5 Deep turns and the dynamics of reorientation in escape response and free crawling of C. elegans Abstract — In the previous chapter, we developed an automated tracking algorithm for C. elegans 2D motor behavior that solves self-occluding body shapes. Here, we apply that algorithm to study the dynamics of deep turning during two important behaviors: the escape response, and abrupt reorientations in freely crawling worms. During the escape response, we find that the worm steers sharply away from the noxious stimulus by 180◦ on average, in a tightly controlled way. Additional reorientations before and after the turn broaden the distribution of reorientation angles, and can be linked to known underlying neural and molecular mechanisms. During free crawling, we find that abrupt turns, appearing in previous literature as ‘omega turns’, are in fact differentiated into two distinct classes. Despite a dramatically different visual appearance, both statically and dynamically, the two turns share similar kinematics in posture space; only the amplitude of a pulse in the third postural eigenmode is a dominant distinguishing feature. The two classes of turns reorient the worm towards different sides: omega turns ventrally; delta turns dorsally, by over-turning through the ventral side. We show that the two turns occur independently, but with approximately equal rates that remain equal during adaptation. Taken together, these observations suggest a shared underlying neural infrastructure, and a more diverse navigational repertoire than previously thought. See the footnote at the start of Chapter 4, p. 97. 125 5 Deep turns and dynamics of reorientation in C. elegans 5.1 Introduction In the previous chapter, we have developed a tracking algorithm for the full 2D motor behavior of C. elegans. This algorithm is capable of tracking not just ‘simple’ postures, but also self-occluding ones. This enables us to now take a closer, quantitative look at behaviors of the worm that feature such self-occluding body shapes, and that have thus far not been fully resolved. In the first part of this chapter, we focus on the escape response of C. elegans. This sequence of orchestrated behavioral motifs is evoked by a noxious stimulus (in our case a localized heating of the worm’s head), and turns the worm away from the stimulus. The ‘omega turn’, a sharp reorientation featuring self-overlapping body shapes, is the crucial part of the sequence, and can now for the first time be resolved. Studying this highly stereotyped behavior also helps us to identify quantitative patterns in the data, which will guide our analysis in the second part. In the second part, we analyze free crawling experiments. Extensive research has been done on the strategies used by the worm to localize food and favorable habitats, but the sharp turns that occur during this behavior have never been fully investigated. Here, we take the first steps towards resolving these turns. 5.2 The C. elegans escape response 5.2.1 Introduction to the escape response C. elegans is capable of responding to a range of potentially harmful conditions. Its sensory neurons, most of which are connected to the outside world through non-motile cilia [1] [2, Ch. 3.1], can detect noxious stimuli of a chemical, mechanical, optical, or thermal nature [3]. Upon encountering a sufficiently strong stimulus, the worm executes a stereotyped escape response: a behavioral sequence that steers the worm away from the observed threat. 126 5.2 The C. elegans escape response (i) (ii) (iii) (iv) (v) stimulus sensory neurons AVB forward locomotion forward locomotion PVC ALM AVM backward locomotion tyramine RIM head movement extra-synaptic tyramine VB VD DD DA VA SMD AVD RIM DB RMD AVA locomotory system asymmetry SMD / RIV deep head swing time Figure 5.1 | Schematic overview of the C. elegans escape response (based on ref. 4). Top row shows worm body shapes extracted from tracking data: (i) forward locomotion and exploratory head motions; (ii) infrared laser stimulus; (iii) reversal phase; (iv) omega turn; (v) resumption of forward locomotion in opposite direction. Diagram below shows sequence of events in the C. elegans nervous system. Time flows from left to right. RIM, SMD, and RIV designate C. elegans neurons. Green plus signs indicate stimulation; red minus signs inhibition. Inset shows the neural network implicated in the escape response [4, 5]. Rectangles: sensory neurons; hexagons: command neurons; circles: motor neurons. Synaptic connections are indicated by triangles, gap junctions by bars, and extra-synaptic diffusion of tyramine by a dashed arrow. Green connections are stimulatory, red ones inhibitory. Details have been omitted; see ref. 4 for a more in-depth overview. 127 5 Deep turns and dynamics of reorientation in C. elegans Fig. 5.1 (top) shows the timeline of the C. elegans escape response. During normal locomotion on an agar surface (i), the worm moves by propagating a sinusoidal wave through the body. This snake-like motion is typically accompanied by exploratory head movements. In our experiments, we elicit an escape response by applying a localized thermal stimulus to the worm’s head, using an infrared laser pulse (ii) (this causes a 0.5 ◦ C temperature increase at the head over 100 ms) [6]. The stimulus causes the worm to abruptly pause, and then back up (iii). During this reversal phase, head movements are suppressed, which has been suggested to increase the worm’s chances of escaping from the ‘snare traps’ of predatory fungi [7]. After roughly 5 to 10 seconds of reversing, the worm executes an omega turn: a sharp reorientation manoeuvre, during which the worm’s body briefly resembles the shape of the Greek letter Ω (iv) [5]. The turn reorients the worm away from the stimulus, and allows it to resume forward locomotion in the opposite direction (v). Some aspects of the escape response sequence have been teased apart at the cellular, molecular, and genetic level. This has been greatly facilitated by the availability of a genetic toolbox for C. elegans [8] as well as the worm’s stereotyped development [9]. The latter means that each individual worm has the exact same body plan, in which each of its roughly 1 000 somatic cells can be named, and each cell develops from a fully predictable lineage, starting at the fertilized egg [9]. Genetic changes can be easily introduced into the worm’s genome, and can often be made to target specific cells [8]. As an example, using these techniques, the nociceptive ASH neurons have been found to express ion channel proteins that are implicated in osmo- and mechanosensation [10]. This makes the ASH neurons multimodal, and downstream signaling therefore relies on signal multiplexing [11]. As the connectivity of the C. elegans nervous system is largely known [1], it has even been possible to partly trace the neural network that drives the escape response (Fig. 5.1, inset) [5, 4]. The escape response is classically studied with a ‘gentle anterior touch’: a touch to the head of the worm using a human hair [12] (typically, an eye- 128 5.2 The C. elegans escape response brow hair). In this scenario, a qualitative description of the mechanisms that orchestrate the different phases of the escape response has emerged from experiments (Fig. 5.1b) [4, 13], and a key element is the neurotransmitter tyramine, released by the RIM interneurons. Tyramine acts through multiple pathways, one of which is extra-synaptic diffusion and binding to G-protein coupled receptors (GPCRs) in distant cells. As this diffusion process has a timescale of seconds (see Methods) — much slower than the millisecond timescale associated with the initial synapse-mediated reversal response [14] — it creates a time separation between the reversal and the subsequent omega turn. At the moment the worm is touched, the signaling cascade starts with the sensory neurons ALM and AVM registering the touch (Fig. 5.1, diagram and inset) [4]. These neurons initiate backward locomotion (through the AVD and AVA command neurons), while inhibiting forward locomotion (through the PVC and AVB command neurons). AVA, in turn, activates the RIM interneurons, which release the neurotransmitter tyramine. Donnelly et al. show that tyramine further inhibits forward locomotion (via AVB), thus prolonging the reversal. Through fast-acting ion channels (LGC-55) in neck muscles and head motor neurons, it additionally suppresses head movements. Crucially, tyramine also diffuses out of the synaptic cleft (‘synaptic spillover’), and activates GPCRs (SER-2) elsewhere in the worm’s body. SER-2 activation by diffusing tyramine sets up an asymmetry in the worm’s locomotory system: it disinhibits the ventral body wall muscles (through suppression of the VD inhibitory motor neuron). After the omega turn has been initiated by a steep ventral head swing (controlled by the SMD neuron [5]), the resulting hypercontraction of the ventral body wall muscles during the body wave produces the characteristic omega turn. The construction and validation of such a description relies on classic methods from biology. Typically, a perturbation is introduced into the worm’s locomotory system, either through genetic knockouts or cell ablations [15]. To assess the impact on the worm’s behavioral phenotype, one 129 5 Deep turns and dynamics of reorientation in C. elegans identifies a behavioral motif affected by the perturbation, and characterizes its temporal or physical appearance. As an example of the latter, part of the role of the SER-2 GPCR was elucidated by linking its absence to a marked decrease in ‘omega angle’, the arc angle between the head and tail during the turn [4]. Worms did not fully ‘close’ their omega turn by touching their head to the body, evidencing a lack of disinhibition of the VD neurons. While detailed, the existing description of the escape response is qualitative. In this section, we take the first steps towards a quantitative description of the escape response. We use the intrinsic coordinate system for C. elegans motor behavior that was introduced in the previous chapter: the low-dimensional space of body postures. Using our new tracking approach, which also resolves self-occluding body shapes, we are, for the first time, able to quantify the full escape response, including the omega turn. We show that the escape response follows stereotyped patterns in posture space, and relate this to the reorientation of the worm during the response. We find that the omega turn produces a tightly-controlled reorientation of 180◦ , but also note that pre- and post-omega phases result in additional orientation changes, broadening the full distribution. Finally, we propose a simple model, in which the worm uses head–body touch as a navigational aid to reorient by 180◦ , away from the stimulus. 5.2.2 Quantifying the escape response The techniques presented in the previous chapter provide a method for quantifying the full postural dynamics of the escape response, including the self-overlapping shapes that are crucial to the reorientation. Through video tracking of the worm’s body postures, and decomposing each posture as a linear combination of postural eigenmodes (or ‘eigenworms’), we obtain a description of the escape response as shown in Fig. 5.2. Here, a1 through a4 represent the first four eigenmodes; at each point in time, the full body posture of the worm can be obtained by adding together the eigenworms 130 5.2 The C. elegans escape response shown as insets, in the relative amounts indicated by the values of ai 1 . The different phases of the qualitative scenario described in the previous section (also shown at the top of Fig. 5.2) correspond to distinct postural features of the time series. For larger datasets, this will allow us to automate the detection of behavioral motifs in the data. What features do we observe? During normal, forward locomotion (segment i, t = 0 . . . 10 s), the worm crawls by propagating a sinusoidal wave through its body. This is reflected in Fig. 5.2 as a pair of phase-locked sinusoidal oscillations in a1 and a2 . In fact, as shown previously [16], both forward and backward crawling correspond to a circular trajectory in the (a1 , a2 ) plane (Fig. 5.3; see also Fig. 4.1f on p. 101). Based on this circular trajectory, a body wave phase angle (ϕ) can be defined as the angle in the (a1 , a2 ) plane. When the worm is stimulated by the infrared laser pulse (mark ii, t = 10 s), it immediately backs up, seen as a reversal of the direction of the (a1 , a2 ) body wave, and thus a decrease in ϕ (iii). An increased ‘floppiness’ of the worm during the reversal, commonly observed in the movie data by eye, shows up as a significant increase in the amplitude of oscillations in a3 and a4 . The omega turn itself (segment iv, gray-shaded area) is preceded by a steep head swing to the ventral side of the body. The resulting highcurvature anterior body bend, observed as peaks in a1 and a4 (red arrows), propagates head-to-tail: this implies another switch of the direction of the body wave, and hence a return to increasing ϕ. The Ω-shaped apex of the turn is marked by a large peak in a3 (red arrow). As the worm exits the turn, the high-curvature body bend reaches the tail, causing a negative peak in a4 (red arrow). Taken together, these quantitative features can be used to define an omega turn: a significant peak in a3 , bounded by the low-curvature body shapes that occur before and after the turn (see 1 More formally, each eigenworm êi is a vector of tangent angles along the ‘backbone’ of the worm, obtained by Principle Component Analysis. The backbone tangent angle P vector at time t is given by θ(t) = i ai êi . See the previous chapter for details. 131 5 Deep turns and dynamics of reorientation in C. elegans (i) (ii)(iii) (iv) (v) 5 a1 0 ‒5 5 a2 0 ‒5 a3 10 0 5 a4 0 ‒5 φ 8π 0 0 5 10 15 Time (s) 20 25 30 Figure 5.2 | Quantification of a typical escape response. Time series for the postural eigenmode decompositions of a tracked worm. a1 to a4 represent the first four postural eigenmodes (also shown as the gray worm shapes in the top-left corners of each graph); ϕ is the body wave phase angle, as defined in the previous chapter (see Fig. 4.1f, p. 101). The infrared laser stimulus takes place at t = 10 s (red line). Phases of the escape response as shown in Fig. 5.1 are shown at the top. Shaded area indicates the omega turn, as defined in the Methods. Red arrows highlight features discussed in the text. 132 5.2 The C. elegans escape response Methods). Finally, as the turn is finished, the worm resumes its forward crawling (v). Interestingly, the increased ‘floppiness’ of the worm (the increased amplitude of the a3 and a4 oscillations) often remains after the omega turn, as is the case in Fig. 5.2. We will come back to this point later. 