Fax: (203) 737-4067
Small bowel carcinoid (enterochromaffin cell) neoplasia exhibits transforming growth factor–β1-mediated regulatory abnormalities including up-regulation of C-Myc and MTA1
Article first published online: 27 APR 2007
Copyright © 2007 American Cancer Society
Volume 109, Issue 12, pages 2420–2431, 15 June 2007
How to Cite
Kidd, M., Modlin, I. M., Pfragner, R., Eick, G. N., Champaneria, M. C., Chan, A. K., Camp, R. L. and Mane, S. M. (2007), Small bowel carcinoid (enterochromaffin cell) neoplasia exhibits transforming growth factor–β1-mediated regulatory abnormalities including up-regulation of C-Myc and MTA1. Cancer, 109: 2420–2431. doi: 10.1002/cncr.22725
- Issue published online: 4 JUN 2007
- Article first published online: 27 APR 2007
- Manuscript Accepted: 9 MAR 2007
- Manuscript Revised: 7 MAR 2007
- Manuscript Received: 29 NOV 2006
- National Institutes of Health. Grant Numbers: R01-CA-097050, R01-CA-115825
- Bruggeman Medical Foundation
- enterochromaffin cell;
- growth inhibition;
- transforming growth factor-β (TGFβ);
Although it is known that small intestinal carcinoids are derived from enterochromaffin (EC) cells, these cells remain poorly characterized and little is known of the growth regulatory mechanisms of these neuroendocrine cells. Down-regulation or loss of the transforming growth factor-β1 (TGFβ1) cytostatic program and activation of TGFβ-mediated transcriptional networks is associated with uncontrolled growth and metastasis in other neural tumors, glioblastomas. Whether this phenomenon is common to small intestinal carcinoid tumors was investigated.
The effects of TGFβ1 on cultured normal EC cells (isolated by FACS sorting) and the neoplastic EC cell line, KRJ-I, was assessed using the MTT assay. The TGFβRII transcript and protein were identified in tumor cells and the effects of TGFβ1 on SMAD2 phosphorylation and nuclear translocation quantified. The time-dependent response of SMAD4, SMAD7, c-Myc, and P21WAF1/CIP1 protein expression and c-Myc and p21WAF1/CIP1 transcript was measured in response to TGFβ1 and the transcript expression of candidate downstream targets, MTA1 and E-cadherin, were assessed.
TGFβ1 inhibited normal EC cell proliferation (IC50 = 17 pM) but stimulated neoplastic EC cell proliferation (EC50 = 22 pM). In tumor cells, significantly decreased transcript (P < .01) of TGFβRII was identified, but no receptor mutations were identified and protein expression was evident. TGFβ1 (1 ng/mL) resulted in SMAD2 phosphorylation and <7% nuclear expression compared with 93% in normal EC cells. In neoplastic cells, TGFβ1 (1 ng/mL) caused a decrease in SMAD4 (>16%, P < .05), whereas SMAD7 and c-Myc transcript and protein were respectively increased >21% (P < .05) and ≈40% (P < .002). TGFβ1 (1 ng/mL) also decreased p21WAF1/CIP1 transcript by 60% (P < .001) and protein that was undetectable at 24 hours. Expression of the downstream targets of the c-Myc pathway, MTA1, was increased (20%) and E-cadherin decreased (30%).
The neoplastic EC cell is characterized by loss of TGFβ-1-mediated growth inhibition and, similar to glioblastomas, utilizes the TGFβ system to induce gene responses associated with growth promotion (c-Myc and the ERK pathway), invasion (E-cadherin), and metastasis (MTA1). Cancer 2007. © 2007 American Cancer Society.
