Automated analysis of long-term grooming behavior in Drosophila using a k-nearest neighbors classifier

Drosophila Contributions towards Understanding Neurofibromatosis 1

Neurofibromatosis 1 (NF1) is a multisymptomatic disorder with highly variable presentations, which include short stature, susceptibility to formation of the characteristic benign tumors known as neurofibromas, intense freckling and skin discoloration, and cognitive deficits, which characterize most children with the condition. Attention deficits and Autism Spectrum manifestations augment the compromised learning presented by most patients, leading to behavioral problems and school failure, while fragmented sleep contributes to chronic fatigue and poor quality of life. Neurofibromin (Nf1) is present ubiquitously during human development and postnatally in most neuronal, oligodendrocyte, and Schwann cells. Evidence largely from animal models including Drosophila suggests that the symptomatic variability may reflect distinct cell-type-specific functions of the protein, which emerge upon its loss, or mutations affecting the different functional domains of the protein. This review summarizes the contributions of Drosophila in modeling multiple NF1 manifestations, addressing hypotheses regarding the cell-type-specific functions of the protein and exploring the molecular pathways affected upon loss of the highly conserved fly homolog dNf1. Collectively, work in this model not only has efficiently and expediently modelled multiple aspects of the condition and increased understanding of its behavioral manifestations, but also has led to pharmaceutical strategies towards their amelioration.

Deep Phenotyping of Sleep in Drosophila

Sleep is an evolutionarily conserved behavior, whose function is unknown. Here, we present a method for deep phenotyping of sleep in Drosophila, consisting of a high-resolution video imaging system, coupled with closed-loop laser perturbation to measure arousal threshold. To quantify sleep-associated microbehaviors, we trained a deep-learning network to annotate body parts in freely moving flies and developed a semi-supervised computational pipeline to classify behaviors. Quiescent flies exhibit a rich repertoire of microbehaviors, including proboscis pumping (PP) and haltere switches, which vary dynamically across the night. Using this system, we characterized the effects of optogenetically activating two putative sleep circuits. These data reveal that activating dFB neurons produces micromovements, inconsistent with sleep, while activating R5 neurons triggers PP followed by behavioral quiescence. Our findings suggest that sleep in Drosophila is polyphasic with different stages and set the stage for a rigorous analysis of sleep and other behaviors in this species.

Capturing continuous, long timescale behavioral changes in Drosophila melanogaster postural data

Animal behavior spans many timescales, from short, seconds-scale actions to circadian rhythms over many hours to life-long changes during aging. Most quantitative behavior studies have focused on short-timescale behaviors such as locomotion and grooming. Analysis of these data suggests there exists a hierarchy of timescales; however, the limited duration of these experiments prevents the investigation of the full temporal structure. To access longer timescales of behavior, we continuously recorded individual Drosophila melanogaster at 100 frames per second for up to 7 days at a time in featureless arenas on sucrose-agarose media. We use the deep learning framework SLEAP to produce a full-body postural data set for 47 individuals resulting in nearly 2 billion pose instances. We identify stereotyped behaviors such as grooming, proboscis extension, and locomotion and use the resulting ethograms to explore how the flies' behavior varies across time of day and days in the experiment. We find distinct circadian patterns in all of our stereotyped behavior and also see changes in behavior over the course of the experiment as the flies weaken and die.

Sensing of a spore surface protein by a Drosophila chemosensory protein induces behavioral defense against fungal parasitic infections

In addition to innate immunity in a physiological context, insects have evolved behavioral defenses against parasite attacks. Here, we report that Drosophila can sense the CFEM (common in fungal extracellular membrane) protein Mcdc9, which acts as a negative virulence factor of the entomopathogenic fungus Metarhizium robertsii. The individual deletions of 18 CFEM genes in Metarhizium followed by fly infection identified three null mutants that could kill the flies more quickly than the wild-type strain, among which Mcdc9 can coat fungal spores and interact with the fly chemosensory protein CheA75a. The deletion of Mcdc9 in the fungus or the knockdown of CheA75a in flies had a similar effect, in which a greater number of fungal spores were left on flies than on the respective controls after topical infection. Thus, similar to the accelerated death of the wild-type flies treated with ΔMcdc9, the CheA75a^RNAi flies succumbed more quickly than the control insects topically challenged with the wild-type strain. The CheA75a gene is highly transcribed in fly legs and wings, and positive electrophysiological responses were evidenced in tarsal sensilla after stimulation with the Mcdc9 protein. The results imply that this CFEM protein could be sensed as a contact elicitor inducing the hygienic behavior of flies against fungal parasitic infection, which reveals a previously unsuspected mechanism of fungus-insect interactions.

