Fax: (706) 721-6608
Expression of cyclic guanosine monophosphate-dependent protein kinase in metastatic colon carcinoma cells blocks tumor angiogenesis
Article first published online: 7 FEB 2008
Copyright © 2008 American Cancer Society
Volume 112, Issue 7, pages 1462–1470, 1 April 2008
How to Cite
Kwon, I.-K., Schoenlein, P. V., Delk, J., Liu, K., Thangaraju, M., Dulin, N. O., Ganapathy, V., Berger, F. G. and Browning, D. D. (2008), Expression of cyclic guanosine monophosphate-dependent protein kinase in metastatic colon carcinoma cells blocks tumor angiogenesis. Cancer, 112: 1462–1470. doi: 10.1002/cncr.23334
- Issue published online: 19 MAR 2008
- Article first published online: 7 FEB 2008
- Manuscript Accepted: 23 OCT 2007
- Manuscript Revised: 27 SEP 2007
- Manuscript Received: 18 JUN 2007
- American Cancer Society. Grant Number: RSG-07-174-01 CSM
- protein kinase;
- cyclic guanosine monophosphate;
Type 1 cyclic guanosine monophosphate (cGMP)-dependent protein kinase (PKG) reportedly has exhibited antitumor properties, and its expression is down-regulated in many tumors.
The authors recently demonstrated that PKG re-expression in metastatic colon carcinoma cells results in decreased tumorigenesis: In the current study, they addressed that mechanism.
Over-expression of PKG in SW620 cells produced smaller, more apoptotic subcutaneous tumors in athymic mice, but the observed effect of PKG expression on growth and apoptosis in vitro was minimal. Closer examination of the subcutaneous xenografts revealed highly vascular tumors produced by the parental SW620 cells, which contrasted greatly with the PKG-expressing tumors, in which cell growth was limited to “islands” surrounding CD31-positive cells. The idea that PKG expression was associated with reduced tumor angiogenesis was supported by decreased levels of vascular endothelial growth factor in these tumors compared with tumors that were derived from parental SW620 cells. Investigation of potential mechanisms revealed that PKG expression was associated with reduced levels of β-catenin compared with parental cells. Moreover, this effect of exogenous PKG on β-catenin expression in SW620 cells also occurred in vitro, where the decrease was associated with reduced T-cell factor-dependent transcription.
Together the findings indicated that PKG down-regulation in colon cancer cells is important for optimal tumor angiogenesis and that regulation of β-catenin expression may be important to this process. Cancer 2008. © 2008 American Cancer Society.
Colorectal carcinoma (CRC) is one of the most destructive cancers in industrialized countries, and the identification of new therapeutic targets remains an important endeavor. It has been suggested that a lower incidence of CRC in developing countries is because of the prevalence of enterotoxigenic Escherichia coli infection in these regions.1, 2 It is believed that the protective effect arises from heat-stable enterotoxins (STa) produced by these bacteria, which mimic endogenous guanylin/uroguanylin by binding to receptor guanylyl-cyclase (GC-C), leading to increased cellular cGMP.3 In the intestinal epithelium, guanylin binding to GC-C is a classic regulator of fluid and electrolyte homeostasis, but it also can contribute to the regulation of proliferation/differentiation along the crypt-villus axis.4, 5 Proliferation in the colonic crypt is controlled by the Wnt/β-catenin pathway, and aberrant activation of this system is a key initiating event for intestinal tumorigenesis.6 In normal cells, cytosolic β-catenin associates with the adenomatous polyposis coli (APC) complex, leading to it's phosphorylation by glycogen synthase kinase 3β (GSK-3β) and subsequent ubiquitination and degradation in proteasomes.7, 8 Most colorectal tumors have either spontaneously derived or inherited, truncating mutations in APC that render it unable to bind β-catenin.9, 10 Increased levels of β-catenin in these cells promote interaction with T-cell factor (TCF)-related transcription factors that enter the nucleus to activate the expression of growth and angiogenesis related target genes.8, 11
In T84 colon cancer cells that express GC-C, it has been reported that cGMP has cytostatic effects that result from activation of cyclic-nucleotide gated calcium channels.12 Although it remains to be determined how widespread this cytostatic effect is, the notion that cGMP has antitumor properties is supported by independent studies with exisulind (Aptosyn; a sulindac derivative). This nonsteroidal anti-inflammatory drug induces apoptosis in diverse tumors,13–15 particularly in the colon, where it is believed that the mechanism involves increasing intracellular cGMP levels by inhibiting phosphodiesterases.16–18 In contrast to the antitumor mechanism of the uroguanylin/STa system,2, 19 work with exisulind in several colon cancer cell lines has underscored a proapoptotic role for PKG.15, 16, 20 Although those in vitro studies have resulted in some debate concerning the roles of cGMP and PKG,21 the potential antitumor effects of this pathway remain an important area of investigation.
