Mammalian anterior gradient 2 (AGR2), an endoplasmic reticulum (ER) protein disulfide-isomerase (PDI), is involved in cancer cell growth and metastasis, asthma and inflammatory bowel disease (IBD). Mice lacking Agr2 exhibit decreased Muc2 protein in intestinal goblet cells, abnormal Paneth cell development, ileitis and colitis. Despite its importance in cancer biology and inflammatory diseases, the mechanisms regulating agr2 expression in the gastrointestinal tract remain unclear. In the present study, we investigated the mechanisms that control agr2 expression in the pharynx and intestine of zebrafish by transient/stable transgenesis, coupled with motif mutation, morpholino knockdown, mRNA rescue and ChIP. A 350 bp DNA sequence with a hypoxia-inducible response element (HRE) and forkhead-response element (FHRE) within a region −4.5 to −4.2 kbp upstream of agr2 directed EGFP expression specifically in the pharynx and intestine. No EGFP expression was detected in the intestinal goblet cells of Tg(HREM:EGFP) or Tg(FHREM:EGFP) embryos with mutated HRE or FHRE, whereas EGFP was expressed in the pharynx of Tg(HREM:EGFP), but not Tg(FHREM:EGFP), embryos. Morpholino knockdown of foxa1 (forkhead box A1) reduced agr2 levels in the pharynx, whereas knockdown of foxa2 or hif1ab decreased intestinal agr2 expression and affected the differentiation and maturation of intestinal goblet cells. These results demonstrate that Foxa1 regulates agr2 expression in the pharynx, whereas both Foxa2 and Hif1ab control agr2 expression in intestinal goblet cells to regulate maturation of these cells.

INTRODUCTION

Anterior gradient 2 (AGR2) protein, a protein disulfide-isomerase (PDI), has been implicated in cancer cell proliferation, metastatic growth and inflammatory diseases, including asthma and inflammatory bowel disease (IBD). AGR2 is an unusual PDI member, as it contains an atypical thioredoxin-like domain (CPHS) for enzymatic activity [1]. AGR2 is known to be an endoplasmic reticulum (ER) protein because of the presence of a C-terminal KTEL motif. AGR2 expression is controlled by the unfolded protein response and is involved in the ER quality control system to maintain ER homoeostasis [2].

The level of AGR2 expression is correlated with the proliferation and survival of breast cancer, and thus the level of AGR2 is used as a prognostic indicator for certain cancers [3,4]. AGR2 interacts with metastasis-associated GPI-anchored C4.4a protein and extracellular α-dystroglycan, indicative of a potential role for this protein in tumour establishment and metastasis [5]. In lung adenocarcinoma cells, AGR2 was shown to induce expression of amphiregulin, an epidermal growth factor receptor (EGFR) ligand, to activate the Hippo signalling pathway to promote cell growth [6].

AGR2 is involved in inflammatory diseases, such as asthma and IBD. In a mouse model of asthma, Agr2 is overexpressed during allergen exposure and is thus considered to be a suitable asthma biomarker [7]. Increased AGR2 accompanied with mucin overexpression was identified in persons with asthma and in a mouse model of asthma [8].

IBD, including Crohn's disease (CD) and ulcerative colitis (UC), represents the occurrence of chronic remittent or progressive inflammation of the entire gastrointestinal tract and is linked with increased risk for colon cancer [9]. Human AGR2 is localized on chromosome 7p21.3, which is a susceptibility region for IBD [10]. Single nucleotide polymorphisms (SNPs) were tested for association with CD and UC patients, and an association signal was identified for a SNP located in the 5′ region of the AGR2 gene. Down-regulated expression of AGR2 was identified in UC patients. In Agr2-knockout mice, disruption of Paneth and goblet cell homoeostasis and enhanced ER stress was identified, implying that Agr2 plays an important role in intestinal homoeostasis [11]. Furthermore, AGR2 is essential for the production of intestinal mucus, through direct modulation of MUC2 processing [12]. Mice lacking Agr2 are highly susceptible to colitis, implicating AGR2 in the aetiology of IBD. Despite its perceived importance in intestinal homoeostasis and aetiology of IBD, little is known about how the expression of Agr2 is regulated in the gastrointestinal tract.

Intestinal epithelial cells of the gastrointestinal tract are located between an anaerobic lumen and a highly metabolic lamina propria. During active IBD, metabolic shifts towards hypoxia are severe; the activation of hypoxia-inducible factor (HIF) acts as an alarm to help resolve inflammation in several mouse models of disease [13]. Under normoxia, hydroxylation at two proline residues in the ODD domain of HIFα by prolyl hydroxylase (PHD) leads to directed ubiquitylation and subsequent degradation of HIF. Under hypoxia, the HIFα subunit is stabilized, allowing it to interact with HIFβ within the nucleus to activate target gene expression by binding to the hypoxia-inducible response element (HRE) consensus motif on DNA [14].

During hypoxia, HIF-1 was shown to promote expression of intestinal trefoil factor, ecto-5′-nucleotidase (CD73) and muc3 to improve intestinal barrier function [1517]. Pharmacological activation of HIF-1α using a PHD inhibitor (FG-4497) also reduces clinical symptoms in a murine chemical colitis model, suggesting the possible therapeutic use of PHD inhibitors against IBD [1820]. The winged helix transcription factors forkhead box A1 and A2 (Foxa1 and Foxa2) are expressed in definitive endodermal cells during embryogenesis and in different digestive organs of adult mice [21]. Knockdown of Foxa1 and Foxa2 by RNA interference in cultured intestinal cells significantly reduced Muc2 expression, indicating that Foxa1 and Foxa2 are important regulators of Muc2 expression in the intestine [22]. Differentiation of goblet and enteroendocrine L- and D-cells was affected in intestine-specific Faxa1/a2 deletion mutant mice; these results demonstrate the essential role of these proteins in regulating secretory cell differentiation in the mammalian gastrointestinal tract [23].

In the present study, we investigated the regulatory mechanisms controlling agr2 expression in the zebrafish pharynx and intestine by coupling transient/stable transgenesis with 5′-end deletion and binding motif mutation analyses. A 350 bp DNA sequence with a HRE and forkhead-response element (FHRE) located within a region −4.5 to −4.2 kbp upstream of agr2 directed EGFP expression specifically in the pharynx and intestine. Furthermore, we used motif mutation, morpholino knockdown and ChIP to demonstrate that Foxa1 regulates agr2 expression in the pharynx, whereas both Foxa2 and Hif1ab control agr2 expression in intestinal goblet cells and regulate the maturation of these cells.

EXPERIMENTAL

Zebrafish strains and maintenance

Zebrafish, including ASAB wild-type and various transgenic fish, were maintained in 20 litre aquaria supplied with filtered fresh water and aeration, or a high-density self circulation system (Aqua Blue), under a photoperiod of 14 h light and 10 h dark. Embryos were cultivated at 28.5°C and staged according to the criteria described previously [24].

Plasmid construction

PCR was performed to amplify different regions of the agr2 upstream sequence. The following primer pairs were used to amplify the agr2 upstream regions indicated in parentheses: (i) 5′-CTCGAGCTCGAGTGTTCAGGAGATCGGTTCAAAACAGT-CC-3′ and 5′-GTCGACGTCGACCTTTTGCGTCTCTCTCTTC-CACCT-3′ (4.2 kbp: −4120 to +1); (ii) 5′-CTCGAGCTCGAG-CTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-GTCG-ACGTCGACCTTTTGCGTCTCTCTCTTCCACCT-3′ (4.5 kbp: −4527 to +1). All forward primers contain two XhoI sites, whereas reverse primers contain two SalI sites (underlined). PCR products were cloned into the Tol2-transposon-EGFP vector (pT2KXIG△in) [25,26]. The −5.3 k-EGFP plasmid was modified from the −6.0 k-EGFP plasmid by digestion with PacI and BbvCI, followed by end blunting using Klenow DNA polymerase. The 350endo-EGFP construct was generated by PCR, using plasmid DNA from the −6.0 k-EGFP construct as template, and the primer pair: 5′-CCTAGGCCTAGGCTATTTGC-ACTTTTGTCTTCAGAATGCC-3′ and 5′-GGCGCGCCGG-CGCGCCATGAATTGCTGAGGTGTCCTTAACC-3′ (350endo: −4527 to −4176); the PCR product was cloned into the Tol2-transposon-EGFP vector with the agr2 basal promoter by digestion with AvrII and AscI. To generate motif mutation constructs (single mutations of either (i) FHRE or (ii) HRE) within the −4.5 to −4.2 kbp agr2 upstream region, we used the −6.0 k-EGFP plasmid DNA or product of first and second PCR as template for nested PCR with the following primer pairs: (i) 5′-CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAATGC-C-3′ and 5′-AAGTGGTCTGGCCGCTTTGTCGTCG-3′; 5′-AAAGCGGCCAGACCACTTCACTGAAC-3′ and 5′-GGC-GCGCCGGCGCGCCATGAATTGCTGAGGTGTCCTTAACC-3′; 5′CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAA-TGCC-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGA-GGTGTCCTTAACC-3′ (FHREM-EGFP); (ii) 5′-CCTAGGC-CTAGGCTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-TGAGAATGTAAATCTCCCGCTCCG-3′; 5′-GAGATTTACA-TTCTCAGGGCGGAG-3′ and 5′-GGCGCGCCGGCGCGCC-ATGAATTGCTGAGGTGTCCTTAACC-3′; 5′CCTAGGCCT-AGGCTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGAGGTGTCCTT-AACC-3′ (HREM-EGFP). To generate double motif mutations within the −4.5 to −4.2 kbp agr2 upstream region, we used the FHREM-EGFP plasmid DNA or product of first and second PCR as template for nested PCR and the following primer pairs: 5′-CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAAT-GCC-3′ and 5′-TGAGAATGTAAATCTCCCGCTCCG-3′; 5′-GAGATTTACATTCTCAGGGCGGAG-3′ and 5′-GGCGCG-CCGGCGCGCCATGAATTGCTGAGGTGTCCTTAACC-3′; 5′-CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAATGC-C-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGAGGT-GTCCTTAACC-3′ (HFM-EGFP). Mutated sequences in the above primers are italicized. PCR products were then cloned into the Tol2-transpon-EGFP vector with the agr2 basal promoter.

