KLF4 activates NFκB signaling and esophageal epithelial inflammation via the Rho-related GTP-binding protein RHOF.

Understanding the regulatory mechanisms within esophageal epithelia is essential to gain insight into the pathogenesis of esophageal diseases, which are among the leading causes of morbidity and mortality throughout the world. The zinc-finger transcription factor Krüppel-like factor (KLF4) is implicated in a large number of cellular processes, such as proliferation, differentiation, and inflammation in esophageal epithelia. In murine esophageal epithelia, Klf4 overexpression causes chronic inflammation which is mediated by activation of NFκB signaling downstream of KLF4, and this esophageal inflammation produces epithelial hyperplasia and subsequent esophageal squamous cell cancer. Yet, while NFκB activation clearly promotes esophageal inflammation, the mechanisms by which NFκB signaling is activated in esophageal diseases are not well understood. Here, we demonstrate that the Rho-related GTP-binding protein RHOF is activated by KLF4 in esophageal keratinocytes, leading to the induction of NFκB signaling. Moreover, RHOF is required for NFκB activation by KLF4 in esophageal keratinocytes and is also important for esophageal keratinocyte proliferation and migration. Finally, we find that RHOF is upregulated in eosinophilic esophagitis, an important esophageal inflammatory disease in humans. Thus, RHOF activation of NFκB in esophageal keratinocytes provides a potentially important and clinically-relevant mechanism for esophageal inflammation and inflammation-mediated esophageal squamous cell cancer.

Introduction

Esophageal diseases are among the leading causes of morbidity and mortality in the U.S. and the world [1]. For example, esophageal cancers, of which approximately 90% are esophageal squamous cell cancer (ESCC) [2], are the 8th most common cause of cancer and the 6th leading cause of cancer-related deaths worldwide [3, 4]. Many diseases of the squamous esophagus, including ESCC and eosinophilic esophagitis (EoE), occur in the setting of chronic inflammation, and a number of these conditions have been effectively modeled in the mouse, leading to new insights into molecular pathogenesis of these diseases [5-10]. In particular, the NFκB signaling pathway has emerged as a critical activator of epithelial inflammation and inflammation-mediated carcinogenesis [11, 12], and in the esophagus, activated NFκB signaling is implicated in the development of ESCC and EoE, among other disorders [7, 13–20]. Moreover, constitutive NFκB activation in murine esophagus promotes inflammation and angiogenesis in vivo [6]. However, to date, the molecular mechanisms governing NFκB pathway activation in esophageal epithelia are not well understood.

In murine esophagus, upregulation of the DNA-binding transcription factor Krüppel-like factor 4 (KLF4) within squamous epithelial cells activates NFκB signaling, leading to chronic inflammation and inflammation-mediated ESCC [7]. KLF4 has other important cell-autonomous functions, including in proliferation and differentiation, and pro-inflammatory effects of KLF4 in esophageal epithelia are consistent with those seen in other cell-types and tissues [21-31]. For example, KLF4 promotes macrophage polarization and signaling, monocyte differentiation, cytokine expression in dendritic cells, vascular inflammation, inflammatory responses in microglia, and intestinal inflammation via NFκB signaling. Classically, the NFκB pathway is stimulated by proinflammatory cytokines or other receptor ligands, leading to the activation of IκB kinase (IKK) which then phosphorylates IκB, leading to IκB degradation and subsequent nuclear translocation of canonical NFκB members [11, 32]. Yet the mechanisms by which the transcription factor KLF4 activates NFκB signaling are not known.

Previously, we conducted a microarray analysis for genes differentially regulated in the presence and absence of Klf4 in murine esophagus and identified the Rho GTPase RhoF as a potential KLF4 target [33]. Rho GTPases can activate NFκB signaling, including in the esophagus and the skin, which like the esophagus is lined by a squamous epithelium, making Rho GTPases intriguing candidates as downstream mediators of KLF4 on NFκB and inflammation [34-38]. RHOF (also known as RIF) is a Rho family member that has been implicated in membrane trafficking, cell migration, and cytoskeletal dynamics [39-43]. Like other Rho GTPases, RHOF cycles between an active GTP-bound state and an inactive GDP-bound state, a process mediated by the interplay between guanine nucleotide exchange factors (GEFs) and the opposing GTPase activating proteins (GAPs) [39, 44, 45]. Interestingly, constitutive RhoF knockout mice have no overt phenotype, suggesting that RHOF may be dispensable for cellular function in vivo under normal conditions [46]. A role for RHOF in inflammation has not previously been reported.