5.2.3 The escape response in posture space A geometric interpretation of the time series from Fig. 5.2 is shown in Fig. 5.3. Here, we plot the same data as a trajectory in the first three dimensions of posture space. The circular motion in the (a1 , a2 ) plane is clearly visible. At the time of the stimulus, the direction of the body wave circle reverses (iii). During the reversal phase, the previously observed ‘increased floppiness’ now shows up as a distinct tilt of the crawling cycle, into a3 (iii, bottom). While the amount of tilting varies from worm to worm (data not shown), it is a frequently observed feature of the escape response trajectory. Additionally, the radius of the crawling cycle increases, until, at (iv), the omega turn is visible as a large excursion along a3 . The abrupt ‘kink’ in the trajectory, highlighted by the black arrow in (iv), hints at the presence of a neural mechanism that triggers the start of the omega turn. The kink also signals the return to a clockwise progression along the crawling cycle. This geometric picture is potentially powerful, as it clearly visualizes the temporal relationships between the different quantitative features of the escape response. It also suggests a simple decomposition of the omega turn, as a superposition of the body wave attractor cycle and a pulse along the third mode. 5.2.4 Reorientation during the escape response In order to further understand these posture space trajectories, and examine their functional significance for the worm, we considered the reorientation of the worm during the response. As the escape response serves 133 5 Deep turns and dynamics of reorientation in C. elegans (i) (iii) (iv) (v) a3 a1 a2 a3 a1 a2 Figure 5.3 | The escape response in posture space. The same data as in Fig. 5.2 is shown, now as a trajectory through the first three dimensions of posture space. Time is color-coded, flowing in the direction of the arrow from blue to green to yellow to red. The phases of the escape response from Fig. 5.1 are indicated at the top. The bottom row shows a rotated view of the plot in the top row, highlighting the tilt of the body wave cycle during the reversal phase (iii). 134 5.2 The C. elegans escape response to turn the worm away from a perceived threat, back to known-safe territory, one might expect a 180◦ turnaround over the course of the response. Indeed, distributions centered around this value have been found before [4, 6]. Our algorithm provides an easily accessible value for the overall orientation: the average tangent angle of the backbone, hθi. In Fig. 5.4d, we confirm that, on average, the N = 91 tested worms change their overall orientation by 180◦ over the course of the full escape response (−0.9π±0.5π rad, mean ± SD; see also Methods). Our tracking algorithm also makes it uniquely possible to track the overall orientation continuously throughout the different phases of the response. In Fig. 5.4a–c, we calculate how much each of the three phases of the escape response reorients the worm: (a) reversal; (b) omega turn (as defined in the preceding sections; also see Methods); and (c) post-omega forward crawling. The omega turn itself results in a sharp turn of around 180◦ (−0.9π±0.4π rad, mean ± SD). Pre- and post-omega phases also show small but significant contributions (0.1π ± 0.3π rad and −0.1π ± 0.2π rad, respectively), with distributions that are asymmetric. Interestingly, this implies that, while the omega turn is an effective manoeuvre for turning away from the stimulus, the full-response orientation change is broadened by the pre-omega (reversal) and post-omega phases. To see if there was an underlying correlation between the pre- and postomega phases, and hence a possible overarching mechanism controlling the two, we simulated a histogram of full-response orientation changes by adding randomly chosen reorientation values from the three phases (see Methods). If there were significant correlations between the pre- and postomega phases, these would be destroyed by the random sampling process. The resulting distribution is shown in Fig. 5.4d (orange). A Kolmogorov– Smirnov test showed that this simulated distribution was not significantly different from the data (p = 0.14). This implies that, if there are any correlations, they are too weak to be observed here. Therefore, the reorientation effected by the full escape response appears to be a combination 135 5 Deep turns and dynamics of reorientation in C. elegans Probability 0.3 Reversal Omega turn 0.5 (a) 0.4 0.2 Post-omega (b) 0.3 0 ‒2π ‒π 0 π 0 0.1 0.1 0.1 ‒2π ‒π π 0 0 ‒2π ‒π 0 π Orientation change (rad) Orientation change (rad) Orientation change (rad) Orientation change (rad) π (e) π/2 0 ‒π/2 ‒π ‒5 Probability 0.3 0 a3 mean (h) 10 20 30 a3 peak amplitude a3 mean 0 0 π (g) 0 ‒π ‒5 0.4 0 5 0 5 a3 mean (j) 0.3 0.2 0.2 5 ‒π ‒π/2 0.4 0 0 ‒2π Orientation change (rad) π/2 0.6 (i) 0.1 ‒5 0 5 0.2 0 π π (f) π/2 0 ‒π/2 ‒π ‒3π/2 Full response (d) 0.2 0.2 0.2 0.1 0.3 0.3 (c) 0.1 5 10 15 20 25 a3 peak amplitude 0 ‒5 a3 mean Figure 5.4 | Reorientation of the worm occurs during all phases of the escape response. (a–c) Change of the worm’s overall orientation (∆hθi) during the three different phases of the escape response, for each of N = 91 worms (see Methods). (d) The distribution of orientation changes across the full response (blue) is composed of independent contributions from the three phases in a–c. To show this, a simulated histogram has been computed from 10 000 random combinations of orientation changes from a–c, producing a statistically indistinguishable distribution (orange; Kolmogorov–Smirnov test, p = 0.14). (e,g) The mean a3 value, versus the resulting orientation change, during the reversal phase and post-omega phase, respectively. The orientation change is strongly correlated with the mean a3 value. (f) Peak amplitude of the a3 peak corresponding to the omega turn, versus the resulting orientation change. (h,j) Histogram of mean a3 values, during the reversal phase and post-omega phase, respectively, showing similarly asymmetric distributions. (i) Histogram of a3 peak amplitudes during the omega turn. Shown is a reconstituted worm image for an a3 value of 15. 136 5.2 The C. elegans escape response of uncorrelated orientation changes from the reversal, omega-turn, and post-omega phases. From previous work on the interpretation of the postural eigenmodes, we know that the third eigenmode (an overall bending of the worm) is linked to reorientation of the worm [16, 17]. We therefore tested if any asymmetry in the fluctuations of a3 during the reversal phase could be linked to the observed reorientations. Such asymmetry is also visible in Fig. 5.2, as a baseline shift of the third mode during the reversal. As shown in Fig. 5.4e, this is indeed the case: the mean a3 value during the reversal is strongly correlated with the resulting orientation change. The opposite correlation is observed during post-omega forward crawling (g). During the omega turn itself, a weak correlation between a3 peak amplitude and reorientation is observed (f). The actual distributions of mean a3 value and a3 peak amplitude are shown in Fig. 5.4h–j. The sum of these observations allows us to link the behavioral output of the worm to the description of the escape response at the molecular level from Donnelly et al. [4]. As the worm enters the reversal phase, release of tyramine sets up an asymmetry in the worm’s body; this would appear as a baseline shift in the fluctuations of the third mode (Fig. 5.4h). Such a nonzero a3 mean, in turn, has been linked to orientation changes [17] (Fig. 5.4e). After the omega turn, lingering effects of the tyramine may produce a similar baseline shift: the distributions of mean a3 values during the reversal and post-omega phases (Figs. 5.4h,j) are similarly asymmetric. When the worm is moving forward instead of backward, however, this now leads to an opposite orientation change (Fig. 5.4g). We may even speculate as to a reason for the broadening of the orientation change distribution by the pre- and post-omega phases, observed above: this could be an inevitable consequence of the action of tyramine in the build-up towards the omega turn. A reason for the previously observed ‘increased floppiness’ of the worm during the reversal and post-omega phases, visible as an increased amplitude of a3 and a4 oscillations, is less clear. No significant correlations 137 5 Deep turns and dynamics of reorientation in C. elegans to other quantitative properties of the escape response were found in the currently available data (including orientation change). As ‘floppiness’ is usually observed during fast backward locomotion, additional data on reversals may help to elucidate a mechanism for the a3 and a4 oscillations. 