Functional connectomics in C . elegans

The complete structure and connectivity of the Caenorhabditis elegans nervous system was first published in 1986. The ‘mind of a worm’ was the first organism to have its nervous system to be reconstructed at the level of synapses, and represented a critical milestone considering today it remains the only organism to be mapped to that level of connection. Recently, the extrasynaptic connectome of neuropeptides and monoamines has been described. This review discusses recent technological advances used to perturb whole-organism neuronal function, such as: whole brain imaging, optogenetics, sonogenetics and mutant analysis, which have allowed for interrogations of both local and global neural circuits, leading to different behaviors. A better understanding of a whole organism requires combining experimental datasets with biophysical neuronal modelling, and behavioral quantification. Combining these approaches will provide a complete understanding of the worm nervous system and shed light into how networks function and interact with the synaptic network to modulate information processing and behavioral output.


Introduction
Extensive research on the Caenorhabditis elegans neural connectome dates back to 1986, leaving them rooted in the minds of neuroscientists ever since.Investigation into the physical connections of the neural pathways by electron microscopy predates the naming of this research field (1).Knowledge of the brain's structure has improved exponentially over the past century; before which, focus was on gross anatomy and discrete brain regions.The shift to how neurons connected at the beginning of the 20 th century with debates between Camillo Golgi and Santiago Ramón y Cajal over the connections between neurons (2).Initially referred to as "collation of connectivity data," the term "connectomics" was popularized in 2005 with regards to the human connectome (3,4).This field was brought to public consciousness by 2012, alongside the "Human Connectome Project" and reinforced the following year with the Brain Research through Advancing Innovative Neurotechnologies® (BRAIN) Initiative (5)(6)(7)(8)(9)(10).One overarching goal in neuroscience is to understand what structural components of a neural circuit relate to certain behaviors.Understanding this relation is not possible with the connectome alone, though it provides a strong foundation.One neuron could be involved in one behavior, but one behavior could be tied to a circuit of neurons.However, all behaviors result from summation of actions that take place within the connectome.

Caenorhabditis elegans
The importance of the connectome comes from the understanding that structure can influence function.Knowing the structure of a network is critical for understanding the function.Historically, most of the brain's functions have only been understood when there was injury to isolated structures; as was the case with Phineas Gage and others whose injuries advanced scientific knowledge (11).Due to ethical considerations, there is a reliance on retrospective studies, dependent on injuries, to understand function.The appeal of model organisms becomes apparent given the cost of care, generally shorter life span, and simpler structure.C. elegans are a microscopic worm that has modeled macroscopic research for six decades (Figure 1).The nematode's wiring circuit, identified by electron microscopy in 1986, was one of the original renderings of a connectome for any organism (1).Over three decades later, it remains the only organism with a completely mapped connectome.
With model organisms the focus therefore becomes microscopic -in scale, not significance because it is possible to generate diseased or genetic homologs.One technique used to understand the shape of connections is microsurgery, which takes place at the level of individual neurons.Immobilizing a C. elegan, it is possible to use a laser to ablate neural bodies (12,13).Laser ablations are possible in C. elegans because all of the neurons have been mapped and they are eutelic, making it possible to ablate the same neuron across multiple animals (14).This technique has power in manipulating structure.Worms can later be observed to see how behavior and function are related to structure.These types of studies are important, as damage of physical destruction to connections allows for the consideration as to how structural changes impact function and signaling between neurons.These changes to neuronal structure can be achieved across generations by using genetic ablations by expressing varying toxins or caspases (15).
One of the most commonly used tools in identifying structures in human models is Magnetic Resonance Imaging, or MRI.First image generated in 1973, this technique utilizes a magnetic field created within the MRI machine to align the protons of water molecules within the sample (16).This information can be analyzed to provide structural data of the sample.Variants of this technique include diffusion MRI, which is of particular interest when studying the brain, as it can differentiate white matter tracts from gray matter.This technique isn't used to understand the structure of individual neurons or small scale circuits, but rather of the whole brain.It has tremendous strength in identifying abnormalities in structure and large pathways between regions (17).
Non-human models make use of many other techniques in determining the structure of the nervous system, however.While MRI relied on the location of protons, electron microscopy measures electron density.This method requires a fixed sample to be cut nanometers thick before staining and photographing the sections.Software then compiles these images into a 3D reconstruction of the image.In the case of C. elegans, a worm less than 1 mm in length from tip to tail, the initial process of imaging rendered 8000 images that had to be compiled (1).It is not difficult to imagine why this technique may take some time to analyze a larger, more complex organism, like a human as the BRAIN Initiative hopes to achieve (7).
Using C. elegans grants a better understanding of the mechanisms and possible treatments for diseases.In neuroscience especially, the central goal is to understand how the nervous system functions under typical and atypical circumstances (18).However, despite knowing the physical connections of the C. elegans nervous system, much remains to be learned about how the nervous system functions as a whole to enact behaviors (19).This involves understanding neuronal dynamics that extend beyond anatomical synapses, such as how neuromodulators can act over longer distances and timescales to shape behavioral responses.Further complexing our understanding of this "functional connectome," is how the same neuron is capable of playing different roles in two different circuits.

