Chromatin remodeling mediated by the FOXA1/A2 transcription factors activates CFTR expression in intestinal epithelial cells

Abstract

The forkhead box A transcription factors, FOXA1 and FOXA2, function as pioneer factors to open condensed chromatin and facilitate binding of other proteins. We showed previously that these factors are key components of a transcriptional network that drives enhancer function at the cystic fibrosis transmembrane conductance regulator (CFTR) locus in intestinal epithelial cells. The CFTR promoter apparently lacks tissue-specific regulatory elements and expression of the gene is controlled by multiple cis-acting elements, which coordinate gene expression in different cell types. Here we show that concurrent depletion of FOXA1 and FOXA2 represses CFTR expression and alters the three-dimensional architecture of the active locus by diminishing interactions between the promoter and intronic cis-acting elements. Reduction of FOXA1/A2 also modifies the enrichment profile of the active enhancer marks H3K27ac and H3K4me2 across the CFTR locus and alters chromatin accessibility at individual cis-elements. Moreover, loss of FOXA1/A2 suppresses the recruitment of other members of the transcriptional network including HNF1 and CDX2, to multiple cis-elements. These data reveal a complex molecular mechanism underlying the role of FOXA1/A2 in achieving high levels of CFTR expression in intestinal epithelial cells.

Introduction

The complex mechanisms that confer cell-specific expression on individual genes are often orchestrated by a series of cis-regulatory elements that lie outside the gene promoter. These elements may lie within introns or distal to the gene itself in intergenic regions. A toolbox of methodologies to investigate the function and organization of cis-acting elements was provided by the technical and conceptual advances made by the ENCODE consortium.^1,2 The identification of regions of open chromatin provides a starting point for these studies, which may identify enhancers and other critical regulatory elements. Upon activation by trans-acting factors, long-range chromatin loops often bring these distal elements into close association with the gene promoters they regulate.³

Cell-specific expression of the cystic fibrosis transmembrane conductance regulator gene (CFTR), which when mutated causes cystic fibrosis (CF), is largely controlled by distal cis-acting elements.⁴ We used open chromatin analysis by DNase-chip and DNase-seq and chromosome conformation capture (3C) to integrate our earlier data on individual CFTR cis-regulatory elements^5-9 with whole locus analysis.^10-13 We showed that several intronic enhancers associate with the active CFTR promoter in intestinal and male genital duct epithelial cells, by a looping mechanism.¹⁰ However, expression of CFTR in the airway epithelium is driven by a different set of cis-elements and a divergent regulatory mechanism.^13,14

In intestinal epithelial cells, a cell-type selective network of transcription factors (TFs) regulates CFTR expression.¹⁵ These factors, forkhead box A family members (FOXA1/A2), hepatocyte nuclear factor 1 (HNF1), and caudal-type homeobox 2 (CDX2), were shown to interact in vitro and in vivo with CFTR regulatory element sequences, and contribute to the intestine-selective enhancer function of a regulatory element located in intron 11 (note in this manuscript we use CFTR legacy nomenclature for numbering introns) at 1811 + 0.8 kb (where 1811 is the last coding base in exon 11).^10,15 Moreover, HNF1 was shown previously to be a key component of the TF complex interacting with cis-elements in the first (185 +10kb)¹² and tenth (1716 + 13.2, + 13.7 and +23 kb)⁸ introns of the locus (see Table 1 for relevant CFTR cis-element genomic locations).

Table 1.CFTR DHS locations.

DHS	Location in CFTR locus	hg19 coordinates
–20.9 kb1	5'	chr7:117099050-117099445
185 + 10 kb	intron 1	chr7:117129649-117130749
1716 + 13.2/13.7 kb	intron 10 (a,b)	chr7:117212349-117213649
1716 + 23 kb	intron 10 (c)	chr7:117222349-117223049
1811 + 0.8 kb	intron 11	chr7:117228049-117229449
+15.6 kb2	3'	chr7:117322649-117324149

¹ Distance upstream of CFTR translational start site. ²Distance downstream of last coding base of CFTR transcript.