5.2.5 A simple model for the escape response The preceding sections have developed a simple picture of the escape response, as a stereotyped trajectory through posture space. The most prominent features are the crawling cycle in the (a1 , a2 ) plane, combined with a distinctive ‘pulse’ along the third eigenmode that corresponds to the omega turn. A particularly striking feature of the escape-response omega turn is how closely the worm controls its reorientation: Fig. 5.4b shows a distribution that is tightly centered around π (180◦ ). This tight control is mirrored in the underlying distribution of a3 peak amplitudes (Fig. 5.4i). The evolutionary benefit of such a behavior seems obvious: it steers the worm away from harm, back to known safety. For a ‘blind’, microscopic organism such as C. elegans, achieving such a feat is not necessarily easy: after all, how does one find their way back, without any visual reference to the outside world? Our data hints at a possible answer for this navigational issue: the worm could use its own body as a ‘guide’ for reorientation. As shown in Fig. 5.4i, the distribution of a3 peak amplitudes lies close to a value of 15: the lowest a3 value that generates a self-touching body shape (inset). A similar observation can be made for the a1 /a4 peaks that precede the a3 peak, and correspond to the deep ventral head swing (data not shown). This suggests that the worm might coil up until it hits its own body, which it then slides along to find its way back. Lacking a neural mechanism for such a ‘body-touch assisted turn’, this simple model remains of course speculative. Forward genetic screens [18] could provide a way forward for testing this hypothesis, and for identifying further neural and molecular substrates for this behavior. Comparison of the escape response across different nematode species [19] may also help validate our proposed model. 138 5.3 C. elegans free crawling behavior 5.3 C. elegans free crawling behavior 5.3.1 Introduction to free crawling behavior In the first half of this chapter, we used a well-defined stimulus to elicit a stereotyped response from C. elegans. While this approach is excellent for teasing apart specific neural circuits within the worm’s nervous system, it does not necessarily shed light on its full range of spontaneous behavior. On the other side of the spectrum of behavioral assays, we can therefore let the organism more naturally explore a more unrestricted space of behavior [20]. An example of such an experiment is the ‘free crawling assay’. When crawling freely on a 2D agar surface, C. elegans shows a mixture of behavioral motifs: bouts of forward crawling are alternated with pauses, reversals, and sharp turns (including the previously encountered omega turn). These motifs are combined into a navigational strategy for locating hospitable habitats, food, and mating partners. Multiple types of sensory cues feed into this strategy: the worm responds, for example, to attractant and repellent odorant molecules (where the type of response is encoded by which of the sensory neurons expresses the molecule’s receptor protein [15]). The worm also actively seeks out a preferred salt concentration [21, 22], temperature [23, 24], and oxygen concentration [25], with fascinating evidence of adaptation [26] and associative learning [23, 27, 28]. So how does the worm actually reach, say, the peak of an attractant chemical gradient? Interestingly, C. elegans employs two parallel mechanisms for this [29]. The first mechanism, klinokinesis, is stochastic: the worm executes a biased random walk up the gradient. Forward crawls are interspersed with so-called pirouettes: bursts of reversal and turning events that reorient the worm towards the gradient [30]. The frequency of pirouette initiation is strongly dependent on the rate of change of attractant concentration (dC/ dt) detected by the worm. If the worm is running up the gradient, pirouettes are suppressed; if, on the other hand, it detects a decrease, the probability of a pirouette increases strongly [30]. The pirouette is similar 139 5 Deep turns and dynamics of reorientation in C. elegans to the bacterial tumble, used by bacteria such as E. coli to locate food (chemotaxis) [31, 32] — albeit more efficient, as the pirouette is not a purely random reorientation [30]. The second mechanism, klinotaxis, is a deterministic strategy. Also called weathervaning in C. elegans, this is a gradual curving of the worm’s trajectory towards the gradient [33]. In this case, the worm samples the local attractant concentration during its normal undulatory head movements, using the sensory neurons in its head. It then uses this information to bias the angle of the head during locomotion [33, 29]. Analysis of C. elegans taxis behavior is historically based on tracking of the worm’s center-of-mass (or centroid) as it crawls around the agar plate [30]; later generations of experiments have added increasingly sophisticated mechanisms for also tracking the worm’s body shape [34–40]. This has provided us with our current understanding of the taxis and kinesis mechanisms, but also leaves some questions unanswered. How does the pirouette reorient the worm up the gradient? And what exactly constitutes a pirouette? Our tracking algorithm now makes it possible to begin filling in those gaps: we can precisely quantify the behavioral dynamics of a freely crawling worm, even during manoeuvres that feature self-overlapping body shapes. In the previous chapter, we showed that omega turns can actually be classified into two different classes, based on their amplitude (as measured by the third postural eigenmode). In this section, we further explore these two classes, and their possible role in C. elegans navigation. 5.3.2 Two types of ventral turns can be discerned during free crawling We previously tracked 12 worms as they crawled freely on a 2D agar surface, each for the duration of 35 minutes (see Methods). As no stimuli or attractant/repellent gradients were present, this represents a basic experiment for studying C. elegans free crawling behavior. We applied our 140 5.3 C. elegans free crawling behavior 0.1 (a) 0.1 0.08 Probability Probability 0.08 0.06 0.04 0.06 0.04 0.02 0.02 0 0 −30 (b) −20 −10 0 10 a3 local extremum 20 30 0 5 Ω δ 10 15 20 25 a3 local extremum 30 Figure 5.5 | C. elegans turns during free crawling can be differentiated into two separate, ventrally-biased classes. (a) Histogram of the amplitude of local extrema in the time series of the third postural eigenmode a3 (for N = 12 freely crawling worms). Colors represent the sign of the a3 amplitude, and hence the dorsal (gray) or ventral (blue) direction of the resulting turn. Insets show reconstructed worm shapes for the indicated a3 amplitudes. (b) As a, but with all negative a3 amplitudes plotted as positive. Two classes of turns show up on the ventral side, corresponding to ‘classic’ Ω (omega) turns, and deep δ (delta) turns. tracking algorithm to quantify the body postures of each worm. As our primary interest was the ‘omega turn’, we focused on the third postural eigenmode (a3 ). As mentioned in the previous section on the escape response, peaks in the third mode are a characteristic