Article Text

Download PDFPDF

Original article
ZEB1 promotes inflammation and progression towards inflammation-driven carcinoma through repression of the DNA repair glycosylase MPG in epithelial cells
  1. Oriol de Barrios1,
  2. Lidia Sanchez-Moral1,
  3. Marlies Cortés1,
  4. Chiara Ninfali1,
  5. Nuria Profitós-Pelejà1,
  6. MC Martínez-Campanario1,
  7. Laura Siles1,
  8. Rosa del Campo2,
  9. María Jesús Fernández-Aceñero3,
  10. Douglas S Darling4,
  11. Antoni Castells5,6,
  12. Joan Maurel7,
  13. Azucena Salas5,
  14. Douglas C Dean8,9,
  15. Antonio Postigo1,9,10
  1. 1 Group of Transcriptional Regulation of Gene Expression, Dept of Oncology and Hematology, IDIBAPS, Barcelona, Spain
  2. 2 Dept of Microbiology, Hospital Ramon y Cajal Health Research Institute (IRYCIS), Spanish Network of Infectious Diseases (REIPI), National Health Institute Carlos III (ISCIII), Madrid, Spain
  3. 3 Dept of Pathology, Hospital Gregorio Marañón, Madrid, Spain
  4. 4 Dept of Oral Immunology and Infectious Diseases and Center for Genetics and Molecular Medicine, University of Louisville, Louisville, Kentucky, USA
  5. 5 Dept of Gastroenterology, Hospital Clínic and IDIBAPS, Barcelona, Spain
  6. 6 Gastrointestinal and Pancreatic Oncology Team, CIBERehd, Barcelona, Spain
  7. 7 Group of Translational Genomics and Targeted Therapeutics in Solid Tumours, Dept of Medical Oncology, Hospital Clínic and IDIBAPS, Barcelona, Spain
  8. 8 Dept of Ophthalmology and Visual Sciences and Birth Defects Center, University of Louisville, Louisville, Kentucky, USA
  9. 9 Molecular Targets Program, James G. Brown Cancer Center, Louisville, Kentucky, USA
  10. 10 ICREA, Barcelona, Spain
  1. Correspondence to Dr Douglas C Dean, James G Brown Cancer Center, University of Louisville, KY 40202, USA; douglas.dean{at}louisville.edu and Dr Antonio Postigo, Group of Transcriptional Regulation of Gene Expression. IDIBAPS. 08036 Barcelona, Spain; idib412{at}clinic.cat

Abstract

Objective Chronic inflammation is a risk factor in colorectal cancer (CRC) and reactive oxygen species (ROS) released by the inflamed stroma elicit DNA damage in epithelial cells. We sought to identify new drivers of ulcerative colitis (UC) and inflammatory CRC.

Design The study uses samples from patients with UC, mouse models of colitis and CRC and mice deficient for the epithelial-to-mesenchymal transition factor ZEB1 and the DNA repair glycosylase N-methyl-purine glycosylase (MPG). Samples were analysed by immunostaining, qRT-PCR, chromatin immunoprecipitation assays, microbiota next-generation sequencing and ROS determination.

Results ZEB1 was induced in the colonic epithelium of UC and of mouse models of colitis. Compared with wild-type counterparts, Zeb1-deficient mice were partially protected from experimental colitis and, in a model of inflammatory CRC, they developed fewer tumours and exhibited lower levels of DNA damage (8-oxo-dG) and higher expression of MPG. Knockdown of ZEB1 in CRC cells inhibited 8-oxo-dG induction by oxidative stress (H2O2) and inflammatory cytokines (interleukin (IL)1β). ZEB1 bound directly to the MPG promoter whose expression inhibited. This molecular mechanism was validated at the genetic level and the crossing of Zeb1-deficient and Mpg-deficient mice reverted the reduced inflammation and tumourigenesis in the former. ZEB1 expression in CRC cells induced ROS and IL1β production by macrophages that, in turn, lowered MPG in CRC cells thus amplifying a positive loop between both cells to promote DNA damage and inhibit DNA repair.

Conclusions ZEB1 promotes colitis and inflammatory CRC through the inhibition of MPG in epithelial cells, thus offering new therapeutic strategies to modulate inflammation and inflammatory cancer.

  • colorectal cancer
  • inflammatory bowel disease
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • Patients with ulcerative colitis (UC) are at a higher risk of developing inflammatory-associated colorectal cancer (CRC).

  • Inflammatory cytokines and reactive oxygen species (ROS) produced by the inflamed microenvironment induce DNA damage in cancer cells and inhibit its repair.

  • Inflammation in the tumour stroma induces an epithelial-to-mesenchymal transition (EMT) of nearby cancer cells.

  • The transcription factor ZEB1 is a key driver of EMT and its expression in CRC determines a poorer prognosis.

  • However, the role of ZEB1 in inflammation and inflammation-induced cancer has not been studied.

Significance of this study

What are the new findings?

  • ZEB1 is not expressed by epithelial cells in the healthy colon, but it was induced in the colonic epithelium of patients with UC and of a mouse model of colitis. Next-generation sequencing revealed similar gut microbiota composition in wild-type and Zeb1-deficient mice.

  • In an experimental model of colitis, Zeb1-deficient mice exhibited lower inflammation.  Downregulation of ZEB1 in CRC cells reduced DNA damage in response to oxidative stress and inflammatory cytokines.  In an experimental model of inflammatory CRC, the tumours developed in Zeb1-deficient mice exhibited lower DNA damage and higher levels of the DNA repair glycosylase N-methyl-purine glycosylase (MPG).

  • ZEB1 inhibits MPG expression through direct binding to its promoter.  Genetic ablation of Mpg in Zeb1-deficient mice reverted their reduced colitis and inflammatory-mediated tumourigenesis.

  • ZEB1 expression in CRC cells induces macrophage production of ROS and interleukin1β with the latter downregulating MPG in CRC cells, thus amplifying a positive loop between both cells to promote DNA damage and inhibit DNA repair.

How might it impact on clinical practice in the foreseeable future?

