Esophageal squamous cell dysplasia and delayed differentiation with deletion of Klf4 in murine esophagus

Abstract

Background & Aims

Klf4 (Krüppel-like factor 4; GKLF) is a DNA-binding transcriptional regulator highly expressed in skin and gastrointestinal epithelia, specifically in regions of cellular differentiation. Homozygous null mice for Klf4 die shortly after birth from skin defects, precluding their analysis at later stages. The aim of this study was to analyze the function of Klf4 in keratinocyte biology and epithelial homeostasis in the adult by focusing on the squamous lined esophagus.

Methods

Using the ED-L2 promoter of Epstein-Barr virus to drive Cre, we obtained tissue specific ablation of Klf4 in the squamous epithelia of the tongue, esophagus, and forestomach.

Results

Mice with loss of Klf4 in esophageal epithelia survived to adulthood, bypassing the early lethality. Tissue-specific Klf4 knockout mice had increased basal cell proliferation and a delay in cellular maturation; these mice developed epithelial hypertrophy and subsequent dysplasia by 6 months of age. Moreover, loss of Klf4 in vivo was associated with increased expression of the pro-proliferative Klf5, and Klf4 downregulated Klf5 both transcriptionally and post-transcriptionally. Using gene expression profiling, we also showed decreased expression of critical late-stage differentiation factors and identified alterations of several genes important in cellular differentiation.

Conclusions

Klf4 is essential for squamous epithelial differentiation in vivo and interacts with Klf5 to maintain normal epithelial homeostasis.

Maintenance of normal esophageal epithelial homeostasis requires cell differentiation and desquamation at the luminal surface to be matched by cell proliferation^1–5. Esophageal keratinocyte proliferation normally occurs in the basal layer and cells differentiate as they migrate through the suprabasal and superficial layers. Dysregulation of the normal balance of proliferation and differentiation may underlie disorders such as gastroesophageal reflux disease, which affects up to 44% of the US population⁶ and esophageal cancer, the 6^th most common cause of cancer death worldwide⁷. Moreover, the processes of keratinocyte differentiation in the esophagus have much in common with those in other squamous tissues, such as the interfollicular skin and the oropharynx, making the esophagus an outstanding model to understand the processes of squamous epithelial homeostasis and disease in general^{2–4, 8}.

Krüppel-like factors (KLFs) are DNA-binding transcriptional regulators which play critical roles in the control of proliferation, differentiation, development, and carcinogenesis in a number of tissues^{9, 10}. Among these, Klf4 (previously known as GKLF or EZF) is highly expressed in differentiating cells of the gastrointestinal tract, including the suprabasal and superficial layers of the esophagus, as well as the skin¹¹. This distribution suggests that Klf4 may function in the switch from proliferation to differentiation in stratified squamous epithelia. In vitro, Klf4 overexpression inhibits proliferation and promotes differentiation of esophageal keratinocytes¹². Furthermore, Klf4 transcriptionally activates keratin 4, a marker of keratinocyte differentiation^{13, 14}. KLF4 is reportedly downregulated in a number of human epithelial cancers, including esophageal, colorectal, gastric, and bladder carcinomas^15–18. However, in other contexts, KLF4 may promote tumorigenesis¹⁹.

Several animal studies have provided information about the role of Klf4 in epithelial cells. Many of these studies have focused upon or been limited to the role of Klf4 during embryogenesis, in part due to the early lethality of Klf4 null mice, which die on postnatal day 1 from skin barrier defects²⁰. Klf4 null mice have abnormal differentiation of skin and of goblet cells in the colon, confirming a function for Klf4 in differentiation of a number of epithelial types^{20, 21}. In a contrasting experiment, ectopic expression of Klf4 using a tetracycline inducible keratin 5 promoter produces accelerated epidermal differentiation in the embryo²². To understand the role of Klf4 later in life, we utilized the Cre-loxP system to obtain tissue-specific deletion of Klf4 in the glandular epithelium of the stomach²³. Loss of Klf4 in gastric glands results in a four-fold increase in cellular proliferation and alterations in the normal differentiation pathways of the stomach. Conditional deletion of Klf4 in the eye also results in increased proliferation but with fewer epithelial layers, due to corneal fragility, and loss of conjunctival goblet cells²⁴. These studies highlight the similarities and differences of Klf4 function in various types of epithelia.