feature of omega turns, and have a known role in reorientation of the worm [17]. In our earlier analysis of the same data (see previous chapter), we only considered a3 amplitudes for turns during which the worm self-overlapped. In Fig. 5.5, we instead look at the full distribution of a3 values for all local extrema in the data (see Methods). In this distribution, negative a3 amplitudes correspond to dorsal turns; ventral turns have strictly positive amplitudes. In Fig. 5.5a, a clear asymmetry can be observed. On top of a symmetric background distribution of shallow turns in both directions, we see, on the ventral side, two distinct additional peaks in the histogram. Drawing 141 5 Deep turns and dynamics of reorientation in C. elegans Ω (omega) δ (delta) t Figure 5.6 | Examples of the two types of turns identified in the data. Stills from a movie of a worm making a classical omega turn (top, yellow), and a deep ‘delta’ turn (bottom, blue). The head is marked with a red dot in the first frames. reconstituted worm images for the center values of these two peaks, we see that the peak at a3 ∼ 15 corresponds to a ‘classic’ Ω shape. The second peak, at a3 ∼ 23, shows a body shape with a much higher curvature. In Fig. 5.5b, we have ‘folded’ the dorsal side of the distribution over the ventral side, highlighting the ventral asymmetry at high a3 amplitudes. As indicated in Fig. 5.5b, we will refer to turns in the lower-amplitude peak as omega turns. We will distinguish these from the higher-amplitude delta (δ) turns in the second peak. (As for the omega turn, the name delta turn is chosen to reflect the δ-like shape of the worm during a typical delta turn.) Returning to the original tracking movies, the presence of these two classes of turns is clearly visible. In Fig. 5.6, we show movie stills for two example turns: one omega turn, and one delta turn. During the omega turn, the worm slides its head along its body, ending up with a roughly 180◦ reorientation — similar to the escape-response omega turn. A delta turn, on the other hand, is much deeper: the worm completely crosses its head over its body, resulting in an ‘over-turn’ of greater than 180◦ . 142 5.3 C. elegans free crawling behavior 5.3.3 Despite disparate appearance, delta- and omega-turn kinematics are similar Omega turns and delta turns are visually distinct, but how is this difference reflected in their posture space trajectories? Most notable is the higher a3 peak amplitude of delta turns; but are they otherwise different? Taking one particular example, shown in Fig. 5.7a, suggests that the other aspects of the delta turn’s kinematics are actually very similar to those of the omega turn (cf. Fig. 5.2). Just like the omega turn, the delta turn starts with a deep ventral head swing, and corresponding peaks in a1 and a4 . This is followed by the peak in a3 , marking the apex of the turn, and a negative peak in a4 for the exit of the high-curvature body wave through the tail. Averaging the eigenmode time series across N = 348 delta turns (grayshaded area in Fig 5.7a; see Methods), we obtain the average response shown in Fig. 5.7b. Comparing this with the average omega turn for the escape response dataset from the previous section (red dashed line), we see that a3 amplitude is indeed the only feature that clearly separates the delta turn from the omega turn. 5.3.4 Delta and omega turns as part of the worm’s navigational strategy So far, we have seen that the worm makes, in addition to shallow turns in both the ventral and dorsal direction, two distinct types of steep ventral turns. These omega and delta turns appear to be different only in their a3 pulse amplitude; turn kinematics are otherwise very similar. Could they perhaps play specific roles in the worm’s navigational strategy, and could this be linked to the difference in a3 amplitude? We first plotted where both types of turns occur in the worm’s trajectory on the agar plate. In Fig. 5.8a,b, we show such a trajectory for one of the twelve worms in the experiment, with omega (yellow/light) and delta (blue/dark) turns marked. The worm starts out at position (0, 0) (black arrow), and shows typical ‘dwelling’ behavior: a high turning fre- 143 5 Deep turns and dynamics of reorientation in C. elegans (a) (b) 10 10 a1 0 a1 0 −10 10 −10 10 a2 0 a2 0 −10 −10 20 a3 10 0 20 a3 10 0 5 a4 0 −5 5 a4 0 −5 0 φ −π −3π 5 6 7 8 0.2 0.4 0.6 Normalized time 0.8 9 10 11 12 13 14 Time (s) Figure 5.7 | Kinematics of the delta turn are similar to those of the omega turn. (a) Typical time series for the postural eigenmodes a1..4 during a deep ‘delta’ turn; ϕ is the body wave phase angle, as defined in Fig. 4.1f (p. 101). Shaded area indicates the delta turn, as defined in the Methods. (b) Average eigenmode time series during a delta turn (blue, N = 348). Gray lines indicate SD. For comparison, the average escape response omega turn is also shown (red dotted line). Time has been normalized with respect to the total length of the turn: 6 ± 2 s (mean ± SD) for delta turns, 7 ± 3 s for escape-response omega turns. 144 1 5.3 C. elegans free crawling behavior (a) (b) Ω δ −14 Y (mm) Y (mm) 10 0 −16 −10 (c) 10 X (mm) 20 8 (d) 0° ve nt 15 12.5 90° d l sa or l ra 270° 180° Mean turning rate (min−1) 0 10 X (mm) 1.5 12 Ω δ 1.25 1.0 0.75 0.5 0.25 10 20 Time (min) 30 Figure 5.8 | The omega and delta turns as part of the navigational strategy of C. elegans. (a,b) Location on the agar plate of one of the 12 tracked worms over the course of a 35-minute tracking experiment (off-food), starting at (0, 0) (black arrow). Ventral omega turns (yellow/light) and ventral delta turns (blue/dark) are highlighted. A blow-up of the area marked in gray is shown in b. (c) Histogram of orientation change due to ventral omega turns (yellow/light) and ventral delta turns (blue/dark). Omega turns dominantly lead to a re-orientation change towards the ventral side of the animal, whereas deep delta turns cause the animal to ‘over-turn’ into the dorsal side of the compass. (d) Average turning rate (across N = 12 worms) during the 35 minutes of the tracking experiment. Ventral omega and delta turns are temporally independent (see text), but occur with approximately equal rate, and show co-adaptation as the worm switches from a ‘dwelling’ to a ‘roaming’ strategy. 145 5 Deep turns and dynamics of reorientation in C. elegans quency, and an exploration of its local environment [5]. After dwelling for some minutes without finding any food, it gradually switches to ‘roaming’ behavior, traversing a wider area of the agar plate [5, 41]. During the full course of the experiment, the two types of turns are statistically independent. Apart from an overall modulation of the turn frequency (discussed later), no temporal correlations are observed. We checked this by computing the mutual information between time-binned, time-shifted time series for both turns (see Methods). With a maximum mutual information less than a few percent of the maximum entropy, the turns have to be considered independent. While independent, the turns result in a different change of orientation — following the known link between the third mode and orientation changes [17]. In Fig. 5.8c, we show how the worm reorients using both omega (yellow/light) and delta (blue/dark) turns. The free-crawling omega turns show a much broader distribution than the previously encountered escape-response ones (cf. Fig. 5.4d). Still, they generally reorient the worm ventrally. In contrast, the deeper delta turns consist of the worm ‘over-turning’ and ending up reorienting towards its dorsal side. The difference in reorientation angle may provide a hint as to why these two distinct turns exist. In the first part of