  • The study uncovers ZEB1 as an important driver of colitis and inflammatory CRC and opens the way for new therapeutic approaches to modulate inflammation and inflammatory cancer in patients with UC.

Introduction

Chronic inflammation is a risk factor for many types of cancers.1 In turn, activation of certain oncogenic pathways triggers an inflammatory and immune-suppressed tumour microenvironment.2 Inflammatory cytokines and reactive oxygen (ROS) and nitrogen (RNS) species released by neutrophils and macrophages both induce DNA alterations in epithelial cells and inhibit their repair.3–5

Base modifications can be repaired by MGMT and the enzymes of the base excision repair (BER) system.6 7 The first step in the BER system involves the recognition and excision of damaged bases by a group of DNA glycosylases, including N-methyl-purine glycosylase (MPG) (also known as N-alkyladenine glycosylase), 8-oxoguanine-DNA glycosylase (OGG1) and mutY DNA glycosylase (MUTYH).5–7 Loss-of-function mutations in humans or targeted deletions in mice of some of these glycosylases result in greater susceptibility to DNA damage, inflammation and spontaneous and/or induced tumourigenesis. For instance, compared with wild-type counterparts, Mpg (−/−) mice accumulate more DNA lesions in their colonic mucosa after chemically induced colitis.8 Mpg (−/−), Ogg1 (−/−) and Mutyh (−/−) mice all exhibit enhanced tumourigenesis in experimental models of inflammatory colorectal cancer (CRC).8–11

Efficient DNA repair requires a tight regulation and balance of all BER components.5 7 12 13 Thus, mice lacking some BER glycosylases paradoxically display increased resistance to certain alkylating agents,7 8 while their overexpression induces frameshift mutations and microsatellite instability, and associates with poorer prognosis in several cancers.6 12 14

CRC is one of the cancers where the link with inflammation has been more extensively studied.15 Patients with IBD, particularly ulcerative colitis (UC), are at a higher risk of developing CRC, which is directly correlated with the cumulative chronic inflammatory burden.16–18 The colonic mucosa of patients with UC and of experimental mouse models of colitis exhibits an array of DNA base lesions including 7,8-dihydro-8-oxo-2’-deoxyguanosine (8-oxo-dG) and εA, εC, εG adducts.8 11 19–21 Alterations in the gut microbiota have been involved in the pathogenesis of UC and inflammatory CRC through DNA damage-dependent and independent mechanisms.22

In many carcinomas, including CRCs, some malignant cells undergo a dedifferentiation gene reprogramming—as part of the process referred as epithelial-to-mesenchymal transition (EMT)—that endows them with stem-like properties that are important in cancer initiation and progression.23 24 Inflammation in the tumour stroma contributes to the EMT of nearby cancer cells and promotes their metastatic dissemination.25 EMT is driven by transcription factors of the ZEB, Snail and Twist families with ZEB1 (also known as ZEB and δEF1) often operating downstream of Snail and Twist.23 26 Expression of ZEB1 in CRC associates with increased cancer cell invasiveness and determines a worse clinical prognosis.23 27 28 ZEB1’s functions depend on a fine threshold of its expression and the deletion of a single Zeb1 allele—as in the Zeb1 (+/−) mouse—is sufficient to trigger developmental defects, increase the susceptibility to some muscular pathologies and, importantly, to inhibit tumour progression in mice.23 27–32 Although inflammatory mediators upregulate ZEB1 expression in cancer cell lines,33 34 its role and mechanism of action in inflammation-induced cancer is not fully understood.

Using human samples of UC, Zeb1 (+/−) and Mpg (−/−) mice—Zeb1 (−/−) mice die perinatally—and mouse models of colitis and inflammatory CRC, we found that ZEB1 promotes both intestinal inflammation and inflammatory CRC through repression of the MPG glycosylase. ZEB1 is not expressed in the epithelial cells of a healthy colon,27 but it was induced in the colonic epithelium of human patients with UC and of mice subjected to an experimental model of colitis for which Zeb1-deficient mice were partially protected. We found that the DNA base lesions (8-oxo-dG) induced in CRC cells by oxidative stress and inflammatory cytokines are mediated, at least in part, by ZEB1 expression. In a mouse model of inflammatory CRC, and compared with wild-type counterparts, Zeb1-deficient mice not only exhibited fewer tumours but also lower levels of 8-oxo-dG and higher levels of MPG. ZEB1 inhibited MPG expression and bound directly to the Mpg promoter, interaction that was enhanced by inflammatory stimuli. This mechanism of action was validated in vivo at the genetic level as the crossing of Zeb1 (+/−) and Mpg (−/−) mice reverted the reduced inflammation and tumourigenesis found in the former to the same levels than those in wild-type counterparts.

These data set ZEB1 as an important driver of intestinal inflammation and inflammatory CRC and uncover a new mechanism of action for ZEB1 in cancer.

Methods

Human samples

Human colon samples were obtained from biopsies collected during routine endoscopy from patients with UC and healthy controls recruited at Hospital Clínic (Barcelona, Spain) and Hospital Clínico San Carlos (Madrid, Spain). Healthy controls were individuals undergoing colonoscopy but that did not present lesions. All human samples were obtained with the informed consent of patients and conformed to the principles of the Helsinki Declaration.

Mouse samples

The following mouse models were examined in the study: CL57BL/6 (referred as wild-type or Zeb1 (+/+) mice) (Jackson Laboratories, Bar Harbor, Maine, USA), Zeb1 (+/−)35 and Mpg (−/−) (obtained from Dr L Samson, MIT, Cambridge, Massachusetts, USA).36

Cell lines and cell culture

SW480 and HCT116 CRC cells were obtained from the Cancer Cell Line Repository at the Barcelona Biomedical Research Park (Barcelona, Spain) and cultured as described elsewhere.27

Antibodies, plasmids, and DNA and RNA oligonucleotides

The identity and source of the antibodies, plasmids and DNA and RNA oligonucleotides used in the study are described in online supplementary information.