To generate human FHRE and/or HRE motif replacement constructs (single replacement of either (i) FHRE or (ii) HRE) within the −4.5 to −4.2 kbp agr2 upstream region, we used the −6.0 k-EGFP plasmid DNA or product of first and second PCR as template for nested PCR with the following primer pairs: (i) 5′-CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAATGC-C-3′ and 5′-TCCCTGGGTTAAAGAGTTGTCGTCG-3′; 5′-AAC-TCTTTAACCCAGGGACACTGAAC-3′ and 5′-GGCGCG-CCGGCGCGCCATGAATTGCTGAGGTGTCCTTAACC-3′; 5′-CCTAGGCCTAGGCTATTTGCACTTTTGTCTTCAGAATGC-C-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGAGGT-GTCCTTAACC-3′ (HuFHRE-EGFP); (ii) 5′-CCTAGGC-CTAGGCTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-TGAGAGCACTAGTGTCCCGCTCCG-3′; 5′-GACACTAGTG-CTCTCAGGGCGGAG-3′ and 5′-GGCGCGCCGGCGCGCC-ATGAATTGCTGAGGTGTCCTTAACC-3′; 5′-CCTAGGCCT-AGGCTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGAGGTGTCCTT-AACC-3′ (HuHRE-EGFP). To generate double motif replacements within the −4.5 to −4.2 kbp agr2 upstream region, we used the HuFHRE-EGFP plasmid DNA or product of first and second PCR as template for nested PCR with the following primer pairs: 5′-CCTAGGCCTAGGCTATTT-GCACTTTTGTCTTCAGAATGCC-3′ and 5′-TGAGAGCACTA-GTGTCCCGCTCCG-3′; 5′-GACACTAGTGCTCTCAGGGCG-GAG-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCTGA-GGTGTCCTTAACC-3′; 5′-CCTAGGCCTAGGCTATTTGCAC-TTTTGTCTTCAGAATGCC-3′ and 5′-GGCGCGCCGGCGC-GCCATGAATTGCTGAGGTGTCCTTAACC-3′ (HuHF-EGFP). Human HRE or FHRE motif sequences in the above primers are italicized. The PCR product was cloned into the Tol2-transposon-EGFP vector with the agr2 basal promoter by digestion with AvrII and AscI. To generate motif SNP mutation constructs within the −4.5 to −4.2 kbp agr2 upstream region, we used the −6.0 k-EGFP plasmid DNA or product of first and second PCR as template for nested PCR with the following primer pairs: (i) 5′-CCTAGGCCTAGG-CTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-TCC-CTGGGGTAAAGAGTTGTCGTCG-3′; 5′-AACTCTTTACCCC-AGGGACACTGAAC-3′ and 5′-GGCGCGCCGGCGCGCC-ATGAATTGCTGAGGTGTCCTTAACC-3′; 5′-CCTAGGCCT-AGGCTATTTGCACTTTTGTCTTCAGAATGCC-3′ and 5′-GG-CGCGCCGGCGCGCCATGAATTGCTGAGGTGTCCTTAA-CC-3′ (HuFHREM-EGFP); (ii) 5′-CCTAGGCCTAGGCTATT-TGCACTTTTGTCTTCAGAATGCC-3′ and 5′-TGAGAGCAC-TGGTGTCCCGCTCCG-3′; 5′-GACACCAGTGCTCTCAGGG-CGGAG-3′ and 5′-GGCGCGCCGGCGCGCCATGAATTGCT-GAGGTGTCCTTAACC-3′; 5′-CCTAGGCCTAGGCTATTTGC-ACTTTTGTCTTCAGAATGCC-3′ and 5′-GGCGCGCCGGCG-CGCCATGAATTGCTGAGGTGTCCTTAACC-3′ (HuHREM-EGFP). Human HRE or FHRE motif with SNP sequences in the above primers are italicized. PCR products were then cloned into the Tol2-transposon-EGFP vector with the agr2 basal promoter by digestion with AvrII and AscI.

Generation of germline transgenic zebrafish

Transposase mRNA was synthesized using a modified T7–TP plasmid [27] and the mMESSAGE mMACHINE T7 kit (Ambion). To conduct transient expression analyses, circular DNA of the relevant construct was diluted to 11.4 ng/μl and mixed with diluted (11.4 ng/μl) transposase mRNA at a 1:1 ratio. Approximately 2.3 nl of DNA/RNA solution was microinjected into the cytoplasm of one- or two-cell zygotes using a Nanoject II automatic injector (Drummond). To generate germline transgenic zebrafish, we raised injected embryos to adulthood (F0). F0 adult fish were crossed with ASAB wild-type to produce F1 embryos; EGFP expression in the F1 generation was examined by fluorescence microscopy at 120 h post fertilization (hpf). Positive F1 embryos from different lines were raised to adulthood to generate the F2 generation.

Morpholino and mRNA injection

Five previously published morpholino oligonucleotides (MOs), foxa1 MO, 5′-CGCCCAACATTATGGAGGAAATCC-3′ (0.9 ng) [28]; foxa2 MO, 5′-CCTCCATTTTGACAGCACCGAGCAT-3′ (0.5 or 0.9 ng) [29]; foxa3 MO, 5′-CTCGTAAGAA-ACGGGATAGTGACTG-3′ (1.8 ng) [30]; hif1ab MO, 5′-GTGACAACTCCAGTATCCATTCCTG-3′ (1.1 or 2.3 ng) [31]; and epas1b MO, 5′-CGCTGTTCTCGCGTAATTCCCGCAG-3′ (0.5 ng) [32], were used. Additionally, four 5-base mismatch control (5 mm) MOs were synthesized by Gene Tools: 5 mm foxa1 MO, 5′-CGgCgAACATaATGcAGcAAATCC-3′ (0.9 ng); 5 mm foxa2 MO, 5′-CCaCgATTTTcACAcCACCcAGCAT-3′ (0.5 or 0.9 ng); 5 mm hif1ab MO, 5′-GTcACAAg-TCCAcTATCgATTgCTG-3′ (1.1 or 2.3 ng); 5 mm epas1b MO, 5′-CTcTgATCcCGCCcTTCTCcCGTAA-3′ (0.5 ng) (mismatched nucleotides are shown in lower case). MOs were individually diluted in Danieau solution, and injected into one- or two-cell stage zygotes using a Nanoject II automatic injector. For rescue experiments, LacZ (23 pg) or agr2 mRNA (11 or 23 pg) were synthesized by using a mMESSAGE mMACHINE SP6 Kit (Ambion), and co-injected with foxa2 MO or hif1ab MO. To evaluate the specificity and efficacy of the MO used, we cloned part of the 5′ UTR and full-length coding region of hif1aa, hif1ab, epas1a, epas1b, hif1al, hif1al2, foxa1, foxa2 or foxa3 into the pcDNA3-EGFP vector.

Whole-mount in situ hybridization, double fluorescence in situ hybridization and signal intensity quantification

Whole-mount in situ hybridization was conducted on embryos treated with 0.003% phenylthiocarbamide, using digoxigenin-labelled antisense RNA probes and alkaline phosphatase-conjugated anti-digoxigenin antibodies, as described previously [33]. The following antisense RNA probes were generated (restriction sites and promoters in parentheses): agr2 (NcoI/SP6); epas1a (KpnI/SP6); epas1b (NotI/T7); foxa1 (SalI/T7); foxa2 (NcoI/SP6); foxa3 (SacII/SP6); hif1aa (KpnI/SP6); hif1ab (NotI/T7); hif1al (EcoRI/SP6); hif1al2 (BamHI/T7); muc2.1 (NcoI/SP6) and shha (BamHI/T7).

Double fluorescence in situ hybridization for EGFP and shha were conducted using fluorescein- or digoxigenin-labelled RNA probes and Tyramide signal amplification (TSA) plus fluorescein or Cyanine 3 reagents (PerkinElmer), as described previously [34].

The expression level of agr2 in the pharynges or intestines was determined by quantifying signal intensities in the red dotted areas by ImageJ software (NIH) using the following steps: (i) from the Analyze menu, ‘set scale’ was selected to define the length; (ii) the region was selected by hand; (iii) from the Analyze menu, ‘Measure’ was selected for calculating the area size; (iv) from the Edit menu, ‘Clear Outside’ was selected; (v) the images were converted into greyscale; (vi) background was eliminated by adjusting the threshold; and (vii) the intensity was quantified by selecting ‘Analyze Particles’ on the Analyze menu.

Whole-mount wheatgerm agglutinin (WGA) staining

Embryos were fixed in 4% paraformaldehyde at room temperature for 4 h. Fixed embryos were washed three times for 5 min with PBST (PBS containing 0.1% Triton X-100) and incubated in PBST at room temperature for 1 h. Embryos were washed with buffer B (10 mM HEPES and 0.15 M NaCl, pH 7.5), and then incubated with rhodamine-conjugated WGA (Vector Laboratories, 1:100, in buffer B) at 4°C overnight. After four washes with PBST (30 min/wash), embryos were fixed in 4% paraformaldehyde and stored at 4°C.

Alcian Blue staining and quantification of goblet cell number and area

Whole-mount Alcian Blue staining of embryos with different treatments was performed as described in [35]. Alcian Bluestained goblet cell areas were quantified using ImageJ software using the following steps: (i) from the Analyze menu, ‘set scale’ was selected to define the length; (ii) the images were converted into greyscale; (iii) background was eliminated by adjusting the threshold; and (iv) the area was quantified by selecting ‘Analyze Particles’ on the Analyze menu. Alcian Blue-stained goblet cells were counted using ImageJ software as follows: (i) the image was loaded in ImageJ; (ii) ‘Cell counter’ was selected from ‘Analyze’ items on the Plugins menu; (iii) ‘Initialize’ was selected; and (iv) cell number was determined.

Photography

Images of embryos were taken using an AxioCam HRC camera on a Zeiss Imager M1 microscope equipped with DIC mode or an Andor Zyla sCMOS camera on a Zeiss Imager A2 microscope.

ChIP and ChIP–quantitative PCR

The full-length foxa1, foxa2 or hif1ab coding sequence was cloned into a pCS2 vector with a C-terminal 5×-Myc tag sequence for the synthesis of foxa1-Myc, foxa2-Myc or hif1ab-Myc tagged mRNA respectively. ChIP was carried out as previously described [36] with some modifications. One- or two-cell stage zygotes were microinjected with 23 pg of foxa1-Myc, foxa2-Myc or hif1ab-Myc mRNA. Approximately 100 foxa1-Myc-injected 58 hpf embryos, foxa2-Myc-injected 96 hpf embryos or hif1ab–Myc-injected 96 hpf embryos were subjected to DNA cross-linking using 1% formaldehyde, and chromatin DNA was harvested and sonicated to an average size of 100–200 bp using a Bioruptor Pico (Diagenode). For ChIP, pre-blocked Protein G–magnetic beads (Invitrogen Dynal AS) were first incubated with 5 μg of mouse anti-Myc (Cell Signaling Technology) or mouse IgG (Invitrogen) antibody overnight at 4°C. Sheared chromatin (10 μg) was then incubated with antibody-bound Protein G–magnetic beads overnight at 4°C. The protein–DNA complexes were washed, eluted, reverse cross-linked and digested with proteinase K. Chromatin DNA was purified by phenol/chloroform extraction and ethanol precipitation, and then used as template for PCR with primer pairs spanning the FHRE or HRE upstream of the agr2 promoter. Next, qPCR was conducted in a 20 μl reaction buffer containing diluted 20× chromatin DNA sample, forward and reverse primer pairs (each at a concentration of 0.25 mM), and SYBR Green I (1× final concentration) using a Roche Light Cycler 480 II. The following PCR programme was used: 95°C for 10 s, then 95°C for 10 s, 60°C for 10 s and 72°C for 10 s for 60 cycles, then cooling to 4°C. The primer pair used to amplify the HRE region indicated in parentheses was: 5′-TGCACTTTTGTCTTCAGAATGCCTC-3′ and 5′-CC-CTGAGAACACGGATCTCCC-3′ (−4522 to −4351). The primer pair used to amplify the FHRE regions indicated in parentheses was 5′-TGCTCGTGGCCGCTGCACTGTG-3′ and 5′-TGAATTGCTGAGGTGTCCTTAACCC-3′ (−4339 to −4177). The primer pair used as a control for HRE amplification (region indicated in parentheses) was 5′-CATTTTCCATGGCCCCTGATTTGTC-3′ and 5′-CCATCAC-CGGTAGAACACTTGATAC-3′ (−5035 to −4863); the primer pair used as a control for FHRE amplification (region indicated in parentheses) was 5′-CGAAGAGGCCTGACTTGATTCCAC-3′ and 5′-CCAGAGCAAGTCATTGTATGTGAG-3′ (−2650 to −2479).