Here we show that RHOF is upregulated by KLF4 and that RHOF promotes inflammation and is required for induction of NFκB signaling by KLF4 in esophageal keratinocytes. In addition, we find that RHOF is increased in a human esophageal inflammatory disease. Thus, we demonstrate that RHOF as an important mediator of esophageal inflammation.

Materials and methods

Immunohistochemistry

Murine primary esophageal keratinocytes isolated from ED-L2/Klf4 mice or ED-L2/Cre;Klf4 mice and age-matched controls, both male and female, were used for experiments [7, 33]. Mice were housed in the barrier facility at the University of Pennsylvania and given ad libitum access to food and water. All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania (IACUC) under protocol #803502. Human subjects were enrolled at the time of diagnostic esophagogastroduodenoscopy at the Hospital of the University of Pennsylvania (IRB #813363). Inclusion criteria for initial recruitment included no other esophageal or chronic inflammatory disease of the gastrointestinal tract. EoE subjects were diagnosed based on 2011 clinical guidelines [47]. Non-EoE subjects are comprised of patients who reported symptoms warranting upper endoscopy, did not carry a previous diagnosis of EoE and demonstrated no histopathologic abnormalities. For immunohistochemistry, human and mouse tissues or organotypic cultures were processed using standard protocols described elsewhere [33, 48]. Paraffin embedded slides were stained using rabbit anti-KLF4 [49] at 1:2,500 dilution and rabbit anti-RHOF antibodies (LS Bio,Lot# 66566, SL C353833) at 1:300 dilution.

Western blots

Western blots were performed as described previously [33, 50]. Briefly, cells were lysed with Triton Lysis buffer, and protease and phosphatase inhibitors (Sigma-Aldrich) were added. From each sample, 15 μg of protein was separated on a NuPAGE 4–12% acrylamide gel (Invitrogen). The following antibodies were used for Western blotting: rabbit anti-KLF4 antibody [49] at 1:10,000 dilution; rabbit anti-RHOF antibody (LSBio LS-C353833) at 1:500 dilution; rabbit anti-phospho-p65 (Cell Signaling, S536) at 1:1,000 dilution; rabbit anti-p65 (Cell Signaling, C22B4) at 1:1,000 dilution; mouse anti-IKK2 (Cell Signaling) at 1:1,000 dilution; mouse anti-β-actin at 1:10,000 dilution; rabbit anti-GAPDH (Cell Signaling) at 1:10,000 dilution; and mouse anti-α-tubulin at 1:15,000 dilution as described previously [50].

RNA analyses and real-time PCR

Total RNA was isolated with the RNeasy Micro Kit (Qiagen) following manufacturer’s protocol, and cDNA was synthesized with the Maxima First Strand cDNA Synthesis Kit for RT-qPCR, with dsDNase (ThermoFisher). Quantitative real-time polymerase chain reaction (qPCR) was performed using SYBR Green Master Mix (ThermoFisher) as described [50]. Relative mRNA expression levels were normalized by GAPDH. The following primer were used to amplify specific target genes: RHOF (human F 5’-AGCAAGGAGGTGACCCTGAAA-3’, R 5’-CCGCAGCCGGTCATAGTC-3’; mouse F 5’-ACTGCTCCTTGTCCTTCCTCA-3’, R 5’-CGACAACGTCCTCATCAAAGTG-3’);

KLF4 (human F 5’-GCGGCAAAACCTACACAAAG-3’, R 5’-CCCCGTGTGTTTACGGTAGT-3’; mouse F 5’-GTGCCCCGACTAACCGTTG-3’ R 5’-GTCGTTGAACTCCTCGGTTCT-3’);

