An emerging role for Cullin-3 mediated ubiquitination in sleep and circadian rhythm

Abstract

Although the neurophysiological correlates of sleep have been thoroughly described, genetic mechanisms that control sleep architecture, long surmised from ethological studies, family histories and clinical observations, have only been investigated during the past decade. Key contributions to the molecular understanding of sleep have come from studies in Drosophila, benefitting from a strong history of circadian rhythm research. For instance, a number of recent papers have highlighted the role of the E3 ubiquitin ligase Cullin-3 in the regulation of circadian rhythm and sleep. We propose that different Cullin-3 substrate adaptors may affect specific molecular pathways and diverse aspects of circadian rhythm and sleep. We have previously shown that mutations in BTBD9, a risk factor for Restless Legs Syndrome (RLS) encoding a Cullin-3 substrate adaptor, lead to reduced dopamine, increased locomotion and sleep fragmentation. Here, we propose that Cullin-3 acts together with BTBD9 to limit the accumulation of iron regulatory proteins in conditions of iron deficiency. Our model is consistent with clinical observations implicating iron homeostasis in the pathophysiology of RLS and predicts that lack of BTBD9 leads to misregulation of cellular iron storage, inactivating the critical biosynthetic enzyme Tyrosine Hydroxylase in dopaminergic neurons, with consequent phenotypic effects on sleep.

It has been just over a decade since an empirical definition of sleep was formalized in Drosophila, and already research in flies has made fundamental contributions to our understanding of the molecular machinery underlying sleep.^1-5 Where the small size of the model organism limits neurophysiological approaches that have been used widely in larger vertebrate models, the rapid generation time, semi-automated population studies and an impressive, and ever growing, genetic tool box have provided powerful new ways to investigate sleep in this system. Following two papers correlating Drosophila activity patterns with sleep,^6,7 the number of laboratories that now actively use this model to study sleep has grown rapidly and has contributed to date over 250 PubMed entries. Several of these studies report the results of mutational screens that have been performed using the Drosophila Activity Monitor (DAM), a system that was originally devised to monitor the circadian rhythmicity in adult flies.^6,7 DAM has been used to record sleep data in a comparable manner to the more intensive ultra-sound and videographic measurements used in higher animals. Relatively minor changes in the recording conditions and analytical parameters have allowed researchers to extract sleep data and to hunt for genes controlling various aspects of sleep such as total sleep time, the number and length of sleep bouts and sleep homeostasis (the naturally conserved drive in animals to make up for “lost sleep”). While gene discovery is the traditional strength of Drosophila (examples discussed below), a number of more recent studies have also begun exploring the neural circuitry that regulates fly sleep owing to remarkable enhancements in the resolution of tissue-specific transgene expression by the GAL4-UAS system. These studies have highlighted the existence of specific dopaminergic circuitry that control various aspects of sleep and arousal, most likely distinct from other such neurons that regulate a myriad of behaviors such as locomotor drive, learning, reward and aggression.^8,9 As sleep studies improve in power and sophistication (e.g., with the advent of continuous video monitoring of sleep and new methods for statistical analysis of sleep distribution^1,10-12) the impact of Drosophila sleep research will continue to contribute in two key areas: the molecular mechanisms that determine the biological requirement of sleep and the molecular pathophysiology of human sleep disorders.

Four recent reports within the last year have suggested that the E3 Ubiquitin ligase Cullin-3 (a member of the Cullin-RING family of Ubiquitin ligases) plays an important role in the regulation of circadian rhythm and sleep.^1,13-15 Cullin-3 is known to bind its specific substrates via a number of adaptor proteins containing a characteristic BTB domain.¹⁶ These BTB domain adaptors contain a protein interaction motif first identified in three Drosophila proteins, Bric-a-brac, Tramtrack, and Broad Complex.¹⁷ Cullin-3 is known to ubiquitinate a diverse array of targets, based on cell-specific interactions with key BTB domain adaptors. For instance, Cullin-3 binds Kelch—a founding member of the KEL repeat and BTB domain containing substrate adaptors—to interact with and organize filamentous actin in the female germ line ring canal in Drosophila.¹⁸

As is often the case in science, studies have converged on sleep-related functions for Cullin-3 in Drosophila from different angles. Nicholas Stavropoulos and Michael Young identified mutations in the gene insomniac (inc) encoding a BTB domain adaptor for Cullin-3 from a classical forward genetic screen for sleep mutants.¹³ inc mutants display severely reduced sleep, as well as loss of sleep consolidation without any effect on circadian rhythm. Further experiments showed that INC binds physically to Cullin-3 and that pan-neuronal knockdown of Cullin-3 phenocopies sleep disruptions observed in inc mutants. These experiments established the sleep-related functions for INC-Cullin-3 and were independently confirmed by Cory Pfeiffenberger and Ravi Allada, who also uncovered strong sleep homeostasis defects in inc mutants and showed that decreased sleep in inc mutants can be rescued by inhibiting the biosynthesis of dopamine.¹⁴ Cullin-3 dependent molecular pathways have also been shown to regulate circadian rhythm in flies. In clock neurons, Cullin-3 activity controls circadian oscillations in PER and TIM and affects the circadian clock.¹⁵ More specifically, Cullin-3 seems to operate in contrast to a Cullin-1 complex member, supernumerary limbs (SLMB), by binding to hypo-phosphorylated TIM and affecting its accumulation in the absence of PER. Such candidate-directed studies were prompted by the observation that in addition to SLMB, Cullin-3 regulates the proteolytic processing of Cubitus Interruptus (CI), another component of the circadian clock.^19-21

