Although transforming growth factor-β (TGF-β) is both a suppressor and promoter of tumorigenesis, its contribution to early tumor suppression and staging remains largely unknown. In search of the mechanism of early tumor suppression, we identified the adaptor protein ELF, a β-spectrin from stem/progenitor cells committed to foregut lineage. ELF activates and modulates Smad4 activation of TGF-β to confer cell polarity, to maintain cell architecture, and to inhibit epithelial-to-mesenchymal transition. Analysis of development of colon cancer in (adult) elf+/−/Smad4+/−, elf+/−, Smad4+/−, and gut epithelial cells from elf−/− mutant mouse embryos pinpoints the defect to hyperplasia/adenoma transition. Further analysis of the role of ELF in human colorectal cancer confirms reduced expression of ELF in Dukes' B1 stage tissues (P < 0.05) and of Smad4 in advanced colon cancers (P < 0.05). This study indicates that by modulating Smad 4, ELF has a key role in TGF-β signaling in the suppression of early colon cancer.

The transforming growth factor-β (TGF-β) signaling pathway has been known to play an important role in gastrointestinal epithelial cell homeostasis; cell differentiation, proliferation, and migration; and modulation of gastrointestinal cancers (13). The enlarging TGF-β superfamily comprises >40 members, which include TGF-β1, TGF-β2, and TGF-β3; bone morphogenetic proteins; activins; Nodal; Lefty-1; Lefty-2; anti-Müllerian hormone; and other growth/differentiation factors (4, 5). Despite the diverse and complex responses they elicit, the basic signaling cascade of TGF-β is composed of TGF-β type I and type II transmembrane serine/threonine kinase receptors (TβRI and TβRII), and the cellular response is controlled by intracellular signaling proteins, Smads (6).

TGF-β ligand binding results in phosphorylation at Gly-Ser (GS) in the cytoplasmic tail domain of TβRI by TβRII, activation of Smad2, and Smad3 phosphorylation at the COOH-terminal serines (7, 8). Subsequent heteromeric complex formation with the L3 loop region phosphoserine-binding pockets of Smad4 facilitates nuclear translocation and TGF-β target gene activation (911). Adaptor proteins are required for functional specificity and Smad modulation. In previous studies, we have shown that ELF, a β-spectrin, is a crucial adaptor protein in TGF-β signaling and is required for Smad3 and Smad4 localization and signaling (12, 13). This finding was interesting because β-spectrins are major dynamic scaffold molecules involved in generating functionally distinct membrane protein domains, conferring cell polarity and regulating endocytic traffic (14, 15).

Originally described for its transforming capability, TGF-β is also a growth inhibitor in epithelial tissues as it is both a suppressor and promoter of tumorigenesis. It has been suggested that nearly all colon cancers, pancreatic cancers, and gastric carcinomas have mutations inactivating some component of TGF-β signaling from TβRII frameshift mutations with microsatellite instability to mutations in Smad4, Smad2, or an as yet untested component of the TGF-β signaling pathway (1619). Genetic studies in mice have provided strong models and further evidence for the role of TGF-β in tumor suppression in early stages. Tgf-β−/−/Rag2−/− mutant mice that live to adulthood rapidly develop colon cancer by 5 months of age, preceded by precancerous lesions with inflammation and hyperplasia (20). Smad4 deficiency in the ApcΔ716 mouse increases adenoma size and promotes cancer progression (2) and Smad4+/− mutant mice develop gastric polyps and carcinomas (17).

To understand the role of ELF in Smad4 tumor suppression, we intercrossed elf+/− mutant mice with Smad4+/− mutant mice and determined a genetic basis for ELF/Smad4 interaction in early colorectal cancers. It was possible that, in addition to its involvement in TGF-β signaling, ELF, as a β-spectrin, could be important in conferring cell polarity and maintaining cell architecture. Support for this hypothesis comes from embryonic day 11.5 (E11.5) elf−/− mutant embryos that display a profoundly abnormal gut phenotype, with flattened gut epithelial cells and loss of villi. We then investigated the abnormal gut phenotype and determined that the distribution of cell polarization markers dependent upon β-spectrin, such as E-cadherin, Na+-K+-ATPase, and microtubule associated protein-2 (MAP-2), was altered. We found that the cellular polarization through abnormal distribution of proteins, such as Na+-K+-ATPase, could be restored by ELF expression in elf−/− mutant cells. However, alterations in MAP-2 expression were not associated with aberrant expression of tubulin or microtubule function. Finally, we extrapolated our findings of abnormal cellular architecture and early colon cancer suppression to human studies. A strikingly reduced expression of ELF alone was seen in Dukes' B1 stage tissues (P < 0.05) and with a concomitant loss of Smad4 expression in advanced colon cancers (Dukes' D stage, P < 0.05). Our results indicate that ELF maintains cellular polarization by localization of a specific subset of proteins and is involved in preservation of cell architecture, the disruption of which may be a key to early colon cancer development.

Generation of elf+/−/Smad4+/− mice.Elf−/− mice die predominantly at E11.5. Mice heterozygous for the elf mutation (elf+/−) are normal and fertile. Elf+/− mice were intercrossed with Smad4+/− mice to generate elf+/−/Smad4+/− mutants to analyze the onset of colon adenomas. Elf+/−/Smad4+/− mutations were maintained on a mixed 129SvEv/Black Swiss background. The presence of mutations was monitored by use of the PCR as described previously (13, 17).

Confocal laser scanning immunofluorescence microscopy. Colocalization studies were done with antibodies against ELF and Smad4 on normal gastric tissues, wild-type mouse gastric antral cells, or mouse embryonic fibroblasts (MEF). Monoclonal mouse and polyclonal goat and rabbit primary antibodies were visualized with tetramethyl rhodamine isothiocyanate–conjugated goat secondary rabbit IgG or FITC-conjugated goat anti-mouse IgG or cyanine (Cy5). The samples were analyzed with a Bio-Rad Radiance 2000 confocal microscope (Bio-Rad, Cambridge, MA) with an ILT model 5470 K laser (Ion Laser Technology, Salt Lake City, UT). Digital images were analyzed using Metamorph (Universal Imaging, Downingtown, PA) and figures were prepared using Adobe Photoshop.