this chapter, we saw that the neural mechanisms that produce the escape-response omega turn are fundamentally asymmetric, producing only ventral turns (through disinhibition of the VD motor neurons) [4]. If the worm uses the same neural infrastructure during free crawling, this would only ever allow it to reorient itself towards its ventral side. Lacking a dorsal ‘copy’ of the same neural infrastructure, the worm could instead hyper-activate the existing infrastructure to produce ventral over-turning’. These over-turns are what we call delta turns, and enable the worm to also reorient towards its dorsal side. Evidence that both turns share at least part of the worm’s neuronal control mechanisms is shown in Fig. 5.8d. Here, we plot the frequency of turning events over the course of the experiment (see Methods). As the 146 5.3 C. elegans free crawling behavior worm switches from ‘dwelling’ to ‘roaming’, in search for food in a larger area, the turn frequency decreases significantly — a well-known adaptation [5, 28, 41]. Interestingly, both types of turns show similar frequencies, and co-adaptation over time. This suggests that both types of turns are regulated by a shared underlying mechanism. One possible realization of such a mechanism is the known regulation of ‘dwelling’ behavior by the dopamine neurotransmitter, which modulates motion-related glutaminergic signaling pathways in the worm’s nervous system [28]. We expect that our observation of the existence of both omega and delta turns will be highly relevant for future investigations of C. elegans ‘pirouettes’. The definition of what exactly constitutes a pirouette has varied from publication to publication [5, 33, 38, 30], but always features a series of one or more omega turns as the course-adjusting motif. We have now shown that these turns are, in fact, two distinct turns, each producing a different reorientation. How the two turns are combined to produce the known reorientation towards an attractant gradient [30], is an interesting avenue for further investigation. 5.3.5 Conclusions and outlook In this chapter, we have taken the first steps towards opening up the ‘black box’ of self-overlapping body shapes that occur during navigation in C. elegans. During the C. elegans escape response, the signature omega turn is the central behavioral motif that steers the worm away from the noxious stimulus. The turn contributes around 180◦ of the reorientation observed across the response, and appears to be highly stereotyped, with a tightly controlled amplitude. The characteristics of the escape response can be linked to previously found mechanisms at the neural and molecular level. A tyramine-mediated build-up of asymmetry in the worm’s locomotory system is quantitatively visible as a baseline shift in the oscillations of the third postural eigenmode a3 . This occurs during the reversal phase, and lingers after the omega turn has completed, resulting in a broadening of 147 5 Deep turns and dynamics of reorientation in C. elegans the distribution of orientation changes across the full response. Finally, we have proposed a simple hypothesis that the worm, lacking visual cues, uses its own body to ‘guide’ itself backwards. In free crawling, a distinction can be made between two separate populations of deep turns: omega turns, featuring the classic Ω shape, and a novel class of deep turns, which we here call ‘delta’ (δ) turns. We show that, despite their distinct visual appearance, the posture space trajectories of delta and omega turns are very similar. Their only clear distinguishing feature is the amplitude of the pulse in the third postural eigenmode a3 , which seems to drive a difference in resulting reorientation. Taken together, this generates a picture of omega and delta turns as a way for the worm to make sharp turns in the ventral and dorsal direction, respectively. The turns are temporally uncorrelated, although their overall frequency appears to be controlled by a shared mechanism. Our first glance into the ‘black box’ of self-overlapping body shapes of C. elegans is of course just that: a first glance. Some important questions remain, as yet, unanswered. For example, posture space trajectories for sharp turns, while geometrically similar, still show significant variability. It remains unclear what the source of this inter- and intra-individual variance is. To answer such questions, a principled way of quantifying differences between trajectories through posture space, and relating those differences to behavioral output, would be needed. Our tracking algorithm could also provide a way forward for elucidating the neural and molecular pathways that drive turning behavior in C. elegans. Behavioral defects induced by genetic mutations or cell ablations are typically detected manually [5, 4]; automated tracking would allow for increased throughput, and detection of more subtle defects than might be detected by eye. Finally, work on other nematodes has shown interesting differences in the way these species modulate their navigational strategies, in this case between a ‘roaming’ and a ‘dwelling’ strategy [19]. Whether omega and delta turns are conserved across species, and contribute similarly to es- 148 5.4 Methods Molecule Molecular weight (g/mol) Free diffusion constant (10−10 m2 /s) Source norepinephrine dopamine leucine threonine 169 153 131 119 5.5 6.0 6.0 7.5 [42] [42] [43] [44] Table 5.1 | Diffusion constants of some small molecules (neurotransmitters and amino acids) that are similar in molecular weight to tyramine (MW = 137 g/mol). cape response and foraging behavior, could be an interesting future line of investigation. 5.4 Methods Tyramine diffusion. During the escape response, extra-synaptic diffusion of tyramine is thought to temporally separate the initial reversal of the worm and the omega turn (see ‘Introduction to the escape response’, p. 126) [4]. We can make a rough, back-of-an-envelope estimate of the time needed for tyramine to diffuse, and show that this is consistent with the reversal phase duration typically observed in our experiments. To our knowledge, no diffusion constant has been published for tyramine. Instead, we can estimate the relevant order of magnitude by considering other molecules with a similar molecular mass. From the data in Table 5.1, we can derive a conservative estimate of the tyramine diffusion constant of D ∼ 5 · 10−10 m2 /s. Using the diffusion equation, h(∆x)2 i = 2Dt, for a length scale that we estimate to be a tenth of the worm’s full-grown length of 1 mm [3] (since the distance between the RIM and VD neurons is far less than 1 mm), we arrive at a typical timescale of t ∼ (10−4 m2 )2 /(2 · 5 · 10−10 m2 /s) ∼ 10 s. This is consistent with the typical reversal time of 5 s observed in our escape response data. 149 5 Deep turns and dynamics of reorientation in C. elegans Data collection. Two datasets were used in this chapter: one with C. elegans escape responses evoked by a thermal stimulus; and one with free crawling assays. The collection of both of these datasets, including the postural tracking, was described in the Methods of the previous chapter (see p. 115). As C. elegans crawls on its side [5], worms can have one of two possible orientations during experiments. Dorsal/ventral orientation was recorded during the experiments by noting the position of the vulva, and results for worms with a right-handed orientation were mirrored. Surprisingly, we found two worms in the free-crawling dataset that changed their dorsal/ventral orientation during the experiment. Both worms showed the asymmetry in a3 peak amplitude that was noted in Fig. 5.5, but this asymmetry switched sign half-way during the