Gastrointestinal (GI) carcinoids (enterochromaffin [EC] cell-derived) are often misdiagnosed, difficult to detect, and usually incurable when metastatic. Of note, their incidence is steadily increasing (700% in the last 3 decades1), and outcome data demonstrate no alteration in survival in this time period.1, 2 Despite these dramatic clinical observations, to our knowledge there has been little investigation of this class of neoplasia. As a consequence, biochemical markers of early disease have not been identified, the mechanisms responsible for the growth or metastasis of these tumors are for the most part unknown, and no molecular tools exist to predict malignant progression.3 In contrast to the small intestinal carcinoids (EC cell-derived), gastric carcinoids (EC-like cell-derived) have been well characterized and, therefore, early diagnosis is possible, the mechanistic basis of disease is established, and a rational therapeutic strategy thus exists with excellent outcome.3 Gastric carcinoids are derived from the ECL cells that proliferate in response to elevated levels of gastrin caused by decreased acid secretion.4, 5 In contrast, the mechanisms of neoplastic transformation of the small intestinal EC cell are unknown and uncharacterized.3
Examining the molecular alterations that characterize malignancy and metastases have identified that EC cell carcinoids exhibit a microsatellite stable phenotype with retention of mismatch repair function,6, 7 but often display chromosomal alterations, particularly loss of chromosome 18, which occurs in 50% to 67% of cases.8 This arises early in small intestinal carcinoid development and is present in both primary tumors and metastases.9 Chromosome 18 encodes SMAD4 (18q21),10 which regulates TGFβ signaling,11 a growth inhibitory factor expressed in most small intestinal carcinoids.12 Because this chromosome is often lost in small intestinal carcinoids, aberrations in the TGFβ-mediated inhibitory pathway constitute a plausible mechanism for neoplastic transformation in these lesions.
The TGFβ-mediated cytostatic pathway is considered a barrier to tumor emergence and progression and loss of this program is considered a hallmark of cancer.11, 13 Evasion of cytostasis is a strong selective advantage and tumor cells can achieve this through a number of mechanisms, including coding sequence mutations in the TGFβRII (in gliomas, gastric, and colorectal cancers14, 15), mutations in SMAD2 (in colorectal and lung cancers14), and SMAD4 (in approximately 10% of all colon cancers and in 25% to 90% of all pancreatic tumors14) or via inactivation of ELF signaling.16 In other types of cancers, TGFβ appears to play a more complex, dual role. At early stages, the TGFβ pathway functions as a tumor suppressor, inhibiting primary tumor growth, but at later stages tumor cells develop the ability to bypass these tumor suppressor functions and TGFβ appears to promote tumor progression.13 A loss of the cytostatic program, despite retention of functional TGFβ receptors and SMAD activity, has been reported in neural cell-derived tumors such as glioblastomas.17, 18 In this system the TGFβ system induces gene responses associated with growth promotion (eg, production of autocrine mitogens18), and invasion and evasion of immune surveillance and metastasis.13 The question remains whether similar mechanisms exist in cells that are classified as neuron-endocrine.
Despite alterations in chromosome 18, to our knowledge no SMAD4 mutations have been identified in small intestinal carcinoids,19 although they exhibit variable expression of TGFβRII (upstream of SMAD4),6 and variable expression of the TGFβ1 cytostatic program targets protein P21WAF1/CIP1.20, 21 In addition, they frequently express c-Myc,22 a TGFβ1 pathway antagonist. This information as well as data that TGFβ1 suppresses neuroendocrine cell proliferation23 has led us to hypothesize that neoplastic EC cells are characterized by down-regulation or loss of the TGFβ1-mediated cytostatic program. We propose that in EC tumor cells (the cell type of small intestinal carcinoids) the TGFβ pathway activates signal transduction and transcriptional programs that result in uncontrolled growth, metastasis, and expansion as noted in neural cell-derived tumors like glioblastomas.17, 18
Because mutations are present in only a small percentage of tumors refractory to TGFβ1 growth inhibition,24–26 we have focused on perturbations in downstream targets of the TGFβ pathway.11, 14 We therefore examined P21,WAF1/CIP1 c-Myc, MTA1 (ac-Myc transcriptional target), E-cadherin, and the ERK pathway. MTA1 was examined because it is a metastasis factor overexpressed in malignant primary GI carcinoids and their metastases27 and has been linked to a worse prognosis in patients with EC cell-derived appendiceal carcinoids.28 E-cadherin is of interest because reduced expression correlates with malignant carcinoid behavior,21 whereas the ERK pathway is relevant because it is active in the majority of GI carcinoids,29 and may be cross-activated either by TGFβ130 or by growth factors directly released by TGFβ1.11
It has not been possible to address these mechanistic questions previously because to our knowledge no homogenous human EC cell preparations or cell lines existed until we developed the methodology to isolate and culture normal human small intestinal EC cells.31 In addition, we have developed and characterized a rapidly growing, long-term, primary tumor-derived, human carcinoid cell line, KRJ-I.31, 32 This small intestinal EC cell-derived in vitro model is ideal to examine the TGFβ1 pathway because the alternatives such as the BON pancreatic tumor cell line,33 the COLO320DM colonic cell line,34 or the GOT1 liver metastases-derived cell line35 are either of a mixed cell origin or are derived from metastases. In this study, using preparations of homogenous human EC cells and the KRJ-I cell line, we examined the proliferative effects of TGFβ1 and then delineated aspects of the TGFβ1 signaling pathway alterations that characterize the EC cell neoplastic phenotype.
