By Aylia Mohammadi a Thesis Submitted in Conformity with the Requirements for the Degree of Doctor of Philosophy Graduate Depart

Home , Grimace scale (animals)

QUANTITATIVE BEHAVIORAL ANALYSIS OF THERMAL NOCICEPTION IN Caenorhabditis elegans: INVESTIGATION OF NEURAL SUBSTRATES SPATIALLY MEDIATING THE NOXIOUS RESPONSE, AND THE EFFECTS OF PHARMACOLOGICAL PERTURBATIONS.

Aylia Mohammadi

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Physics University of Toronto

Quantitative behavioral analysis of thermal nociception in Caenorhabditis elegans: investigation of neural substrates spatially mediating the noxious response, and the effects of pharmacological perturbations.

Aylia Mohammadi Doctor of Philosophy Graduate Department of Physics University of Toronto 2013

The nematode Caenorhabditis elegans possesses a relatively simple nervous system of only 302 neurons, but is able to perform an impressive range of complex behaviors. This dissertation aims to understand the neurobiology of behavior by quantifying, at the systems-level, the sensorimotor response to carefully controlled stimuli. Through neuronal or genetic perturbations to the system, we can begin to uncouple the behavior from the underlying circuitry. The behavior studied here is thermal nociception, an escape response designed to protect an organism from potential tissue damage or harm from noxious heat. Vertebrates and invertebrates alike possess sensory neurons called nociceptors that detect noxious stimuli and relay the stimulus information to elicit an appropriate escape response. C. elegans is known to perform a reversal or forward response when presented with noxious stimuli at the head or tail, respectively. In this work, we develop a novel thermal stimulus assay with precise spatiotemporal control of an infrared pulse that targets small regions along the worm to spatially dissect the noxious response. We comprehensively quantify the nociceptive behavior, and identify key metrics that scale with intensity, such as speed in the escape state and the probability of certain behavioral states after the stimulus. Furthermore, we have mapped the behavioral receptive ﬁeld of the worm along its body, and show a previously unreported probabilistic midbody behavior distinct from the head and tail responses. Surprisingly, the worm is able to differentiate localized

ii stimuli at the midbody that are as close as 80 microns. We identified PVD as the thermal nociceptor for the midbody response using calcium imaging, genetic ablation and laser ablation. This suggests PVD could be used as a model to study spatial discrimination at the level of a single nociceptor. This spatial specificity further extends to pharmacological perturbations of the system. In particular, the application of clinically used painkillers to the worm results in a knockdown of this nociceptive response, but does so in a spatially specific manner. These results are promising for future studies building upon the techniques developed here, as they evidentiate the use of C. elegans as a model organism to study pain.

iii Acknowledgements

First and foremost, I would like to thank my supervisor, William Ryu. Will’s breadth of knowledge and dedication to science truly inspires. Will gave me enough freedom to start this project from scratch and take it to new, unexpected, exciting places that really motivated and engaged me as a scientist. Not only did Will teach me a lot regarding the technical aspects of this work, but I also learned invaluable skills as an experimentalist in general. Will showed me the satisfaction that comes with a beautifully designed, simple experiment to test a hypothesis, and that if we let our curiosity guide us, we can learn new things when things don’t make sense at ﬁrst.

I would also like to thank the members of my committee, Asher Cutter and Anton Zilman, for their support during this process.

I would also like to express my heartfelt gratitude to all the members of the Ryu lab: your friendship has made coming to the lab every day genuinely enjoyable. I know you all will continue to do great work, wherever life takes you. In particular, I would like to thank my co-authors, Jarlath Byrne Rodgers and Ippei Kotera, without whom this work would not have been possible.

On a personal note, there are some very important people who have been instrumental in my accomplishing this goal. My in-laws, Vito and Christine Barbera, have been a second set of parents to me, offering me unconditional support, as only parents can, throughout the years. My sister-in-law, Lisa Barbera, has been a stellar example to me of a successful female scientist, and has given me much (solicited) advice to help with tough decisions. My sister, Shaista Raza, has been a silent but indispensable partner in everything I do. Words cannot begin to describe my appreciation of my two biggest cheerleaders, my parents Ali and Kulsoom Mohammadi, and how they have supported me my entire life, believing I am capable of achieving anything. This has been a long road, and they have been my pillars of strength every step of the way. My father was the ﬁrst scientist I knew in life, and his passion for discovery has deﬁnitely rubbed off on me.