Our laboratory recently reported that type 1 PKG is expressed in normal colon epithelium but is greatly reduced in colon tumors.22 That study also demonstrated that the silencing of PKG expression is likely to be necessary for tumor progression, because expression of PKG in a colon cancer cell line blocked tumor growth and invasiveness in athymic mice.22 In the current work, we examined the effects of ectopic PKG in tumors further and demonstrated that PKG-expressing xenografts are defective in angiogenesis. In addition, the results suggested that a PKG-dependent reduction of β-catenin levels is a likely mechanism.
MATERIALS AND METHODS
The mifepristone, 4,6-diamidino-2-phenylindole (DAPI), and 8-Br-cGMP were obtained from Calbiochem (San Diego, Calif). NP-40 and Tween-20 were obtained from Sigma (St. Louis, Mo); and, unless specified, all other chemicals were obtained from Fisher Scientific (Pittsburgh, Pa). The antibodies against poly-ADP-ribose polymerase (cleaved PARP), and β-catenin (N-terminus) were obtained from Cell Signaling (Beverly, Mass), and the β-actin antibody was obtained from Sigma. The antivascular endothelial growth factor (anti-VEGF) antibody was obtained from AbCam (Cambridge, Mass). The polyclonal anti-PKG antibodies were raised in rabbits against a peptide corresponding to type 1 PKG C-terminus and have been described previously.23
Tissue Culture and Growth Assays
We previously described the creation of inducible SW620 colon carcinoma cell lines for PKG expression using the Gene-Switch system (Invitrogen).22 These clones respond to nanomolar levels of mifepristone with induction of type 1 PKG expression. The cell lines were maintained at atmospheric CO2 levels in L15 medium containing 10% fetal bovine serum and 300 μg/mL each of hygromycin and zeosin. In vitro measurement of growth and apoptosis was performed by using cells that were seeded at low density (approximately 20% confluence) on 6-well plates using medium in which hygromycin and zeosin were excluded. The growth medium was replaced every 48 hours with fresh medium with or without 10 nM mifepristone/100 μM 8-Br-cGMP. At intervals, the cells were rinsed in phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde (for 10 minutes), extracted in 0.1% NP-40, then stained with DAPI (1 μg/mL; 1 minute). The relative number of nuclei was measured by quantification of DAPI fluorescence in triplicate wells. Apoptosis in cell populations treated with mifepristone for 48 hours was measured by Western blot analysis to detect caspase-mediated cleavage of PARP (see below).
Assessment of SW620 growth as xenografts followed animal protocols that were approved by the Institutional Animal Care and Use Committee at the Medical College of Georgia and were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Flank tumors were generated by subcutaneously injection of 107 SW620 cells in 100 μL of Hank balanced salt solution into the right flank of athymic BALB/c (nu/nu) mice (Charles River Laboratories, Wilmington, Mass) for n = 6 per group. To induce PKG expression in vivo, mifepristone was administered to animals that harbored either parental or PKG-inducible SW620 cells every 48 hours by intraperitoneal injections (100 μL of 5 μM in olive oil), as described previously.22 Tumors were harvested for analysis after 25 days.
