|Home | About | Journals | Submit | Contact Us | Français|
Esophageal squamous cell cancer accounts for more than 90% of cases of esophageal cancers. Its pathogenesis involves chronic epithelial irritation, although the factors involved in the inflammatory process and the mechanisms of carcinogenesis are unknown. We sought to develop a mouse model of this cancer.
We used the ED-L2 promoter of Epstein-Barr virus to overexpress the transcriptional regulator Krüppel-like factor 4 (Klf4) in esophageal epithelia of mice; we used mouse primary esophageal keratinocytes to examine the mechanisms by which KLF4 induces cytokine production.
KLF4 was an epithelial-specific mediator of inflammation; we developed a new mouse model of esophageal squamous dysplasia and inflammation-mediated squamous cell cancer. KLF4 activated a number of proinflammatory cytokines, including TNF-α, CXCL5, G-CSF and IL-1α, within keratinocytes in an NF-κB– dependent manner. KLF4 was not detected in proliferating or cancer cells, indicating a non-cell autonomous effect of KLF4 on proliferation and carcinogenesis.
KLF4 has distinct functions in carcinogenesis; upregulation of Klf4 specifically in esophageal epithelial cells induces inflammation. This mouse model might be used to determine the molecular mechanisms of esophageal squamous cell cancer and inflammation-mediated carcinogenesis.
Esophageal cancer is the sixth leading cause of cancer death in the world, and worldwide more than 90% of esophageal cancers are squamous cell cancers.1-3 Ingestion of alcohol or use of tobacco, chronic irritants responsible for up to 90% of esophageal squamous cell cancers, as well as other dietary risk factors, set off a process that generally plays out over decades.3 Because most patients do not show symptoms of disease before the development of frank cancer, the early phenotypic changes and molecular events preceding the development of cancer are not well known.3,4 Alterations in a number of genes have been linked to human esophageal squamous cell cancer, but these genetic alterations were identified by examining existing esophageal squamous cell cancers.3 Thus, little direct evidence has linked genetic alterations functionally to the development of esophageal squamous cell cancer. Compounding this, esophageal squamous cell cancer has been extraordinarily difficult to model genetically in vivo.3
To further our understanding of the development of esophageal squamous cell cancer, we chose to examine the role of the transcription factor Krüppel-like factor 4 (KLF4) in esophageal homeostasis and carcinogenesis. KLF4 is an important regulator of epithelial proliferation and differentiation in numerous epithelia and activates a group of epithelial-specific keratin genes including the keratinocyte differentiation marker keratin 4 in vitro.5-7 Consistent with this, KLF4 promotes induction of the epidermal barrier in mice.8 Recently, we functionally linked Klf4 loss in esophageal epithelial cells to hyperplasia and squamous cell dysplasia in vivo.9 Yet the role of KLF4 in carcinogenesis is complex.10 KLF4 is down-regulated in some cancers, including stomach, colon, and esophagus, but expression is increased in others, such as breast, skin, and oropharyngeal.5,11 Although KLF4 inhibits survival of esophageal squamous cancer cells in vitro, expression of human KLF4 in mouse skin blocks the proliferation– differentiation switch and results in squamous cell dysplasia.12-14 KLF4 inhibits differentiation of embryonic stem cells and, in combination with other factors, de-differentiates adult cells into cells with stem-like properties,15,16 leading some to suggest that KLF4 may contribute to the self-renewal of cancer stem cells.17
Here, we have used the regulatory elements of the ED-L2 promoter to target increased Klf4 expression to esophageal epithelial cells. ED-L2/Klf4 mice develop marked esophageal epithelial inflammation and squamous dysplasia by 6 months of age and esophageal squamous cell cancers by 2 years of age. The keys to carcinogenesis in ED-L2/Klf4 mice appear to lie not in the direct effects of KLF4 on cell proliferation and differentiation pathways but through novel effects of KLF4 on cytokine activation within epithelial cells. As such, ED-L2/Klf4 mice provide a durable model for inflammation-mediated esophageal squamous cell cancer.