recording. Inspecting the movies, we found that the time of switching corresponded to moments when the worms moved off or into the agar, escaping the 2D constraints of the experiment. This was corrected for by mirroring tracking results for only the relevant part of the experiment. Reorientation during escape response. For the analysis of the worm’s reorientation during the escape response (Fig. 5.4), N = 91 escape responses were analyzed. Each 30-second recording was segmented by first finding the omega turn, as described below. After identification of the omega turn, the reversal phase was simply defined as the first frame after the stimulus with a negative body wave phase velocity dϕ/ dt, up until the start of the omega turn. The ‘post-omega’ phase was any data after the end of the omega turn until the end of the recording at t = 30 s. To generate the simulated histogram for the orientation change of the full response, we picked a random worm for the reversal phase, a random worm for the omega turn (possibly the same), etc. The orientation changes of these three phases were summed, and added to the simulated distribution. If any of the chosen recordings did not have a successfully detected omega turn, this was skipped. A total of 106 iterations were performed. Definition of omega and delta turn. For the escape response data, the largest peak in a3 between t = 10 s (the time of the stimulus) and t = 29 s was identified as the apex of the omega turn. To locate the end of the omega turn, the first root (zero) of a4 after the apex was found; any point after that root that had a3 < 3 was considered to be the end of the omega turn. This ensured that the negative peak in a4 , representing a high-curvature state of the 150 5.4 Methods tail at the end of the omega turn, had finished, and that the worm had reached a relatively ‘straight’ shape. For such straight shapes, the overall orientation hθi has a straightforward, intuitive interpretation. The same criterion was used, in the opposite direction, to find the start of the omega turn. If no starting point and/or end point of the omega turn could be found, the recording was excluded from the analysis. (In the escape response dataset, this was the case for 15 out of 91 recordings). In the free crawling dataset, the same quantitative criterion was used to find both omega and delta turns. Local extrema in a3 . For the free crawling dataset, we analyzed the amplitudes of local minima and maxima in the third postural eigenmode a3 (Fig. 5.5). As the tracking was originally performed on a segmented version of this data, tracking data first had to be ‘stitched together’ again. Segmentation of the 33 600-frame movies was described in the ‘Methods’ of the previous chapter. Briefly, each segment of a movie was chosen such, that it contained a consecutive series of non-crossed frames, followed by a consecutive series of crossed frames (a turn), followed by another series of non-crossed frames. The last series of non-crossed frames in segment j overlapped with the first series of non-crossed frames in segment j +1. This facilitated the stitching process: two segments could be joined together at a frame that occurred in both segments j and j + 1, and that had the exact same tracking solution in both segments within a given margin of error (0.1). Tracking errors could cause violations of this margin of error. All such stitching problems were manually inspected; if a tracking error was found, that segment of the data was excluded from the analysis. In total, across all twelve worms, 878 out of 936 segments (94%) produced tracking results of sufficient quality. For detection of local extrema in a3 , a standard peak-finding algorithm was used to detect both minima and maxima (based on MATLAB’s findpeaks function, which defines a peak as a data point with a greater value than its immediate neighbors). Only extrema with a minimum prominence of 0.5 were kept. Some a3 peaks featured smaller sub-peaks in their shoulders; such sub-peaks were discarded. Average delta-turn eigenmode time series. To generate Fig. 5.7b, delta turns as defined above were cut from the free crawling dataset (N = 348). Each five-dimensional time series (modes a1...5 ) for each delta turn was mapped onto 151 5 Deep turns and dynamics of reorientation in C. elegans a 50-point, linearly-spaced ‘normalized time’ axis (tN ∈ [0, 1]), using linear interpolation of the time series where necessary. This normalized time corresponded to an actual duration of the delta turns of 6 ± 2 s (mean ± SD). Data across delta turns was then averaged per normalized-time point, giving the trajectories shown in the Figure. The same procedure was followed for omega turns in the escape response dataset (N = 91). For this data, the normalized time axis corresponds to an actual duration of the omega turns of 7 ± 3 s (mean ± SD). Mutual information between omega and delta turns. For calculating the mutual information between the omega and delta turns during free crawling, we followed the procedure from ref. 45. We created binarized time series for each type of turn, by binning turning events into bins of 2, 4, 10, or 20 s. The mutual information was calculated for different time shifts, ranging from −60 to +60 s. Mutual information across time shifts never exceeded 3% of the maximum entropy of each time series, hence precluding a significant correlation between the two types of turns. Turn frequency adaptation. In Fig. 5.8d, we show how the average turn frequencies for omega and delta turns change over the course of the 35-minute free-crawling experiments. Turns were detected in the stitched data by using the peak detection algorithm outlined above (see ‘Local extrema in a3’ before). Using the boundaries identified in Fig. 5.5b, a3 extrema with an absolute value between 10 and 20 were classified as ‘omega turns’, while extrema with an absolute value greater than 20 were considered to be ‘delta turns’. We also distinguished between ventral turns, with a positive amplitude, and dorsal turns, with a negative amplitude. For the Figure, we counted the average number of turns per unit time, across the 12 experiments, in a 10-minute sliding window, shifted across the data in 5-minute increments. The first 200 seconds of each experiment were discarded, as the worms showed signs of adaptation to their new environment (the agar plate without food): the average turn frequency, somewhat erratically, increased during this period. The population of ‘omega turns’ thus found consists, as can be seen in Fig. 5.5b, of two sub-populations: a tail of the symmetric distribution of ‘shallow turns’, and the actual population of ventrally-biased omega turns. We therefore counted the number of a3 peaks with an amplitude between −20 and −10 in each time 152 5.5 References window, and subtracted this from the total number of a3 peaks with an amplitude between +10 and +20. This gave us the number of ‘true’, ventrally-biased omega turns. This number showed excellent agreement with the number of delta turns (Fig. 5.8d). 5.5 References [1] J. G. White, E. Southgate, J. N. Thomson, and S. Brenner. The Structure of the Nervous System of the Nematode Caenorhabditis elegans. Philosophical Transactions of the Royal Society B: Biological Sciences, 314(1165):1–340, 1986. [2] B. Prevo. Kinesin Illuminated. PhD thesis, VU University Amsterdam, 2015. [3] Z. Altun and D. Hall. Nervous system, general description. In WormAtlas. 