Introduction

Intrauterine growth retardation (IUGR), a common complication of pregnancy, has been linked to the later development of diseases in adulthood such as type 2 diabetes [1]. We have demonstrated that the abnormal intrauterine milieu associated with IUGR limits the supply of critical substrates and hormones and affects the development of the fetus by permanently modifying gene expression and function of susceptible cells, such as the pancreatic beta cell leading to the development of diabetes in adulthood [2–4].

Pdx1 encodes a homeobox transcription factor critically important for pancreatic beta cell function and development and plays a pivotal role in the development of diabetes in humans and rodents. Even a relatively modest decrease in Pdx1 expression impairs the compensatory response to insulin resistance [2, 4–13]. Expression of Pdx1 is permanently reduced in IUGR beta cells and in previous studies we demonstrated that aberrant histone modifications and later gain of DNA methylation at the proximal promoter of Pdx1 are responsible for decreased Pdx1 transcription [14].

The long-acting GLP-1 analogue, exendin-4 (Ex-4) increases Pdx1 expression in IUGR beta cells [4, 15–18] and is being used to treat humans with type 2 diabetes. The molecular mechanisms by which Ex-4 increases Pdx1 transcription in IUGR animals are unknown. In previous studies, we found that administration of Ex-4 during the prediabetic neonatal period (postnatal days 1–6 [PD 1–6]) prevents the development of diabetes in IUGR animals by restoring expression of Pdx1 to normal levels [4].

One of the earliest molecular events involved in silencing the Pdx1 promoter in islets of IUGR animals is the loss of binding of the critical activator, USF1 [14]. However, USF1 protein levels are not reduced in IUGR animals, and this leads us to posit that IUGR may induce post-translational modifications (e.g. phosphorylation), which prevent the association of USF1 with the Pdx1 promoter. Transcriptional activity of USF1 is dependent on phosphorylation at Ser257 [19–23]; and USF1 purified from rat liver, spleen and kidney is primarily present in the phosphorylated form in the nucleus [20].

In the present study, we hypothesised that normalisation of Pdx1 transcription by Ex-4 treatment is mediated by phosphorylation of USF1, which in turn mediates epigenetic changes involving histone modifications and DNA methylation at the proximal promoter of Pdx1.

Methods

For a complete description of the methods, please see the electronic supplementary material (ESM).

Animal model

Four experimental groups from our previously reported rat model of IUGR were studied [2, 3]: (1) control pups treated with vehicle (PBS); (2) control pups treated with Ex-4 (Bachem, King of Prussia, PA, USA); 1 nmol/kg body weight injected subcutaneously for 6 days starting on day of life 1; (3) IUGR pups treated with vehicle; and (4) IUGR pups treated with Ex-4. On day of life 7, islets were harvested for neonatal experiments. For neonatal experiments, islets were harvested from 16 litters of IUGR and 16 litters of control animals and pooled from one litter for each experiment. Islets were used from 22 male adult animals from different litters for each treatment group. These studies were approved by the Animal Care Committee of The Children’s Hospital of Philadelphia.

Pdx1 mRNA

Total RNA was isolated from islets (n = 5, all from different litters, per group) using RNAzol B (Tel-Test, Friendswood, TX, USA). Quantitative PCR was performed as previously described [4]. Pdx1 expression was normalised to β-actin expression after assessment of a panel of housekeeping genes found this to be the most stable.

Chromatin immunoprecipitation (ChIP) assay

ChIP was performed as previously described [14] using approximately 1,000 neonatal and 500 adult islets per experiment per group. Quantitative PCR was used to measure binding of USF1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), acetylated H3 (Millipore, Billerica, MA, USA), H3K9me2 (Millipore), p300 (Millipore), DNA methyltransferase (Dnmt) 1 (Millipore) and H3K4me3 (AbCAM, Cambridge, MA, USA) at the promoters of Pdx1 and β-actin. Results were expressed as IP per total input for each experimental group and each experiment was normalised to the IP per total input value for the control vehicle experimental group for each antibody. Control vehicle groups were not compared across age groups. Standard methods of ChIP data normalisation were used to normalise individual sample immunoprecipitated DNA to both total input DNA for each experimental group and to a control sequence [24]. For the adult experiments, ChIP was performed at 3–4 months for PCAF, 6 months for Dnmt1 and 9–12 months for AcH3, H3K4me3 and H3K9me2. For the histone modifications, data were independently analysed for the 9- and 12-month ChIP experiments (ESM Fig. 1, ESM Tables 1–3). The individual histone modification profiles were consistent at 9 and 12 months and thus these results were combined for the manuscript.

