2316?The?Journal?of?Clinical?Investigation? ? ? http://www.jci.org? ? ? Volume 118? ? ? Number 6? ? ? June 2008
Development of type 2 diabetes
following intrauterine growth retardation
in rats is associated with progressive
epigenetic silencing of Pdx1
Jun H. Park,1 Doris A. Stoffers,2,3 Robert D. Nicholls,4 and Rebecca A. Simmons1,3
1Department of Pediatrics, Children’s Hospital of Philadelphia, 2Division of Endocrinology, Diabetes, and Metabolism, Department of Medicine, and
3Institute for Diabetes, Obesity, and Metabolism, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA.
4Birth Defects Laboratories, Division of Medical Genetics, Department of Pediatrics, Children’s Hospital of Pittsburgh, and
Department of Human Genetics, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.
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). It has been hypothesized
that the molecular mechanisms underlying this phenomenon
may in part be related to epigenetic modulation of expression of
key developmental genes (2, 3). Epigenetic modifications provide
a mechanism that allows the stable propagation of gene activity
states from 1 generation of cells to the next. Epigenetic states can
be modified by environmental factors, which may contribute to the
development of abnormal phenotypes. In mammals, DNA meth-
ylation and histone modifications represent the major epigenetic
mechanisms implicated in the regulation of gene transcription.
We have developed an animal model of IUGR caused by utero-
placental insufficiency, which limits the supply of critical sub-
strates and hormones to the fetus (4, 5). This abnormal metabolic
intrauterine milieu affects the development of the fetus by per-
manently modifying gene expression and function of susceptible
cells, such as the β cell (4–6), and leads to the development of dia-
betes in adulthood (4–6).
Pdx1 is a pancreatic and duodenal homeobox 1 transcription
factor that regulates pancreas development and β cell differentia-
tion. Both genetic and acquired reductions in Pdx1 expression in
humans and in animal models have been shown to cause type 2 dia-
betes, β cell dysfunction (7–14), and impaired islet compensation
in the presence of insulin resistance (7, 8). In IUGR rats, we have
observed that Pdx1 mRNA levels are reduced by more than 50% in
IUGR fetuses (6). At birth, β cell mass is normal but Pdx1 mRNA
levels are decreased in IUGR rats (6). In IUGR adult rats, β cell mass
is markedly decreased and Pdx1 expression is nearly absent (6).
IUGR leads to the permanent suppression of Pdx1 in islets,
suggesting that an epigenetic mechanism may be responsible via
changes in DNA methylation, histone modifications, or chroma-
tin protein binding (15). The region spanning the proximal 5ʹ pro-
moter and all of exon 1 of Pdx1 contains a highly conserved CpG
island. Deletion of the proximal promoter completely abolishes
Pdx1 promoter activity (16). This region is also heavily acetylated
at histone H3 and H4 in a β cell tumor line (17). However, the epi-
genetic characteristics of the Pdx1 gene have not been examined in
primary islet tissue in normal or disease states.
In the present study, we hypothesized that epigenetic changes involv-
ing histone modifications, DNA methylation, chromatin remodeling,
and transcription were responsible for Pdx1 silencing during the tran-
sition from IUGR to diabetes in adults. We determined that in early
postnatal life, prior to the onset of diabetes, the IUGR state induces
deacetylation of histones H3 and H4, which is facilitated by recruit-
ment of histone deacetylase 1 (HDAC1) and Sin3A to the proximal
Nonstandard?abbreviations?used: 5-AzaC, 5-aza-2ʹ-deoxycytidine; Dnmt, DNA
methyltransferase; HDAC, histone deacetylase; H3K4, histone 3 lysine 4; H3K9,
histone 3 lysine 9; H3K4me3, trimethylation of lysine 4 at H3; H3K9me2, dimethyl-
ation of lysine 9 at H3; IUGR, intrauterine growth retardation; TSA, trichostatin A.
Conflict?of?interest: The authors have declared that no conflict of interest exists.
Citation?for?this?article: J. Clin. Invest. 118:2316–2324 (2008). doi:10.1172/JCI33655.
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
promoter of Pdx1. Loss of acetylation is accompanied by loss of bind-
ing of the key transcription factor, USF-1. As the disease state pro-
gresses, histone 3 lysine 4 (H3K4) is demethylated and histone 3 lysine
9 (H3K9) is methylated. Finally, once diabetes occurs, methylation of
the CpG island in the proximal promoter of Pdx1 ensues, locking in
marked suppression of Pdx1. These are the first studies to our knowl-
edge to describe the ontogeny of chromatin remodeling in vivo from
the fetus to the onset of a disease in adulthood.
Pdx1 transcription is reduced in IUGR islets. In previous studies, we
found that Pdx1 mRNA levels were decreased by 50% in fetuses
and 2-week-old pups and 80% in adult IUGR animals (6). To
determine whether this reduction in Pdx1 mRNA levels was sec-
ondary to a reduction in transcription of the Pdx1 gene or to an
increase in mRNA stability, we induced IUGR in Pdx1-LacZ in
vivo transcriptional reporter mice. The transgene was regulated
by the 4.6-kb XbaI-XbaI fragment containing 4.5 kb of the mouse
Pdx1 promoter and most of the 5ʹ untranslated region (UTR).
Most of the important Pdx1 enhancers described to date are
included within this promoter fragment (17–19). Endogenous
Pdx1 mRNA levels are reduced by 50.4% and transgene LacZ
mRNA is reduced in parallel by approximately 57.1% (P < 0.05
versus control). This indicates that IUGR regulates Pdx1 mRNA
by reducing the activity of transcriptional promoter or enhancer
elements in the proximal 4.5-kb Pdx1 5ʹ region.