RESULTS

A 350 bp DNA sequence within the −4.5 to −4.2 kbp region upstream of agr2 confers specific EGFP expression in the gastrointestinal tract of zebrafish embryos

Previously, we generated a stable Tg(−6.0k agr2:EGFP) transgenic zebrafish line that expressed EGFP in a pattern that recapitulated that of endogenous agr2 [37]. Within the gastrointestinal tract of Tg(−6.0k agr2:EGFP), EGFP expression was detected in the anterior endoderm at 48 hpf, in the pharynx at 72 hpf, and in intestinal goblet cells in the mid intestinal region and anus at 96 and 120 hpf respectively. However, weak non-specific EGFP fluorescence was also observed lining the whole intestine from 72 hpf. Additionally, EGFP expression is also observed in the otic vesicles and skin mucous cells from 24 to 120 hpf, and in the swim bladder at 72 hpf (Figure 1B). To confirm that EGFP is expressed in the goblet cells of pharynx and intestine, we stained mucin glycoprotein synthesized by goblet cells in the pharynx and intestine with rhodamine-conjugated WGA. Rhodamine fluorescence was observed to be restricted to apical cup-shaped cytoplasm (theca), which is surrounded by EGFP in the goblet cells of pharynx of 58 hpf embryos and intestines of 96 hpf embryos (Supplementary Figure S1). Since shha is known to be expressed in the oesophagus and pharynx from 34 to 96 hpf [38,39], we performed double fluorescence in situ hybridization against EGFP and shha in Tg(−6.0k agr2:EGFP) transgenic embryos; we observed that these two genes are co-localized in the pharynx of 58 hpf embryos (Supplementary Figure S2).

EGFP expression in embryos (F0) injected with the indicated constructs containing various lengths of the 5′-upstream region of agr2, and in the F1 generation

Figure 1
EGFP expression in embryos (F0) injected with the indicated constructs containing various lengths of the 5′-upstream region of agr2, and in the F1 generation

At 120 hpf, EGFP expression can be detected in the otic vesicles, pharynges, skin mucous cells, intestinal goblet cells and muscles of embryos injected with −6.0 k-EGFP (A), −5.3 k-EGFP (C) or −4.5 k-EGFP (E). In −4.2 k-EGFP-injected embryos, EGFP is expressed only in the otic vesicles, skin mucous cells and muscles (G). EGFP expression is observed in the otic vesicles, pharynges, intestinal goblet cells in the mid intestinal region and skin mucous cells in F6 embryos from the Tg(−6.0 k agr2:EGFP) line (B), F1 embryos from the Tg(−5.3 k agr2:EGFP) line (D) or F1 embryos from the Tg(−4.5 k agr2:EGFP) (F) line. However, EGFP is expressed only in the otic vesicles of F1 embryos from the Tg(−4.2 k agr2:EGFP) transgenic line (H). Percentages of embryos injected with the respective constructs showing EGFP expression in the intestinal goblet cells, otic vesicles, pharynges, skin mucous cells or muscles (I). Percentages of F1 embryos from different transgenic lines expressing EGFP in the intestinal goblet cells, otic vesicles, pharynges or skin mucous cells are shown (J). Error bars indicate the S.E.M. Igc, intestinal goblet cells; Mc, skin mucous cells; Ms, muscle; Ov, otic vesicle; Ph, pharynx.

Figure 1
EGFP expression in embryos (F0) injected with the indicated constructs containing various lengths of the 5′-upstream region of agr2, and in the F1 generation

At 120 hpf, EGFP expression can be detected in the otic vesicles, pharynges, skin mucous cells, intestinal goblet cells and muscles of embryos injected with −6.0 k-EGFP (A), −5.3 k-EGFP (C) or −4.5 k-EGFP (E). In −4.2 k-EGFP-injected embryos, EGFP is expressed only in the otic vesicles, skin mucous cells and muscles (G). EGFP expression is observed in the otic vesicles, pharynges, intestinal goblet cells in the mid intestinal region and skin mucous cells in F6 embryos from the Tg(−6.0 k agr2:EGFP) line (B), F1 embryos from the Tg(−5.3 k agr2:EGFP) line (D) or F1 embryos from the Tg(−4.5 k agr2:EGFP) (F) line. However, EGFP is expressed only in the otic vesicles of F1 embryos from the Tg(−4.2 k agr2:EGFP) transgenic line (H). Percentages of embryos injected with the respective constructs showing EGFP expression in the intestinal goblet cells, otic vesicles, pharynges, skin mucous cells or muscles (I). Percentages of F1 embryos from different transgenic lines expressing EGFP in the intestinal goblet cells, otic vesicles, pharynges or skin mucous cells are shown (J). Error bars indicate the S.E.M. Igc, intestinal goblet cells; Mc, skin mucous cells; Ms, muscle; Ov, otic vesicle; Ph, pharynx.

To investigate the underlying mechanism controlling agr2 expression in organs of the gastrointestinal tract, such as the pharynx and intestine, we used PCR to generate several 5′-end deletion constructs of this gene, and cloned them separately into a Tol2-transposon-EGFP vector. Individual constructs were microinjected into one- or two-cell stage zygotes, and the EGFP expression patterns were inspected at 120 hpf. EGFP expression was detected in 41.2–43.7% of pharynges and 21.9–31.8% of intestinal goblet cells of embryos injected with 6.0 k-EGFP, 5.3 k-EGFP or 4.5 k-EGFP constructs (Figures 1A, 1C, 1E and 1I). However, no EGFP expression was observed in the pharynges or intestines of embryos injected with the 4.2 k-EGFP construct (Figures 1G and 1I). EGFP expression was also detected in the otic vesicles, skin mucous cells and muscles of high proportions of injected embryos, regardless of the construct injected (Figure 1I). As the transient injection of GFP reporter construct usually results in a high percentage of non-specific GFP expression in the muscle, we proceeded to raise the F1 generation of injected embryos for comparison. EGFP expression was detected in the pharynges and intestinal goblet cells of 22.4 or 14.5% of F1 embryos from the Tg(−5.3 k agr2:EGFP) or Tg(−4.5 k agr2:EGFP) line at 120 hpf respectively (Figures 1D, 1F and 1J). However, EGFP expression was not detected in the pharynges or intestines of any F1 embryos from the Tg(−4.2 k agr2:EGFP) transgenic line at the same developmental stage; only EGFP expression in the otic vesicles was detected in 19.6% of F1 embryos from this transgenic line (Figures 1H and 1J). These results indicate that a DNA sequence between −4.5 and −4.2 kbp upstream of agr2 directs EGFP expression specifically in the pharynges, intestinal goblet cells and skin mucous cells of zebrafish embryos.

Intact hypoxia-inducible response and forkhead-response elements are required for EGFP expression in the gastrointestinal tract of zebrafish embryos

The JASPAR database was used to identify the presence of a HRE (CCGTG) and FHRE (TATTGACTCGGC) within the −4.5 to −4.2 kbp region upstream of agr2 (Figure 2K). We then cloned a DNA segment containing HRE and FHRE motifs into the Tol2-transpon-EGFP vector with the agr2 basal promoter (350endo-EGFP). Subsequently, three motif mutation constructs were generated, which include a construct with HRE motif mutation (HREM-EGFP), a construct with FHRE motif mutation (FHREM-EGFP) and a construct with both HRE and FHRE motif mutations (HFM-EGFP) (Figure 2L). In embryos injected with one of the four constructs, most of them expressed EGFP in the skin mucous cells and muscles at 120 hpf (Figures 2A, 2C, 2E, 2G and 2I). In a small portion of embryos that expressed EGFP in the pharynges and/or the intestinal goblet cells, the expression percentages were especially low in embryos injected with HREM-EGFP, FHREM-EGFP or HFM-EGFP (0–5.2% of which expressed EGFP in the pharynges and 0–2.6% of which expressed EGFP in the intestinal goblet cells) as compared with embryos injected with 350endo-EGFP (34.7% of which expressed EGFP in the pharynges and 19.6% of which expressed EGFP in the intestinal goblet cells) (Figure 2I). Injected embryos were raised to adulthood and crossed with wild-type fish to generate the F1 and F2 generation. EGFP was detected in the intestinal goblet cells, pharynges and skin mucous cells of 16% of F1 and roughly 50% of F2 embryos of Tg(350endo:EGFP) transgenic lines (Figure 2B and results not shown). No EGFP expression was detected in the intestinal goblet cells of embryos from Tg(HREM:EGFP), Tg(FHREM:EGFP) or Tg(HFM:EGFP) transgenic lines, although weak non-specific EGFP fluorescence was observed to line the whole intestine of Tg(HREM:EGFP) embryos. EGFP was observed in the pharynges and skin mucous cells of 4.1% of F1 embryos and approximately 50% F2 embryos from the Tg(HREM:EGFP) transgenic line (Figures 2D, 2F, 2H and 2J, and results not shown). These results indicate that HRE and FHRE motifs are required to drive EGFP expression in intestinal goblet cells, whereas the FHRE motif is also essential for directing EGFP expression in the pharynx and skin mucous cells.

HRE and FHRE motifs within the −4.5 to −4.2 kbp region upstream of agr2 are required for directing EGFP expression in the intestinal goblet cells, pharynx and/or skin mucous cells

Figure 2
HRE and FHRE motifs within the −4.5 to −4.2 kbp region upstream of agr2 are required for directing EGFP expression in the intestinal goblet cells, pharynx and/or skin mucous cells

At 120 hpf, EGFP expression is detected in the pharynges, intestinal goblet cells, skin mucous cells and muscles of embryos injected with 350endo-EGFP construct containing both HRE and FHRE motifs (A). EGFP is expressed in skin mucous cells and muscles of the majority of embryos injected with the constructs, containing HRE motif mutation (HREM-EGFP, C), FHRE motif mutation (FHREM-EGFP, E) or both HRE and FHRE motif mutations (HFM-EGFP, G). EGFP is expressed in the pharynges, skin mucous cells and intestinal goblet cells in the mid-intestinal region of F1 embryos of the Tg(350endo:EGFP) transgenic line (B). In F1 embryos of the Tg(HREM:EGFP) transgenic line, EGFP expression is observed in the pharynges and skin mucous cells; however, weak non-specific EGFP fluorescence lining the whole intestine is also detected (D). No EGFP expression can be detected in the F1 embryos of the Tg(FHREM:EGFP) (F) or Tg(HFM:EGFP) (H) transgenic lines. Insets in (A), (B), (D), (F) and (H) show enlarged mid- and posterior intestines. Percentages of embryos expressing EGFP in the intestinal goblet cells, pharynges, skin mucous cells or muscles following injection with the indicated construct are shown (I). Percentages of F1 embryos of the indicated transgenic lines expressing EGFP in the intestinal goblet cells, pharynges or skin mucous cells are shown (J). HRE (red, underlined) and FHRE (blue, underlined) motifs within the −4386 to −4220 bp region upstream of agr2 (K). Mutated HRE (red, italicized) or FHRE (blue, italicized) motif in the indicated constructs (L). Error bars indicate the S.E.M. Igc, intestinal goblet cells; Mc, skin mucous cells; Ms, muscle; Ph, pharynx.