In this study, we used the ED-L2 promoter of the Epstein-Barr virus, previously shown to target expression to squamous epithelia of the tongue, esophagus and forestomach^{25, 26}, to produce tissue specific gene ablation of Klf4. This approach allowed us to investigate the role of Klf4 in esophageal epithelial homeostasis while bypassing the early lethality of Klf4 null mice, resulting from Klf4 loss in the skin²⁰. Mice with loss of Klf4 in esophagus developed epithelial hypertrophy, with increased proliferation, altered cell morphology, and evidence of delayed cellular maturation. Similar changes were seen with Klf4 loss in tongue and with patchy deletion of Klf4 in the skin of the ventral neck. Klf4 conditional null mice also developed squamous cell dysplasia by 6 months of age. Thus, expression of Klf4 is essential for normal squamous epithelial differentiation and homeostasis in vivo.

Methods

ED-L2-Cre/Klf4^loxp/loxpmice

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania. Mutant mice were homozygous for floxed Klf4 and hemizygous for the Cre transgene. All mice used for the experiments were on a mixed genetic backgound. For all experiments with ED-L2-Cre/Klf4^loxp/loxp mice, sex-matched littermate Klf4^loxp/loxp mice lacking the Cre transgene served as controls. Additional details are provided in the supplementary methods.

Cell Culture and Treatment

The isolation and culture of mouse primary esophageal keratinocytes have been described elsewhere²⁷. Cells were cultured and infected or transfected and reporter assays or quantitation were performed as described in the supplementary methods.

Immunohistochemistry/Immunofluorescence/Western blots

Immunohistochemistry, immunofluorescence, and Western blots were performed using standard protocols. For descriptions of the protocols and antibodies used, see the supplementary methods.

RNA analyses

RNA was extracted from esophagi or primary esophageal keratinocytes using the RNeasy Mini Kit (Qiagen, Valencia, CA) following manufacturer’s instructions. Reverse transcription was performed with random hexamers and SuperScript II Reverse Transcriptase (Invitrogen). Quantitative real-time PCR (qPCR) analysis was performed in triplicate using an ABI Prism 7000 sequence detection system (Applied Biosystems, Foster City, CA) and SyBr Green PCR master mix (Applied Biosystems). The TATA box binding protein gene (TBP) was used as the internal control. We used GeneChip Mouse Expression Arrays MOE430A v2 (Affymetrix, Santa Clara, California) for microarray analyses and identified differentially expressed genes using a fold change cutoff of greater than or equal to 2.0 and a false discovery rate of 5%, as described¹². Microarray results have been deposited in GEO under the accession number GSE17447.

Transmission Electron Microscopy

Ultrathin (~80nm) sections of the trans-sectional plane of each esophagus were examined with a FEI Tecnai-T12 transmission electron microscope operated at 80kv. Digital images were captured with a Gatan camera at a resolution of x2048. Additional details are provided in the supplementary methods.

Results

To generate tissue-specific recombination in esophagus, we utilized the EBV ED-L2 promoter, which targets expression specifically to the tongue, esophagus, and forestomach, but not the skin^{25, 26}. Using Gt(ROSA)26^tm1Sor reporter mice²⁸, we demonstrated that ED-L2/Cre directed Cre-mediated recombination in esophageal epithelia. While no β-galactosidase expression was seen in esophageal tissues of mice containing Gt(ROSA)26^tm1Sor without the ED-L2/Cre transgene (Fig. 1A), ED-L2/Cre;Gt(ROSA)26^tm1Sor littermates (Fig. 1B) had staining throughout the epithelia. Staining patterns were similar in proximal and distal esophagus. Expression was mosaic in the epithelia of the tongue and forestomach (not shown), and Cre-mediated recombination was demonstrated in two independent lines.

Fig. 1.