MATERIALS AND METHODS
Normal EC Cells
EC cells from the terminal ileum (from right hemicolectomy resections: n = 5) were isolated using pronase digestion, Nycodenz gradient centrifugation, and FACS sorting of acridine orange-labeled cells as described.31 FACS sorted cells (≈2 × 104 cells/100 μL) were cultured in Ham F12 medium (Gibco, Gaithersburg, Md) supplemented with 10% fetal calf serum and antibiotics (100 U penicillin/mL + 100 μg streptomycin/mL; Sigma-Aldrich, St. Louis, Mo) (2 × 104 cells/well 96-well collagen I-coated plates; Becton Dickinson, San Jose, Calif) in a humidified atmosphere at 37°C in 5% carbon dioxide for 72 hours.
Neoplastic EC Cells (KRJ-I Cell Line)
Cells were seeded in 96-well plates at 2 × 104 cells/well in Ham F12 (100 U penicillin/mL + 100 μg streptomycin/mL).31, 36 TGFβ1 (Santa Cruz Biotechnology, Santa Cruz, Calif [hBA-112, sc-4561]) was added at concentrations ranging from 1 pM to 100 pM (n = 8 wells for each concentration)38 and cells were incubated for 72 hours. MTT was added (final concentration 0.5 mg/mL per well) and incubated for 3 hours (37°C at 5% carbon dioxide), and the reaction stopped by adding acid-isopropanol, and the formazan dye solubilized. Optical density was read at 595 nanometers (nm) using a microplate reader (Bio-Rad, Hercules, Calif; 3500).36
Real-Time Polymerase Chain Reaction
RNA was extracted from 1 × 106 acutely isolated normal EC cells (n = 3) or 5 × 106 KRJ-I cells in log phase growth (n = 3) (TRIZOL, Invitrogen, La Jolla, Calif) and cleaned (Qiagen RNeasy kit and DNeasy Tissue kit, Qiagen, Valencia, Calif) to minimize contaminating genomic DNA. RNA (2 μg) was converted to cDNA (High Capacity cDNA Archive Kit; Applied Biosystems [ABI], Foster City, Calif).6, 39 Real-time reverse transcriptase-polymerase chain reaction (RT-PCR) analysis was performed using Assays-on-Demand products and the ABI 7900 Sequence Detection System according to the manufacturer's suggestions.6 Cycling was performed under standard conditions (TaqMan Universal PCR Master Mix Protocol) and the Standard Curve Method (ABI User Bulletin #2) used to determine relative transcript levels. Products included TGFβ1 (Hs00171257_m1), TGFβRI (Hs00610319_m1), TGFβRII (Hs00559661_m1), SMAD4 (Hs00232068_m1), SMAD7 (Hs00178696_m1), ELF (Hs00162271_m1), p15INK4b (Hs00365249_m1), p21WAF1/CIP1 (Hs00355782_m1), and c-myc (Hs00153408_m1). Data was normalized using geNorm40 and expression of the novel house-keeping genes ALG9, TFCP2, and ZNF410.41
MTA1 and E-Cadherin Transcript Analysis
The effect of TGFβ1 on MTA1 and E-cadherin transcript was measured in short-term cultured normal and neoplastic cells. Cells were stimulated with TGFβ1 (1 ng/mL) for 0, 30, 60, 120, and 240 minutes and RNA isolated as described above. Real-time RT-PCR analysis was performed as described above for MTA1 (Hs00183042_m1) and E-cadherin (Hs00170423_m1) and data were normalized using geNorm and expression of ALG9, TFCP2, and ZNF410.41
Genomic DNA was extracted from KRJ-I cells (2 × 106 KRJ-I cells in log phase growth: n = 3) using a Qiagen DNeasy tissue kit.6 The BAT-RII repeat sequence and (GT)3 microsatellite region of TGFβRII were analyzed as previously described.6 For SMAD mutations, SMAD2 (exons 4, 8, 9, 10, and 11) and SMAD4 (exons 4, 10, 11, 12, and 13) were examined as described.42 Purified PCR products were sequenced and analyzed by the W.M. Keck Biotechnology Resource Laboratory at Yale University using an automated Applied Biosystems 373A Stretch DNA sequencer (Perkin-Elmer, Norwalk, CT). PCR products were initially sequenced using the forward primer. If ambiguous peaks were present, the sequence was confirmed with the reverse primer.6