iv Finally, none of this would be possible without the unconditional love and support of my husband, Paul Barbera. If there is anyone who rivals my parents for my biggest fan, it is Paul. He has supported me in every sense of the word from the very beginning, and I look forward to our future adventures. Our life together has brought me so much genuine and profound joy, it has made even the toughest situations bearable. Thank you, Paul, for everything.

v Contents

Abstract ii

Acknowledgements iv

Table of Contents viii

List of Tables ix

List of Figures x

1 General Introduction 1 1.1 Quantitative framework for studying thermal sensorimotor behavior ...... 2

1.2 Nociception and pain in mammals ...... 5

1.3 Noxious signal transduction ...... 7

1.4 Attempts to map the spatial sensitivity of nociception ...... 8

1.5 C. elegans as a model organism for the study of nociception and pain ...... 9

2 Thermal stimulus experiment and behavioral phenotyping 14 2.1 Introduction ...... 14

2.2 Instrumentation ...... 15

2.2.1 Worm Zapper V1 ...... 15

2.2.2 Worm Zapper V2 ...... 18

2.3 Characteristics of the noxious thermal pulse on a moving worm ...... 22

vi 2.4 High-content phenotyping ...... 25

3 Behavioral response of C. elegans to localized thermal stimuli 38 3.1 Abstract ...... 38

3.2 Background ...... 39

3.3 Results ...... 41

3.3.1 Novel assay for quantifying the noxious response and mapping the behavioral receptive ﬁeld ...... 41

3.3.2 Multi-parameter, high-content phenotyping of N2 noxious response for the head, midbody, and tail ...... 43

3.3.3 The noxious response is elicited by a temporal temperature gradient rather than a temperature threshold...... 45

3.3.4 Spatial sensitivity of the midbody response ...... 47

3.3.5 Mutant behavioral analyses identify neurons involved in the midbody and tail responses ...... 47

3.3.6 PVD is required for the midbody and tail thermal noxious response . . 51

3.3.7 PVD responds differently to spatially localized heat pulses targeted at different locations near the midbody ...... 53

3.3.8 Mutant strains show defective noxious behavior suggesting molecules involved in sensing heat at the midbody ...... 54

3.4 Discussion ...... 57

3.5 Conclusion ...... 60

3.6 Materials and Methods ...... 60

4 C. elegans as a model system for pain 65 4.1 Introduction ...... 65

4.1.1 Animal models in thermal pain research ...... 67

4.1.2 Opioid pharmacology ...... 69

vii 4.2 Evidence of a mu opioid receptor in C. elegans ...... 70 4.2.1 The thermal avoidance response is modulated by morphine and naloxone at the midbody ...... 70 4.2.2 Preliminary mutant screen identiﬁes MOR candidates in C. elegans . . 73 4.3 C. elegans with defective receptor NMUR-2 display spatially sensitive impaired nociception ...... 76 4.4 Materials and Methods ...... 79 4.5 Future directions ...... 79 4.6 Concluding remarks ...... 82

References 84

viii List of Tables

3.1 Mutant strains used for thermal nociception assay...... 55

4.1 Results from BLASTP search identifying top matches between C. elegans receptors and the human MOR ...... 74

ix List of Figures

1.1 Schematic of transmission of nociceptive signals from peripheral target tissue to the dorsal horn of the spinal cord...... 6

1.2 Confocal image of the PVD neuron in C. elegans...... 12

2.1 Worm Zapper V1 ...... 16

2.2 Temperature proﬁle of an anesthetized worm on an agar plate...... 17

2.3 Effect of the lid on the beam temperature at the surface of the agar ...... 18

2.4 Schematic of localized thermal stimulus assay...... 19

2.5 Reducing beam size for Worm Zapper V2 with 100mm lens ...... 19

2.6 Sample pattern on receipt paper using IR laser ...... 20

2.7 Thermocouple measurements of 150mA 133ms pulse ...... 21

2.8 Thermal proﬁle of 133 ms IR pulse measured using a thermal camera . . . . . 22

2.9 Spatial and temporal temperature distributions for 150mA pulse ...... 24

2.10 Example of raw data passed through the skeletonization algorithm to reduce the image to 41 points (x,y pixel locations) along the body of the worm. . . . . 26