Methylation status of the Pdx1 gene promoter in IUGR and control islets.
The proximal 5ʹ flanking region of the Pdx1 gene and its first exon
include a highly conserved CpG island that encompasses nucleo-
tides –360 to +200 relative to the translational start site (Figure
1A). Active promoters are associated with unmethylated CpG
islands and open chromatin structure, whereas inactive promot-
ers are typically characterized by a repressed chromatin structure
and hypermethylated CpGs. We determined the DNA methylation
Methylation status of proximal promoter of Pdx1. (A) Map of Pdx1.
The proximal promoter of Pdx1 is part of a more extensive CpG island.
Filled circles represent CpG dinucleotides, and numbers indicate posi-
tions relative to the transcriptional start site. The position of the E-box
bound by USF-1 is indicated. (B) Methylation of CpG dinucleotides
within the promoter of Pdx1. Islets were isolated from 5 IUGR and 5
control animals at 6 months of age, and Pdx1 DNA methylation was
quantified by pyrosequencing of bisulfite-treated DNA. We did not
detect methylation at any CpG of Pdx1 in controls. The shaded bars
represent data obtained and averaged from 5 IUGR animals and pre-
sented as percent methylation at each CpG site. Data are ± SEM.
Quantitative analysis of Dnmt1, Dnmt3a, and Dnmt3b
bound at the Pdx1 promoter region. ChIP analysis of
cross-linked chromatin from islets of IUGR and control
animals at 2 weeks and 6 months of age IP with anti-
body to Dnmt1, Dnmt3a, and Dnmt3b. Control (Con) and
IUGR samples were run on the same gel. Input DNA (In)
represents PCR products without prior IP. The IgG IP
showed negligible PCR product, indicating little or no IP
in the absence of primary antibody. The relative amount
of Dnmt1-, Dnmt3a-, and Dnmt3b-bound Pdx1 promoter
was measured by genomic quantitative real-time PCR
(Q-PCR) and normalized to input DNA. Data are repre-
sented as percent of control values. n = 3 experiments,
data are ± SEM; *P < 0.05 versus controls.
2318?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
status of the portion of the CpG island that is located in the Pdx1
promoter in islets of IUGR and control animals using bisulfite
treatment and pyrosequencing analyses. Evaluation of DNA from
islets of 5 IUGR and 5 control animals revealed that at age 2 weeks
none of the 14 CpG sites in the Pdx1 promoter were methylated
in islets from either IUGR or control animals. At 6 months of age
(n = 5 animals per group), Pdx1 DNA methylation levels across the
CpG island averaged 51.3% ± 10.3% compared with no CpG meth-
ylation in controls (P < 0.05 vs. controls). No single CpG site was
consistently methylated (Figure 1B).
Binding of DNA methyltransferases at Pdx1 in IUGR islets. Mammalian
CpG methylation is mediated by 3 active DNA methyltransferases
(Dnmts), Dnmt1, Dnmt3a, and Dnmt3b. Dnmt1 mediates replica-
tion-coupled maintenance of DNA methylation patterns, whereas
Dnmt3a and Dnmt3b are considered to be de novo DNA methylases
(20, 21). Surprisingly, we observed some association of Dnmt1 to
Pdx1 in islets of 2-week-old IUGR pups, despite the lack of DNA
methylation of CpGs (Figure 2). However, there was no binding of
Dnmt3a or Dnmt3b to Pdx1 at this age. At 6 months of age, when
DNA methylation at Pdx1 is increased in IUGR islets, Dnmt1 bind-
ing was significantly enhanced (Figure 2) and moderate levels of
Dnmt3a, but not Dnmt3b, were also associated with Pdx1 (Figure
2). No binding of Dnmt1 or Dnmt3a and -3b to Pdx1 was detected
in control islets at either age.
Characterization of the histone code at the Pdx1 promoter. Whereas
DNA methylation is associated with transcriptional silencing,
histone acetylation is associated with active chromatin, and
deacetylation results in repressed chromatin structure (15). The
impact of IUGR on the acetylation status of core histones H3 and
H4, which are physically associated with the Pdx1 promoter, was
determined by ChIP analyses using primers that encompass the
region of the CpG island at Pdx1, shown in Figure 1A. Similar to
findings in a β cell tumor line (17), this region of Pdx1 was heavily
acetylated at H3 and H4 in islets from control animals (Figure 3).
Histone acetylation and methylation at the Pdx1 pro-
moter. ChIP analysis of cross-linked chromatin from
islets of IUGR and control animals at fetal day 21 (A),
2 weeks (B), and 6 months of age (C) IP with antibody
to acetylated H3 (AceH3), acetylated H4 (AceH4),
H3K4me3, and H3K9me2. Input DNA represents PCR
products without prior IP. The IgG IP showed negligible
PCR product, indicating little or no IP in the absence of
primary antibody. The relative amount of acetylated H3–,
acetylated H4–, H3K4me3-, and H3K9me2-bound Pdx1
promoter was measured by Q-PCR and normalized to
input DNA. Data are represented as percent of control
values. n = 3 experiments, data are ± SEM; *P < 0.05
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
IUGR significantly reduced the abundance of acetylated H3 and
H4 and deacetylation progressed with age. By 6 months of age, H3
and H4 acetylation was absent (Figure 3).