HAT activity

Islet HAT activity was measured from pooled islet samples from 1-week-old animals, approximately 1,000 islets per group, via colorimetric assay using cofactor acetyl CoA (Kamiya, Seattle, WA, USA). Measurements were made at baseline and at 1, 2 and 3 h. Total HAT activity was computed as area under the curve and normalised to control vehicle.

USF1 phosphorylation and PCAF coimmunoprecipitation

Islets were harvested from animals at 1 week of age and nuclear protein extracts were then prepared from islets. Nuclear extracts (10 μg of protein/reaction) were first immunoprecipitated with normal rabbit serum (negative control) or anti-USF1. Membranes were incubated with anti-phosphoserine (1:500 dilution) or PCAF (1:500 dilution) antibody, and immunoreactive proteins were visualised by incubation with horseradish peroxidase-linked donkey anti-rabbit secondary antibody (1:5,000 dilution) and the ECL Plus western blotting system (GE Healthcare, Piscataway, NJ, USA) according to the manufacturer’s instructions. Densitometric analyses were then performed and values were normalised to control samples treated with vehicle.

Methylation analysis of the Pdx1 CpG island

Genomic DNA was extracted from islets from 6–9 month old animals. Bisulfite modification was done using the Zymo Research EZ Methylation Gold kit (Irvine, CA, USA). Pyrosequencing analysis was performed by EpigenDx (Worchester, MA, USA).

Statistical analyses

Statistical analyses were performed using analysis of variance and Student’s unpaired t test not requiring equal variance between groups. A p value less than 0.05 was considered significant.

Results

Ex-4 treatment restores Pdx1 mRNA levels in IUGR islets

Similar to our previous findings, Ex-4 treatment of newborn IUGR rats significantly increased Pdx1 expression [4]. At 14 days of age, Pdx1 mRNA levels were reduced by 75% in IUGR vehicle-treated rats compared with controls (p < 0.05 vs control vehicle, Fig. 1a). We have previously shown that beta cell mass remains normal at 14 days in the IUGR rat [4], indicating that the reduction in Pdx1 mRNA level at this age is not due to decreased beta cell mass. Neonatal Ex-4 treatment led to a restoration of Pdx1 mRNA levels in IUGR rats at 14 days (Fig. 1a), an effect that persisted into adulthood (Fig. 1b).

Fig. 1
figure 1

Pdx1 expression is normalised by neonatal Ex-4 treatment. a Pdx1 mRNA levels at 2 weeks of age were measured by quantitative PCR as described in Methods (n = 5 per treatment group). b Pdx1 mRNA levels in adult animals (n = 5 per treatment group). *p < 0.05 IV vs IEx; †p < 0.05 IV vs CV; ‡ p < 0.05 IV vs CEx treatment groups; p > 0.10 CV vs CEx and IEx vs CV