IKK1 (human F 5’-CTCCGAGACTTTCGAGGAAATAC-3’, R 5’-GCCATTGTAGTTGGTAGCCTTCA-3’; mouse F 5’-GTCTCGGAATTGAGCGTGAAA-3’, R 5’-TCCCTGTCTCTGACAGAAGCTCCTGA-3’); IKK2 (mouse F 5’-TCTAAATGGCCTTTTCCTGCTAAT-3’, R 5’-TGACTCTCCCAAAGTTAGATGCA-3’);

IKK3 (mouse F 5’-CTGGAAGATCTGAGGCAACA-3’, R 5’-CCAGGGCCTCCTCAGCTTGC-3’); GAPDH (human F 5’-GAAGGTGAAGGTCGGAGTCA-3’ R 5’-AATGAAGGGGTCATTGATGG-3’; mouse F 5’-CGGCCGCATCTTCTTGTG-3’ R 5’-ACCGACCTTCACCATTTTGTCT-3’.

Cell culture and treatment

Murine primary esophageal keratinocytes were grown as described [7, 33, 50]. Established primary human esophageal keratinocytes (EPC2-hTERT cells) [51] were grown in keratinocyte serum-free medium (K-SFM, Invitrogen) supplemented with bovine pituitary extract and epidermal growth factor (Invitrogen) [48]. Cells were infected and transfected as described [48]. For cytokine studies, cells were stimulated with 10 ng/ml IL-4, IL-13, or TGFβ (R&D Systems) for 24 hours.

shRNAs and expression plasmids and transfection

Full-length constitutively active RhoF (myc Rif-QL) [43] was a gift from Harry Mellor (Addgene plasmid #38768) and was overexpressed in mouse epithelial cells using lentiviral vectors psPAX2 (gift from Didier Trono, Addgene plasmid #12260) and pMD2.G (gift from Didier Trono, Addgene plasmid #12259. Two distinct shRNAs against RhoF (Sigma, TRCN0000077678 and TRCN0000447186) were used for knockdown experiments. FuGENE 6 (Promega) was used for lentiviral generation.

RHOF activation assay

RHOF activation assays were performed using protocols described elsewhere [42, 52]. The GST-MDia1 plasmid (gift of Harry Mellor) was used to generate GST-MDia1 protein, which was used as a probe to specifically isolate the active form of RHOF. Esophageal keratinocytes from ED-L2/Klf4 mice were seeded on 100 mm culture dishes and lysed. RHOF protein was pulled down using GST beads, and the beads were washed three times with washing buffer. Activated RHOF bound to the beads or total RHOF in cell lysates was detected by Western blot using rabbit anti-RHOF antibody (LSBio) at 1:1,000 dilution.

Organotypic culture

Organotypic culture was performed as previously described [48, 53]. Human esophageal keratinocytes were seeded on a collagen fibroblast layer with and without human peripheral blood mononuclear cells (PBMC), which were stimulated by inflammatory cytokines (IL-2, IL-7, and IL-15; Human Immunology Core, University of Pennsylvania) as described [54]. To model a TH1 inflammatory environment, the pro-inflammatory cytokines IL-7 (10 ng/mL; Cell Signaling) and IL-15 (20 ng/mL; Prospec-Tany Technogene) were included in cell culture media, and IL-2 (10 U/mL, BD Biosciences) was added to support PBMC viability. Organotypic cultures were processed and stained according to standard protocols [48].

Time-lapse video microscopy

Single cells were plated as described previously and cultured on 8-well chamber slides [55, 56]. Cells were kept at 37°C and 5% CO2 for the duration of the 24 hour time-lapse recording. Serial phase-contrast images were captured at 10 minute intervals and built into a movie using MetaMorph software (Molecular Devices). Approximately 10 cells per field for a total of three fields per sample were highlighted and their movements followed over the 24 hour period. Distance from the origin was computed using MetaMorph as previously described [55, 57]

Statistical analysis

Statistical significance between groups of data was calculated by Student's t test, using Prism 4 (GraphPad), and differences were considered significant for p <0.05.