Klf4 was deleted successfully in esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice. (A–B) Esophageal sections from Gt(ROSA)26^tm1Sor control mice (A) showed no epithelial staining for β-galactosidase, while ED-L2/Cre;Gt(ROSA)26^tm1Sor mice (B) had staining throughout the epithelia. (C–D) Both control (C) and ED-L2/Cre;Klf4^loxP/loxP mice (D) demonstrated nuclear staining for Klf4 in the skin. (E–F) In esophageal epithelia, control mice (E) had nuclear Klf4 expression in cells of the suprabasal layer while Klf4 expression was lost in ED-L2/Cre;Klf4^loxP/loxP mice (F). Scale bars (A–F), 10 μm.

To produce deletion of Klf4 in esophageal epithelia of adult mice, we crossed the ED-L2/Cre transgenic mice with Klf4^loxP/loxP mice²³. We documented Klf4 gene ablation by PCR (data not shown) and demonstrated the targeted deletion of Klf4 by performing immunohistochemistry. Staining for Klf4 revealed expression in the skin of the back of both littermate control (Fig. 1C) and ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 1D). While staining was seen in esophageal suprabasal cells in control mice (Fig. 1E), Klf4 expression was absent in esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 1F), confirming Cre-mediated deletion of Klf4 in the esophagus.

ED-L2/Cre;Klf4^loxP/loxP mice were born at appropriate Mendelian ratios and survived into adulthood. Grossly, the esophagi and other organs appeared normal, with the exception of the skin of the face and ventral neck, as detailed below. Heterozygous mice had no overt phenotype. Up to one month of age, esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice appeared histologically normal. However, compared to littermate controls (Fig. 2A–B), esophageal epithelia of 1 month-old ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 2C–D) had evidence of hypertrophy with altered keratinocyte morphology. This phenotype was more evident in ED-L2/Cre;Klf4^loxP/loxP mice at 3 months of age (Fig. 2G–H), compared to littermate controls (Fig. 2E–F). Changes were even more dramatic at 6 months of age. Compared to littermate controls (Fig. 2I–J), esophageal epithelia of 6 month-old of ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 2K–L) showed phenotypes ranging from hyperplasia to moderate dysplasia, with loss of surface keratinization, and an apparent delay in keratinocyte maturation. No evidence of invasion was seen in the ED-L2/Cre;Klf4^loxP/loxP mice, although there were many epithelial buds pushing into the lamina propria.

Fig. 2.

ED-L2/Cre;Klf4^loxP/loxP mutant mice had altered esophageal epithelial homeostasis. Compared to 1 month-old (A–B) and 3 month-old littermate controls (E–F), H&E stained esophageal epithelia from ED-L2/Cre;Klf4^loxP/loxP mice were hypertrophic and keratinocytes showed altered morphology (C–D, G–H). The basal layers at these time points also appeared thickened and irregular. By 6 months of age, compared to littermate controls (I–J), ED-L2/Cre;Klf4^loxP/loxP mice had evidence of delayed keratinocyte maturation and showed moderate dysplasia (K–L). Note the budding of epithelia into the lamina propria and the lack of surface keratinization in ED-L2/Cre;Klf4^loxP/loxP mice. Scale bars (A, C, E, G, I, K), 50 μm; (B, D, F, H, J, L), 10 μm.

Even though Klf4 was not generally deleted in skin, some patchy deletion occurred most predominantly on the skin of the face and ventral neck (Fig. S1). While this was not sufficient to cause lethality, allowing us to study the effects of Klf4 loss on esophagus in adults, these lesions in the skin generally necessitated euthanasia of ED-L2/Cre;Klf4^loxP/loxP mice by 6 months of age for humane reasons. We observed crusted skin lesions and moistness of the skin of the ventral neck starting at 3 months of age. No lesions were observed in littermate controls. To rigorously investigate the effect of Klf4 in other squamous epithelia, the histology of tongue, forestomach, and skin was analyzed (Fig. S2), and the pattern of Klf4 deletion in these tissues was confirmed by immunohistochemistry (Fig. S3). Evidence of hypertrophy was also seen in the tongue and the skin on the ventral neck, where Klf4 was deleted in ED-L2/Cre;Klf4^loxP/loxP mice. No differences in histology were seen in the skin of the back, where Klf4 was not deleted. In additional, no significant changes were observed in the forestomach of ED-L2/Cre;Klf4^loxP/loxP mice, likely due to incomplete Klf4 deletion in this tissue.

