cells transfected with AQP1 and AQP3 expression constructs were used as positive controls for the Western blots of AQPs 1 and 3, respectively. H293, a normal kidney cell line, was used as a positive control for the Western blot of AQP5. NIH3T3 cells were used as a negative control.

Probe construction for in situ hybridization

Human AQP1 complementary DNA samples were obtained by PCR using the same primers as those used to obtain the RT–PCR product in Figure 1a. Human AQP5 complementary DNA was obtained by PCR using sense primer, 5′-IndexTermCGTTTGGCCTGGCCATAGGCA, and antisense primer, 5′-IndexTermCGATTCATGACCACCGCAGGG. The cDNA specificity was confirmed by gel sequencing. The cDNA was introduced into the plasmid pCR®II-TOPO, and this construction was used as a template to generate sense and antisense probes. During transcription, nonradioactive labeling of single-strand RNA probes was performed using digoxigenin-UTP (DIG RNA labeling kit, Boehringer Mannheim, Germany). The probes were divided into aliquots with RNAse inhibitor and stored at −80°C until use.

In situ hybridization of tissue sections with AQPS riboprobes

For each case tested, paraffin-embedded tissue sections 4 (4 μm) were cut onto silane-coated slides (Sigma). The sections were deparaffinized in xylene, rehydrated in gradually decreasing concentrations of ethanol, and treated with 0.2 N HCl. Sections were then treated with proteinase K for 15 min at 37°C. Sections were washed thrice with 1 × PBS, postfixed in 4% paraformaldehyde for 5 min at room temperature, and rerinsed with 1 × PBS. Then sections were acetylated in 0.25% acetic anhydride and 0.1 M triethanolamine for 10 min. The sections were dehydrated in gradually increasing concentrations of ethanol and air-dried prior to hybridization. Sections were prehybridized for 1 h at 42°C in hybridization buffer (20 × SSC, 50% deionized formamide, 2.5 mg of predenatured salmon sperm DNA, 1 g of dextran sulfate, 2% 100 × Denhart's solution, 2% DTT, and 4 mg of yeast tRNA). Hybridization was performed at 42°C for 4 h in hybridization buffer containing 400 ng/ml of the probe. The tissue was twice washed in 2 × SSC for 5 min at room temperature and then treated with RNase A (40 μg/ml) and RNase T1 (10 U/ml) in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.5 M NaCl for 30 min at 37°C. Then sections were incubated with agitation for 2 h with 2 × SSC containing 0.05% Triton X-100 and 2% normal sheep serum at room temperature. The sections were rinsed in buffer 1 (0.1 M maleic acid and 0.15 M NaCl, pH 7.5) for 5 min at room temperature and then incubated with 2% normal sheep serum, 0.3% Triton X-100 in buffer 1 for 30 min also at room temperature. Slides were then incubated overnight at 4°C with antidigoxigenin antibody (1 : 500). After two rinses in buffer 1, slides were rinsed briefly in buffer 3 (100 mM Tris-HCl, 100 mM NaCl, and 50 mM MgCl2, pH 9.5). Alkaline phosphatase was detected using 5-bromo-4-chloro-3-indolyl phosphate and nitro-blue tetrazolium chloride. Slides were then rinsed in buffer 3 (10 mM Tris-HCl, and 1 mM EDTA, pH 8) and mounted with aqua mounting medium (Fischer). Sections pretreated with RNase to prevent hybridization or incubated with the digoxigenin-labeled sense probe in the same conditions were used as negative controls.

References

  1. Agre P, Bonhivers M and Borgnia MJ . (1998). J. Biol. Chemother., 273, 14659–14662.

  2. Agre P, Brown D and Nielsen S . (1995). Curr. Opin. Cell Biol., 7, 472–483.

  3. Bai C, Fukuda N, Song Y, Ma T, Matthay MA and Verkman AS . (1999). J. Clin. Invest., 103, 555–561.

  4. Calamita G, Kempf B, Bonhivers M, Bishai WR, Bremer E and Agre P . (1998). Proc. Natl. Acad. Sci. USA, 95, 3627–3631.

  5. Deen PMT, Verdijk MAJ, Knoers NVA, Wieringa B, Monnens LAH, van Os CH and van Oost BA . (1994). Science, 264, 92–94.

  6. Fearon ER and Vogelstein B . (1991). Cell, 61, 759–762.

  7. Heymann JB, Agre P and Engel A . (1998). J. Struct. Biol., 121, 191–206.

  8. King LS and Agre P . (1996). Annu. Rev. Physiol., 158, 619–648.

  9. King LS, Nielsen S and Agre P . (1997). Am. J. Physiol., 273, C1541–C1548.

  10. King LS, Yasui M and Agre P . (2000). Mol. Med. Today, 6, 60–65.

  11. Ma T and Verkman AS . (1999). J. Physiol., 517, 317–326.

  12. Marples D . (2000). Lancet, 355, 1571–1572.

  13. Maurel C . (1997). Annu. Rev. Plant Physiol. Plant Mol. Biol., 48, 399–429.

  14. Moon C, King LS and Agre P . (1997). Am. J. Physiol., 273, C1562–C1570.

  15. Moon C, Preston GM, Griffin CA, Jabs EW and Agre P . (1993). J. Biol. Chem., 268, 15772–15778.

  16. Moon C, Williams JB, Preston GM, Copeland NG, Gilbert DJ, Nathans D, Jenkins NA and Agre P . (1995). Genomics, 30, 354–357.

  17. Nakhoul NL, Davis BA, Romero MF and Boron WF . (1998). Am. J. Physiol., 274, C543–C548.

  18. Nielsen S, King LS, Christensen BM and Agre P . (1997). Am. J. Physiol., 273, C1549–C1561.

  19. Preston GM, Carroll TP, Guggino WB and Agre P . (1992). Science, 256, 385–387.

  20. Preston GM, Smith BL, Zeidel ML and Moulds JJ . (1994). Science, 265, 1585–1587.

  21. Saadoun S, Papadopoulos MC, Davies DC, Bell BA and Krishna S . (2002). Br. J. Cancer, 87, 621–623.

  22. Splinter PL, Masyuk AI and LaRusso NF . (2002). J. Biol. Chem., 278, 6268–6274.

  23. Zheng X and Chen X . (2001). FEBS Lett., 489, 4–7.

Download references

Acknowledgements

We thank Dr Li K Su, who provided total RNA for the three colon cell lines used in this study. The study was supported in part by NIH/NCI Grant P50 CA96784-01 (to CM), American Cancer Society Grant RPG-98-054 (to LM), Fondation de France, AP-HP and Lilly Fondation Grant (to J-CS), Cancer Center Grant P30 CA 16620 (to MD Anderson Cancer Center), and Tobacco Research Fund from State of Texas (to MD Anderson Cancer Center).

Author information

Affiliations

Authors

Corresponding author

Correspondence to Chulso Moon.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Moon, C., Soria, JC., Jang, S. et al. Involvement of aquaporins in colorectal carcinogenesis. Oncogene 22, 6699–6703 (2003). https://doi.org/10.1038/sj.onc.1206762

Download citation

Keywords

  • aquaporin
  • in situ hybridization
  • RT–PCR
  • colon cancer
  • carcinogenesis

Further reading

Search

Quick links