Full size image

Ex-4 treatment restores histone acetylation at the Pdx1 promoter

Histone acetylation is associated with active chromatin structure, whereas histone deacetylation is associated with a repressed chromatin state [25]. To determine whether the restoration of Pdx1 transcription in IUGR Ex-4 treated animals was due to increased acetylation of H3 at the proximal promoter of Pdx1, we performed ChIP analyses on pancreatic islets using primers that encompass the region of the CpG island shown in Fig. 2. This area is obligate for Pdx1 transcription. At 1 week of age, there was a significant reduction in histone H3 acetylation at the proximal promoter of Pdx1 in IUGR islets compared with control animals (Fig. 3a). Treatment with Ex-4 increased histone H3 acetylation by approximately ninefold (IUGR vehicle (IV) vs IUGR Ex-4 (IEx); p = 0.042) in IUGR animals at 1 week of age, but had no significant effect in control animals (control vehicle (CV) vs control Ex-4 (CEx): p = 0.46; Fig. 3a) consistent with the lack of change in expression of Pdx1 in response to Ex-4 treatment in control animals. The effects of Ex-4 on H3 acetylation persisted into adulthood in IUGR animals, resulting in a fourfold increase in H3 acetylation (IV vs IEx: p = 0.026; Fig. 3b). Similar to 1-week-old animals, there was no effect of Ex-4 in controls (CV vs CEx; p = 0.177).

Fig. 2
figure 2

Map of promoter region of Pdx1. The proximal promoter of Pdx1 is part of a more extensive CpG island. Filled circles represent CpG dinucleotides and numbers indicate positions relative to the transcriptional start site. The position of the E-box bound by USF1 is indicated

Full size image
Fig. 3
figure 3

Histone H3 acetylation at the proximal promoter of Pdx1. ChIP analysis of cross-linked chromatin from islets of IUGR and control animals with and without Ex-4 treatment at (a) 1 week of age, and (b) adult, immunoprecipitated with antibody to acetylated H3. The IgG and no antibody immunoprecipitate showed negligible PCR product, indicating little or no immunoprecipitation in the absence of primary antibody (see ESM). The relative amount of acetylated H3 bound at the Pdx1 promoter was measured by QPCR and normalised to total input DNA for each experimental group. Data are represented as fold change relative to control vehicle values. n = 6 experiments, data are ±SEM. a *p = 0.004 IV vs CV; †p = 0.042 IV vs IEx; p > 0.05, IEx vs CV and CV vs CEx. b *p = 0.004 IV vs CV; †p = 0.026 IV vs IEx; p > 0.10 IEx vs CV and CV vs CEx

Full size image

Effects of Ex-4 on trimethylation of H3K4

Trimethylation of lysine 4 at H3 (H3K4me3) is associated preferentially at the promoters of active genes and is influenced by the density of acetylated histone H3 [26–28]. In islets from controls at all ages, there was a robust association of H3K4me3 at the Pdx1 promoter. At 1 week, the abundance of H3K4me3 was markedly reduced at the proximal promoter of Pdx1 in IUGR islets, and this reduction was completely prevented by neonatal Ex-4 as shown by the 27-fold increase (IV vs IEx: p = 0.04) in Fig. 4a. This effect persisted with a sixfold increase of H3K4me3 in adulthood (IV vs IEx: p = 0.033; Fig. 4b). Ex-4 had no significant effect on trimethylation of H3K4 in controls at 1 week of age (Fig. 4a; CV vs CEx: p = 0.244). In adults, Ex-4 controls showed an approximate twofold increase in enrichment of H3K4 trimethylation compared with controls (p = 0.035; Fig. 4b).

Fig. 4
figure 4

H3K4 trimethylation and H3K9 dimethylation at the proximal promoter of Pdx1. ChIP analysis of cross-linked chromatin from islets of IUGR and control animals with and without Ex-4 treatment immunoprecipitated with antibody to H3K4me3 and H3K9me2: (a) H3K4me3 at 1 week of age and (b) H3K4me3 adult; (c) H3K9me2 at 1 week of age and (d) H3K9me2 adult. The IgG and no antibody immunoprecipitate showed negligible PCR product, indicating little or no immunoprecipitation in the absence of primary antibody. The relative amount of H3K4me3 and H3K9me2 bound at the Pdx1 promoter was measured by QPCR and normalised to total input DNA for each experimental group. Data are represented as fold change relative to CV values. n = 6 experiments, data are ±SEM. a *p = 0.04 IV vs IEx; †p = 0.001 IV vs CV; p > 0.05 IEx vs CV and CV vs CEx. b *p = 0.033 IV vs IEx; †p = 0.002 IV vs CV; ‡p = 0.035 IEx vs CV; § p = 0.035 CV vs CEx. c There was minimal binding of H3K9me2 for all groups at 1 week of age. d *p = 0.044 IV vs IE; †p = 0.043 IV vs CV; p > 0.10 IEx vs CV and CV vs CEx