Results

KLF4 activates NFkB signaling in esophageal keratinocytes

Initially, we sought to validate that KLF4 activates NFκB in esophageal epithelial cells. Consistent with our prior work [7], we found that primary murine esophageal keratinocytes from ED-L2/Klf4 mice, which overexpress Klf4 in esophageal epithelial cells, had increased expression of the kinases Ikk1, Ikk2, and Ikk3, which activate NFκB signaling, compared to control cells (Fig 1A), and in primary murine esophageal keratinocytes from mice with genetic ablation of Klf4 (ED-L2/Cre;Klf4 mice), Ikk1, Ikk2, and Ikk3 expression was decreased (Fig 1B). Moreover, NFκB signaling was also decreased as demonstrated by reduction in phosphorylation of the p65 subunit of NFκB, in primary murine esophageal keratinocytes from ED-L2/Cre;Klf4 mice (Fig 1C). Klf4 overexpression in mice activates NFκB [7], and KLF4 knockdown in primary human esophageal keratinocytes reduced p65 phosphorylation (Fig 1D). Thus, KLF4 upregulates Ikk1, Ikk2, and Ikk3 transcription to activate pro-inflammatory NFκB signaling in esophageal epithelial cells, and KLF4 loss decreases NFκB pathway activation. However, the mechanism of this NFκB pathway activation is not known.

Fig 1

KLF4 activates NFκB signaling in esophageal epithelial cells.

(A) By qPCR, Ikk1, Ikk2, and Ikk3 expression was increased in primary esophageal keratinocytes from mice overexpressing Klf4 (ED-L2/Klf4 mice) compared to keratinocytes from control mice. (B) By qPCR, expression of Ikk1, Ikk2, and Ikk3 was reduced in primary esophageal keratinocytes from mice with loss of Klf4 (ED-L2/Cre;Klf4 mice) compared to control keratinocytes from Klf4 mice without Cre. (C) Compared to control keratinocytes from Klf4 mice without Cre, primary esophageal keratinocytes from ED-L2/Cre;Klf4 mice had less phosphorylated p65 on Western blots but no change in total p65. (D). Compared to a scrambled shRNA control (shSCR), shRNA directed against KLF4 reduced p65 phosphorylation in primary human esophageal keratinocytes, as seen on Western blot.

KLF4 increases expression and activity of the small GTPase RHOF

To define the factors in esophageal epithelial cells that mediate NFκB pathway activation by KLF4, we examined a list of candidate KLF4 targets previously identified by microarray analysis [33], focusing on those with potential functions in NFκB activation. Certain Rho GTPases can activate NFκB signaling [34-38], and both RhoF (also known as Rif) and Argef17, which encodes RhoGEF17, a Rho-specific guanine nucleotide exchange factor (GEF) [58], are differentially expressed by microarray in esophageal epithelia of mice with Klf4 deletion [33]. RHOF functions in cytoskeletal remodeling but has not previously been implicated in inflammatory signaling [37, 41, 59]. Nonetheless, we postulated that RHOF might mediate NFκB pathway activation by KLF4. By qPCR, we demonstrated that RhoF and Argef17 were downregulated in esophageal epithelial cells from mice with Klf4 loss (Fig 2A) and increased in esophageal epithelial cells from with Klf4 overexpression (Fig 2B). Rho factors cycle between an active GTP-bound state and an inactive GDP-bound conformation [44], and to determine whether upregulation of RhoF by KLF4 also results in an increase in activated RHOF, we determined the levels of total and GTP-bound RHOF using a GTP pull-down assay [52] in esophageal epithelial cells from control mice and mice with esophageal epithelial Klf4 overexpression. We found that Klf4 overexpression results not only in more total RHOF but also in a dramatic increase in activated RHOF (Fig 2C). Using ED-L2/Klf4 mice, which have epithelial-specific Klf4 overexpression [7], we demonstrated that KLF4 also regulates RHOF expression in vivo, as RHOF is increased specifically within esophageal epithelia of ED-L2/Klf4 mice (Fig 2D). Thus, KLF4 upregulates RHOF in esophageal epithelia, resulting in increased levels of activated RHOF.

Fig 2

KLF4 increases RhoF expression and activity in esophageal epithelial cells.

(A) By qPCR, Klf4, RhoF, and the guanine-exchange factor Arhgef17 were significantly decreased in esophageal epithelial cells from mice with Klf4 deletion, compared to cells from control mice. (B) When Klf4 was increased in esophageal epithelial cells from mice with Klf4 overexpression, RhoF and Arhgef17 were increased on qPCR, compared to control cells. (C) Esophageal epithelial cells from mice with Klf4 overexpression also demonstrated an increase in the amount of activated RHOF, indicated on a RHOF activation assay. (D) Staining for RHOF (brown) was low in esophageal epithelia of control mice and increased markedly in mice with Klf4 overexpression. Magnification = 100x.