All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania. To express Klf4 in esophageal epithelia, we subcloned the complete coding sequence of murine Klf4 into the pL2 plasmid (a gift from Dr Anil Rustgi, University of Pennsylvania), containing 782 base pairs of the Epstein–Barr virus ED-L2 promoter.18 ED-L2/Klf4 mice were back-crossed to C57BL/6 mice (Charles River, Wilmington, MA) for 10 generations and then used for experiments. Additional details are provided in the Supplementary Materials and Methods section.
Mouse primary esophageal keratinocytes were isolated as described.19 Cells were cultured and infected or transfected, and reporter assays or quantitation were performed as described in the Supplementary Materials and Methods section.
Ultrathin (~80-nm) sections of the trans-sectional plane of each esophagus were examined with an FEI Tecnai-T12 transmission electron microscope (FEI, Hillsboro, OR) operated at 80 kv. Additional details are provided in the Supplementary Materials and Methods section.
Viability of primary esophageal keratinocytes was monitored by Trypan blue exclusion. Results were expressed as a percentage of viable cells ± standard error of the mean. Additional details are provided in the Supplementary Materials and Methods section.
RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA) following the manufacturer’s instructions. Ribonuclease protection assays were performed as described previously using 1 μg of total RNA per sample.20 Quantitative real-time polymerase chain reaction (PCR) analyses were performed as described.9 Additional details are provided in the Supplementary Materials and Methods section.
Cytokines released by mouse primary esophageal epithelial cells were determined using the Panomics Cytokine Mouse Antibody Array (Affymetrix, Fremont, CA). Additional details are provided in the Supplementary Materials and Methods section.
Results were expressed as mean ± standard error of the mean, with statistical significance of differences between experimental conditions established at 95%. Two-way analysis of variance and Student t test were used to indicate the statistical difference between the groups using GraphPad Prism version 4.0 (GraphPad Software, San Diego, CA).
To evaluate KLF4 function in esophageal epithelial homeostasis and carcinogenesis, we targeted Klf4 to esophageal epithelia using the Epstein–Barr virus ED-L2 promoter, which drives epithelial-specific expression in the tongue, esophagus, and squamous forestomach.18 Transgene expression was confirmed in the esophagus of 3 ED-L2/Klf4 transgenic lines by ribonuclease protection assays and immunohistochemistry (Supplementary Figure 1A–C). ED-L2/Klf4 mice were born at the appropriate Mendelian ratio and appeared grossly normal. Compared with littermate controls (Figure 1A, Supplementary Figure 2A), 3-month-old ED-L2/Klf4 mice (Figure 1B, Supplementary Figure 2B) had normal-appearing esophageal epithelia. However, by 6 months of age, although littermate controls maintained normal esophageal epithelia (Figure 1C, Supplementary Figure 2C), ED-L2/Klf4 mice developed hyperplasia, dysplasia, and a prominent inflammatory infiltrate in the epithelia and lamina propria (Figure 1D, Supplementary Figure 2D). Hyperplasia and inflammation also were observed in tongue and forestomach of ED-L2/Klf4 mice at 6 months of age (Supplementary Figure 3). By 2 years of age, 6 of 6 ED-L2/Klf4 mice had esophageal dysplasia and 3 of 6 ED-L2/Klf4 mice developed invasive esophageal squamous cell carcinoma (Figure 1F) compared with 0 of 6 controls (Figure 1E). There were no unexpected deaths in the control group, and no cancers were seen in other tissues of control or ED-L2/Klf4 mice.