Finally, our own work has implicated Cullin-3 in the regulation of sleep consolidation. In this case, reverse genetic analysis of the gene BTBD9, a risk factor for Restless Legs Syndrome (RLS) or Willis-Ekbom Disease (WED), showed that loss of the Drosophila homolog CG1826 leads to increased sleep fragmentation (without altering total sleep time or circadian rhythm) and reduced dopamine in the fly brain.¹ Although BTBD9 co-localizes with Cullin-3 in fly neurons, a specific genetic interaction between Cullin-3 and BTBD9 in the regulation of sleep remains to be shown. However, homology modeling of BTBD9 supported the notion that this protein is an adaptor for Cullin-3 since it possesses a characteristic Cullin-3 binding “3-box”²². Taken together, these studies provide considerable new evidence for the involvement of Cullin-3 dependent mechanisms in the regulation of the circadian clock and sleep in flies (Fig. 1A).

Figure 1. The role of Cullin-3 dependent mechanisms in circadian rhythm and sleep. (A) Summary of Cullin-3 functions in the regulation of circadian rhythm and sleep in Drosophila. The BTB adaptors Insomniac and BTBD9 regulate sleep architecture, but their targets are unknown. Conversely, Cullin-3 regulates TIM stability, but the BTB adaptor for this interaction has not been identified. (B) Model depicting predicted functions for BTBD9 in the regulation of iron homeostasis. BTBD9 is suggested to function like a clamp to limit IRP2 accumulation under low iron conditions. In this manner a Cullin-3 dependent pathway regulates ferritin levels inside a cell in conjunction with known Cullin-1 mediated mechanisms that control degradation of IRP2. (C) Measurement of metal ions using atomic absorption spectrometry²³ in whole flies in two BTBD9 null mutant alleles (named wanderlust or wlst) as compared with control flies of similar genetic background.¹ A trend for increased iron and decreased manganese accumulation in mutants as compared with controls did not reach statistical significance in 4–6 replicate measurements. Both copper and zinc remained remarkably unchanged in the same samples. Alterations in iron levels are relevant to RLS since many patients display low serum ferritin levels suggesting aberrant iron homeostasis²⁴ and an increase in free serum iron was found in the BTBD9 knockout mice.²⁵

Except for TIM, that has been shown to co-immunoprecipitate with Cullin-3,¹⁵ specific targets for Cullin-3/BTB adaptors involved in sleep and circadian rhythm regulation are currently unknown (Fig. 1A). Identification of these substrates is a pre-requisite for understanding cell-specific functions for Cullin-3 in different neuronal classes in the context of sleep regulation. Three parallel strategies can be employed to detect these proteins. First, biochemical approaches such as co-immunoprecipitation followed by mass-spectrometry might help in discovering protein binding partners for BTB domain Cullin-3 adaptors such as BTBD9 or INC. Second, unbiased phenotypic screens aimed at isolating genetic modifiers of these adaptor mutations or more broadly, Cullin-3, are likely to reveal substrates. Finally, knowledge of particular molecular pathways that are influenced by a specific Cullin-3/BTB adaptor combination might also point toward likely candidates. For example, connections between brain iron metabolism and the dopaminergic neural network in RLS pathology^26-30 and between Drosophila iron homeostasis and circadian rhythm^31,32 led us to investigate if human BTBD9 could affect iron metabolism in HEK cell lines. We have shown that BTBD9 influences iron homeostasis by regulating ferritin, potentially through the iron-regulatory protein IRP2.^1,33,34 One possibility is that BTBD9 directly binds to IRP2 and regulates its stability through Cullin-3 dependent ubiquitination or, alternatively, BTBD9 might regulate IRP2 indirectly via other effectors. In either case, the observation that BTBD9 impacts on IRP2 turnover (and thereby on ferritin accumulation) is particularly relevant in the context of RLS/WED since low serum ferritin has been clinically associated with RLS/WED.²⁴

Incidentally, IRP2 stability in response to extracellular iron concentration has been shown to be dependent on the Cullin-1 complex through the F-box substrate-specific adaptor protein FBXL5.^35,36 Thus, iron binds to a hemerythrin-like domain in FBXL5 and stabilizes this protein, which in complex with Cullin-1 and Skp1 degrade IRP2, permitting ferritin translation (Fig. 1B). Given this pathway for iron responsive regulation of IRP2, we wondered what purpose, if any, could be served by the independent mechanism to degrade IRP2 we have identified. We think that Cullin-3 and BTBD9 may serve to limit the extent to which IRP2 levels increase inside a cell. Consistent with this idea, our experiments in HEK cells have shown that forced overexpression of BTBD9 can limit the increase in IRP2 under iron deprivation.¹ As a result, the Cullin-3/BTBD9 complex might prevent over-accumulation of IRP2 and consequently a precipitous decrease in ferritin. This “upper limit clamp” might also be necessary to prevent IRP2 levels from increasing in a runaway fashion, such that when iron supply is transiently elevated, the iron-responsive Cullin-1/Skp1/FBXL5 system can cope with the relatively limited amount of IRP2 in the cell and degrade it quickly to trigger ferritin translation (Fig. 1B).