Protein and RNA expression, and chromatin immunoprecipitation assays

Protein and RNA expression were assessed as described in online supplementary information. Analysis of ZEB1 binding to DNA by chromatin immunoprecipitation assays were performed as detailed in online supplementary information.

Gut microbiota composition

Analysis of the gut microbiota composition in mice was determined as indicated in online supplementary information.

Statistical analysis

Statistical analysis of the data shown in this study was performed using Prism for Mac (GraphPad Software, La Jolla, California, USA) except for gut microbiota distribution where R software (https://www.r-project.org/) was used. Statistical significance was assessed with non-parametric Mann-Whitney U test. Histograms throughout the manuscript are the mean with SE. Where appropriate, relevant comparisons were labelled as either significant at the p≤0.001 (***), p≤0.01 (**) or p≤0.05 (*) levels, or non-significant for values of p>0.05.

Results

ZEB1 is induced in epithelial cells during intestinal inflammation

To investigate a possible role of ZEB1 in intestinal inflammation, we first examined its expression in a gene microarray of UC.37 Compared with healthy controls, ZEB1 was upregulated in active stages of UC and declined during remission periods, although it still remained above the levels found in healthy colon (figure 1A). As controls, we also examined IL1B and COL1A2—whose expression increase during active UC and return to basal levels in remission—and REG4—whose levels increase in active stages but remain elevated even after the inflammation has been resolved.37 The upregulation of ZEB1 in colonic samples from patients with UC was validated by real-time PCR (figure 1B).

Figure 1

ZEB1 is upregulated in the colonic epithelium of patients with UC. (A) ZEB1 increases in active stages (outbreaks) of UC and declines during the inactive periods (remission). Heat map of ZEB1, SNAI1, TWIST1, COL1A2, REG4 and IL1B expression in the colonic mucosa of 13 healthy individuals, 15 patients with UC in active stage and 8 patients with UC in remission from the gene expression array GSE38713.37 (B) ZEB1, COL1A2 and REG4 mRNA levels in five colonic samples obtained from healthy individuals, and patients with UC in active and remission stages were assessed by qRT-PCR with respect to ACTB. (C) The expression of miR-200a, miR-200b and miR-200c decreases in active stages of UC and increases in the remission periods. Heat map of the expression of the three microRNAs in the colonic mucosa of 10 healthy individuals, 10 patients with active UC (outbreak) and 7 patients with inactive UC (remission) from the gene expression array GSE48957.38 (D) Representative colonic samples from healthy donors and patients with UC were stained for ZEB1 (clone H-102). Scale bar: 10 µm. (E) Quantification of ZEB1 in epithelial and stromal cells of colonic samples from healthy donors and patients with UC in (D). *P≤0.05, **p≤0.01, ***p≤0.001. ns, non-significant.

Immunohistochemistry analysis of colitis samples in rats and humans found no expression of Twist while that of Snail is restricted to stromal cells.38 Our analysis of SNAI1 and TWIST1 in the gene expression array of UC did not find a clear association with inflammation in the case of the former and with fewer cases than for ZEB1 in the latter (figure 1A).

ZEB1 represses and is repressed by microRNAs of the miR-200 family.23 24 Accordingly, in a microRNA array of UC,39 we found that miR-200a, miR-200b and miR-200c were downregulated during active stages of UC in a reverse pattern with respect to ZEB1 (figure 1C).

Next, we examined by immunostaining the cellular compartment that accounted for the upregulation of ZEB1 in UC. As a dedifferentiation/EMT marker, ZEB1 is not expressed by intestinal epithelial cells and in a healthy human colon or the large intestine of wild-type mice, ZEB1 is restricted to scattered cells in the stroma.23 27 ZEB1 has been only described within the intestinal epithelial compartment in dedifferentiated malignant cells in CRCs.23 27 40 Notably, we found here that ZEB1 is induced in the majority of epithelial cells in the colonic mucosa of patients with UC (figure 1D and E). The number of ZEB1+ stromal cells also increased in UC samples, although only slightly (figure 1D and E).

The expression of ZEB1 in the inflamed intestinal epithelium was also examined in an experimental mouse model of colitis induced by dextran sodium sulfate (DSS).41 In parallel with the results in human UC, ZEB1 mRNA was upregulated in the mouse intestinal mucosa in response to DSS (figure 2A) with ZEB1 protein induced in most epithelial intestinal cells (figure 2B and C, and supplementary figure S2A). Altogether, these data indicate that ZEB1 is induced in epithelial cells during human and mouse intestinal inflammation.

Figure 2

ZEB1 is induced and required for intestinal inflammation in a mouse model of colitis. (A) Zeb1 increases in a mouse model of colitis. Wild-type mice were either untreated (exposed only to plain drink water) or subjected to 3% DSS in the drinking water as described in online supplementary information. At the end of the protocol, colon samples were assessed for Zeb1 levels by qRT-PCR with respect to Actb as reference gene. Five mice were examined for each condition. (B) As in (A), intestinal samples from wild-type mice exposed to plain water or to 3% DSS were stained for ZEB1 (H-102). Scale bars in upper and lower panels: 50 and 10 µm, respectively. See online supplementary figure S2A for lower magnification captures. (C) Quantification of ZEB1 in epithelial and stromal cells of colonic samples from wild-type mice either untreated or exposed to 3% DSS in (B). (D) ZEB1 promotes and is required for DSS-induced colitis. Wild-type and Zeb1 (+/−) mice subjected to DSS-induced colitis as in (A) were assessed for colon length shortening (left panel) and total body weight loss (right panel). Both parameters were calculated with respect to untreated counterparts of the same genotype. Sixteen mice were analysed for each genotype. (E) Compared with DSS-treated wild-type counterparts, the intestine of Zeb1 (+/−) mice retained a higher number of colonic crypts. A representative colon sample of each genotype was stained for H&E. Scale bar: 100 µm. (F) Colon samples of mice from both genotypes treated with 3% DSS were assessed by qRT-PCR for the expression of the indicated pro-inflammatory markers using Actb as reference gene. *P≤0.05, **p≤0.01, ***p≤0.001. ns,  non-significant.