Members of the FOXA family of TFs serve as pioneer factors and have the ability to open condensed chromatin independent of SWI/SNF chromatin remodelers, in part due to structural homology with the H1-linker histone.^16-18 Here, we show that FOXA1/A2 regulate intestinal expression of CFTR by altering the chromatin structure of the active locus, modifying histone marks, and facilitating recruitment of other cell-specific TFs. Moreover, using siRNA-mediated depletion of FOXA1/A2 we demonstrate that these factors are critical for the formation of long-range chromatin loops that bring distal cis-regulatory elements in close proximity to the transcriptional complex at the CFTR promoter.

Results

FOXA1/A2 contribute to expression of CFTR in intestinal epithelial cells

FOXA1/A2, which bind the same consensus sequence, were shown previously to associate with the CFTR locus in the intestinal epithelial cell lines Caco2 and HT29,¹⁵ which express high levels of CFTR. In both lines FOXA1 and FOXA2 were highly enriched at DNase I hypersensitive sites (DHS) in introns 10 (1716 + 13.2/13.7 kb, DHS10a,b) and 11 (1811 + 0.8 kb).¹⁵ Slight enrichment was also measured at other CFTR cis-regulatory elements, including those in introns 1 (185 + 10 kb) and 23 (4374 + 1.3 kb; FOXA2 only).¹⁵

To determine the contribution of bound FOXA1/A2 to CFTR expression, pools of three siRNAs were used to deplete each protein in Caco2 cells. Reduction of FOXA1 or FOXA2 alone was ineffective at modulating CFTR expression (data not shown). However simultaneous targeting of both factors, which reduced levels of each by 80–90% in Caco2 cells (Fig. 1A and B), caused a 40% reduction in CFTR expression (Fig. 1B). These data confirm the functional importance of FOXA1/A2 in facilitating intestinal cell expression of CFTR.

Figure 1. FOXA1/A2 contribute to CFTR expression in vivo and influence long-range interactions across the CFTR locus. Caco2 cells were reverse-transfected for 72 h with negative control (NC) siRNA or siRNAs targeting FOXA1 or FOXA2 (in combination). Western blots of whole-cell lysate for total FOXA1, FOXA2, or β-tubulin (A). FOXA1 and FOXA2 protein levels were normalized to β-tubulin, and are shown relative to the NC samples (B, left panel). CFTR mRNA levels were normalized to 18S rRNA, and are shown relative to the NC samples (B, right panel). Values are means ± S.E.M., n = 9; ***P < 0.001 and N.S., not significant, using an unpaired Student’s t test. CFTR promoter and DHS interactions in Caco2 cells treated with NC (gray bars) or FOXA1/A2 specific (black bars) siRNAs for 72 h, as measured by q3C (C). A forward primer and TaqMan probe for a HindIII fragment spanning the CFTR promoter were used as bait (dotted line). Reverse primers were designed in other HindIII fragments spanning the CFTR locus. The y-axis indicates interaction frequency between two HindIII fragments relative to ubiquitously expressed ERCC3. Values are means ± SEM for duplicate PCRs; n = 3 with results shown from a representative 3C experiment.

Looping of distal cis-regulatory elements to the CFTR promoter requires FOXA1/A2

We previously showed by quantitative 3C (q3C) that the active CFTR locus is organized into a looped structure. Strong interactions were detected between the CFTR promoter and the middle of the locus encompassing introns 10 and 11, and also a region containing the +6.8 kb and +15.6 kb insulator elements (distances 3′ to the last coding base of the CFTR transcript).^10,11 These interactions were seen in Caco2 cells, and to a lesser extent in HT29 and primary epididymis cells.¹⁰

To determine whether FOXA1/A2 depletion in Caco2 cells altered the conformation of the locus q3C was performed. Simultaneous siRNA-mediated depletion of both factors caused a marked decrease in the interaction frequency between the CFTR promoter and distal regions of the locus. In the middle of the locus (introns 10 and 11) a 31–67% decrease was evident and the 3′ insulator region showed a 47% decrease (Fig. 1C). These data show that the establishment and/or maintenance of CFTR looping in intestinal epithelial cells is dependent on FOXA1/A2. Next, to elucidate the causative mechanism, we first investigated whether FOXA1/A2 enrichment at cis-regulatory elements in the locus modulates histone marks and subsequently the interactions of other cell-type specific TFs.

