. 27
( 51 .)


forms a complete barrier to glucocorticoid access (White et al., 1997; Kotelevtsev
et al., 1999). In the placenta, however, the enzyme is not a complete barrier to mater-
nal steroids (Benediktsson et al., 1997) and in rodents the peak of the circadian
rhythm of plasma corticosterone is able to penetrate the 11 -HSD-2 barrier to
some extent (Venihaki et al., 2000), presumably adding to the provision of glucocor-
ticoids to the fetus for normal key developmental processes such as maturation of the
lung. However, dexamethasone readily passes the placenta as it is a poor 11 -HSD-2
substrate (Albiston et al., 1994; Brown et al., 1996b). Betamethasone is presumed a
similarly poor substrate, but 11 -HSD-2 rapidly inactivates prednisolone to inert

Uterine artery Maternal

11 -HSD-1


Umbilical vein
Active cortisol (corticosterone)

Inert cortisone (11-dehydrocorticosterone)

Figure 5.2 Physiology: placental 11 -HSD-2 converts glucocorticoids to inert forms thus excluding
active maternal steroids from the fetal compartment
148 J. R. Seckl et al.

Placental 11 -HSD-2 and birth weight
Observational studies have suggested that placental 11 -HSD-2 relates to birth
weight. The activity of placental 11 -HSD-2 near term shows considerable inter-
individual variation in humans and rats (Benediktsson et al., 1993; Stewart et al.,
1995). A relative de¬ciency of 11 -HSD-2, with consequent reduced placental
inactivation of maternal steroids, may lead to overexposure of the fetus to gluco-
corticoids, retard fetal growth and programme responses leading to later disease
(Edwards et al., 1993). Studies in rats have demonstrated that lower placental
11 -HSD-2 activity is seen in the smallest fetuses with the largest placentas
(Benediktsson et al., 1993). Similar associations have been reported in humans
(Stewart et al., 1995; Shams et al., 1998; McTernan et al., 2001; Murphy et al., 2002),
although not all studies have reproduced this ¬nding (Rogerson et al., 1996; 1997).
Additionally, markers of fetal exposure to glucocorticoids such as cord blood levels
of osteocalcin (a glucocorticoid-sensitive osteoblast product that does not cross
the placenta), also correlate with placental 11 -HSD-2 activity (Benediktsson
et al., 1995).
Rare human cases of 11 -HSD-2 de¬ciency are described, with homozygotes (or
compound heterozygotes) substantially de¬cient in 11 -HSD-2 activity due to
mutations in the encoding gene. Whilst children and adults exhibit ˜apparent miner-
alocorticoid excess™ due to illicit activation of renal MR by cortisol (Stewart et al.,
1988), and an identical adult phenotype is seen in 11 -HSD-2 knockout mice
(Kotelevtsev et al., 1999), affected individuals have very low birth weight (Dave-
Sharma et al., 1998), averaging 1.2 kg less than their heterozygote siblings. Though an
initial report suggested that 11 -HSD-2 null mice have normal fetal weight in late
gestation (Kotelevtsev et al., 1999), this appears to have re¬‚ected the ˜genetic noise™ of
the crossed (129 MF1) strain background of the original 11 -HSD-2 null mouse.
Indeed preliminary data suggest that in congenic mice on the C57Bl/6 strain back-
ground 11 -HSD-2 nullizygosity lowers birth weight (Holmes et al., 2002).
Additionally, there may also be species differences. Thus, the mouse shows dramatic
late gestational loss of placental 11 -HSD-2 gene expression (Brown et al., 1996a),
whereas in humans, placental 11 -HSD-2 activity increases through gestation
(Stewart et al., 1995).

An introduction to experimental studies of early life glucocorticoid exposure
Given the plausible links between glucocorticoid excess and the epidemiological ¬nd-
ings, a causal role was hypothesized soon after the initial human observations were
reported (Edwards et al., 1993). This notion has been addressed in a variety of experi-
mental models. Importantly, the animal data have also been used to re-examine
mechanisms and the phenotype of human low-birth-weight populations to ˜complete
149 Prenatal glucocorticoids and the programming of adult disease

Diet Stress Disease
% conversion active ’ inert
35 Placental

• Cortisol
10 • Other placental
and maternal
22 18 9
Maternal dietary protein intake (%)

Figure 5.3 Placental 11 -HSD-2 is downregulated by maternal dietary protein restriction. This may
be one common mechanism linking the maternal and fetal environments. Adapted from
Langley-Evans et al. (1996b)

the circle™. Broadly, two approaches have been employed. Most workers have used
synthetic glucocorticoids such as dexamethasone and betamethasone that relatively
freely cross the placenta as they are poor substrates for 11 -HSD-2. Such agents are
also used clinically in obstetric practice. In addition, some studies have exploited
drugs such as carbenoxolone that inhibit 11 -HSD-2 thus increasing feto-placental
exposure to endogenous steroids. The data from both approaches are in general
complementary and here the key ¬ndings are reviewed.
It is noteworthy that a common mechanism may underlie fetal programming
through maternal undernutrition and glucocorticoid exposure (Figure 5.3). Dietary
protein restriction during rat pregnancy selectively attenuates 11 -HSD-2, but appar-
ently not other placental enzymes (Langley-Evans et al., 1996b; Bertram et al., 2001;
Lesage et al., 2001). Indeed in the maternal protein restriction model, offspring
hypertension can be prevented by treating the pregnant dam with glucocorticoid
synthesis inhibitors, and can be recreated by concurrent administration of cortico-
sterone, at least in female offspring (Langley-Evans, 1997).
As the maternal glucocorticoid levels are much higher than those of the
fetus, subtle changes in placental 11 -HSD-2 activity may have profound effects
on fetal glucocorticoid exposure (Lopez-Bernal et al., 1980; Lopez Bernal and
Craft, 1981). A relative de¬ciency of placental 11 -HSD-2 therefore has far
greater potential consequences in terms of the fetal glucocorticoid load than
any alteration in fetal adrenal steroid production, once the capacity of the fetal
hypothalamic“pituitary“adrenal (HPA) axis to suppress fetal adrenal output has
been overwhelmed.
150 J. R. Seckl et al.

Peripheral programming

Blood pressure and cardiovascular control
The human epidemiology began by showing low birth weight associates with adult
heart disease mortality and hypertension (Barker et al., 1989a, b; RichEdwards et al.,
1997). These associations remain arguably the most robust in the literature
(Huxley et al., 2000). It is unsurprising therefore that investigators have addressed
the role of antenatal insults, including glucocorticoid excess, upon cardiovascular
parameters. From these studies it appears that prenatal glucocorticoid exposure usu-
ally produces permanently elevated offspring blood pressure in later life, as assessed
by the direct (semi-restrained animals with chronically catheterized vessels) and indi-
rect (tail cuff) methods employed. It should be noted that these techniques involve
an inherent component of stress in their measurement and true basal pressures
(using ˜gold standard™ telemetric approaches) remain as yet unreported.
In utero, cortisol infusion into the fetus elevates blood pressure in sheep
(Tangalakis et al., 1992). Betamethasone given to pregnant baboons has similar
hypertensive effects on the fetuses (Koenen et al., 2002). Excess cortisol also directly
elevates blood pressure at birth in humans (Kari et al., 1994) and sheep (Berry et al.,
1997). Such effects appear to persist.
Thus, treatment of pregnant rats with dexamethasone, a synthetic glucocorticoid
used in obstetric practice which readily crosses the placenta, reduces birth weight, a
de¬cit reversed by weaning. Both male and female adult offspring of dexamethasone-
treated pregnancies have elevated blood pressures (Benediktsson et al., 1993).
Similarly, adult hypertension is produced in sheep exposed to excess glucocorticoid
in utero, either as maternally administered dexamethasone or as a maternal cortisol
infusion (Dodic et al., 1998; 1999; 2002a, b; Jensen et al., 2002). The timing of gluco-
corticoid exposure appears to be important; exposure to glucocorticoids during the
¬nal week of pregnancy in the rat is suf¬cient to produce permanent adult hyperten-
sion (Levitt et al., 1996; Sugden et al., 2001), whereas the sensitive window for such
effects in sheep are earlier in gestation (Gatford et al., 2000). Such differences may be
primarily due to the complex species-speci¬c patterns of expression of GR, MR and
the isoenzymes of 11 -HSD, which are crucial in both the regulation of maternal
glucocorticoid transfer to the fetus, and in modulating glucocorticoid action at the
tissue level. So, excess exposure to exogenous glucocorticoid can programme car-
diovascular physiology, but outside obstetric pharmacotherapy does this matter to
the majority of low-birth-weight babies?
Inhibition of 11 -HSD by treatment of pregnant rats with carbenoxolone has
effects similar to dexamethasone, leading to offspring of modestly reduced birth
weight. This associates with increased passage of maternal corticosterone to the
fetal plasma. Although the weight de¬cit is typically regained by weaning, as with
151 Prenatal glucocorticoids and the programming of adult disease

dexamethasone, prenatal carbenoxolone-exposed rats develop adult hypertension
(Lindsay et al., 1996a). These effects of carbenoxolone are independent of changes
in maternal blood pressure or electrolytes, but require the presence of maternal
glucocorticoids; the offspring of adrenalectomized pregnant rats are protected
from carbenoxolone actions upon birth weight or adult physiology. It must be
noted that carbenoxolone is non-selective and inhibits the other 11 -HSD isozyme
(type 1) (HSD-1) and related dehydrogenases. However, congenic 11 -HSD-2 null
mice also have low birth weight and preliminary data suggest that they show
programming of CNS development and adult functions such as anxiety-related
behaviours (see below).

Mechanisms of cardiovascular programming by prenatal glucocorticoids
Exploring such rodent models, the mechanisms of glucocorticoid-programmed
adult hypertension have been studied. These are thought to involve a variety of
processes that are also likely to have distinctive windows of sensitivity. Thus, pre-
natal glucocorticoids lead to reductions in nephron number (Ortiz et al., 2001) which
are largely irreversibly determined around birth. In addition, antenatal glucocorti-
coid exposure affects fetal and adult vascular responses to vasoconstrictors, enhancing
endothelin-induced vasoconstriction in association with abnormal endothelium-
dependent relaxation at least in sheep (Molnar et al., 2002; 2003), indicating
microvascular dysfunction. Analogous ¬ndings occur in rats (Hadoke et al., unpub-
lished data). The vascular changes may re¬‚ect the programming of receptors and
post-receptor mechanisms in the vascular wall and other cardiovascular structures.
These effects appear to be vascular bed speci¬c (Docherty et al., 2001), underlining
the exquisite complexity of the systems involved. Also, renin“angiotensin system
(RAS) parameters including receptor density and tissue RAS component synthesis
are affected by antenatal steroid exposure (Dodic et al., 2001), notably within the
fetal kidney (Moritz et al., 2002) where angiotensinogen, the AT1 and AT2 receptors
are increased after dexamethasone, accompanied by a reduced glomerular ¬ltration
rate response to angiotensin II. Finally, key brain stem barocontrol centres are altered
by prenatal glucocorticoid exposure (Dodic et al., 1999). These actions may combine
to form an adult with multiple processes contributing to hypertension. Which
processes are key remains to be discerned and may, of course, differ between species
and the timing of the exposure. Thus, the same apparent adult phenotype may
clearly be underpinned by distinct ˜programmed™ processes, a notion we return to
below when addressing neuroendocrine programming.

Programming the heart?
A key component of the human early life origins phenomenon is an increased risk of
cardiovascular death in adults who were of low birth weight (Barker et al., 1989b;
152 J. R. Seckl et al.

RichEdwards et al., 1997). This may merely re¬‚ect the sum of increased cardiovascu-
lar risk factors such as hypertension and metabolic disorders, but primary cardiac
programming may also be involved. In support of the latter possibility, prenatal gluco-
corticoid exposure alters the trajectory of development of cardiac noradrenergic
innervation and sympathetic activity (Bian et al., 1993), increases cardiac adenylate
cyclase reactivity to a range of stimuli (Bian et al., 1992) and alters key cardiac meta-
bolic regulators such as the glucose transporter 1, akt/protein kinase B, speci¬c uncou-
pling proteins (UCP) and the nuclear receptor for fatty acids PPAR (Langdown et al.,
2001a, b). Perhaps crucially, given the documented association between overexpres-
sion of cardiac calreticulin and cardiac dysfunction and death, antenatal glucocor-
ticoid exposure increases calreticulin levels markedly in the adult heart (Langdown
et al., 2003). Thus, these experimental models teach that increased coronary heart
disease in low-birth-weight populations may re¬‚ect both an increased prevalence
of major cardiovascular risk factors as well as primary cardiac dysfunction.

Programming of glucose“insulin homoeostasis and metabolic functions
Prenatal glucocorticoid overexposure also ˜programmes™ permanent hyperglycaemia
and, particularly, hyperinsulinaemia in the adult offspring in the rat (Nyirenda et al.,
1998; Sugden et al., 2001), effects delimited to the last third of gestation. Gestational
11 -HSD inhibition has similar adult hyperglycaemic effects (Lindsay et al., 1996b).
Earlier gestational dexamethasone exposures or post-partum steroids do not pro-
gramme hypergylcaemia/hyperinsulinaemia in the rat, de¬ning a tight window for
this effect (Nyirenda et al., 1998; 2001). Maternal glucocorticoid administration has
an effect on cord glucose and insulin levels in the ovine fetus (Sloboda et al., 2002b).
Adult glucose“insulin dyshomoeostasis also occurs in sheep exposed to dexametha-
sone in utero (Dodic et al., 1998; Gatford et al., 2000), though the sensitive ˜windows™
again appear to be earlier than in the rat and can be dissociated from those produc-
ing with hypertension. Importantly, in the ovine model, antenatal glucocorticoid
exposure alters adult glucose metabolism whether or not there is prior fetal growth
restriction (Moss et al., 2001). Speci¬cally, maternal but not fetal injections of
betamethasone restrict fetal growth (Newnham et al., 1999); however, offspring of
both the groups have altered adult glucose dynamics (Moss et al., 2001). Thus, it
appears that the programming effects on glucose“insulin homoeostasis in this
model relate to fetal exposure to excess glucocorticoids in utero, rather than any pri-
mary effect of intrauterine growth retardation per se.

Mechanisms of glucocorticoid-programmed hyperglycaemia/insulin resistance
Several important hepatic metabolic systems are regulated by glucocorticoids, includ-
ing key enzymes of carbohydrate metabolism such as phosphoenolpyruvate car-
boxykinase (PEPCK), a rate-limiting enzyme in gluconeogenesis. In rats, exposure
153 Prenatal glucocorticoids and the programming of adult disease



. 27
( 51 .)