Figure 2
HRE and FHRE motifs within the −4.5 to −4.2 kbp region upstream of agr2 are required for directing EGFP expression in the intestinal goblet cells, pharynx and/or skin mucous cells

At 120 hpf, EGFP expression is detected in the pharynges, intestinal goblet cells, skin mucous cells and muscles of embryos injected with 350endo-EGFP construct containing both HRE and FHRE motifs (A). EGFP is expressed in skin mucous cells and muscles of the majority of embryos injected with the constructs, containing HRE motif mutation (HREM-EGFP, C), FHRE motif mutation (FHREM-EGFP, E) or both HRE and FHRE motif mutations (HFM-EGFP, G). EGFP is expressed in the pharynges, skin mucous cells and intestinal goblet cells in the mid-intestinal region of F1 embryos of the Tg(350endo:EGFP) transgenic line (B). In F1 embryos of the Tg(HREM:EGFP) transgenic line, EGFP expression is observed in the pharynges and skin mucous cells; however, weak non-specific EGFP fluorescence lining the whole intestine is also detected (D). No EGFP expression can be detected in the F1 embryos of the Tg(FHREM:EGFP) (F) or Tg(HFM:EGFP) (H) transgenic lines. Insets in (A), (B), (D), (F) and (H) show enlarged mid- and posterior intestines. Percentages of embryos expressing EGFP in the intestinal goblet cells, pharynges, skin mucous cells or muscles following injection with the indicated construct are shown (I). Percentages of F1 embryos of the indicated transgenic lines expressing EGFP in the intestinal goblet cells, pharynges or skin mucous cells are shown (J). HRE (red, underlined) and FHRE (blue, underlined) motifs within the −4386 to −4220 bp region upstream of agr2 (K). Mutated HRE (red, italicized) or FHRE (blue, italicized) motif in the indicated constructs (L). Error bars indicate the S.E.M. Igc, intestinal goblet cells; Mc, skin mucous cells; Ms, muscle; Ph, pharynx.

Hif1ab regulates intestinal agr2 expression and modulates the differentiation and maturation of intestinal goblet cells

Six hypoxia-inducible factor α (hifα) genes exist in zebrafish due to teleost lineage-specific genome duplication [40]. Through whole-mount in situ hybridization, we found that hif1ab (previous name: hif1α) is expressed in the intestine, pharyngeal arches and brain regions at 72 and 102 hpf, epas1b (endothelial PAS domain protein 1b, previous name: hif2α) is expressed in the intestine, pharyngeal arches and optic tectum regions at 72 hpf and in the foregut, pharyngeal arches and optic tectum at 102 hpf and hif1aa is expressed in the foregut and brain regions, whereas hif1al, hif1al2 and epas1a are only expressed at low levels in the pharyngeal arches at 102 hpf (Supplementary Figure S3).

In order to identify which Hifα binds to the HRE motif located within the −4.5 to −4.2 kbp region upstream of agr2 to regulate agr2 expression in the gastrointestinal tract, we performed morpholino knockdowns of the hif1ab and epas1b genes. To demonstrate the specificity and efficacy of the hif1ab and epas1b MOs [31,32], we fused the EGFP reporter gene in-frame with part of the 5′ UTR sequence and the full-length coding sequence (without the stop codon) of each of the six hifα genes (to generate six different constructs). EGFP expression was not detected in embryos co-injected with 22.3 pg Hif1ab-EGFP and 2.3 ng hif1ab MO or 22.3 pg Epas1b-EGFP and 0.5 ng epas1b MO as compared with embryos co-injected with Hif1ab-EGFP or Epas1b-EGFP and the respective 5 mm control MO at 24 hpf. This finding indicates that hif1ab MO and epas1b MO can efficiently prevent the translation of Hif1ab-EGFP and Epas1b-EGFP fusion proteins respectively. However, EGFP expression was observed in 24 hpf embryos injected with Hif1ab-EGFP plasmid alone and in embryos co-injected with 2.3 ng hif1ab MO and Hif1aa-EGFP, Espa1a-EGFP, Epas1b-EGFP, Hif1al-EGFP or Hif1al2-EGFP. Similarly, EGFP expression was detected at 24 hpf in embryos injected with Epas1b-EGFP plasmid alone and in embryos co-injected with 0.5 ng epas1b MO and Hif1aa-EGFP, Hif1ab-EGFP, Epas1a-EGFP, Hif1al-EGFP or Hif1al2-EGFP (Supplementary Figure S4). These results indicate that the hif1ab and epas1b MOs used in this study are specific and efficacious.

We then injected 2.3 ng of hif1ab MO or 0.5 ng of epas1b MO into one- or two-cell zygotes and performed whole-mount in situ hybridization to examine agr2 expression in the intestinal goblet cells at 102 hpf. Since agr2 is also strongly expressed in the pharynx and otic vesicles at 102 hpf, it is difficult to quantify agr2 expression level in the intestine by quantitative reverse transcription-PCR (qRT-PCR). We then quantified intestinal agr2 expression levels based on signal intensity quantification using ImageJ. We detected a significant decrease in agr2 expression in intestinal goblet cells in hif1ab morphants, but not epas1b morphants, as compared with wild-type and 5 mm hif1ab MO-injected embryos at 102 hpf (Figures 3A, 3C, 3E and 3G and Supplementary Figure S5). In the mammalian intestinal tract, differentiation of goblet cells includes the initial formation of pre-goblet cells (oligomucous cells) containing a small number of mucous granules filled with neutral mucins carrying low amounts of sialic acid. Later, these cells become mature goblet cells containing greater amounts of mucous granules, which in turn possess mucins with abundant sialic acid in the supranuclear cytoplasm [41,42]. Histochemical staining with Alcian Blue, a tetravalent cationic dye, was used to visualize acid mucins within intestinal goblet cells, as it was previously reported to enable immature goblet cells to be distinguished from mature cells in the colons of ER-deficient Min/+mice [43]. Since Agr2 is required for the maturation of intestinal goblet cells in zebrafish embryos, we next analysed the differentiation of goblet cells in the intestine by Alcian Blue staining [34]. We used ImageJ to measure the area of Alcian Blue-stained goblet cells within the intestines of different wild-type embryos. Alcian Blue-stained goblet cells with an area equal to or larger than 26.6±0.4 μm2 (n=128) were defined as mature goblet cells, and those with an area less than this value as immature goblet cells. We then counted both mature and immature goblet cells in the entire intestine of embryos under different treatments. A significant reduction in total goblet cell number in the intestines was detected in hif1ab morphants as compared with wild-type and control embryos at 102 hpf (Figures 3B, 3D, 3F and 3J). Approximately 61.5–69.7% of the goblet cells were immature in the intestines of embryos injected with 1.1 or 2.3 ng of hif1ab MO. Furthermore, co-injection of different amounts of agr2 mRNA, but not LacZ mRNA, significantly rescued the number and maturity of intestinal goblet cells in hif1ab morphants (Figures 3H–3J). Similarly, a substantial reduction in intestinal goblet cell number (with the majority (52%) of the remaining cells being immature) was identified in epas1b morphants as compared with control embryos at 102 hpf (Supplementary Figure S5). Together, these results indicate that Hif1ab, but not Epas1b, regulates agr2 expression in intestinal goblet cells. Both Hif1ab and Epas1b are required for the differentiation and maturation of intestinal goblet cells.

Intestinal agr2 expression and differentiation and maturation of intestinal goblet cells are regulated by Hif1ab

Figure 3
Intestinal agr2 expression and differentiation and maturation of intestinal goblet cells are regulated by Hif1ab

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm hif1ab MO (C) or hif1ab MO (E) at 102 hpf. Expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression level was detected in hif1ab morphants as compared with wild-type or 5 mm hif1ab MO-injected embryos by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with the majority of the remaining goblet cells being immature) was identified in hif1ab morphants (F) as compared with wild-type (B) or 5 mm hif1ab MO-injected control embryos (D) at 102 hpf. Insets in (B) show an immature and a mature goblet cell. A substantial increase in mature goblet cell number was detected in hif1ab morphants co-injected with agr2 mRNA (I) but not with LacZ mRNA (H) at 102 hpf. Intestinal goblet cell numbers in wild-type embryos and embryos injected with the indicated constructs are shown (J). Student's t test was conducted to compare intestinal goblet cell numbers in hif1ab MO-injected embryos with those of wild-type or 5 mm hif1ab MO-injected embryos, and to compare numbers in hif1ab MO and LacZ mRNA co-injected embryos with those of embryos co-injected with hif1ab MO and different amounts of agr2 mRNA; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Figure 3
Intestinal agr2 expression and differentiation and maturation of intestinal goblet cells are regulated by Hif1ab

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm hif1ab MO (C) or hif1ab MO (E) at 102 hpf. Expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression level was detected in hif1ab morphants as compared with wild-type or 5 mm hif1ab MO-injected embryos by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with the majority of the remaining goblet cells being immature) was identified in hif1ab morphants (F) as compared with wild-type (B) or 5 mm hif1ab MO-injected control embryos (D) at 102 hpf. Insets in (B) show an immature and a mature goblet cell. A substantial increase in mature goblet cell number was detected in hif1ab morphants co-injected with agr2 mRNA (I) but not with LacZ mRNA (H) at 102 hpf. Intestinal goblet cell numbers in wild-type embryos and embryos injected with the indicated constructs are shown (J). Student's t test was conducted to compare intestinal goblet cell numbers in hif1ab MO-injected embryos with those of wild-type or 5 mm hif1ab MO-injected embryos, and to compare numbers in hif1ab MO and LacZ mRNA co-injected embryos with those of embryos co-injected with hif1ab MO and different amounts of agr2 mRNA; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Foxa2 modulates intestinal agr2 expression and regulates the differentiation and maturation of intestinal goblet cells whereas Foxa1 regulates agr2 expression in the pharynx

Three winged helix transcription factors of the Foxa family exist in zebrafish. In order to investigate which Foxa protein binds to the FHRE motif within the −4.5 to −4.2 kbp region upstream of agr2 to modulate agr2 expression in the gastrointestinal tract, we performed morpholino knockdown. Whole-mount in situ hybridization revealed that foxa1, foxa2 and foxa3 were expressed in the pharynx and intestine at 58 hpf and in the intestine at 102 hpf (Supplementary Figure S6). To examine the specificity and efficacy of the foxa1, foxa2 or foxa3 MOs [2830], we fused the EGFP reporter gene in-frame with part of the 5′ UTR sequence and the full-length coding sequence (lacking the stop codon) of three foxa genes, individually. EGFP expression was not detected at 24 hpf in embryos co-injected with 22.3 pg of Foxa1-EGFP and 0.9 ng foxa1 MO, 22.3 pg of Foxa2-EGFP and 0.9 ng of foxa2 MO or 22.3 pg of Foxa3-EGFP and 1.8 ng of foxa3 MO. This finding indicates that the translation of Foxa1-EGFP, Foxa2-EGFP and Foxa3-EGFP fusion proteins was efficiently blocked by foxa1, foxa2 or foxa3 MOs respectively. EGFP expression was observed in embryos co-injected with Foxa1-EGFP or Foxa2-EGFP and the respective 5 mm control MOs. EGFP expression was also detected in embryos injected with Foxa1-EGFP, Foxa2-EGFP or Foxa3-EGFP plasmid alone, Foxa1-EGFP plasmid and foxa2 MO or foxa3 MO, Foxa2-EGFP and foxa1 MO or foxa3 MO, or with Foxa3-EGFP and foxa1 MO or foxa2 MO (Supplementary Figure S7). These results demonstrate the specificity and efficacy of foxa1 MO, foxa2 MO or foxa3 MO.