2011. [4] J. L. Donnelly, C. M. Clark, A. M. Leifer, J. K. Pirri, M. Haburcak, M. M. Francis, A. D. T. Samuel, and M. J. Alkema. Monoaminergic orchestration of motor programs in a complex C. elegans behavior. PLoS Biology, 11(4):e1001529, 2013. [5] J. M. Gray, J. J. Hill, and C. I. Bargmann. A circuit for navigation in Caenorhabditis elegans. Proceedings of the National Academy of Sciences of the United States of America, 102(9):3184–91, 2005. [6] A. Mohammadi, J. Byrne Rodgers, I. Kotera, and W. S. Ryu. Behavioral response of Caenorhabditis elegans to localized thermal stimuli. BMC Neuroscience, 14(1):66, 2013. [7] S. M. Maguire, C. M. Clark, J. Nunnari, J. K. Pirri, and M. J. Alkema. The C. elegans Touch Response Facilitates Escape from Predacious Fungi. Current Biology, 21(15):1326–1330, 2011. [8] T. Boulin and O. Hobert. From genes to function: The C. elegans genetic toolbox. Wiley Interdisciplinary Reviews: Developmental Biology, 1(1):114–137, 2012. [9] J. Sulston, E. Schierenberg, J. White, and J. Thomson. The embryonic cell lineage of the nematode Caenorhabditis elegans. Developmental Biology, 100(1):64–119, 1983. [10] D. M. Tobin and C. I. Bargmann. Invertebrate nociception: Behaviors, neurons and molecules. Journal of Neurobiology, 61(1):161–174, 2004. [11] J. E. Mellem, P. J. Brockie, Y. Zheng, D. M. Madsen, and A. V. Maricq. Decoding of polymodal sensory stimuli by postsynaptic glutamate receptors in C. elegans. Neuron, 36(5):933–944, 2002. [12] M. Chalfie, J. E. Sulston, J. G. White, E. Southgate, J. N. Thomson, and S. Brenner. The neural circuit for touch sensitivity in Caenorhabditis elegans. Journal of Neuroscience, 5(4):956–964, 1985. [13] J. K. Pirri, A. D. McPherson, J. L. Donnelly, M. M. Francis, and M. J. Alkema. 153 5 Deep turns and dynamics of reorientation in C. elegans A Tyramine-Gated Chloride Channel Coordinates Distinct Motor Programs of a Caenorhabditis elegans Escape Response. Neuron, 62:526–538, 2009. [14] T. H. Lindsay, T. R. Thiele, and S. R. Lockery. Optogenetic analysis of synaptic transmission in the central nervous system of the nematode Caenorhabditis elegans. Nature communications, 2(May):306, 2011. [15] E. R. Troemel, B. E. Kimmel, and C. I. Bargmann. Reprogramming chemotaxis responses: Sensory neurons define olfactory preferences in C. elegans. Cell, 91(2):161–169, 1997. [16] G. J. Stephens, B. Johnson-Kerner, W. Bialek, and W. S. Ryu. Dimensionality and dynamics in the behavior of C. elegans. PLoS Computational Biology, 4(4):e1000028, 2008. [17] G. J. Stephens, B. Johnson-Kerner, W. Bialek, and W. S. Ryu. From modes to movement in the behavior of Caenorhabditis elegans. PLoS One, 5(11):e13914, 2010. [18] E. M. Jorgensen and S. E. Mango. The art and design of genetic screens: Caenorhabditis elegans. Nature Reviews Genetics, 3(5):356–369, 2002. [19] S. J. Helms, L. Avery, G. J. Stephens, and T. S. Shimizu. Modeling the ballistic-todiffusive transition in nematode motility reveals low-dimensional behavioral variation across species. 2015, arXiv:1501.00481. [20] G. J. Stephens, L. C. Osborne, and W. Bialek. Searching for simplicity in the analysis of neurons and behavior. Proceedings of the National Academy of Sciences of the United States of America, 108 Suppl:15565–71, 2011. [21] S. Ward. Chemotaxis by the nematode Caenorhabditis elegans: identification of attractants and analysis of the response by use of mutants. Proceedings of the National Academy of Sciences of the United States of America, 70(3):817–821, 1973. [22] C. I. Bargmann and H. R. Horvitz. Chemosensory neurons with overlapping functions direct chemotaxis to multiple chemicals in C. elegans. Neuron, 7(5):729–42, 1991. [23] E. M. Hedgecock and R. L. Russell. Normal and mutant thermotaxis in the nematode Caenorhabditis elegans. Proceedings of the National Academy of Sciences of the United States of America, 72(10):4061–4065, 1975. [24] D. A. Clark, C. V. Gabel, H. Gabel, and A. D. T. Samuel. Temporal activity patterns in thermosensory neurons of freely moving Caenorhabditis elegans encode spatial thermal gradients. Journal of Neuroscience, 27(23):6083–90, 2007. [25] J. M. Gray, D. S. Karow, H. Lu, A. J. Chang, J. S. Chang, R. E. Ellis, M. A. Marletta, and C. I. Bargmann. Oxygen sensation and social feeding mediated by a C. elegans guanylate cyclase homologue. Nature, 430(6997):317–322, 2004. [26] A. J. Chang, N. Chronis, D. S. Karow, M. A. Marletta, and C. I. Bargmann. A 154 5.5 References [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] distributed chemosensory circuit for oxygen preference in C. elegans. PLoS Biology, 4(9):e274, 2006. M. Gomez, E. De Castro, E. Guarin, H. Sasakura, A. Kuhara, I. Mori, T. Bartfai, C. I. Bargmann, and P. Nef. Ca2+ signaling via the neuronal calcium sensor-1 regulates associative learning and memory in C. elegans. Neuron, 30(1):241–248, 2001. M. de Bono and A. Villu Maricq. Neuronal Substrates of Complex Behaviors in C. elegans. Annual Review of Neuroscience, 28(1):451–501, 2005. S. R. Lockery. The computational worm: spatial orientation and its neuronal basis in C. elegans. Current Opinion in Neurobiology, 21(5):782–90, 2011. J. T. Pierce-Shimomura, T. M. Morse, and S. R. Lockery. The fundamental role of pirouettes in Caenorhabditis elegans chemotaxis. Journal of Neuroscience, 19(21):9557–69, 1999. E. F. Keller and L. A. Segel. Model for chemotaxis. Journal of Theoretical Biology, 30(2):225–34, 1971. H. C. Berg and D. A. Brown. Chemotaxis in Escherichia coli analysed by threedimensional tracking. Nature, 239(5374):500–504, 1972. Y. Iino and K. Yoshida. Parallel use of two behavioral mechanisms for chemotaxis in Caenorhabditis elegans. Journal of Neuroscience, 29(17):5370–80, 2009. J. H. Baek, P. Cosman, Z. Feng, J. Silver, and W. R. Schafer. Using machine vision to analyze and classify Caenorhabditis elegans behavioral phenotypes quantitatively. Journal of Neuroscience Methods, 118(1):9–21, 2002. Z. Feng, C. J. Cronin, J. H. Wittig, P. W. Sternberg, and W. R. Schafer. An imaging system for standardized quantitative analysis of C. elegans behavior. BMC Bioinformatics, 5(1):115, 2004. C. J. Cronin, J. E. Mendel, S. Mukhtar, Y.-M. Kim, R. C. Stirbl, J. Bruck, and P. W. Sternberg. An automated system for measuring parameters of nematode sinusoidal movement. BMC Genetics, 6:5, 2005. E. Fontaine, J. Burdick, and A. Barr. Automated tracking of multiple C. elegans. In Engineering in Medicine and Biology Society, 2006. EMBS ’06. 28th Annual International Conference of the IEEE, volume 2, pages 3716–3719, 2006. L. C. M. Salvador, F. Bartumeus, S. A. Levin, and W. S. Ryu. Mechanistic analysis of the search behaviour of Caenorhabditis elegans. Journal of the Royal Society Interface, 11(92):20131092, 2014. N. Roussel, J. Sprenger, S. J. Tappan, and J. R. Glaser. Robust tracking and quantification of C. elegans body shape and locomotion through coiling, entanglement, and omega bends. Worm, 3(4):e982437, 2014. S. Nagy, M. Goessling, Y. Amit, and D. Biron. A generative statistical algorithm for automatic detection of complex postures. PLoS Computational Biology, 2015 (in press). 155 5 Deep turns and dynamics of reorientation in C. elegans [41] N. Srivastava, D. A. Clark, and A. D. T. Samuel. Temporal analysis of stochastic turning behavior of swimming C. elegans. Journal of Neurophysiology, 102(2):1172– 9, 2009. [42] G. Gerhardt and R. N. Adams. Determination of diffusion coefficients by flow injection analysis. Analytical Chemistry, 54(14):2618–2620, 1982. [43] A. Polson. On the diffusion constants of the amino-acids. Biochemical Journal, 31(10):1903–1912, 1937. [44] Y. Ma, C. Zhu, P. Ma, and K. T. Yu. Studies on the Diffusion Coefficients of Amino Acids in Aqueous Solutions. Journal of Chemical & Engineering Data, 50(4):1192–1196, 2005. [45] S. Strong, R. Koberle, R. de Ruyter van Steveninck, and W. Bialek. Entropy and Information in Neural Spike Trains. Physical Review Letters, 80(1):197–200, 1998. 156
© Copyright 2026 Paperzz