FOXA1/A2 alter histone modifications at active CFTR cis-regulatory elements

It is known that FOXA TFs, as pioneer factors, can facilitate the binding of additional factors by modifying chromatin accessibility.¹⁷ Hence, we examined the chromatin landscape across the CFTR locus in Caco2 cells by chromatin immunoprecipitation (ChIP), following siRNA-mediated depletion of FOXA1/A2. Since FOXA1/A2 bind the same consensus sequences and show similar enrichment patterns across the CFTR locus,¹⁵ FOXA2 binding alone was evaluated following siRNA treatment. Initial experiments confirmed that FOXA2 enrichment decreased at cis-regulatory elements across the CFTR locus in FOXA1/A2 knockdown cells (Fig. 2A). Sites with the highest FOXA2 enrichment in control cells (5–10-fold), including DHS –20.9 kb, an element 5′ to the CFTR translational start site, which encompasses an enhancer-blocking insulator,⁹ enhancers associated with DHS in introns 1 and 11,^10,12 and other cis-elements in intron 10 (10a,b)^7,10 showed a 29–67% reduction in FOXA2 enrichment following siRNA treatment. Concurrently, histone modifications that mark enhancer elements were examined. These included histone H3 lysine 27 acetylation (H3K27ac), which marks active enhancers and histone H3 lysine 4 dimethylation (H3K4me2), which marks both active and poised enhancers.^19,20 H3K27ac levels fell by 33–72% at cis-elements at -20.9 kb, in introns 10 (a,b, and c) and 11, and at +15.6 kb, though only by 10% at the intron 1 enhancer (Fig. 2B). H3K4me2 levels increased, decreased, or remained the same at individual sites after FOXA1/A2 knockdown (Fig. 2C). Specifically, the levels of H3K4me2 remained unchanged at the –20.9 kb and +15.6 kb sites, while levels decreased at the intron 10a,b DHS by 50% (Fig. 2C). Interestingly, at the CFTR promoter and the cis-elements in introns 1, 10c, and 11, H3K4me2 levels increased by 20–40% following FOXA1/A2 knockdown. These data suggest that FOXA1/A2 are critical for the function of multiple enhancers across CFTR and that loss of these factors drives specific CFTR enhancers from an active to a poised state.

Figure 2. Changes in transcription factor recruitment and associated histone modifications upon FOXA1/A2 depletion. ChIP with FOXA2- (A), HNF1α/β- (B), CDX2- (C), H3K3me2- (D), or H3K27ac- (E), specific antibodies and chromatin from Caco2 cells treated with NC (gray bars) or FOXA1/A2 (black bars) specific siRNAs for 72 h. Primer sets spanning the CFTR locus were used to assay enrichment by SYBR Green qPCR analysis. Results were normalized to 18s rRNA levels and compared with IgG enrichment at the same site (broken line) (A–C), or shown as percent input (D and E). Values are means ± S.E.M., for duplicate PCRs; n = 3 with data shown from a representative ChIP experiment. See Table 1 for genomic locations of the DHS shown.

FOXA1/A2 are required for recruitment of other intestinal transcription factors to multiple CFTR cis-elements

