. 38
( 51 .)


nals in the ME into the hypophysial vasculature in an activity-dependent manner.
Of course, the ¬nal component that we have to consider in this schema is corticos-
terone, although the effects of corticosterone upon overall HPA function have been
extensively documented (e.g. Dallman et al., 1987; Jones and Gillham, 1988), the
speci¬c mechanisms by which it controls CRH neuronal function remain unclear.
As corticosterone can control a wide range of cellular processes, it is possible that
it can indirectly regulate mechanisms in all six compartments of the CRH neuron.
However, its direct actions within CRH neurons will be limited to effects on recep-
tors, signal transduction cascades, and genes (Figure 7.5; Brann et al., 1995; Burke
et al., 1997; Reichardt et al., 1998; Kovács et al., 2000). We will consider the speci¬cs
of these mechanisms later.
Figure 7.5 shows that two processes are targeted by processes that control the
activity of CRH neurons: (a) changes in membrane excitability (potential) and (b)
CRH heteronuclear (hn) RNA levels. However, each contributes quite differently
to CRH neuronal function: CRH hnRNA is an initial component in the path lead-
ing to peptide biosynthesis; membrane excitability is critical for stimulus“release
coupling that controls CRH release at the neuroendocrine terminal. In broader
terms, membrane excitability can be seen as controlling the immediate response of
CRH neurons to a stimulus; while changes in the expression of the CRH gene “ or
indeed any other gene “ can be seen as an adaptive response.
Again, it is worth remembering that there is great cell speci¬city about how
genes and their products are controlled. For example, the way corticosterone regu-
lates the CRH gene in the PVH is very different from the way this regulation occurs
in the CEAl (Watts and Sanchez-Watts, 1995a), the cerebral cortex, the human pla-
centa, or in vitro cell lines. This cell speci¬city emphasizes that where possible, we
should use in vivo models to examine regulatory mechanisms, and when focusing on
one cell type, we should use care if we infer mechanisms using results derived from
others (e.g. Dumont et al., 2002; King et al., 2002).

Multiple levels of interaction
As we have discussed, glucocorticoids can stimulate, repress or have no effect on
CRH gene expression in the brain depending on the cell type examined (Beyer et al.,
212 A. G. Watts

1988; Swanson and Simmons, 1989; Frim et al., 1990; Watts and Sanchez-Watts,
1995a). This shows that we cannot look solely to the CRH gene to explain gluco-
corticoid effects on CRH gene expression. Indeed, a great deal of evidence points
to the fact that the cellular actions of glucocorticoid are highly complex and utilize
mechanisms other than direct gene interactions (Beato et al., 1995; Brann et al.,
1995). In this way, there must be multiple levels at which glucocorticoids can act,
and this next section considers four possible levels of glucocorticoid interaction
with CRH control mechanisms: gene, cell, neural network, and time domains.

Actions on the gene
The inhibitory actions of glucocorticoids on CRH gene expression in the PVH
appear to be effected by glucocorticoid receptor (GR) rather than a mineralocor-
ticoid receptor (MR)-dependent mechanisms (Watts, 1996; Reichardt et al., 1998),
and a functional GR is an absolute requirement for glucocorticoid inhibition of
CRH gene expression in the PVH (Reichardt et al., 1998; Kretz et al., 1999). With
this in mind, the most direct way that glucocorticoids can control CRH gene
expression is to affect the rate of transcription through direct interactions between
the ligand-activated GR and GR-control elements on the CRH gene. Although
some studies suggest that the CRH gene does not contain a consensus glucocorti-
coid-regulatory element (GRE), there is evidence that regions of the CRH gene
will bind glucocorticoids. Malkoski and co-workers have identi¬ed a negative-
glucocorticoid-response element that mediates glucocorticoid repression of cyclic
adenosine monophosphate (cAMP)-stimulated but not basal CRH gene expres-
sion in transfected AtT20 cells (Malkoski et al., 1997; Malkoski and Dorin, 1999).
Furthermore, King et al. (2002) have recently used transfected AtT20 cells to iden-
tify a second cAMP-response element (CRE) on the CRH gene that is distinct from
the consensus CRE. They conclude that different regions of the CRH gene confer
the inhibitory and stimulatory actions of glucocorticoids.
However, it remains to be determined whether these mechanisms actually oper-
ate in the neuroendocrine PVH in vivo. Indeed evidence for a direct action of the
ligand-activated GR binding to the CRH gene in the PVH is currently inconclusive.
The fact that corticosterone applied directly to the PVH in vivo appears to have lit-
tle effect on CRH mRNA levels in ADX rats (Kovács and Mezey, 1987) is consistent
with more indirect actions. Furthermore, some intriguing evidence suggests that
glucocorticoid downregulation of CRH gene expression in the PVH in vivo may
not in fact involve DNA binding at all. Thus, mutant mice that have a GR incapable
of binding to DNA (GRdim/dim) have normal CRH peptide levels in the ME
(Reichardt et al., 1998). In contrast, proopiomelanocortin (POMC) gene expres-
sion in these same mice is markedly upregulated showing that DNA binding is
required for the GR regulation of this particular gene.
213 Glucocorticoids and the ups and downs of neuropeptide gene expression