Generation of mouse embryo–derived fibroblasts. Mouse embryo–derived fibroblasts harboring the null allele elf as well as wild-type, Elf knockout (elf−/−), and elf wild-type (elf+/+) MEFs were derived as previously described (13). Briefly, embryos E14.5 were tritrurated in 0.25% trypsin/1 mmol/L EDTA and genotyped as previously described. The lines were propagated in DMEM supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 50 μg/mL streptomycin to establish wild-type and elf−/− fibroblasts that were cultured over multiple passages to obtain sufficient cells to perform the experiments. The fibroblasts used for the experiments were at passages 3 to 25. Three different elf−/− and wild-type fibroblast lines were tested in different experiments and the results obtained were also independent of passage number. Representative data are shown.

Histology and immunohistochemical staining. Colon carcinomas detected in mutant mice were fixed in 10% neutral buffered formalin and paraffin-embedded, and later, H&E staining was used to confirm histologic characterization. An indirect immunoperoxidase procedure was used for immunohistochemical localization of ELF and Smad4 proteins in colorectal cancer tissue samples. Serial, sagittal sections of colorectal cancer tissues were immersed in xylene to remove paraffin, then dehydrated in graded alcohol, and rinsed in 1× PBS. Endogenous peroxide was quenched using 3% hydrogen peroxide (Sigma, St. Louis, MO). Nonspecific binding sites were blocked using 1 mL PBS containing 5% goat serum and 1 mg/mL bovine serum albumin (BSA). The sections were incubated overnight at 4°C in a humidor with primary antibodies, monoclonal antibody to Smad4 linker region (Zymed, San Francisco, CA), and peptide-specific antibody to ELF (13) diluted to 2.5 to 5 μg/mL in 1× PBS containing 1 mg/mL BSA. Primary antibodies to Na+-K+-ATPase, vimentin, E-cadherin, and ankyrin (Santa Cruz Biotechnology, Santa Cruz, CA) were used to for cell polarization studies. In addition, to examine cell proliferation and apoptosis in mouse colon tissues, primary antibodies of anti-p-histone H3 (Ser10), a mitosis maker (Upstate, Charlottesville, VA), and anti-active caspase 3 (Promega, Madison, WI) were used. All further steps were carried out at room temperature. Four 5-minute rinses with 1× PBS followed each successive step. The sections were then incubated with peroxidase-conjugated secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA) as previously described (13). The secondary antibody was diluted in 1× PBS containing 1% goat serum. After rinses, 200 to 500 μL of the insoluble peroxidase substrate 3,3′-diaminobenzidine (Sigma) was added to cover the entire tissue on the slide, and we monitored color development under the microscope. After rinsing in distilled water for 2 minutes, we counterstained with modified Harris hematoxylin solution (Sigma) for 1 minute followed by a rinse in distilled water for 3 minutes. Sections were dehydrated by passage through graded alcohol concentrations and finally xylene. Coverslips were mounted using DPX (Sigma) before observation.

Cell polarity assays. Antibodies to Na+-K+-ATPase, vimentin, E-cadherin, and tubulin were used for immunohistologic characterization of the gut in elf−/− and wild-type mice as described above. In transient transfection assays, elf−/− and wild-type MEFs were seeded at a density of 2 × 105 cells/well in six-well dishes. They were then transfected with a construct that contains the full-length of elf or vector DNA alone (3 μg of DNA per well). For localization studies, transfections with pEYFP-Mem (encoding GAP-43, which contains a signal for posttranslational palmitoylation of cysteines 3 and 4 that targets membranes), pEYFP-Nuc (encoding a gene with three copies of the nuclear localization signal of the SV40 large antigen fused at its COOH terminus), and pECFP-Golgi (encoding the NH2-terminal human β-1,4-galactosyltransferase that helps in targeting the fusion protein to the trans-medial region of the Golgi apparatus) were used. Transfected cells were washed 2× with DMEM after 12 to 18 hours and then treated with 5 μg/mL TGF-β and incubated for an additional 24 hours. All experiments were repeated at least thrice and similar results were obtained each time. Cells were then fixed and analyzed by confocal microscopy as above.

Luciferase assays.Elf−/− and elf+/+ MEFs were plated 1 day before transfection, in 12-well plates, at a density of 1.5 × 105 to 2.5 × 105 cells per well in DMEM medium (10% fetal bovine serum, 1% P/S, 1% l-glutamine). For TGF-β response assays, the cells were transfected with p3TP-lux (1.5 μg) in controls and in cells treated with placitaxel (a microtubule-stabilizing agent; Sigma) or nocodazole (a microtubule-disrupting agent; Sigma). The cells were subsequently incubated for 18 hours with or without TGF-β1 (100 pmol/L). In mutant rescue assays, the elf+/+ and elf−/− cells were transfected with a construct that contains the full-length of elf or vector DNA alone and subsequently treated with TGF-β1 (100 pmol/L). Protein normalized luciferase activity in cell lysates was measured in a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA) by using substrate prepared in accordance with Promega luciferase assay system. All assays were carried out in duplicate or triplicate.

TGF-β–responsive regions of the human PAI-1 promoter contain DNA elements homologous to the consensus binding site for the transcription factor AP1. The p3TP-Lux luciferase reporter is a well-described and widely used artificial promoter construct that was empirically designed to have maximal responsiveness to TGF-β. Interestingly, using a variety of TGF-β–dependent luciferase gene reporters, it has been found that only Smad3 can directly interact with Smad binding elements found in the promoters of many TGF-β–responsive genes. Thus, Smad3 activates transcription of the 3TP-Lux, a reporter that contains elements of the PAI-l promoter. Functional effects of reduced Smad3 expression are seen in our model as reduced basal and TGF-β1–stimulated levels of luciferase expression using the TGF-β1–responsive reporter construct 3TP-Lux in the elf null-phenotype cells compared with control cells.