Trimethylation of lysine 4 at H3 (H3K4me3) preferentially marks
promoters, is associated with active genes, and is influenced by
density of H3 acetylation (22, 23). In islets from control animals
at all ages, there was a high association of H3K4me3 at Pdx1. In
contrast, despite the marked decrease in histone acetylation in
IUGR fetuses, there was no difference in the methylation status of
H3K4 between IUGR and control fetal animals. However, in IUGR
animals at 2 weeks of age, the abundance of H3K4me3 was signifi-
cantly reduced, and by 6 months of age, H3K4me3 was absent at
the Pdx1 proximal promoter region (Figure 3).
Loss of H3K4 trimethylation can initiate dimethylation of lysine
9 at H3 (H3K9me2), a histone modification that is involved in the
establishment and maintenance of silent heterochromatin regions
(24). ChIP assays demonstrated that by 2 weeks of age, H3K9me2
at the Pdx1 promoter was present in IUGR islets and H3K9me2
binding increased with age (Figure 3). We did not observe methyla-
tion of H3K27, another silencing methyl mark (data not shown) in
IUGR or control islets at any age.
Histone modifications inhibit USF-1 binding at Pdx1. A number of
studies have demonstrated that USF-1 binds to and functionally
regulates the Pdx1 gene promoter (25, 26). A reduction in endog-
enous USF-1 binding leads to a significant decrease in Pdx1 pro-
moter activity, which, in turn, results in marked reductions in Pdx1
mRNA and protein levels (25). The USF-1 binding site is highly
conserved and is located at a CpG site within the CpG island (Fig-
ure 1A). As expected, we observed strong association of USF-1 at
Pdx1 in control islets at all ages examined (Figure 4A). In contrast,
USF-1 binding was abolished in IUGR islets at all ages. Western
blot analyses showed no difference in USF-1 protein abundance
between IUGR and controls indicating that decreased USF-1 asso-
ciation with Pdx1 was not due to a decrease in USF-1 protein levels
in IUGR islets (Figure 4B).
Recruitment of transcription repressors to the Pdx1 promoter in IUGR
islets. The reduction of histone H3 and H4 acetylation in IUGR
islets suggested that HDACs might be involved in facilitating
the observed histone modifications. In fetal islets from IUGR
animals, there was modest binding of HDAC1, but not mSin3A,
at Pdx1. Neither protein was associated with the proximal pro-
moter of Pdx1 in control fetal islets (Figure 5A). At 2 weeks and 6
months of age, we observed a robust association of HDAC1 and
mSin3A at Pdx1, which was concomitant to complete deacety-
lation of H3 and H4 in IUGR islets (Figure 5, B and C). There was
no binding of HDAC2 at Pdx1 in IUGR or control islets at any
age (data not shown).
Inhibition of HDAC partially restores Pdx1 expression. To determine
whether either DNA methylation or histone deacetylation is suf-
ficient for silencing of Pdx1 expression in IUGR animals, islets
were isolated from 2-week-old IUGR and control animals. At this
age (as opposed to later in life), there is significant replication of
β cells in rodents making it the ideal age to determine whether
stable propagation of histone modifications occurs during cell
division and whether these changes can be reversed (27). Islets
from IUGR and control animals were cultured for 5 days in the
presence of 5-aza-2ʹ-deoxycytidine (5-AzaC) and/or trichostatin
A (TSA), which inhibit Dnmts and HDACs, respectively. TSA
treatment alone (and in combination with 5-AzaC) normalized
acetylation of H3 and restored methylation of H3K4 at the Pdx1
promoter in IUGR islets, further substantiating the dependence
of H3K4 methylation upon levels of H3 acetylation (Figure 6, A
USF-1 in IUGR and control islets. (A)
ChIP analysis of cross-linked chro-
matin from islets of IUGR and control
animals at fetal day 21, 2 weeks, and
6 months of age IP with antibody to
USF-1. Input DNA represents PCR
products without prior IP. The IgG IP
showed negligible PCR product, indi-
cating little or no IP in the absence
of primary antibody. The relative
amount of USF-1 bound Pdx1 pro-
moter was measured by Q-PCR and
normalized to input DNA. Data are
represented as percent of control
values. n = 3 experiments, data are ±
SEM; *P < 0.05 versus controls. (B)
Western blot and densitometric anal-
yses of islets isolated from IUGR and
control rats at fetal day 21, 2 weeks,
and 6 months of age. Blots were
probed with anti–USF-1 and stripped
and probed with anti-actin as a con-
trol. Data are ± SEM.
2320?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
and B). However, Pdx1 expression (Figure 6C) was largely but not
completely restored in IUGR islets, indicating that other relatively
minor mechanisms of repression might exist in addition to the
recruitment of a histone deacetylation complex. While we did not
observe any DNA methylation at the CpG island in IUGR islets,
it is possible that other CpGs located in distal enhancer regions
could contribute to Pdx1 repression. However, treatment with
5-AzaC alone did not alter Pdx1 mRNA levels in either IUGR or
control islets (Figure 6C), suggesting that DNA methylation does
not contribute to Pdx1 silencing at this young age.
Surprisingly, TSA, but not 5-AzaC, treatment resulted in nor-
malization of USF-1 binding to Pdx1 in IUGR islets (Figure
6D). Thus, reversal of histone modifications by TSA normalized
association of USF-1 to Pdx1 in the IUGR animal. Neither agent
altered USF-1 protein levels (data not shown). USF-1 is a meth-
ylation-sensitive transcription factor and the USF site contains
an evolutionary, conserved CpG. However, our data suggest that
histone modifications, and not DNA methylation, are respon-
sible for loss of USF-1 association with the proximal promoter
of Pdx1 in IUGR islets.