Functional Focus
C. elegans neurons have been grouped into three functional "classes" of neurons: sensory neurons, motor neurons, and interneurons.The sensory neurons account for a third of neurons, and have more connections pre-synaptic than post-synaptic.Contrarily, motor neurons, another third of all neurons, have more post-synaptic connections.The remaining neurons are considered to be interneurons, with large numbers of both preand post-synaptic connections (20).These structural components can be understood in the context which they aid function.With more pre-synaptic neurons, sensory neural bodies are able to gather more signals from surrounding neurons.Post-synaptic connections within motor neurons allows them more synaptic connections with muscles to therefore produce fine-tuned actions.Understanding the connections alone between these neuron classes helps increase our understanding of how a signal is transduced, processed, to ultimately produce actions, as seen in previous work on C. elegans' navigation (21).However, these actions cannot be understood by the static connections alone.
It is well documented how in the C. elegans connectome, a single neuronal class can be involved in the sensation of diverse stimuli or elicit different behaviors.For example, the polymodal, nociceptive neuron, ASH, detects a myriad of different mechano-, osmo-, and chemo-stimuli that result in aversive behavior (22)(23)(24)(25)(26)(27)(28).However, not all stimuli utilize the same pathways and connections as one might expect given detection by a single sensory neuron serving as progenitor for these circuits.
The diversity in neuronal circuitries may be due to the intracellular machinery used within individual neurons.C. elegans is equipped with a large set of G protein subunits that exhibit overlapping expression, rendering particular intracellular pathways important in different behavioral circuits (29).The nematodes genome codes for 21 Gα protein subunits, and 2 subunits of both Gβ and Gγ proteins (29).Of the 21 Gα subunits, 16 are expressed throughout the chemosensory neurons, and many overlap in their expression profiles (29).For example, on its own, ASH expresses ten different Gα subunits, while ASE expresses only three (29).
One of the earliest methods to monitor neuron function was patch clamp electrophysiology, a technique first used on ex vivo giant squid axons (30).This technique lends itself to understanding functional connectomics as it monitors the flow of ions across neuronal membranes (31).In C. elegans it was used to understand the role of graded potentials, in opposition to mammals all or nothing action potentials (19) (31).This method is still rather invasive, requiring fixed samples of individual neurons for testing (32).There are however, emerging techniques that allow for measurements of single cells within a larger neuronal structure.
It is possible to elucidate the functional connectome in less invasive ways than when determining the structural connectome.However, this often means that the images focus on individual neurons but the sample would not be permanently modified by the imaging procedures.A variant of MRI, functional MRI (fMRI), connects structure and function within in vivo samples (33).In these scans, a subject can be asked to perform a task or to think of a topic, and the blood flow within their brain can be tracked (34).This set up can also be manipulated to expose a sample to a particular stimuli and the responses monitored.In this way, it is even possible to use a non-human model.
Techniques used pan-neuronally effect all neurons in a region surrounding the area of interest with the use of chemicals or induced electrical stimulation.Manipulation of a single neuron, such as in laser ablation, permanently alter structure before testing for a change in function.Optogenetics offers temporal control of activity of individual neurons utilizing light-controlled ion channels.This modification requires the insertion of opsin genes, responsible for proteins such as channelrhodopsin-2, natively found in green algae or Chlamydomonas reinhardtii (35).The activity of a single neuron can be controlled by this method.Similarly, calcium imaging allows for the activity of a single neuron to be monitored (36).As in optogenetic studies, genetic manipulation is involved, but in this case using fluorescent tags such as GFP or mCherry fused to the calcium binding protein, calmodulin, resulting in Genetically Encoded Calcium Indicators (GECIs).These GECIs are ideally localized to the neural body and or whole neuron.The technique is possible as it is known which genes localize to different neuron types and locations.In this technique, C. elegans is immobilized in a microfluidic device, and calcium levels are measured via changes in fluorescence of the GECI (37,38).