Cancer specimens. Formalin-fixed and paraffin-embedded colorectal cancer and colon specimens were obtained from the Department of Pathology, University of Pennsylvania Medical Center-Presbyterian. Twenty-one colorectal cancers collected from patients with varying grades and stages of colorectal cancer, identified by the Dukes' classification, were analyzed for ELF and Smad4 expression. All the specimens were collected after colectomy. Tissues were collected randomly at various zones of cancer manifestation, including the rectum, and the ascending, transverse, descending, and sigmoid colon. Tumor grade was determined by histology and markers, such as carcinoembryonic antigen (CEA; Table 1). In staging the tumors, Dukes' classification B1, C1, and D stages represent dysplasia, adenomas, and invasive carcinomas, respectively. Two independent blinded pathologists evaluated the tumors used in the study. The control samples of normal colon tissue used in the present investigation were taken from the borders of the surgical specimens.

Table 1.

Clinical classification of 21 colorectal cancer tissues used in the study

Dukes' stageCEAMetastaticHistology
3.9 Moderately diffused carcinoma 
B1 <1.0 Moderate to well-diffused carcinoma 
B1 3.1 Moderately diffused adenocarcinoma 
C1 2.7 CT na Adenocarcinoma 
B1 1.2 Moderately diffused invasive adenocarcinoma 
B1 Moderately diffused invasive adenocarcinoma 
B1 1.3 Two invasive moderately diffused adenocarcinoma 
B1 3.5 Moderately diffused adenocarcinoma 
C1 3.5 CT na Invasive moderately diffused adenocarcinoma 
907 Moderate to poorly diffused adenocarcinoma and metastasis to liver 
12 Moderately diffused invasive adenocarcinoma 
B1 <1.0 Moderately diffused adenocarcinoma 
C1 3.1 Invasive adenocarcinoma 
C1 3.5 Invasive adenocarcinoma 
B1 1.6 Moderately diffused adenocarcinoma 
2,915 Moderately diffused adenocarcinoma with metastasis to the ovary 
C2 1.7 Moderately diffused adenocarcinoma 
5,814 Invasive moderately diffused adenocarcinoma 
N/A No cancer Tubular adenoma–high-grade dysplasia 
B1 na Moderately diffused adenocarcinoma 
N/A No cancer No cancer in this segment of the bowel 
B1 na Moderately diffused adenocarcinoma 
Dukes' stageCEAMetastaticHistology
3.9 Moderately diffused carcinoma 
B1 <1.0 Moderate to well-diffused carcinoma 
B1 3.1 Moderately diffused adenocarcinoma 
C1 2.7 CT na Adenocarcinoma 
B1 1.2 Moderately diffused invasive adenocarcinoma 
B1 Moderately diffused invasive adenocarcinoma 
B1 1.3 Two invasive moderately diffused adenocarcinoma 
B1 3.5 Moderately diffused adenocarcinoma 
C1 3.5 CT na Invasive moderately diffused adenocarcinoma 
907 Moderate to poorly diffused adenocarcinoma and metastasis to liver 
12 Moderately diffused invasive adenocarcinoma 
B1 <1.0 Moderately diffused adenocarcinoma 
C1 3.1 Invasive adenocarcinoma 
C1 3.5 Invasive adenocarcinoma 
B1 1.6 Moderately diffused adenocarcinoma 
2,915 Moderately diffused adenocarcinoma with metastasis to the ovary 
C2 1.7 Moderately diffused adenocarcinoma 
5,814 Invasive moderately diffused adenocarcinoma 
N/A No cancer Tubular adenoma–high-grade dysplasia 
B1 na Moderately diffused adenocarcinoma 
N/A No cancer No cancer in this segment of the bowel 
B1 na Moderately diffused adenocarcinoma 

Abbreviations: N/A, not applicable; na, not available; CT, computed tomography scan.

Statistical analysis. Global χ2 test was used to test the hypothesis that the coefficient of each variable was equal to 0. Tissue sample sets of immunohistochemical data were compared with assess the significance. A P value <0.05 was required for statistical significance, and all tests were two-sided. All tests were done with SPSS 10.1 software (SPSS, Inc., Chicago, IL).

Elf+/−/Smad4+/− intercrosses establish the synergistic role of ELF and Smad4 in colorectal cancers. To determine the role of ELF and Smad4 in colorectal cancer, we intercrossed elf+/− and Smad4+/− mice. Out of 19 elf+/−/Smad4+/− mice, 3 developed colon adenomas as early as 8 to 13 months of age (Fig. 1A) and almost all of them developed gastric tumors at 12 months of age,7

7

Unpublished observations.

whereas in Smad4+/− heterozygotes late onset carcinomas were seen (17). In elf+/−, late-onset hepatocellular cancers are seen. In contrast, we did not find any colorectal polyps, tumors, and adenomas in either elf+/− or Smad4+/− mice at the age of 8 to 12 months. H&E staining of the colon adenoma sections revealed aberrant crypts with loss of normal cellular structure in elf+/−/Smad4+/− mice (Fig. 1B and C) compared with normal wild-type colon tissues (Fig. 1D and E). Interestingly, we also noted aberrations in expression patterns of vimentin (Fig. 1F-K).

Figure 1.

Histologic analysis of colon polyps in wild-type and elf+/−/Smad4+/− mice. A, macroscopic analysis of colon tumor development in elf+/−/Smad4+/− mice. B and C, H&E-stained sections of colon polyps in elf+/−/Smad4+/− mice. D and E, H&E-stained sections of normal wild-type colon tissues. Immunostaining with vimentin in E11.5 wild-type (F) and elf−/− mutant (G) gut tissues. Immunohistochemical detection of E-cadherin in E11.5 wild-type (H) and elf−/− mutant (I) gut tissues, adult wild-type normal colon tissues (J), and elf+/−/Smad4+/− mutant colon tissues (K).

Figure 1.

Histologic analysis of colon polyps in wild-type and elf+/−/Smad4+/− mice. A, macroscopic analysis of colon tumor development in elf+/−/Smad4+/− mice. B and C, H&E-stained sections of colon polyps in elf+/−/Smad4+/− mice. D and E, H&E-stained sections of normal wild-type colon tissues. Immunostaining with vimentin in E11.5 wild-type (F) and elf−/− mutant (G) gut tissues. Immunohistochemical detection of E-cadherin in E11.5 wild-type (H) and elf−/− mutant (I) gut tissues, adult wild-type normal colon tissues (J), and elf+/−/Smad4+/− mutant colon tissues (K).