As the phenotype was well-established by 3 months of age and mice were euthanized by 6 months of age, we chose to focus on the 3 and 6 month time points for further experiments. To examine the effects of Klf4 deletion at the ultrastructural level, we performed transmission electron microscopy on esophageal epithelia from 3 month-old mice. In littermate controls (Fig. 3A), cells progressing from the basal to suprabasal layers reoriented and underwent stratification. In contrast, the initial layers of suprabasal cells in ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 3B) did not undergo stratification, with these cells instead maintaining a longitudinal orientation. Basal cells of ED-L2/Cre;Klf4^loxP/loxP mice also demonstrated nuclear distortion, suggesting an expansion of proliferative cells, and superficial cells had increased nuclear to cytoplasmic ratios (Fig. 3C–D; Fig. S4) compared to controls, suggesting incomplete differentiation.

Fig. 3.

Ultrastructural abnormalities in Klf4-deficient esophageal epithelia from 3 month-old mice. Assessment of ultrastructure of littermate controls (A, C) and ED-L2/Cre;Klf4^loxP/loxP mutant mice (B, D). (A–B) Compared to controls (A), esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice (B) demonstrated changes in the orientation of cells as they progressed from the basal to suprabasal layers. (C–D) In the superficial layer, compared to controls (C), cells from ED-L2/Cre;Klf4^loxP/loxP mice (D) had increased nuclear to cytoplasmic ratios. Typically, cells in this layer have small compacted nuclei. Scale bars (A–D), 2 μm.

To analyze whether these apparent defects of cell maturation in Klf4 conditional null mice resulted in increased proliferation, we pulse-labeled cells in S-phase with 5-bromo-2-deoxyuridine (BrdU) and determined the proliferative index by counting the number of BrdU-labeled cells per 100 basal cells. Compared to controls (Fig. 4A), ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 4B) showed an increased number of proliferating cells, with changes in proliferation confined to the basal layer. A statistically significant 1.9-fold increase (Fig. 4C) in proliferating cells was seen in ED-L2/Cre;Klf4^loxP/loxPmice at 3 months of age, comp ared to littermate controls (p=0.006). We next investigated whether ED-L2/Cre;Klf4^loxP/loxP mice had altered apoptosis in esophageal epithelia by TUNEL staining. Compared to littermate controls (Fig. 4D), which had only rare apoptotic cells in esophageal epithelia, ED-L2/Cre;Klf4^loxP/loxP mice had increased apoptosis at 3 months of age (Fig. 4E). This may represent an attempt to balance hyperproliferation and apoptosis to maintain homeostasis, as has been described in other epithelia including skin²⁹. By counting the number of TUNEL positive cells per section (Fig. 4F), we confirmed a 2.7–fold increase in apoptosis in ED-L2/Cre;Klf4^loxP/loxP mice compared to controls (p=0.001). Nonetheless, this increased apoptosis was insufficient to prevent the progressive epithelial hypertrophy seen in ED-L2/Cre;Klf4^loxP/loxPmice.

Fig. 4.

ED-L2/Cre;Klf4^loxP/loxP mice had increased proliferation and apoptosis in esophageal epithelia. (A–B) At 3 months of age, proliferating cells were confined to the basal layer in both control (A) and ED-L2/Cre;Klf4^loxP/loxP mice (B). (C) Quantitation of BrdU-labeled cells revealed a statistically significant 1.9-fold increase in cell proliferation in ED-L2/Cre;Klf4^loxP/loxP mice (p=0.006). (D–E) Compared to littermate controls (D), TUNEL staining of esophageal epithelia from ED-L2/Cre;Klf4^loxP/loxP mice (E) demonstrated increased numbers of apoptotic cells in the superficial layers at 3 months of age. (F) By quantitation of TUNEL positive cells, apoptosis was increased 2.7-fold in ED-L2/Cre;Klf4^loxP/loxP mice (p=0.001). Scale bars (A–B, D–E), 50 μm.

