Conserved properties of Drosophila Insomniac link sleep regulation and synaptic function

Sleep is an ancient animal behavior that is regulated similarly in species ranging from flies to humans. Various genes that regulate sleep have been identified in invertebrates, but whether the functions of these genes are conserved in mammals remains poorly explored. Drosophila insomniac (inc) mutants exhibit severely shortened and fragmented sleep. Inc protein physically associates with the Cullin-3 (Cul3) ubiquitin ligase, and neuronal depletion of Inc or Cul3 strongly curtails sleep, suggesting that Inc is a Cul3 adaptor that directs the ubiquitination of neuronal substrates that impact sleep. Three proteins similar to Inc exist in vertebrates—KCTD2, KCTD5, and KCTD17—but are uncharacterized within the nervous system and their functional conservation with Inc has not been addressed. Here we show that Inc and its mouse orthologs exhibit striking biochemical and functional interchangeability within Cul3 complexes. Remarkably, KCTD2 and KCTD5 restore sleep to inc mutants, indicating that they can substitute for Inc in vivo and engage its neuronal targets relevant to sleep. Inc and its orthologs localize similarly within fly and mammalian neurons and can traffic to synapses, suggesting that their substrates may include synaptic proteins. Consistent with such a mechanism, inc mutants exhibit defects in synaptic structure and physiology, indicating that Inc is essential for both sleep and synaptic function. Our findings reveal that molecular functions of Inc are conserved through ~600 million years of evolution and support the hypothesis that Inc and its orthologs participate in an evolutionarily conserved ubiquitination pathway that links synaptic function and sleep regulation.


78
Mutations of the Drosophila insomniac (inc) gene [11] severely curtail the duration and 79 consolidation of sleep but do not alter its circadian regulation [11,12]. inc encodes a protein of the 80 Bric-à-brac, Tramtrack, and Broad / Pox virus zinc finger (BTB/POZ) superfamily [13], which 81 includes adaptors for the Cullin-3 (Cul3) E3 ubiquitin ligase complex [14][15][16][17]. Cul3 adaptors have a 82 modular structure, in which the BTB domain binds Cul3 and a second distal domain recruits substrates 83 to the Cul3 complex for ubiquitination [14][15][16][17]. The BTB domain also mediates adaptor self-84 association, enabling the oligomerization of Cul3 complexes and the efficient recruitment and 85 ubiquitination of substrates [18,19]. Biochemical and genetic evidence supports the hypothesis that Inc 86 is a Cul3 adaptor [11,12]. Inc and Cul3 physically interact in cultured cells [11,12] and Inc is able to 87 self-associate [12]. In vivo, neuronal RNAi against inc or Cul3 strongly reduces sleep [11,12], and 88 reduction in the levels of Nedd8, a protein whose conjugation to Cullins is essential for their activity, 89 also decreases sleep [11]. While Inc is thus likely to function as a Cul3 adaptor within neurons to 90 promote sleep, the neuronal mechanisms through which Inc influences sleep are unknown.

91
Three proteins similar to Inc-KCTD2, KCTD5, and KCTD17-are present in vertebrates 92 [11,20,21], but their functions in the nervous system are uncharacterized and their functional 93 5 conservation with Inc has not been addressed. KCTD5 can self-associate and bind Cul3 [20], 94 suggesting that it may serve as a Cul3 adaptor, yet no substrates have been identified among its 95 interacting partners [22,23]. One KCTD17 isoform has been shown to function as a Cul3 adaptor for 96 trichoplein, a regulator of primary cilia [24]. However, trichoplein-binding sequences are not present 97 in Inc, KCTD2, KCTD5, or other KCTD17 isoforms, and trichoplein is not conserved in Drosophila.

98
Thus, it remains unclear whether Inc and its vertebrate homologs have conserved molecular functions, 99 particularly within neurons and cellular pathways relevant to sleep.