2.11 Examples of raster plots generated from high-content phenotyping tool . . . . . 29

2.12 Examples of uses for high-content phenotyping tool ...... 30

2.13 Analysis of center-of-skeleton trajectory of a worm crawling on an agar plate, stimulated with a noxious thermal pulse at the head...... 33

2.14 Normalized (z-scores) of behavioral feature vectors for 16 strains for head, midbody, and tail stimulation...... 35

x 2.15 Correlation amongst strains using feature vectors from Figure 2.14...... 36

3.1 Assay for the spatial dissection of the thermal noxious response...... 42 3.2 Spatial high-content phenotyping of N2 thermal noxious response...... 44 3.3 Withdrawal behavior is dependent on the ramp rate of the thermal stimulus. . . 46 3.4 Spatial sensitivity of the midbody thermal noxious response...... 48 3.5 Strains exhibiting spatially defective thermal avoidance behaviors...... 50 3.6 The PVD sensory neurons mediate the thermal noxious response in the midbody and tail...... 52 3.7 The PVD neurons show a spatially dependent response to localized noxious heat pulses at the midbody...... 54 3.8 Potential molecules spatially mediating the thermal avoidance response at the midbody...... 56

4.1 Localized effect of morphine and naloxone on wild type C. elegans...... 72 4.2 Preliminary measurement of morphine analgesia on midbody thermal noxious response for candidate MOR strains ...... 75 4.3 NMUR-2 exhibits defective nociceptive behavior ...... 78

xi Chapter 1

General Introduction

Imagine touching a very hot stove. Your initial reaction to the sharp increase of heat is to rapidly withdraw your hand away from the hot surface, and maybe say “ouch”. The cascade of molecular cues triggered by an acute, possibly harmful stimulus are designed to elicit a withdrawal response to protect an organism from potential tissue damage or environmental threat (Pirri and Alkema, 2012; Bromm and Treede, 1980; Julius and Basbaum, 2001). This detection of a noxious stimulus – the process of nociception – and the resulting behavior is critical for an animal’s survival (Wittenburg and Baumeister, 1999; Glauser et al., 2011; Maroteaux et al., 2012). Although the rapid, coordinated response happens seemingly in an instant, an optimal protective sensorimotor program must determine the location of the noxious stimulus to initiate proper withdrawal direction, and evaluate the level of the threat so the organism can remain in “escape mode” until it is deemed safe again. The entry point to this nociceptive behavior is a particular type of sensory neuron, ﬁttingly called the nociceptor. Vertebrates and invertebrates both possess nociceptors that transduce various kinds of noxious stimuli, such as harsh touch or acute heat (Caterina et al., 1997; Chatzigeorgiou et al., 2010; Xu et al., 2006). Nociception is considered to be the purely physiological, non-psychological aspect of pain. The Interna- tional Association for the Study of Pain (IASP) deﬁnes pain as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage...”(Bonica, 1979). The

1 CHAPTER 1. GENERAL INTRODUCTION 2 inclusion of emotion in the perception of pain suggests that invertebrates which lack higher cortical function cannot feel pain per se, although they do share similar nociceptive building blocks and pathways that may be precursors to mammalian systems, and therefore useful in understanding the mechanisms of nociception (Julius and Basbaum, 2001). In this thesis, I will primarily focus on nociception in the invertebrate system C. elegans. However, I will use the term pain when discussing the sensorimotor behavior that can be perturbed by clinically relevant pharmacological agents that are known modulators of pain (for example, opiates).

1.1 Quantitative framework for studying thermal sensori-

motor behavior

The ability to sense and react strategically to environmental cues in order to ﬁnd favorable conditions or avoid negative situations is common to organisms ranging from tiny bacteria (Berg, 2004), to lower vertebrates (Sneddon et al., 2003), to higher vertebrate animals, including humans (Bromm and Treede, 1980; Brockmann et al., 2006). In order to study complex behaviors, an important task at hand is to identify meaningful measurements that characterize these processes. The question arises, how do we describe a multifaceted behavior in a more than qualitative way so we can begin to investigate the underlying mechanisms? These behavioral events are the result of organism-environment interactions that vary in time. Is our snapshot of behavior that we use as a phenotype an accurate representation of a dynamic, ﬂexible system with rapid and reversible modulation? In theory, we can simplify a multidimensional, complex behavior that is time variant by saying it is a sequential series of events, and look at the probabilities of these events occurring over time. Behavior can then be seen on the level of the animal as a whole moving through and sensing its environment. With careful instrumentation designed to prudently control environmental stimuli, we can begin to uncover the genetic and/or neural correlates of the sensorimotor programs mediating a particular behavior.