The translational genetics of ADHD and related phenotypes in model organisms

Attention-deficit/hyperactivity disorder (ADHD) is a highly prevalent neurodevelopmental disorder resulting from the interaction between genetic and environmental risk factors. It is well known that ADHD co-occurs frequently with other psychiatric disorders due, in part, to shared genetics factors. Although many studies have contributed to delineate the genetic landscape of psychiatric disorders, their specific molecular underpinnings are still not fully understood. The use of animal models can help us to understand the role of specific genes and environmental stimuli-induced epigenetic modifications in the pathogenesis of ADHD and its comorbidities. The aim of this review is to provide an overview on the functional work performed in rodents, zebrafish and fruit fly and highlight the generated insights into the biology of ADHD, with a special focus on genetics and epigenetics. We also describe the behavioral tests that are available to study ADHD-relevant phenotypes and comorbid traits in these models. Furthermore, we have searched for new models to study ADHD and its comorbidities, which can be useful to test potential pharmacological treatments.

Behavioral Repertoire on a Vertical Rod-An Ethogram in Dermacentor reticulatus Ticks

Ticks are important vectors of pathogens that endanger humans and animals. Study of their behavior under laboratory conditions is important for both predicting their behavior in natural conditions and understanding their involvement in transmission cycles of pathogens, which may lead to effective prevention of tick-borne disease transmission or establishment of effective preventive measures. The aim of our study was to describe the behavior of D. reticulatus ticks using laboratory assay. We focused on the description of individual behavioral units during their vertical movement. The assay consisted of glass beakers filled with sand and an embedded glass rod. We observed 10 different behavioral units, 4 of which have not yet been described: body posturing called "jogger", leg grooming, and body or leg jerking. The most frequent tick behavior observed was an upwards positioning of the two front legs while the body remained motionless (88.9%). Other common observations were both horizontal (63%) and vertical (58.0%) body posturing with all legs lowered, followed by questing behavior (51.9%). Ticks spent the most time questing (75.2%), crawling (54.7%), and grooming legs on the right side (23%). We did not observe any differences between males and females.

Effects of plant growth retardant daminozide (Alar) on neuromuscular co-ordination behavior in Drosophila melanogaster

Daminozide (alar), a plant growth retardant, is used in different fruit orchard to make fruits attractive and reduce pre-harvest losses. Previously data demonstrated that acute daminozide exposure affected reproductive fitness and produced neurodegeneration in Drosophila melanogaster. The goal of this study was to determine whether continuous exposure to daminozide affects neuromuscular co-ordination in D. melanogaster as manifested in various behavioral responses. Fruit flies were exposed to 200 or 400 mg/L concentration of daminozide for two successive generations. Treated D. melanogaster were examined for the behaviors indicative of neuromuscular coordination and cognitive abilities, that include climbing, social interaction, adult grooming, migration, flight, male aggression, and adult courtship. Aberrant behavioral responses were noted among treated D. melanogaster of both sexes as evidenced by the following parameters: reduction in flight duration, abnormal social interaction, altered copulatory acts, and over-aggressiveness. Data suggest that daminozide produces impairment in neuromuscular coordination and cognitive ability in Drosophila, which was reflected as altered behavioral patterns. As Drosophila is considered as a reliable in vivo model utilized in toxicity testing, our findings may help us to anticipate and monitor potential daminozide-induced toxicity in animals and humans.

Central and Peripheral Clock Control of Circadian Feeding Rhythms

Many behaviors exhibit ~24-h oscillations under control of an endogenous circadian timing system that tracks time of day via a molecular circadian clock. In the fruit fly, Drosophila melanogaster, most circadian research has focused on the generation of locomotor activity rhythms, but a fundamental question is how the circadian clock orchestrates multiple distinct behavioral outputs. Here, we have investigated the cells and circuits mediating circadian control of feeding behavior. Using an array of genetic tools, we show that, as is the case for locomotor activity rhythms, the presence of feeding rhythms requires molecular clock function in the ventrolateral clock neurons of the central brain. We further demonstrate that the speed of molecular clock oscillations in these neurons dictates the free-running period length of feeding rhythms. In contrast to the effects observed with central clock cell manipulations, we show that genetic abrogation of the molecular clock in the fat body, a peripheral metabolic tissue, is without effect on feeding behavior. Interestingly, we find that molecular clocks in the brain and fat body of control flies gradually grow out of phase with one another under free-running conditions, likely due to a long endogenous period of the fat body clock. Under these conditions, the period of feeding rhythms tracks with molecular oscillations in central brain clock cells, consistent with a primary role of the brain clock in dictating the timing of feeding behavior. Finally, despite a lack of effect of fat body selective manipulations, we find that flies with simultaneous disruption of molecular clocks in multiple peripheral tissues (but with intact central clocks) exhibit decreased feeding rhythm strength and reduced overall food intake. We conclude that both central and peripheral clocks contribute to the regulation of feeding rhythms, with a particularly dominant, pacemaker role for specific populations of central brain clock cells.

Spatial Comparisons of Mechanosensory Information Govern the Grooming Sequence in Drosophila

Animals integrate information from different sensory modalities, body parts, and time points to inform behavioral choice, but the relevant sensory comparisons and the underlying neural circuits are still largely unknown. We use the grooming behavior of Drosophila melanogaster as a model to investigate the sensory comparisons that govern a motor sequence. Flies perform grooming movements spontaneously, but when covered with dust, they clean their bodies following an anterior-to-posterior sequence. After investigating different sensory modalities that could detect dust, we focus on mechanosensory bristle neurons, whose optogenetic activation induces a similar sequence. Computational modeling predicts that higher sensory input strength to the head will cause anterior grooming to occur first. We test this prediction using an optogenetic competition assay whereby two targeted light beams independently activate mechanosensory bristle neurons on different body parts. We find that the initial choice of grooming movement is determined by the ratio of sensory inputs to different body parts. In dust-covered flies, sensory inputs change as a result of successful cleaning movements. Simulations from our model suggest that this change results in sequence progression. One possibility is that flies perform frequent comparisons between anterior and posterior sensory inputs, and the changing ratios drive different behavior choices. Alternatively, flies may track the temporal change in sensory input to a given body part to measure cleaning effectiveness. The first hypothesis is supported by our optogenetic competition experiments: iterative spatial comparisons of sensory inputs between body parts is essential for organizing grooming movements in sequence.

Zhang et al. find that Drosophila covered with dust compare sensory inputs from mechanosensory bristles on different body parts during grooming. The ratio of anterior:posterior sensory input and its dynamics, rather than the rate of dust removal from the anterior, drives the anterior-to-posterior grooming sequence.

An automatic behavior recognition system classifies animal behaviors using movements and their temporal context

Animals can perform complex and purposeful behaviors by executing simpler movements in flexible sequences. It is particularly challenging to analyze behavior sequences when they are highly variable, as is the case in language production, certain types of birdsong and, as in our experiments, flies grooming. High sequence variability necessitates rigorous quantification of large amounts of data to identify organizational principles and temporal structure of such behavior. To cope with large amounts of data, and minimize human effort and subjective bias, researchers often use automatic behavior recognition software. Our standard grooming assay involves coating flies in dust and videotaping them as they groom to remove it. The flies move freely and so perform the same movements in various orientations. As the dust is removed, their appearance changes. These conditions make it difficult to rely on precise body alignment and anatomical landmarks such as eyes or legs and thus present challenges to existing behavior classification software. Human observers use speed, location, and shape of the movements as the diagnostic features of particular grooming actions. We applied this intuition to design a new automatic behavior recognition system (ABRS) based on spatiotemporal features in the video data, heavily weighted for temporal dynamics and invariant to the animal's position and orientation in the scene. We use these spatiotemporal features in two steps of supervised classification that reflect two time-scales at which the behavior is structured. As a proof of principle, we show results from quantification and analysis of a large data set of stimulus-induced fly grooming behaviors that would have been difficult to assess in a smaller dataset of human-annotated ethograms. While we developed and validated this approach to analyze fly grooming behavior, we propose that the strategy of combining alignment-invariant features and multi-timescale analysis may be generally useful for movement-based classification of behavior from video data.

Drosophila melanogaster grooming possesses syntax with distinct rules at different temporal scales

Mathematical modeling of behavioral sequences yields insight into the rules and mechanisms underlying sequence generation. Grooming in Drosophila melanogaster is characterized by repeated execution of distinct, stereotyped actions in variable order. Experiments demonstrate that, following stimulation by an irritant, grooming progresses gradually from an early phase dominated by anterior cleaning to a later phase with increased walking and posterior cleaning. We also observe that, at an intermediate temporal scale, there is a strong relationship between the amount of time spent performing body-directed grooming actions and leg-directed actions. We then develop a series of data-driven Markov models that isolate and identify the behavioral features governing transitions between individual grooming bouts. We identify action order as the primary driver of probabilistic, but non-random, syntax structure, as has previously been identified. Subsequent models incorporate grooming bout duration, which also contributes significantly to sequence structure. Our results show that, surprisingly, the syntactic rules underlying probabilistic grooming transitions possess action duration-dependent structure, suggesting that sensory input-independent mechanisms guide grooming behavior at short time scales. Finally, the inclusion of a simple rule that modifies grooming transition probabilities over time yields a generative model that recapitulates the key features of observed grooming sequences at several time scales. These discoveries suggest that sensory input guides action selection by modulating internally generated dynamics. Additionally, the discovery of these principles governing grooming in D. melanogaster demonstrates the utility of incorporating temporal information when characterizing the syntax of behavioral sequences.

Analysis of temporally rich behavioral sequences provides a quantitative description of the rules underlying their generation. Drosophila melanogaster grooming behavior consists of many complex sequences involving repetitions of well-characterized actions. In this paper, we leverage advances in machine vision to automatically annotate over 40 hours of video data of flies covered in dust and develop mathematical models that reveal the existence of syntax in D. melanogaster grooming. We find that sequence organization depends on grooming action identity, as has been well-established, and, more surprisingly, grooming action duration. The discovery of duration-dependent action selection leads us to conclude that, although sensory input informs grooming decisions on long time scales, internal dynamics also guide individual transitions between grooming actions. Therefore, incorporating action duration into our models allows us to uncover multi-scale temporal dynamics that suggest the existence of neural circuits dedicated to partially sensory-independent decision-making. Our approach highlights the importance of incorporating temporal information into sequential models, as doing so reveals the relative contributions of sensory input and internal dynamics to behavioral sequence generation.

Courtship behavior induced by appetitive olfactory memory

Reinforcement signals such as food reward and noxious punishment can change diverse behaviors. This holds true in fruit flies, Drosophila melanogaster, which can be conditioned by an odor and sugar reward or electric shock punishment. Despite a wide variety of behavior modulated by learning, conditioned responses have been traditionally measured by altered odor preference in a choice, and other memory-guided behaviors have been only scarcely investigated. Here, we analyzed detailed conditioned odor responses of flies after sugar associative learning by employing a video recording and semi-automated processing pipeline. Trajectory analyses revealed that multiple behavioral components were altered along with conditioned approach to the rewarded odor. Notably, we found that lateral wing extension, a hallmark of courtship behavior of D. melanogaster, was robustly increased specifically in the presence of the rewarded odor. Strikingly, genetic disruption of the mushroom body output did not impair conditioned courtship increase, while markedly weakening conditioned odor approach. Our results highlight the complexity of conditioned responses and their distinct regulatory mechanisms that may underlie coordinated yet complex memory-guided behaviors in flies.

Most sleep does not serve a vital function: Evidence from Drosophila melanogaster

Sleep appears to be a universally conserved phenomenon among the animal kingdom, but whether this notable evolutionary conservation underlies a basic vital function is still an open question. Using a machine learning-based video-tracking technology, we conducted a detailed high-throughput analysis of sleep in the fruit fly Drosophila melanogaster, coupled with a lifelong chronic and specific sleep restriction. Our results show that some wild-type flies are virtually sleepless in baseline conditions and that complete, forced sleep restriction is not necessarily a lethal treatment in wild-type D. melanogaster. We also show that circadian drive, and not homeostatic regulation, is the main contributor to sleep pressure in flies. These results offer a new perspective on the biological role of sleep in Drosophila and, potentially, in other species.

Fast animal pose estimation using deep neural networks

The need for automated and efficient systems for tracking full animal pose has increased with the complexity of behavioral data and analyses. Here we introduce LEAP (LEAP estimates animal pose), a deep-learning-based method for predicting the positions of animal body parts. This framework consists of a graphical interface for labeling of body parts and training the network. LEAP offers fast prediction on new data, and training with as few as 100 frames results in 95% of peak performance. We validated LEAP using videos of freely behaving fruit flies and tracked 32 distinct points to describe the pose of the head, body, wings and legs, with an error rate of <3% of body length. We recapitulated reported findings on insect gait dynamics and demonstrated LEAP's applicability for unsupervised behavioral classification. Finally, we extended the method to more challenging imaging situations and videos of freely moving mice.