RHOF is upregulated in inflammation and regulates expression of pro-inflammatory cytokines

To define the function of RHOF in esophageal epithelial cells, we initially examined the effects of RhoF knockdown on the actin cytoskeleton and esophageal epithelial cell migration, since RHOF is implicated in cytoskeletal remodeling [41] and Rho GTPases are critical for cell migration [60]. In esophageal epithelial cells, expression of constitutively active RHOF promoted actin remodeling and significantly increased single cell migration (S1 Fig). To define the role of RHOF in esophageal mucosal inflammation, we first knocked down RhoF in esophageal epithelial cells using shRNA and examined the consequences on pro-inflammatory genes. RhoF knockdown significantly decreased expression of Ikk1, Ikk2, and Ikk3 (Fig 3A), consistent with a function for RHOF in mediating KLF4 activation of NFκB signaling. In addition, RhoF knockdown reduced expression of the key pro-inflammatory genes TNFα, IL-1α, CXCL5, and G-CSF (Fig 3B), each of which is also upregulated in esophageal epithelial cells with Klf4 overexpression [7]. Finally, RHOF was upregulated in a model of esophageal inflammation (Fig 3C), in which PBMCs stimulated with interleukins are co-cultured with esophageal epithelial cells in organotypic culture [53]. Thus, RHOF is induced in esophageal epithelia during inflammation and activates NFκB signaling to promote inflammation.

Fig 3

RHOF is upregulated in inflammation and activates pro-inflammatory cytokines.

(A) By qPCR, expression of the NFκB activators Ikk1, Ikk2, and Ikk3 decreased when RhoF was knocked down by shRNA in primary esophageal epithelial cells from wild-type mice, compared to similar cells infected with scrambled controls (shScr). (B) Compared to cells with shScr, primary mouse esophageal epithelial cells with shRNA against RhoF also had reduced expression of pro-inflammatory TNFα, IL-1α, CXCL5, and G-CSF. (C) When human primary esophageal epithelial cells were grown in organotypic culture with PBMCs that were stimulated with IL-2, IL-7, and IL-15, RHOF levels increased markedly within epithelial cells, compared to cells grown without PBMCs.

RHOF mediates NFκB pathway activation by KLF4

To delineate whether RHOF was required for KLF4 induction of NFκB signaling in esophageal epithelial cells, we infected primary esophageal epithelial cells from mice with Klf4 overexpression to express shRNA directed against RhoF or a scrambled shRNA control. Compared to control esophageal epithelial cells (with endogenous Klf4 expression) and ED-L2/Klf4 cells (that overexpress Klf4), ED-L2/Klf4 cells with RhoF knockdown had dramatically decreased NFκB activation (Fig 4A) and decreased expression of Ikk1 and Ikk2 (Fig 4B). Based on these findings, we propose a model (Fig 4C) in which KLF4 acts via RHOF to induce NFκB signaling, leading to esophageal epithelial inflammation and esophageal squamous cell cancer [7].

Fig 4

Knockdown of RhoF prevents KLF4-mediated activation NFκB.

(A) By Western blot, RhoF knockdown with shRNA blocked p65 phosphorylation in primary esophageal epithelial cells from mice with Klf4 overexpression. (B) By qPCR, RhoF knockdown in mouse esophageal epithelial cells with Klf4 overexpression reduced Ikk1 and Ikk2 expression (*p≤0.05). (C) KLF4 upregulates RHOF and increases RHOF activation in esophageal epithelia, leading to NFκB activation, inflammation, and cancer.

RHOF is upregulated in human eosinophilic esophagitis

To determine whether RHOF might have a role in human esophageal inflammation, we examined the expression of RHOF in an in vitro model of the human inflammatory disease eosinophilic esophagitis (EoE) and in human EoE samples. When primary human esophageal epithelial cells were treated with IL-4, IL-13, or TGFβ, cytokines that are upregulated in human EoE [61, 62], RHOF increased significantly, including a nearly two-fold increase following TNFβ treatment (Fig 5A). In addition, RHOF was found at much higher levels in esophageal epithelia from humans with EoE, compared to controls (Fig 5B). Thus, elevated RHOF expression is observed in human EoE and in vitro EoE models.

