Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Oct 19.
Published in final edited form as: Oncogene. 2006 Mar 27;25(36):4986–4997. doi: 10.1038/sj.onc.1209505

Silencing of epithelial CXCL12 expression by DNA hypermethylation promotes colonic carcinoma metastasis

MK Wendt 1, PA Johanesen 1, N Kang-Decker 2, DG Binion 3, V Shah 2, MB Dwinell 1
PMCID: PMC4610155  NIHMSID: NIHMS728357  PMID: 16568088

Abstract

Cellular metastasis is the most detrimental step in carcinoma disease progression, yet the mechanisms that regulate this process are poorly understood. CXCL12 and its receptor CXCR4 are co-expressed in several tissues and cell types throughout the body and play essential roles in development. Disruption of either gene causes embryonic lethality due to similar defects. Post-natally, CXCL12 signaling has a wide range of effects on CXCR4-expressing cells, including the directed migration of leukocytes, lymphocytes and hematopoietic stem cells. Recently, this signaling axis has also been described as an important regulator of directed carcinoma cell metastasis. We show herein that while CXCR4 expression remains consistent, constitutive colonic epithelial expression of CXCL12 is silenced by DNA hypermethylation in primary colorectal carcinomas as well as colorectal carcinoma-derived cell lines. Inhibition of DNA methyltransferase (Dnmt) enzymes with 5-aza-2′-deoxycytidine or genetic ablation of both Dnmt1 and Dnmt3b prevented promoter methylation and restored CXCL12 expression. Re-expression of functional, endogenous CXCL12 in colorectal carcinoma cells dramatically reduced metastatic tumor formation in mice, as well as foci formation in soft agar. Decreased metastasis was correlated with increased caspase activity in cells re-expressing CXCL12. These data constitute the unique observation that silencing CXCL12 within colonic carcinoma cells greatly enhances their metastatic potential.

Keywords: chemokine, intestinal mucosa, epithelial cell, epigenetic

Introduction

Chemokines are small chemotactic cytokines, which direct cellular migration through receptor specific interactions on target cells (Arya et al., 2003). The homeostatic chemokine–chemokine receptor pair CXCL12 and CXCR4 is widely expressed throughout the body (Bleul et al., 1996). CXCL12, formerly known as stromal cell-derived factor-1, is an alpha type 7.8 kDa CXC chemokine (Shirozu et al., 1995). Originally described as a growth factor for bone marrow developing B cells (Nagasawa et al., 1994), CXCL12 was subsequently characterized as a chemoattractant for T cells and monocytes (Bleul et al., 1996). Genetic ablation of CXCR4 or CXCL12 results in embryonic lethality (Nagasawa et al., 1996). Similar embryonic defects in either of those chemokine receptor or chemokine gene-deficient animals has revealed roles for CXCR4–CXCL12 signaling in cardiovascular, neuronal and hematopoietic stem cell development as well as gastrointestinal vascularization (Tachibana et al., 1998; Zou et al., 1998). Previous studies by our group have established a role for CXCL12 and CXCR4 in gut vascularization, a key process in mucosal immunity and homeostasis (Heidemann et al., 2004).

In addition to endothelial expression, the cells of the human colonic epithelium also express both CXCL12 and CXCR4 (Dwinell et al., 1999; Jordan et al., 1999; Agace et al., 2000). Moreover, using an in vitro wound healing assay we have shown that non-transformed intestinal epithelial cells migrate across a denuded surface in response to CXCL12, a key component of the rapid healing ability of the mucosal epithelial surface (Smith et al., 2005). The role of CXCL12–CXCR4 signaling in mucosal wound healing is consistent with other physiologic processes utilizing this signaling axis such as organogenesis and immune surveillance. Thus, our data demonstrate an important role for the combined expression of both CXCR4 and CXCL12 by the cells of the mucosal epithelium. More broadly, these processes of epithelial wound healing, enterocyte migration and vascular angiogenesis, which we have shown in healthy gut mucosa, are known to be dysregulated in colorectal cancer as well as chronic inflammatory diseases. Recent evidence indicates that CXCR4 expression by carcinoma cells may also participate in the metastasis of various cancer types including breast, prostate, non-small-cell lung and colon (Muller et al., 2001; Phillips et al., 2003; Sun et al., 2003; Zeelenberg et al., 2003). Notably, several studies linking chemokine receptor signaling to cancer cell metastasis suggest that aberrant regulation of CXCR4 expression plays an important role in this process (Muller et al., 2001; Haviv et al., 2004). In contrast, studies defining CXCL12 expression in various carcinomas are more limited.

Given the important functional roles and consistent dual expression of both CXCR4 and CXCL12 by human intestinal epithelium (Dwinell et al., 1999; Agace et al., 2000) we hypothesized that perturbations in epithelial CXCL12 expression would contribute to colorectal carcinoma disease progression, possibly by allowing carcinoma cells to more readily sense CXCL12 from exogenous sources, aiding metastasis. This novel hypothesis is supported by evidence from developing hematopoietic stem cells exiting the bone marrow, in which a disruption of CXCL12–CXCR4 signaling is required for stem cell mobilization (Christopherson et al., 2004; Gazitt, 2004). Our data herein suggest that a disruption in CXCR4 autocrine-signaling results from the silencing of CXCL12 in human colonic carcinoma cells. We define a mechanism of CXCL12 silencing in human colorectal carcinoma by DNA methyltransferase (Dnmt) enzyme-mediated promoter hypermethylation. Consistent with our hypothesis, re-establishment of endogenous CXCL12 expression in colonic carcinoma cells dramatically reduced in vivo metastatic tumor formation. Our data demonstrate a previously unrecognized mechanism of CXCL12 silencing in colorectal carcinoma, which significantly impacted the metastatic properties of those cells. Further, we suggest a new paradigm in which the epigenetic silencing of one arm of the CXCL12–CXCR4-signaling axis promotes tumor cell metastasis.

Results

Constitutive CXCL12 expression in the human colonic epithelium is absent in colonic carcinoma cells

To establish CXCL12 expression in the colonic mucosa we isolated the crypt epithelium from resectioned human colonic tissue (Figure 1a). Reverse transcriptase–polymerase chain reaction (RT–PCR) analysis of these colonic crypts showed that both CXCL12 and CXCR4 were expressed by normal human colonic epithelium (Figure 1b and c). Immunohistochemistry verified CXCL12 protein expression in human colonic epithelium (Figure 1d) indicating that CXCL12 mRNA expression did not reflect altered gene regulation resulting from crypt isolation (Agace et al., 2000). In agreement with our prior data CXCR4 protein expression was similarly observed in isolated crypts from human colonic tissues (data not shown). Isolated human peripheral blood monocytes assayed as a control did not express CXCL12 (Kimura et al., 2003). Those data, combined with the amplification of villin and CD45 as control transcripts, indicate that CXCL12 and CXCR4 mRNA expression observed in our mucosal preparations solely reflected epithelial expression and was not the result of contaminating immune cells.

Figure 1.