ZEB1 promotes intestinal inflammation

To investigate whether ZEB1 plays a pathogenic role in intestinal inflammation, wild-type and Zeb1 (+/−) mice were treated with DSS and the severity of inflammation was assessed in both genotypes by macroscopic, histological and gene expression parameters. At the macroscopic level, intestinal inflammation is manifested inter alia by colon length shortening, ulceration, faecal bleeding and loss of body weight. We found that DSS-induced colitis was more intense in wild-type mice; the colon shortening and body weight loss in wild-type mice doubled those observed in Zeb1 (+/−) counterparts (figure 2D and online supplementary figure S2B). At the histological level, the intestinal mucosa of DSS-treated mice from both genotypes exhibited loss of the normal epithelial architecture. However, inflammation was less severe in the intestine of Zeb1 (+/−) mice that retained a higher number of colonic crypts (figure 2E). Compared with wild-type counterparts, the inflammatory infiltration and the expression of inflammatory markers (eg, Il1b, Tnf and Ccl2) were reduced in the colon of Zeb1 (+/−) mice (figure 2F and online supplementary figure S2C). As expected, Zeb1 and miR-200a/b/c exhibited a reverse expression pattern in the colon of wild-type and Zeb1 (+/−) mice treated with DSS (online supplementary figure S2D). These results indicate that ZEB1 promotes intestinal inflammation while its downregulation confers protection against it.

ZEB1 promotes inflammatory CRC tumourigenesis

The combination of the alkylating agent azoxymethane (AOM) with DSS is a widely used experimental model of colitis-associated CRC42 (figure 3A). On the other hand, the administration of multiple cycles of AOM in the absence of DSS generates non-inflammation-induced CRCs42 (figure 3A). As noted above, ZEB1 is expressed in dedifferentiated malignant epithelial cells of colorectal and other carcinomas.23 24 40 Accordingly, we found that ZEB1 was expressed in the dedifferentiated areas of inflammatory (AOM/DSS) and non-inflammatory (AOM) CRCs (figure 3B).

Figure 3

ZEB1 promotes and is required for inflammatory intestinal tumourigenesis. (A) Mouse models of intestinal tumourigenesis. Mice were subjected to AOM/DSS or AOM protocols to induce inflammatory and non-inflammatory colorectal cancer (CRC), respectively, as detailed in online supplementary information. (B) Expression of ZEB1 (H-102) in colonic samples of wild-type mice subjected to AOM/DSS or AOM protocols were assessed by immunohistochemistry. Scale bar: 50 µm. (C) and (D) ZEB1 promotes AOM/DSS-induced inflammatory intestinal tumourigenesis. Wild-type and Zeb1 (+/−) mice were subjected to AOM/DSS and their colon were assessed for the total number of tumours (C) and colon length shortening (D). Both parameters were calculated with respect to untreated counterparts of the same genotype. Twelve wild-type and 13 Zeb1 (+/−) mice were analysed. (E) Zeb1 (+/−) mice exhibited a lower inflammatory component in response to AOM/DSS. At the end of the AOM/DSS protocol, colonic samples were assessed by qRT-PCR for Tnf and Il1b expression with respect to Gapdh as reference gene. Data are the mean of a least four mice for each genotype. (F) The tumours generated by AOM/DSS in Zeb1 (+/−) mice display a more differentiated phenotype than those developed in wild-type counterparts. The tumours developed in both genotypes were stained for H&E. The captures shown are representative of mice in panels C–E. Scale bar: 100 µm. (G) Microbiota profile in wild-type and Zeb1 (+/−) mice. Average relative abundance of the top phyla present in colonic mucosa and faeces of 9 wild-type and 10 Zeb1 (+/−) mice. *P≤0.05, **p≤0.01, ***p≤0.001.

ZEB1 promotes tumour initiation and progression through the regulation of different hallmarks.23 However, the mechanism of action of ZEB1 in inflammation-induced cancer has not been studied. When wild-type and Zeb1 (+/−) mice were treated with the AOM/DSS protocol, Zeb1 (+/−) mice developed fewer tumours than wild-type counterparts (figure 3C), and they exhibited a lower inflammatory component with reduced colon shortening and lower expression of inflammatory cytokines (figure 3D and E, and online supplementary figure S3A). Colon tumours in Zeb1 (+/−) mice also displayed a more benign adenoma-like histology—lower mitotic activity, greater preservation of cellular polarisation and gland structure—than in wild-type mice (figure 3F). Altogether, these data show that ZEB1 is required to drive tumourigenesis in the AOM/DSS model and to maintain its inflammatory component.

Similar gut microbiota profile in wild-type and Zeb1 (+/−) mice

Alterations in the gut microbiota have been linked to the pathogenesis of UC and CRC.22 43 Consequently, we examined whether the milder colitis and reduced progression towards inflammatory CRC found in Zeb1 (+/−) mice were related to changes in their gut microbiota. Colonic mucosa and faecal samples from wild-type and Zeb1 (+/−) mice were subjected to next-generation sequencing to determine their microbiota distribution. Bioinformatic and linear discriminant analysis effect size analyses revealed very similar microbiota distribution in both genotypes at the phyla level (figure 3G) and only minor variations at the family and genera level (online supplementary figures S3B and S3C), differences that are unlikely to explain the reduced susceptibility of wild-type and Zeb1 (+/−) mice to colitis and inflammatory CRC.

Downregulation of ZEB1 reduces DNA damage and upregulates the MPG glycosylase during colitis and inflammatory CRC

Inflammatory cytokines and ROS produced by immune cells induce DNA damage and inhibit DNA repair in neighbouring epithelial cells and contribute to their malignant transformation and tumour progression.3–5 We investigated whether a lower DNA damage rate and/or more efficient DNA repair account for the reduced tumourigenesis found in AOM/DSS-treated Zeb1 (+/−) mice.