Full size image

In previous studies, we observed increased abundance of H3K9me2 at the Pdx1 proximal promoter in islets from adult IUGR rats. Thus, to determine whether Ex4 prevented H3K9 dimethylation, ChIP assays were performed in both 1 week and adult islets. There was very little association of H3K9me2 at the Pdx1 promoter in all groups at 1 week of age (Fig. 4c). However, in adults, H3K9me2 was abundant at the Pdx1 promoter in IUGR islets and Ex-4 prevented this silencing histone modification (Fig. 4d; IV vs IEx: p = 0.044).

USF1 binding is restored in IUGR Ex-4 treated animals

USF1 is a homeobox transcription factor that binds to and functionally regulates the Pdx1 gene promoter and its binding is obligate for Pdx1 transcription [29, 30]. The USF1 binding site at Pdx1 is highly conserved and is located at a CpG site within the CpG island at the proximal region of the promoter (Fig. 1) [31, 32]. USF1 binding at Pdx1 was markedly reduced in IUGR islets from 1-week-old animals compared with control islets (Fig. 5). Ex-4 significantly increased association of USF1 at Pdx1 in IUGR islets from 1-week-old animals compared with vehicle-treated animals (Fig. 5: IEx vs IV: 32-fold increase, p = 0.032). Control animals treated with Ex-4 showed no change in USF1 binding (CEx vs CV, p = 0.374).

Fig. 5
figure 5

USF1 binding at the proximal promoter of Pdx1 in 1-week-old islets. ChIP analysis of cross-linked chromatin from islets of IUGR and control animals with and without Ex-4 treatment at 1 week of age immunoprecipitated with antibody to USF1. The IgG immunoprecipitate showed negligible PCR product, indicating little or no immunoprecipitation in the absence of primary antibody. The relative amount of USF1 bound at the Pdx1 promoter was measured by QPCR and normalised to total input DNA. Data are represented as fold change relative to control vehicle values. n = 6 experiments, data are ±SEM. *p = 0.032 IV vs IEx; †p = 0.034 IV vs CV; p > 0.10 IEx vs CV and CV vs CEx

Full size image

USF1 phosphorylation

PKA-dependent serine phosphorylation of USF1 has been implicated in modulating its DNA binding and transcriptional activation [19–23]. We found no differences in USF1 protein levels between the four study groups; however, endogenous levels of serine phosphorylated USF1 were very low in islets from 1-week-old IUGR animals compared with CV-treated animals (Fig. 6). Ex-4 treatment significantly increased phosphorylated USF1 in IUGR islets, but not controls.

Fig. 6
figure 6

Immunoprecipitation/western blot analyses of phosphorylated USF proteins in islets. Nuclear extracts were prepared from islets at 1 week of age from CV-, CEx-, IV- and IEx-treated animals. Extracts (10 μg of protein/reaction) were first immunoprecipitated with anti-USF1 (top panel) and then subjected to western blot analyses using anti-phosphoserine antibodies (anti-PS; bottom panel). Densitometric analyses of the anti-USF1 and anti-phosphoserine coimmunoprecipitation were performed. Values were measured as densitometry units and normalised to CV; n = 3 experiments, data are ±SEM. IV = 0.147 ± 0.06; IEx = 1.22 ± 0.73; CV = 1.00 ± 0.35; CEx = 1.11 ± 0.02. *p < 0.05 between IV and all other groups