Figure 1

Open in a new tab

Constitutive CXCL12 expression is absent in colonic carcinoma cell lines. (a) Representative samples of crypt and epithelial sheet preparations from normal human colonic mucosa. (b) CXCL12 and CXCR4 mRNA is expressed in normal crypt (NC) epithelium from seven separate surgical specimens. Epithelial enrichment was verified by reverse transcriptase–polymerase chain reaction (RT–PCR) amplification of villin. Mucosal leukocyte contamination was assessed by RT–PCR amplification of CD45. glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified as a loading control. (c) CXCL12 mRNA expression normalized to GAPDH in isolated colonic epithelial preparations indicated consistent CXCL12 expression in the normal colon. (d) Immunohistochemistry showing areas of CXCL12 staining (arrows) in normal colonic surface epithelium relative to isotype antibody control staining (Ig). (e) CXCL12 mRNA expression in HT29, HCT116, Caco2 and T84 colonic carcinoma cell lines was not detectable using RT–PCR analysis. cDNA derived from human intestinal microvascular endothelium (HIMEC) was used as a positive control. Data in (a–e) are representative of 2–3 independent analyses.

In marked contrast to normal colonic epithelium, CXCL12 was minimally, if at all, expressed in HT29, HCT116, Caco2 and T84 colonic carcinoma cell lines assessed using identical RT–PCR conditions as those used for normal samples (Figure 1e). However, as a positive control, CXCL12 mRNA was expressed in human intestinal microvascular endothelial cells (HIMEC) (Heidemann et al., 2004). In contrast to CXCL12, CXCR4 was consistently expressed in both normal colonic epithelium and colonic carcinoma cell lines (Figure 1b and e) (Dwinell et al., 1999; Heidemann et al., 2004; Smith et al., 2005). Taken together, these results suggest that CXCL12, unlike CXCR4 and other epithelial-expressed chemokines (Dwinell et al., 2001; Izadpanah et al., 2001), is significantly downregulated in those human colonic carcinoma cell lines.

Identification of a putative CXCL12 promoter encompassed in a CpG island

We next sought to define the mechanism by which CXCL12 is specifically downregulated in colonic carcinoma cells. The immediate 5′ genomic region of CXCL12 lacks a true TATA-box and is rich in G/C content in the form of CpG dinucleotides, which is characteristic of homeostatic gene promoters (Shirozu et al., 1995; Garcia-Moruja et al., 2005). Using promoter prediction software (Scherf et al., 2000) we examined 10 kb of genomic sequence encompassing the CXCL12 gene, including 2 kb of sequence upstream of the 5′ untranslated region. We identified an area of sequence extending from −493 to −168 as a putative promoter region, relative to transcriptional start, +1 (Figure 2a). Our identification of this region as the CXCL12 promoter by computational analysis was recently verified experimentally (Garcia-Moruja et al., 2005). Using CpG analysis software we determined that this promoter was surrounded by a large CpG island extending from −840 to −852 (Li and Dahiya, 2002) (Figure 2a). Three additional CpG islands were also identified further upstream extending from −1877 to −1581, −1391 to −1231 and −1123 to −899 (not shown).

Figure 2.

Figure 2

Open in a new tab

The CXCL12 promoter region is methylated in human colorectal carcinoma (CRC). (a) The 5′ region of the CXCL12 gene contains a large CpG island (gray area) encompassing a predicted promoter region, the 5′ untranslated region (5′-UTR) and exon one (E1). (b) Methylation-specific polymerase chain reaction (PCR) indicated the CXCL12 promoter in HT29, HCT116, T84 and Caco2 carcinoma cells was methylated (m). Data from cell lines are representative of 3–5 independent analyses. (c) In contrast, the CXCL12 promoter was homozygous unmethylated (u) in normal human colonic (NC) crypts. The lack of promoter methylation in normal crypt epithelium is representative of 19 separate non-cancerous colonic tissues. Similar to the carcinoma cell lines, several primary human CRC tissues showed methylation of the CXCL12 promoter. Data from primary CRC tissues are representative of 21 independent samples and indicate a 62% methylation frequency in CRC relative to 0% observed methylation in NC. (d) Immunohistochemistry of a representative methylated CRC sample from panel (c), indicated CXCL12-specific staining was restricted to normal appearing epithelium and was absent in adjoining cancerous epithelium. CXCR4 staining was consistently observed in normal and cancerous tissues. H&E and immunoglobulin G (IgG) control images are shown at × 100 magnification with the boxed areas indicating the × 400 images shown for CXCL12 and CXCR4 staining on the left-most panels. Data in panel (d) representative of three separate methylated CRC specimens.

The CXCL12 promoter is hypermethylated in human colorectal carcinoma

Next, we used methylation-specific PCR (MSP) and bisulfite-sequencing PCR (Table 1) to determine the methylation status of the CXCL12 promoter as a potential mechanism for its transcriptional repression in colorectal carcinoma cells (Clark et al., 1994; Herman et al., 1996). In agreement with its mRNA and protein expression in normal human colonic epithelium, the CXCL12 promoter region in normal colonic epithelial isolates was consistently homozygous unmethylated (Figure 2c and Table 2).

Table 1.

Primers used for MSP, BSSP and RT–PCR analyses

Application and specificity Template DNA Forward primer (5′–3′) Reverse primer (5′–3′) Product
size (bp) 		Annealing
temperature (°C)
MSP
1 CXCL12-M BS DNA ggagtttgagaaggttaaaggtc ttaacgaaaaataaaaatagacgat 241 63
2 CXCL12-U BS DNA gagtttgagaaggttaaaggttgg taacaaaaaataaaaatacaacaat 242 57
BSSP
3 CXCL12 BS DNA gggattaatttgtttgttttttattg aactacctccacccccactatat 711 56
4 CXCL12 No. 3 PCR ggggttttgttatagggataataag aactacctccacccccactatat 595 58
RT–PCR
5 Villin cDNA aggcacctcccgaactaacaactt ccgctaccacccttcccacacca 189 63
6 CD45 cDNA catcccgcgggtgttcag tggtcccaaatcatcctccaga 252 63
7 eGFP cDNA acggccacaagttcagc cgtcgccgatgggggtgttct 504 63
8 GAPDH cDNA accacagtccatgccatcac tccaccaccctgttgctgta 452 63
Open in a new tab

Abbreviations: BS DNA, bisulfite-converted genomic DNA; BSSP, bisulfite sequencing PCR; MSP, methylation-specfic PCR; M, specfic for methylated bisulfite-converted DNA; RT–PCR, reverse transcriptase–polymerase chain reaction; U, specific for unmethylated bisulfite-converted DNA.

Table 2.