We next evaluated enrichment of other members of the intestine-selective TF network (CDX2 and HNF1) that we previously showed to interact with CFTR enhancers.¹⁵ At many of the sites that lost FOXA2 after FOXA1/A2 depletion (Fig. 2A), we also observed a decrease in HNF1 and CDX2 (Fig. 2D and E). HNF1 enrichment fell by 58–78% at the –20.9 kb and +15.6 kb insulators and at cis-elements in intron 10 (at DHS10a,b and DHS10c) (Fig. 2D). However, at the intron 1 enhancer, which binds relatively modest amounts of FOXA2 but is highly enriched for HNF1, FOXA1/A2 depletion reduces FOXA2 binding but has no impact on HNF1 enrichment (Fig. 2A and D). Depletion of FOXA1/A2 had no effect on HNF1 protein levels (Fig. S1). Loss of FOXA1/2 also diminishes CDX2 enrichment by 80% at the DHS10a,b elements and 23–41% at the DHS10c element, the intron 11 enhancer, and the +15.6 kb insulator (Fig. 2E). CDX2 protein levels are reduced ~60% by combined FOXA1/A2 knockdown in Caco2 cells (Fig. S1). However, siRNA-mediated depletion of CDX2 protein (>90%) has a lesser effect on CDX2 occupancy measured by ChIP across the CFTR locus than does loss of FOXA1/A2 (data not shown). These results suggest that recruitment of HNF1 to CFTR cis-elements is dependent on FOXA1/A2 at the majority of cis-regulatory elements that bind these pioneer factors, while recruitment of CDX2 to CFTR cis-elements is likely both directly and indirectly controlled by FOXA1/A2.

CpG methylation in DHS11 and DHS10a,b does not correlate with FOXA1/A2 binding

An inverse relationship was described recently between DNA methylation and FOXA1 binding at cell-type specific enhancers genome-wide.²¹ To determine whether DNA methylation influenced the FOXA1/A2-dependent regulation of CFTR expression, we evaluated the methylation status of CpGs within the DHS10a,b and DHS11 elements. Two CpGs were investigated in DHS10a,b (hg19 chr7:117212602–117212603 and 117212678–117212679), which are between two predicted FOXA binding sites, together with two CpGs in DHS11 (hg19 chr7:117228597–117228598 and 117228958–117228959) close to the binding sites for key TFs.¹⁵ Bisulfite modified genomic DNA from several cell types (skin fibroblasts, A549 lung cancer cells, NHBE primary airway epithelial cells, Caco2, or MCF7) was sequenced. However, no correlation was found between FOXA1/A2 enrichment, CFTR expression levels, or DNA methylation at any of the CpG sites examined (Fig. 3). In Caco2 cells, which are strongly enriched for FOXA1 at DHS10a,b and DHS11,^15,22 CpGs are methylated in intron 10 (Fig. 3A) and unmethylated in intron 11 (Fig. 3B). In NHBE cells, which express low levels of CFTR and FOXA1/A2, though these factors do not bind to the cis-elements at DHS10a,b and DHS11 in these cells,²² the CpGs examined were either unmethylated or partially methylated (Fig. 3). Thus, DNA methylation, measured by modification of CpGs, does not appear to play a major role in regulating the activity of FOXA-dependent CFTR enhancers.

Figure 3. DNA methylation status of CpGs within CFTR intronic regulatory regions. Skin fibroblast, A549, NHBE, Caco2, or MCF7 genomic DNA was bisulfite converted, and regions spanning DHS10(a,b) (A) or DHS11 (B) were PCR-amplified, cloned, and sequenced to determine methylation status of indicated CpGs. FOXA consensus binding sites (as predicted by MatInspector). The number of clones sequenced and average percent CpG methylation for various cell types is indicated, all clones had >99% conversion rate. CFTR expression levels determined by qRT-PCR, FOXA1/A2 protein expression determined by western blot (not shown).

FOXA1/A2 influence chromatin accessibility of DHS in intron 10 of CFTR

Since FOXA family members have chromatin-remodeling activity, we next looked for changes in the DNase I hypersensitivity profile of the CFTR locus following FOXA1/A2 depletion. The DNase I accessibility of the CFTR promoter and of cis-elements that are either strongly enriched for (DHS10a,b and DHS11) or have little or no FOXA1/A2 binding (DHS1 and DHS10c) (Fig. 2A)¹⁵ was examined (Fig. 4A). In FOXA1/A2 depleted Caco2 cells DHS10a,b loses all DNase I hypersensitivity in comparison to negative control cells (Fig. 4B and C). In contrast, only a slight reduction was observed in Caco2 cells with reduced CDX2 levels (Fig. S2). No appreciable changes in hypersensitivity of the CFTR promoter or cis-regulatory elements at DHS1, DHS10c, or DHS11 were observed with depletion of either factor (Fig. 4B and C; Fig. S2). These data demonstrate that FOXA1/A2 are crucial for the establishment and/or maintenance of the open chromatin state of the cis-regulatory elements at DHS10a,b.