To characterize changes in the proliferation-differentiation equilibrium in ED-L2/Cre;Klf4^loxP/loxP mice, we stained 3 month-old mice for keratin 14, which normally marks undifferentiated cells in the basal layer, and keratins 4 and 13, markers of keratinocyte differentiation³⁰. While keratin 14 was localized to the esophageal basal layer of control mice (Fig. 5A), the region of keratin 14 staining was expanded to include at least the first several layers of suprabasal cells in ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 5B), consistent with delayed differentiation of keratinocytes in the suprabasal layer. In control mice, keratin 13 (Fig. 5C) and keratin 4 (Fig. 5E) were found in band-like patterns throughout the suprabasal and superficial layers. In contrast, ED-L2/Cre;Klf4^loxP/loxP mice had a marked shift in keratin 13 expression towards the lumen with a loss of the normal band-like pattern (Fig. 5D) and nearly complete extinction of keratin 4 expression in the suprabasal layer (Fig. 5F), further indicating a delay in keratinocyte maturation with Klf4 loss.

Fig. 5.

Klf4 deletion altered esophageal keratinocyte differentiation. (A–B) Staining for keratin 14 (red), which is normally confined to cells in the basal layer, as in controls (A), revealed expansion of the expression domain to include cells within the first few layers of suprabasal cells in ED-L2/Cre;Klf4^loxP/loxP mice (B). (C–D) Compared to controls (C), expression of keratin 13, a marker of keratinocyte differentation, was decreased and shifted toward the luminal surface in ED-L2/Cre;Klf4^loxP/loxP mice (D). (E–F) Keratin 4, another keratinocyte differentiation marker, was strongly expressed in the suprabasal layers of controls (E), but expression was markedly decreased in ED-L2/Cre;Klf4^loxP/loxP mice (F). DAPI (blue) was used as a nuclear counterstain for immunofluorescence (A–B, E–F). Scale bars, 10 μm.

The closely related KLF family member Klf5 binds to the same cis-regulatory elements as Klf4 and has been shown to have pro-proliferative properties in a number of contexts¹¹. Moreover, transgenic expression of Klf5 in esophagus results in increased basal cell proliferation¹². However, while Klf5 has been shown to transcriptionally regulate Klf4 in vitro³¹, similar regulation of Klf5 by Klf4 has not previously been reported. To evaluate whether Klf5 might contribute to the phenotype of ED-L2/Cre;Klf4^loxP/loxP mice, we examined the expression of Klf5 in esophageal epithelia. As shown in Figure 6A, nuclear Klf5 staining was seen only in basal cells in esophagus of 3 month-old control mice but was found in both basal and suprabasal layers of ED-L2/Cre;Klf4^loxP/loxP mice (Fig. 6B). Consistent with this, Klf5 mRNA levels in esophageal epithelial cells isolated from ED-L2/Cre;Klf4^loxP/loxP mice were increased by 50% compared to control mice (Fig. 6C). To determine whether Klf5 was a direct transcriptional target for Klf4, we examined the 5′ regulatory region of Klf5 for putative Klf4-binding sites, using the computational program TESS³² and identified a putative Klf4 binding site ~600bp from the translation start site. By ChIP assays in mouse primary esophageal keratinocytes, we demonstrated binding of Klf4 to Klf5 between −601 to −419 (Fig. 6D). To confirm that Klf4 functionally inhibited Klf5, we transfected mouse primary esophageal keratinocytes infected with pBabe-puro control or pBabe-Klf4 with a luciferase reporter containing the 1 kb region immediately upstream of the Klf5 translational start site. Expression of Klf4 by pBabe-Klf4 resulted in a 53% decrease in activity (Fig. 6E). Moreover, Klf4 overexpression resulted in a similar suppression of Klf5 mRNA (Fig. 6F), confirming that Klf5 is transcriptionally repressed by Klf4. At the protein level, Klf5 downregulation by Klf4 was even more dramatic (Fig. 6F), suggesting that Klf4 may also exert effects on post-transcriptional regulation of Klf5.