Methylation patterns of the CXCL12 putative promoter in CRC cell lines and normal tissues

CpG number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Position from +1 −275 −258 −250 −245 −235 −211 −206 −199 −194 −177 −166 −143 −140 −131 −129 −127 −115 −109 −98 −92 −86 −76 −74 −72 −67 −62 −57 −35
HT29 + + + + + + + + + + + + + + + + + + + + + + + + + + + +
HCT116 + + + + + + + + + + /− − + + + + + + + + + + + + + + + + +
T84 + − − − − − − − − − + /− − − − + /− − − − − − − − − − − − − −
Caco2 − − − − − − + + + + − − − − − − − − − − − − − − − − − −
HIMEC − − − − − − − − − − − − − − − − − − − − − − − − − − − −
NC3 − − − − − − − − − − − − − − − − − − − − − − − − − − − +
NC9 × − − − − − − − − − − − − − − − − − − − − − − − − − − −
NC10 + − − + − − − − − − − − − − − − − − − − − − − − − − − −
NC11 + − − − − − − − − − − − − − − − − − − − − − − − + /− + /− − ×
NC12 − − − − − − − − − − − − − − − − − − − − − − − − − − − −
NC14 − − + /− − − + /− + /− − − − + /− − − − − − − − − − − − − − − − − −
PBMC × − − + − − − − − − − − − − − − − − − − − − − − − − − −
Open in a new tab

Abbreviations: BSSP, bisulfite sequencing PCR; CRC, colorectal carcinoma; PBMC, peripheral blood mononuclear cell. +, methylated; −, unmethylated; + /−, heterozygous; × , undetermined. Direct sequencing of the 595 bp product of BSSP primer set no. 4 indicated the methylaton status of 28 CpG dinucleotides. The location of each CpG dinucleotide according to the CXCL12 transcriptional start (+ 1) is indicated. Data are representative of sequence analysis from two to five independent BSSP analyses.

Methylation of the CXCL12 promoter was further examined by bisulfite sequencing PCR (BSSP). Nearly all CpG’s from −275 to −35 were methylated in HCT116 and HT29 carcinoma cells. Caco2 cells showed methylation of five CpG dinucleotides from −211 to −177 while T84 cells were heterozygous methylated at two CpG’s in the examined promoter region. These latter data suggest additional sites of methylation or other mechanisms of gene silencing may also play a role in CXCL12 transcriptional regulation in those cell lines. In marked contrast to the carcinoma cell lines, this region of the promoter was consistently unmethylated in CXCL12-expressing normal epithelial samples and cultured HIMEC (Table 2).

Promoter methylation and repression of epithelial CXCL12 protein expression in primary colonic carcinoma

Next, we used our diagnostic MSP to analyse the methylation status of the CXCL12 promoter in several primary colorectal carcinoma tissues to ensure hyper-methylation was not solely a property of colonic carcinoma cell lines. Aberrant hypermethylation of the CXCL12 promoter was readily observed with increasing frequency in several carcinomas, but not normal mucosa, suggesting chemokine gene silencing during in vivo disease progression (Figure 2c and Figure S1 in Supplementary Information). Consistent with the heterogeneous nature of colorectal tumors, we determined that hypermethylation of the CXCL12 promoter was observed in more than half of the colorectal tumors analysed by MSP, with 62% of 21 separate colonic carcinomas possessing methylated CXCL12 alleles (Figure 2c and Figure S1 in Supplementary Information). Moreover, our observed frequency for CXCL12 methylation is consistent with other genes previously shown to be silenced by DNA hypermethylation and involved with cancer in vivo (Robertson, 2001).

Carcinomas determined to contain methylated alleles of CXCL12 were further analysed by immunohistochemistry to define CXCL12 protein expression in cancerous versus normal human colonic mucosa. As shown in Figure 2d for a representative sample, CXCL12 was not detectable in the disorganized cancerous epithelium, while adjacent regions of organized normal epithelium maintained expression of that homeostatic chemokine. Parallel sections of the same tissues indicated that both the normal, CXCL12 expressing and dysplastic, CXCL12-null, epithelium-expressed CXCR4 (Figure 2d). Taken together these results suggest that methylation of the CXCL12 promoter is a pathological event in vivo and may play a role in its transcriptional repression in human colorectal carcinoma.

DNA methyltransferase enzymes are over-expressed in colorectal carcinoma cell lines

To better define the mechanism leading to CXCL12 hyper-methylation in colonic carcinoma cells we determined Dnmt expression in those cells as compared to normal epithelium. Consistent with CXCL12 promoter hyper-methylation, HCT116, HT29, T84 and Caco2 colonic carcinoma cell lines strongly expressed Dnmt enzymes. In sharp contrast, the proteins were not detectable in normal mucosal samples or the IEC6 cell line, a model of normal non-transformed intestinal epithelium (Figure 3a). Importantly, the extractable levels of Dnmt 1 could be mechanistically ablated by the Dnmt inhibitor, 5-aza-2′-deoxycytidine (5-aza) (Figure 3b) (Jones and Taylor, 1980).

Figure 3.

Figure 3

Open in a new tab

DNA methyltransferase-1 (Dnmt1) and Dnmt3b are over-expressed in colonic carcinoma cell lines. (a) Immunoblot analysis showing Dnmt1 and Dnmt3b expression in the CXCL12-null HCT116, HT29, T84 and Caco2 cells compared to CXCL12-expressing normal colonic (NC) epithelial preparations and cultured non-transformed IEC6 cells. (b) Immunoblot analysis showing extractability of Dnmt1 was inhibited in HCT116 cells treated with increasing concentrations of 5-aza-2′-deoxycytidine (5-aza). Levels of actin indicated equal protein loading between non-stimulated (NS) and 5-aza-treated HCT116 cells. Data in (a–b) are representative of three independent experiments.

Inhibition of DNA methyltransferase enzymes restores CXCL12 expression in colonic carcinoma cells

We next sought to determine whether inhibition of Dnmt enzymes could re-establish CXCL12 expression in colorectal carcinoma cells. Treatment of either HT29 or HCT116 colonic carcinoma cells with 5-aza dose-dependently restored CXCL12 mRNA expression (Figure 4a). Furthermore, treatment of those cells with the optimal 2.5 μm concentration of 5-aza restored CXCL12 expression, with maximal re-expression noted after 5 days (Figure 4a). Treatment of Caco2 and T84 cells with 5-aza, similarly fostered gene re-expression, suggesting that these heterozygous methylated cell populations harbor functional alleles of CXCL12 that are silenced by DNA hypermethylation (data not shown). In parallel with restored CXCL12 mRNA expression, protein expression was detected by immunofluorescence microscopy upon 5-aza treatment (Figure 4b).

Figure 4.

Figure 4

Open in a new tab

Inhibition of DNA methyltransferase (Dnmt) enzymes restored CXCL12 expression. (a) Treatment of HT29 (top) and HCT116 (bottom) cells with the indicated concentrations of 5-aza-2′-deoxycytidine (5-aza) for 5 days or with 2.5 μM 5-aza for 3, 4, 5 or 6 days (D3–D6) resulted in re-expression of CXCL12 relative to non-stimulated (NS) cells. CXCR4 mRNA levels remained unchanged with 5-aza treatment. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified as a loading control. (b) Immunofluorescence analysis indicated the recovery of CXCL12-specific staining (red) after 3 days of 2.5 μM 5-aza treatment (bi). No CXCL12-specific staining was observed in non-stimulated control HCT116 cells (bii). Immunoglobulin G (Ig) staining for 5-aza treated (biii) and non-stimulated HCT116 cells (biv) were performed to define levels of nonspecific immunostaining. (c) CXCL12 methylation-specific polymerase chain reaction (PCR) indicating the presence of unmethylated alleles of CXCL12 in HCT116 cells treated with 2.5 μM 5-aza for 3 days. Data in (a–c) are representative of three independent experiments.