Histologic Analysis of Xenografts
Excised tumors were fixed in 10% formalin and stored for at least 1 day in ethanol before paraffin embedding, sectioning, and staining with hematoxylin and eosin (MCG histology core). For immunohistology, the slides were deparaffinized with 3 xylene washes followed by rehydration using a series of alcohol dilutions. Antigen unmasking for staining with anti-β-catenin and anti-CD31 antibodies was achieved by boiling for 10 minutes in citrate buffer (0.01 M sodium citrate, pH 6.5) in a microwave oven. Slides were allowed to cool for 20 minutes, washed in PBS, then blocked with 5% calf serum in PBS containing 0.05% Tween 20 (PTS buffer) at 37°C for 1 hour, and incubated either with 1:100 antihuman β-catenin N-terminus antibodies (Cell Signaling) or with anti-CD31/platelet endothelial cell adhesion molecule (PECAM) antibodies (goat; Santa Cruz Biotechnology Inc., Santa Cruz, Calif) overnight at 4°C. The specific signal was detected by using a peroxidase-antiperoxidase (Sternberger Monoclonals Inc., Luthersville, Md) amplification system and was observed by using an indocarbocyanine (Cy3)-conjugated secondary antibody (Jackson ImmunoResearch Laboratories). Immunohistology of cleaved caspase 3 (1:100; Cell Signaling) or PKG (1:50; described previously24) required antigen unmasking with chymotrypsin treatment, as described in detail elsewhere (AbCam Inc.). Visualization of these antibodies was accomplished with the Cy3-conjugated secondary antibody directly without amplification. For confocal imaging, a Zeiss 510 laser-scanning confocal microscope was used (Zeiss LTM software). Nonspecific signal was determined by processing parallel slides without the primary antibody, and these slides were used as a basis for establishing exposure times for each experiment. Once established, the same microscope settings were used for the remaining slides in that experiment.
Western Blot Analysis
Cells cultured in vitro were placed on ice, and the medium was replaced with 250 μL ice-cold lysis buffer (50 mM N-[2-hydroxethyl] piperazine-NA-2-ethanesulfonic acid, pH 8.0; 150 mM NaCl; 1% Nonidet P-40; 0.25% deoxycholate) supplemented with phosphatase and protease inhibitor cocktails (Calbiochem, La Jolla, Calif). Dissected tumors were dissociated in lysis buffer by using a tissue homogenizer (Omni International, Marietta, Ga). Cell suspensions were agitated for 20 minutes at 4 °C followed by clarification of the extracts by centrifugation. For electrophoretic analysis, 30 μL of homogenates were mixed with 10 μL 5 × polyacrylamide gel electrophoresis sample buffer, boiled for 10 minutes, then 10 μL were loaded per lane. Electrophoresis of proteins was performed on 10% mini-gels followed by electrophoretic transfer to nitrocellulose. The blots were blocked with 5% bovine serum albumin in PTS buffer for 20 minutes at room temperature; then, antibodies were added overnight at 4 °C. After addition of 1:3000 peroxidase-conjugated secondary antibody (Bio-Rad Laboratories, Hercules, Calif) for 1 hour, the bands on the blots were observed by using chemiluminescence according to the manufacturer's instructions (Pierce, Rockford, Ill). At least 3 washes (5 minutes each) using excess PTS buffer were performed between each incubation step.
Quantification of β-Catenin/TCF-dependent Transcription
For TCF-dependent gene transcription activity, the TOPflash system was used according to the manufacturer's protocols (Promega). The assays were performed by cotransfecting cells (in triplicate) in 12-well plates using Lipofectamine 2000 reagent (Invitrogen) with 0.2 μg TOPflash luciferase reporter per well (or mutant control FOPflash vector) and 0.2 μg cytomegalovirus/β-galactosidase control. After 16 hours, the medium was changed, and cells were stimulated with 100 μM 8-Br-cGMP for an additional 6 to 8 hours before enzyme assays. Cell extracts were prepared and analyzed for luciferase and β-galactosidase activities as described elsewhere.25 The luciferase activity was standardized by dividing by the respective β-galactosidase activity, and the TCF-specific luciferase was expressed as the net TOPflash luciferase after subtraction of the associated FOPflash activity.