To this end, the feature-rich behavior I sought out to dissect using novel instrumentation CHAPTER 1. GENERAL INTRODUCTION 3 and quantitative techniques is thermal nociception in C. elegans. C. elegans has emerged as a powerful model organism since its “parts list” (genes and neurons) is well characterized. With the ability to specifically perturb the biology (for example, with targeted laser ablation of neurons or specific genetic mutations), the next goal is to make biology predictive and quantifiable by observing behavior at the systems level. Statistical and mathematical tools to analyze behavior provide this framework. Pursuant to this, another benefit to using C. elegans instead of other animal models – such as the common rodent model – for behavioral studies, is that the locomotion of the worm is quantifiable using relatively straightforward machine vision techniques (discussed in Chapter 2). Additionally, its locomotory response to stimuli is a reliable measure of output for most, if not all, sensory inputs. We therefore can obtain an exhaustive measure of the behavioral output as a comprehensive feature space while varying sensory input, in a relatively high-throughput way. Essentially, by watching how the worm moves, we can read its “body language” to assess how it is reacting to our presented stimuli. One example of a sophisticated sensory behavior in C. elegans that was observed through its locomotion originally in 1975 by Hedgecock and Russell (Hedgecock and Russell, 1975), is thermotaxis: the worm’s behavioral tendency to migrate through temperature gradients towards a preferred temperature near its cultivation temperature, then track this isotherm within 0.05 of a degree. Further quantitative analysis of movements of the worm performing thermotaxis advanced our understanding of this process, revealing mechanisms underlying the behavior. Notably, the worms migrate down gradients to the cultivation temperature but not up gradients, and do so by modulating run duration(Ryu and Samuel, 2002). When close to the cultivation temperature (the distribution of temperatures that the worm will track is not necessarily centered on the cultivation temperature), the worm then engages a different mechanism to track the isotherm, and does so by controlling run orientation (Ryu and Samuel, 2002). Another example of C. elegans sensory behavior via locomotion is the slowing of movement when the worm feels the mechanical sensations of crawling on its food, in order to maximize its time in the presence of food (Sawin et al., 2000). More modular stereotyped sensorimotor behaviors have been CHAPTER 1. GENERAL INTRODUCTION 4 shown in C. elegans as well, such as the thermal avoidance response to a noxious thermal pulse heating the whole body of a freely crawling worm (Ghosh et al., 2012): the worm halts its forward movement to initiate a reversal, reorients itself so its head is pointed away from the stimulus, then resumes its forward motion in that new direction. While this stereotyped behavior is deterministic, multidimensional phenotyping of the response discussed in greater detail in Chapters 2 and 3 reveals strategies underlying the graded behavior, and the worm’s simple neural circuitry can control several aspects of the response advantageously.

More generally, we can analyze this directed, self-propelled escape of the worm by mathe- matically characterizing the worm’s trajectory itself. An elegant example of how an organism may employ a particular movement strategy to sample its space advantageously is the biased random walk for bacterial chemotaxis (Berg, 2004; Berg and Brown, 1972). Brieﬂy, the bac- terium E. coli samples its world by executing a random walk composed of two alternate modes: runs (smooth swimming, average length about 1 second) and tumbles (abrupt direction changing events with little net displacement, shorter length of about 0.1s). If the cell is traveling in a favorable direction, such as that of an increasing chemical attractant spatial gradient, the tumbling probability decreases and the runs are lengthened; this biases the random walk towards the favorable region and enables chemotaxis (Berg and Brown, 1972). Segall et al. studied the chemical impulse response and demonstrated that the tiny cell is able to make measurements of chemical gradients by comparing the concentration over the past 1 second with that over the previous 3 seconds as it moves, and responds accordingly (Segall et al., 1986). Therefore, since E. coli does not store spatial information about the gradient, only temporal, when performing negative chemotaxis (traveling up a spatial gradient of chemorepellent), the cell shortens its run durations and increases frequency of tumbling, but does so with no preferred direction. This is different from our observed avoidance response in C. elegans. As I will show in Chapter 3, the direction of movement of the worm after a noxious stimulus is very much dependent on stimulus location, demonstrating that C. elegans has the ability to store spatial information about environmental stimuli. While the timescale of the nociceptive behavior I am capturing is short CHAPTER 1. GENERAL INTRODUCTION 5

(the time series is only 15 seconds post stimulus), we still can ﬁnd an underlying signature in the behavior by looking at the mean-square displacement (MSD) between two points of the centroid’s trajectory. Our results will be discussed further in Chapter 2.

