Last updated April 2007
The Arnsten Lab studies molecular influences on the higher cognitive functions
of the prefrontal cortex (PFC), with the overarching goal of developing rational
treatments for cognitive disorders and mental illness. The lab uses a multi-disciplinary
approach to understand mechanisms influencing working memory at the cellular and
behavioral levels. Research has focused on how the catecholamines norepinephrine (NE) and
dopamine (DA), powerfully and dynamically modulate PFC cognitive function and physiology
through intracellular signaling mechanisms. Our data explain how exposure to stress
causes the rapid loss of PFC cognitive abilities, and how genetic mutations in molecules
that regulate these pathways can lead to symptoms of mental illness. Dysregulation of
these pathways also contributes to cognitive decline with normal aging. Understanding
these mechanisms has led to successful new treatments for patients with PFC dysfunction,
including medications for Attention Deficit Hyperactivity Disorder, Post-Traumatic Stress
Disorder and a potential treatment for schizophrenia and bipolar disorder.
Disclosure statements: Dr. Amy F.T. Arnsten and Yale University have license agreements with Shire Pharmaceuticals for the development of guanfacine for the treatment of Attention Deficit Hyperactivity Disorder and other disorders that involve prefrontal cortical dysfunction, and with Marinus Pharmaceuticals for the development of chelerythrine for the treatment of schizophrenia, bipolar disorder, post-traumatic stress disorder and related conditions. Dr. Arnsten also consults with these companies to help translate the basic research into treatments for human cognitive disorders.
Patricia Goldman-Rakic considered working memory to be a fundamental building block of cognition: the ability to represent information no longer in the environment, and to use this representational knowledge to guide behavior, thought and affect. This representational knowledge is used to overcome distraction or prepotent responses, and to maintain correct information in the face of interference. Goldman-Rakic used a spatial working memory paradigm to uncover the neural basis of these working memory abilities, and found that representational knowledge is encoded by networks of prefrontal cortical (PFC) pyramidal cells with shared stimulus properties, engaged in recurrent excitation (schematically illustrated in this drawing by Dr. S. Mark Williams, Pyramis Studios). Spatial tuning is heightened through GABAergic, inhibitory connections between networks with dissimilar spatial properties (e.g. Rao et al, J. Neurosci 20: !
485, 2000). The working memory abilities of the PFC are also highly dependent on the neuromodulatory environment, whereby loss of catecholamines in PFC is as detrimental as destruction of the PFC itself (Brozoski et al, Science 205: 929 1979).
Neurochemical Regulation of Prefrontal Cortex
The Arnsten Lab has shown that both dopamine (DA) and norepinephrine (NE) have “inverted U” influences on PFC cognitive functions. Thus, either too little or too much DA or NE impairs working memory at the cellular (shown) and behavioral levels. DA exhibits both beneficial and detrimental effects at D1/D5 receptors depending upon the amount of stimulation and the cognitive operation required, (Arnsten and Goldman-Rakic, 1990; Zahrt et al., 1997; Vijayraghavan et al, 2007). In contrast, NE has beneficial effects at post-synaptic α2A receptors, and detrimental actions at high levels of a1 receptor stimulation. This dissociation of receptor type with NE has allowed more expedient drug development for the treatment of PFC dysfunction.
Previous research has shown that the beneficial effects of NE α2-receptor stimulation result from actions at receptors POST-synaptic to NE neurons (Arnsten and Goldman-Rakic, 1985; Cai et al., 1993), and of the α2A subtype (Franowicz et al, 2002). Blockade of α2 receptors in PFC with yohimbine markedly impairs working memory and impulse control (Li and Mei, 1994; Ma et al, 2003) and reduces delay-related firing (Li et al., 1999), while stimulation of these receptors with guanfacine improves a variety of PFC cognitive functions and enhances delay-related firing (reviewed in Arnsten and Li, 2005). NE appears to not have significant actions at α1 receptors during nonstress conditions, as α1 antagonists have little effect under these conditions. However, NE appears to have marked effects at these receptors during stress exposure, as α1 antagonists protect PFC functions from stress exposure, !
while α1 agonists mimic the stress response (e.g. Arnsten et al., 1999; Birnbaum et al., 1999; Mao et al, 1999).
Under optimal conditions for PFC cognitive function- i.e. alert, nonstressed, interested- tonic firing of NE neurons is low, and phasic firing to relevant stimuli is enhanced (see the work of Gary Aston-Jones). Under these conditions, moderate levels of NE release engage high affinity α2A receptors, inhibit cAMP-HCN signaling, and strengthen network connectivity between neurons with shared characteristics, thus increasing “signal” (see below). Similarly, the work of Wolfram Schultz shows that DA neurons have low tonic firing under these conditions, and increase their firing to the cue preceding the delay period in a delayed response task. Our work indicates that optimal levels of DA D1 receptor stimulation activate cAMP-HCN signaling to reduce nonpreferred inputs, thus reducing “noise”.