Apoptosis of Metastatic Colon Carcinoma Cells Is Associated With PKG Expression in Tumors but Not When Grown in Vitro
Several in vitro studies with normal and cancer cell lines have described potentially antitumorigenic effects of the cGMP/PKG pathway16–18, 26, 27; however, to our knowledge, our recent report of decreased PKG expression in colon tumors is the first in vivo evidence.22 The same report demonstrated that re-expression of PKG in SW620 colon carcinoma cells caused reduced tumor growth and invasiveness in athymic mice. In the current work, we sought to determine the mechanism underlying the antitumor properties of PKG in vivo with a more detailed analysis of SW620 xenografts engineered for inducible PKG expression. The ability of PKG expression to affect the tumor growth was tested by using a mifepristone-inducible clone (referred to as clone H3Z6b). The growth kinetics of xenografts derived from parental SW620 colon cancer cells was almost identical to the H3Z6b clone of SW620 cells in the absence of inducer (Fig. 1). After a 2-week lag phase with slower increases in tumor volume, the growth rates increased dramatically during the remainder of the study period. Induction of PKG expression in the H3Z6b cells with mifepristone treatment significantly reduced the postlag increase in xenograft growth (P = .017), such that the final tumor volumes were approximately 50% of the volumes in the uninduced H3Z6b cells and the parental SW620 cells. Histologic observation of hematoxylin and eosin-stained sections from subcutaneous tumors derived from parental and PKG-expressing SW620 cells revealed striking differences (Fig. 1). The “parental tumors” were solid and homogeneous, and this contrasted dramatically with the “PKG tumors,” which were fragile and generally less organized. Higher magnifications revealed that the SW620 cells in “parental tumors” grew as large, parallel layers of multinucleate columnar cells, whereas growth in the PKG tumors appeared to be restricted to “islands” with a clearly visible lumen (compare Fig. 1b,c with Fig. 1e,f).
The morphology of the PKG-expressing tumors ostensibly agreed with previous in vitro studies, which indicated a proapoptotic role for PKG in some colon cancer cell lines.16, 26–28 Therefore, apoptosis in our xenografts was examined by Western blot analysis to detect PARP cleavage as a measure of caspase activation in extracts from the excised tumors (Fig. 2A). This approach revealed that tumors induced to express PKG underwent notably more apoptosis (PARP cleavage product) than the parental tumors. Immunofluorescent staining of tumor sections for activated caspase 3 revealed that the majority of apoptotic cells were located in the tissue between the growth islands in the PKG-expressing tumors, but specific staining was absent in the parental tumors (Fig. 2B). This observation was strengthened by the pycnotic nuclei observed in these intervening regions by hematoxylin and eosin staining (see Fig. 1f). Although it was more concentrated in growth islands, the expression of PKG in these tumors was variable and also was detected in the intervening regions.
To further investigate the effect of PKG expression on apoptosis, we grew several clones of inducible SW620 cells in vitro and observed that the basal level of PARP cleavage was elevated only slightly compared with the parental cells (Fig. 3A). However, comparing the ratio of full-length PARP to cleavage product, this slight increase in apoptosis observed in cells that expressed PKG in vitro was not comparable to the level observed in the same cells that were growing as tumors. To determine whether the slight increase in apoptosis observed in vitro may have cumulative effects that lead to more extensive PARP cleavage in the longer term, we compared the growth of parental and PKG-expressing SW620 cells over a 7-day period in tissue culture (Fig. 3B). Although a slight decrease in growth rate was noted in the PKG-expressing cells compared with the uninduced or parental cells, this difference was not significant.