It has been suggested that CXCL12 expression can be enhanced by tissue and DNA damage (Ponomaryov et al., 2000). To ensure that 5-aza was restoring CXCL12 transcription by inhibiting methylation and not by nonspecific DNA damage, we used MSP to analyse DNA from 5-aza-treated cells. Methylation of the CXCL12 promoter was inhibited by 5-aza treatment in HCT116 (Figure 4c) and HT29 cells (data not shown). Taken together these data indicate that pharmacological Dnmt inhibition prevents methylation of the CXCL12 promoter and allows mRNA and protein expression in colonic carcinoma cells.

DNA methyltransferase 1 and DNA methyltransferase 3b silence CXCL12 in colorectal carcinoma cells

We next utilized HCT116 cells lacking Dnmt1 (MT1kO), Dnmt3b (3bkO), or both enzymes (Dko) to define the mechanism by which CXCL12 becomes silenced in colorectal cancer. Specific gene deficiency in these cell lines has previously been shown (Rhee et al., 2002) and was verified in our laboratory using RT–PCR and immunoblot analyses (data not shown). Consistent with CXCL12 being silenced by DNA hypermethylation the transcript was expressed in HCT116 cells lacking both Dnmt1 and Dnmt3b, but not the single knockout cell line (Figure 5a). Further, MSP analysis showed that CXCL12 was homozygous unmethylated in Dko cells as opposed to the homozygous methylated profile of the wild-type parent, or similarly, the cells lacking only Dnmt1 or Dnmt3b (Figure 5b). These data agree with the Dnmt expression and methylation patterns of our cultured cell lines and suggest that either Dnmt1 or Dnmt3b can silence CXCL12 expression in carcinogenic but not normal colonic epithelium.

Figure 5.

Figure 5

Open in a new tab

CXCL12 expression in DNA methyltransferase (Dnmt)-deficient HCT116 carcinoma cells. (a) CXCL12 and CXCR4 mRNA expression analysed by reverse transcriptase–polymerase chain reaction (RT–PCR) in WT: wild-type; MT1ko: Dnmt1 knockout; 3bko: Dnmt3b knockout; or Dko HCT116 cells: Dnmt1/Dnmt3b double knockout. (b) CXCL12 methylation-specific PCR (MSP) analysis showing the DNA methylation profile of the Dnmt-deficient HCT116 cells. CXCL12 is homozygous methylated in the WT parent, MT1ko and 3bko cells that lack CXCL12 mRNA transcripts. Dko cells that express the CXCL12 transcript were homozygous unmethylated. Data are representative of three independent experiments.

Endogenous CXCL12 expression by colonic carcinoma cells decreases metastatic tumor formation

Having established an epigenetic mechanism for CXCL12 silencing in human colorectal carcinoma, we sought to define the significance of this event in disease progression. We therefore generated HT29 and HCT116 colonic carcinoma cells which stably expressed CXCL12 in order to recapitulate the CXCR4-signaling axis of the normal colonic epithelium (Dwinell et al., 1999). Reverse transcriptase–polymerase chain reaction analysis showed specific expression of CXCL12 or enhanced green fluorescent protein (eGFP) used as a vector control, in both HT29 and HCT116 carcinoma cells (Figure 6a and b). CXCR4 mRNA expression (Figure 6a and b) and total protein levels (Figure 7d) were comparable in the wild-type parent cell lines, CXCL12-expressing, and vector control cells. CXCL12 protein was detected in the supernatant of CXCL12-transfected HT29 and HCT116 colonic carcinoma cells, but not parent or vector control cells, and was within the normal physiological range for that chemokine (Figure 6c and d) (Derdeyn et al., 1999). Further, CXCL12 secreted from both HT29 and HCT116 stable transfectants was functional, as assessed by the ability to stimulate chemotaxis of CXCR4-expressing U937 monocytic cells (Figure 6e and f). Further, chemotaxis of those cells was specific for epithelial CXCL12-leukocyte CXCR4 as the receptor antagonist AMD3100 potently blocked migration of those cells across the filter (Figure 6e and f).

Figure 6.

Figure 6

Open in a new tab

Stable re-expression of functional CXCL12 in colonic carcinoma cells. (a and b) HT29 (a) and HCT116 (b) cells were stably transfected with plasmid vectors encoding CXCL12 or enhanced green fluorescent protein (eGFP) as a control and specific gene expression verified by reverse transcriptase–polymerase chain reaction (RT–PCR). CXCR4 mRNA levels remained unchanged following re-expression of CXCL12. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified as a loading control. (c and d) CXCL12 protein was detectable in the supernatant of several independent CXCL12 stable transfectant clones in both HT29 (c) and HCT116 (d) cells, but not the WT or vector control cell lines. (e and f) CXCL12 produced by HT29 (e) and HCT116 (f) cells stimulated the chemotaxis of CXCR4-expressing U937 monocytic cells (black bars). Chemotaxis was specifically inhibited by the CXCR4 antagonist AMD3100 (5 μg/ml) (white bars). Data are representative of three independent experiments.

Figure 7.

Figure 7

Open in a new tab

Endogenous CXCL12 expression in CXCR4-expressing colonic carcinoma cells reduced in vivo metastatic tumor formation in severe combined immunodeficiency (SCID) mice. (a) CXCL12 or eGFP-expressing HT29 human colonic carcinoma cells were injected into the hepatic portal vein of SCID mice and allowed to form metastases for 5 weeks. Enhanced green fluorescent protein (eGFP)-expressing carcinoma cells formed noticeably larger metastases (white tissue) than did CXCL12-expressing HT29 cells. Data are representative of five separate mice per treatment group. (b) Tumors dissected from normal liver tissue were weighed and data presented as percent of body weight. Asterisk equals significant decrease in tumor size (P≤0.05) between tumors resulting from CXCL12 stable carcinoma cells and those of the eGFP vector control cells. Data in (b) are mean±s.d. of five separate mice per group. (c) Reverse transcriptase–polymerase chain reaction (RT–PCR) analyses of the dissected tumor fractions (T) were positive for villin, a marker of intestinal epithelial cells, while normal liver tissue (L) lacked villin expression similar to wild-type (WT) mouse liver tissue from a non-injected animal. In vivo transgene expression of human CXCL12 (h-CXCL12) and eGFP were verified for each of the two tumor types. (d) Immunoblot analysis showing consistent CXCR4 expression in CXCL12 and eGFP-expressing HT29 cells both before (d0) and after tumor dissection (d35) following injection into SCID mice (data representative of three separate mice per treatment group).

To determine the in vivo impact of endogenous CXCL12 expression on colonic carcinoma disease progression we utilized a severe combined immunodeficiency (SCID) mouse model of tumor cell metastasis (Panis et al., 1990). Carcinoma cells stably re-expressing CXCL12 or control eGFP cells were injected into the portal vein to assess the ability of those cells to invade and colonize the liver. Consistent with the notion that CXCL12-silencing aids in carcinoma cell metastasis, colon carcinoma cells in which CXCL12 was endogenously re-expressed formed significantly smaller meta-static lesions than did eGFP-expressing vector control cells (Figure 7a and b). Resultant dissected metastatic tumors from mice injected with CXCL12 stable cell lines maintained expression of human CXCL12 as verified using RT–PCR (Figure 7c). However, CXCL12 expression was absent in those tumors resulting from the eGFP vector control. The native CXCL12 locus remained methylated in resultant tumors for all cases (data not shown). Primers specific for the human villin transcript were used as a marker for the presence of intestinal epithelial cells in the developed tumors and was absent in both dissected normal liver and wild-type non-injected liver tissue (Figure 7c). Further, CXCR4 expression in HT29-stable CXCL12 and eGFP carcinoma cells (Figure 7c and d) was maintained at comparable levels and was not significantly altered after implantation into SCID mice. These data were not unique to HT29 cells as our HCT116 cells stably reexpressing CXCL12 (Figure 6) formed similar small metastatic tumors relative to the control cells in SCID mice (data not shown).