Close modal

Increased cell proliferation and suppressed apoptosis in elf+/−/Smad4+/− colonic mucosa. The occurrence of precancerous colonic polyps in these double heterozygotes compared with none in the elf+/− or Smad4+/− colons suggests that disruption of elf, in addition to disruption of Smad4, results in hyperplasia of the colonic mucosa. Increased cell proliferation was observed in elf+/−/Smad4+/− colon tissues (Figs. 2C,, arrows, and S1C), compared with normal colon tissues (Figs. 2A  and S1A), when using immunochemical labeling by the anti-p-Histone H3 (Ser10) antibody (a mitosis marker; ref. 21). This indicated that colonic epithelial cell proliferation was stimulated by the disruption of elf.

Figure 2.

Cell proliferation and apoptosis in wild-type, Smad4+/−, and elf+/−/Smad4+/− colonic mucosa. Immunohistochemical detection of mitotic cells in normal wild-type (A), Smad4+/− (B), and elf+/−/Smad4+/− (C) mutant colon tissues. D to F, identification of apoptotic cells in normal wild-type, Smad4+/−, and elf+/−/Smad4+/− mutant colon tissues, using antiactive caspase 3 antibody. Arrows, mitotic (B and C) and apoptotic cells (D and E).

Figure 2.

Cell proliferation and apoptosis in wild-type, Smad4+/−, and elf+/−/Smad4+/− colonic mucosa. Immunohistochemical detection of mitotic cells in normal wild-type (A), Smad4+/− (B), and elf+/−/Smad4+/− (C) mutant colon tissues. D to F, identification of apoptotic cells in normal wild-type, Smad4+/−, and elf+/−/Smad4+/− mutant colon tissues, using antiactive caspase 3 antibody. Arrows, mitotic (B and C) and apoptotic cells (D and E).

Close modal

Loss of response to TGF-β signaling in elf+/−/Smad4+/− cells could indicate that colonic epithelial cell apoptosis may also be altered in the elf+/−/Smad4+/− mutants. We further examined epithelial apoptosis in the colonic tissue with anticaspase 3 labeling. In wild-type control mice, apoptosis was noted in crypt epithelial cells (Figs. 2D,, arrows, and S1D) but few apoptotic cells were seen in elf+/−, Smad4+/−, and elf+/−/Smad4+/− mutant colonic epithelium (Figs. 2F, and S1F). Therefore, we observed significantly decreased apoptotic cells in elf+/−/Smad4+/− colon tissues (Figs. 2F, and S1F) compared with normal colon tissues (Figs. 2D , arrows, and S1D), when using the apoptotic marker, antiactive caspase 3. The suppressed apoptosis indicates that elf may be important in the TGF-β induction of apoptosis in crypt epithelial cells and may contribute to the epithelial cell hyperplasia in the elf+/−/Smad4+/− colonic mucosa.

Loss of polarity in elf−/− gut epithelial cells. We first investigated abnormalities in intestinal epithelial cell morphology in the elf−/− mutant embryos, then analyzed the role of ELF/TGF-β in organelle formation and gut epithelial cell polarity. H&E-stained sections showed gut epithelial cells to be severely flattened with loss of villi in E11.5 elf−/− mutant embryos. Further visualization of plasma membrane, Golgi, and nuclei by transfection of pEYFP-Mem, pECFP-Golgi, and pEYFP-Nuc (Clontech, Palo Alto, CA) into elf−/− mutant MEFs revealed all three to be normal.7 Strikingly, we observed a marked distortion in expression patterns of Na+-K+-ATPase, MAP-2, and actin with increased vimentin, decreased E-cadherin, but normal ankyrin B and G expression in the elf−/− and decreased E-cadherin expression in the elf+/−/Smad4+/− gut epithelial cells (Figs. 1F,-K and S2). Na+-K+-ATPase labeling showed irregular and punctate intracellular localization, and it seemed absent at the apical (brush border) plasma membrane in the elf+/−/Smad4+/− colonic mucosa and aberrantly localized in elf−/− cells (Fig. 3C and E) compared with normal, wild-type control, as well as Smad4+/− colonic mucosa (Fig. 3A , B, and D).

Figure 3.

ELF rescues loss of polarity and Na+-K+-ATPase expression in elf−/− mutants. A to C, immunohistochemical labeling of wild-type (A), Smad4+/− (B), and elf+/−/Smad4+/− (C) mutant colon tissues with anti–Na+-K+-ATPase. D, immunofluorescent confocal microscopy showing normal Na+-K+-ATPase distribution (rhodamine) in wild-type MEFs (arrow) transfected with pcDNA3.1 DNA only. E, Na+-K+-ATPase expression is decreased and aberrant in elf−/− MEFs. F, Na+-K+-ATPase expression is rescued by overexpression of ELF (arrow). G, Western blots show ELF expression in elf−/− MEFs transfected with a construct containing the full-length of elf cDNA (lane 3) restored back to its endogenous level, when compared with the ELF expression in both wild-type (lane 1) and elf−/− (lane 2) MEFs transfected with pcDNA3.1 only. H, mutant rescue assays demonstrating restoration of TGF-β response in elf−/− MEFs on transfection with full-length elf. I, schematic representation of the role of ELF in the TGF-β-Smad signaling pathway.

Figure 3.

ELF rescues loss of polarity and Na+-K+-ATPase expression in elf−/− mutants. A to C, immunohistochemical labeling of wild-type (A), Smad4+/− (B), and elf+/−/Smad4+/− (C) mutant colon tissues with anti–Na+-K+-ATPase. D, immunofluorescent confocal microscopy showing normal Na+-K+-ATPase distribution (rhodamine) in wild-type MEFs (arrow) transfected with pcDNA3.1 DNA only. E, Na+-K+-ATPase expression is decreased and aberrant in elf−/− MEFs. F, Na+-K+-ATPase expression is rescued by overexpression of ELF (arrow). G, Western blots show ELF expression in elf−/− MEFs transfected with a construct containing the full-length of elf cDNA (lane 3) restored back to its endogenous level, when compared with the ELF expression in both wild-type (lane 1) and elf−/− (lane 2) MEFs transfected with pcDNA3.1 only. H, mutant rescue assays demonstrating restoration of TGF-β response in elf−/− MEFs on transfection with full-length elf. I, schematic representation of the role of ELF in the TGF-β-Smad signaling pathway.