Cell: actions on mechanisms
Receptor mechanisms
CRH neurons express a host of transmitter (gamma amino butyric acid (GABA),
glutamate, monoamine) and peptide receptors (Whitnall, 1993; Herman et al.,
2002). In turn, corticosterone can modify receptor function in the PVH (Figure
7.5), for example angiotensin receptors (Aguilera et al., 1995; Shelat et al., 1998),
neuropeptide Y (NPY) receptors (Akabayashi et al., 1994), and adrenoreceptors
(Jhanwar-Uniyal and Leibowitz, 1986; Day et al., 1999).

Signal transduction mechanisms
There is strong evidence that glucocorticoids can downregulate CRH gene expres-
sion by modifying those signal transduction mechanisms that directly control
transcription (Figure 7.5). Currently the best-de¬ned direct transcriptional regu-
lator of the CRH gene is cAMP, which regulates transcription in many cell types
using the CRE-binding protein (CREB). The CRH gene contains a functional CRE,
and CRH gene transcription in vitro is increased by agents, such as forskolin, that
increase cAMP using protein kinase (PK) A, rather than PKC-associated mecha-
nisms (Majzoub et al., 1993). Much evidence suggests that unlike many other tran-
scription factors that act by binding to a DNA promoter sequence, CREB is usually
constitutively bound to the appropriate promoter of the target gene. However, CREB
does not apparently bind constituently to the CRH gene, but is phosphorylated to
form pCREB by a stimulus-initiated PKA-dependent cascade. Only then does
pCREB bind to a CREB-binding protein (CBP) to interact with the CRE on the
CRH gene so that the pCREB/CBP/CRE complex initiates CRH gene transcription
(Wol¬‚ et al., 1999).
Glucocorticoids can repress the stimulatory actions of cAMP and CREB on
CRH gene expression in vitro (Majzoub et al., 1993; Guardiola-Diaz et al., 1996;
Malkoski et al., 1997). But the fact that in these same systems glucocorticoid has
no effect on the unstimulated rates of CRH gene transcription suggests that the
actions of glucocorticoid requires some form of coincidence between the receptor
activation and the appropriate signal transduction pathway to exert its effects.
Direct in vivo evidence for the transcriptional activation of the CRH gene by
way of a pCREB-dependent mechanism, or indeed any other signal transduction
mechanism, remains scant. However, data do suggest that the activation of signaling
molecules and CRH gene expression can occur very rapidly following appropriate
stimulation. In one experiment, Kovacs and Sawchenko (1996a) used a pulse-chase
design to show that a brief episode of ether anesthesia triggers a cascade of cellular
events beginning within 5 min of the stressor with the concurrent accumulation
of immunocytochemically detectable pCREB and the CRH primary transcript;
increases in CRH mRNA levels and AVP hnRNA followed later.
214 A. G. Watts

Increases in other signaling processes are similarly rapid. For example, Khan and
Watts (2004) showed that intravenous 2-deoxyglucose (2-DG) elevates CRH gene
transcription together with the phosphorylation of the mitogen-activated protein
(MAP) kinases, Erk 1/2, within 10 min in CRH neurons. Similar increases in Erk
1/2 phosphorylation are seen after local norepinephrine injections into the region
of the PVH (Khan and Watts, 2003) suggesting that these signaling kinases can act
as intermediaries between catecholaminergic inputs and CRH gene expression fol-
lowing 2-DG. Although little work has been performed in vivo to determine how
glucocorticoids interact with CREB and MAP kinase signaling systems, evidence
shows that the GR agonist dexamethasone modulates stress-induced accumulation
of pCREB in CRH neurons (Legradi et al., 1997).


Evidence presented in the previous two sections (Gene and Cell) strongly support
the notion that the inhibitory actions of glucocorticoid on CRH gene expression in
the PVH are complex. If this is the case then we also need to look outside the PVH
for answers and examine potential mechanisms that operate at the network level.
In this section I will brie¬‚y discuss a model of the afferent inputs to CRH neuro-
endocrine neurons, and then use this as a framework to examine glucocorticoid

Efferent organization
Afferent control processes involve a range of neural and hormonal components
that operate with bewildering complexity. To provide a framework for understand-
ing the functional organization of CRH neuronal-control processes, for experi-
mental design, and for interpreting data, we consider that a hierarchically ordered
model of neural and corticosterone-dependent-control processes is a useful work-
ing hypothesis (Figure 7.6). This model is based on the accepted notion that CRH
neuroendocrine neurons are motor neurons, since they control the activity of non-
neuronal cells outside the brain: that is, corticotropes. Hierarchical models have
long proved useful for explaining neural control of the somatic motor system, and
it seems reasonable to use this type of organization as a framework for exploring
the neural control of CRH gene expression, at least as a ¬rst approximation
(Schneider and Watts, 2002; Watts and Swanson, 2002). The advantages of this
approach is that it provides a useful and manageable way for organizing afferent
inputs, and also encourages us to think about the different control mechanisms in
a more integrative way, rather than considering each as being an isolated system.
To do this we have taken those neural systems known to control CRH neural func-
tion and divided them into two broad categories: Level 1 neurons make synaptic
215 Glucocorticoids and the ups and downs of neuropeptide gene expression