First, we induced oxidative stress in SW480 cells—a CRC cell line expressing high levels of ZEB123 44 (supplementary figure S4A)—by treating them with IL1β or hydrogen peroxide (H2O2) and assessed DNA damage by their expression of 8-oxo-dG, a marker of oxidative DNA damage with well-established mutagenic properties5 45 (online supplementary figure S4B). Next, we examined the effect of IL1β and H2O2 in SW480 cells that had been stably knocked down for ZEB1 with a validated short hairpin RNA (shRNA) against ZEB1 (SW480-shZEB1 cells)44 or transfected with a control shRNA (SW480-shCtrl cells). It should be noted that 8-oxo-dG is found not only in the nucleus but also in the cytoplasm.46 47 Interestingly, and compared with SW480-shCtrl cells, induction of 8-oxo-dG by IL1β and H2O2 was virtually abrogated in SW480-shZEB1 cells (figure 4A and B, and online supplementary figures S4C-E).

Figure 4

ZEB1 promotes DNA damage and inhibits the expression of MPG. (A) Upper panel: SW480-shCtrl and SW480-shZEB1 cells were either untreated or treated for 30 min with 40 ng/mL of human recombinant IL1β and assessed by immunofluorescence for 8-oxo-dG (15A3) and counterstained with NucBlue (Hoechst 33342) for nuclear staining. Representative captures from five independent experiments. Individual staining pictures are shown in online supplementary figure S4C. Scale bar: 25 µm. Lower panel: as in the upper panel, but SW480-shCtrl and SW480-shZEB1 cells were either untreated or treated for 30 min with 2 mM of hydrogen peroxide (H2O2). Individual staining pictures are shown in online supplementary figure S4E. (B) Quantification of the nuclear mean fluorescence intensity for 8-oxo-dG in (A). (C) ZEB1 promotes 8-oxo-dG lesions in inflammatory intestinal tumours. The intestinal tumours developed in AOM/DSS-treated wild-type and Zeb1 (+/−) mice were examined for 8-oxo-dG (15A3). Scale bar: 50 µm. (D) ZEB1 binds to the MPG promoter. Upper panel: scheme of 1.1 kb of the human MPG promoter. Consensus binding sites for ZEB1 are represented by vertical red lines. The regions of the MPG promoter between −300/–126 bp (containing four ZEB1 consensus binding sites and referred here as ZBS region) and between −685/–517 bp (with no ZEB1 binding sites and referred as NBS region) are labelled with horizontal blue lines. Lower panel: chromatin immunoprecipitation (ChIP) analysis of ZEB1 binding to the ZBS and NBS regions of the MPG promoter as in the upper panel. qRT-PCR amplification of both regions following immunoprecipitation of DNA from SW480 CRC cells—untreated or treated with 40 ng/mL of human recombinant IL1β for 24 hours—either without antibody, with an antibody against ZEB1 (C-20) or goat IgG serum control. Histograms represent the amplification of the ZBS and NBS regions in the MPG promoter—reflecting the binding of ZEB1 to each region—relative to input DNA under each condition. Values shown are representative of four independent experiments. (E) Human recombinant IL1β downregulated MPG expression in a ZEB1-dependent manner. SW480-shCtrl and SW480-shZEB1 cells were treated for 48 hours with 40 ng/mL of human recombinant IL1β and MPG mRNA levels were determined by qRT-PCR with GAPDH as reference gene. Data are the mean of at least seven independent experiments. (F) As in (D) but showing the mean of the binding of ZEB1 to the ZBS region in four independent experiments. (G) Knockdown of ZEB1 in CRC cells upregulates the expression of MPG mRNA. SW480 cells were transiently interfered with siCtrl or with a validated small interfering RNA (siRNA) against ZEB1 and assessed for MPG mRNA expression by qRT-PCR using GAPDH as reference gene. (H) As in (G) but cell lysates from the different conditions were blotted for MPG (11 481–2-AP), ZEB1 (H-102) along with α-tubulin (B5-1-2) as loading control. See online supplementary figure S4F for the full unedited blots. (I) Overexpression of ZEB1 in CRC cells downregulates MPG mRNA levels. ZEB1 and MPG mRNA expression were assessed in HCT116-vector and HCT116-ZEB1 cells by qRT-PCR using GAPDH as reference gene. (J) ZEB1 inhibits MPG expression during colitis. The colon of DSS-treated wild-type and Zeb1 (+/−) mice were assessed for MPG (11 481–2-AP) expression by immunohistochemistry. Left panel: a representative picture of five mice for each genotype is shown. Scale bar: 100 µm. Right panel: quantification of MPG+ cells in representative pictures as in the left panel. (K) ZEB1 inhibits MPG expression in inflammatory CRC. The colonic tumours of AOM/DSS-treated wild-type and Zeb1 (+/−) mice were assessed for MPG (11 481–2-AP) expression by immunohistochemistry. Left panel: a representative picture of five mice for each genotype is shown. Scale bar: 50 µm. Right panel: quantification of MPG+ cells in representative pictures as in the left panel. (L) The colonic tumours of AOM/DSS-treated wild-type and Zeb1 (+/−) mice were assessed for MPG mRNA by qRT-PCR using GAPDH as reference gene. Data shown are the mean of four mice for each genotype. *P≤0.05, **p≤0.01, ***p≤0.001. ns, non-significant.

In addition, we examined whether the in vivo downregulation of Zeb1 also has an effect in the expression of 8-oxo-dG. It was found that the tumours developed by Zeb1 (+/−) mice treated with AOM/DSS exhibited lower levels of 8-oxo-dG than those formed in wild-type mice (figure 4C). Altogether, two inter-related conclusions can be drawn from the above data: first, ZEB1 expression in CRC cells promotes DNA damage in cell line-based systems and in vivo while its knockdown reduces it and, second, the DNA damage induced by these DNA damage promoting stimuli is mediated, at least in part, through ZEB1.

To avoid abnormal base removal and/or the accumulation of toxic/mutagenic intermediates, DNA repair requires a precise transcriptional and post-transcriptional regulation of the different enzymes involved.7 13 Analyses of the activity of serial deletions of the human and rat MPG promoter indicate the existence of several negative regulatory regions.48 49 ZEB1 represses transcription by direct binding to E-box-like sequences in the regulatory regions of its target genes.50 Examination of the human MPG promoter revealed multiple high-affinity consensus sites for ZEB1 (figure 4D, upper panel). We found that an antibody against ZEB1, but not its corresponding matched specie IgG control, immunoprecipitated a region of the MPG promoter containing several ZEB1 binding sites (referred as ZBS region in figure 4D), but not a region lacking consensus binding sites for ZEB1 (NBS region in figure 4D).

Inflammation induces DNA damage and it can inhibit the transcription of genes involved in the DNA repair machinery.3–5 ZEB1 is induced by inflammatory cytokines, including IL1β.34 51 We therefore examined whether IL1β modulates MPG expression in SW480 CRC cells. It was found that SW480-shZEB1 cells express higher levels of MPG than SW480-shCtrl and that IL1β downregulated the steady-state levels of MPG mRNA in SW480-shCtrl but not in SW480-shZEB1 cells (figure 4E). These results suggest that inhibition of MPG by IL1β is mediated, at least in part, by ZEB1. Accordingly, we also found that incubation of SW480 cells with IL1β increased ZEB1 binding to the MPG promoter (lower panel of figure 4D, and figure 4F). Collectively, the above data suggest that MPG is under transcriptional repression by ZEB1.

We then examined whether modulation of ZEB1 expression in CRC cell lines alters MPG expression. Transient knockdown of ZEB1 in SW480 CRC cells with specific small interfering RNAs against ZEB1 27 44 resulted in the upregulation of MPG mRNA and protein (figure 4G and H, and online supplementary figure S4F). Conversely, overexpression of ZEB1 in HCT116—a CRC cell line that, compared with SW480, expresses low levels of ZEB1 (online supplementary figure S4A)27 45—downregulated MPG expression (figure 4I). Lastly, we explored whether ZEB1 regulates MPG in mice subjected to experimental colitis and inflammatory CRC. When wild-type and Zeb1 (+/−) were treated with DSS, Zeb1 (+/−) mice expressed higher levels of MPG than wild-type counterparts (figure 4J). Likewise, the intestinal tumours developed by Zeb1 (+/−) mice in response to AOM/DSS expressed higher levels of MPG than those formed in wild-type mice (figure 4K and L). The above results suggest that ZEB1 inhibits MGP during DSS-induced colitis and AOM/DSS-induced inflammatory CRC.

ZEB1’s role promoting colitis and inflammatory CRC depends on its inhibitory effect on MPG expression

Compared with wild-type counterparts, Mpg (−/−) mice display enhanced colitis in response to DSS and develop a higher number of tumours following treatment with AOM/DSS.8 To investigate whether the role of ZEB1 promoting colitis and inflammatory CRC is mediated by MPG, wild-type and Zeb1 (+/−) mice were crossed with Mpg-deficient mice and the resulting progeny was examined for their response to DSS and AOM/DSS (figure 5A).

Figure 5

ZEB1’s role promoting colitis and inflammatory CRC depends on its inhibition of MPG and a IL1β/ROS-dependent feedback loop with macrophages. (A) Wild-type and Zeb1 (+/−) mice were crossed with Mpg (−/−) mice and subjected to DSS-induced colitis or AOM/DSS-induced inflammatory intestinal tumourigenesis. (B) Deletion of Mpg in the Zeb1 (+/−) mouse reverted its reduced DSS-induced colitis to similar levels than in mice with full levels of Zeb1. Mice of the six genotypes were treated with DSS and shortening of their colon length was assessed with respect to untreated counterparts of the same genotype. The number of mice examined were as follows: Zeb1 (+/+);Mpg (+/+) 18 mice, Zeb1 (+/−);Mpg (+/+) 17 mice, Zeb1 (+/+);Mpg (+/−) 6 mice, Zeb1 (+/−);Mpg (+/−) 8 mice, Zeb1 (+/+);Mpg (−/−) 8 mice, Zeb1 (+/−);Mpg (−/−) 6 mice. (C) Deletion of Mpg in the Zeb1 (+/−) mouse reverted its reduced AOM/DSS-induced inflammatory tumourigenesis to similar levels than in mice with full levels of Zeb1. Left panel: mice of the six genotypes were treated with AOM/DSS and assessed for the number of colonic tumours. Right panel: colon shortening in each mouse was quantified with respect to untreated counterparts of the same genotype. The number of mice examined were as follows: Zeb1 (+/+);Mpg (+/+) 14 mice, Zeb1 (+/−);Mpg (+/+) 15 mice, Zeb1 (+/+);Mpg (+/−) 10 mice, Zeb1 (+/−);Mpg (+/−) 6 mice, Zeb1 (+/+);Mpg (−/−) 4 mice, Zeb1 (+/−);Mpg (−/−) 7 mice. (D) Deletion of both Mpg alleles in the background of the Zeb1 (+/−) mouse reverted the histology of its AOM/DSS-induced tumours towards the more dedifferentiated pattern found in mice with AOM/DSS-treated mice with wild-type levels of Zeb1. Representative samples of the tumours formed in the four genotypes were stained for H&E. See online supplementary figure S5A for the histological pattern of tumours in Zeb1 (+/+);Mpg (+/−) and Zeb1 (+/−);Mpg (+/−) mice. Scale bar: 50 µm. (E) ZEB1 expression in CRC cells elicits higher ROS production by macrophages. Left panel: scheme of the experiment. See online supplementary information for details. Macrophage production of ROS following their culture with conditioned medium (CM) from SW480-shCtrl and SW480-shZEB1 cells was assessed by FACS analysis of CH2-DCFDA fluorescence. Central panel: quantification of the CH2-DCFDA mean fluorescence intensity in macrophages from six mice. Right panel: a representative FACS plot is shown. (F) As in (E) but macrophages were incubated with the CM from HCT116-vector and HCT116-ZEB1 cells. (G) and (H) As in (E) and (F), respectively, but macrophages were assessed by qRT-PCR for Il1b mRNA levels using Gapdh as reference gene. *P≤0.05, **p≤0.01, ***p≤0.001. ns, non-significant.

Importantly, the ablation of one (Mpg (+/−)) or both (Mpg (−/−)) Mpg alleles in the background of the Zeb1 (+/−) mouse (Zeb1 (+/−);Mpg (+/−) and Zeb1 (+/−);Mpg (−/−) mice, respectively) reverted its reduced DSS-induced colitis—assessed by colon shortening—to similar levels than in mice with full levels of Zeb1 (Zeb1 (+/+);Mpg (+/−) and Zeb1 (+/+);Mpg (−/−) mice, respectively) (figure 5B). We also examined the effect of Mpg downregulation in the inflammatory tumourigenesis of wild-type and Zeb1 (+/−) mice following treatment with AOM/DSS. The deletion of one or both Mpg alleles in the background of the Zeb1 (+/−) mouse reverted the fewer number of tumours and reduced colon shortening observed in Zeb1 (+/−) mice in response to AOM/DSS and brought them up to similar levels than those found in Mpg-deficient mice where both alleles of Zeb1 are intact (figure 5C). In addition, Mpg deletion in the background of the Zeb1 (+/−) mouse also reverted the more differentiated histological pattern of tumours in this mouse as described in figure 3F (figure 5D and online supplementary figure S5A). Thus, the tumours formed in Zeb1 (+/−);Mpg (+/−) and Zeb1 (+/−);Mpg (−/−) mice displayed a less differentiated gland architecture that resembled that of tumours developed in mice where both alleles of Zeb1 are wild-type (figure 5D and online supplementary figure S5A). Altogether, these results suggest that the effect of ZEB1 promoting colitis and inflammatory-induced tumourigenesis is mediated, at least in part, by its repressor effect on MPG expression.

ZEB1 expression in CRC cells elicits higher production of ROS and IL1β by macrophages

Epithelial and stromal cells in the intestinal mucosa influence each other during homeostasis and in disease. Soluble factors released by intestinal epithelial cells play key immunoregulatory roles to limit inflammation and to maintain the tolerogenic state of mucosal immune cells against pathogens and the gut microbiota.52 We investigated whether the expression of ZEB1 in CRC cells can modulate the inflammatory and ROS response of macrophages. To that effect, macrophages were incubated with the conditioned medium (CM) collected from either SW480-shCtrl or SW480-shZEB1 cells (figure 5E, left panel). ROS production by macrophages was assessed by both FACS and luminescence using either fluorescent (6-carboxy-2',7'-dichlorodihydrofluorescein diacetate) or chemoluminescent (L-012) probes, respectively. We found that macrophages incubated with CM from SW480-shZEB1 cells released lower levels of ROS than macrophages incubated with CM from SW480-shCtrl cells (figure 5E and online supplementary figure S5B). Conversely, the incubation of macrophages with the CM of HCT116 cells that overexpressed ZEB1 (HCT116-ZEB1 cells) resulted in a higher production of ROS compared with macrophages that were cultured in the presence of CM from HCT116 that harbour the empty expression vector alone (HCT116-vector cells) (figure 5F). These results indicate that ZEB1 expression in epithelial cells stimulates ROS production by macrophages.

Parallel results were observed with regard to IL1β; the CM from SW480-shZEB1 cells triggered lower production of Il1b in macrophages than that from SW480-shCtrl cells (figure 5G), while the CM from HCT116-ZEB1 cells induced higher macrophage production of Il1b than the CM of HCT116-vector cells (figure 5H). Altogether, the above data suggest that ZEB1 expression in inflamed or malignant epithelial cells would induce higher ROS and inflammatory cytokine production by mucosal immune cells, thus reinforcing the positive feedback loop between both cell types.

Discussion

We show here that ZEB1 is upregulated in the epithelial cells of patients with UC and of mouse models of colitis where its expression promotes intestinal inflammation and inflammatory tumourigenesis. ZEB1 exerts these functions, at least in part, through inhibition of the MPG DNA glycosylase (figure 6).

Figure 6

ZEB1 promotes colitis and inflammatory CRC through repression of MPG. See main text for details.

MPG recognises and excises a wide range of damaged bases, including methylated purines (e.g, 3-methyladenine (3meA) 7-methylguanine (7meG)), deaminated purines, ε-based adducts (εA, 1,N2-εG) and although less efficiently than OGG1, 8-oxo-dG.7 8 36 53 Hydrolysation of the glycosylic bond of the damaged base by MPG generates an abasic site that is subsequently processed by APEX1 to generate 3’-hydroxyl and 5’-deoxyribose termini. The latter is subsequently removed by the DNA polymerase β and the nick is eventually ligated to restore DNA integrity.6 7 At every step of this repair process, the BER system generates intermediates whose accumulation can be more toxic and mutagenic than the original DNA base lesion and, as a result, upregulation of DNA glycosylases can paradoxically increase spontaneous mutagenesis and the sensitivity to alkylating agents.12 54 Thus, MPG and APEX1 expression is higher in areas of the colonic mucosa of patients with UC displaying microsatellite instability suggesting an adaptive response of the BER system to inflammation that can paradoxically contribute to tumourigenesis.14 Likewise, several BER genes, including some DNA glycosylases, are overexpressed in CRC and associate with a poorer survival.7

The genetic ablation and overexpression of Mpg in mice illustrate the opposing effects arising from imbalances in the BER system.7 For instance, and compared with wild-type counterparts, the intestine of Mpg (−/−) mice displayed increased levels of ε-base and oxidised base lesions in response to DSS and AOM/DSS as well as higher levels of intestinal inflammation and tumourigenesis.8 11 On the other hand, the overexpression of Mpg in transgenic mice promotes the accumulation of mutagenic BER intermediates and increases the sensitivity of most tissues to the cytotoxic effects of many alkylating agents from whom Mpg (−/−) mice are protected.7 55–57 Nevertheless, this enhanced resistance of Mpg (−/−) mice is dependent on the alkylating agent and is tissue-specific as embryonic stem cells and mouse embryo fibroblasts from these mice are more sensitive to alkylating agents.7 36 58

Along these lines, ZEB1 determines enhanced—rather than reduced—homologous recombination DNA repair and clearance of double-strand breaks in radiotherapy-resistant breast cancer cells.59 Mechanistically, radiotherapy-resistant cells harbour hyperactivated ATM that stabilises ZEB1 protein levels that, in turn, promotes the USP7-mediated stabilisation of CHK1.59 In addition, ZEB1 transcriptionally represses some DNA damage response-related genes to enhance chemotherapy resistance in LoVo CRC cells.60 It remains to be determined whether ZEB1-mediated chemoresistance and/or radioresistance also involve its regulation of MPG.