PKG Expression Blocks Angiogenesis in Subcutaneous Xenografts
The striking difference between the effects of PKG expression on SW620 cell growth and apoptosis in vitro, compared with growth as subcutaneous xenografts, suggested a role for PKG in the interaction of tumor cells with host tissues. One of the most important interactions between tumor and host tissues is angiogenesis, which is essential for efficient tumor progression. This idea led to the hypothesis that the growth islands observed in the PKG-expressing xenografts were viable tumor cells surrounding blood vessels and that more apoptosis was observed at more distant sites. Because CD31 (PECAM) is a well characterized vascular marker, we stained tumor sections with CD31-specific antibodies to reveal blood vessels (Fig. 4A). Microscopic analysis of the tumor vasculature revealed several differences between the parental and PKG-expressing tumors; the most notable was the lack of CD31 staining in the PKG tumors, which was abundant throughout the tumors derived from parental SW620 cells. The CD31 staining observed in the PKG-expressing xenografts, as hypothesized, corresponded to the luminal surface of the “islands” of cancer cell growth, but surrounding regions appeared to be devoid of capillaries, which were abundant and distributed evenly in the parental tumors. A key mediator of tumor angiogenesis in colorectal cancer is VEGF. By using Western blot analysis of whole tumor extracts, we were able to detect VEGF protein in all of the tumors. Although there was significant variation between tumors from each group, VEGF clearly was more abundant in the parental SW620 tumors compared with the PKG-expressing tumors when normalized to the β-actin loading control (Fig. 4B). Because the anti-VEGF antibody used was not species-specific, reverse transcriptase-polymerase chain reaction analysis was used to confirm that the VEGF protein detected was derived from the human SW620 cells and not from mouse host tissues (not shown).
PKG Down-regulates β-Catenin Function in Colon Cancer Cells
A common denominator of colon tumors is elevated β-catenin signaling, which is associated with increased TCF-dependent gene expression.29, 30 Recent studies have identified several TCF binding sites in the VEGF promoter and have determined that elevated β-catenin signaling is important to VEGF-A synthesis in colorectal tumors from both mice and humans.31, 32 Because it has been suggested that β-catenin down-regulation is central to the effects of exisulind on SW480 cells,26 it is possible that the effects of PKG on angiogenesis in the current studies also may involve β-catenin. To test this idea, Western blot analysis was used to measure β-catenin expression in the homogenates from SW620 xenografts (Fig. 5). This work revealed that β-catenin was present in all the tumors, but the levels were notably less in most of the PKG-expressing specimens compared with the parental controls (Fig. 5A). Immunofluorescence analysis of xenografts supported this finding and further revealed that the β-catenin expressed in the PKG tumors was predominantly in the growth islands. This contrasted with the β-catenin in the parental tumors, which was more abundant and was distributed uniformly (Fig. 5B).
It was believed that angiogenesis in colorectal cancer relied on VEGF production as a direct result of elevated β-catenin/TCF signaling in these tumors, as discussed above. Therefore, in the current study, the decreased β-catenin levels in the PKG-expressing tumors could contribute to the reduced angiogenesis in the SW620 xenografts. Although it was difficult to address directly in the tumors, we sought to determine whether PKG expression could reduce β-catenin levels and associated TCF-dependent transcriptional activity in SW620 cells grown in vitro. It was discovered that induction and activation of PKG in several different SW620 cell clones resulted in a reduction in the level of β-catenin protein (Fig. 6A). Moreover, when transfected with a β-catenin/TCF-responsive luciferase reporter, we observed that activation of PKG resulted in a 60% inhibition of the basal activity in the inducible SW620 cells (Fig. 6B). It is noteworthy that the addition of 8Br-cGMP alone or the induction of PKG in the absence of activation did not inhibit TCF-reporter activity.
There have been several reports of antiproliferative or proapoptotic effects of cGMP signaling in cell lines grown in vitro; however, none of those reports specifically addressed PKG function in cancer. We recently reported that type 1 PKG is silenced in colorectal tumors compared with normal tissue from matched specimens and that PKG overexpression in metastatic SW620 colon cancer cells results in reduced tumor growth.22 The current work has further examined the effects of PKG expression in SW620 xenografts to better understand the antitumor mechanism.