We next examined whether proliferation and differentiation were altered in ED-L2/Klf4 mice. Although there was no significant difference in proliferation within the esophageal epithelia of control and ED-L2/Klf4 mice at 3 months of age, 6-month-old ED-L2/Klf4 mice showed a 3.8-fold increase in the number of proliferating cells compared with controls (Supplementary Figure 4). We hypothesized that KLF4 expression in esophageal progenitor cells might be responsible for this increased proliferation. However, KLF4 did not colocalize with the proliferation markers Ki-67 (data not shown) or 5-bromo-2-deoxyuridine (BrdU) in esophageal epithelia of control (Figure 2A) or ED-L2/Klf4 mice (Figure 2B); nuclear KLF4 expression was confined to nonproliferating suprabasal cells whereas BrdU staining was restricted to the proliferative zone. These findings suggested a non–cell autonomous effect of KLF4 on proliferation.
To examine differentiation, we performed immunofluorescence for keratin 14, a basal cell marker, keratin 4, which is found in the differentiated layers of esophageal epithelia, and involucrin, which marks terminally differentiated keratinocytes.21 Although the expression domains of keratin 14 and keratin 4 were unchanged in ED-L2/Klf4 mice at 3 months of age (data not shown), compared with control mice (Figure 2C), the area of keratin 14 staining was expanded in 6-month-old ED-L2/Klf4 mice (Figure 2D). The expression domain of keratin 4 was also altered in ED-L2/Klf4 mice (Figure 2F) compared to controls (Figure 2E). However, the reciprocal staining patterns of keratin 4 and keratin 14 seen in esophageal epithelia of ED-L2/Klf4 mice indicated that keratinocytes switched appropriately from proliferative to differentiating. Consistent with this, involucrin staining was unchanged between control and ED-L2/Klf4 mice (data not shown). Given our recent identification of Klf5 as a transcriptional target for KLF4,22 we also examined the expression of KLF4 and KLF5 in 3- and 6-month-old ED-L2/Klf4 mice (Supplementary Figure 1D and E). Although KLF4 was increased in 3-month-old ED-L2/Klf4 mice before the onset of inflammation, KLF4 was decreased at 6 months of age. In contrast, KLF5 was decreased in ED-L2/Klf4 mice at 3 months of age but increased at 6 months of age. KLF5 promotes keratinocyte proliferation,19 and the findings at 6 months of age likely resulted from increased KLF5 in the expanded basal layer (Supplementary Figure 1F and G), where KLF4 was not expressed (Figure 2B).
All ED-L2/Klf4 mice examined at 6 months, 1 year (data not shown), and 2 years of age had esophageal epithelial dysplasia, and half of the animals developed squamous cell cancer at 2 years of age. However, KLF4 is lost in human esophageal squamous cell cancer,23,24 and homozygous deletion of Klf4 produces squamous cell dysplasia in mice.9 We therefore examined whether KLF4 was expressed in esophageal cancers from ED-L2/Klf4 mice. Although KLF4 was localized to esophageal suprabasal cells in littermate controls (Figure 3A), KLF4 was absent from the tumors (Figure 3B), which arose from esophageal epithelia of ED-L2/Klf4 mice, and yet was still present in the suprabasal layers of adjacent epithelia from ED-L2/Klf4 mice (Figure 3C). Ki-67 staining showed that, compared with esophageal epithelia from littermate control mice (Figure 3D), esophageal tumors from ED-L2/Klf4 mice were highly proliferative (Figure 3E).