Full size image

USF1 binding and histone modifications at the β-actin promoter

Given the dramatic changes in USF1 binding, H3 acetylation and H3K4me3 binding in islet chromatin at the proximal Pdx1 promoter, we considered the possibility that these changes may have represented non-specific transcription factor/protein binding to chromatin induced by tissue preparation technique. To exclude this possibility, we measured binding of USF1 and enrichment of acetyl H3 and H3K4me3 at the β-actin promoter, a gene whose expression does not change in IUGR islets. In 1-week-old islets there was no change in USF1 binding at the β actin promoter in any experimental group (ESM Fig. 2; p > 0.10 for IV vs IEx, CV and CEx). In addition, we did not detect any differences in acetyl H3 or H3K4me3 enrichment at the β-actin promoter in adult animals in any of the four experimental groups (ESM Fig. 2: p > 0.10 for IV vs IEx, CV and CEx for both acetyl H3 and H3K4Me3 in adult islets). In order to ensure that the transcription factor/protein binding alterations that we observed at the Pdx1 promoter were not due to non-specific antibody binding, we measured IgG and no antibody binding in experiments using antibodies to acetyl H3, H3K4me3, USF1 and PCAF. There was a negligible amount of immunoprecipitate when the ChIP experiments were run without antibody or with IgG antibody, indicating that we were successfully measuring the proteins of interest compared with non-specific protein binding at the Pdx1 promoter (ESM Table 4).

Ex-4 treatment increases histone acetyl transferase activity in IUGR islets

Given the increased levels of histone acetylation at the proximal Pdx1 promoter induced by Ex-4 treatment, we hypothesised that Ex-4 increased HAT activity and thus we measured generalised HAT activity in islet chromatin from 1-week-old animals. Figure 7a shows an approximate threefold increase in HAT activity in IUGR Ex-4 islets compared with IUGR vehicle islets (p = 0.003). There was no significant effect of Ex-4 on HAT activity in control animals (CV vs CEx; p = 0.074) at 1 week of age.

Fig. 7
figure 7

a HAT activity: islets from IUGR and control animals with and without treatment with Ex4 were isolated. HAT activity was measured via colorimetric assay with the cofactor acetyl CoA at baseline, 1, 2, and 3 h. Integrated HAT activity was computed as area under the curve and represented as per cent of CV values. n = 3 experiments, data are ±SEM. *p = 0.003 IV vs IEx; †p = 0.002 IV vs CV; p > 0.10 IEx vs CV and p > 0.05 CV vs CEx. b, c PCAF binding at the proximal promoter of Pdx1. ChIP analysis of cross-linked chromatin from islets of IUGR and control animals with and without Ex-4 treatment at (b) 1 week of age and (c) 4 months of age, immunoprecipitated with antibody to PCAF. The IgG immunoprecipitate showed negligible PCR product, indicating little or no immunoprecipitation in the absence of primary antibody. The relative amount of PCAF bound at the Pdx1 promoter was measured by QPCR and normalised to total input DNA. Data are represented as fold change relative to control vehicle values. n = 6 experiments, data are ±SEM. b *p = 0.037 IV vs IEx; †p = 0.015 IV vs CV; ‡p = 0.018 IEx vs CV; p > 0.05 CV vs CEx. c IV vs IEx: p = 0.97; IV vs CV: p = 0.461; IEx vs CV: p = 0.148; CV vs CEx: p = 0.781