Pdx1 is a critical regulator of β cell growth and function in
both fetal and postnatal developmental stages, and even a rela-
tively modest decrease in expression impairs the compensatory
response to insulin resistance (8, 28). Decreased Pdx1 expres-
sion plays a pivotal role in the development of diabetes in IUGR
animals, as normalization of Pdx1 expression is associated
with long-term maintenance of β cell mass and normal glucose
homeostasis (6). Uteroplacental insufficiency, the most common
cause of IUGR, limits the supply of critical substrates such as
oxygen, glucose, and amino acids to the fetus, resulting in an
altered redox state and oxidative stress (4, 5). Here we have shown
that this altered metabolic milieu decreases Pdx1 transcription
by mediating a cascade of epigenetic modifications culminating
in silencing of Pdx1. These current studies are the first to our
knowledge to characterize the histone code at Pdx1 in vivo in pri-
mary islets and link the progression of epigenetic modifications
at a key gene to the development of diabetes.
Chromatin modification mechanisms serve a critical function
in affecting the transcriptional status of genes. Our data demon-
Recruitment of repressor complex at
Pdx1 in IUGR islets. ChIP analysis of
cross-linked chromatin from islets of
IUGR and control animals at fetal day
21 (A), 2 weeks (B), and 6 months of
age (C) IP with antibody to HDAC1 and
mSin3A. Input DNA represents PCR
products without prior IP. The IgG IP
showed negligible PCR product, indi-
cating little or no IP in the absence of
primary antibody. The relative amount
of HDAC1 and mSin3A bound Pdx1
promoter was measured by Q-PCR
and normalized to input DNA. Data
are represented as percent of control
values. n = 3 experiments, data are ±
SEM; *P < 0.05 versus controls.
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
strate that the open chromatin domain marked by histone H3
and H4 acetylation at the proximal promoter of Pdx1 is essential
for transcription. Robust Pdx1 expression in islets from control
animals is coincident with the presence of acetylated histones H3
and H4 as well as H3K4me3. Loss of these marks results in Pdx1
silencing and reversal of IUGR-induced epigenetic modifications
normalizes Pdx1 expression. These data suggest that histone modi-
fications can be stably propagated throughout life.
The first epigenetic mark that is modified in β cells of IUGR
animals is histone acetylation (Figure 7). Islets isolated from IUGR
fetuses show a significant decrease in H3 and H4 acetylation at the
proximal promoter of Pdx1. These changes in H3 and H4 acetyla-
tion are associated with a loss of binding of USF-1 to the proximal
promoter of Pdx1. USF-1 is a critical activator of Pdx1 transcrip-
tion, and decreased binding markedly decreases Pdx1 transcription
(25, 26). After birth, histone deacetylation progresses and is fol-
lowed by a marked decrease in H3K4 trimethylation and a signifi-
cant increase in H3K9me2 in IUGR islets (Figure 7). Progression
of these histone modifications parallels the progressive decrease in
Pdx1 expression as glucose homeostasis deteriorates and oxidative
stress increases in IUGR animals. Nevertheless, in the IUGR pup
(at 2 weeks of age), these silencing histone modifications alone
suppress Pdx1 expression, since there is no appreciable methyla-
tion in the CpG island and reversal of histone deacetylation in
IUGR islets (in the presence of active β cell replication) is sufficient
to nearly normalize Pdx1 mRNA levels.
The incomplete restoration of Pdx1 mRNA levels associated with
the complete reversal of histone deacetylation in newborn IUGR
islets suggests that there may be an additional repressor protein(s)
such as SIRT4 or a microRNA like miR9 (both are expressed in
islets; refs. 29, 30) that may be involved in chromatin silencing.
Identifying such factors will require extensive future experiments.
Alternatively, remodeling of active chromatin may only occur in
β cells capable of replication, and there may be a minority of cells
for which this does not occur, accounting for the extensive but
incomplete reactivation of Pdx1 expression.
The initial mechanism by which IUGR silences Pdx1 is by recruit-
ment of corepressors, including HDAC1 and mSin3A, which cata-
lyze histone deacetylation — the first repressive mark observed at
Pdx1 in IUGR islets. Binding of these deacetylases in turn facili-
tated loss of H3K4me3, further repressing Pdx1 expression (Figure
7). Our observation that inhibition of HDAC activity by TSA treat-
ment normalized H3K4me3 levels at Pdx1 in IUGR islets suggests
that the association of HDAC1 at Pdx1 in IUGR islets likely serves
as a platform for the recruitment of a demethylase, which catalyzes
demethylation of H3K4. Lysine demethylases in the Jumonji class
remove H3Kme3 and H3Kme2 (reviewed in ref. 31), while LSD1
removes H3Kme1/2 (32). Klose and coworkers have recently dem-
onstrated that the retinoblastoma binding protein, RBP2, contains
a JmjC domain, which can specifically demethylate H3K4me3 (33).
However, enzymes in the Jumonji class may not catalyze H3K4me3
demethylation in IUGR islets, as activity of these proteins is depen-
Reversal of histone modifications partially restores Pdx1 expression in IUGR islets. The relative amount of acetylated histone H3 (A), H3K4me3
(B), and Pdx1 mRNA (C) levels measured by Q-PCR and normalized to Gapdh. n = 3 experiments, data are ± SEM; *P < 0.05 versus controls;
#P < 0.05 versus pretreated IUGR islets. (D) USF-1–bound Pdx1 promoter, as detected by ChIP, was measured by Q-PCR and normalized to
input DNA. n = 3 experiments, data are ± SEM; *P < 0.05 versus controls; #P < 0.05 versus pretreated IUGR islets.
2322?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
dent upon the presence of an iron-binding domain (34, 35), which
would be inactivated under conditions of oxidative stress that are
present in IUGR islets (5). These results imply that another class
of histone demethylases may exist in β cells.