At 102 hpf, foxa2 morphants, but not foxa1 or foxa3 morphants, showed substantially decreased agr2 expression in the intestinal goblet cells (but not pharynges) as compared with wild-type and 5 mm foxa2 MO-injected embryos (Figures 4A, 4C, 4E and 4G, and Supplementary Figures S8 and S9). Alcian Blue staining was performed to investigate whether the differentiation of intestinal goblet cells was affected in foxa2 morphants. A significant reduction in the number of goblet cells (with most (69.5%) of the remaining cells being immature) was identified in foxa2 morphants as compared with wild-type and control embryos (Figures 4B, 4D, 4F and 4J). Furthermore, co-injection with various amounts of agr2 mRNA, but not LacZ mRNA, substantially rescued the number and maturity of intestinal goblet cells in foxa2 morphants (Figures 4H, 4I and 4J). Similarly, total cell number and maturity of intestinal goblet cells were significantly affected in foxa1 and foxa3 morphants as compared with control embryos at 102 hpf (Supplementary Figures S8 and S9).

Foxa2 modulates intestinal agr2 expression and the differentiation and maturation of intestinal goblet cells

Figure 4
Foxa2 modulates intestinal agr2 expression and the differentiation and maturation of intestinal goblet cells

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm foxa2 MO (C) or foxa2 MO (E) at 102 hpf. Expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression level was detected in foxa2 morphants as compared with wild-type or 5 mm foxa2 MO-injected embryos by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with most of the remaining goblet cells being immature) was detected in foxa2 morphants (F) as compared with wild-type (B) or 5 mm foxa2 MO-injected control embryos (D) at 102 hpf. Insets in (B) show an immature and a mature goblet cell. Intestinal goblet cell numbers in the indicated embryos are shown (J). Student's t test was conducted to compare intestinal goblet cell numbers in foxa2 MO-injected embryos with those of wild-type or 5 mm foxa2 MO-injected embryos, and to compare numbers in foxa2 MO and LacZ mRNA co-injected embryos with those of embryos co-injected with foxa2 MO and different amounts of agr2 mRNA; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Figure 4
Foxa2 modulates intestinal agr2 expression and the differentiation and maturation of intestinal goblet cells

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm foxa2 MO (C) or foxa2 MO (E) at 102 hpf. Expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression level was detected in foxa2 morphants as compared with wild-type or 5 mm foxa2 MO-injected embryos by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with most of the remaining goblet cells being immature) was detected in foxa2 morphants (F) as compared with wild-type (B) or 5 mm foxa2 MO-injected control embryos (D) at 102 hpf. Insets in (B) show an immature and a mature goblet cell. Intestinal goblet cell numbers in the indicated embryos are shown (J). Student's t test was conducted to compare intestinal goblet cell numbers in foxa2 MO-injected embryos with those of wild-type or 5 mm foxa2 MO-injected embryos, and to compare numbers in foxa2 MO and LacZ mRNA co-injected embryos with those of embryos co-injected with foxa2 MO and different amounts of agr2 mRNA; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Strong expression of agr2 in the pharynges of embryos at 102 hpf may hinder the detection of decreased expression of agr2 after knockdown with different foxa MOs. Because development of the pharynx and oesophagus can be detected at 58 hpf, we knocked down individual foxa genes and analysed their effect on agr2 expression in the pharynx at this developmental stage [44]. First, we observed that foxa1, foxa2 and foxa3 are expressed in the pharynx at 58 hpf (Supplementary Figure S6). We then knocked down foxa1, foxa2 or foxa3 using the respective specific MO, and harvested morphants at 58 hpf for whole-mount in situ hybridization. foxa1 morphants exhibited a significant decrease in agr2 expression level in pharynges (based on signal intensity quantification) as compared with wild-type and control embryos (Figures 5A–5C and 5G), whereas agr2 expression in the pharynges was not affected in embryos injected with foxa2 MO or foxa3 MO at 58 hpf (Figures 5D–5F and 5G). The specific effect of foxa1 was confirmed by the finding that expression of shha in the pharynges was not affected in embryos injected with different foxa MO at 58 hpf (Figures 5H–5M).

Knockdown of foxa1 reduces agr2 expression in the pharynx

Figure 5
Knockdown of foxa1 reduces agr2 expression in the pharynx

Expression of agr2 in the pharynx is diminished in embryos injected with foxa1 MO (C) as compared with wild-type (A) or embryos injected with 5 mm foxa1 MO (B). Similar expression levels of agr2 in the pharynx were observed in wild-type (A) and embryos injected with 5 mm foxa2 MO (D), foxa2 MO (E) or foxa3 MO (F). Expression of agr2 in the pharynx was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A)–(F). A significant decrease in agr2 expression in the pharynx was detected in foxa1, but not foxa2 or foxa3 morphants (G). Student's t test was conducted to compare expression levels in foxa1 MO-injected embryos with those of wild-type or 5 mm foxa1 MO-injected embryos; ***P<0.001. Insets in (B)–(F) show enlarged pharynx regions. Expression of shha in the pharynges of foxa1 (J), foxa2 (L) or foxa3 (M) morphants is similar to that of wild-type (H) and embryos injected with 5 mm foxa1 MO (I) or 5 mm foxa2 MO (K) at 58 hpf. Ph, pharynx; WT, wild-type.

Figure 5
Knockdown of foxa1 reduces agr2 expression in the pharynx

Expression of agr2 in the pharynx is diminished in embryos injected with foxa1 MO (C) as compared with wild-type (A) or embryos injected with 5 mm foxa1 MO (B). Similar expression levels of agr2 in the pharynx were observed in wild-type (A) and embryos injected with 5 mm foxa2 MO (D), foxa2 MO (E) or foxa3 MO (F). Expression of agr2 in the pharynx was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A)–(F). A significant decrease in agr2 expression in the pharynx was detected in foxa1, but not foxa2 or foxa3 morphants (G). Student's t test was conducted to compare expression levels in foxa1 MO-injected embryos with those of wild-type or 5 mm foxa1 MO-injected embryos; ***P<0.001. Insets in (B)–(F) show enlarged pharynx regions. Expression of shha in the pharynges of foxa1 (J), foxa2 (L) or foxa3 (M) morphants is similar to that of wild-type (H) and embryos injected with 5 mm foxa1 MO (I) or 5 mm foxa2 MO (K) at 58 hpf. Ph, pharynx; WT, wild-type.

Taken together, these results demonstrate that Foxa1 regulates agr2 expression in the pharynx, whereas Foxa2 modulates agr2 expression in intestinal goblet cells. The differentiation and maturation of intestinal goblet cells are modulated by Foxa1, Foxa2 and Foxa3.

Hif1ab and Foxa2 function synergistically in the regulation of agr2 expression and the differentiation and maturation of intestinal goblet cells

Based on the close proximity of the HRE and FHRE motifs upstream of the agr2 promoter, as well as the finding that injection of hif1ab MO or foxa2 MO reduced agr2 expression in intestinal goblet cells in 102 hpf embryos (Figures 3 and 4), we wondered whether Hif1ab co-operates with Foxa2 to regulate agr2 expression in the intestinal goblet cells. To test this hypothesis, we co-injected reduced amounts of hif1ab MO (1.1 ng) and foxa2 MO (0.5 ng) into one- or two-cell stage zygotes and analysed their effects on agr2 expression and the differentiation of goblet cells in the intestine at 102 hpf. A significant decrease of agr2 expression in the intestinal goblet cells was identified based on signal intensity quantification in 102 hpf embryos co-injected with reduced amounts of hif1ab MO and foxa2 MO (Figures 6E and 6G). Similarly, decreased numbers of intestinal goblet cell (with the majority of remaining cells being immature goblet cells) were detected in more (72.4%) co-injected embryos (Figures 6F and 6H) than embryos injected with hif1ab MO (69.7%) or foxa2 MO (69.5%) (Figures 3J and 4J). These results indicate that Hif1ab and Foxa2 function synergistically in the regulation of agr2 expression in the intestinal goblet cells, as well as the differentiation and the maturation of goblet cells in the intestine.

Co-injection of low amounts of hif1ab and foxa2 MO has synergistic effects on the differentiation and maturation of intestinal goblet cells and intestinal agr2 expression

Figure 6
Co-injection of low amounts of hif1ab and foxa2 MO has synergistic effects on the differentiation and maturation of intestinal goblet cells and intestinal agr2 expression

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm hif1ab MO and 5 mm foxa2 MO (C) or hif1ab MO (1.1 ng) and foxa2 MO (0.5 ng) (E) at 102 hpf. The expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression levels was detected in embryos co-injected with hif1ab MO and foxa2 MO as compared with wild-type or embryos co-injected with 5 mm hif1ab MO and 5 mm foxa2 MO by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with the majority of the remaining goblet cells being immature) was detected in embryos co-injected with hif1ab MO and foxa2 MO (F) as compared with wild-type (B) or 5 mm hif1ab MO and 5 mm foxa2 MO-co-injected control embryos (D). Insets in (B) show an immature and a mature goblet cell. Intestinal goblet cell numbers in the indicated embryos are shown (H). Student's t test was conducted to compare intestinal goblet cell numbers in hif1b MO and foxa2 MO-co-injected embryos with those of wild-type or embryos co-injected with 5 mm hif1ab MO and 5 mm foxa2 MO-injected embryos; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Figure 6
Co-injection of low amounts of hif1ab and foxa2 MO has synergistic effects on the differentiation and maturation of intestinal goblet cells and intestinal agr2 expression

Whole-mount in situ hybridization using agr2 antisense RNA probe was conducted on wild-type embryos (A) and embryos injected with 5 mm hif1ab MO and 5 mm foxa2 MO (C) or hif1ab MO (1.1 ng) and foxa2 MO (0.5 ng) (E) at 102 hpf. The expression of agr2 in intestinal goblet cells was analysed by quantification of signal intensity within the areas surrounded by red dotted lines in (A), (C) and (E). A significant decrease in intestinal agr2 expression levels was detected in embryos co-injected with hif1ab MO and foxa2 MO as compared with wild-type or embryos co-injected with 5 mm hif1ab MO and 5 mm foxa2 MO by Student's t test (G). A decrease in the total number of Alcian Blue-stained goblet cells (with the majority of the remaining goblet cells being immature) was detected in embryos co-injected with hif1ab MO and foxa2 MO (F) as compared with wild-type (B) or 5 mm hif1ab MO and 5 mm foxa2 MO-co-injected control embryos (D). Insets in (B) show an immature and a mature goblet cell. Intestinal goblet cell numbers in the indicated embryos are shown (H). Student's t test was conducted to compare intestinal goblet cell numbers in hif1b MO and foxa2 MO-co-injected embryos with those of wild-type or embryos co-injected with 5 mm hif1ab MO and 5 mm foxa2 MO-injected embryos; ***P<0.001. Igc, intestinal goblet cells; WT, wild-type.