Figure 4. Changes in CFTR DNase I hypersensitivity profile upon FOXA1/A2 KD in Caco2. Relative DNase I hypersensitivity of CFTR cis-regulatory regions (A) in Caco2 cells treated with NC (B) or FOXA1/A2- (C) specific siRNAs for 72 h. Following purification of genomic DNA digested with increasing amounts of DNase I, SYBR Green qPCR was performed comparing the DNase I hypersensitivity of known Caco2 hypersensitive sites to that of a non-hypersensitive region in intron 1 (DHS1+5 kb). Values are means ± SEM for duplicate PCRs; n = 3 with results shown from a representative experiment.

Functional interaction of FOXA-binding elements within different DHS in the middle of the CFTR locus

Four DHS in the middle of the CFTR locus, within introns 10 and 11, are associated with high levels of gene expression. The element in the intron 11 DHS is a strong, cell-type specific enhancer, regulated by HNF1, CDX2, and FOXA1/A2.^10,15 This DHS is rarely evident in isolation and is usually seen with one or more DHS in intron 10 (DHS10a,b and DHS10c).²³ The intron 10 DHS are not associated with enhancer activity,^7,10 though as observed above, chromatin accessibility at the DHS10a,b elements is highly dependent on FOXA1/A2 in Caco2 cells (Figs. 1C, 2, and 4). Moreover, we also showed previously that another member of the intestinal TF network, HNF1, binds to this cluster of elements in vitro⁸ and in vivo.^10,15 To determine whether there was a functional interaction between the DHS10a,b cis-elements and the enhancer in DHS11, we used transient transfections of a reporter gene construct in which the CFTR basal promoter drives luciferase expression and cis-elements are cloned into the enhancer site. A 1.5 kb sequence encompassing DHS10a,b, which has no independent enhancer activity, caused an ~7-fold increase in luciferase expression from the DHS11 enhancer when combined with it. In contrast a 1.5 kb fragment from intron 3 (405 + 13.1 kb) of CFTR, that encompasses no DHS, had no effect when combined with either the DHS10a,b or DHS11 elements in the same construct (Fig. 5). Mutation of two of three predicted FOXA binding sites in DHS10a,b and one critical FOXA binding site in DHS11¹⁵ did not reduce luciferase activity of the combined construct (data not shown). However three additional predicted FOXA sites in DHS11 may substitute for the mutated sites. These data suggest that the cis-acting elements in DHS10a,b recruit factors that facilitate the enhancer function of the DHS element in intron 11.

Figure 5. A regulatory element in DHS10a,b works cooperatively with the intron 11 enhancer element. Caco2 cells were transfected with luciferase reporter vectors containing the 787 bp basal CFTR promoter (pGL3B.245) and 1.5 kb regions spanning a fragment of intron 3 that lack a DHS or the DHS in introns 10a,b (DHS10a,b) or 11 cloned into the enhancer site of the vector in forward orientations, alone or in combination. Luciferase expression is shown relative to the CFTR basal promoter-only vector. Error bars represent SEM (n = 9). *** denotes P < 0.001; N.S. denotes not significant using an unpaired t test.

Discussion

In earlier work, we showed that the pioneer TFs FOXA1/A2 are enriched at CFTR cis-regulatory elements in intestinal epithelial cells and are key regulators of a transcriptional enhancer located in intron 11 of the gene.¹⁵ Here we describe a FOXA1/A2-dependent mechanism by which high levels of CFTR expression are maintained in intestinal cells through epigenetic chromatin modifications at cis-regulatory sites.