Western Blot Analysis
KRJ-I cells (2 × 106 KRJ-I cells in log phase growth: n = 3) were prepared as previously described.43 Cells were snap-frozen in liquid nitrogen and homogenized using a mortar and pestle before the addition of RIPA buffer. Protein concentration was assayed using the Bio-Rad protein assay (Bio-Rad). Samples were boiled in Laemmli reducing buffer and equal protein resolved on 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. Protein was immobilized on nitrocellulose membranes by electrotransfer. After blocking (5% nonfat milk for 60 minutes at room temperature [RT]), gels were immunoblotted with rabbit polyclonal antibodies TGFβRII (#sc-220, 1:500; Santa Cruz Biotechnologies), pSMAD2 (#3101, 1:1000; Cell Signaling Technology, Beverly, Mass), SMAD4 (IMG-565, 1:500; IMGENEX, San Diego, Calif), SMAD7 (IMG-531, 1:1500; IMGENEX), and β-actin (sc-1616R. 1:4000; Santa Cruz Biotechnology) or mouse anti-P21WAF1/CIP1 (0.5 mg/mL; Becton Dickinson) overnight at 4°C and then horseradish peroxidase-conjugated goat secondary antibodies (R&D Systems, Minneapolis, Minn; 1:4000 for 30 minutes at RT). Membrane-bound antibodies were detected using a luminol-based chemiluminescence system (Roche, Nutley, NJ). Blots were exposed on X-OMAT-AR film. Gels were stripped and reprobed with anti-β-actin to confirm equivalent protein loading. Densitometry analysis of bands was performed by Scion Image (Scion, Frederick, Md).
Normal EC cells or KRJ-I cells (5 × 104 cells) were fixed in methanol and pipetted onto frosted microscope slides stained with a series of antibodies including anti-TGFβRII (1:1000, sc-220), neuroendocrine-specific antibodies (mouse chromogranin A [1:1000, DakoCytomation, Dako, Carpinteria, Calif] or mouse VMAT1 [1:100; Chemicon, Temecula, Calif]), EC cell-specific antibodies (mouse TPH [1 μg/mL, Calbiochem, La Jolla, Calif], mouse serotonin [Dako, 1:20], or rabbit substance P [1:200, GeneTex, San Antonio, Tex], and DAPI (4′,6′diamidino-2-phenylindole) (20 mg/mL) overnight at 4°C.31, 36 After washing (0.1% Tween/PBS), cells were stained with secondary antibody (FITC-antimouse/rabbit, 1:100; Promega, Madison, Wis) (1 hour at RT). Fluorescence microscopy was used to visualize and count the number of positive cells, with 1 × 103 cells counted for each antibody. As a control for nonspecific staining, primary antibodies were excluded in matched groups.
ERK1/2 Pathway Analysis
The effect of TGFβ1 on the ERK1/2 signaling pathway was measured in neoplastic cells as described.44 Cells were stimulated with TGFβ1 (1 ng/mL) and cultured overnight (24 hours). To assess the specificity of ERK pathway activation, cells were also preincubated with PD98059 (30 μM) (an inhibitor of ERK1/2 phosphorylation) for 30 minutes before the addition of TGFβ1. ERK1/2 phosphorylation was measured using an enzyme-linked immunoadsorbent assay (ELISA) approach (SuperArray CASE, Frederick, Md) as per the manufacturer's protocol. Briefly, stimulated cells were fixed (4% formaldehyde), and stained with either primary antibodies against the nonphosphorylated or phosphorylated forms of each protein (60 minutes at RT). After washing and secondary antibody application (60 minutes at RT), cells were incubated with color developer (10 minutes at RT) and plates read at 450 nm. Thereafter, protein was assayed in each well (protein development: reading at 595 nm). The results were calculated as antibody (450 nm) per protein concentration (595 nm) and normalized to unstimulated cells. Phosphorylated signal was compared with total nonphosphorylated signal.