Foxa1, Foxa2 and Hif1ab directly bind to FHRE or HRE motifs located upstream of agr2

In order to investigate whether Foxa1, Foxa2 or Hif1ab binds directly to the respective FHRE or HRE motifs within the −4.5 to −4.2 kbp region upstream of agr2, we performed ChIP. As no antibodies against zebrafish Foxa1, Foxa2 or Hif1ab are available, we microinjected foxa1-Myc, foxa2-Myc or hif1ab-Myc mRNA into one- or two-cell stage zygotes and used an anti-Myc antibody to immunoprecipitate cross-linked chromatin from injected embryos at 58 or 96 hpf. In vivo binding of Foxa1 or Foxa2 to the FHRE motif located within the −4275 to −4263 bp region upstream of agr2 was identified in foxa1-Myc-injected 58 hpf or foxa2-Myc-injected 96 hpf embryos, whereas no binding was observed in another region (−2650 to −2479) of the agr2 promoter (Figures 7A and 7B). Similarly, in vivo binding of Hif1ab to the HRE motif located within the −4364 to −4360 bp region upstream of agr2 was detected in hif1ab-Myc-injected 96 hpf embryos, whereas no binding was detected at another region (−5035 to −4863) of the agr2 promoter (Figure 7C). These results demonstrate that Foxa1, Foxa2 and Hif1ab directly bind to FHRE or HRE motifs located within the −4.5 to −4.2 kbp region upstream of agr2.

Foxa1, Foxa2 and Hif1ab directly bind to FHRE or HRE motifs located upstream of the agr2 promoter

Figure 7
Foxa1, Foxa2 and Hif1ab directly bind to FHRE or HRE motifs located upstream of the agr2 promoter

Chromatin DNA isolated from 58 hpf embryos injected with foxa1-Myc mRNA (A) or 96 hpf embryos injected with foxa2-Myc mRNA (B) or with hif1ab-Myc mRNA (C) were immunoprecipitated using anti-Myc antibody or anti-IgG antibody as a control. The immunoprecipitates were subjected to PCR using specific primers spanning the FHRE (−4339 to −4177 bp) and negative control primers against a −2650 to −2479 bp region (A and B), or using specific primers spanning the HRE (−4522 to −4351 bp) and negative control primers against a −5035 to −4863 bp region (C) in the agr2 promoter. FHRE (A and B) and HRE (C) motifs were significantly enriched by anti-Myc antibody as compared with IgG control. Error bars indicate the S.E.M.

Figure 7
Foxa1, Foxa2 and Hif1ab directly bind to FHRE or HRE motifs located upstream of the agr2 promoter

Chromatin DNA isolated from 58 hpf embryos injected with foxa1-Myc mRNA (A) or 96 hpf embryos injected with foxa2-Myc mRNA (B) or with hif1ab-Myc mRNA (C) were immunoprecipitated using anti-Myc antibody or anti-IgG antibody as a control. The immunoprecipitates were subjected to PCR using specific primers spanning the FHRE (−4339 to −4177 bp) and negative control primers against a −2650 to −2479 bp region (A and B), or using specific primers spanning the HRE (−4522 to −4351 bp) and negative control primers against a −5035 to −4863 bp region (C) in the agr2 promoter. FHRE (A and B) and HRE (C) motifs were significantly enriched by anti-Myc antibody as compared with IgG control. Error bars indicate the S.E.M.

DISCUSSION

Transient/stable transgenic analyses coupled with 5′-end deletion revealed that a 350 bp sequence containing HRE and FHRE motifs within the −4.5 to −4.2 kbp region upstream of agr2 is able to direct EGFP expression in the gastrointestinal tract and skin mucous cells. Furthermore, we used motif mutation, morpholino knockdown and ChIP to demonstrate that Foxa1 regulates agr2 expression in the pharynx, whereas both Foxa2 and Hif1ab control agr2 expression in intestinal goblet cells.

Foxa1 and Foxa2 are essential for the differentiation of goblet and enteroendocrine cells in the mammalian gastrointestinal tract, as demonstrated by examining the effects of deleting both Foxa1 and Foxa2 specifically in the small intestine and colon of Villin-Cre mice [23]. Furthermore, ChIP and electrophoretic mobility shift assays were used to show that Foxa1 and Foxa2 directly bind to the Muc2 promoter, and inhibition of the expression of Foxa1 and Foxa2 by RNA interference in cultured intestinal cells reduced Muc2 mRNA levels [22]. Consistent with these reports, knockdown of zebrafish foxa1, foxa2 or foxa3 caused defects in the differentiation and maturation of goblet cells in the intestine (Figure 4, and Supplementary Figures S8 and S9). The differentiation of enteroendocrine cells in the intestines was also affected in zebrafish foxa1 and foxa2 morphants but not foxa3 morphants (results not shown). However, muc2.1 expression in the intestinal goblet cells was not affected in foxa1, foxa2 or foxa3 morphants at 102 hpf (Supplementary Figure S10); this indicates that, unlike their homologues in mice, zebrafish Foxa1, Foxa2 and Foxa3 do not regulate muc2.1 expression. In contrast, we discovered that Foxa1 and Foxa2 bind an FHRE motif upstream of the agr2 promoter; Foxa1 regulates agr2 expression in the pharynx, whereas Foxa2 modulates intestinal agr2 expression in the intestinal goblet cells of zebrafish embryos (Figures 2, 4, 5 and 7). This result is supported by recent findings that expression of AGR2 in breast cancer cells is dependent on FOXA1, and up-regulation of FOXA2 induces expression of AGR2 and MUC2 in Barrett's oesophagus [45,46]. Since Agr2 is required for the terminal differentiation of intestinal goblet cells [34], reduced intestinal agr2 expression in foxa2 morphants is likely to have resulted in the observed defects of goblet cell maturation.

In Caenorhabditis elegans, PHA-4, a FoxA homologue, was shown to be the central regulator of pharynx development, because no pharyngeal primordium formed in pha-4 deficient embryos [47]. Unlike C. elegans PHA-4, Foxa1 is not important for regulating pharynx development in zebrafish embryos, as evidenced by the observation that the pharynx exhibited normal development and shha expression in foxa1 morphants at 58 hpf (Figure 5). Both FoxA1 and FoxA2 are considered to be pioneer factors due to their ability to actively open up condensed chromatin and make it competent for the binding of other transcription factors [48]. Cell-lineage-specific activities were identified at FOXA1-bound enhancers, which occur through co-recruitment of a second cell-specific factor: AR in LNCaP prostate adenocarcinoma cells or ESR1 in MCF7 breast cancer cells [49]. In C. elegans, PHA-4 is considered to be the selector gene of the pharynx, as (i) prevention of PHA-4 binding in intestinal cells eliminates target gene expression in the intestine; and (ii) PHA-4 binding within the pharynx is limited by a negative regulator (EMR-1/Emerin) [50]. Based on these previous findings, we speculate that FoxA1 and FoxA2 modulation of agr2 expression specifically in the pharynx and intestinal goblet cells (respectively) may be attributed to the differing cellular environments of these two organs and the presence of cell-specific factors or negative regulators.

Mammalian Hif1α has been shown to maintain intestinal barrier function during hypoxia by induction of intestinal trefoil factor and Ecto-5′-nucleotidase (Cd73) [15,16]. Moreover, Hif1α was shown to be essential for intestinal barrier protection during mucosal inflammation; mice lacking intestinal epithelium-specific Hif1α displayed severe clinical symptoms during induction of colitis by 2,4,6-trinitrobenzene sulfonic acid (TNBS) treatment, whereas von Hippel–Lindau (VHL)-deficient mice exhibited increased levels of HIF-1-regulated barrier-protective genes, resulting in the attenuation of colitis symptoms [51]. In the present study, we demonstrated that the differentiation and maturation of intestinal goblet cells was affected in hif1ab-deficient zebrafish embryos (Figure 3). In addition, we identified that Hif1ab directly regulates intestinal expression of agr2, which is required for the maturation of intestinal goblet cells (Figures 2 and 7). Therefore, Hif1ab is required for the normal development of goblet cells to maintain proper intestinal barrier function in zebrafish embryos.

Down-regulated AGR2 expression was detected in patients with UC, and some SNPs located upstream of the AGR2 genes exhibited an association signal with UC and CD [10]; these observations suggest that susceptibility to human IBD is associated with the AGR2 gene. Since we observed that a 350 bp DNA sequence containing HRE and FHRE motifs within the −4.5 to −4.2 kbp region upstream of agr2 confers specific EGFP expression in the gastrointestinal tract of zebrafish embryos, we wondered whether the region upstream of human AGR2 gene contains similar HRE and FHRE motifs. We also investigated whether HRE and FHRE motifs from the human AGR2 gene can replace those of zebrafish agr2 gene to direct EGFP expression in the pharynx and intestines of zebrafish embryos, and whether incorporation of SNPs into HRE and/or FHRE motifs of human AGR2 gene prevents EGFP expression in zebrafish embryos. We then manually searched the region upstream of the human AGR2 gene for the presence of SNPs, HRE and FHRE motifs (no homology was detected between the 350 bp region in zebrafish and the human AGR2 upstream region by Nucleotide Blast), thereby identifying a 360 bp DNA sequence containing three predicted FHRE and two HRE motifs within the −3590 to −3230 bp region upstream of the human AGR2 translation initiation codon. The sequences of a pair of FHRE (−3337 bp to −3326 bp) and HRE (−3261 bp to −3257 bp) motifs separated by 65 bp DNA sequence within this 360 bp DNA segment were used to replace the respective FHRE and HRE motif sequences in the 350endo-EGFP plasmid to direct EGFP expression. In addition, we also incorporated an SNP located in either the human FHRE or HRE motif sequence into constructs to direct EGFP expression in zebrafish embryos.