Full size image

GLP-1 and its long-acting agonist, Ex-4, increase cAMP levels [33, 34]. The transcriptional effects of cAMP are mediated by the binding of the cAMP response element-binding protein (CREB), which is facilitated by the coactivators CBP and p300. p300/CBP have intrinsic acetyltransferase activity on histones [34]. Surprisingly, we found little association of p300 at the proximal promoter of Pdx1 in control or IUGR islets and there were no significant differences in binding of p300 to Pdx1 between any of the treatment groups. We then examined PCAF binding to the proximal promoter of Pdx1 by ChIP assay. PCAF acetylates free histones or nucleosomes, primarily on histone H3 through a mechanism that can be independent of the CBP/p300 complex-mediated pathway [35]. In contrast to p300/CBP, we found a dramatic increase in PCAF binding at Pdx1 in Ex-4 IUGR islets compared with IUGR vehicles at 1 week of age (Fig. 7b; IV vs IEx: tenfold increase, p = 0.037). There was no significant effect of Ex-4 on PCAF binding in controls (CV vs CEx; p = 0.94). Similar to USF1, Ex-4 had no effect on PCAF protein levels in controls or IUGR animals at 1 week of age. After establishing that Ex-4 treatment increased binding of PCAF at Pdx1, we investigated whether Ex-4 treatment enhanced PCAF and USF1 interaction in 1-week-old islets from IUGR animals treated with Ex-4 using coimmunoprecipitation experiments (Table 1). As measured by densitometry, neonatal Ex-4 treatment increased USF1 and PCAF interaction in 1-week-old islets (Table 1), whereas there was no change in interaction of USF1 and PCAF in control groups. Finally, we sought to determine whether neonatal Ex-4 treatment permanently increased the association of PCAF at the Pdx1 promoter or whether the effects of Ex-4 were transient. In ChIP experiments performed in islets harvested from adult rats, we found no change in PCAF binding among experimental groups (Fig. 7c; p > 0.05 in IV vs IEx, CV and CEx).

Table 1 PCAF association with USF1 at the proximal promoter of Pdx1
Full size table

Methylation status of the Pdx1 gene promoter in IUGR and control islets

The proximal 5′ flanking region of the Pdx1 gene and its first exon include a highly conserved CpG island that encompasses nucleotides −275 to +200 relative to the transcriptional start site (ESM Fig. 3). Using pyrosequencing, we evaluated the DNA methylation status of the portion of the CpG island located in the Pdx1 promoter in islets of IUGR and control animals. At 1 week of age, none of the 11 CpG sites in the Pdx1 promoter were methylated in either group [14]. However, in adulthood, Pdx1 promoter DNA methylation levels across the CpG island in IUGR vehicle-treated animals averaged 51.3 ± 10.3% compared with an absence of CpG methylation in control vehicles (Fig. 8a; IV vs CV; p < 0.05). No single CpG site on Pdx1 was consistently methylated in IUGR vehicle-treated islets. Neonatal Ex-4 treatment completely prevented DNA methylation at all CpG sites in Pdx1 in adult IUGR animals. Pdx1 DNA methylation levels across the CpG island in adult IUGR Ex-4 treated animals averaged 3.9 ± 0.64% and this was not significantly different from adult control animals treated with Ex-4, averaging 3.46 ± 0.52% (ESM Fig. 3 and Fig. 8a; IEx vs CEx; p = 0.63).

Fig. 8
figure 8

a Methylation status of the Pdx1 gene promoter in adult IUGR and control islets. The proximal promoter of Pdx1 is part of a more extensive CpG island. We performed pyrosequencing of bisulfite-treated DNA to determine the DNA methylation status of the portion of the CpG island that is located in the Pdx1 promoter in islets of IUGR and control animals. Evaluation of islet DNA from IUGR and control animals at 1 week of age revealed that none of the CpG sites in the Pdx1 promoter were methylated in either group. CpG site DNA methylation levels of DNA isolated from adult islets averaged across the proximal promoter of Pdx1 in the four treatment groups. n = 5 experiments. *p = 0.016 IV vs IEx; †p = 0.016 IV vs CV; p > 0.10 IEx vs CV and CV vs CEx. b Dnmt1 binding at the proximal promoter of Pdx1 at 6 months. ChIP analysis of cross-linked chromatin from islets of IUGR and control animals with and without Ex-4 treatment at 6 months of age immunoprecipitated with antibody to Dnmt1. The relative amount of Dnmt1 bound at the Pdx1 promoter was measured by QPCR and normalised to total input DNA. Data are represented as per cent of control vehicle values. n = 6 experiments, data are ±SEM; *p = 0.016 IV vs IEx; †p = 0.021 IV vs CV; p > 0.10 IEx vs CV and CV vs CEx