Proliferation Studies: Effect of TGFβ1 on Normal EC Cells and Neoplastic (KRJ-I) Cells
In short-term cultured (72 hours) normal human EC cells, TGFβ1 significantly inhibited MTT uptake with a maximal effect of 28 ± 2.4% at 100 pM (Fig. 1A) and an estimated IC50 = 17.4 pM. In contrast, TGFβ1 stimulated MTT uptake in the neoplastic EC cell line over this time period (Fig. 1B). The maximal effect was noted at 100 pM (31 ± 3.7%); EC50 = 20.18 pM. This indicates that the TGFβ1-mediated cytostatic program is lost in neoplastic EC cells.
Transcript Levels of TGFβ1 Signaling Components in Normal and Neoplastic EC Cells
Using a real-time quantitative PCR approach and normalization of expression levels using geNorm and 3 housekeeping genes (ALG9, TFCP2 and ZNF410),41 we evaluated the expression of genes involved in the TGFβ1 signaling network in normal (n = 3) and neoplastic (n = 3) EC cells. Significant differences were apparent (Table 1). Normal EC cells expressed both TGFβI and II receptors, ELF, p15INK4b, and p21WAF1/CIP1, indicating that components of the TGFβ1 signaling network associated with the cytostatic response are present. The tumor cell line, however, had significantly lower (2–100-fold) transcript levels of TGFβRI and RII, ELF, and the cytostatic genes, p15INK4b and p21WAF1/CIP1 (P < .01) and high expression (>1000-fold) of SMAD4 and SMAD7 (P < .01). These data indicate that a number of components of the TGFβ1 signaling network are substantially altered in neoplastic EC cells.
|Normal EC||0.0058 ± 0.003||0006*± 0.002||0.071*± 0.041||0.0004 ± 0.0001||0.006 ± 0.003||8.42*± 3.44||0.246*± 0.122||0.008*± 0.004||0.006 ± 0.002|
|KRJ-I||0.597*± 0.055||0.003 ± 0.001||0.00005 ± 0.000003||0.494*± 0.018||0.698*± 0.08||0.66 ± 0.047||0.00009 ± 0.000003||0.002 ± 0.001||0.004 ± 0.002|
Examination of TGFβRII Expression in KRJ-I Cells
TGFβRII protein was identified in KRJ-I cells by Western blot analysis (Fig. 2A) and immunohistochemistry (Fig. 2B). Western blot analysis identified the mature form of the protein (band of ≈87 kDa) and the precursor (nonglycosylated) forms.45 Immunohistochemistry identified membrane staining for TGFβRII and confirmed the transcript expression of this receptor.
We examined 2 repeat regions in the TGFβRII gene in genomic DNA from KRJ-I to assess whether TGFβRII may encode activating mutations, as has been demonstrated in loss of responsiveness to TGFβ in gliomas, gastric, and colorectal cancers.14, 15 No frameshift mutations in the poly(A)10 (BAT-II microsatellite region) or (GT)3 regions were detected.
Examination of TGFβ-Mediated SMAD2 Phosphorylation and Translocation
To examine whether the TGFβ signaling pathway was intact, we measured TGFβ1-mediated SMAD2 phosphorylation and relocation to the nucleus. Protein levels of pSMAD2 were measured by Western blot analysis in KRJ-I cells in response to TGFβ1 (1 ng/mL). PSMAD2 was detectable after 1-hour stimulation (Fig. 3A). This suggests that TGFβRII in the neoplastic cell was functional and that TGFβ1 stimulated SMAD phosphorylation. Immunohistochemical examination of phosphorylated SMAD2, 1 hour after TGFβ1 (1 ng/mL), demonstrated that nuclear colocalization occurred in only 6.3 ± 0.7% of tumor cells, whereas normal EC cells studied under the same experimental conditions exhibited nuclear colocalization in 92.7 ± 3.6% (P < .01) (Fig. 3B). In neoplastic EC cells, pSMAD2 was almost exclusively sequestrated in the cytoplasm.