Close modal

To further explore the role of ELF in cell polarization and to investigate the possibility of rescuing localization of Na+-K+-ATPase signaling in the elf−/− mutants by rescuing Na+-K+-ATPase membrane localization through restoration of ELF activity, we transiently transfected full-length elf into the elf−/− mutant MEFs (transfection efficiency of 43.8 ± 2.5 %; Fig. 3D,-F, arrows). Correction of Na+-K+-ATPase localization at the membrane by transient transfection of elf−/− MEFs with full-length elf was documented by confocal immunofluorescent microscopy (Fig. 3F; the frequency of correction being 41 ± 1.8%). These data suggest that a functional ELF spectrin with inherent dynamic stability that is responsive to environmental cues may represent a key regulatory element for Na+-K+-ATPase modulation.

Microtubules have been shown to modulate TGF-β–induced Smad signaling (22). We noted an aberrant expression of MAP-2 in elf−/− mutant embryonic and in the elf+/−/Smad4+/− colonic mucosa adult gut tissue (Fig. 4B and E) compared with normal wild-type embryonic and adult elf control tissues (Fig. 4A and C). Interestingly, both β-spectrin and MAP-2 have been shown to be important for microtubule bundling and function (23, 24). To exclude abnormalities in microtubule function as a cause of the observed phenotype seen in elf−/− mutant embryos, we analyzed microtubule distribution and function in wild-type and elf−/− mutant MEFs and in embryos. Immunofluorescence confocal microscopy determined that the subcellular distribution of β-tubulin is unaltered in the mutant embryos.7 Furthermore, neither a microtubule-stabilizing agent (placitaxel) nor a microtubule-disrupting agent (such as nocodazole) corrected TGF-β signaling in elf−/− mutant fibroblasts (Fig. 4F). Lack of response to TGF-β stimulation in mutant cells suggests that microtubule modulation of Smads may be less relevant and secondary to ELF spectrins. This was further supported by transient transfection of the reporter construct p3TP-Lux, into wild-type and elf−/− cultured MEFs. When we treated transfected wild-type MEFs with TGF-β1, the luciferase activity induced was seven to eight times that of elf−/− MEFs (Fig. 3H). However, in MEFs derived from elf−/− mouse embryos, TGF-β1–dependent induction of p3TP-Lux was abolished, as it was in vector controls, which indicated that the TGF-β1 response needs ELF (Fig. 3H). In mutant rescue assays, restoration of elf in elf−/−-cultured MEFs dramatically induced a 7- to 8-fold increase in luciferase activity, almost as in the wild type (elf+/+).

Figure 4.

Loss of polarity in Smad4+/− and elf+/−/Smad4+/− colon tissues. Immunohistochemical detection of MAP-2 in E11.5 wild-type (A) and elf−/− mutant (B) gut tissues. Loss of polarized expression of MAP-2 is also observed in Smad4+/− (D) and, to a greater extent, in elf+/−/Smad4+/− (E) colon tissues when compared with normal wild-type tissues (C). F, analysis of the TGF-β response in control elf+/+ and elf−/− cell lines treated with placitaxel or nocodozole under transient transfection conditions.

Figure 4.

Loss of polarity in Smad4+/− and elf+/−/Smad4+/− colon tissues. Immunohistochemical detection of MAP-2 in E11.5 wild-type (A) and elf−/− mutant (B) gut tissues. Loss of polarized expression of MAP-2 is also observed in Smad4+/− (D) and, to a greater extent, in elf+/−/Smad4+/− (E) colon tissues when compared with normal wild-type tissues (C). F, analysis of the TGF-β response in control elf+/+ and elf−/− cell lines treated with placitaxel or nocodozole under transient transfection conditions.

Close modal

ELF localization in human colon cancers. ELF expression was observed along the tubular crypts of Leiberkühn in the mucosa of control colon tissues (Fig. 5A). Strong labeling was observed at the base of the crypt and a gradient of moderate to intense labeling in the apical regions. ELF expression was mostly cytoplasmic. The muscularis mucosa beneath the deep end of the crypts also exhibited faint labeling (Fig. 5A). Of the normal colons, 66% showed intense labeling and 14.2% showed moderate labeling (P ≤ 0.05) of the basal crypts (Tables 1 and 2). The lamina propria separating the crypts with loose connective tissue, capillaries, and strands of smooth muscle did not label strongly for ELF. Labeling for ELF in colorectal cancer tissue samples showed reduction or loss of expression in aberrant crypt foci (Fig. 5B and C). In Dukes' B1, an early stage of cancer, ELF expression was reduced and abnormal compared with that of controls, which suggested a down-regulation of this gene in early neoplasia (Fig. 5B; Table 2). Abnormal ELF expression was also observed in the epithelial outpockets from the basal crypts that form adenomatous polyclonal polyps, an obligatory step for tumor progression. In adenomas, 14.2% showed moderate labeling and 23.8% showed intense labeling; almost 57.1% exhibited disruption of ELF. About 24% and 15% of carcinomas showed moderate and intense labeling for ELF, respectively (P ≤ 0.05). ELF expression was strikingly absent in 9 of 10 B1 and 2 of 4 C1 cancers (Fig. 5B and C; Tables 1 and 2). Aberrant or reduced ELF expression or loss of it was observed in almost all B1 colon cancers. In C1 and D1 late-stage tumors, we observed loss of expression in the crypt epithelial cells and a simultaneous positive labeling in the stromal tissue.

Figure 5.

Immunohistochemical analysis of human colorectal cancers. A, ELF immunostaining in normal human colon tissue. Labeling at the base of the crypts of Leiberkühn and at the tips is observed. B, ELF expression in Dukes' B1 is significantly reduced or lost in adenomatous polyps. C, a marked decrease in ELF expression in Dukes' D tumors. D, anti-Smad4-labeled normal colon controls. E and F, reduced or loss of Smad4 expression in the basal crypts of B1 and D stage tumors, respectively. G, negative control (normal colonic tissue with secondary antibody only).

Figure 5.