Comprehensively analyzing the behavior of the worm using this systems-level approach was the ﬁrst step towards building a behavioral receptive ﬁeld map for the underlying thermal nociceptors of the worm. From there, I had the necessary framework to quantitatively assess the effects of genetic and neuronal perturbations to the system.

1.2 Nociception and pain in mammals

The concept of a special type of primary afferent neuron being the entry point to cutaneous nociception and pain has been around at least since 1903, when Nobel laureate Sir Charles Scott Sherrington stated “There is considerable evidence that the skin is provided with a set of nerve endings whose speciﬁc ofﬁce it is to be amenable to stimuli that do the skin injury, stimuli that in continuing to act would injure it still further” (Sherrington, 1903). Since then, we have developed a much broader and deeper understanding of these nociceptors, and how they act as specialized sensors for noxious stimuli.

The cell bodies of nociceptors in mammals are found in the dorsal root ganglia (DRG) for the body, and the trigeminal ganglion (TG) for the face. These primary afferent nociceptors innervate the target organ (for example, the skin), and convey the noxious signal to projection neurons in the spinal cord (Figure 1.1) (Basbaum et al., 2009; Zeilhofer, 2005). The mechanisms of dorsal horn and supraspinal processing of the noxious signal into the thalamus and further ascending regions of the brain is beyond the scope of this introduction, but it is interesting to note that there are very precise targets for the afferent ﬁbers in the spinal neurons, and different afferents terminate within speciﬁc, discrete laminae in the dorsal horn of the spinal cord (Basbaum et al., 2009). Furthermore, another interesting aspect that makes nociceptors different from other somatosensory neurons is that they are bidirectional (Basbaum et al., CHAPTER 1. GENERAL INTRODUCTION 6

Figure 1.1: Schematic of transmission of nociceptive signals from peripheral target tissue to the dorsal horn of the spinal cord. The cell body is in the dorsal root ganglia (DRG), and the nerve has a single axon that bifurcates to one part that innervates the target tissue, and a second part that forms the presynaptic terminal on to second order neurons in the dorsal horn of the spinal cord. Adapted from (Zeilhofer, 2005).

2009). As seen in Figure 1.1, there is one single axon that bifurcates and innervates both the peripheral and central terminals. The proteins synthesized in the DRG cell are distributed to both the recipient branch (peripheral terminal, or dendrite in prototypical neurons) and the transmission branch (central terminal); either end can therefore send or receive signals (Basbaum et al., 2009). This bidirectionality in the transmission of the pain signal (that is, the central and peripheral terminals can both transmit and receive pain signals) is of particular interest when developing novel pain therapeutics, since both ends of the afferent mediate sensitivity to pain (Basbaum et al., 2009). Nociceptors are categorized into two main classes based on their com- position and function (Burgess and Perl, 1967; Bessou and Perl, 1969; Yeomans and Proudfit, 1996; Yeomans et al., 1996; Dubin and Patapoutian, 2010; Lawson, 2002). First, there exist lightly myelinated, medium diameter Ad fibers whose faster conductance velocities (5-40 m/s) mediate the localized, acute, withdrawal-inducing “fast pain” response (Julius and Basbaum, 2001; Lawson, 2002; Yeomans et al., 1996; Price and Dubner, 1977). The majority of nociceptors, however, are made up of the small diameter, slower conducting (~1 m/s) unmyelinated C-fibers (Dubin and Patapoutian, 2010; Price and Dubner, 1977; Lawson, 2002). These nerves mediate the dull, diffuse, “slow pain”, that you feel a few seconds after the acute pain that initiates withdrawal (Price and Dubner, 1977). Both fibers are polymodal, meaning they can CHAPTER 1. GENERAL INTRODUCTION 7 respond to mechanical, thermal, and chemical stimulation (Julius and Basbaum, 2001). One additional interesting functional segregation is selectivity in pharmacological perturbations in the Ad and C fibers (Cuellar et al., 2010; Basbaum et al., 2009). In particular, opioids seem to selectively reduce pain in C fiber nociceptors, but not Ad fibers (Lu et al., 1997; Cooper et al., 1986; Jones et al., 2003). It has been suggested that the difference in antinociceptive effects of opioids on different kinds of clinical pain (for example, acute pain from coughing or dressing change versus dull pain from recovery from surgery) may be due to the differential distribution of opioid receptors at the central terminals of the C versus Ad fibers (Lu et al., 1997; Cuellar et al., 2010; Cooper et al., 1986). This suggests the utility in using pharmacological perturbations of the nociceptive response to reveal underlying mechanisms in the sensorimotor program at the level of the nociceptor. This was one main motivation for my pharmacology experiments in Chapter 4.