Divergent functions within a neuronal class
The sensation of diverse stimuli as understood via the connectome is also seen in work on multisensory integration.While multisensory integration work has been conducted with humans using techniques such as magnetoencephalography (MEG) and fMRI, and also with model organisms such as Drosophila melanogaster, the results are science still needs time to finish its next connectome and will also have to overcome the obstacle of much more complex neural circuitry (13,42,43).For example recent work on the multilevel multimodal convergence circuit for rolling in Drosophila, which relies on multisensory integration, was limited by the lack of knowledge surrounding multisensory neuronal convergence (44).All of these factors, especially the comparatively well understood C. elegans connectome, combined with what is otherwise often costprohibitive methodology, make C. elegans an ideal choice for multisensory work.
Multisensory methodology utilizing C. elegans falls into two broad categories; coexposure to two stimuli, one that is aversive and one that is agreeable, or exposure to one stimuli in conjunction with an environmental indicator, both of which can be done via avoidance assays (45).The avoidance assay, also commonly known as the drop test was pioneered by in the early 2000s, and has transformed how researchers can approach multisensory integration research in C. elegans (46).The stimuli is dropped on the tail of the animal and through capillary action, this droplet quickly reaches the anterior amphid sensory organs, ultimately resulting in the C. elegan either continuing movement in the same direction or moving backwards to avoid the stimuli (46).Both methods of multisensory exposure within the scope of avoidance assays are comparatively simple and inexpensive, which also allows for more robust data.Furthermore, with only 60 ciliated sensory neurons, including 16 pairs of chemosensory neurons neuronal ablation techniques such as the laser ablation method previously discussed as well as genetic ablation become effective tools for understanding the diverse roles of individual sensory neurons with respect to the connectome as a whole (47,48).
Examining one neuron in particular, ASH, studies have revealed differential use of both intra-and inter-signaling molecules by the neuron upon detection of various stimuli, yet still resulting in the same behavioral outcome: avoidance (24,27,28,(49)(50)(51).For example, nose touch avoidance, which is assayed by allowing the animal to run into an eyelash positioned perpendicularly to the animal's movement, requires expression of itr-1 in ASH (Figure 2A) ( 49).Yet itr-1 have not been found to be necessary for osmotic aversive responses mediated the same neuron (Figure 2A) (49).This implies that specific G protein subunits are utilized by the same neuron in response to individual stimuli, in turn activating unique pathways within the neuron and ultimately deciding which synapses relay the response, establishing the functional connections.Indeed, response to nose touch and benzaldehyde require itr-1 in ASH neurons (49).However, only nose touch requires IP3 production via phospholipase C enzymes egl-8 and plc-3 (49).It is therefore likely that the upstream Gαq subunit, egl-30, is only involved in nose touch, but not benzaldehyde detection (49).
Downstream of the initial sensation of stimuli, differences in functional postsynaptic connections have also been observed.As with itr-1, the glutamate receptor, glr-1, is utilized primarily in nose touch avoidance, as well as regulating subtle reactions within of osmosensation (Figure 2A) (28,50,51).Conversely, specific genes within ASH, such as osm-10, are specific to osmotic detection, and not tactile response.These genes have been shown to have specific downstream targets, such as nmr-1 (Figure 2A) ( 27, 51).Thus, it is possible different stimuli evoke different intra-signaling pathways which lead to varied synaptic release profiles to enact specific downstream targets in a functional circuit.Supporting this notion is the presence of both clear synaptic and dense core vesicles in ASH (1).Furthermore, it has been shown in ASI neurons, that depending on the signaling molecule, different neuropeptides are released from distinct neuronal compartments and asymmetrically between the neurons in the pair (52).
While certain intracellular components and synaptic connections are vital in some behaviors, they may be irrelevant in other behavioral circuits which utilize the same neurons.One example of this is the amphid sensory neuron, ADL, and its involvement in the response to ascaroside #3 (ascr#3).Ascarosides, are small-molecule signals, which serve diverse functions in inter-organismal chemical signaling (53).
Hermaphrodites are observed to avoid ascr#3 through ADL chemical synaptic transmission, presumably, to the backward command interneurons AVA and AVD (1,54).
Promotion of ADL response to ascr#3 is achieved through the gap junction hub-andspoke RMG circuit, wherein the interneuron RMG serves as a hub to modulate sensory neuron responses (54,55).RMG, through the activity level of the neuropeptide receptor npr-1, and input from the sensory neuron ASK, can inhibit ADL triggered avoidance by altering gap junction properties (54,55).Thus, chemical synapses are involved in the avoidance to ascr#3, whereas gap junctions are necessary for modulating the response in an npr-1 dependent manner to elicit aggregation or attraction (Figure 2B).