Given the non–cell autonomous effects of KLF4 on proliferation in ED-L2/Klf4 mice, we hypothesized that the inflammation might be driving the hyperproliferation, dysplasia, and cancer. To understand the early changes in ED-L2/Klf4 mice, we performed ultrastructural analyses of esophageal epithelia of mice at 3 months of age, before the onset of overt inflammation, hyperplasia, and dysplasia. In contrast to controls (Supplementary Figure 5A), ED-L2/Klf4 mice (Supplementary Figure 5B), despite apparently normal stratification, developed marked intercellular edema or spongiosis (see arrows) within the suprabasal layers. KLF4 establishes the epithelial barrier in the squamous epithelia of the skin,8 but these findings in ED-L2/Klf4 mice suggest that increased Klf4 expression in esophageal epithelia leads to changes in barrier integrity. In fact, ED-L2/Klf4 mice at 3 months of age had increased messenger RNA (mRNA) expression of several junctional proteins, including claudin-14, claudin-15, occludin, and desmocollin-2 (Supplementary Figure 5C). It is not clear whether these increases were a direct effect of KLF4 or a compensatory response to tissue damage. Notably, similar findings were seen in a rat model of reflux esophagitis, in which an increase of junctional proteins was an early molecular event, and expression of these proteins declined as basal cell hyperplasia developed25; dilated intercellular spaces also were observed in this model. Although it is possible that KLF4 expression must be titrated carefully to maintain the barrier, inflammation also can alter barrier integrity and produce changes in the microenvironment.26,27
By using a number of cell lineage markers, we characterized the inflammatory infiltrate in ED-L2/Klf4 mice. No infiltrate was seen in control or ED-L2/Klf4 mice at 3 months of age (data not shown). However, compared with 6-month-old control mice (Figure 4A and C), ED-L2/Klf4 had a large number of neutrophils throughout the epithelia and lamina propria by staining with anti-7/4 antibody (Figure 4B) and increased numbers of F4/80-positive macrophages (Figure 4D), predominantly in the lamina propria just below the basal layer. Staining with antibody to the T-lymphocyte marker CD3 revealed, compared with controls (Figure 4E), increased numbers of T lymphocytes in the esophageal lamina propria and the presence of intraepithelial lymphocytes in ED-L2/Klf4 mice (Figure 4F). CD45R-positive B cells, not seen in control mice (Supplementary Figure 6A), were observed in the esophageal mucosa of ED-L2/Klf4 mice (Supplementary Figure 6B).
The production of proinflammatory cytokines can precede or be the product of an inflammatory response.27 We hypothesized that Klf4 overexpression in esophageal keratinocytes could lead directly to the release of cytokines and chemokines in esophageal epithelia of ED-L2/Klf4 mice, resulting in inflammation. Initially, we examined the expression of the potent neutrophil chemoattractant CXCL5,28 and the proinflammatory cytokine tumor necrosis factor-α (TNFα) by quantitative real-time PCR in esophageal cells of ED-L2/Klf4 mice and controls. At 6 months of age, ED-L2/Klf4 mice had a 100-fold increase in CXCL5 expression (Figure 5A) and a 13-fold increase in TNFα expression (Figure 5B), compared with littermate controls. Interestingly, CXCL5 and TNFα expression already were increased by 2.9- and 3.6-fold, respectively, in ED-L2/Klf4 mice at 3 months of age, before the onset of overt inflammation. To confirm that these changes in expression of CXCL5 and TNFα were the direct result of Klf4 overexpression, we isolated primary esophageal keratinocytes from ED-L2/Klf4 mice and littermate controls and determined the expression of CXCL5 and TNFα by quantitative real-time PCR. Compared with control cells, primary esophageal keratinocytes from ED-L2/Klf4 mice had a 12-fold increase in CXCL5 and a 31-fold increase in TNFα expression (Figure 5C). This induction in CXCL5 and TNFα mRNA also was validated in wild-type primary esophageal keratinocytes, infected with Klf4-expressing retrovirus or control (Supplementary Figure 7).
To determine whether other proinflammatory cytokines were up-regulated by Klf4 overexpression, we performed cytokine antibody arrays and identified up-regulation of TNFα, granulocyte colony–stimulating factor (G-CSF), and interleukin (IL)-1α proteins in conditioned media from ED-L2/Klf4 keratinocytes, compared with controls (Supplementary Figure 8). G-CSF and IL-1α secretion by primary keratinocytes was correlated with an increase in mRNA expression of G-CSF by 9-fold and IL-1α by 17-fold in vitro (Figure 5D). This up-regulation of G-CSF and IL-1α also was confirmed by infecting primary esophageal keratinocytes from wild-type mice with Klf4-expressing retrovirus (Supplementary Figure 7) and in vivo because 6-month-old Klf4 transgenic mice had a 44-fold increase in G-CSF mRNA levels (Figure 5E) and a 2-fold increase in IL-1α mRNA (Figure 5F), compared with littermate controls. No changes were seen in the expression of G-CSF and IL-1α at 3 months of age, suggesting that up-regulation of these cytokines in vivo follows the earlier induction of CXCL5 and TNFα by KLF4. Next, we treated naive keratinocytes from wild-type mice with conditioned media from keratinocytes isolated from control and ED-L2/Klf4 mice. Relative to conditioned media from control cells, conditioned media from ED-L2/Klf4 mice increased proliferation of these naive keratinocytes (Supplementary Figure 9). Thus, secreted factors regulated by KLF4, likely including but not limited to the previously identified cytokines, increased proliferation of esophageal keratinocytes in vitro, consistent with the non–cell autonomous role for KLF4 in the regulation of esophageal epithelial proliferation in vivo.