We next sought to define possible biochemical mechanisms preventing in vivo metastatic tumor formation by colonic carcinoma cells in which CXCL12 was re-established. In agreement with our SCID mouse model, several clones of HT29 cells stably re-expressing CXCL12 failed to invade the matrix and establish foci in soft agar (Figure 8a). The inability of CXCL12-stable transfectants to form foci suggested a decreased ability of CXCL12-expressing cells to invade and populate their surrounding microenvironment. Consistent with this notion we observed an increase in apoptosis in CXCL12-expressing cells compared to eGFP clones, as assessed by active caspase-3/7 levels (Figure 8b). HCT116 cells stably expressing CXCL12 similarly had increased caspase-3/7 activity (not shown). Together, these data suggest that autocrine CXCL12–CXCR4 signaling, a process subverted by colon cancer cells through the silencing of CXCL12, increases caspase-3/7 activity inhibiting the ability of carcinoma cells to metastasize in vivo.

Figure 8.

Figure 8

Open in a new tab

Endogenous CXCL12 secretion by colonic carcinoma cells reduces in vitro soft agar foci formation and increases caspase-3/7 activity. (a) HT29 cells stably transfected with CXCL12 form smaller foci when grown 2 weeks in soft agar as compared to vector control (enhanced green fluorescent protein (eGFP)) or wild-type (WT) parent cells. Brightfield microscopy images at × 40 (left) and × 200 (right). (b) eGFP or CXCL12-expressing HT29 cells (1 × 104) were analysed using a caspase-3/7 luminescence assay. Data were expressed as relative luciferase units (RLU) and showed greater caspase activity in CXCL12-expressing cells than eGFP cells. Images in (a) are representative of three independent experiments. Values in (b) are the mean±s.d. of replicate samples and representative of three independent experiments.

Discussion

The chemokine receptor CXCR4 has been shown to be expressed on cancer cells and play a significant role in the process of metastasis (Muller et al., 2001). In addition to this role, CXCR4 signaling is also a key regulator of organogenesis as well as lymphopoiesis and myelopoiesis (Nagasawa et al., 1994; Zou et al., 1998). In previous studies we and others have defined the concurrent expression of both CXCR4 and CXCL12 by the cells of the human intestinal epithelium (Dwinell et al., 1999; Jordan et al., 1999). We subsequently determined that that signaling axis regulated enterocyte migration (Smith et al., 2005), a key process in the establishment and ongoing repair of the healthy mucosal epithelial barrier. Given this dichotomy between the physiologic and pathophysiologic functions of CXCR4, we hypothesized that changes in the constitutive epithelial expression of CXCL12 may play a pivotal role in determining the function for CXCR4 signaling in the human intestinal mucosa. Our results show that in marked contrast to normal colonic epithelium, CXCL12 is absent in several colorectal cancer cell lines and primary carcinoma tissues, while CXCR4 expression is maintained. We defined DNA hypermethylation as a mechanism for CXCL12 gene silencing in colorectal carcinoma. Further, re-establishing endogenous expression of CXCL12 in colonic carcinoma cells profoundly reduced in vivo metastatic tumor formation, reflecting, in part, increased caspase-3/7 activity in those cells. Current models suggest that CXCR4 expression by tumor cells drives those cells to migrate to ectopic sites of CXCL12 expression. Our data add to this paradigm, wherein the epigenetic silencing of constitutive CXCL12 expression in carcinoma cells, elicits a metastatic phenotype enabling tumor cells to pathologically utilize the chemokine system, exacerbating disease.

Given recent evidence demonstrating the pro-meta-static roles of CXCR4–CXCL12 signaling in carcinoma cells, one may have expected over-expression of CXCL12 in carcinoma cells to result in increased metastasis (Muller et al., 2001; Kang et al., 2005). The inherent differences between exogenous CXCL12 stimulation and endogenous expression of the protein are undoubtedly responsible for our results compared to previous reports examining the role for CXCR4 in colonic tumorigenesis. We propose that exogenous stimulation of cells with CXCL12 is representative of carcinoma cells which do not produce their own CXCL12 and can thus respond to chemokines produced by distal tissue sites, resulting in pro-tumorigenic-signaling processes. We believe our stable re-expression model system is more representative of normal in vivo colonic epithelial cells undergoing autocrine and/or local paracrine CXCL12–CXCR4 signaling, participating in the maintenance of the epithelial barrier, a process requiring cellular migration and apoptosis. Directed cellular migration is dependent on a cell responding to a CXCL12 gradient, a process facilitated in leukocytes and metastatic carcinoma cells by the expression of CXCR4. Our prior reports indicate that CXCL12 signaling through CXCR4 is an important regulator of mucosal wound healing by inducing intestinal epithelial cell migration (Smith et al., 2005). The CXCL12 migratory response of immune cells is much higher, however, than that of intestinal epithelial cells, an observation consistent with the intrinsic absence of CXCL12 expression in immune cells (Kimura et al., 2003). We show herein that similar to immune cells, carcinoma cells lack expression of CXCL12 but maintain expression of CXCR4. These data are consistent with reports indicating the importance of CXCR4–CXCL12 signaling in the homing of cancer cells to sites of metastasis in which CXCR4-expressing tumor cells pathologically follow endocrine CXCL12 chemotactic gradients, enter the vascular or lymphatic circulation, resist apoptosis and actively invade ectopic tissues (Bleul et al., 1996; Muller et al., 2001; Schrader et al., 2002). As shown herein, re-establishing CXCL12 expression in carcinoma cells restored the normal epithelial phenotype preventing pathological utilization of this signaling axis, resulting in reduced in vivo metastatic tumor formation.

Physiologic DNA methylation is achieved by the activity of several Dnmt enzymes. The Dnmt3 family of enzymes is believed to act as the de novo methyltransferases, while Dnmt1 is believed to act as the maintenance methyltransferase (Bestor, 2000). Many lines of evidence challenge these definitive categorizations making it difficult to predict which Dnmt enzyme is responsible for hypermethylation and gene silencing in cancer (Robertson, 2001). CpG island methylation and gene silencing in the absence of Dnmt1 has been shown, in contrast to reports suggesting Dnmt1 is required to maintain CpG methylation (Rhee et al., 2000; Robert et al., 2003). Recent evidence suggests that Dnmt1 and Dnmt3b act cooperatively to silence genes in carcinoma (Rhee et al., 2002). Consistent with this notion, our results are the first to indicate CXCL12 can be pathologically silenced in colorectal carcinoma by Dnmt1 or Dnmt3b enzymes which are markedly over-expressed in those cells relative to normal epithelium. Further, our data strengthen the importance of epigenetic gene regulation in the processes responsible for changes in cell growth in metaplasia, but also tumor cell invasion and metastasis.