Sex Differences
As an androdioecious species, C. elegans are primarily hermaphroditic, with only 0.1% of the population being males (56,57).As such, most connectome efforts and studies have identified 302 neurons in hermaphrodite (13).Despite the male connectome not being completely mapped, much has been studied in regards to the sex-specific differences in the 385 neuronal cells present in the male (58).The increase in neurons has been shown to be localized in the tail and head of the animal, and largely function in finding and reproducing with a mate (58)(59)(60).Furthermore, sex-specific circuits have been identified that govern the male response to sex pheromones, demonstrating the importance of fully mapping neural circuits in both hermaphrodites and males (61).
Interestingly, the sex of the animal can establish the synaptic connection and function of a neuron.Ascr#3 is also sensed by ADF, but only in males, and hermaphrodites which have been masculinized through expression of fem-3, which inhibits the sexual regulator gene, tra-1 (63,64).Neuronal activation of ADF by ascr#3 requires mab-3, which is naturally inhibited in hermaphroditic animals (63).As ADL is still activated in males, masculinized ADF inhibits the aversive response to ascr#3.This inhibition may be taking place via extrasynaptic connections, or direct serotonin signaling on downstream neuronal target of ADL (Figure 2B).Sex can also result in different physical circuits, where synapses between certain neurons are only present in males, and pruned in hermaphrodites (65).This highlights the necessity of investigating how sex results in specific connections underlying a behavioral circuit, not merely the requisite neuron, in order to generate a more complete functional connectome.