Nuclear factor-κ B (NF-κB) is a key regulator of inflammation and has a central role in the control of proliferation and immune response genes.29,30 We postulated that activation of NF-κB by Klf4 overexpression might underlie the increased expression of CXCL5, TNFα, G-CSF, and IL-1α. To determine whether NF-κB was activated after Klf4 overexpression in vivo, we stained sections from ED-L2/Klf4 mice and littermate controls for the p65 subunit of NF-κB. At 3 months of age, little p65 staining was seen within esophageal keratinocytes of control mice (Supplementary Figure 10A), whereas nuclear p65 was prominent in esophageal epithelia of ED-L2/Klf4 mice (Supplementary Figure 10B). NF-κB activation at 3 months of age also was confirmed by Western blot for phospho-p65 (Supplementary Figure 10C). These findings suggest that NF-κB activation was an early event in the inflammatory response because it precedes recruitment of inflammatory cells. At 6 months of age, p65 staining in esophageal mucosa of control mice (Supplementary Figure 10D) was minimal compared with ED-L2/Klf4 mice (Supplementary Figure 10E), where strong nuclear p65 staining was identified in esophageal keratinocytes and immune cells. Compared with keratinocytes isolated from control mice, ED-L2/Klf4 keratinocytes also had a 2.9-fold increase in NF-κB activity, which was blocked by treatment with an inhibitor of nuclear factor kappa-B kinase subunit beta inhibitor (Figure 6A). Infection of wild-type keratinocytes with Klf4-expressing retrovirus also increased NF-κB activity, compared with control (Supplementary Figure 11A).
To determine if NF-κB activation was required for the induction of CXCL5, TNFα, G-CSF, and IL-1α by KLF4, primary esophageal keratinocytes from control and ED-L2/Klf4 transgenic mice were cultured in the presence or absence of IKK2 inhibitor. Inhibition of NF-κB signaling abolished the up-regulation of CXCL5 (Figure 6B), TNFα (Figure 6C), G-CSF (Figure 6D), and IL-1α mRNA (Figure 6E) in ED-L2/Klf4 keratinocytes, indicating that NF-κB activation was upstream of these proinflammatory cytokines in esophageal keratinocytes. NF-κB has been shown to promote cell survival by suppression of apoptosis and necrosis, including in premalignant cells.30 Thus, we evaluated the viability of primary esophageal keratinocytes from control and ED-L2/Klf4 mice treated with or without IKK2 inhibitor. IKK2 inhibition did not affect cell viability in keratinocytes from control mice, but treatment of ED-L2/Klf4 keratinocytes with IKK2 inhibitor led to a statistically significant 22% decrease in cell viability (Figure 6F).