We produced the following constructs: (i) HuHRE-EGFP (the zebrafish HRE motif was replaced by the human HRE (TAGTG) motif in 350endo-EGFP); (ii) HuFHRE-EGFP (the zebrafish FHRE motif was replaced by the human FHRE (TCTTTAACCCAG) motif); (iii) HuHF-EGFP (both zebrafish HRE and FHRE motifs were replaced by human HRE and FHRE motifs); (iv) HuHREM-EGFP (an SNP (07AGR1N17) was introduced in the HuHRE motif); and (v) HuFHREM-EGFP (an SNP (07AGRNP53) was incorporated into the HuFHRE motif). We then injected individual constructs into one- or two-cell stage zygotes, and the EGFP expression patterns were inspected at 120 hpf. EGFP expression at 120 hpf was only observed in skin mucous cells and muscles of embryos injected with HuHRE-EGFP, HuHREM-EGFP, HuFHRE-EGFP, HuFHREM-EGFP or HuHF-EGFP (Supplementary Figure S11). Injected embryos were raised to adulthood and crossed with wild-type fish to generate the F1 generation. Interestingly, EGFP expression at 120 hpf was detected in the pharynges, intestinal goblet cells and skin mucous cells of 0.6 or 3.5% of F1 embryos from Tg(HuHRE:EGFP) or Tg(HuHREM:EGFP) transgenic lines respectively (Supplementary Figure S11). EGFP expression in the pharynx and skin mucous cells in Tg(HuHRE:EGFP) or Tg(HuHREM:EGFP) transgenic F1 embryos is directed by the binding of Foxa transcription factor to intact zebrafish FHRE motif. EGFP expression in the intestinal goblet cells in Tg(HuHRE:EGFP) or Tg(HuHREM:EGFP) transgenic F1 embryos is attributed to the binding of Hif1ab transcription factor to HuHRE (TAGTG) or HuHREM (CAGTG) motifs, which possess only one or two nucleotide differences from the zebrafish HRE (CCGTG). No EGFP expression was observed at 120 hpf in F1 embryos from Tg(HuFHRE:EGFP), Tg(HuFHREM:EGFP) or Tg(HuHF:EGFP) transgenic lines, indicating that the failure of zebrafish Foxa transcription factors to bind can be attributed to sequence divergence between zebrafish FHRE and human FHRE motifs. Further investigation using human intestinal cell lines is required to determine whether the predicted HuHRE and HuFHRE motifs located upstream of the human AGR2 gene can be bound by Hif1α or Foxa transcription factor to regulate AGR2 expression in the gastrointestinal tract.

We demonstrated that Hif1ab and Foxa2 directly bind to HRE and FHRE motifs respectively to regulate agr2 expression in intestinal goblet cells. Moreover, the differentiation and maturation of intestinal goblet cells is also regulated by Hif1ab or Foxa2 (Figures 24 and 7). Synergistic effects on intestinal agr2 expression and the differentiation and maturation of intestinal goblet cells were identified in embryos injected with reduced amounts of hif1ab or foxa2 MO (Figure 6), suggesting that Hif1ab may co-operate with Foxa2 to modulate intestinal agr2 expression and facilitate terminal differentiation of intestinal goblet cells. Since Agr2 primarily modulates intestinal goblet cell maturation, the factor(s) that is/are regulated by Hif1ab or Foxa2 and is/are responsible for the differentiation of intestinal goblet cells remain(s) to be identified. Nevertheless, our result is consistent with a previous finding that co-operation between HIF-1α and FoxA2 in neuroendocrine tissue transactivates HIF-regulated genes (including Hes6, Sox9 and Jmjd1a) to promote neuroendocrine prostate tumour development [52].

In conclusion, through dissecting the regulatory mechanisms of agr2 expression in the gastrointestinal tract, we have demonstrated that Hif1ab may co-operate with Foxa2 to transactivate agr2 expression in intestinal goblet cells, thereby promoting the terminal differentiation of these cells, whereas agr2 expression in the pharynx is activated by Foxa1.

AUTHOR CONTRIBUTION

Yun-Ren Lai, Yu-Fen Lu and Huang-Wei Lien performed experiments and analysed data; Yun-Ren Lai, Chang-Jen Huang and Sheng-Ping Hwang developed the concepts, and prepared and edited (pre-submission) the paper.

We thank Professor Chin-Hwa Hu for providing epas1b MO. We thank the Taiwan Zebrafish Core Facility (TZCAS) for providing the ASAB strain. We are also grateful to the Core Facility of the Institute of Cellular and Organismic Biology, Academia Sinica for ChIP assistance and Ms Mei-Chen Chen for fish maintenance.

FUNDING

This work was supported by the Ministry of Science and Technology [grant numbers NSC-102-2628-B-001-001-MY3 (to S.P.H.) and NSC 98-2321-B-001-028-MY3 (to S.P.H.)]; and the Innovative Translational Agricultural Research Program, Academia Sinica [grant number 542300 (to S.P.H.)].

Abbreviations

     
  • AGR2

    anterior gradient 2

  •  
  • CD

    Crohn's disease

  •  
  • ER

    endoplasmic reticulum

  •  
  • FHRE

    forkhead-response element

  •  
  • Foxa

    forkhead box A

  •  
  • HIF

    hypoxia-inducible factor

  •  
  • hpf

    hours post fertilization

  •  
  • HRE

    hypoxia-inducible response element

  •  
  • IBD

    inflammatory bowel disease

  •  
  • 5 mm

    5-base mismatch

  •  
  • MO

    morpholino oligomer

  •  
  • PDI

    protein disulfide-isomerase

  •  
  • PHD

    prolyl hydroxylase

  •  
  • SNP

    single nucleotide polymorphisms

  •  
  • UC

    ulcerative colitis

  •  
  • WGA

    wheatgerm agglutinin

References

References
1
Shishkin
S.S.
Eremina
L.S.
Kovalev
L.I.
Kovaleva
M.A.
AGR2, ERp57/GRP58, and some other human protein disulfide isomerases
Biochemistry (Mosc.)
2013
, vol. 
78
 (pg. 
1415
-
1430
)
[PubMed]
2
Higa
A.
Mulot
A.
Delom
F.
Bouchecareilh
M.
Nguyen
D.T.
Boismenu
D.
Wise
M.J.
Chevet
E.
Role of pro-oncogenic protein disulfide isomerase (PDI) family member anterior gradient 2 (AGR2) in the control of endoplasmic reticulum homeostasis
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
44855
-
44868
)
[PubMed]
3
Salmans
M.L.
Zhao
F.
Andersen
B.
The estrogen-regulated anterior gradient 2 (AGR2) protein in breast cancer: a potential drug target and biomarker
Breast Cancer Res.
2013
, vol. 
15
 pg. 
204
 
[PubMed]
4
Brychtova
V.
Vojtesek
B.
Hrstka
R.
Anterior gradient 2: a novel player in tumor cell biology
Cancer Lett.
2011
, vol. 
304
 (pg. 
1
-
7
)
[PubMed]
5
Fletcher
G.C.
Patel
S.
Tyson
K.
Adam
P.J.
Schenker
M.
Loader
J.A.
Daviet
L.
Legrain
P.
Parekh
R.
Harris
A.L.
Terrett
J.A.
hAG-2 and hAG-3, human homologues of genes involved in differentiation, are associated with oestrogen receptor-positive breast tumours and interact with metastasis gene C4.4a and dystroglycan
Br. J. Cancer
2003
, vol. 
88
 (pg. 
579
-
585
)
[PubMed]
6
Dong
A.
Gupta
A.
Pai
R.K.
Tun
M.
Lowe
A.W.
The human adenocarcinoma-associated gene, AGR2, induces expression of amphiregulin through Hippo pathway co-activator YAP1 activation
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
18301
-
18310
)
[PubMed]
7
Di Valentin
E.
Crahay
C.
Garbacki
N.
Hennuy
B.
Gueders
M.
Noel
A.
Foidart
J.M.
Grooten
J.
Colige
A.
Piette
J.
Cataldo
D.
New asthma biomarkers: lessons from murine models of acute and chronic asthma
Am. J. Physiol. Lung Cell. Mol. Physiol.
2009
, vol. 
296
 (pg. 
L185
-
L197
)
[PubMed]
8
Schroeder
B.W.
Verhaeghe
C.
Park
S.W.
Nguyenvu
L.T.
Huang
X.
Zhen
G.
Erle
D.J.
AGR2 is induced in asthma and promotes allergen-induced mucin overproduction
Am. J. Respir. Cell Mol. Biol.
2012
, vol. 
47
 (pg. 
178
-
185
)
[PubMed]
9
Kaser
A.
Zeissig
S.
Blumberg
R.S.
Inflammatory bowel disease
Annu. Rev. Immunol.
2010
, vol. 
28
 (pg. 
573
-
621
)
[PubMed]
10
Zheng
W.
Rosenstiel
P.
Huse
K.
Sina
C.
Valentonyte
R.
Mah
N.
Zeitlmann
L.
Grosse
J.
Ruf
N.
Nurnberg
P.
, et al. 
Evaluation of AGR2 and AGR3 as candidate genes for inflammatory bowel disease
Genes Immun.
2006
, vol. 
7
 (pg. 
11
-
18
)
[PubMed]
11
Zhao
F.
Edwards
R.
Dizon
D.
Afrasiabi
K.
Mastroianni
J.R.
Geyfman
M.
Ouellette
A.J.
Andersen
B.
Lipkin
S.M.
Disruption of Paneth and goblet cell homeostasis and increased endoplasmic reticulum stress in Agr2−/− mice
Dev. Biol.
2010
, vol. 
338
 (pg. 
270
-
279
)
[PubMed]
12
Park
S.W.
Zhen
G.
Verhaeghe
C.
Nakagami
Y.
Nguyenvu
L.T.
Barczak
A.J.
Killeen
N.
Erle
D.J.
The protein disulfide isomerase AGR2 is essential for production of intestinal mucus
Proc. Natl. Acad. Sci. U.S.A.
2009
, vol. 
106
 (pg. 
6950
-
6955
)
[PubMed]
13
Colgan
S.P.
Taylor
C.T.
Hypoxia: an alarm signal during intestinal inflammation
Nat. Rev. Gastroenterol. Hepatol.
2010
, vol. 
7
 (pg. 
281
-
287
)
[PubMed]
14
Hirota
S.A.
Beck
P.L.
MacDonald
J.A.
Targeting hypoxia-inducible factor-1 (HIF-1) signaling in therapeutics: implications for the treatment of inflammatory bowel disease
Recent Pat. Inflamm. Allergy Drug Discov.
2009
, vol. 
3
 (pg. 
1
-
16
)
[PubMed]
15
Furuta
G.T.
Turner
J.R.
Taylor
C.T.
Hershberg
R.M.
Comerford
K.
Narravula
S.
Podolsky
D.K.
Colgan
S.P.
Hypoxia-inducible factor 1-dependent induction of intestinal trefoil factor protects barrier function during hypoxia
J. Exp. Med.
2001
, vol. 
193
 (pg. 
1027
-
1034
)
[PubMed]
16
Synnestvedt
K.
Furuta
G.T.
Comerford
K.M.
Louis
N.
Karhausen
J.
Eltzschig
H.K.
Hansen
K.R.
Thompson
L.F.
Colgan
S.P.
Ecto-5′-nucleotidase (CD73) regulation by hypoxia-inducible factor-1 mediates permeability changes in intestinal epithelia
J. Clin. Invest.
2002
, vol. 
110
 (pg. 
993
-
1002
)
[PubMed]
17
Louis
N.A.
Hamilton
K.E.
Canny
G.
Shekels
L.L.
Ho
S.B.
Colgan
S.P.
Selective induction of mucin-3 by hypoxia in intestinal epithelia
J. Cell. Biochem.
2006
, vol. 
99
 (pg. 
1616
-
1627
)
[PubMed]
18
Robinson
A.
Keely
S.
Karhausen
J.
Gerich
M.E.
Furuta
G.T.
Colgan
S.P.
Mucosal protection by hypoxia-inducible factor prolyl hydroxylase inhibition
Gastroenterology
2008
, vol. 
134
 (pg. 
145
-
155
)
[PubMed]
19
Semenza
G.L.
Hypoxia-inducible factors in physiology and medicine
Cell
2012
, vol. 
148
 (pg. 
399
-
408
)
[PubMed]
20
Cummins
E.P.
Doherty
G.A.
Taylor
C.T.
Hydroxylases as therapeutic targets in inflammatory bowel disease
Lab. Invest.
2013
, vol. 
93
 (pg. 
378
-
383
)
[PubMed]
21
Besnard
V.
Wert
S.E.
Hull
W.M.
Whitsett
J.A.
Immunohistochemical localization of Foxa1 and Foxa2 in mouse embryos and adult tissues
Gene Expr. Patterns
2004
, vol. 
5
 (pg. 
193
-
208
)
[PubMed]
22
van der Sluis
M
Vincent
A.
Bouma
J.
Korteland-Van Male
A.
van Goudoever
J.B.
Renes
I.B.
Van Seuningen
I.
Forkhead box transcription factors Foxa1 and Foxa2 are important regulators of Muc2 mucin expression in intestinal epithelial cells
Biochem. Biophys. Res. Commun.
2008
, vol. 
369
 (pg. 
1108
-
1113
)
[PubMed]
23
Ye
D.Z.
Kaestner
K.H.
Foxa1 and Foxa2 control the differentiation of goblet and enteroendocrine L- and D-cells in mice
Gastroenterology
2009
, vol. 
137
 (pg. 
2052
-
2062
)
[PubMed]
24
Kimmel
C.B.
Ballard
W.W.
Kimmel
S.R.
Ullmann
B.
Schilling
T.F.
Stages of embryonic development of the zebrafish
Dev. Dyn.
1995
, vol. 
203
 (pg. 
253
-
310
)
[PubMed]
25
Kawakami
K.
Takeda
H.
Kawakami
N.
Kobayashi
M.
Matsuda
N.
Mishina
M.
A transposon-mediated gene trap approach identifies developmentally regulated genes in zebrafish
Dev. Cell.
2004
, vol. 
7
 (pg. 
133
-
144
)
[PubMed]
26
Urasaki
A.
Morvan
G.
Kawakami
K.
Functional dissection of the Tol2 transposable element identified the minimal cis-sequence and a highly repetitive sequence in the subterminal region essential for transposition
Genetics
2006
, vol. 
174
 (pg. 
639
-
649
)
[PubMed]
27
Peng
K.C.
Pan
C.Y.
Chou
H.N.
Chen
J.Y.
Using an improved Tol2 transposon system to produce transgenic zebrafish with epinecidin-1 which enhanced resistance to bacterial infection
Fish Shellfish Immunol.
2010
, vol. 
28
 (pg. 
905
-
917
)
[PubMed]
28
Chatterjee
S.
Bourque
G.
Lufkin
T.
Conserved and non-conserved enhancers direct tissue specific transcription in ancient germ layer specific developmental control genes
BMC Dev. Biol.
2011
, vol. 
11
 pg. 
63
 
