Cancer Research
Letter to the Editor
GLI1 Modulates EMT in Pancreatic Cancer—Response Simon Joost1, Luciana L. Almada2, Verena Rohnalter1, Philipp S. Holz1, Maite G. Fernandez-Barrena2, Martin E. Fernandez-Zapico2, and Matthias Lauth1
In a recent publication (1), we showed that low levels of the GLI1 transcription factor sensitize pancreatic ductal adenocarcinoma (PDAC) cells to induce an epithelial-to-mesenchymal transition (EMT). Analysis of the mechanism identified Ecadherin (encoded by the CDH1 gene) as a direct target of GLI1 in PDAC cells. We showed direct binding of GLI1 to the CDH1 promoter and transcriptional activation of the CDH1 gene by GLI1 in pancreatic cancer cells using expression and reporter studies. Finally, we were able to show a positive correlation between GLI1 and E-cadherin expression in cultured PDAC cells and in patient samples. Our data were questioned in a letter to the editor from Inaguma and colleagues who recently published that GLI1 induces mucin MUC5AC expression in PDAC cells leading to Ecadherin protein destabilization (2). In their letter, the authors raise concerns about several points of our manuscript, and we would like to reply to the major points of criticism in this response letter. One of these points refers to the fact that we did not show changes in E-cadherin levels upon GLI1 overexpression in several PDAC cell lines, a somewhat astonishing criticism given the fact that the authors themselves never properly showed alterations in E-cadherin expression upon full-length GLI1 overexpression in their own manuscript (2). However, we assume that the transcriptional machinery that controls the GLI1–CDH1 axis might be saturated in these cells and therefore no change in CDH1 levels upon further increase of GLI1 by means of transfection can be seen. One exception was MiaPaca-2 cells, which upregulated CDH1 expression upon GLI1 transfection, despite the reported hypermethylation of the CDH1 promoter in these cells. In agreement with our data, others have also reported induction of CDH1 in MiaPaca-2 cells upon transfection with transcription factor expression plasmids, suggesting that the promoter methylation does not fully prevent transcriptional activity at this gene locus under certain circumstances (3). Also not discussed in the letter by Inaguma and colleagues is our finding that a transcriptionally inactive
Authors' Affiliations: 1Institute of Molecular Biology and Tumor Research (IMT), Philipps University, Marburg, Germany; and 2Schulze Center for Novel Therapeutics, Division of Oncology Research. Mayo Clinic, Rochester, Minnesota Note: Current address for S. Joost: Karolinska Institutet, Center for Biosciences, Department of Biosciences and Nutrition, 141 83 Huddinge, Sweden. Corresponding Author: Matthias Lauth, Institute of Molecular Biology and Tumor Research (IMT), Philipps University, Emil-Mannkopff-Str. 2, Marburg 35032, Germany. Phone: 49-6421-2866727; Fax: 49-6421-2865932; E-mail:
[email protected] doi: 10.1158/0008-5472.CAN-12-1468 Ó2012 American Association for Cancer Research.
3704
Cancer Res; 72(14) July 15, 2012
GLI1 lacking the DNA-binding domain is able to suppress E-cadherin expression and induce an EMT in Panc1 cells. We interpret these data as dominant-negative behavior of the mutant GLI1 on the wild-type form. Taken together, we concluded that in most of the cell lines analyzed, the effects of GLI1 on CDH1 are saturated, and only a reduction in GLI1 is translated into effects on CDH1. A slightly alternative possible scenario would be that GLI1 is required to maintain CDH1 expression once the promoter has been activated by a second factor. In such a case, the second factor would define the level of CDH1 expression and a reduction in the maintenance factor (GLI1) would decrease the CDH1 expression but increasing GLI1 levels would not further stimulate it. Another point of concern that Inaguma and colleagues pointed out was that our clinical samples were not microdissected and therefore the positive correlation between GLI1 and CDH1 might have been caused by contaminating noncancerous cells. While it is correct that our samples contained surrounding stroma, the epithelial cancer cells are most likely the major source for CDH1 within the tumor. If, as the authors of the letter imply, GLI1 expression was linked to CDH1 repression, one would not expect a positive GLI1–CDH1 correlation. On the contrary, in this case, one would rather observe a negative GLI1–CDH1 correlation. In light of the finding that the CDH1-negative fibroblastic stroma is the most dominant contributor of GLI1 expression (4) and is highly abundant in PDAC, turning a negative correlation into a positive one would require the systematic contamination with stromal cells, an unlikely scenario. Additional support for a positive GLI1– CDH1 correlation comes from our in vitro analysis of 15 (stroma-free) PDAC cell lines that show a highly significant positive correlation (1) and that was unfortunately ignored by Inaguma and colleagues in their letter. We would furthermore like to point out that our report on a positive GLI1–CDH1 axis is not unprecedented: GLI expression has been associated with increased epithelial morphology and with reduced migratory potential in keratinocytes (5, 6). Moreover, ectopic GLI1 expression in ovarian cancer cells increases E-cadherin protein levels (7), and Hedgehog/GLI signaling promotes epithelial differentiation and E-cadherin expression in gastric cells (8). Inaguma and colleagues concluded their letter with experiments overexpressing suppressor of fused (SUFU, a negative regulator of GLI activity) where they observe an inhibition of MUC5AC expression and a slight increase in Ecadherin protein levels. This is surprising as the same authors previously showed that SUFU is present but is functionally inactivated in PDAC cells, at least with respect to inhibition of GLI (9). Hence, one possible mechanism of the effects on MUC5AC could potentially stem from GLIindependent functions of SUFU.