100
Here, we assess the functional conservation of Inc and its mammalian orthologs and elucidate 101 a neuronal mechanism through which they may impact sleep. Inc and each of its orthologs bind Cul3 102 and self-associate, supporting a universal role for these proteins as Cul3 adaptors. Inc and its orthologs 103 furthermore exhibit biochemical interchangeability within Inc-Inc and Inc-Cul3 complexes, indicating 104 that the oligomeric architecture of Inc-Cul3 complexes is highly conserved. Strikingly, KCTD2 and 105 KCTD5 can functionally substitute for Inc in vivo and restore sleep to inc mutants, indicating that 106 these Inc orthologs readily engage the molecular targets through which Inc impacts sleep. Our studies 107 furthermore reveal that Inc and its orthologs localize similarly within fly and mammalian neurons and 108 traffic to synapses. Finally, we show that inc mutants exhibit defects in synaptic structure and 109 physiology, indicating that inc is essential for both sleep and synaptic function. Our findings 110 demonstrate that molecular functions of Inc are conserved from flies to mammals, and support the 111 hypothesis that Inc and its orthologs direct the ubiquitination of conserved neuronal proteins that link 112 sleep regulation and synaptic function.

114
The mammalian orthologs of Insomniac are expressed in the nervous system 115 Inc functions in neurons to impact sleep [11,12]. To assess whether Inc orthologs are 116 expressed in the mammalian nervous system, we performed RT-PCR on mouse brain RNA using 117 primers specific for KCTD2, KCTD5, and KCTD17. All three genes are expressed in the brain (Fig   118   1A), and in situ hybridizations reveal expression within cortex, thalamus, striatum, pons, and 119 cerebellum among other brain regions (S1 Fig). Cloning of RT-PCR products revealed single 120 transcripts for KCTD2 and KCTD5, encoding proteins of 263 and 234 residues respectively, and two 121 alternatively spliced transcripts for KCTD17, encoding proteins of 225 and 220 residues with distinct 122 C-termini ( Fig 1B and S2A Fig)

128
We next assessed the expression of KCTD2, KCTD5, and KCTD17 proteins, using a 129 polyclonal anti-KCTD5 antibody that cross-reacts with mouse Inc orthologs and Drosophila Inc (S3A 130 Fig). This antibody detected a strongly reactive species of ~26 kD and additional species of ~28 to 29 131 kD in extracts from mouse and rat brain, cultured rat cortical neurons, and human 293T cells (Figs 1C 132 and 5D,and S3A Fig), consistent with the range of molecular weights predicted for KCTD2, KCTD5, 133 and KCTD17 (Fig 1B). The size of these immunoreactive species and their biochemical properties, 134 described further below, indicate that they correspond to one or more isoforms of KCTD2, KCTD5, 135 and KCTD17. The expression of Inc orthologs in the mammalian brain and Inc in the fly brain [11], 136 together with the similarity of their primary sequences, suggests that Inc defines a protein family that 137 may have conserved functions in the nervous system. Insomniac family members self-associate and bind Cul3 in an evolutionarily conserved 140 and interchangeable manner 141 Inc and KCTD5 are able to bind Cul3 and to self-associate [11,12,20], key attributes of BTB 142 adaptors [14][15][16][17]. To determine whether these attributes are universal to Inc orthologs and their 143 isoforms expressed in the nervous system, we first examined the physical interactions of these proteins 144 with mouse Cul3. Co-immunoprecipitations revealed that KCTD2, KCTD5, KCTD17.2, and 145 KCTD17.3 are able to associate with mouse Cul3 (Fig 2A and   To assess the extent to which the Inc-Cul3 interface is evolutionarily conserved, we tested 151 whether Drosophila Inc is able to associate with mouse Cul3 in a cross-species manner. We observed 152 that fly Inc and mouse Cul3 interact (Fig 2A), indicating that Inc readily assembles into mammalian 153 Cul3 complexes. Conversely, we tested whether mouse KCTD2, KCTD5, and KCTD17 can associate 154 with fly Cul3 in Drosophila S2 cells, and observed that each Inc ortholog associated with fly Cul3 in a 155 manner indistinguishable from Inc ( Fig 2B). The interchangeable biochemical associations of Inc 156 family members and Cul3 indicate that the Inc-Cul3 interface is functionally conserved from flies to 8 of Inc family proteins. The presence of three Inc orthologs in mammals and their likely co-expression 164 in brain regions such as thalamus and cortex (S1 Fig) led us to test whether these proteins can also 165 heteromultimerize. We observed robust heteromeric associations between all pairwise combinations of 166 Inc orthologs (Fig 3A), a property that may enable functional redundancy in vivo or the assembly of 167 functionally distinct complexes. To further probe the multimeric self-associations of Inc family 168 members, we tested whether KCTD2, KCTD5, and KCTD17 can heteromultimerize with Drosophila 169 Inc. We observed that each Inc ortholog associates readily with Inc in both mammalian and 170 Drosophila cells (Fig 3A and 3B). Thus, the multimerization interface of Inc family members is highly 171 conserved through evolution. Together with the interchangeable associations of Inc family members 172 and Cul3 (Fig 2)    Myc-Inc trafficked both to medial dendritic structures and to lateral puncta located in the same regions 266 as presynaptic termini of these neurons (Fig 6D and 6E). In PDF + neurons, Myc-Inc localized to 267 ipsilateral ventral projections to accessory medulla that are of a largely dendritic character ( ]. Neuronal inc activity is essential for normal sleep [11,12], but whether inc impacts synaptic 291 function is not known. To test whether the distribution of Inc at synaptic termini reflects a synaptic 292 function, we assessed the anatomical and physiological properties of inc mutants at the NMJ. inc 1 and 293 inc 2 null mutants both exhibited significantly increased bouton number with respect to wild-type 294 animals ( Fig 7A and 7B), indicating that Inc is essential for regulation of synaptic growth or plasticity.