Optimal neurochemical conditions for PFC cognitive function: NE stimulates α2A receptors, inhibits cAMP-HCN signaling, and strengthens network connectivity between neurons with shared characteristics, thus increasing delay-related firing for the preferred direction (i.e. “signal”). Modest levels of DA D1 receptor stimulation activate cAMP-HCN signaling on spines receiving nonpreferred inputs, thus reducing “noise”. The width of tuning can thus be dynamically modulated by DA based on current cognitive demands.
From Arnsten, Cerebral Cortex, epub 2007, doi: 10.1093/cercor/bhm033
EM studies by Constantinos Paspalas have shown that HCN channels are localized on the heads (A) and necks (B) of dendritic spines in layer II-III of PFC. These channels are often co-localized with α2A receptors, e.g. on spine heads (see C) or spine necks (D). Thus, α2A receptors are positioned to control the local concentration of cAMP, and determine whether HCN channels are more likely to be in a closed or open state. From Wang et al., 2007.
A model illustrating how HCN channels on PFC spines can gate information coming into the neuron based on the level of cAMP in the spine compartment. Under conditions of high cAMP production, the open probability of HCN channels is increased, and incoming information would be shunted due to reduced membrane resistance. When cAMP production is inhibited by α2A adrenoceptor stimulation near the HCN channels, the channels close and allow information to pass into the cell. From Wang et al., 2007.
Support for this model can be observed at the cellular and behavioral levels. The collapse in firing induced by α2 receptor blockade (yohimbine) can be rescued by blocking HCN channels with ZD7288 (above). Either blockade or knockdown of HCN channels in rat PFC improves spatial working memory performance (below). From Wang et al., 2007.
Changes in these signaling pathways likely contribute substantially to declines in PFC cognitive function with normal aging. Our data suggest that cAMP signaling is disinhibited in the aged PFC (Ramos et al., Neuron 40: 835, 2003), and contributes to working memory deficits. This finding is consistent with data showing reduced α2 receptor binding in the aged PFC in animals with cognitive deficits (Moore et al., 2005). There are also substantial decreases in DA in the aged PFC, and loss of VTA neurons, whereas NE neurons remain but are likely underactive due to loss of orexin inputs.
Genetic changes in extracellular molecules that weaken catecholamine transmission have been linked to ADHD. Inadequate catecholamine modulation of PFC function may be normalized by stimulant (methylphenidate (MPH), amphetamine) and nonstimulant (atomoxetine (ATX) medications that block NE and DA transporters, and by guanfacine (GFC)) (reviewed in Arnsten, Neuropsychopharm 3: 2376, 2006). Importantly, Berridge has shown that low doses of methylphenidate have larger effects on NE than DA in PFC (Berridge et al, 2006). Similarly, Li has shown that blockade of α2 receptors in PFC is sufficient to recreate symptoms of ADHD, including poor impulse control and locomotor hyperactivity (reviewed in Arnsten and Li, 2005).
Based on research from the Arnsten lab, the α2A adrenoceptor agonist, guanfacine (GFC), is being developed by Shire Pharmaceuticals for the treatment of ADHD. Guanfacine is also being tested in a number of additional disorders that involve weakened PFC function.
Exposure to even quite mild uncontrollable stress impairs PFC cognitive functioning (reviewed in Arnsten, Science 280: 1711,1998). This may have survival value when we are in danger, but is often detrimental in the Information Age when we depend on PFC functions to steer us through massive interference. It also leaves us vulnerable to mental illnesses such as depression and Post-Traumatic Stress Disorder (PTSD).
We have now learned many of the receptor and second messenger mechanisms contributing to stress-induced PFC dysfunction: High levels of DA and NE release engage D1 receptor activation of cAMP signaling, and α1 activation of phosphotidyl inositol protein kinase C (PKC) signaling (NE may also impair via β1 receptors coupled to Gs). These actions markedly and rapidly impair PFC cognitive function. The detrimental effects of these pathways on PFC neuronal firing are shown below.
Hormones influence the stress response as well. Gluco-corticoid release during stress also contributes to impaired PFC function (e.g. the work of S. Lupien in humans, and B. Roozendal and J. Taylor in rats), and may due so in part by inhibiting uptake of catecholamines by extraneuronal catecholamine transporters (Arnsten, 2000). Studies in rats indicate that stress-induced PFC dysfunction is exacerbated by estrogen (Shansky et al. 2004; 2006), which may account for the greater prevalence of depression in human females during cycling years.