Fig. 6.

Klf4 inhibited expression of pro-proliferative Klf5. (A) Klf5 staining was seen in nuclei of cells only in the basal layer of the esophagus. (B) In contrast, ED-L2/Cre;Klf4^loxP/loxP mice showed expression of Klf5 in both basal and suprabasal cells. (C) By qPCR, Klf5 expression was increased by 54% in esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice (p=0.01). (D) ChIP assays of mouse primary esophageal keratinocytes demonstrated Klf4 binding to the 5′ regulatory region of Klf5 between −601 and −419. Cells were also treated with calcium to induce differentiation. Input was DNA extracted before immunoprecipitation. Anti-IgG antibody was used as a negative control. (E) Luciferase reporter assays with the 1 kb region immediately upstream of the Klf5 translational start site revealed a 53% decrease in activity (p<0.05) when cells were co-transfected with a Klf4 expression vector, compared to control. (F) Overexpression of Klf4 in mouse primary esophageal keratinocytes resulted in a 45% decrease in Klf5 mRNA by qPCR (p<1×10⁻⁹). (G) Western blotting and subsequent quantitation revealed a 98% decrease in Klf5 protein when Klf4 was overexpressed by 7-fold in these cells. A model for Klf4 and Klf5 in the switch from proliferation to differentiation in esophageal epithelia is shown in (H).

To identify other potential Klf4 targets in esophageal epithelium, we isolated esophageal epithelial cells from control and ED-L2/Cre;Klf4^loxP/loxP mice at 3 months of age and performed gene expression profiling. We identified a total of 121 genes which were differentially regulated more than two-fold in esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice compared to littermate controls (Table S1). Many of these genes are involved in cornified cell envelope formation, keratinization, and keratinocyte differentiation. Among the most highly upregulated genes in esophageal epithelia of the ED-L2/Cre;Klf4^loxP/loxP mice were those encoding the small proline-rich proteins (Sprr2d, Sprr2f, Sprr2g, Sprr2h, Sprr2i, Sprr2j), which serve as precursors of the cornified cell envelope³³. Consistent with the findings of our microarray analyses, increased expression of Sprr2d, Sprr2f, Sprr2g, Sprr2h, Sprr2i, Sprr2j was reported in the skin of neonatal Klf4-null mice³³. Interestingly, decreased expression of late-stage differentiation markers, such as envoplakin, Klk5, and late cornified envelope 1A1, 1A2, 1B, and 1D was also observed in esophageal epithelia of the ED-L2/Cre;Klf4^loxP/loxP mice.

From the total number of differentially expressed genes, we selected 16 genes for validation by qPCR. These genes were chosen based upon their likely roles in keratinocyte differentiation, implication in clinically relevant diseases involving squamous epithelia, and/or large fold-changes in ED-L2/Cre;Klf4^loxP/loxP mice versus controls. Of the 16 genes tested, 14 had statistically significant changes confirmed by qPCR which were in the same direction as the microarray results (Table 1). The relevance and importance of a number of these genes are outlined in the discussion below. One additional gene, Ephb6, was similarly decreased by qPCR but did not reach statistical significance (p=0.09). The final gene, Ltbp1, showed no change on qPCR and was likely a false-positive on microarray. Thus, the qPCR data confirmed the validity of the gene expression profiling.

Table 1.

Differentially regulated genes on microarray confirmed by qPCR

Gene name	Fold change by microarray*	Fold change by qPCR*	p-value
Msln	3.2	20.4	0.0002
Ctsc	2.2	2.3	0.0003
Evpl	−2.0	−3.6	0.008
Lce1d	−2.0	−3.6	0.0004
Timp2	−2.1	−1.7	0.01
Tcf23	−2.2	−4.3	0.0005
Lce1a2	−2.2	−3.6	0.0002
Lce1a1	−2.2	−4.0	0.002
Igfbp6	−2.6	−2.4	0.0002
Lce1b	−2.9	−4.8	0.0003
Wnt5a	−3.0	−2.9	0.0001
Dkkl1	−5.0	−2.1	0.005
Klk5	−5.0	−5.6	0.0005
Slurp1	−9.0	−7.1	0.0004