Immunohistochemical analysis of human colorectal cancers. A, ELF immunostaining in normal human colon tissue. Labeling at the base of the crypts of Leiberkühn and at the tips is observed. B, ELF expression in Dukes' B1 is significantly reduced or lost in adenomatous polyps. C, a marked decrease in ELF expression in Dukes' D tumors. D, anti-Smad4-labeled normal colon controls. E and F, reduced or loss of Smad4 expression in the basal crypts of B1 and D stage tumors, respectively. G, negative control (normal colonic tissue with secondary antibody only).

Close modal
Table 2.

Immunohistochemical labeling results for 21 colorectal cancers

Cancer typeLabeling for
ELF
Smad4
+++NAP+++NAP
Normal 14 − 0.0043 0.0401 
Adenoma 11 0.011 0.0504 
Carcinoma 10 0.016 11 0.0304 
Cancer typeLabeling for
ELF
Smad4
+++NAP+++NAP
Normal 14 − 0.0043 0.0401 
Adenoma 11 0.011 0.0504 
Carcinoma 10 0.016 11 0.0304 

NOTE: Data obtained were statistically significant at P ≤ 0.05. ++, intense; +, moderate; −, loss of or reduced labeling.

Abbreviation: NA, not available.

Smad4 localization in relation to ELF expression in human colon cancers. In the control colorectal tissues, we observed labeling of Smad4 in the basal one third of the crypts of Leiberkühn similar to ELF labeling (Fig. 5A and D). Prominent ELF and Smad4 expression with an increased staining at the base of the crypt was an important correlative feature observed (Fig. 5A and D). A high level of expression was seen in the submucosa and muscularis mucosa at the deep end of the crypts but not in the lamina propria that separates the crypts (Fig. 5D). Of the normal tissues, 28% showed moderate labeling, whereas 20% and 38% showed intense and loss of labeling, respectively (P ≤ 0.05; Tables 1 and 2). In these normal samples, labeling was mostly in the cytoplasm.

We observed loss of Smad4 expression in B1, C1, and D tumors, as well as marked loss of expression in the late stages (Dukes' D cancers; Fig. 5E and F). In other types of tumors, 42.8% of adenoma had moderate staining, whereas 52.3% of carcinomas showed loss of Smad4 expression (P ≤ 0.05).

Adult elf+/−/Smad4+/− mutant mice bred onto a 129SvEv/Black Swiss background develop cancerous tumors of the colon. Here, we show that ELF suppresses early events in colon cancer formation. These studies are similar to findings from Tgf1−/− mice, in which the tumor suppressor activity of TGF-β was not directed at cell proliferation, suppression of inflammation or maintenance of genetic stability, or via regulation of adenomatous polyposis coli (APC) levels (2). Progression of tumors from adenomatous to in situ and invasive carcinomas may result from an inability to maintain normal tissue architecture (20). Spectrins are involved in the generation of cell polarity and protein sorting (2326); hence, it is conceivable that abnormalities in spectrin function could result in the partial or complete loss of cellular polarity, a characteristic feature of tumor cells (27).

Spectrins are key proteins involved in the support of general membrane integrity, stabilization of cell-to-cell interactions, axonal growth, and the formation of the sarcoplasmic reticulum (24, 28). They are also known to be involved in the generation of cell polarity and protein sorting (23, 29, 30). Spectrins create a multifunctional scaffold on which membrane proteins, cytoplasmic signaling molecules, and structural elements are organized in distinct domains; in this way, the general cytoarchitecture and tissue integrity of cells are maintained (24, 30).

Resistance to TGF-β is commonly associated with late events in tumorigenesis and probably is secondary to inactivating mutations in TβRII, Smad2, or Smad4. Our recent studies that show that the disruption of TGF-β signaling by inactivation of the adaptor protein β-spectrin, encoded by elf (14, 31), results in Smad4 localization and activation (13). Hence, aberrations in ELF expression should also affect the otherwise normal downstream events leading to abrogation of Smad4 activity. Studies have established the role of Smad4 as a tumor suppressor, and failure of Smad4 expression has been associated with advanced stage disease, the presence of lymph node metastasis, and a significant shorter overall survival (31, 32). The essential and adaptor protein ELF, a β-spectrin, may therefore play a similar role in suppressing gastrointestinal tumors.

Loss of β-spectrins, such as ELF, an important protein necessary for maintaining the structural integrity of epithelial cells, may be pivotal for epithelial cell integrity and maintenance of tissue architecture in the early stages. Establishment of spatial coordinates during differentiation of polarized cells involves a positional cue from cadherins that results in targeting of β-spectrin to a discrete plasma membrane domain (33). The spectrin tetramer is then able to capture and stabilize additional membrane interacting proteins, such as ankyrin and the basolateral Na+-K+-ATPase, to form the characteristic profile of a polarized membrane domain (34, 35). To facilitate simultaneous reception and transmission of positional information, one of the two binding sites on the tetramer for ankyrin would allow for interaction with a cell adhesion molecule for a positional clue, whereas the second ankyrin allows for acquisition of these studies indicate the requirement of β-spectrin for this interaction. The lack of polarized distribution of the Na+-K+-ATPase in the elf β-spectrin mutant phenotype resembles the Drosophila β-spectrin mutant, which in turn is reminiscent of the Drosophila labial phenotype, particularly in the gut (36). Control of the homeotic gene labial is dependent upon extracellular gradients of wingless and decapentaplegic (the Drosophila homologue of TGF-β during embryogenesis; ref. 37), which suggests that the relationship between elf and TGF-β is important for gut epithelial cell formation and is conserved through evolution.

Here, we show that loss of ELF expression correlates with B1 colon cancers and that a strict correlation exists between ELF and Smad4 expression in the normal epithelial cells of the crypts of Leiberkühn (Fig. 5A and D). Previous coimmunoprecipitation studies indicate an interaction between ELF and Smad4 that is essential for TGF-β–mediated gene expression in mutant mice, as well as in human gastric cancers (13).8

8

Unpublished observations.

Smad4 expression is prominent in normal colonic crypts with typically diffuse epithelial staining. Stronger labeling is observed in the villus apex and the bottom one third of the crypts. Loss of Smad4 expression in advanced colorectal tumors substantiates earlier reports and further supports the role of Smad4 as a gastrointestinal tumor suppressor and a potential marker in colorectal cancer (38).