The shift to a metastatic cellular phenotype by the epigenetic downregulation of CXCL12 expression is paralleled by previous reports noting the absence of CXCL12 expression in isolated primary colonic adenomas as well as other carcinoma cell lines (Begum et al., 1996). Similarly, renal tumor cells have been shown to display diminished CXCL12 mRNA expression relative to adjacent normal tissue (Schrader et al., 2002). Taken together with these studies, our data suggest that silencing the ligand arm of this signaling axis changes the homeostatic autocrine and paracrine CXCR4 signaling to a strictly endocrine communication arc that facilitates metastasis of those carcinomas.

It has recently been shown that methylation-specific markers can be used as a non-invasive diagnostic indicator of tumor progression for colorectal cancer (Jubb et al., 2003; Lenhard et al., 2005). The use of CXCL12 as a methylation marker is promising given the strong correlation shown here between methylation, gene silencing and disease. The effectiveness of CXCL12 in a panel of methylation markers as an indicator of tumor progression, however, remains to be established through a comprehensive assessment of CXCL12 expression and methylation status in primary tissues of known tumorigenic status (Fearon and Vogelstein, 1990).

In summary, the homeostatic expression of CXCL12, but not CXCR4, is a target for gene silencing in colorectal cancer, via DNA hypermethylation by Dnmt1 and Dnmt3b. Silencing of this immunosurveillance chemokine likely aids in carcinoma disease progression, as our data indicates that re-establishment of normal CXCL12 expression in colonic carcinoma markedly reduced tumor cell metastasis in vitro and in vivo. These findings are consistent with and expand upon previous data concerning the role of CXCR4 signaling in carcinoma cell metastasis and maintenance of the human colonic epithelium. Our results, together with recent findings emphasizing the importance of CXCR4 signaling in cancer cell migration and invasion, constitute a unique observation that loss of endogenous CXCL12 expression plays a role in the increased metastasis of cancer cells.

Materials and methods

Human colorectal carcinoma cell lines

HT29 (HTB-38), HCT116 (CCL-247), Caco2 (HTB-37), T84 (CCL-248) colonic carcinoma cells were purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA) and maintained as previously described (Dwinell et al., 1999; Smith et al., 2005). Dnmt1, Dnmt3b or Dnmt1/Dnmt3b double knockout HCT116 cells were the kind gift of Dr Bert Vogelstein (John Hopkins University School of Medicine) and were maintained similarly to wild-type HCT116 cells. In some experiments, carcinoma cells were treated with 5-aza (EMD Biosciences, La Jolla, CA, USA) every 24 h for the indicated number of days.

Human mucosal samples

Colonic epithelium and HIMEC were obtained from surgical remnants from colonic resections or carcinoma biopsy in accordance with protocols approved by the Medical College of Wisconsin human research review committee institutional review board. HIMEC samples were isolated as described previously (Binion et al., 1997; Heidemann et al., 2004). To isolate colonic crypts the muscularis externa was detached from surgical specimens and the resulting mucosal strips were washed, minced and incubated 90 min at room temperature in phosphate-buffered saline (PBS) containing 3 mm EDTA and 1 μm DTT and the tube was shaken to liberate crypts. The resulting supernatant, containing epithelial crypts, was transferred to a clean centrifuge tube, and the shaking step repeated for a total of four times. Supernatants were combined, filtered through a sterile gauze pad, and the isolated crypt epithelium collected by centrifugation (Whitehead et al., 1999).

Immunohistochemistry

Full thickness normal colonic specimens were fixed in 4% (w/v) paraformaldehyde/PBS overnight as detailed previously (Heidemann et al., 2004). CXCL12 protein expression in human colonic epithelium was determined using mouse monoclonal antibody (mAb), clone K15C (Amara et al., 1997), or murine isotype control mAb (R&D Systems, Minneapolis, MN, USA) and visualized using the alkaline-phosphatase antialkaline phosphatase method as described by the manufacturer (DAKO, Carpentaria, CA, USA) or diaminobenzidine as described by the manufacturer (Vector Labs, Burlingame, CA, USA).

Reverse transcriptase–polymerase chain reaction analysis

Total RNA was isolated from cultured cells and colonic crypt epithelium using TRIzol reagent (Invitrogen, Carlsbad, CA, USA), DNase-treated (Ambion, Austin, TX, USA) and 2 μg of total RNA was converted to cDNA via reverse transcription using random priming in a 40 μl volume. CXCL12, CXCR4 and β-actin mRNA transcripts were amplified using previously described PCR primers and conditions (Heidemann et al., 2004; Smith et al., 2005). Other PCR analyses were conducted using listed primer pairs and conditions (Table 1). RNA was excluded in cDNA synthesis reactions as a negative control.

CpG island and promoter analysis

A putative promoter region was identified from 10 kb of genomic sequence encompassing the CXCL12 gene (GenBank Ac#AL390792) including 2 kb flanking the gene both 5′ and 3′. This sequence was entered into the Genomatix promoter prediction program ‘Promoter Inspector’ (Scherf et al., 2000), or into Methprimer for analysis of CpG dinucleotide content and CpG island identification (Li and Dahiya, 2002). CpG islands were defined as regions of DNA greater than 200 bp, containing a guanine/cytosine content greater than 50% and an observed to expected CpG ratio above 0.6.

Methylation-specific polymerase chain reaction and bisulfite sequencing polymerase chain reaction

Genomic DNA from cell lines and colonic crypt preparations was isolated using the Genomic DNA wizard kit (Promega, Madison, WI, USA) or TRIzol according to the manufacturer’s instructions. Genomic DNA was isolated from paraffin-embedded carcinoma tissues by deparaffinization followed by Proteinase K digestion, phenol chloroform extraction, and precipitation in ethanol containing sodium acetate and gylcogen as a carrier. DNA (2 μg) was denatured in 0.3 m NaOH at 42°C for 20 min in a 111 μl reaction volume. To this 1.2 ml of sodium bisulfite solution (4.5 m NaHSO3, 0.02 m hydroquinone, and pH 5.0) was added and incubated 16 h at 55°C. This reaction was desalted using DNA Purification Wizard (Promega) and the DNA was desulfonated in 0.3 m NaOH at 37°C for 20 min. The converted DNA was precipitated at −20°C, overnight in 75% ethanol containing 0.7 m ammonium acetate and 0.05 mg/ml of glycogen, reconstituted in 50 μl of water, and 4.0 μl used in each 50 μl PCR reaction using 0.5 μm of specific MSP primers (Table 1). The same bisulfite-converted genomic DNA was separately analysed by BSSP using seminested primers (Table 1). Polymerase chain reaction products obtained with these BSSP primers were directly sequenced using Big Dye Terminator v3.1 (Applied Biosystems, Foster City, CA, USA).

Immunofluorescence microscopy

HCT116 epithelial cells were plated to glass chamber slides and grown for 3 days in untreated or 5-aza-containing media. Cells were stained with mouse mAb specific for CXCL12 (clone no. 79018.111, R&D Systems) or a nonspecific immunoglobulin G (IgG) control antibody (BD Pharmingen, San Jose, CA, USA). CXCL12 protein and IgG background staining was visualized using Alexa Fluor 594 conjugated goat anti-mouse antibody (Molecular Probes, Eugene, OR, USA). Nuclei were visualized with a 4,6-diamidino-2-phenylindole counterstain.