1.3 Noxious signal transduction

The transient receptor potential (TRP) channel superfamily of cation channels has an important role in signal transduction for sensory behavior (Montell, 2005; Patapoutian et al., 2009). The TRPs are molecular detectors of many sensory stimuli, such as smell, taste, vision, mechanosensation, and pain (Montell, 2005). The original TRP channel was found as a detector for light in Drosophila melanogaster (Montell and Rubin, 1989). Of relevance to this thesis are the so-called ThermoTRPs (for review, see (Patapoutian et al., 2009)). A major advance in understanding the transduction of thermal pain signals came from the discovery that the mammalian subgroup of TRP channels, TRPV(anilloid)1 is activated by both noxious heat (> ~43ºC) and capsaicin, the compound that makes chili peppers hot (Caterina et al., 1997; Caterina et al., 1999). This functionally confirmed that TRPV channels are in the thermal pain pathway. However, when TRPV1 deficient mice were stimulated with noxious heat, they still showed a normal behavioral noxious response (Woodbury et al., 2004). This suggests an CHAPTER 1. GENERAL INTRODUCTION 8 additional pathway in the pain response that is TRPV1 independent (Woodbury et al., 2004; Basbaum et al., 2009). Potential candidates (such at TRPV3 or TRPV4) have been suggested, but a definitive TRPV1-independent component of the pain pathway is still not known (Bas- baum et al., 2009).

Another temperature-activated TRP channel is TRPA1 (Patapoutian et al., 2009; Basbaum et al., 2009). In addition to being activated by certain chemical compounds such as those found in wasabi (Jordt et al., 2004) and garlic (Bautista et al., 2005), the TRPA1 channel is also activated by noxious cold (< ~15ºC)(Story et al., 2003; Kwan et al., 2006). This suggests that nociceptors mediate different painful stimuli using distinct molecular machinery, a discovery that may be useful when designing pharmacological agents targeting distinct manifestations of pain (Patapoutian et al., 2009).

1.4 Attempts to map the spatial sensitivity of nociception

One of Sherrington’s initial observations about the relaying of a decidedly painful message from an afferent neuron, was that the behavioral response due to a potentially harmful stimulus depends on the location of the stimulus. For example, stepping on a thorn: “the foot is withdrawn from the offending stimulus. Instead of wounding itself further it escapes from the threatened wounding”(Sherrington, 1903). The assaulted organism must assess the location of the stimulus and execute a motor response that withdraws it from danger. Also, the aforementioned idea of receptor localization affecting the antinociceptive properties of morphine- modulated behaviors (Yeomans et al., 1996; Lu et al., 1997) further highlights the importance of investigating the spatial sensitivity in the pain response. Extensive research has been done in mammalian models to create so-called “nociceptive maps” using psychophysical measurements or functional imaging (fMRI) of higher cortical processes (Mancini et al., 2012; Apkar- ian et al., 1999; Baumgärtner et al., 2010; Bingel et al., 2004; Bornhövd et al., 2002; Chen et al., 2011; Coghill et al., 1999; Weissman-Fogel et al., 2012), but these measurements are CHAPTER 1. GENERAL INTRODUCTION 9 not at the level of the nociceptor. This research on the CNS mechanisms of spatial information processing does not give us any insight if and how spatial information could be processed and relayed in subcortical regions – in particular the peripheral terminal – to elicit a coordinated motor response. The current view of peripheral involvement stems from classical psychophysical models which are explained by peripheral innervation density and receptive field size, rather than any spatial encoding mechanism in the nociceptor itself (Kandel et al., 2000; Arthur and Shelley, 1959; Johnson and Phillips, 1981). However, recent work in human volunteers combining psychophysics with histology strongly suggests that this classic notion is incorrect (Mancini et al., 2013). In this study, Mancini and colleagues deliver small-diameter painful laser pulses to the fingertips and hand, respectively, and show a higher spatial resolution for pain at the fingertips compared to the hand. Yet, when the skin biopsies were performed, the histology showed that there was a much lower innervation density of nociceptors on the fingertips compared to the hand (Mancini et al., 2013). The surprising conclusion is that peripheral innervation density alone cannot account for spatial resolution of the pain response outside the CNS. Could there be spatial discrimination at the level of a single nociceptor? Chapter 3 in this thesis establishes the C. elegans nociceptor PVD as a single neuron model for the spatial discrimination of the nociceptive response.