These data indicate that TGFβ1-mediated signal transduction is blocked at the level of SMAD nuclear translocation in neoplastic EC cells. This is consistent with the decreased transcript expression of ELF and increased SMAD7 expression noted in KRJ-I cells (Table 1) because ELF is critical for SMAD2 translocation,46 whereas SMAD7 is an antagonist of this process.11
To assess whether pSMAD mutations resulted in the inhibition of nuclear targeting of this transcription factor, we examined mutational hotspots in SMAD2 (exons 4, 8, 9, 10, and 11)42 and SMAD4 (exons 4, 10, 11, 12, and 13),42 but no mutations in either of these genes was identified in DNA from KRJ-I cells. We interpreted the absence of inactivating mutations in these TGFβ1 transcriptional activators as evidence that cytoplasmic SMAD2 sequestration was not due to a mutational event. Alternatively, low endogenous expression of SMAD2 in KRJ-I cells could be a factor in the decreased nuclear translocation, but this is unlikely as staining levels were similar in normal and neoplastic cells (Fig. 3B).
Examination of TGFβ-Mediated SMAD4 and SMAD7 in KRJ-I
To examine the effect of TGFβ1 on SMAD4 and SMAD7 expression, we measured protein levels (Western blot analysis) of SMAD4 and SMAD7 at 0, 1, 4, and 8 hours in KRJ-I cells in response to TGFβ1 (1 ng/mL). SMAD4 expression was not changed by TGFβ1 at 1 hour but was decreased at 4–8 hours (Fig. 4). In contrast, SMAD7 was significantly elevated at 1, 4, and 8 hours (Fig. 4). Densitometric quantitation analysis of bands (Scion Image software) confirmed that SMAD4 was decreased 16% to 31% (P < .05 vs 0 hour: no stimulation by TGFβ1) and demonstrated that SMAD7 was elevated 18% to 24% (P < .05 vs 0 hour). These results indicate that TGFβ1 decreased expression of SMAD4—an obligatory factor for SMAD complex nuclear targeting, whereas SMAD7—an inhibitor of SMAD nuclear targeting, was increased. This provides a potential mechanism for blockade of the TGFβ1-cytostatic program; pSMAD2 nuclear targeting would be predicted to be decreased after these SMAD alterations.
Effect of TGFβ1 on P21WAF1/CIP1 and c-Myc Expression
To confirm evidence of alterations in TGFβ signaling, we next assessed transcript and protein levels of p21WAF1/CIP1 using real-time PCR at 0, 30, 60, 120, and 240 minutes and Western blot analysis at 0 1, 4, 8, and 24 hours. Levels of p21WAF1/CIP1 were decreased by TGFβ1 (1 ng/mL) from 60 minutes and did not return to prestimulated levels (Fig. 5A). Protein levels, in contrast, did not change between 0 and 8 hours, but by 24 hours P21WAF1/CIP1 was no longer detectable.
Evidence That TGFβ1 Activates Non-SMAD Pathways in KRJ-I Cells
Phosphorylation and activation of the ERK1/2 pathway results in inhibition of TGFβ1 growth inhibitory signals by preventing nuclear accumulation of SMADs.50 We measured activation of ERK1/2 in KRJ-I cells after incubation with TGFβ1 (1 ng/mL) for 0–240 minutes. TGFβ1 resulted in a significant initial increase in phosphorylated ERK1/2 at 30 minutes that was sustained for 2 hours and thereafter declined by 4 hours (Fig. 6). Preincubation with PD98059 (30 μM) inhibited TGFβ-activated ERK1/2 phosphorylation, which demonstrates that TGFβ1 activates the MAPK growth regulatory cascade via phosphorylation of ERK1/2 in neoplastic cells. This may also contribute to the decreased nuclear pSMAD2 translocation noted in KRJ-I cells incubated with TGFβ1.
TGFβ1-Mediated Increases in Expression of Metastasis-Associated Genes
To examine whether TGFβ1 signaling was associated with transcriptional regulation of malignancy-associated genes in the KRJ-I cell line, we examined the effects of TGFβ1 on MTA1 and E-cadherin expression and compared this to normal EC cells. TGFβ1 stimulated expression of MTA1 transcript (Fig. 7A) and decreased E-cadherin expression (Fig. 7B) in neoplastic cells but had no significant effect on gene expression in normal cells. Stimulating neoplastic cells with TGFβ1 therefore alters the expression of 2 well-characterized malignancy-associated factors, an observation consistent with an additional role (beyond proliferation) of TGFβ1 signaling in neoplastic EC cells.
The investigation of TGFβ pathway alterations in tumor cells has yielded significant mechanistic information regarding the regulation of different kinds of neoplasia and provided mechanistic explanations as to the basis of the evolution of a malignant phenotype.11, 13, 30 In the current study, we examined the effect of TGFβ1 on a neoplastic small intestinal EC cell-derived cell line, KRJ-I, and demonstrated that in neoplastic EC cells TGFβ1: 1) increased cell proliferation as opposed to inhibiting it; 2) decreased expression of SMAD4 but increased expression of the inhibitor of SMAD nuclear translocation, SMAD7; 3) down-regulated p21WAF1/CIP1 transcript, a regulator of the cytostatic program; 4) increased expression of c-Myc, an oncogene involved in evasion of TGFβ1-mediated growth inhibition and a transcription factor that up-regulates malignancy-associated genes; 5) resulted in phosphorylation and activation of ERK1/2, a growth stimulatory pathway that also is involved in evasion of TGFβ1-mediated growth inhibition via negative regulation of pSMAD nuclear targeting; and 6) caused downstream activation of the malignancy-defining genes MTA1 and E-cadherin.
A cytostatic role for TGFβ1 on normal endocrine or neuroendocrine cell proliferation (eg, lactotrope, pituitary, parathyroid, and chromaffin cell) has been identified.51–53 In contrast, tumors from these organs demonstrate a loss of cytostasis and respond to TGFβ1 with proliferation.52 Our examination of normal neuroendocrine (EC) cells from the small intestine and the small intestinal EC cell line, KRJ-I, demonstrate a similar phenomenon: normal cells can be growth-inhibited by TGFβ1, whereas neoplastic neuroendocrine cells lose TGFβ1-mediated cytostasis. Further examination of the TGFβ1 pathway in the neoplastic cell line demonstrated low expression levels (mRNA and protein) of nonmutated TGFβRII but SMAD2 could be phosphorylated in response to TGFβ1-mediated stimulation. These results indicate an intact TGFβ1:TGFβRII interaction. However, in contrast to normal EC cells, which demonstrated nuclear targeting of pSMAD2 after TGFβ1 stimulation, nuclear pSMAD2 positivity was barely detectable in KRJ-I cells (<7%). These data suggest that TGFβ1-mediated signal transduction in a small intestinal carcinoid cell line, as in glioma cell lines,54 is blocked at the level of SMAD nuclear translocation.
To investigate this, we examined transcript expression of a number of candidate factors in this pathway. Both SMAD4 and SMAD7 transcript levels were elevated in KRJ-I compared with normal EC cells, whereas ELF was decreased. These differences provide a potential mechanism for a decrease in nuclear pSMAD2 targeting because ELF is critical for SMAD2 translocation,46 whereas SMAD7 is an antagonist of this process.11
To examine this in more detail, we then determined the effect of TGFβ1 on SMAD4 and SMAD7 protein expression. These quantitative studies demonstrated that TGFβ1 decreased SMAD4 expression, whereas it increased SMAD7 levels. TGFβ1-mediated alterations in SMAD expression have been noted in the human lymphoma cell line HT58, where SMAD7 expression was stimulated by TGFβ1.55 In addition, other cell lines (head and neck squamous cell carcinoma) that are resistant to TGFβ1-mediated cytostatic also exhibit high expression of SMAD7.56 An increased expression of this inhibitory SMAD may be important because SMAD7 blocks TGFβ-induced growth inhibition by inhibiting TGFβ-mediated down-regulation of c-Myc, CDK4, and Cyclin D1, and suppresses the expression of p21WAF1/CIP1.57 SMAD7 up-regulation has also been associated with increased tumorigenicity of human colonic carcinoma FET cells.57
To examine the importance of alterations in SMADs, we analyzed transcription of P21WAF1/CIP1 and c-Myc in the KRJ-I cell line. Our studies demonstrated that loss of TGFβ1 responsiveness in KRJ-I was associated with decreased p21WAF1/CIP1 and increased c-Myc transcription. P21WAF1/CIP1 expression levels are increased in cell lines treated with TGFβ1 that have intact TGFβ1-mediated cytostatic pathways,58, 59 and may be unaltered in glioma cell lines that have lost this responsiveness.54 In the current study, p21WAF1/CIP1 transcription and protein was decreased. KRJ-I cells express time-dependent increases in both transcript and protein expression of c-Myc in response to TGFβ1 treatment. A similar inability to down-regulate c-Myc has been demonstrated in esophageal cells line that have lost responsiveness to TGFβ1.60
Elevations in c-Myc may be responsible for a loss of p21WAF1/CIP1 transcriptional response because induction of c-Myc is associated with blockade of TGFβ1-mediated p21WAF1/CIP1 transcription and down-regulation of basal P21WAF1/CIP1 levels.61 In addition, repression of P21WAF1/CIP1 has been demonstrated to play a role in c-Myc-dependent cell cycle progression.61 Based on our studies and this information, we postulate that neoplastic EC cells use the TGFβ1 transcriptional pathway to up-regulate c-Myc and inhibit P21WAF1/CIP1 with a resultant increase in cell proliferation and concomitant activation of c-Myc transcriptional targets.