Full size image

To determine the underlying mechanisms responsible for prevention of DNA methylation in Ex-4 IUGR islets, we examined the association of Dnmt1 to Pdx1. Mammalian CpG methylation is mediated by Dnmts. Dnmt1 mediates replication-coupled maintenance of DNA methylation patterns [36, 37] and previous studies showed marked association of Dnmt1 at Pdx1 in IUGR islets at 6 months of age [14]. ChIP assays were performed on adult islets when DNA methylation at Pdx1 was first observed in IUGR islets. Dnmt1 binding was significantly enhanced in IUGR vehicle islets (17.1 ± 2.1-fold change over CV, p = 0.021; Fig. 8b), and we did not detect any significant association of Dnmt1 at Pdx1 in IUGR Ex-4, CV or CEx islets (Fig. 8b).

Discussion

The major finding of our study was that a short treatment course of Ex-4 in the newborn period permanently reverses the aberrant epigenetic modifications at the proximal promoter of Pdx1 in IUGR islets, an animal model of type 2 diabetes. Normalisation of USF1 binding, H3 acetylation and H3K4 trimethylation at the Pdx1 proximal promoter allows for an open chromatin domain, creating an environment permissive of Pdx1 transcription that is necessary for proper beta cell function and development. These are the first studies to show that Ex-4, a drug now commonly used for treatment of type 2 diabetes, permanently reverses epigenetic modifications of a key beta cell gene, Pdx1, leading to increased transcription.

USF1 is a critical activator of Pdx1 transcription and decreased binding of USF1 markedly reduces or abolishes Pdx1 transcription [29, 30]. Other authors and we have shown that loss of USF1 binding leads to an increase in repressive chromatin modifications such as methylation of H3K9 and H3K27 [14, 38]. Furthermore, USF1 binding is important for generating and maintaining adjacent histone modifications associated with active chromatin structure [39]. Here, we show that neonatal treatment with Ex-4 restores binding of USF1 thus allowing for an open chromatin domain and restoration of Pdx1 transcription in IUGR islets. In the beta cell, Ex-4 activates various signalling pathways in pancreatic beta cells, in particular cAMP, protein kinase A (PKA), Ca2+ and protein kinase B (PKB/Akt) leading to phosphorylation of a number of proteins [40]. We found that IUGR markedly decreases serine phosphorylation of nuclear USF1, which is normalised by Ex-4 treatment. Increased phosphorylation in turn is associated with increased USF1 binding at Pdx1 in IUGR islets. As a number of studies have shown that serine phosphorylation is obligate for USF1 binding to DNA, it is likely that this is the mechanism underlying Ex-4 induced increased association at Pdx1 in IUGR islets [19–22].

The first epigenetic mark that is modified in pancreatic beta cells of IUGR animals is histone acetylation [14]. Islets isolated from IUGR fetuses show a significant decrease in H3 acetylation at the proximal promoter of Pdx1 [14]. After birth, histone deacetylation progresses in IUGR animals. In this study, we found that Ex-4 administration during the newborn period normalises H3 acetylation to levels equal to or greater than those of control animals at 1 week of age. Ex-4 mediates this process by increasing HAT activity in IUGR islets to control levels. Our data show that PCAF is the primary histone acetyl transferase that catalyses acetylation at Pdx1. Ex-4 treatment of IUGR islets significantly increases PCAF binding at the proximal promoter of Pdx1. PCAF, also known as (p300/CBP [CREB-binding protein]-associated factor), was initially described as a member of an activating complex containing CREB and p300. However, in islets, PCAF binds to Pdx1 in the absence of p300. Other investigators have also shown that PCAF possesses HAT activity independent of CREB and p300 [35].

The mechanism by which Ex-4 increases PCAF binding at Pdx1 in IUGR islets is unclear. It is likely that during the period of increased beta cell replication that takes place in the neonatal period, Ex-4 increases association of PCAF and USF1 at the Pdx1 promoter and that these proteins together are responsible for increasing HAT activity and recruiting histone activating marks to the chromatin. This is in keeping with other studies that have recently shown that USF1 interacts directly with and recruits PCAF to the proximal promoter regions of a number of genes [22, 38, 41]. Our data indicate that in adulthood, the relative increase in PCAF binding in IUGR vehicle-treated islets is no longer present. However, the resulting increase in histone activating marks at the Pdx1 promoter is permanent thus leading to the permanent increase in Pdx1 expression. This implies that once histone acetylation is established, a mediator such as PCAF no longer needs to be present at the promoter.