Fig 5

RHOF is upregulated in human EoE.

(A) Primary human esophageal epithelial cells stimulated with IL-4, IL-13, or TGF-β, cytokines that are physiologically relevant for human EoE, had significant increases in RHOF expression by qPCR, compared to unstimulated cells. (B) By immunohistochemistry, RHOF levels increased dramatically in EoE compared to normal esophageal epithelia.

Discussion

In the esophagus, activation of proinflammatory pathways within esophageal squamous epithelial cells can promote inflammation throughout the mucosa, providing a microenvironment favorable for the development of ESCC [5-8]. Previously, we demonstrated that transgenic overexpression of Klf4 within esophageal keratinocytes activates NFκB signaling, which is associated with the development numerous inflammatory diseases and cancers [11-20], resulting in inflammation-mediated ESCC [7]. We also showed that, consistent with this, activation of NFκB signaling within esophageal keratinocytes by transgenic Ikkβ expression promotes inflammation and angiogenesis, features of inflammatory diseases and the tumor microenvironment [5, 63, 64]. The Rho GTPases interact with the NFκB pathway and are involved in the pathogenesis of a number of human cancers and other inflammatory diseases [35]. Here, we link the Rho family member RHOF to KLF4-mediated NFκB activation in esophageal keratinocytes and to the development of inflammation and a human esophageal inflammatory disease, EoE.

Tumor-promoting inflammation is an “enabling characteristic” of cancers, including ESCC [5, 63], and to date, a number of important murine models for inflammation-mediated ESCC have been developed, including mice with Klf4 overexpression, p120 catenin knockout, or conditional Sox2 knockout [7, 8, 65]. Both Klf4 overexpressing mice and p120 catenin knockout have robust NFκB activation that is an early event, and in Sox2 knockout mice, tumor progression correlates with inflammation. Interestingly, p120 deletion in murine esophagus and epidermis results in inflammation, hyperproliferation, and squamous cell cancer that appear to be mediated by aberrant activation of RHOA upstream of NFκB [36, 38, 66], and Rho GTPases also function in immune cell migration and the tumor microenvironment [67, 68]. Yet similar functions for RHOF in inflammation and tumorigenesis have not previously been reported. Of note, global RhoF deletion in mice has no overt phenotype, raising the possibility that other Rho family members may compensate for RHOF function in vivo under normal conditions [46].

Interestingly, KLF4 enhances RHOF protein activation, seemingly to a greater extent than can be explained through KLF4 upregulation of RhoF expression alone. The mechanism for RHOF activation by KLF4 is not known but may be related to effects of KLF4 on the GEFs and GAPs, the interplay between which regulates the activity of Rho GTPases [37, 39, 44, 45]. In fact, KLF4 does upregulate one of these factors, Arhgef17, and increased levels of RhoGEF17 would be expected to increase Rho factor activation. shRNA knockdown of RhoF results in a dramatic decrease in phosphorylated p65, some of which is likely related to the effects of KLF4, both endogenous and transgenic, on NFκB signaling, although these data also raise the intriguing possibility that RHOF might promote esophageal inflammation and disease independent of KLF4. Taken together, we conclude that RHOF activates NFκB signaling and esophageal epithelial inflammation, and thus RHOF, and potential activators of RHOF such as RhoGEF17, could emerge as therapeutic targets for inflammatory diseases.

RHOF promotes actin reorganization and single cell migration of esophageal epithelial cells.

(A) Compared to control cells, esophageal epithelial cells that expressed constitutively active RhoF (green) had reorganization of F-actin to the cell surface with small actin-rich surface projections as indicated by staining for phalloidin (red). DAPI staining is in blue. (B) Expression of constitutively active RhoF also increased single-cell migration of esophageal epithelial cells, as assessed by time-lapse microscopy.

(EPS)

Click here for additional data file.

Completed ARRIVE Guidelines Checklist.

(PDF)

Click here for additional data file.