Loss of H3K4me3 was concomitant with a marked increase
in H3K9me2 at Pdx1 in 2-week-old IUGR animals, suggesting
that K4 methylation precludes methylation at lysine 9. These in
vivo findings support several in vitro studies showing that active
chromatin states are maintained by H3K4 methylation, which
opposes the lysine methylations that characterize inactive chro-
matin (36, 37). Since restoration of histone acetylation by TSA
treatment of IUGR islets reversed H3K9me2, this also demon-
strates that histone acetylation prevents methylation of H3K9.
Thus, IUGR-induced histone modifications are mutually rein-
forcing and interdependent.
DNA methylation of a CpG island in the promoter is a key
mechanism for silencing gene expression. Most CpG islands
remain unmethylated in normal cells; however, under some nor-
mal circumstances, such as for imprinted genes, genes on the
inactive X chromosome in females, and for some disease pro-
cesses such as cancer (38) and oxidative stress (39), CpG islands
can become methylated de novo. This is particularly relevant to
type 2 diabetes, as there are now substantial data that show that
oxidative stress plays a significant role in the progression of β cell
deterioration (40–44). Further, IUGR induces mitochondrial dys-
function in the β cell leading to increased production of ROS and
oxidative stress (5). It is not known why particular CpG islands
are susceptible to aberrant methylation. A recent study by Feltus
et al. (45) suggests that there is a “sequence signature associated
with aberrant methylation.” Of particular relevance to this study
is their finding that Pdx1 and a flanking gene, Cdx2 (also encod-
ing a homeobox protein), were 2 of only 15 genes (a total of 1,749
genes with CpG islands were examined) that were methylation
susceptible under conditions of increased methylation induced
by overexpression of DNMT1.
The molecular mechanism responsible for DNA methylation
in IUGR islets is likely to involve H3K9 methylation. A number
of studies have shown that methylation of H3K9 precedes DNA
methylation (21, 46). It has also been suggested that Dnmts may
act only on chromatin that is methylated at lysine 9 on histone H3
(H3K9) (47). Histone methyltransferases bind to the DNA meth-
ylases DNMT3A and DNMT3B, thereby initiating DNA methyla-
tion (21). Surprisingly, we found that DNMT1 was associated with
Pdx1 prior to CpG methylation. Since DNMT1 can be recruited by
interaction with a HDAC such as HDAC1 (48), which is already
associated with the Pdx1 promoter in fetal IUGR islets, we suggest
that this occurs in the IUGR islet prior to the alterations in histone
methylation that only occur after birth (Figure 7). As H3K9 meth-
ylation occurs during postnatal life in IUGR islets, this would then
allow recruitment of the de novo methyltransferase DNMT3A (21)
(but not DNMT3B). Subsequent to onset of DNA methylation,
DNMT1 would then be positioned to maintain the methylated
state, locking in Pdx1 silencing in adult IUGR islets (Figure 7).
In conclusion, our results demonstrate that IUGR induces a self-
propagating epigenetic cycle, in which the mSin3A/HDAC com-
plex is first recruited to the Pdx1 promoter, histone tails are sub-
jected to deacetylation, and Pdx1 transcription is repressed. At the
neonatal stage, this epigenetic process is reversible and may define
an important developmental window for therapeutic approaches.
However, as H3K9me2 accumulates, DNMT3A is recruited to the
promoter and initiates de novo DNA methylation, which locks in
the silenced state in the IUGR adult pancreas, resulting in diabe-
tes. We believe our studies indicate novel mechanisms of epigen-
etic regulation of gene expression in vivo, which link gene silencing
in the β cell to the development of type 2 diabetes and suggest
therapeutic agents that we believe are novel for the prevention of
common diseases with late-onset phenotypes.
Animal model. Surgical methods have been previously described (4, 5, 49). In
brief, time-dated Sprague-Dawley pregnant rats were individually housed
under standard conditions and allowed free access to standard rat chow
and water. On day 18 of gestation (term is 22 days), pregnant rats were anes-
thetized with intraperitoneal xylazine (8 mg/kg) and ketamine (40 mg/kg),
and both uterine arteries were ligated. This results in an approximately
50% reduction in uteroplacental blood flow and induces IUGR (4, 49).
Summary of epigenetic changes at Pdx1 in IUGR rats during the
development of type 2 diabetes. In pancreatic β-cells (top row), the
Pdx1 proximal promoter is normally found in an unmethylated (open
circles) open chromatin state, allowing access to transcription factors
such as USF-1 and associated with nucleosomes characterized by
acetylated (Ac, octagons) histones H3 and H4 and with H3K4me3
(Me3, hexagons). In IUGR fetal and 2 week islets (second and third
rows), histone acetylation is progressively lost through association with
a mSin3A-HDAC1-DNMT1 repressor complex, with H3K4me3 disap-
pearing and H3K9me2 (Me2, triangles) appearing after birth. IUGR
adult islets (fourth row) are characterized by inactive chromatin with
H3K9me2 and extensive DNA methylation (filled circles) locking in the
transcriptionally silent state of Pdx1.
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
Controls underwent sham surgery. Rats recovered within a few hours and
had ad libitum access to food and water. The pregnant rats were allowed
to deliver spontaneously, and the litter size was randomly reduced to 8 at
birth to assure uniformity of litter size between IUGR and control litters.
IUGR and control pups were fostered to unoperated normal female rats
and remained with their foster mothers until they were weaned. Only male
animals were studied to avoid the potentially confounding hormonal vari-
ables associated with female rats. While it is possible that the changes that
we observed may differ in female rats as gender specific effects on histone
acetylation and DNA methylation have been reported (reviewed in refs. 50,
51), we have not observed gender specific effects on β cell function in IUGR
rats. Thus, it is unlikely that epigenetic modifications at Pdx1 may differ
between male and female animals.
Islets were harvested from 10 litters each of 2-week-old IUGR and con-
trol animals, and islets were pooled from a litter. Islets were also harvested
from 10 IUGR and 10 control animals at 6 months of age (each animal
was from a different litter). In rats, β cells compose 95% of the cellular
mass of the islet (52).
These studies were approved by the Animal Care Committee of the Chil-
dren’s Hospital of Philadelphia and the University of Pennsylvania.
Pdx1 reporter gene analysis. The generation and characterization of the
Pdx1-LacZ transgenic reporter mouse was previously reported (53). A
modification of the IUGR procedure was performed on gestation day
17 in maternal pregnant Pdx1-LacZ mice. Pancreatic islets were isolated
from transgene-positive offspring at 3 weeks and RNA was isolated for
real-time PCR quantification of Pdx1 and LacZ mRNA levels.
RNA isolation and real-time PCR. Total RNA was extracted from islets using
RNAzol B (Tel-Test Inc.). Real-time PCR was carried out using an ABI
7900HT Real-Time PCR system with SYBR Green Master Mix from Applied
Biosystems. Results were normalized to glyceraldehyde-3-phosphate dehy-
drogenase expression (primers are listed in Supplemental Table 1).
Methylation analysis of the Pdx1 CpG island. Genomic DNA was extracted
from islets of 2-week-old (n = 8 litters of 6 pups for each group), 6-month-
old IUGR (n = 10), and control (n = 8) rats by phenol-chloroform-isoamyl
alcohol and was collected by ethanol precipitation. We denatured 10 μg
aliquots of XbaI-digested genomic DNA in 0.3 M NaOH at 37°C. Sulfo-
nation and hydrolytic deamination reactions were then carried out by
adding 2.5 M sodium bisulfite (Na2S2O5)/10 mM hydroquinone solution
(pH 5.0) to the digested genomic DNA and then incubating at 50°C for 4
hours in the dark. Bisulfite-converted DNA was purified using a QIAquick
spin purification kit (QIAGEN) and eluted with tris/EDTA buffer (TE).
Desulfonation reactions were performed in 0.3 M NaOH at 37°C for 15
minutes, and DNA was precipitated with 3 M sodium acetate/ethanol and
resuspended in water.
The PCR reaction buffer for the pyrosequencing of Pdx1 contained
2.5 mM MgCl2, 0.2 mM each of deoxyribonucleotide triphosphate (dNTP),
0.2 mM each of primer, and 2 U AmpliTaq Gold (Applied Biosystems) in
a 25 ml reaction mixture. The PCR amplification was done for 50 cycles
with an annealing temperature of 55°C (primers are listed in Supplemen-
tal Table 1). All the reactions were constructed as recommended by the
manufacturer’s instructions (Biotage AB). The pyrosequencing reaction
was performed automatically using a PSQ 96MA system along with a SNP
Reagent kit. As a control, plasmid DNA, which is unmethylated, was ampli-
fied and then subjected to the same analysis in parallel in order to confirm
the completion of the bisulfite reaction.
ChIP assays. Islets were trypsin digested and suspensions were fixed in
phosphate-buffered saline containing 1% formaldehyde and protease
inhibitor cocktail. Sonication was performed to yield DNA fragments
ranging in size from 200–1,000 bp, followed by centrifugation for 10 min-
utes. Soluble chromatin (150 μg) was diluted in IP buffer and protease
inhibitors. Supernatants were collected and cleared by incubation with
protein A–sepharose, sonicated salmon sperm DNA, and 10 μl of preim-
mune serum. An aliquot of supernatant was collected and used as the
input DNA. IP was carried out overnight with IgG (negative control) or
primary antibody to the modified histone (acetylated histone H3 and H4,
trimethyl-histone H3 [Lys4], acetylated H3K4, and dimethyl-histone H3
[Lys9]), HDAC1 (Upstate Biotech), USF-1, Dnmt1, Dnmt3a, and Dnmt3b
(Abcam), and mSin3A (Santa Cruz Biotechnology Inc.). Immune com-
plexes were isolated by incubation with protein A–agarose, and the com-
plexes were then serially washed in low-salt buffer, high-salt buffer, LiCl
buffer, and finally TE. Bound complexes were eluted from protein A with
a 1% SDS buffer and cross-links were reversed in the presence of Rnase
A and extensively digested with proteinase K. The released DNA was
extracted, precipitated, and resuspended in water. DNA sequences in the
“input” and the IP samples were quantified relative to each other by real-
time PCR (primers are listed in Supplemental Table 1).
To determine whether reversal of epigenetic modifications would restore
Pdx1 expression in IUGR animals, islets were harvested from 2-week-old
IUGR and control pups and cultured for 5 days in RPMI 1640 culture
medium containing 11 mM glucose, supplemented with 10% fetal bovine
serum, 2 mM l-glutamine, 50 IU/ml penicillin, and 50 mg/ml streptomy-
cin (Gibco BRL) in the presence and absence of 10 μM TSA and/or 1 μM
5-AzaC, which inhibited HDACs and Dnmt1, respectively.
Statistics. Statistical analyses were performed using analysis of vari-
ance and the 2-tailed Student’s unpaired t test. A P value of less than
0.05 was considered significant.
This study was supported by the NIH grant DK55704 (to R.A. Sim-
mons) and DK062965 (to R.A. Simmons and D.A. Stoffers). We
thank Hongshun Niu for his expert technical assistance.