After birth, in addition to persistent histone deacetylation, the IUGR state is further characterised by a marked decrease in H3K4me3 and an increase in H3K9me2, additional marks of repressed chromatin structure [14]. In our studies of IUGR animals treated with Ex-4, by the adult age H3K4me3 at Pdx1 is restored to control levels and dimethylation of H3K9 is inhibited. Histone acetylation is the initial step in chromatin remodelling leading to increased gene transcription, which is followed by histone methylation (specifically H3K4 trimethylation) [27]. We have previously shown that H3K4 methylation precludes methylation at lysine 9. Our in vivo findings support several in vitro studies showing that active chromatin states are maintained by H3K4 methylation, which opposes the lysine methylations that characterise inactive chromatin [42, 43].

The prevention of the aberrant chromatin structure observed in IUGR animals treated with Ex-4 is permanent. Increased acetylation of H3 and trimethylation of H3K4 in islets isolated from adult IUGR animals receiving a neonatal course of Ex-4 was maintained well into adulthood. In the normal rat, replication of existing beta cells and formation of new beta cells are higher during the newborn period than at any other time in postnatal life [44–46]. Thus, these histone modifications are somehow maintained during active beta cell replication. It is not known how histone modifications are propagated in daughter cells and it remains to be determined if histone modifications can serve as templates for reproducing these same structures on newly incorporated nucleosomes following cell replication.

Of major significance was our finding that neonatal Ex-4 treatment prevented DNA methylation at the CpG island within the proximal promoter of Pdx1, the final step in IUGR induced chromatin remodelling. In IUGR newborn islets, there is no DNA methylation at Pdx1; however, by 6 months of age, high levels of DNA methylation are found at the proximal promoter of Pdx1 in IUGR but the Pdx1 promoter remains unmethylated in control animals. DNA methylation is thought to be an essential step in ‘locking in’ the aberrant chromatin changes characteristic of IUGR animals.

The mechanisms by which Ex-4 prevents DNA methylation in IUGR islets are likely to be related to the prevention of H3K9 dimethylation, which in turn prevents recruitment of Dnmt1, the primary DNA methyltransferase that is associated with Pdx1 in IUGR islets [14]. A number of studies have shown that methylation of H3K9 precedes DNA methylation [37, 47]. It has also been suggested that DNA methyltransferases may act only on chromatin that is methylated at lysine 9 on histone H3 (H3K9) [48]. Histone methyltransferases bind to DNA methylases thereby initiating DNA methylation [37].

Of particular interest is the finding that neonatal Ex-4 treatment increases H3K4 trimethylation at the Pdx1 promoter in adult islets, but that this was not associated with an increase in Pdx1 expression. This apparent dissociation between the epigenetic mark and gene expression indicates that there are other yet unidentified mechanisms that are in place to prevent overexpression of any particular gene thus preventing changes in beta cell mass and preserving glucose homeostasis. Such braking mechanisms could include additional epigenetic regulators at the chromatin level or could function through the saturation of GLP-1 or other receptors. Further studies are needed to explore these possible regulatory mechanisms.

In conclusion, our results demonstrate that Ex-4 interrupts the self-propagating aberrant epigenetic cycle in IUGR islets by inducing PCAF and USF1 binding to the Pdx1 promoter, which in turn restores histone acetylation and H3K4 methylation, thereby preventing H3K9 and DNA methylation. At the neonatal stage, this epigenetic malprogramming is reversible and may define an important developmental window for innovative therapeutic approaches to prevent the development of adult onset diseases. Our studies indicate novel mechanisms by which Ex-4 functions to regulate gene expression in vivo. By increasing HAT activity, Ex-4 is able to reverse epigenetic modifications induced by IUGR and normalises Pdx1 expression.