Cell-specific factors bind to enhancers to facilitate their direct physical interactions with gene promoters by looping mechanisms, which are well characterized in a number of genes and result in enhanced transcription.²⁰ 3C experiments have confirmed that this model applies to expression of the CFTR gene.^10-13,24 In intestinal cells, the CFTR promoter physically associates with regulatory elements located in the middle of the gene, and with flanking insulators.¹⁰ FOXA1 is important for chromatin loop formation at other loci, and is detected at numerous estrogen receptor-mediated chromatin loops in breast cancer cells.^25-27 Moreover, a breast cancer risk-associated SNP alters the binding affinity of FOXA1 at an enhancer element, which loops to the TOX high mobility group box family member 3 (TOX3) gene promoter. The SNP decreases both enhancer activity and TOX3 expression.²⁸ Here we demonstrate that FOXA1 and FOXA2 directly mediate looping between the CFTR promoter and cis-elements in introns 10 and 11 of the CFTR gene, in Caco2 cells. This is in contrast to previous data, which shows an indirect role for FOXA protein in mediating long-range chromatin loops that are established or maintained by other factors.^25-27

In previous work we showed that several intestinal enhancers on the CFTR locus interact with the same network of TFs including FOXA1/2, HNF1, and CDX2. Consistent with these data are the current observations that FOXA1/A2 are necessary for the activity of multiple CFTR cis-regulatory elements, as measured by H3K27ac enrichment levels. H3K4me2, a chromatin mark associated with both active and poised enhancers,^29,30 is important for the recruitment of FOXA2 to developmental enhancers.³¹ However, we observed no consistent changes in H3K4me2 enrichment at the enhancers in CFTR. These data are consistent with other reports of H3K4me1/2 enrichment at FOXA1-bound sites following silencing of FOXA1 in MCF7 and LNCaP cells.³¹

As observed at other loci,^25,31,32 we demonstrate a dependence of other cell-specific TF (HNF1 and CDX2) binding at FOXA1/2-bound enhancer elements. Also of interest are the relatively modest changes that are seen after FOXA1/2 depletion in the histone modification profiles at the –20.9 kb and +15.6 kb enhancer-blocking insulator elements and the intron 1 enhancer. These elements also bind FOXA1/A2 but are apparently less dependent on FOXA1/A2 than the DHS10a,b and DHS11 elements, suggesting that they function by mechanisms independent or upstream of FOXA1/A2. Our previous findings that the insulators interact with CTCF and RXR-VDR¹¹ and that the activity of the intron 1 enhancer is largely dependent on HNF1^5,7,12 support this conclusion. Alternatively, the changes observed at these sites following FOXA1/A2 depletion might be indirect, and reflect the changes to the conformation of the locus in these cells.

The FOXA factors have chromatin remodeling abilities, and are required, but not sufficient, for the opening and activation of specific developmental enhancers.³³ At the CFTR cis-elements, only DHS10a,b DNase I hypersensitivity appears to be directly impacted by levels of FOXA1/A2. Hypersensitivity of these elements also appears to be partially dependent on CDX2, suggesting that though FOXA1/A2 are critical, a complex of factors may maintain the accessibility of these sites. Additional factors, including CTCF and GATA3, may precede FOXA binding at some loci^27,34 and it is possible that other factors facilitate direct FOXA binding throughout the CFTR locus. However, our data show that the cluster of cis-elements in the middle of the CFTR gene apparently work together to coordinate intestinal expression of CFTR in a FOXA-dependent manner.

Materials and Methods

Cell culture

Human colorectal carcinoma Caco2 and HT29,³⁵ A549 lung carcinoma,³⁶ and MCF7 breast carcinoma³⁷ cell lines were grown in DMEM (Dulbecco’s modified Eagle’s medium) supplemented with 10% FBS (fetal bovine serum). Primary human skin fibroblasts (Coriell GM08333) were grown in MEM (minimal essential media) with 15% FBS. NHBE, a mixture of primary human bronchial and tracheal epithelial cells (Lonza, CC-2541), were grown according to manufacturer’s recommendations in BEGM (bronchial epithelial cell growth medium).