One of the most interesting observations in this study was the striking difference in the effect of PKG on SW620 cells in vitro compared with growth as xenografts in vivo. Our results demonstrate that PKG does not directly affect apoptosis in SW620 cells and are consistent with previous work, which identified a role for PKG in anoikis-related apoptosis but not under normal tissue culture conditions.33 However, these results ostensibly conflict with other studies, which documented proapoptotic or cytostatic effects of PKG in colon cancer cells grown in vitro.16, 26–28 For example, work with phosphodiesterase inhibitors (eg, exisulind) has demonstrated that PKG can induce apoptosis through several pathways that include Janus kinase and β-catenin.17 More recently, however, the proapoptotic effects of the cGMP/PKG pathway in the context of this drug have been contested; consequently, the role of PKG in these cells remains controversial.34 Another line of investigation that centers on activation of the guanylyl-cyclase C receptor has described cytostatic effects of cGMP in T84 colon cancer cells.2, 4 That work has suggested that calcium dysregulation is part of the mechanism, but the role of PKG in that system is not fully understood. Our measurements did not reveal any effect of PKG overexpression or of physiologic levels of exogenous 8Br-cGMP on the growth of SW620 cells in vitro. The reason for this discrepancy is not clear but may reflect the different cell types used, because apoptosis has been observed mostly in nonmetastatic cell lines, such as SW480, but not in the metastatic T84 and SW620 cell lines, which may be less sensitive to PKG.
In any event, the current finding that PKG does not directly cause apoptosis of SW620 cells in vitro is consistent with the xenograft studies, in which immunostaining of apoptotic cells did not correspond specifically with PKG expression. Taken together, these observations highlight the tumor microenvironment as an important component of the suppressive effects of PKG in this system. Consistent with this idea, we observed that the tumors induced to express PKG were much less vascular than the tumors from parental cells, suggesting that the apoptosis observed in vivo was a secondary effect resulting from decreased tumor angiogenesis. This notion is supported by several observations of the PKG-expressing tumors, including 1) decreased blood vessels and the absence of capillaries, 2) decreased VEGF protein, 3) growth limited to sites surrounding blood vessels, and 4) extensive intervascular cell death. The reduced angiogenic potential of PKG-expressing SW620 cells can explain the decreased tumor size compared with the parental controls and also the absence of PKG-induced apoptosis in vitro.
To our knowledge, the inhibition of tumor angiogenesis by PKG-dependent signaling has not been documented, and it is important to understand the mechanism. Angiogenesis is a complex process and requires many intercellular factors that mediate interactions between tumor and host tissues.35 However, in colon cancer cells, β-catenin is important for the proper expression of proangiogenic factors, including VEGF and endothelin, as discussed previously.31, 32 The current work supports those studies, because both β-catenin and VEGF protein levels were reduced in the SW620 xenografts that expressed PKG. It is difficult to prove that the reduced levels of β-catenin in the PKG-expressing tumors were responsible for the decreased VEGF; however, the ability of PKG to block TCF-dependent gene expression in vitro strongly supports this idea. In light of the large number of factors that control angiogenesis, more work will be required to determine whether other PKG-regulated mechanisms may be important in addition to VEGF.
The reduced β-catenin levels observed in the PKG-expressing xenografts is particularly intriguing, because, as discussed above, this may have different consequences, depending on the nature of the tumor cell. For example, a decrease in the level of β-catenin may have a more pronounced effect on the growth of more primary cells than in metastatic cells, which may be less reliant on the β-catenin pathway for growth but still require it for angiogenesis. The mechanism of β-catenin down-regulation by PKG was not addressed in our studies; however, work with exisulind has indicated that PKG can mimic the effects of GSK3β and, by direct phosphorylation, causes increased proteasomal degradation.16, 17, 26 However, those studies did not address the specific phosphorylation sites labeled by PKG. It is noteworthy that examination of the β-catenin amino-acid sequence reveals 2 potential PKG phosphorylation sites, serine 552 (Ser552) and Ser675; however, these residues are far removed from the amino-terminal residues that are targeted by GSK3β and casein kinase 1, which lead to ubiquitination and subsequent degradation. The effect of phosphorylation of these sites in colon cancer cells is not known, but it was demonstrated recently that modification of these residues by protein kinase A can augment the gene expression function of β-catenin with no effect on expression.25 The effect of PKG on the regulation of β-catenin is made even more complex by a recent study indicating that PKG can inhibit GSK3β by direct phosphorylation.36 Clearly, there is much room for future investigations of β-catenin regulation by PKG-dependent pathways, and our finding that the down-regulation occurs in vitro sets the stage for this work.
- 14[No authors listed] Exisulind: Aptosyn, FGN 1, Prevatac, sulindac sulfone. Drugs R D. 2004; 5: 220–226.