Given the increased expression of TNFα in ED-L2/Klf4 keratinocytes, we examined whether NF-κB was activated by TNFα in esophageal keratinocytes, perhaps as part of a proinflammatory feedback loop, by culturing keratinocytes in the presence or absence of TNFα neutralizing antibody and assaying for NF-κB transcriptional activity. In control esophageal keratinocytes treated with recombinant TNFα, addition of TNFα neutralizing antibody successfully blocked TNFα downstream effects on NF-κB (Supplementary Figure 11B). Without addition of recombinant TNFα, TNFα neutralization had no effect on NF-κB transcriptional activity in control keratinocytes; however, in keratinocytes from ED-L2/Klf4 mice, inhibition of TNFα decreased NF-κB transcriptional activity (Figure 7A). These results suggested that after KLF4 over-expression, NF-κB was activated, leading to increased expression of TNFα, which subsequently induced NF-κB via a positive feedback loop. Finally, we determined if TNFα drove the increased expression of CXCL5, TNFα, G-CSF, and IL-1α within keratinocytes. In contrast to IKK2 inhibition, TNFα neutralization did not block the mRNA induction of CXCL5 (Figure 7B), TNFα itself (Figure 7C), G-CSF (Figure 7D), or IL-1α (Figure 7E). Taken together, these results indicate that, in the context of Klf4 overexpression in esophageal keratinocytes, CXCL5, TNFα, G-CSF, and IL-1α are targets of activated NF-κB.
KLF4 has a well-established role in transcriptional regulation of proliferation and differentiation, and altered expression has been found in a number of cancers.5,10,11 Yet the function of KLF4 depends greatly on context.5 Recently, we showed that deletion of Klf4 in murine esophagus results in squamous cell dysplasia and delayed differentiation.22 Here, we identify a novel proinflammatory role for KLF4 via activation of the IKK/NF-κB pathway within epithelial cells. Interestingly, increased nuclear localization of NF-κB and increased levels of TNFα and CXCL5 precede recruitment of inflammatory cells to the sites of this inflammation. Moreover, as shown by complementary experiments in primary esophageal keratinocytes in culture, this initial activation of the NF-κB pathway is KLF4-dependent and does not require bacterial infection or injury.
So how does KLF4 activate NF-κB? Preliminary studies do not suggest a direct effect of KLF4 on NF-κB transcription (data not shown), and, given the diversity of stimuli leading to NF-κB activation,31 the specific targets of KLF4 may be difficult to discern. Previously, deletion of p120 catenin was shown to similarly initiate inflammation in skin keratinocytes through indirect activation of NF-κB by a complex mechanism, possibly involving induction of RhoA guanosine triphosphatase activity.32 In macrophages, KLF4 interacts with the p65 subunit of NF-κB to cooperatively induce the inducible nitric oxide synthase promoter,33 but we did not find evidence that KLF4-p65 binding in keratinocytes was altered by increased KLF4 expression (data not shown). Thus, precisely how KLF4 influences NF-κB activity remains to be determined.
At first, the development of squamous cell dysplasia with both loss22 and overexpression of Klf4 might seem to be confusing. Nonetheless, we believe that the critical event leading to dysplasia and squamous cell cancer in ED-L2/Klf4 mice is inflammation. KLF4, which is lost in esophageal squamous cell cancers in mice and human beings,23,24 is likely to have tumor-suppressive properties in esophageal keratinocytes, as suggested in several other studies.12,13,34 The mechanism of KLF4 loss in esophageal cancers is not known but may parallel loss in other gastrointestinal cancers, in which both hypermethylation and hemizygous deletion have been implicated.11
Inflammation is linked to carcinogenesis in a number of epithelial tissues, and various cytokines and other inflammatory agents can act as tumor promoters in the context of chronic inflammation.27,30 For example, TNFα is an important regulator of the early stages of tumor promotion in the skin.35 CXCL5 is overexpressed in head and neck squamous cell cancer and its down-regulation inhibits squamous carcinogenesis by decreasing invasion and cell proliferation.36 NF-κB signaling is involved in epidermal development, squamous cell homeostasis, chronic inflammatory diseases, and cancers, including esophageal squamous cell carcinomas.29,37,38 However, in some mouse models, NF-κB inhibition in epithelial cells results in the spontaneous development of severe inflammation.29 Thus, a careful balance of NF-κB activation is required for the maintenance of normal epithelial and immune homeostasis, and much more investigation is needed, using animal models such as ED-L2/Klf4 mice, before targeting of NF-κB or other inflammatory pathways could be considered in the chemoprevention of esophageal squamous cell cancer.