There are some “design” similarities between the mechanism we propose and the Cullin-1/Cullin-3 dependent regulation of PER and TIM in the circadian machinery¹⁵ and CI in the Hedgehog signaling pathway,^19,20 suggesting that perhaps there are other instances where Cullin-1 and Cullin-3 cooperate to regulate the stability of the same protein or signaling pathway components. It is intriguing that RNA interference of ferritin in a small subset of clock neurons disrupts circadian activity to the same extent as RNA interference of the Cullin-1 adaptor SLMB (Ph.D. Thesis of K. Mandilaras, 2012) and these two proteins have been shown to interact with each other.³⁷ It is, therefore, plausible that the two signaling pathways may act in different cell types, but through similar molecules to affect different behaviors. In this context, it remains to be seen whether BTBD9 activity or expression might be responsive to changes in iron supply and what the targets of Drosophila BTBD9 are, since flies use IRP-1A to regulate ferritin translation.^38,39

Low serum ferritin in RLS/WED patients suggests low iron availability,²⁴ an idea that is supported by studies that have found global reduction in iron levels in the brain and cerebrospinal fluid (CSF).^28,40-44 Additionally, the at-risk SNP in human BTBD9 also correlates with low serum ferritin.⁴⁵ If the presence of this allele leads to reduced BTBD9 expression (a prediction that needs confirmation in patients), then it would be consistent with our observations in HEK cells where increased expression of BTBD9 elevates ferritin levels.¹ From this, it might also be surmised that loss of BTBD9 in animal models could lead to decreased iron, especially in the nervous system, although a strict correlation between ferritin and iron can potentially be disrupted through multiple mechanisms, especially in RLS/WED. Paradoxically, a recent report demonstrates that in mice, a global knockout of BTBD9 leads to elevated serum iron but no change in iron levels in the striatum.²⁵ Similarly, our experiments to measure whole body iron in the fly model also suggest a trend of increased iron levels in BTBD9 mutants (Fig. 1C) along with reduced manganese (while copper and zinc remain unchanged). It is important to note that our measurements of iron in the fly are not tissue-specific, so we do not have independent estimates of iron in the nervous system. However, similar observations of a slight systemic iron overload in mice and flies with a BTBD9 mutant background suggest that homeostatic and feedback mechanisms that ultimately control iron supply, availability and uptake in specific tissues require further investigation in these models. Since flies lack a genuine IRP2, but express ferritin heavy and light chains in an iron-responsive manner,^39,46,47 it is also conceivable that BTBD9 influences IRP-1A in a similar fashion or that its molecular substrate is different in flies and mammals. Whatever the case, it seems reasonable to conclude that BTBD9 affects iron storage in dopaminergic cells by inducing the primary iron storage protein ferritin. A resulting decrease in bio-available iron inside cells is expected to adversely affect the functioning of several iron requiring enzymes, among them the rate limiting enzyme for dopamine biosynthesis, Tyrosine Hydroxylase (TH). A decrement in TH function could explain the observed reduction in dopamine in BTBD9 mutant flies. Elucidation of the precise molecular pathway by which BTBD9 controls iron homeostasis will require a much better understanding of the targets of BTBD9. Similar considerations apply to INC and the as yet unknown adaptor for Cullin-3 in circadian rhythm regulation. Whether these Cullin-3 dependent mechanisms are conserved in mammals remains to be seen.

In summary, recent studies have focused attention on the parallel, non-redundant functions of E3 ubiquitin ligases Cullin-3 and Cullin-1 in the regulation of sleep and circadian rhythms. Future studies will be aimed at identifying additional specific substrate adaptors, such as INC and BTBD9 for Cullin-3, as well as cell-specific substrates for these adaptors, such as iron metabolism proteins. Genetic and phenotypic analysis of these additional components will reveal both conserved principles and the diversity of Cullin-3 dependent mechanisms in the regulation of sleep and circadian rhythms.

Glossary

Abbreviations:

RLS

restless legs syndrome

WED

Willis-Ekbom Disease

DAM

Drosophila activity monitor

CSF

cerebrospinal fluid

INC

insomniac

PER

period

TIM

timeless

SLMB

supernumerary limbs

CI

cubitus interruptus

IRP2

iron regulatory protein 2

HEK

human embryonic kidney

TH

tyrosine hydroxylase

Freeman A, Pranski E, Miller RD, Radmard S, Bernhard D, Jinnah HA, et al. Sleep fragmentation and motor restlessness in a Drosophila model of Restless Legs Syndrome. Curr Biol. 2012;22:1142–8. doi: 10.1016/j.cub.2012.04.027.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.