The above evidence highlights that an efficient and safe DNA base repair process requires the tightly regulated expression of all the components of the BER system. Previous studies showed that the MPG promoter contains negative regulatory regions48 49 and we showed here that MPG expression is inhibited by ZEB1. ZEB1 is a transcription factor particularly suitable for a fine-tuning regulation of MPG expression as it can function as either a transcriptional repressor or an activator in a tissue-depending and/or cell status-depending manner.23 61

Since the GI tract is continuously exposed to antigenic stimuli, the activation of the mucosal immune system must be controlled under homeostatic conditions. Intestinal epithelial cells secrete cytokines and factors that contribute to maintaining immune cell tolerance against commensal bacteria.52 Central to the control of steady-state inflammation in the intestinal mucosa is also the equilibrium between anti-inflammatory T regulatory (TReg) and pro-inflammatory T helper 17 (Th17) responses.62 63 This equilibrium is altered in UC where TReg cells are primed to differentiate towards Th17 cells in the context of high levels of inflammatory cytokines.64–66 Interestingly, ZEB1 is upregulated in lung cancer cell lines in response to IL17 as well as during the differentiation of naive CD4+ T cells into Th17 cells.67 68 Conversely, ZEB1 knockdown reduces the expression of IL17 and other Th17 cell-associated cytokines.67

Damaged DNA is secreted extracellularly and activates a pro-inflammatory response in macrophages.69–71 IL1β is upregulated in the mucosa of patients with UC and contributes to colitis, although its role in its progression towards inflammatory CRC remains unclear.64 We found here that the production of ROS and IL1β by macrophages depended on the expression of ZEB1 by CRC cells. ZEB1 would therefore amplify a positive loop between CRC cells and macrophages to promote inflammation, oxidative stress and DNA damage, and inhibit DNA repair.

ZEB1 is not expressed in healthy intestinal epithelial cells but we showed here that it is induced in the colonic epithelium of patients with UC during inflammatory outbreaks and declines in the remission stage. Being ZEB1 associated with poorer survival in CRC, the cumulative expression of ZEB1 during UC outbreaks over time may play a role in UC and it can be a prognostic biomarker of the risk for the progression of UC towards CRC.

This study has provided evidence for a causal in vivo role for ZEB1 in colitis and inflammatory CRC and open new opportunities to interfere with its expression and function in inflammatory-driven cancers.

Acknowledgments

The authors are indebted to all researchers who provided mouse models and materials (see supplementary information) and regret that some articles were only cited indirectly through reviews due to space limitations. The authors are grateful to Dr L Meira (University of Surrey, UK) for helpful comments on the manuscript. The authors would like to thank the Tumour Bank and the Cytometry and Cell Sorting facilities at IDIBAPS for their help with tissue processing and technical assistance with FACS analyses, respectively. The study was conducted at the IDIBAPS’ Centre de Recerca Biomèdica Cellex.

References

View Abstract

Footnotes

  • OdB, LS-M and MC contributed equally as co-first authors to the work.

  • 36 CN, NP-P, MCM-C and LS contributed equally as co-second authors to the work.

  • Contributors OdB, LS-M and MC performed most of the experimental work in the study. CN and LS assisted in some analyses of mRNA expression; NP-P and MCM-C carried out some immunostaining experiments. RdC performed the microbiota next-generation sequencing and its bioinformatics analysis. MJF-A assisted with the procurement of some of the human samples as well as with the pathology analyses of human and mouse tissue immunostaining. DSD, AC, JM, AS and DCD provided important materials and insights to the study. AP conceived and supervised the study, and wrote the manuscript. All authors provided critical comments to the manuscript.

  • Funding The different parts of this study were independently funded by grants to AP from Catalan Agency for Management of University and Research Grants (AGAUR) (2014-SGR-475 and 2017-SGR-1174), Ministry of Economy and Competitiveness (SAF2014-52874-R and SAF2017-84918-R, National Scientific and Technical Research and Innovation 2013–2016 Plan, which is co-financed by the European Regional Development Fund of the European Union Commission), Fundació La Marató de TV3 (201330.10) and Avon Foundation SAU (AFSAU). The IDIBAPS Institute is partly funded by the CERCA Programme of Generalitat de Catalunya. OdB was subsequently supported by a fellowship from the Spanish Ministry of Education, Culture and Sports (MECS) (FPU programme, AP2010-5489) and by grants 2014-SGR-475, 201330.10 and AFSAU. LS-M is recipient of a PhD scholarship from MECS (FPU Programme, FPU14/06217). MC is supported by grants 201330.10 and SAF2014-52874-R to AP. CN was recipient of a PhD scholarship from AGAUR (AGAUR-2015-FI_B-00121) and later supported by grant 2017-SGR-1174 to AP. NPP and MCMC are recipients of PhD scholarships from the Spanish Ministry of Economy and Competitiveness (FPI Programme, BES-2015-073163 and BES-2015-075757, respectively). LS was supported by grants 201330.10, AVFSAU and 2017-SGR-1174 to AP.

  • Competing interests None declared.

  • Ethics approval The use of human samples was approved by the Clinical Ethics Research Committee at the Hospital Clinic of Barcelona under reference HCB/2018/0157. The use of mouse models followed the guidelines of the Animal Experimentation Ethics Committee at the University of Barcelona (Spain) and was approved under reference UB/301/16 and UB/302/16.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Correction notice This article has been corrected since it published Online First. The text in the abstract has been amended.

  • Patient consent for publication Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.