Received for publication March 20, 2007, and accepted in revised
form March 18, 2008.
Address correspondence to: Rebecca A. Simmons, Biomedical
Research Building II/III 1308, University of Pennsylvania School
of Medicine, 421 Curie Boulevard, Philadelphia, Pennsylvania
19104, USA. Phone: (215) 746-5139; Fax: (215) 573-7627; E-mail:
1. Barker, D.J. 1999. The fetal origins of type 2 diabe-
tes mellitus. Ann. Intern. Med. 130:322–324.
2. Gluckman, P.D., and Hanson, M.A. 2004. Living
with the past: evolution, development, and pat-
terns of disease. Science. 305:1733–1736.
3. Gluckman, P.D., and Hanson, M.A. 2007. Devel-
opmental plasticity and human disease: research
directions. J. Intern. Med. 261:461–471.
4. Simmons, R.A., Templeton, L., Gertz, S., and Niu,
H. 2001. Intrauterine growth retardation leads to
type II diabetes in adulthood in the rat. Diabetes.
5. Simmons, R.A., Suponitsky-Kroyter, I., and Selak,
M.A. 2005. Progressive accumulation of mitochon-
drial DNA mutations and decline in mitochon-
drial function lead to beta-cell failure. J. Biol. Chem.
6. Stoffers, D.A., Desai, B.M., Ng, D.D., and Simmons,
R.A. 2003. Neonatal exendin-4 prevents the devel-
opment of diabetes mellitus in the intrauterine
growth retarded rat. Diabetes. 52:734–740.
7. Brissova, M., et al. 2005. Reduced PDX-1 expres-
sion impairs islet response to insulin resistance
and worsens glucose homeostasis. Am. J. Physiol.
Endocrinol. Metab. 288:E707–E714.
8. Kulkarni, R.N., et al. 2004. PDX-1 haploinsuffi-
ciency limits the compensatory islet hyperplasia
that occurs in response to insulin resistance. J. Clin.
9. Holland, A.M., Gonez, L.J., Naselli, G., MacDonald,
R.J., and Harrison, L.C. 2005. Conditional expres-
sion demonstrates the role of the homeodomain
transcription factor Pdx1 in maintenance and
regeneration of beta-cells in the adult pancreas.
10. Jonsson, J., Gonez, L.J., Naselli, G., MacDonald, R.J.,
2324?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 118 Number 6 June 2008
and Harrison, L.C. 1994. Insulin-promoter-factor
1 is required for pancreas development in mice.
11. Offield, M.F., et al. 1996. PDX-1 is required for pan-
creatic outgrowth and differentiation of the rostral
duodenum. Development. 122:983–995.
12. Hui, H., and Perfetti, R. 2002. Pancreas duodenum
homeobox-1 regulates pancreas development during
embryogenesis and islet cell function in adulthood.
Eur. J. Endocrinol. 146:129–141.
13. Ahlgren, U., Jonsson, J., Jonsson, L., Simu, K., and
Edlund, H. 1998. Beta-cell-specific inactivation of
the mouse Ipf1/Pdx1 gene results in loss of the
beta-cell phenotype and maturity onset diabetes.
Genes Dev. 12:1763–1768.
14. Stoffers, D.A., Zinkin, N.T., Stanojevic, V., Clarke,
W.L., and Habener, J.F. 1997. Pancreatic agenesis
attributable to a single nucleotide deletion in the
human IPF1 gene coding sequence. Nat. Genet.
15. Jaskelioff, M., and Peterson, C.L. 2003. Chromatin
and transcription: histones continue to make their
marks. Nat. Cell Biol. 5:395–399.
16. Sharma, S., et al. 1996. Pancreatic islet expression of
the homeobox factor STF-1 relies on an E-box motif
that binds USF. J. Biol. Chem. 271:2294–2299.
17. Gerrish, K., Van Velkinburgh, J.C., and Stein,
R. 2004. Conserved transcriptional regula-
tory domains of the pdx-1 gene. Mol. Endocrinol.
18. Gannon, M., Gamer, L.W., and Wright, C.V. 2001.
Regulatory regions driving developmental and tis-
sue-specific expression of the essential pancreatic
gene pdx1. Dev. Biol. 238:185–201.
19. Stoffers, D.A., Stanojevic, V., and Habener, J.F.
1998. Insulin promoter factor-1 gene mutation
linked to early-onset type 2 diabetes mellitus
directs expression of a dominant negative isopro-
tein. J. Clin. Invest. 102:232–241.
20. Bird, A.P., and Wolffe, A.P. 1999. Methylation-
induced repression-belts, braces, and chromatin.
21. Li, H., et al. 2006. The histone methyltransferase
SETDB1 and the DNA methyltransferase DNMT3A
interact directly and localize to promoters silenced
in cancer cells. J. Biol. Chem. 281:19489–19500.
22. Liang, G., et al. 2004. Distinct localization of
histone H3 acetylation and H3-K4 methylation to
the transcription start sites in the human genome.
Proc. Natl. Acad. Sci. U. S. A. 101:7357–7362.
23. Nightingale, K.P., et al. 2007. Cross talk between
histone modifications in response to HDAC
inhibitors: MLL4 links histone H3 acetylation
and histone H3K4 methylation. J. Biol. Chem.
24. Huang, Y., Fang, J., Bedford, M.T., Zhang, Y., and
Xu, R.M. 2006. Recognition of histone H3 lysine-4
methylation by the double tudor domain of JMJD2A.
25. Qian, J., Kaytor, E.N., Towle, H.C., and Olson, L.K.
1999. Upstream stimulatory factor regulates Pdx-1
gene expression in differentiated pancreatic β-cells.