Disruption of the epithelial barrier can elicit an inflammatory response,26 and we did observe ultrastructural changes in esophageal epithelia of ED-L2/Klf4 mice, with dilated paracellular spaces or spongiosis. Spongiosis occurs in patients with gastroesophageal reflux disease and in animals exposed to various damaging agents.39,40 However, spongiosis also can occur in response to cytokine stimulation alone.41 Because KLF4 induces a proinflammatory response in a sterile environment in vitro, we believe that activation of the NF-κB pathway is the initial event after Klf4 overexpression in vivo, which subsequently induces dilatation of paracellular spaces. However, we cannot exclude completely that KLF4 directly regulates barrier function independent of NF-κB activity. Further investigations will be needed to determine whether defects in barrier function occur independent of the NF-κB pathway in our model.
ED-L2/Klf4 mice develop cancers at an advanced age, consistent with the time-course of esophageal squamous cell cancer development in human beings.1 Although only 3 of 6 ED-L2/Klf4 mice developed esophageal squamous cell cancer, we believe that the development of cancer is a stochastic process that arises in the setting of squamous cell dysplasia, the inflammatory milieu, and additional genetic events, perhaps including loss of KLF4. Moreover, although many associate inflammation with esophageal adenocarcinoma, the major risk factor for esophageal squamous cell cancer is chronic irritation, most frequently by alcohol and tobacco, alone or in combination, as well as by achalasia, esophageal diverticuli, and frequent consumption of extremely hot beverages.1,3 Furthermore, the incidence of esophageal squamous cell carcinoma increases more than 1000-fold after caustic ingestion with stricture formation, and the typical interval between injury and the development of squamous cell carcinoma in human beings is more than 30 years.42 Exposure to nitrosamines is strongly linked to human esophageal squamous cell cancer,1,3 and in mice a prominent subepithelial inflammatory infiltrate is observed during carcinogenesis with methyl-N-amylnitrosamine.43 These factors suggest that inflammation plays a considerable role in the development of esophageal squamous cell cancers. Yet gastroesophageal reflux and reflux esophagitis do not appear to be risk factors for esophageal squamous cell cancer,1 indicating that the causative role of inflammation in esophageal squamous cell cancer still requires further study. Thus, ED-L2/Klf4 mice provide an important model for human esophageal squamous dysplasia and inflammation-mediated squamous cell cancer.
In sum, we propose a model in which Klf4 overexpression promotes proliferation through cytokine activation within esophageal epithelial cells with subsequent recruitment of inflammatory cells and likely disruption of the epithelial barrier, promoting a pro-proliferative, procarcinogenic inflammatory milieu (Figure 7F). The lack of KLF4 expression in proliferating cells of ED-L2/Klf4 mice certainly supports a non–cell autonomous role for KLF4 in proliferation. Further, the absence of KLF4 within the tumors suggests that KLF4 might, in fact, function as a tumor suppressor in esophageal squamous cells, as previously proposed and as seen in several other epithelia.10,11 However, whether the tumors in ED-L2/Klf4 mice arise from cells that do not express Klf4 or if Klf4 is later silenced, either epigenetically or by other means, is not known. Nonetheless, given the phenotypes of mice with Klf4 loss22 and gain of function, it is clear that KLF4 expression must be tightly regulated in normal esophageal epithelia.
This work was supported by National Institutes of Health National Institute for Diabetes and Digestive and Kidney Diseases R01 DK069984 (J.P.K.) and by the University of Pennsylvania Center for Molecular Studies in Digestive and Liver Diseases (National Institutes of Health National Institute for Diabetes and Digestive and Kidney Diseases P30 DK050306) through the Morphology Core, the Molecular Biology Core, and the Transgenic and Chimeric Mouse Facility and by National Institutes of Health National Cancer Institute P01 CA098101.
Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2010.08.048.
Conflicts of interest
The authors disclose no conflicts.