In contrast, activation of these same neurochemical pathways strengthens the affective functions of the amygdala (e.g. the work of Roozendaal). Thus, during stress, the balance shifts from the intelligent PFC to more primitive amygdala regulation of behavior and emotion (reviewed in Arnsten, 1998).
Chronic exposure to stress accentuates this situation. Chronic stress leads to potentiated NE transmission (Miner et al, 2006) and dendritic spine loss in the PFC (Radley et al, 2006; Liston et al, 2006), but increased dendritic complexity in the amygdala (Mitra et al., 2005). These changes likely contribute to clinical disorders such as depression and Post-Traumatic Stress Disorder (PTSD).
Both schizophrenia and bipolar disorder involve profound dysfunction of the PFC, with schizophrenia associated with bilateral dysfunction of the dorsolateral PFC (eg Weinberger et al, Arch Gen Psych 1986), and mania with hypofunctioning of the right ventral PFC (e.g. Blumberg et al, Arch Gen Psych 2003). Recent genetic studies find linkages with a variety of molecules involved in brain development and glutamate transmission; however, some of these molecules also serve as “intracellular brakes” on the signaling cascades that induce loss of PFC function during stress. In particular, DISC1 normally serves to inhibit cAMP signaling, and RGS4 inhibits Gq signaling. Thus, loss of function of these molecules would render patients particularly vulnerable to stress exposure. The work of Manji and Lenox has also indicated that phosphotidyl inositol signaling may be disinhibited in patients with bipolar disorder, and that inhibition of PKC m!
ay be helpful in the treatment of mania. Indeed, the recent publication of the bipolar disorder genome showed that the molecule most altered in this disorder is DAG kinase (Baum et al, 2007), an enzyme that prevents DAG from activating PKC. Thus, alteration in this molecule would lead to excessive PKC signaling.
DISC1 has been found in spines in human PFC (Kirkpatrick et al, 2006). Under conditions of high cAMP production (e.g. stress), DISC1 activates PDE4B (phosphodiesterase 4B) to destroy cAMP (Millar et al, 2005). This in turn should lead to HCN channels
closing, and PFC networks connecting.
Genetic alternations in DISC1 result in loss of function of this molecule (Millar et al, 2000). This would lead to insufficient regulation of PDE4B, and excessive cAMP levels, especially under conditions such as stress. This would result in HCN channel opening, and networks disconnecting.
Stress-Induced Collapse of PFC Networks in Mental Illness
Detrimental neurochemical conditions for PFC cognitive function: During stress, high levels of DA and NE are released in the PFC. High levels of DA D1 receptor stimulation would activate cAMP-HCN signaling throughout the dendrite, thus disconnecting all inputs and causing network collapse. High levels of NE may also conotribute to this mechanism via b1 receptor stimulation of cAMP production. The disconnection of PFC networks is schematically illustrated in the top left graph. The middle left graph shows actual collapse in PFC neuronal firing in the presence of excessive cAMP levels (Wang et al, 2007). Mutations in DISC1 in patients with mental illness would increase vulnerability for network collapse (lower left graph). The right graphs illustrate the effects of high levels of NE release engaging lower affinity a1 receptors, which in turn activates phosphotidyl inositol protein kinase C (PKC) signaling. As shown !
in the right middle graph, this pathway reduces PFC neuronal firing. Genetic or environmental events that weaken RGS4 or DAG kinase would disinhibit this pathway, and lead to marked impairment of PFC function (lower right graph; Birnbaum et al, 2004). In summary, genetic disruptions in intracellular signaling that weaken regulation of the intracellular response to stress may lead to more profound alterations in PFC regulation of thought and affect in these serious mental disorders. From Arnsten, Cerebral Cortex, epub 2007, doi: 10.1093/cercor/bhm033
Based on this research, the PKC inhibitor, chelerythrine (CHEL), is now in preclinical safety testing for the potential treatment of bipolar disorder and schizophrenia. It is noteworthy that the atypical antipsychotics all block a1 and 5HT2A receptors, thus reducing signaling through this pathway. The work of Manji has shown that both lithium and valproate inhibit PKC activity via indirect, intracellular actions. Thus, an agent that inhibits PKC directly, especially at the site where DAG activates PKC, may be useful in treating these disorders.
Extracellular blockade of this pathway is also helpful in treating stress disorders. Raskind and Taylor have shown that the a1 antagonist, prazosin, (PRAZ) ameliorates the symptoms of PTSD. Prazosin is effective in both recent and long-established illness- e.g. it is being given to troops returning from Iraq, and to survivors of the Holocost.
Direct link to:
