The Orexins/Hypocretins and Schizophrenia

Abstract

Advances in molecular biology have led to new peptides and proteins being discovered on a regular basis, including the isolation of a number of neurotransmitter candidates. Rarely, however, do these immediately capture the attention of the scientific community. The isolation and characterization of the orexin/hypocretin peptides a decade ago resulted in a slew of studies that have helped clarified their diverse functions, including prominent roles in arousal and appetitive behavior. A number of recent studies have detailed the role of the orexins/hypocretins in attention and cognition and uncovered an involvement in schizophrenia and the mechanisms of action of antipsychotic drugs (APDs). This issue of Schizophrenia Bulletin presents several articles that review our current understanding and point to future directions for the study of the orexins/hypocretins in schizophrenia and APD actions.

Introduction

New signaling molecules are now discovered on an almost daily basis, and the identification of molecules advanced as neurotransmitters occurs at an only slightly less frenetic rate. But few of these transmitter candidates have provoked as much excitement and interest as the discovery of the orexins/hypocretins almost a decade ago. de Lecea et al,¹ using subtraction hybridization to identify genes enriched in the hypothalamus, uncovered a gene that was expressed only in the lateral hypothalamus and adjacent perifornical area (LH/PFA). This gene encoded a peptide that was named hypocretin, “hypo” referring to its hypothalamic localization and “cretin” to its structural similarity to the incretin family of peptides. At the same time, Yanagisawa and associates, using a cell-based reporter system to uncover orphan G protein–coupled receptors and their endogenous ligands, identified a peptide that upon intracranial administration elicited feeding in sated mice²; they named the peptide orexin. We will use the term orexin to refer to the peptide, simply because it is less cumbersome than orexin/hypocretin.

While the initial description of orexin's roles of arousal and appetitive behavior in animals sparked considerable interest, the discovery that the loss of orexin neurons is the proximate cause of narcolepsy^3–5 dramatically upped interest in the peptide, with a resultant explosion in information about the orexins: more than 1200 published research articles on the orexins since the first publication in 1998. These studies have revealed a remarkably wide array of functional roles for the orexins, including those in feeding, energy homeostasis, sleep, arousal, reward, substance abuse, stress, and sympathetic and cardiovascular functions (see Winsky-Sommerer et al,⁶ Berridge and Espana,⁷ Jones,⁸ Spinazzi et al,⁹ Zeitzer et al,¹⁰ Carr and Kalivas,¹¹ Saper,¹² Harris and Aston-Jones,¹³ Siegel and Boehmer,¹⁴ Bingham et al,¹⁵ Scammell and Saper,¹⁶ and Sakurai¹⁷).

This issue of Schizophrenia Bulletin presents several articles that focus on still another role of the orexins: their involvement in the cognitive dysfunction of schizophrenia and the mechanisms of action of antipsychotic drugs (APDs). We will first briefly review the pharmacology and anatomy of orexin systems in brain and then explore the development of ideas that have led to studies of orexin in schizophrenia.

Orexin Structure and Receptors

The human preproorexin gene is located on chromosome 17q21 and is composed of 616 nucleotides contained in 2 exons and 1 intron. The first exon contains the initial part of the secretory signal sequence, with the somewhat smaller second exon containing the rest of the signal sequence as well as the coding region of orexin.¹⁸ A 3.2-kb promoter region directs expression specifically to the lateral hypothalamus.

The orexin prohormone is proteolytically cleaved into 2 peptides, orexin A and orexin B, which are coexpressed in neurons of the LH/PFA. Orexin A is comprised of 33 amino acids and is highly conserved across mammalian species.² The 28-residue orexin B peptide is identical in the rat and the mouse, with human orexin B displaying 2 amino acid substitutions.^2,18

Two orexin receptors, designated OX₁R and OX₂R, have been identified.² Both are G protein–coupled receptors and have significant homology at the transcript level. OX₁R signals via activation of phospholipase C (PLC) and increases intracellular Ca²⁺. In contrast, OX₂R couples to G_q and G_i/o, and the activation of the OX₂R results in increased PLC and decreased adenylyl cyclase activity.¹⁹

Both orexins A and B have affinities of approximately 30 nM for the OX₂R. The 2 orexin receptors differ critically in that orexin A also binds the OX₁R at approximately 30 nM, whereas orexin B has a very low affinity (approximately 420 nM) at the OX₁R.² Several specific orexin receptor antagonists have been developed and have figured prominently in studies detailing the regional effects of orexins.

Anatomical Organization of Central Orexin Systems

Neurons expressing preproorexin mRNA are restricted to the LH/PFA. Immunohistochemical studies indicate that orexins are expressed in only about 70 000 cells in humans,²⁰ with about 3000 orexin neurons in the rat.²¹ Despite the small number of orexin cells, orexin axons are found in almost all areas of brain, a notable exception being the cerebellar cortex. The widespread distribution of orexin axons can be attributed to the fact that single orexin neurons often collateralize to innervate multiple targets.²² Although orexin axons are found in most of the brain, the relative density of orexin fibers varies considerably, with particularly high densities seen in the locus coeruleus, paraventricular nucleus of the thalamus, septum and diagonal band complex, and the infralimbic and prelimbic aspects of the prefrontal cortex.²³

As would be expected on the basis of the widespread distribution of orexin axons, there is a correspondingly wide distribution of orexin receptors in the brain. The relative densities of the 2 orexin receptors vary across brain areas, with some neurons expressing only one of the 2 receptors. OX₁R mRNA is enriched in neurons of the hypothalamus, thalamus, locus coeruleus, cerebral cortex, hippocampal formation, and basal ganglia,²⁴ with the distribution of OX₁R-binding sites and immunoreactivity paralleling the distribution of OX₁R mRNA. OX₂R mRNA is enriched in the pons, medulla oblongata, hypothalamus, and thalamus.²⁵ Within regions, further segregation is seen. For example, there is a moderately high abundance of OX₁R in layers II–V of the cortex, where OX₂R mRNA is more abundant in layer VI. Both receptors appear to be expressed at roughly comparable levels in the hippocampus, thalamus, and hypothalamus.^25,26

Several of the functional attributes of the orexins, including roles in arousal, reward, and substance abuse, suggest that there may be interactions between orexin and dopamine. Because orexin neurons send projections widely across the brain, it is not surprising that orexin and dopamine intersect in certain brain sites. Both orexin inputs to the midbrain dopamine cell group regions and orexin projections to forebrain sites that receive dopamine inputs have been described.²⁷ Orexin neurons of hypothalamus provide what is arguably the most dense input from any single source to the ventral tegmental area (VTA), site of the A10 dopamine neurons that innervate forebrain sites, including the prefrontal cortex.²⁷ In contrast, there is a sparse orexin innervation of the substantia nigra (SN). Although close appositions between orexin axons and VTA dopamine neurons are seen,^27–29 orexin axons form synapses infrequently with either dopamine- or γ-aminobutyric acid–containing neurons in the VTA,³⁰ suggesting that orexin may modulate VTA dopamine neurons through both volume and synaptic transmission.³¹

In the forebrain, a rich orexin innervation of the PFC that is in register with the cortical dopamine innervation is seen, but almost no orexin axons are present in the dorsolateral striatum, which receives dopamine inputs from the SN.²⁷ The thalamic paraventricular nucleus also receives moderately dense orexin and dopamine inputs. There is some overlap of orexin and dopamine in the lateral septum and certain amygdala nuclei. However, in the nucleus accumbens, which is richly invested with dopamine axons, orexin fibers are largely confined to the septal pole region. Thus, while orexin and dopamine axons can be found intermingled throughout the forebrain, the major sites of overlap of orexin and dopamine appear to be the PFC and thalamic paraventricular nucleus (PVT). Interestingly, the PVT projects heavily on to the PFC,^32,33 and those PVT neurons that innervate the PFC are activated by orexin.³⁴

Leibowitz and Brown³⁵ suggested the presence of a dopamine innervation of the LH/PFA, but only recently has the introduction of contemporary neuroanatomical methods allowed the characterization of a relatively dense dopaminergic input to the vicinity of orexin cells in the hypothalamus.^36–38 Despite this dopamine innervation, very few LH/PFA cells express any of the 5 dopamine receptors ³⁹, suggesting that dopamine inputs to the region may modulate the activity of orexin cells either through interactions with a noncognate receptor or via release of a cotransmitter from dopamine axons.

We have discussed the anatomical organization of orexin neurons with reference to mesotelencephalic dopamine neurons. This is predicated on the long history linking schizophrenia to changes in dopamine function and the fact that APDs target dopamine D2 receptors. There are, however, similarly intricate interactions between orexin and other monoamines, including norepinephrine, serotonin, histamine, and acetylcholine. The reader is referred to several excellent reviews for discussions on the anatomical interactions between orexins and classical transmitter systems.^23,40–42

Narcolepsy Points the Path to Schizophrenia

Narcolepsy, the proximate cause of which is loss of orexin cells, is marked by daytime somnolence, marked disturbances in sleep architecture, and cataplexy (see Scammell⁴³ and Siegel⁴⁴). Sleep disturbances are also well known in schizophrenia.⁴⁵ The observation that the excessive daytime drowsiness of narcolepsy can be effectively treated with modafinil, a drug that activates orexin neurons,^46,47 prompted several clinicians to determine if modafinil could counter the sedation associated with APD treatment. The results were gratifying, even with drugs as sedating as clozapine.^48–50

At the same time, case reports and small studies suggested that narcolepsy can on occasion be misdiagnosed as schizophrenia,^51–56 probably because of the intrusion of the daytime hypnogogic hallucinations that are sometimes seen in narcolepsy. However, testing for the HLA-DR15 autoimmune marker of narcolepsy and examination of cerebrospinal fluid (CSF) orexin levels in schizophrenic patients with significant sleep disorders did not reveal evidence of narcolepsy,⁵⁷ suggesting that in some cases both narcolepsy and schizophrenia are present but that such rare co-occurrences are sporadic.⁵⁵

Modafinil and Schizophrenia

Narcolepsy is not just a disorder of sleep: other disturbances are cognitive changes.^58–62 Teitelman,⁴⁸ in a case report on the use of modafinil to overcome APD-induced sedation, noted that modafinil might also improve attention and reduce cognitive dysfunction. Studies in normal adult subjects soon followed, with one reporting that modafinil does not affect cognition but does affect mood,⁶³ while another more extensive study concluded that modafinil enhances cognitive function, including spatial memory, delayed match-to-sample, and stop-signal reaction time, in normal adult subjects.⁶⁴ The clinical trials of modafinil in schizophrenia emerged at about the same time. Rosenthal and Bryant⁶⁵ reported that modafinil was an effective treatment for schizophrenia. Several double-blind placebo-controlled studies in schizophrenia followed, with most (but not all) reporting that modafinil as an adjunctive medication results in significant improvement, particularly in cognitive function.^66–69

Most of these articles suggested that modafinil appears to benefit a restricted subset of patients. Similarly, the imaging studies of modafinil in schizophrenia found that modafinil caused an overall increased cognitive function with increases in frontal cortical activation, but these effects were seen only in certain patients.^66,70 Pierre et al⁶⁹ examined schizophrenic or schizoaffective subjects with prominent negative symptoms and found that modafinil augmentation did not significantly improve negative symptoms but did result in global improvement. It will clearly be important to define the specific responses across various symptom domains to modafinil and to characterize those patients who best respond. In this issue of Schizophrenia Bulletin, Morein-Zamir et al review studies on the effects of modafinil on cognition, including attention, in schizophrenia.

Orexin and APDs: Therapeutic and Adverse Effects

The interplay between dopamine and orexin systems suggests that dopaminergic drugs, including dopamine receptor antagonists, may modify orexin function. Bubser et al³⁹ found that both D1- and D2-like dopamine receptor agonists activate orexin cells of the LH/PFA, as reflected by induction of the immediate-early gene protein product Fos but that full activation requires concomitant D1/2 receptor activation. Consistent with this observation was the finding that the ability of the mixed D1/2 receptor agonist apomorphine to activate orexin cells was blocked by combined administration of D1- and D2-like antagonists but not D1- or D2-like antagonists alone.³⁹

Fadel et al⁷¹ examined the effects of a number of APDs on activation of orexin cells and noted that certain but not all APDs induced Fos in the orexin cells. Importantly, they found that those drugs that activated orexin neurons were all associated with significant weight gain liability, while those that did not cause weight gain did not activate orexin neurons. Thus, the degree to which APDs activated orexin cells was directly correlated with the amount of weight gained by subjects treated with these APDs. Because chronic treatment with the indirect dopamine agonist amphetamine also activates orexin neurons but causes weight loss, Fadel et al⁷¹ compared the effects of amphetamine and the atypical APD clozapine on activation of those orexin cells that innervate the PFC. Orexin neurons that project to the PFC were strongly activated by clozapine but were unaffected by amphetamine, consistent with the presence of 2 separable populations of orexin cells.

Because the weight gain liability of APDs is positively correlated with therapeutic response,^72–74 weight gain and therapeutic response to certain APDs may be inextricably linked. The animal studies examining activation of orexin cells suggested that PFC-mediated cognitive function is likely to respond to orexins. This suggestion was bolstered by 2 observations. First, both intraventricular and intra-VTA orexin infusions increase extracellular dopamine levels in the PFC but not in the nucleus accumbens, which contains few orexin axons.⁷⁵ Because executive function and in particular working memory appear to be dependent on frontal cortical dopamine, the ability of orexin to increase extracellular dopamine tone in the PFC is consistent with promoting cognitive function. Moreover, preliminary data emerged that suggested that clozapine's actions on the PFC are blocked by orexin antagonists,³⁶ and raised the specter of orexin agonists in the treatment of the cognitive dysfunction in schizophrenia. In this issue of Schizophrenia Bulletin, Rasmussen et al describe their findings that orexin antagonists blocked the effects of haloperidol and olanzapine on midbrain dopamine neurons, and also blocked haloperidol-induced catalepsy, consistent with the hypothesis that orexin promotes the actions of APDs but are not likely to cause extrapyramidal side effects.

While rodent studies indicate that orexin neurons are activated by clozapine and certain other APDs, there is a paucity of clinical data addressing this issue. One clinical study reported that CSF levels of orexin A are significantly lower in schizophrenic patients treated with haloperidol than unmedicated subjects but that orexin A levels did not differ between haloperidol-treated patients and another group of patients treated with either clozapine or olanzapine.⁷⁶ The authors suggested that the clozapine-induced activation of orexin neurons seen in the rat is not manifested in human schizophrenic subjects. However, the source of orexin measured in CSF samples is most likely the rich orexin innervation of the spinal cord, with a relatively small proportion from brain sites, including the PFC. Because orexin neurons collateralize to innervate various areas in a target-specific fashion,^22,71 with one population of orexin axons innervating the PFC and a separate group the spinal cord, it is likely that CSF orexin concentrations are not indicative of the activity of orexin neurons that project to the PFC.

It is somewhat surprising that modafinil reduces a side effect of clozapine (sedation) while at the same time appearing to contribute a therapeutic benefit in the treatment of schizophrenia. This may reflect the fact that the receptor-binding profile of clozapine is very broad, and modafinil may modify clozapine's actions at one receptor but not another. Alternatively, there may be distinct orexin neurons that lead to improvement of cognition and attention in schizophrenia and the amelioration of sedation induced by clozapine and other APDs. The most likely candidate for the system through orexin may improve attention and certain other cognitive changes in schizophrenia is a thalamocortical projection involving the medial thalamus and PFC. Lambe et al, in this issue of Schizophrenia Bulletin, describe an elegant series of studies describing how thalamostriatal activation via orexin results in dendritic remodeling in the prefrontal cortex. Because dystrophic changes in the dendrites of PFC pyramidal cells are one of the pathological features of schizophrenia,^77,78 these data lead one to suspect that orexin may reverse the dendritic spine loss seen in schizophrenia.