GLI1 Modulates EMT in Pancreatic Cancer
How then can the different results (particularly in the siRNA transfections) be explained? Not considering aspects such as cell line drift, culture media composition, or different siRNA sequences, we could envision cell density as a major criterion of the experimental outcome. Inaguma and colleagues observe their MUC5AC-dependent effects only in confluent cultures, whereas our GLI1–CDH1 effects were most prominent under subconfluent conditions. Hence, it might even be conceivable that both mechanisms are separately operational under different culture conditions.
Taken together, we feel confident about the interpretation of our own results but are nevertheless satisfied to conclude that there seems to be at least one point on which we can agree with Inaguma and colleagues. Further research is required to fully address the open issues discussed here. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Received April 25, 2012; accepted April 29, 2012; published OnlineFirst July 3, 2012.
References 1. Joost S, Almada LL, Rohnalter V, Holz PS, Vrabel AM, FernandezBarrena MG, et al. GLI1 inhibition promotes epithelial-to-mesenchymal transition in pancreatic cancer cells. Cancer Res 2012;72:88–99. 2. Inaguma S, Kasai K, Ikeda H. GLI1 facilitates the migration and invasion of pancreatic cancer cells through MUC5AC-mediated attenuation of E-cadherin. Oncogene 2011;30:714–23. 3. Song Y, Washington MK, Crawford HC. Loss of FOXA1/2 is essential for the epithelial-to-mesenchymal transition in pancreatic cancer. Cancer Res 2010;70:2115–25. 4. Tian H, Callahan CA, DuPree KJ, Darbonne WC, Ahn CP, Scales SJ, et al. Hedgehog signaling is restricted to the stromal compartment during pancreatic carcinogenesis. Proc Natl Acad Sci U S A 2009; 106:4254–9. 5. Snijders AM, Huey B, Connelly ST, Roy R, Jordan RC, Schmidt BL, et al. Stromal control of oncogenic traits expressed in response to the overexpression of GLI2, a pleiotropic oncogene. Oncogene 2009;28: 625–37.
www.aacrjournals.org
6. Neill GW, Harrison WJ, Ikram MS, Williams TD, Bianchi LS, Nadendla SK, et al. GLI1 repression of ERK activity correlates with colony formation and impaired migration in human epidermal keratinocytes. Carcinogenesis 2008;29:738–46. 7. Liao X, Siu MK, Au CW, Wong ES, Chan HY, Ip PP, et al. Aberrant activation of hedgehog signaling pathway in ovarian cancers: effect on prognosis, cell invasion and differentiation. Carcinogenesis 2009;30: 131–40. 8. Xiao C, Ogle SA, Schumacher MA, Schilling N, Tokhunts RA, OrrAsman MA, et al. Hedgehog signaling regulates E-cadherin expression for the maintenance of the actin cytoskeleton and tight junctions. Am J Physiol Gastrointest Liver Physiol 2010;299: G1252–65. 9. Kasai K, Inaguma S, Yoneyama A, Yoshikawa K, Ikeda H. SCL/TAL1 interrupting locus derepresses GLI1 from the negative control of suppressor-of-fused in pancreatic cancer cell. Cancer Res 2008;68: 7723–9.
Cancer Res; 72(14) July 15, 2012
3705