Modulation
Behavioral circuits are dependent on the state of the animal.While receptor expression profiles and the sex of the animal are set variables, more flexible stateslargely the physiological state of the animal -shape and modulate these functional circuits.Sensory networks are altered by neuromodulators (neurotransmitters and neuropeptides) in a context specific manner; over varying distances and timescales.The effect of these modulations varies based on site of release and local concentration as governed by release, degradation, and reuptake of neuromodulators.
Serotonin (5-HT) has been shown to have a large role in behaviors related to foraging, egg laying, and locomotion, dependent on the presence or absence of food, as expression levels are correlated with being either fed or starved.For example, when food is present, 5-HT acts via GPA-11 to sensitize ASH to 30% octanol aversion (66).
Interestingly, when dissecting the role of 5-HT, it was found that the site of release is important and can have opposite effects.5-HT released from NSM sensitizes ASH to reverse quicker to 30% octanol.However, 5-HT released from ADF acts on ASH and results in shorter reversals, followed by forward locomotion (67).This highlights how a single neurotransmitter, in the same circuit, can give rise to different synaptic strengths and fine-tuned behavioral outputs, revealing that it is tantamount to not just consider neurotransmitters on a global scale.Moreover, the same stimulus does not necessarily utilize the same circuit at different concentrations.Different functional circuits are realized when animals are responding to 100% versus 30% octanol (66).At 100% octanol, ADL and AWB form electrical synapses onto command interneurons via GLR-1, are important when animals are starved, whereas 30% octanol aversion, regardless of food presence, is only mediated by ASH (66).
Furthermore, timescales of stimulus detection appear to be programmed into the response.As seen in copper avoidance, a cross-talk inhibition circuit between ASI and ASH fine tunes the behavioral response, with ASH responding quickly and robustly in comparison to a slower, weaker response by ASI which inhibits further ASH activation (68).Whereas this is a short-term reciprocal inhibition state, long-term behavioral states also exist that shape functional circuits.For example, roaming and dwelling states in the presence of food alternate, and last for minutes at a time.This switch is achieved via two opposing neuromodulators, dwelling is promoted by serotonergic neurotransmitter signaling, whereas the roaming state is established by the neuropeptide PDF (Figure 2C) (69).Strikingly, this functional circuit acts in a seemingly unorthodox manner, as it defies classical circuit logic of sensory to motor organization: motor and interneurons modulate the activity of sensory neurons (69).This largely extrasynaptic, long-term timescale circuit has many potential inputs that can bias signaling of one state over another.Interestingly, the only overlap between these two circuits involves the interneuron AIY (69).Perhaps, the odor of food biases the switch between dwelling and roaming while on food.
In fact, an odor detection switch in local search behaviors intersects with AIY as well (70).AWC detects food depletion in a dose dependent way: as food is removed, AWC is disinhibited, allowing for inhibition of AIY (70,71).Thus, it is plausible that short timescale detection by AWC recognizes changing concentrations of food, and relays this information to AIY, biasing a switch between the long-term roaming and dwelling states when on food.Together, functional connectomes can vary and take shape in drastically different ways than wiring diagrams suggest, with particular synaptic importance being dictated by physiological states and timescales.Additionally, functional circuits do not work in isolation, the final behavioral output is a readout of the fine tuning of multiple functional circuits creating a functional connectome.
Functional circuits to one stimuli do not act alone, just as the internal state modulates the response to a particular cue, the presence of multiple stimuli are integrated into larger networks.One example of this is the "flip/flop" model of integration of contradictory "good" and "bad" stimuli.Previous research regarding multisensory integration in C. elegans highlights the importance of applying accessible methodologies of avoidance assays while considering neural circuitry more broadly such as through the connectome.For example, a study identifying a "flip-flop" circuit in C. elegans that is vital for the multisensory processing of aversive and attractive stimuli simultaneously would not have been as successful without this circuitry knowledge (72).Similarly, recent work on an inhibitory circuit essential for adaptive avoidance was able to explore intermediate functions because there was already a comprehensive understanding of synaptic connections and work on the functional mapping of said connections (68).
An example of this model occurs with high levels of repulsive quinine (sensed by ASH) and low levels of the attractant diacetyl (sensed by AWA) animals do not exhibit pharyngeal pumping.However, as the concentration of quinine is decreased, the pumping rate, in a steep sigmoidal fashion, increases, displaying a "flip" to increased pharyngeal pumping (72).Likewise, if quinine levels remain unchanged, a non-linear switch in pharyngeal pumping can be seen as diacetyl concentration increases (72).This flip/flop around a particular threshold requires 5-HT and tyramine signaling between the RIM, RIC, and NSM interneurons via a serotonin channel (MOD-1) and tyramine receptor (SER-2) (72).Interestingly, animals that lack MOD-1 and SER-2 still show a decrease in pumping as quinine is increased, but in a linear fashion instead of a flip/flop, on/off switch at a particular threshold.Thus, these two sites of action for the neuromodulators are required for fine tuning the response around critical levels, but not for the overall integration of the two stimuli.
How integration of cues allows for the modulation of circuits can further be exemplified by looking at the tolerance of threats.Expectantly, C. elegans that are well fed are not willing to cross a high osmotic barrier to chemotax towards diacetyl, the risk is not worth reward.However, animals which are deprived of food will cross the same osmotic barrier, presumably as the risk is worth the reward (73).This modulation requires slow accumulation of tyramine, as it increases along with extended times of starvation, thereby desensitizing ASH to the osmotic stressor, and requires a few hours to reach a level which allows for the switch to decide to cross the osmotic barrier (73).
The aforementioned examples showcase the complexity underlying functional circuits, as there seem to be multiple levels of neuronal processing acting in parallel to finely adjust how the animal responds, including specific intercellular machinery allows for rapid adjustment of neuronal responses, thereby affecting the output, and these modulations can take place over long time scales -not merely minutes, but instead hours.
Thus, to truly decipher a functional circuit, many different time points and stimulus concentrations would have to be investigated, as there is likely hidden information that acts at a deeper level.

Figure 1 .
Figure 1.Since its discovery, C. elegans has played a pivotal role in expanding what is known

Figure 2 .
Figure 2. Differential use of neurons and pathways to create functional circuits.A) The same