Questions and Future Directions

The possible links between orexin and schizophrenia have only very recently been explored, and there are more questions at this point than answers concerning the relation of the orexins and psychiatric disorders. There are, however, several key issues that need to be resolved if potential therapeutic benefits of modulating central orexin system are to be realized.

Preclinical data suggest that drugs activating orexin neurons may be useful in treating the cognitive, including attention, deficits in schizophrenia. Clinical studies using modafinil are consistent with this possibility. However, despite extensive investigation, the mechanism of action of modafinil is unknown. Modafinil does not bind with any significant affinity to either of the 2 orexin receptors, suggesting that any effect on orexin neurons is indirect. Similarly, modafinil does not bind to dopamine receptors, although it has a low affinity for the dopamine transporter.⁷⁹ While we have little insight at this point into the receptors through which modafinil operates, it is clear that a wide array of brain sites are activated by the drug, ranging from the frontal cortical areas observed in imaging studies to histamine neurons of the caudal hypothalamus. Careful dose-response studies in rats have reported that neurons in the extended amygdala and the preoptic region, both of which provide inputs to the orexin cell areas of the LH/PFA, are activated by low doses of modafinil.⁴⁷ The pattern of neuronal activation is much more extensive after treatment with higher doses of modafinil.^46,47 The involvement of the amygdala and preoptic region in schizophrenia is poorly understood, although the extended amygdala's role in substance abuse (and possible comorbid schizophrenia and substance abuse) is well known.⁶ Development of new and more targeted therapeutic approaches based on the promising initial results with modafinil will require identification of the receptor mechanisms through which modafinil exerts its effects and will be bolstered significantly by the development of brain-penetrant orexin agonists.

Clinical trials with modafinil have mainly reported global improvement but have often also commented that the effects appear to be best realized in undefined subsets of patients. This is no different from any other drug used in the treatment of schizophrenia and is a clear call to better define the precise symptom domains that best respond to modafinil. The animal studies point to thalamocortical systems as a major target, and as such attention and working memory would be expected to be most responsive (see the contributions Lambe et al and Morein-Zamir et al in this issue of Schizophrenia Bulletin). Because of the relatively poor response of these deficits to most APDs,⁸⁰ further exploration of this area is clearly needed.

Recent data have suggested that orexin plays a critical role in the rewarding properties of drugs of abuse and the augmentation of the actions of such drugs seen upon repeated intermittent administration.^11,16

The data with modafinil are impressive in that reported side effects have been quite benign. However, adverse effects are associated with all therapeutic drugs, and because modafinil is used as an adjunct to treatment with APDs, the risk for emergence of adverse interactions is significant while benefit remains unclear (see Glick et al⁸¹). The use of modafinil in normal control subjects is consistent with a significant increase in attention and other cognitive functions, but these effects are not dose dependent. While animal data strongly suggest that the orexin cells, histamine neurons, and 2 key afferent structures are strongly activated at low doses, higher doses cause widespread activation, and it is reasonable to assume that the risk of side effects increases in parallel. Interestingly, however, Rasmussen et al in this issue note that orexin antagonists block catalepsy, an animal model of extrapyramidal side effects.

It is easy to be impatient with the progress of scientific advances, particularly when introducing new therapeutic approaches to clinical practice. It is striking that the orexins were discovered only a decade ago and that so much information, both preclinical and clinical, has accrued since then. At this pace, we should not have to wait very long until some of the key questions concerning orexins and schizophrenia are answered.