Other inputs Other inputs

Level 2 Level 2
afferents afferents

Level 1 3
Level 1 Level 1
afferents afferents afferents

CRH neuron CRH neuron CRH neuron
ACTH release ACTH release ACTH release

(a) (b) (c) Corticosterone
Figure 7.6 The neural afferents controlling the activity of CRH neurons can be categorized
hierarchically. (a) Level 1 neurons, either individually or as part of a more complex
network, synapse directly upon CRH neurons; (b) while Level 2 neurons control CRH
neurons indirectly by way of Level 1 neurons and (c) glucocorticoids can regulate
CRH neuronal activity using both hierarchical levels

contact with CRH neurons and control their function directly (Figure 7.6(a)).
These can be considered analogous to pre-motor neurons in the somatic motor
system. Examples include NPY/GABAergic neurons in the arcuate nucleus
(Schneider and Watts, 2002), and local glutamatergic neurons (Herman et al.,
2002). Catecholaminergic inputs that originate in the hindbrain and encode
interosensory information and provide important regulatory control of CRH gene
expression (Swanson and Sawchenko, 1983; Ritter et al., 2003) also fall into this
It is important to note that Level 1 afferents are not organized as a parallel array
of independently acting inputs. Interactions between them will form networks that
allow for a more sophisticated level of control. Figure 7.6(a) illustrates a simple
example, where one set of afferents collateralizes with another either at its cell body
or at its terminal; interactions between catecholaminergic inputs to the para-
ventricular and the dorsomedial nucleus of the hypothalamus (PVH and DMH,
respectively) (Thompson and Swanson, 1998) are probably arranged in this
Level 2 neurons in¬‚uence CRH neuronal function indirectly by altering the
signaling properties of Level 1 neurons. These interactions may occur outside the
PVH “ for example, hypothalamic projections into the DMH (Thompson and
Swanson, 1998); amygdalar projections to the bed nucleus of the stria terminalis
216 A. G. Watts

(BST) (Dong et al., 2001) “ or they may occur more proximally, for example at the
pre-synaptic terminal of Level 1 neurons (Figure 7.6(b)). Level 2 neurons provide
opportunities for a wide range of neural in¬‚uences to control CRH function indi-
rectly. For example, ventral subicular neurons would be considered Level 2 neurons
because they affect CRH function by way of the BST (Cullinan et al., 1993). The
subiculum, in turn, processes information from many cortical areas that ultimately
affect CRH neuronal function (Swanson, 2000).
Stressors and other modulatory in¬‚uences are going to control CRH neurons by
engaging distinct ˜afferent sets™. Each set will consist of arrays of Levels 1 and 2 con-
trol neurons, the constituency of which being determined by the sensory com-
position of the stimulus. Sets encoding different stimuli may contain distinct or
common individual afferent groups. For example, the afferent set encoding the
effects of dehydration on CRH gene expression will contain both similar and dis-
tinct afferents to the sets encoding the effects of hypovolemia or starvation (Watts,
1996; Watts and Sanchez-Watts, 2002).
Importantly, this arrangement also offers a useful framework for thinking
about the way glucocorticoids affect CRH neuronal function. We suggest that there
are at least three spatial domains in which corticosterone can operate (Figure

(1) Direct actions on CRH neurons (as just discussed in the Gene and Cell
(2) Actions by way of corticosterone-sensitive afferents.
(3) Corticosterone-sensitive physiological processes, whose effects on CRH neu-
rons are then mediated by way of neural afferents, of which its effects on energy
metabolism are an example (Laugero et al., 2001).

Time domains
In a classic review Keller-Wood and Dallman (1984) discussed the importance of
different time domains (short-term, intermediate, and long-term) when consider-
ing how glucocorticoids regulate ACTH release. They also noted that each of these
domains involved different mechanisms ranging from actions on the corticotrope
to alterations on the neural systems that regulated secretogogue release. There is
evidence to suggest that similar time-domains are important when considering
glucocorticoids actions on gene expression.
A basic property of the way corticosterone regulates the overall level of CRH
gene transcription is the time required for transcription to respond to changes in
circulating corticosterone. Evidence suggests that the time frames for its actions
on AVP and CRH gene expression are very different. Using intra-peritoneal (i.p.)
bolus injections of supra-physiological doses of corticosterone, Ma et al. (1997)
217 Glucocorticoids and the ups and downs of neuropeptide gene expression

400 1500

CRH mRNA (% sham ADX)


. 38
( 51 .)