^*ED-L2/Cre;Klf4^loxP/loxP mice compared to littermate controls

Given the importance of the Wnt pathway in controlling cellular homeostasis, the reported link between Klf4 and β-catenin³⁴, and the identification of several potentially relevant genes by gene expression profiling, we examined whether nuclear localization of β-catenin was altered in ED-L2/Cre;Klf4^loxP/loxP mice versus controls. Unlike in the intestine, Klf4 does not appear to play a role in canonical Wnt signaling, based on the lack of changes in nuclear β-catenin (Figure S5). However, loss of Klf4 does appear to alter both β-catenin levels and localization. This finding may be more important with respect to the role of β-catenin as a structural protein at the adherens junctions³⁵. Moreover, we cannot exclude an effect of Klf4 on non-canonical Wnt signaling. Of note, Wnt5a (see Table 1) is involved in non-canonical Wnt signaling and has been implicated in tumor progression, although the evidence as to whether Wnt5a has a tumor-promoting or tumor-suppressing role is conflicting³⁶.

Discussion

Under normal conditions, the adult tissue architecture is maintained by a careful balance between the rates of cell production and loss, which is in turn regulated by distinct patterns of transcriptional control. Diseases of the esophagus are among the leading causes of morbidity and mortality in the United States and throughout the world^{37, 38}, and in the esophagus and other tissues, disruption of the factors governing squamous epithelial homeostasis may lead to uncontrolled cell proliferation, inappropriate injury responses, dysplasia, and/or cancer⁵. While a number of important pathways have been identified for the development and differentiation of the skin⁴, many details remain to be elucidated, and less is known about the regulatory pathways in the squamous epithelia of the esophagus.

In this study, we examine the role of the transcriptional regulator Klf4 in esophageal epithelial homeostasis. By utilizing the ED-L2 promoter to delete Klf4, we bypass the early lethality of Klf4 null mice²⁰ and are uniquely qualified to evaluate the function of Klf4 in esophageal epithelial homeostasis in the adult. Loss of Klf4 in squamous epithelia of the esophagus, as well as tongue and skin of the ventral neck, results in hypertrophy, hyperplasia, defects in differentiation, and dysplasia, all of which occur in the absence of any significant inflammatory response. Previously, transgenic expression of human KLF4 was reported to induce squamous epithelial dysplasia in the skin³⁹. Our findings do not exclude that Klf4, as a critical mediator of the proliferation-differentiation equilibrium, must be carefully titrated and that both too much or too little can result in dysregulation and dysplasia.

Klf4 is down-regulated in esophageal squamous cell carcinoma and in a number of esophageal cancer cell lines^{16, 17, 40}. These observations suggest that Klf4 may contribute to the initiation and/or progression of esophageal cancer. Unfortunately, due to patchy deletion of Klf4 in the skin of the ventral neck of ED-L2/Cre;Klf4^loxP/loxP mice, leading to development of severe skin lesions, we have not been able to age the conditional Klf4 knockout mice beyond 6 months. Therefore, the functional role of Klf4 in esophageal carcinogenesis cannot be fully addressed in this study. Our experiments were performed on mice of a mixed genetic background. We are currently crossing Klf4^Loxp/Loxp mice and ED-L2/Cre mice to different inbred strains in an attempt to minimize the effects of this patchy deletion of Klf4 in the skin and to allow for a more systematic evaluation of the consequence of loss of Klf4 in the esophagus.