MTA1 is an important component in c-Myc activity as 1 of its first downstream targets and appears essential for the transformation potential of this oncogene.62 In the current studies, MTA1 transcription was up-regulated by TGFβ1 in a time frame consistent with c-Myc activation. We postulate that this metastasis factor, recently shown to be overexpressed in malignant primary GI carcinoids and their metastases27, 63 and linked to a worse prognosis in EC cell-derived appendiceal tumors,28 is up-regulated by TGFβ1 or the TGFβ1-pathway in neoplastic EC cells and plays a role in malignant transformation and metastasis.
E-cadherin expression was reduced by TGFβ1 in the KRJ-I cell line. TGFβ1 treatment has been shown to decrease E-cadherin levels and cause redistribution of β-catenin in intestinal epithelial cells,64 suggesting that TGFβ1 functions to regulate cell migration and metastasis. Reduced expression of E-cadherin correlates with malignant behavior of GI carcinoids21 and it appears that TGFβ1 pathway alteration of the E-cadherin expression represents a mechanistic component of the malignant evolution.
This study demonstrates that the classical TGFβ1-mediated cytostatic pathway is not functional in the KRJ-I cell line. Our data indicates that neoplastic EC cells utilize a relatively intact TGFβ1 transcriptional network to drive cell proliferation via c-Myc activation. An alternative or additional mechanism is provided by the ability of TGFβ1 to cross-associate/activate another pathway, the ERK pathway.11 Our experiments confirm this event and, because phosphorylation of ERK1/2 results in activation of the MAPK growth regulatory cascade, it is plausible that the growth stimulatory effects of TGFβ1 noted in neoplastic EC cells reflects such cross-activity. It is notable that a recent report suggests that an activated ERK pathway is a feature of many GI carcinoids.29
In summary, the results of the current study demonstrate that alterations in expression and function of SMADs in the KRJ-I cell line result in a decreased expression of p21WAF1/CIP1 and up-regulation of c-Myc and altered regulation of carcinoid malignancy-associated factors. These data support the hypothesis that an escape from TGFβ1-mediated growth inhibition occurs in small intestinal carcinoids. The molecular delineation of TGFβ1-mediated proliferative regulation of this small intestinal carcinoid cell line provides an opportunity to identify alterations in TGFβ1 signaling. This information may be of some use to predict prognosis in small intestinal carcinoids, as has been described with regard to expression levels of SMAD4 in head and neck squamous cell, breast, and hepatocellular carcinomas.65–67 More important, it appears likely that further examination of the ubiquitous TGFβ pathway in neuroendocrine tumors may provide alternative targets to regulate proliferation.
Supported in part by Grants R01-CA-097050 and R01-CA-115825 from the National Institutes of Health and the Bruggeman Medical Foundation.
- 37Subculturing Human Diploid Fibroblast Cultures. New York: Academic Press; 1973..
- 41GeneChip, geNorm and gastrointestinal neoplasia: novel reference genes for real-time PCR. Physiol Genomics. In press., , , et al.