Biochem. J. 341:315–322.
26. Sharma, S., et al. 1996. Pancreatic islet expres-
sion of the homeobox factor STF-1 (Pdx-1) relies
on an E-box motif that binds USF. J. Biol. Chem.
27. Scaglia, L., Cahill, C.J., Finegood, D.T., and Bonner-
Weir, S. 1997. Apoptosis participates in the remod-
eling of the endocrine pancreas in the neonatal rat.
28. Del Guerra, S., et al. 2005. Functional and molecu-
lar defects of pancreatic islets in human type 2 dia-
betes. Diabetes. 54:727–735.
29. Plaisance, V., et al. 2006. MicroRNA-9 controls the
expression of Granuphilin/Slp4 and the secretory
response of insulin-producing cells. J. Biol. Chem.
30. Haigis, M.C., et al. 2006. SIRT4 inhibits gluta-
mate dehydrogenase and opposes the effects of
calorie restriction in pancreatic beta cells. Cell.
31. Shi, Y., and Whetstine, J.R. 2007. Dynamic regula-
tion of histone lysine methylation by demethylases.
Mol. Cell. 25:1–14.
32. Forneris, F., et al. 2006. A highly specific mechanism
of histone H3-K4 recognition by histone demethyl-
ase LSD1. J. Biol. Chem. 281:35289–35295.
33. Klose, R.J., et al. 2007. The retinoblastoma bind-
ing protein RBP2 is an H3K4 demethylase. Cell.
34. Trewick, S.C., Henshaw, T.F., Hausinger, R.P., Lin-
dahl, T., and Sedgwick, B. 2002. Oxidative demeth-
ylation by Escherichia coli AlkB directly reverts
DNA base damage. Nature. 419:174–178.
35. Cloos, P.A., et al. 2006. The putative oncogene
GASC1 demethylates tri- and dimethylated lysine
9 on histone H3. Nature. 442:307–311.
36. Litt, M.D., Simpson, M., Gaszner, M., Allis, C.D.,
and Felsenfeld, G. 2001. Correlation between
histone lysine methylation and developmental
changes at the chicken beta-globin locus. Science.
37. Noma, K., and Grewal, S.I. 2002. Histone H3 lysine
4 methylation is mediated by Set and promotes
maintenance of active chromatin states in fission
yeast. Proc. Natl. Acad. Sci. U. S. A. 99:16438–16445.
38. He, B., et al. 2003. SOCS-3 is frequently silenced
by hypermethylation and suppresses cell growth
in human lung cancer. Proc. Natl. Acad. Sci. U. S. A.
39. Cerda, S., and Weitzman, S.A. 1997. Influence of
oxygen radical injury on DNA methylation. Mutat.
40. Ihara, Y., et al. 1999. Hyperglycemia causes oxi-
dative stress in pancreatic beta-cells of GK rats, a
model of type 2 diabetes. Diabetes. 48:927–932.
41. Silva, J.P., et al. 2000. Impaired insulin secretion
and β-cell loss in tissue specific knockout mice with
mitochondrial diabetes. Nat. Genet. 26:336–340.
42. Kaneto, H., et al. 2001. Activation of the hexos-
amine pathway leads to deterioration of pancreatic
beta-cell function through the induction of oxida-
tive stress. J. Biol. Chem. 276:31099–31104.
43. Sakuraba, H., et al. 2002. Reduced beta-cell mass
and expression of oxidative stress-related DNA
damage in the islet of Japanese Type II diabetic
patients. Diabetologia. 45:85–96.
44. Sakai, K., et al. 2003. Mitochondrial reactive oxygen
species reduce insulin secretion by pancreatic β-cells.
Biochem. Biophys. Res. Commun. 300:216–222.
45. Feltus, F.A., Lee, E.K., Costello, J.F., Plass, C., and
Vertino, P.M. 2003. Predicting aberrant CpG
island methylation. Proc. Natl. Acad. Sci. U. S. A.
46. Bachman, K.E., et al. 2003. Histone modifications
and silencing prior to DNA methylation of a tumor
suppressor gene. Cancer Cell. 3:89–95.
47. Kouzarides, T. 2002. Histone methylation in
transcriptional control. Curr. Opin. Genet. Dev.
48. Fuks, F., Burgers, W.A., Brehm, A., Hughes-Davies,
L., and Kouzarides, T. 2000. DNA methyltrans-
ferase Dnmt1 associates with histone deacetylase
activity. Nat. Genet. 24:88–91.
49. Ogata, E.S., Bussey, M., and Finley, S. 1986. Altered
gas exchange, limited glucose, branched chain
amino acids, and hypoinsulinism retard fetal
growth in the rat. Metabolism. 35:950–977.
50. Kaminsky, Z., Sun-Chong, W., and Petronis, A.
2006. Complex disease, gender, and epigenetics.
Ann. Med. 38:530–544.
51. Ke, X., et al. 2006. Uteroplacental insufficiency
affects epigenetic determinants of chromatin struc-
ture in brains of neonatal and juvenile IUGR rats.
Physiol. Genomics. 25:16–28.
52. Finegood, D.T., Scaglia, L., and Bonner-Weir, S.
1995. Dynamics of beta-cell mass in the growing
rat pancreas. Estimation with a simple mathemati-
cal model. Diabetes. 44:249–256.
53. Stoffers, D.A., Heller, R.S., Miller, C.P., and Habener,
J.F. 1999. Developmental expression of the home-
odomain protein IDX-1 in mice transgenic for an
IDX-1 promoter/lacZ transcriptional reporter.