Immunoblot analysis

For the detection of Dnmt1, Dnmt3b and actin, whole cell lysates were prepared as previously described (Smith et al., 2005), and 10 or 25 μg of protein size separated using reducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Equal protein loading was confirmed by Coomassie blue staining. Proteins were electro-transferred to PVDF (Immobilon-P; Millipore, Bedford, MA, USA) for immunoblot analysis in which blots were incubated with goat antibodies specific for human or rat Dnmt1, Dmnt3b or actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and visualized with HRP-conjugated donkey anti-goat antibodies (Santa Cruz Biotechnology) and SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA).

Construction of stable epithelial cell lines

HT29 or HCT116 colonic carcinoma cells were transfected with pcDNA3.1. (Invitrogen) encoding the CXCL12α mRNA transcript, or eGFP, using Nova-Fector (Venn Nova, Pompano Beach, FL, USA). Stable plasmid integration was selected using G418 sulfate (EMD Biosciences). Stable gene expression was verified by RT–PCR, fluorescence microscopy, or enzyme-linked immunosorbent assay (ELISA), using a matched antibody pair (R&D Systems).

Leukocyte chemotaxis assay

HT29 or HCT116 cells were plated and grown to confluence in 24-well dishes. Cells were then serum starved for 2 days. Calcein-AM (Molecular Probes) loaded U937 monocytes (5 × 105) were plated to the upper well of a Transwell chemotaxis chamber (5 μm pore size, Corning Costar, Corning, NY, USA), with the serum-starved epithelial monolayers in the bottom chamber. U937-epithelial co-cultures were supplemented with the CXCR4 receptor antagonist AMD3100 (5 μg/ml) or remained untreated as a control. U937 cells were incubated without epithelial cells or, as a separate control, epithelial cells alone, were used to establish constitutive chemotaxis or auto-fluorescence, respectively. U937 cells were incubated with epithelial cells for 3 h and the number of migrated calcein-loaded monocytes in the bottom chamber quantified by fluorescence spectroscopy (Victor Wallac, Perkin Elmer, Turku, Finland).

Severe combined immunodeficiency mouse portal vein injection

Using protocols approved by the institutional review board of the Mayo Clinic, HT29 colonic carcinoma cells (1 × 106 cells) stably expressing either CXCL12 or eGFP were suspended in a 100 μl volume of sterile PBS and injected into the portal vein of anesthetized 8-week-old male SCID mice (cr-Prdkcscid, Charles Rivers, Wilmington, MA, USA). Prior to injection all cell lines were ~90% viable as assessed by trypan blue exclusion. Mice were allowed to recover and monitored for tumor development. Tumor-bearing mice were weighed and sacrificed by CO2 inhalation. Livers were removed, tumors dissected from normal tissue and the excised tumors weighed. Total RNA or total cell protein was isolated from normal liver and excised tumor tissue using TRIzol or lysis buffer for RT–PCR and immunoblot analyses, respectively.

Soft agar invasion and caspase assays

Equal numbers of cells were plated on a layer of 0.7%. agar made up in full growth media, covered with 0.35% warm agar and cultured for 2 weeks at which point foci were photographed under bright-field microscopy. For caspase assays HT29 or HCT116 cells were grown to near confluence and then serum starved 48 h at which point supernatants from cells were combined with trypsinized monolayers and cells. Cells (1 × 104) were subjected to the caspase-3/7 assay according the manufacturers instructions (Promega). Luminescence was measured as a quantification of caspase activity (Victor Wallac).

Supplementary Material

Supplemental Figure

Acknowledgements

We thank Dr Richard Komorowski (Froedtert Lutheran Memorial Hospital) and Dr Nita Salzman (Medical College of Wisconsin) for assisting in the analysis of primary carcinoma tissues under approval of the Medical College of Wisconsin Institutional Review Board. Portions of the immunohistochemical staining were completed in cooperation with the laboratory of Dr Martin Kagnoff (University of California, San Diego, USA). We appreciate the kind assistance of Dr Bert Vogelstein (John Hopkins University School of Medicine) for providing us with the Dnmt knockout cell lines. Support was provided by National Institutes of Health Grants DK062066 and DK002808, as well as a FIRST Award from the Crohn’s and Colitis Foundation of America and a grant from Medical College of Wisconsin Cancer Center.

Footnotes

Supplementary Information accompanies the paper one Oncogene website (http://www.nature.com/onc)