Esophageal epithelia of ED-L2/Cre;Klf4^loxP/loxP mice have a 1.9-fold increase in proliferation, and we have identified the pro-proliferative Klf5 as a novel target for Klf4 in vitro and in vivo. Klf5 has been shown to regulate esophageal epithelial proliferation in vitro and in vivo^{12, 27}, and mice with transgenic expression of Klf5 in esophagus, like ED-L2/Cre;Klf4^loxP/loxP mice, have increased proliferation confined to the basal layer of the esophagus. We believe that this is due to the normal repression of Klf5 by Klf4 in transit amplifying cells as they exit the esophageal basal layer. Moreover, Klf4 may compete with Klf5 for binding to target genes^{31, 41}. Nonetheless, transgenic expression of Klf5 alone does not alter epithelial differentiation¹², indicating that Klf4 drives differentiation through other mechanisms. Also of note, the effect Klf4 on Klf5 is directly relevant to normal homeostasis but not necessarily to carcinogenesis since Klf5 may play a different role in cancer than in normal epithelia^{40, 42}. Overall, we suggest that when Klf4 expression is turned on in the suprabasal layer of the esophagus, Klf4 represses Klf5 both transcriptionally and post-transcriptionally and at the same time competes with Klf5 for binding to the promoters of other key regulatory genes, thus switching cells from the proliferation to differentiation program (Fig. 6H).

Loss of Klf4 also results in a 2.7-fold increase in apoptosis. While we cannot exclude a direct effect of Klf4 on apoptosis, these findings are typical of efforts to maintain homeostasis in rapidly renewing epithelia, such as esophagus, intestine, and skin^{2, 29, 43}. Although these numerical changes might seem to suggest that the increase in proliferation is balanced by changes in apoptosis, proliferation is much more common in the esophagus than apoptosis, such that a nearly 3-fold increase in apoptosis does not offset the effects of a 2-fold increase in proliferation. Nonetheless, we cannot exclude that loss of Klf4 in vivo alters the rate of cell migration and that this may also influence the development of epithelial hypertrophy and hyperplasia. Of note, Klf5 has been shown to promote esophageal epithelial cell migration in vitro⁴⁴. Thus, we are currently investigating the role of Klf4 in cell migration.

Other potential Klf4 targets identified through our gene expression profiling may also be critical for Klf4 function in squamous epithelia. For example, envoplakin is a cytoskeletal linker protein and a component of the epidermal cornified envelope⁴⁵. Epidermis of envoplakin null mice has a higher proportion of immature cornified envelopes than that of control mice. Similarly, Lce1A1, Lce1A2, Lce1B, and Lce1D are major cornified envelope components induced during keratinocyte terminal differentiation⁴⁶. Also, among the genes that were differentially expressed is Slurp1, encoding a secreted member of the LY6/PLAUR family of proteins⁴⁷, which was decreased between 7- to 9-fold in ED-L2/Cre;Klf4^loxP/loxP mice. SLURP1 is a marker of late keratinocyte differentiation expressed in the granular layer of the skin⁴⁷ and, as a key ligand for the α7 nicotinic acetylcholine receptor⁴⁸, is important for terminal differentiation of epidermal keratinocytes, for homeostasis, and for the formation of the skin barrier⁴⁹. Moreover, mutations in SLURP1 are the cause of Mal de Meleda⁵⁰, a rare autosomal recessive disease characterized by transgressive palmoplantar keratoderma or hyperkeratosis of the palms and soles⁵¹. Of note, tylosis, or focal non-epidermolytic palmoplantar keratoderma, is strongly associated with esophageal squamous cell cancer⁵². Interestingly, envoplakin, has been mapped to the tylosis esophageal cancer (TOC) locus, a region on chromosome 17q25 commonly deleted in sporadic esophageal cancers⁵³. While these implications are intriguing, the roles of these genes in esophageal epithelial homeostasis, hyperproliferation, and carcinogenesis remain to be elucidated, and investigations of expression and function of these factors in esophageal epithelial cells are currently underway.

In conclusion, using a novel conditional knockout mouse model for Klf4, we demonstrate that Klf4 is required for normal homeostasis in tongue, esophagus, and skin. In the esophagus specifically, loss of Klf4 produces epithelial hypertrophy, increased proliferation, altered cell morphology with evidence of delayed cellular maturation, and eventually esophageal epithelial dysplasia by six months of age. Thus, we show here that Klf4 is necessary for differentiation and maintenance of normal homeostasis in esophageal epithelia.

Abbreviations

BrdU

5-bromo-2-deoxyuridine

Klf4

Krüppel-like factor 4

Klf5

Krüppel-like factor 5

qPCR

quantitative real-time PCR

TBP

TATA box binding protein gene

TOC

tylosis esophageal cancer