The basal crypt region of the colonic epithelial cells is known to harbor stem cells, which proliferate and migrate toward the villus while differentiating into cell types (39). On reaching the villus apex, cells become apoptotic and are shed into the gut lumen. These repetitive cycles of proliferation, differentiation, and shedding are events that maintain the integrity of normal colonic epithelia. Expression of ELF and Smad4 in normal colon, particularly in the basal crypt region, suggests a functional role via the TGF-β signaling pathway. As the cells reach the midcrypt region, ELF activity is reduced along with cell differentiation. Absence of ELF expression, especially in the basal crypts of B1 and C1 tumors, suggests an inactivation of the TGF-β signaling pathway via the abrogation of Smad4 functions. These events may modulate the differentiation signals at the stem cell compartment (basal crypt) whereby the proliferative stage is maintained and may lead to tumor formation.

Interestingly, compared with Smad4, ELF expression was significantly lost or reduced in Dukes' B1 cancers, especially in the basal crypts correlating with loss of differentiation and the onset of colonic neoplasia. With further invasiveness and metastasis of the tumors, both ELF and Smad4 expression was diminished or lost particularly in the bottom third compartment and this may serve as a prognostic factor for a poor outcome. Our findings suggest that in normal colon samples, after TGF-β stimulation, ELF interacts with Smad4 in the cytoplasm and localizes to the nucleus for transcriptional control. Aberrant nuclear and cytoplasmic labeling of ELF in B1, C1, and D tumors also suggests mislocalization of ELF, which ablates TGF-β–induced transcriptional response.

Indeed, β-spectrins and ELF bind to E-cadherin via the β-catenin (40), an important mediator of the Wnt signaling pathway (41) that has a central role in colorectal carcinoma, controlling the switch between proliferation and differentiation in intestinal epithelial cells (42). Disruption of the G1 phase of the cell cycle blocks a genetic program that is physiologically active in the proliferative compartment of colon crypts. Regulation of intracellular β-catenin signaling through APC, originally identified from familial adenomatous polyposis patient studies, and by p53 potentially closely links the Wnt signaling to TGF-β signaling pathways (43, 44). With this scenario, it will be interesting to delineate the cross-talk between signaling pathways that involves ELF and the key players in colorectal carcinogenesis, such as Wnt signaling. In view of the present findings, we suggest that loss of ELF potentially contributes to the events that lead to onset of colorectal cancer and ELF probably plays an important role in tumor suppressor mechanisms in colon cancer that can be recognized as a potential early marker in colorectal carcinoma.

Note: Y. Tang and V. Katuri contributed equally to this work.

Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

Grant support: Sally Funderberg award, NIH grants RO1 DK56111 (L. Mishra) and RO1 DK58637 (B. Mishra), and Veterans Affairs Merit Award (L. Mishra).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Barnard JA, Beauchamp RD, Coffey RJ, Moses, HL. Regulation of intestinal epithelial cell growth by transforming growth factor type β.
Proc Natl Acad Sci U S A
1989
;
86
:
1578
–82.
2
Engle SJ, Hoying JB, Boivin GP, Ormsby I, Gartside PS, Doetschman T. Transforming growth factor β1 suppresses nonmetastatic colon cancer at an early stage of tumorigenesis.
Cancer Res
1999
;
59
:
3379
–86.
3
Oft M, Heider KH, Beug H. TGF-β signaling is necessary for carcinoma cell invasiveness and metastasis.
Curr Biol
1998
;
8
:
1243
–52.
4
Shi Y, Massague J. Mechanisms of TGF-β signaling from cell membrane to the nucleus.
Cell
2003
;
113
:
685
–700.
5
Sporn MB. TGF-β: 20 years and counting.
Microbes Infect
1999
;
1
:
1251
–3.
6
Piek E, Heldin CH, Ten Dijke P. Specificity, diversity, and regulation in TGF-β superfamily signaling.
FASEB J
1999
;
13
:
2105
–24.
7
Dore JJ Jr, Edens M, Garamszegi N, Leof EB. Heteromeric and homomeric transforming growth factor-β receptors show distinct signaling and endocytic responses in epithelial cells.
J Biol Chem
1998
;
273
:
31770
–7.
8
Kawabata M, Inoue H, Hanyu A, Imamura T, Miyazono K. Smad proteins exist as monomers in vivo and undergo homo- and hetero-oligomerization upon activation by serine/threonine kinase receptors.
EMBO J
1998
;
17
:
4056
–65.
9
Greene RM, Nugent P, Mukhopadhyay P, Warner DR, Pisano MM. Intracellular dynamics of Smad-mediated TGF-β signaling.
J Cell Physiol
2003
;
197
:
261
–71.
10
Penheiter SG, Mitchell H, Garamszegi N, Edens M, Dore JJ Jr, Leof EB. Internalization-dependent and -independent requirements for transforming growth factor β receptor signaling via the Smad pathway.
Mol Cell Biol
2002
;
22
:
4750
–9.
11
Qin BY, Lam SS, Correia JJ, Lin K. Smad3 allostery links TGF-β receptor kinase activation to transcriptional control.
Genes Dev
2002
;
16
:
1950
–63.
12
Derynck R, Zhang YE. Smad-dependent and Smad-independent pathways in TGF-β family signalling.
Nature
2003
;
425
:
577
–84.
13
Tang Y, Katuri V, Dillner A, Mishra B, Deng CX, Mishra L. Disruption of transforming growth factor-β signaling in ELF β-spectrin-deficient mice.
Science
2003
;
299
:
574
–7.
14
Mishra L, Cai T, Yu P, Monga SP, Mishra B. Elf3 encodes a novel 200-kD β-spectrin: role in liver development.
Oncogene
1999
;
18
:
353
–64.
15
Tang Y, Katuri V, Iqbal S, et al. ELF, a β-spectrin, is a neuronal precursor cell marker in developing mammalian brain; structure and organization of the elf/beta-G spectrin gene.
Oncogene
2002
;
21
:
5255
–67.
16
Villanueva A, Garcia C, Paules AB, et al. Disruption of the antiproliferative TGF-β signaling pathways in human pancreatic cancer cells.
Oncogene
1998
;
17
:
1969
–78.
17
Xu X, Brodie SG, Yang X, et al. Haploid loss of the tumor suppressor Smad4/Dpc4 initiates gastric polyposis and cancer in mice.
Oncogene
2000
;
19
:
1868
–74.
18
Grady WM, Myeroff LL, Swinler SE, et al. Mutational inactivation of transforming growth factor β receptor type II in microsatellite stable colon cancers.
Cancer Res
1999
;
59
:
320
–4.
19
Woodford-Richens KL, Rowan AJ, Gorman P, et al. SMAD4 mutations in colorectal cancer probably occur before chromosomal instability, but after divergence of the microsatellite instability pathway.
Proc Natl Acad Sci U S A
2001
;
98
:
9719
–23.
20
Engle SJ, Ormsby I, Pawlowski S, et al. Elimination of colon cancer in germ-free transforming growth factor β1-deficient mice.
Cancer Res
2002
;
62
:
6362
–6.
21
Siegel PM, Shu W, Cardiff RD, Muller WJ, Massague J. Transforming growth factor β signaling impairs Neu-induced mammary tumorigenesis while promoting pulmonary metastasis.
Proc Natl Acad Sci U S A
2003
;
100
:
8430
–5.
22
Dong C, Li Z, Alvarez R Jr, Feng XH, Goldschmidt-Clermont PJ. Microtubule binding to Smads may regulate TGF-β activity.
Mol Cell
2000
;
5
:
27
–34.
23
Bennett V, Baines AJ. Spectrin and ankyrin-based pathways: metazoan inventions for integrating cells into tissues.
Physiol Rev
2001
;
81
:
1353
–92.
24
Nelson WJ, Hammerton RW, Wang AZ, Shore EM. Involvement of the membrane-cytoskeleton in development of epithelial cell polarity.
Semin Cell Biol
1990
;
1
:
359
–71.
25
Thomas GH, Williams JA. Dynamic rearrangement of the spectrin membrane skeleton during the generation of epithelial polarity in Drosophila.
J Cell Sci
1999
;
112
:
2843
–52.
26
Stein M, Wandinger-Ness A, Roitbak T. Altered trafficking and epithelial cell polarity in disease.
Trends Cell Biol
2002
;
12
:
374
–81.
27
Bennett V, Gilligan DM. The spectrin-based membrane skeleton and micron-scale organization of the plasma membrane.
Annu Rev Cell Biol
1993
;
9
:
27
–66.
28
De Matteis MA, Morrow JS. Spectrin tethers and mesh in the biosynthetic pathway.
J Cell Sci
2000
;
113
:
2331
–43.
29
Dubreuil RR, Maddux PB, Grushko TA, MacVicar GR. Segregation of two spectrin isoforms: polarized membrane-binding sites direct polarized membrane skeleton assembly.
Mol Biol Cell
1997
;
8
:
1933
–42.
30
Mishra L, Cai T, Levine A, et al. Identification of elf1, a β-spectrin, in early mouse liver development.
Int J Dev Biol
1998
;
42
:
221
–4.
31
Xiangming C, Natsugoe S, Takao S, et al. Preserved Smad4 expression in the transforming growth factor β signaling pathway is a favorable prognostic factor in patients with advanced gastric cancer.
Clin Cancer Res
2001
;
7
:
277
–82.
32
Xie W, Rimm DL, Lin Y, Shih WJ, Reiss M. Loss of Smad signaling in human colorectal cancer is associated with advanced disease and poor prognosis.
Cancer J
2003
;
9
:
302
–12.
33
Piepenhagen PA, Nelson WJ. Biogenesis of polarized epithelial cells during kidney development in situ: roles of E-cadherin-mediated cell-cell adhesion and membrane cytoskeleton organization.
Mol Biol Cell
1998
;
9
:
3161
–77.
34
Davis JQ, Bennett V. The anion exchanger and Na+K(+)-ATPase interact with distinct sites on ankyrin in in vitro assays.
J Biol Chem
1990
;
265
:
17252
–6.
35
Dubreuil RR, Wang P, Dahl S, Lee J, Goldstein LS. Drosophila β spectrin functions independently of α spectrin to polarize the Na,K ATPase in epithelial cells.
J Cell Biol
2000
;
149
:
647
–56.
36
Dubreuil RR, Grushko T, Baumann O. Differential effects of a labial mutation on the development, structure, and function of stomach acid-secreting cells in Drosophila melanogaster larvae and adults.
Cell Tissue Res
2001
;
306
:
167
–78.
37
Yu X, Hoppler S, Eresh S, Bienz M. Decapentaplegic, a target gene of the wingless signalling pathway in the Drosophila midgut.
Development
1996
;
122
:
849
–58.
38
Salovaara R, Roth S, Loukola A, et al. Frequent loss of SMAD4/DPC4 protein in colorectal cancers.
Gut
2002
;
51
:
56
–9.
39
Potten CS, Loeffler M. Stem cells: attributes, cycles, spirals, pitfalls and uncertainties. Lessons for and from the crypt.
Development
1990
;
110
:
1001
–20.
40
Pradhan D, Lombardo CR, Roe S, Rimm DL, Morrow JS. α-Catenin binds directly to spectrin and facilitates spectrin-membrane assembly in vivo.
J Biol Chem
2001
;
276
:
4175
–81.
41
Bienz M, Clevers H. Linking colorectal cancer to Wnt signaling.
Cell
2000
;
103
:
311
–20.
42
Van de Wetering M, Sancho E, Verweij C, et al. The β-catenin/TCF-4 complex imposes a crypt progenitor phenotype on colorectal cancer cells.
Cell
2002
;
111
:
241
–50.
43
Abraham SC, Wu TT, Klimstra DS, et al. Distinctive molecular genetic alterations in sporadic and familial adenomatous polyposis-associated pancreatoblastomas: frequent alterations in the APC/β-catenin pathway and chromosome 11p.
Am J Pathol
2001
;
159
:
1619
–27.
44
Cagatay T, Ozturk M. P53 mutation as a source of aberrant β-catenin accumulation in cancer cells.
Oncogene
2002
;
21
:
7971
–80.

Supplementary data