References

  1. Agace WW, Amara A, Roberts AI, Pablos JL, Thelen S, Uguccioni M, et al. Curr Biol. 2000;10:325–328. doi: 10.1016/s0960-9822(00)00380-8. [DOI] [PubMed] [Google Scholar]
  2. Amara A, Gall SL, Schwartz O, Salamero J, Montes M, Loetscher P, et al. J Exp Med. 1997;186:139–146. doi: 10.1084/jem.186.1.139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arya M, Patel HR, Williamson M. Curr Med Res Opin. 2003;19:557–564. doi: 10.1185/030079903125002216. [DOI] [PubMed] [Google Scholar]
  4. Begum NA, Coker A, Shibuta K, Swanson RS, Chen LB, Mori M, et al. Biochem Biophys Res Commun. 1996;229:864–868. doi: 10.1006/bbrc.1996.1893. [DOI] [PubMed] [Google Scholar]
  5. Bestor TH. Hum Mol Genet. 2000;9:2395–2402. doi: 10.1093/hmg/9.16.2395. [DOI] [PubMed] [Google Scholar]
  6. Binion DG, West GA, Ina K, Ziats NP, Emancipator SN, Fiocchi C. Gastroenterology. 1997;112:1895–1907. doi: 10.1053/gast.1997.v112.pm9178682. [DOI] [PubMed] [Google Scholar]
  7. Bleul CC, Fuhlbrigge RC, Casasnovas JM, Aiuti A, Springer TA. J Exp Med. 1996;184:1101–1109. doi: 10.1084/jem.184.3.1101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Christopherson KW, Hangoc G, Mantel CR, Broxmeyer HE. Science. 2004;305:1000–1003. doi: 10.1126/science.1097071. [DOI] [PubMed] [Google Scholar]
  9. Clark SJ, Harrison J, Paul CL, Frommer M. Nucl Acids Res. 1994;22:2990–2997. doi: 10.1093/nar/22.15.2990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Derdeyn CA, Costello C, Kilby JM, Sfakianos G, Saag MS, Kaslow R, et al. AIDS Res Hum Retroviruses. 1999;15:1063–1071. doi: 10.1089/088922299310359. [DOI] [PubMed] [Google Scholar]
  11. Dwinell MB, Eckmann L, Leopard JD, Varki NM, Kagnoff MF. Gastroenterology. 1999;117:359–367. doi: 10.1053/gast.1999.0029900359. [DOI] [PubMed] [Google Scholar]
  12. Dwinell MB, Lugering N, Eckmann L, Kagnoff MF. Gastroenterology. 2001;120:49–59. doi: 10.1053/gast.2001.20914. [DOI] [PubMed] [Google Scholar]
  13. Fearon ER, Vogelstein B. Cell. 1990;61:759–767. doi: 10.1016/0092-8674(90)90186-i. [DOI] [PubMed] [Google Scholar]
  14. Garcia-Moruja C, Alonso-Lobo JM, Rueda P, Torres C, Gonzalez N, Bermejo M, et al. J Mol Biol. 2005;348:43–62. doi: 10.1016/j.jmb.2005.02.016. [DOI] [PubMed] [Google Scholar]
  15. Gazitt Y. Leukemia. 2004;18:1–10. doi: 10.1038/sj.leu.2403173. [DOI] [PubMed] [Google Scholar]
  16. Haviv YS, van Houdt WJ, Lu B, Curiel DT, Zhu ZB. Mol Cancer Ther. 2004;3:687–691. [PubMed] [Google Scholar]
  17. Heidemann J, Ogawa H, Rafiee P, Lugering N, Maaser C, Domschke W, et al. Am J Physiol Gastrointest Liver Physiol. 2004;286:G1059–G1068. doi: 10.1152/ajpgi.00417.2003. [DOI] [PubMed] [Google Scholar]
  18. Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. Proc Natl Acad Sci USA. 1996;93:9821–9826. doi: 10.1073/pnas.93.18.9821. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Izadpanah A, Dwinell MB, Eckmann L, Varki NM, Kagnoff MF. Am J Physiol Gastrointest Liver Physiol. 2001;280:G710–G719. doi: 10.1152/ajpgi.2001.280.4.G710. [DOI] [PubMed] [Google Scholar]
  20. Jones PA, Taylor SM. Cell. 1980;20:85–93. doi: 10.1016/0092-8674(80)90237-8. [DOI] [PubMed] [Google Scholar]
  21. Jordan NJ, Kolios G, Abbot SE, Sinai MA, Thompson DA, Petraki K, et al. J Clin Invest. 1999;104:1061–1069. doi: 10.1172/JCI6685. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jubb AM, Quirke P, Oates AJ. Ann NY Acad Sci. 2003;983:251–267. doi: 10.1111/j.1749-6632.2003.tb05980.x. [DOI] [PubMed] [Google Scholar]
  23. Kang H, Watkins G, Parr C, Douglas-Jones A, Mansel RE, Jiang WG. Breast Cancer Res. 2005;7:R402–R410. doi: 10.1186/bcr1022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kimura R, Nishioka T, Ishida T. Genes Immun. 2003;4:356–361. doi: 10.1038/sj.gene.6363978. [DOI] [PubMed] [Google Scholar]
  25. Lenhard K, Bommer GT, Asutay S, Schauer R, Brabletz T, Goke B, et al. Clin Gastroenterol Hepatol. 2005;3:142–149. doi: 10.1016/s1542-3565(04)00624-x. [DOI] [PubMed] [Google Scholar]
  26. Li LC, Dahiya R. Bioinformatics. 2002;18:1427–1431. doi: 10.1093/bioinformatics/18.11.1427. [DOI] [PubMed] [Google Scholar]
  27. Muller A, Homey B, Soto H, Ge N, Catron D, Buchanan ME, et al. Nature. 2001;410:50–56. doi: 10.1038/35065016. [DOI] [PubMed] [Google Scholar]
  28. Nagasawa T, Hirota S, Tachibana K, Takakura N, Nishikawa S, Kitamura Y, et al. Nature. 1996;382:635–638. doi: 10.1038/382635a0. [DOI] [PubMed] [Google Scholar]
  29. Nagasawa T, Kikutani H, Kishimoto T. Proc Natl Acad Sci USA. 1994;91:2305–2309. doi: 10.1073/pnas.91.6.2305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Panis Y, Nordlinger B, Delelo R, Herve JP, Infante J, Kuhnle M, et al. J Hepatol. 1990;11:53–57. doi: 10.1016/0168-8278(90)90271-r. [DOI] [PubMed] [Google Scholar]
  31. Phillips RJ, Burdick MD, Lutz M, Belperio JA, Keane MP, Strieter RM. Am J Respir Crit Care Med. 2003;167:1676–1686. doi: 10.1164/rccm.200301-071OC. [DOI] [PubMed] [Google Scholar]
  32. Ponomaryov T, Peled A, Petit I, Taichman RS, Habler L, Sandbank J, et al. J Clin Invest. 2000;106:1331–1339. doi: 10.1172/JCI10329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rhee I, Bachman KE, Park BH, Jair KW, Yen RW, Schuebel KE, et al. Nature. 2002;416:552–556. doi: 10.1038/416552a. [DOI] [PubMed] [Google Scholar]
  34. Rhee I, Jair KW, Yen RW, Lengauer C, Herman JG, Kinzler KW, et al. Nature. 2000;404:1003–1007. doi: 10.1038/35010000. [DOI] [PubMed] [Google Scholar]
  35. Robert MF, Morin S, Beaulieu N, Gauthier F, Chute IC, Barsalou A, et al. Nat Genet. 2003;33:61–65. doi: 10.1038/ng1068. [DOI] [PubMed] [Google Scholar]
  36. Robertson KD. Oncogene. 2001;20:3139–3155. doi: 10.1038/sj.onc.1204341. [DOI] [PubMed] [Google Scholar]
  37. Scherf M, Klingenhoff A, Werner T. J Mol Biol. 2000;297:599–606. doi: 10.1006/jmbi.2000.3589. [DOI] [PubMed] [Google Scholar]
  38. Schrader AJ, Lechner O, Templin M, Dittmar KE, Machtens S, Mengel M, et al. Br J Cancer. 2002;86:1250–1256. doi: 10.1038/sj.bjc.6600221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Shirozu M, Nakano T, Inazawa J, Tashiro K, Tada H, Shinohara T, et al. Genomics. 1995;28:495–500. doi: 10.1006/geno.1995.1180. [DOI] [PubMed] [Google Scholar]
  40. Smith JM, Johanesen PA, Wendt MK, Binion DG, Dwinell MB. Am J Physiol Gastrointest Liver Physiol. 2005;288:316–326. doi: 10.1152/ajpgi.00208.2004. [DOI] [PubMed] [Google Scholar]
  41. Sun YX, Wang J, Shelburne CE, Lopatin DE, Chinnaiyan AM, Rubin MA, et al. J Cell Biochem. 2003;89:462–473. doi: 10.1002/jcb.10522. [DOI] [PubMed] [Google Scholar]
  42. Tachibana K, Hirota S, Iizasa H, Yoshida H, Kawabata K, Kataoka Y, et al. Nature. 1998;393:591–594. doi: 10.1038/31261. [DOI] [PubMed] [Google Scholar]
  43. Whitehead RH, Demmler K, Rockman SP, Watson NK. Gastroenterology. 1999;117:858–865. doi: 10.1016/s0016-5085(99)70344-6. [DOI] [PubMed] [Google Scholar]
  44. Zeelenberg IS, Ruuls-Van Stalle L, Roos E. Cancer Res. 2003;63:3833–3839. [PubMed] [Google Scholar]
  45. Zou YR, Kottmann AH, Kuroda M, Taniuchi I, Littman DR. Nature. 1998;393:595–599. doi: 10.1038/31269. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Figure
NIHMS728357-supplement-Supplemental_Figure.jpg (388.3KB, jpg)

RESOURCES