SPINDLE: End-to-end learning from EEG/EMG to extrapolate animal sleep scoring across experimental settings, labs and species.

Understanding sleep and its perturbation by environment, mutation, or medication remains a central problem in biomedical research. Its examination in animal models rests on brain state analysis via classification of electroencephalographic (EEG) signatures. Traditionally, these states are classified by trained human experts by visual inspection of raw EEG recordings, which is a laborious task prone to inter-individual variability. Recently, machine learning approaches have been developed to automate this process, but their generalization capabilities are often insufficient, especially across animals from different experimental studies. To address this challenge, we crafted a convolutional neural network-based architecture to produce domain invariant predictions, and furthermore integrated a hidden Markov model to constrain state dynamics based upon known sleep physiology. Our method, which we named SPINDLE (Sleep Phase Identification with Neural networks for Domain-invariant LEearning) was validated using data of four animal cohorts from three independent sleep labs, and achieved average agreement rates of 99%, 98%, 93%, and 97% with scorings from five human experts from different labs, essentially duplicating human capability. It generalized across different genetic mutants, surgery procedures, recording setups and even different species, far exceeding state-of-the-art solutions that we tested in parallel on this task. Moreover, we show that these scored data can be processed for downstream analyzes identical to those from human-scored data, in particular by demonstrating the ability to detect mutation-induced sleep alteration. We provide to the scientific community free usage of SPINDLE and benchmarking datasets as an online server at https://sleeplearning.ethz.ch. Our aim is to catalyze high-throughput and well-standardized experimental studies in order to improve our understanding of sleep.

Designed cell consortia as fragrance-programmable analog-to-digital converters.

Synthetic biology advances the rational engineering of mammalian cells to achieve cell-based therapy goals. Synthetic gene networks have nearly reached the complexity of digital electronic circuits and enable single cells to perform programmable arithmetic calculations or to provide dynamic remote control of transgenes through electromagnetic waves. We designed a synthetic multilayered gaseous-fragrance-programmable analog-to-digital converter (ADC) allowing for remote control of digital gene expression with 2-bit AND-, OR- and NOR-gate logic in synchronized cell consortia. The ADC consists of multiple sampling-and-quantization modules sensing analog gaseous fragrance inputs; a gas-to-liquid transducer converting fragrance intensity into diffusible cell-to-cell signaling compounds; a digitization unit with a genetic amplifier circuit to improve the signal-to-noise ratio; and recombinase-based digital expression switches enabling 2-bit processing of logic gates. Synthetic ADCs that can remotely control cellular activities with digital precision may enable the development of novel biosensors and may provide bioelectronic interfaces synchronizing analog metabolic pathways with digital electronics.

Synthetic mammalian trigger-controlled bipartite transcription factors.

Synthetic biology has significantly advanced the design of synthetic control devices, gene circuits and networks that can reprogram mammalian cells in a trigger-inducible manner. Prokaryotic helix-turn-helix motifs have become the standard resource to design synthetic mammalian transcription factors that tune chimeric promoters in a small molecule-responsive manner. We have identified a family of Actinomycetes transcriptional repressor proteins showing a tandem TetR-family signature and have used a synthetic biology-inspired approach to reveal the potential control dynamics of these bi-partite regulators. Daisy-chain assembly of well-characterized prokaryotic repressor proteins such as TetR, ScbR, TtgR or VanR and fusion to either the Herpes simplex transactivation domain VP16 or the Krueppel-associated box domain (KRAB) of the human kox-1 gene resulted in synthetic bi- and even tri-partite mammalian transcription factors that could reversibly program their individual chimeric or hybrid promoters for trigger-adjustable transgene expression using tetracycline (TET), γ-butyrolactones, phloretin and vanillic acid. Detailed characterization of the bi-partite ScbR-TetR-VP16 (ST-TA) transcription factor revealed independent control of TET- and γ-butyrolactone-responsive promoters at high and double-pole double-throw (DPDT) relay switch qualities at low intracellular concentrations. Similar to electromagnetically operated mechanical DPDT relay switches that control two electric circuits by a fully isolated low-power signal, TET programs ST-TA to progressively switch from TetR-specific promoter-driven expression of transgene one to ScbR-specific promoter-driven transcription of transgene two while ST-TA flips back to exclusive transgene 1 expression in the absence of the trigger antibiotic. We suggest that natural repressors and activators with tandem TetR-family signatures may also provide independent as well as DPDT-mediated control of two sets of transgenes in bacteria, and that their synthetic transcription-factor analogs may enable the design of compact therapeutic gene circuits for gene and cell-based therapies.

Circadian VIPergic Neurons of the Suprachiasmatic Nuclei Sculpt the Sleep-Wake Cycle.

Although the mammalian rest-activity cycle is controlled by a "master clock" in the suprachiasmatic nucleus (SCN) of the hypothalamus, it is unclear how firing of individual SCN neurons gates individual features of daily activity. Here, we demonstrate that a specific transcriptomically identified population of mouse VIP+ SCN neurons is active at the "wrong" time of day-nighttime-when most SCN neurons are silent. Using chemogenetic and optogenetic strategies, we show that these neurons and their cellular clocks are necessary and sufficient to gate and time nighttime sleep but have no effect upon daytime sleep. We propose that mouse nighttime sleep, analogous to the human siesta, is a "hard-wired" property gated by specific neurons of the master clock to favor subsequent alertness prior to dawn (a circadian "wake maintenance zone"). Thus, the SCN is not simply a 24-h metronome: specific populations sculpt critical features of the sleep-wake cycle.

Dynamic- and Frequency-Specific Regulation of Sleep Oscillations by Cortical Potassium Channels.

Primary electroencephalographic (EEG) features of sleep arise in part from thalamocortical neural assemblies, and cortical potassium channels have long been thought to play a critical role. We have exploited the regionally dynamic nature of sleep EEG to develop a novel screening strategy and used it to conduct an adeno-associated virus (AAV)-mediated RNAi screen for cellular roles of 31 different voltage-gated potassium channels in modulating cortical EEG features across the circadian sleep-wake cycle. Surprisingly, a majority of channels modified only electroencephalographic frequency bands characteristic of sleep, sometimes diurnally or even in specific vigilance states. Confirming our screen for one channel, we show that depletion of the KCa1.1 (or "BK") channel reduces EEG power in slow-wave sleep by slowing neuronal repolarization. Strikingly, this reduction completely abolishes transcriptomic changes between sleep and wake. Thus, our data establish an unexpected connection between transcription and EEG power controlled by specific potassium channels. We postulate that additive dynamic roles of individual potassium channels could integrate different influences upon sleep and wake within single neurons.

The forebrain synaptic transcriptome is organized by clocks but its proteome is driven by sleep.

Neurons have adapted mechanisms to traffic RNA and protein into distant dendritic and axonal arbors. Taking a biochemical approach, we reveal that forebrain synaptic transcript accumulation shows overwhelmingly daily rhythms, with two-thirds of synaptic transcripts showing time-of-day-dependent abundance independent of oscillations in the soma. These transcripts formed two sharp temporal and functional clusters, with transcripts preceding dawn related to metabolism and translation and those anticipating dusk related to synaptic transmission. Characterization of the synaptic proteome around the clock demonstrates the functional relevance of temporal gating for synaptic processes and energy homeostasis. Unexpectedly, sleep deprivation completely abolished proteome but not transcript oscillations. Altogether, the emerging picture is one of a circadian anticipation of messenger RNA needs in the synapse followed by translation as demanded by sleep-wake cycles.