1.5 C. elegans as a model organism for the study of nocicep-

tion and pain

Since the 1960s when Sydney Brenner introduced the small nematode C. elegans as a model organism to the scientiﬁc community, this little roundworm has continued to gain recognition as an attractive model for studying many fundamental biological phenomena(Brenner, 2002). What makes C. elegans especially powerful as a model system is the wiring diagram of the ~300 neurons in the worm is known, and the entire genome of the animal is sequenced (White et al., 1986; Bargmann, 1998). Knowing this parts list is only half the story, however, in using CHAPTER 1. GENERAL INTRODUCTION 10

C. elegans to study sensorimotor behavior. While the connectivity of the neurons is known, this anatomical description is not enough to explain the complexity of behaviors in the organism, especially for those sensory behaviors that are mediated by overlapping circuits. An example of multiple circuits underlying behaviors from a single sensory neuron comes from the study of the polymodal nociceptor ASH. ASH initiates different behaviors through the selective activation of distinct circuits and molecules based on the environmental conditions and stimulus modality. One behavior mediated by ASH is the avoidance response to three different modalities, namely mechanosensation (nose touch), osmotic shock, and chemorepulsion (Bargmann et al., 1990; Kaplan and Horvitz, 1993; Hart et al., 1999). These nociceptive responses are all initiated by a reversal; ASH directly chemically synapses onto command interneuron AVA which is necessary for reversals (White et al., 1986; Chalﬁe et al., 1985). ASH uses distinct molecules to transduce these different stimulus modalities. For example, the glutamate receptor GLR-1 is required in neurons postsynaptic to ASH for withdrawal from mechanical stimuli but not osmotic avoidance (Maricq et al., 1995), and OSM-10 preferentially mediates they hy- perosmolarity response (Hart et al., 1999). Even further, the role of ASH in octanol avoidance depends on the feeding status of the animal; well-fed worms detect the volatile repellant primarily through ASH, but after 10 minutes off food the animals distribute the sensory input across 3 different neurons, ASH, ADL and AWB (Chao et al., 2004). The behavior resulting from these different circuits differs as well– the avoidance response is elicited more rapidly on food than off food (Chao et al., 2004). The mechanism for this differential behavior is not completely understood, but the suggested model is that the higher levels of serotonin (when on food) act directly on ASH to increase neuronal sensitivity or synaptic output (Chao et al., 2004).

Furthermore, in addition to the avoidance reversal response mediated by the chemical synapse to AVA, ASH also mediates another completely different behavior, aggregation, by forming an electrical synapse to the RMG inter/motor neuron in a hub-and-spoke circuit (Ma- cosko et al., 2009). When the neuropeptide receptor NPR-1 (expressed in RMG) activity is CHAPTER 1. GENERAL INTRODUCTION 11 low, aggregation occurs without affecting the avoidance response, even though both behaviors are triggered by ASH(Macosko et al., 2009). ASH therefore mediates different behaviors depending on the internal states of the worm.