[PubMed]
29
Rastegar
S.
Albert
S.
Le Roux
I.
Fischer
N.
Blader
P.
Muller
F.
Strahle
U.
A floor plate enhancer of the zebrafish netrin1 gene requires Cyclops (Nodal) signalling and the winged helix transcription factor FoxA2
Dev. Biol.
2002
, vol. 
252
 (pg. 
1
-
14
)
[PubMed]
30
Dal-Pra
S.
Thisse
C.
Thisse
B.
FoxA transcription factors are essential for the development of dorsal axial structures
Dev. Biol.
2011
, vol. 
350
 (pg. 
484
-
495
)
[PubMed]
31
Barriga
E.H.
Maxwell
P.H.
Reyes
A.E.
Mayor
R.
The hypoxia factor Hif-1alpha controls neural crest chemotaxis and epithelial to mesenchymal transition
J. Cell Biol.
2013
, vol. 
201
 (pg. 
759
-
776
)
[PubMed]
32
Ko
C.Y.
Tsai
M.Y.
Tseng
W.F.
Cheng
C.H.
Huang
C.R.
Wu
J.S.
Chung
H.Y.
Hsieh
C.S.
Sun
C.K.
Hwang
S.P.
, et al. 
Integration of CNS survival and differentiation by HIF2alpha
Cell Death Differ.
2011
, vol. 
18
 (pg. 
1757
-
1770
)
[PubMed]
33
Peng
M.Y.
Wen
H.J.
Shih
L.J.
Kuo
C.M.
Hwang
S.P.
Myosin heavy chain expression in cranial, pectoral fin, and tail muscle regions of zebrafish embryos
Mol. Reprod. Dev.
2002
, vol. 
63
 (pg. 
422
-
429
)
[PubMed]
34
Chen
Y.C.
Lu
Y.F.
Li
I.C.
Hwang
S.P.
Zebrafish Agr2 is required for terminal differentiation of intestinal goblet cells
PLoS One
2012
, vol. 
7
 pg. 
e34408
 
[PubMed]
35
Chen
Y.H.
Lu
Y.F.
Ko
T.Y.
Tsai
M.Y.
Lin
C.Y.
Lin
C.C.
Hwang
S.P.
Zebrafish cdx1b regulates differentiation of various intestinal cell lineages
Dev. Dyn.
2009
, vol. 
238
 (pg. 
1021
-
1032
)
[PubMed]
36
Tsai
M.Y.
Lu
Y.F.
Liu
Y.H.
Lien
H.W.
Huang
C.J.
Wu
J.L.
Hwang
S.L.
Modulation of p53 and met expression by Kruppel-like factor 8 regulates zebrafish cerebellar development
Dev. Neurobiol.
2014
, vol. 
75
 (pg. 
908
-
926
)
[PubMed]
37
Tang
C.H.
Lai
Y.R.
Chen
Y.C.
Li
C.H.
Lu
Y.F.
Chen
H.Y.
Lien
H.W.
Yang
C.H.
Huang
C.J.
Wang
C.Y.
, et al. 
Expression of zebrafish anterior gradient 2 in the semicircular canals and supporting cells of otic vesicle sensory patches is regulated by Sox10
Biochim. Biophys. Acta
2014
, vol. 
1839
 (pg. 
425
-
437
)
[PubMed]
38
Zeng
L.
Childs
S.J.
The smooth muscle microRNA miR-145 regulates gut epithelial development via a paracrine mechanism
Dev. Biol.
2012
, vol. 
367
 (pg. 
178
-
186
)
[PubMed]
39
Schwend
T.
Ahlgren
S.C.
Zebrafish con/disp1 reveals multiple spatiotemporal requirements for Hedgehog-signaling in craniofacial development
BMC Dev. Biol.
2009
, vol. 
9
 pg. 
59
 
[PubMed]
40
Rytkonen
K.T.
Prokkola
J.M.
Salonen
V.
Nikinmaa
M.
Transcriptional divergence of the duplicated hypoxia-inducible factor alpha genes in zebrafish
Gene
2014
, vol. 
541
 (pg. 
60
-
66
)
[PubMed]
41
Karam
S.M.
Lineage commitment and maturation of epithelial cells in the gut
Front. Biosci.
1999
, vol. 
4
 (pg. 
D286
-
D298
)
[PubMed]
42
Specian
R.D.
Oliver
M.G.
Functional biology of intestinal goblet cells
Am. J. Physiol.
1991
, vol. 
260
 (pg. 
C183
-
C193
)
[PubMed]
43
Cho
N.L.
Javid
S.H.
Carothers
A.M.
Redston
M.
Bertagnolli
M.M.
Estrogen receptors alpha and beta are inhibitory modifiers of Apc-dependent tumorigenesis in the proximal colon of Min/+mice
Cancer Res.
2007
, vol. 
67
 (pg. 
2366
-
2372
)
[PubMed]
44
Wallace
K.N.
Pack
M.
Unique and conserved aspects of gut development in zebrafish
Dev. Biol.
2003
, vol. 
255
 (pg. 
12
-
29
)
[PubMed]
45
Wright
T.M.
Wardell
S.E.
Jasper
J.S.
Stice
J.P.
Safi
R.
Nelson
E.R.
McDonnell
D.P.
Delineation of a FOXA1/ERalpha/AGR2 regulatory loop that is dysregulated in endocrine therapy-resistant breast cancer
Mol. Cancer Res.
2014
, vol. 
12
 (pg. 
1829
-
1839
)
[PubMed]
46
Wang
D.H.
Tiwari
A.
Kim
M.E.
Clemons
N.J.
Regmi
N.L.
Hodges
W.A.
Berman
D.M.
Montgomery
E.A.
Watkins
D.N.
Zhang
X.
, et al. 
Hedgehog signaling regulates FOXA2 in esophageal embryogenesis and Barrett's metaplasia
J. Clin. Invest.
2014
, vol. 
124
 (pg. 
3767
-
3780
)
[PubMed]
47
Mango
S.E.
The C. elegans pharynx: a model for organogenesis. WormBook
2007
(pg. 
1
-
26
)
[PubMed]
48
Zaret
K.S.
Carroll
J.S.
Pioneer transcription factors: establishing competence for gene expression
Genes Dev.
2011
, vol. 
25
 (pg. 
2227
-
2241
)
[PubMed]
49
Eeckhoute
J.
Lupien
M.
Meyer
C.A.
Verzi
M.P.
Shivdasani
R.A.
Liu
X.S.
Brown
M.
Cell-type selective chromatin remodeling defines the active subset of FOXA1-bound enhancers
Genome Res.
2009
, vol. 
19
 (pg. 
372
-
380
)
[PubMed]
50
Fakhouri
T.H.
Stevenson
J.
Chisholm
A.D.
Mango
S.E.
Dynamic chromatin organization during foregut development mediated by the organ selector gene PHA-4/FoxA
PLoS Genet.
2010
, vol. 
6
 pg. 
e1001060
 
[PubMed]
51
Karhausen
J.
Furuta
G.T.
Tomaszewski
J.E.
Johnson
R.S.
Colgan
S.P.
Haase
V.H.
Epithelial hypoxia-inducible factor-1 is protective in murine experimental colitis
J. Clin. Invest.
2004
, vol. 
114
 (pg. 
1098
-
1106
)
[PubMed]
52
Qi
J.
Nakayama
K.
Cardiff
R.D.
Borowsky
A.D.
Kaul
K.
Williams
R.
Krajewski
S.
Mercola
D.
Carpenter
P.M.
Bowtell
D.
Ronai
Z.A.
Siah2-dependent concerted activity of HIF and FoxA2 regulates formation of neuroendocrine phenotype and neuroendocrine prostate tumors
Cancer Cell
2010
, vol. 
18
 (pg. 
23
-
38
)
[PubMed]

Supplementary data