Transient siRNA knockdown experiments

siRNA knockdown experiments were conducted in Caco2 using 20nM hCDX2 (Santa Cruz Biotechnology, sc-43680), hFOXA1 (Santa Cruz Biotechnology, sc-37930), hFOXA2 (Santa Cruz Biotechnology, sc-35569), or 20nM or 40nM control (Santa Cruz Biotechnology, sc-37007) siRNA as previously described.¹⁵ Cells were harvested after 48 h for CDX2 knockdown, or 72 h for FOXA1/A2 knockdown.

Western blot analysis

Cells were lysed by standard protocols and protein levels were assayed by western blot analysis with antibodies against FOXA1 (Abcam, ab5089), FOXA2 (Santa Cruz Biotechnology, sc-6554x), HNF1 (Santa Cruz Biotechnology, sc-8986x), CDX2 (Bethyl Laboratories, A300-692A), and β-tubulin (Sigma-Aldrich, T4026). Protein quantification was performed using ImageJ software (NIH).

qRT-PCR (quantitative reverse transcription-PCR)

TaqMan® qRT-PCR to measure CFTR expression using primers listed in Table S1 was performed as previously described.^15,38

Quantitative chromosome conformation capture (q3C)

q3C was performed using primers listed in Table S1 as described previously.^11,39

Chromatin immunoprecipitation (ChIP)

ChIP was performed using standard methods as previously described.¹⁵ Immunoprecipitations were performed with antibodies against FOXA2, CDX2 (Bethyl Laboratories, A300-691A), HNF1, H3K4me2 (Millipore, 07-030), H3K27ac (Abcam, ab4729), normal goat IgG (Santa Cruz Biotechnology, sc-2028), or normal rabbit IgG (Millipore, 12-370). TF enrichments were analyzed relative to IgG and histone modification enrichments were analyzed as percent input using SYBR^® Green qPCR (quantitative PCR) with primers listed in Table S1.

DNA methylation analysis

Purified genomic DNA was bisulfite converted using the EZ DNA Methylation^TM Kit (Zymo Research), according to the manufacturer’s recommendations. Amplification of DHS10(a,b) and DHS11 was performed with nested PCR using primers listed in Table S1 and OneTaq® Hot Start DNA Polymerase (New England Biolabs). Amplicons were cloned using the StrataClone PCR Cloning Kit (Agilent Technologies), sequenced, and analyzed using BiQ Analyzer.⁴⁰

DNase I-qPCR

Cells were trypsinized and nuclei isolated as previously reported.⁴¹ Resuspended nuclei were digested with increasing amounts of DNase I (New England Biolabs) for 5–10 min at 37 °C. Digests were stopped by addition (to final concentration) of 75 mM NaCl, 12.5 mM EDTA, and 0.5% SDS. Samples were purified following treatment with proteinase K (Sigma), and DNase I digestion was assessed by agarose gel. SYBR® Green qPCR with primers listed in Table S1 was performed. DNase I sensitivity was quantified relative to undigested DNA for each primer set, and shown relative to digestion of a non-DNase I hypersensitive amplicon, DHS1+5 kb, for each DNase I sample.

Plasmids and transient luciferase assays

The intron 3 (405 + 13.1 kb), DHS10(a,b), and DHS11 fragments¹⁰ were PCR amplified and cloned into the pGL3B luciferase vector (Promega) containing the 787 bp minimal CFTR promoter^5,7 using primers listed in Table S1, and sequence verified. Caco2 transfections and luciferase assays were conducted as previously reported.^12,15

Glossary

Abbreviations:

3C

chromosome conformation capture

CDX2

caudal-type homeobox 2

CF

cystic fibrosis

CFTR

cystic fibrosis transmembrane conductance regulator gene

ChIP

chromatin immunoprecipitation

DHS

DNase I hypersensitive site

FOXA

forkhead box A

H3K27ac

histone H3 lysine 27 acetylation

H3K4me2

histone H3 lysine 4 dimethylation

HNF1

hepatocyte nuclear factor 1

NC

negative control

q3C

quantitative 3C

qPCR

quantitative PCR

qRT-PCR

quantitative reverse transcription PCR

TF

transcription factor

TOX3

TOX high mobility group box family member 3