295
To assess whether these anatomical defects are associated with altered synaptic transmission, we 296 recorded postsynaptically from muscle in control and transheterozygous inc 1 /inc 2 animals. While the 297 amplitude of spontaneous miniature postsynaptic potentials was not significantly altered in inc 298 mutants, their frequency was reduced (Fig 7C-7E). The amplitude of evoked postsynaptic potentials 299 triggered by presynaptic stimulation was significantly reduced in inc 1 / inc 2 mutants, and quantal 300 content was similarly decreased (Fig 7F-7H). The attenuation of evoked potentials and increased 301 bouton number in inc mutants suggest that a compensatory increase in synaptic growth may arise in 302 response to defects in synaptic transmission, though this increase does not compensate for the 303 decreased strength of inc synapses. These data indicate that inc is vital for normal synaptic structure 304 and physiology and suggest, together with the ability of Inc and its orthologs to localize to synapses, 305 that Inc family members may direct the ubiquitination of proteins critical for synaptic function.

RT-PCR and in situ hybridization 372
Total RNA was isolated with TRIZOL (ThermoFisher) from a single brain hemisphere of a 373 mixed C57BL/6 background mouse. 5 µg total RNA was annealed to random hexamer primers and 374 reverse transcribed with Thermoscript (ThermoFisher) according to the manufacturer's protocol.

539
Alexa 680 and Alexa 790 secondary antibodies were used as described above.

546
Samples were then blocked at room temperature for 30 min in PBS containing 7.5% normal donkey 547 serum and 0.05% Triton X-100, incubated overnight at 4°C in primary antibody cocktail prepared in 548 PBS containing 5% normal donkey serum and 0.05% Triton X-100, and washed 3×5 min in PBS at 549 room temperature. Secondary antibody cocktails were prepared similarly and incubated with samples  were Alexa 488 donkey anti-rabbit and Alexa 568 donkey anti-mouse. Neuromuscular junctions were 585 imaged with a confocal microscope and z-stacks were captured using 40× or 63× oil objectives at 586 27 512×512 resolution. Boutons were counted offline using a manual tally counter while manipulating z-587 stacks in 3-dimensional space using Zen software (Zeiss); each axon branch was counted separately to 588 avoid undercounting or duplicate counts and counts were performed three times to ensure consistency. Crosses were set with five virgin females and three males on cornmeal, agar, and molasses 612 food. One to four day old male flies eclosing from LD-entrained cultures raised at 25°C were loaded in 613 glass tubes containing cornmeal, agar, and molasses food. Animals were monitored for 5-7 days at 614 25°C in LD cycles using DAM2 monitors (Trikinetics). The first 36-48 hours of data was discarded to 615 permit animals to acclimate to glass tubes, and an integral number of days of data (3-5) were analyzed 616 using custom Matlab software as described previously [11]. Locomotor data were collected in 1 min 617 bins, and sleep was defined by inactivity for 5 minutes or more; a given minute was assigned as sleep