These examples of the complexity of the ASH response make it very clear that there is a lot more ﬂexibility in neural circuits controlling behavior than what could be determined from the wiring diagram alone. C. elegans has remarkable behavioral complexity despite a relatively simple anatomy, and sensory context is very important. A couple major motivations in the design of our experiments for the worm are clear: 1) techniques to comprehensively quantify complex behaviors in the animal must be developed to successfully dissect the underlying circuitry (important differences can be missed by simply qualitative analysis), and 2) experiments must control the environmental conditions so the internal states of the worm are (to the best of our ability) consistent when presenting carefully controlled stimuli. In the worm, sensory behaviors are generally mediated by locomotion; we therefore quantify the movement of the worm in various ways to this end, as discussed in Chapters 2 and 3.

In addition to the ASH-mediated avoidance response, substantial work has been done in characterizing the molecules involved in the thermal noxious response (Liu et al., 2012). In particular, recent work identiﬁed TRPV1 channels contribute to the thermal noxious response through the FLP head neurons and PHC tail neurons, and cGMP signaling further contributes to the noxious response at the head through the AFD neuron (Liu et al., 2012). However, as discussed in Chapter 3, the thermal stimulus in these experiments was not sharply localized along the body the of worm, and the phenotyping of the response was coarsely done; both of these issues were barriers to the spatial dissection of the thermal noxious response. Our work addresses this limitations with a novel assay for localized thermal stimuli and high-content phenotyping of the response. Through this, we uncovered a previously unreported noxious thermal response at the midbody.

The noxious responses to harsh touch and cold shock along the body of the worm are mediated by the pair of polymodal PVD nociceptors (Way and Chalﬁe, 1989; Chatzigeorgiou et al., CHAPTER 1. GENERAL INTRODUCTION 12

Figure 1.2: Confocal image of the left PVD neuron (PVDL) in C. elegans, showing extensive branching along body of the worm. Scale bar is 15mm. Taken from (Chatzigeorgiou et al., 2010).

2010). PVD has an extensive dendritic arbor that blankets the body of the worm, a morphol- ogy resembling mammalian nociceptors (Figure 1.2). Recent work identiﬁed distinct sets of molecules involved in sensing different noxious stimuli PVD, namely, the DEG/ENaC channels for harsh touch and the TRPA-1 ion channel for cold shock (Chatzigeorgiou et al., 2010). Using mutant analysis, laser ablation, and genetic ablation, I show in Chapter 3 that the PVD neurons are also required for sensing noxious heat at the midbody, and this may be mediated by the TRPV channel homologue OCR-2. This would be analogous to the mammalian polymodal nociceptors described above, with the TRPV1 channel in the pain pathway for noxious heat, and the TRPA1 channel mediating cold shock (Patapoutian et al., 2009). Unlike the polymodal nociceptor ASH which possesses multiple chemical and electrical targets as discussed earlier, PVD has no known gap junctions and chemically synapses to only two command interneurons, AVA for reverse locomotion and PVC for forward locomotion (White et al., 1986). Chapter 3 also discusses the ability of this neuron to not only detect noxious heat, but to discern the location of of the noxious stimulus in order to initiate an advantageous withdrawal direction.

Having established the quantitative behavioral receptive ﬁeld of C. elegans for thermal nociception, the next logical question to see if the nociceptive behavior can also be described as a pain response is to examine the effects of clinically relevant analgesics. Therefore, my next CHAPTER 1. GENERAL INTRODUCTION 13 task was to investigate pharmacological perturbations to the system. In Chapter 4, I administer an opioid agonist, morphine, and inverse agonist, naloxone, to the worm, and use the spatially localized thermal stimulus assay to this end. The results are promising; I discovered the worm has morphine-induced analgesia at the midbody, as well as naloxone-induced hyperalgesia at the midbody. The spatially-sensitive observations would not have been exposed without the localized stimulus assay characterized in Chapter 2. The major goal for my contribution to this ﬁeld is advancing C. elegans from an already useful model for the study of cellular and molecular mechanisms of nociception, to a quantitative behavioral model for pain research. In Chapter 2, I describe novel instrumentation and high-content phenotyping of the sensorimotor response of C. elegans to noxious thermal stimuli. In Chapter 3, I employ these techniques to reveal a previously unreported midbody response, and further investigate the underlying neural mechanisms of thermal nociception at the midbody. Chapter 4 harnesses the utility of our spatially localized assay to demonstrate the use of C. elegans as a model organism for pain and pharmacology. Chapter 2

Thermal stimulus experiment and behavioral phenotyping

2.1 Introduction

C. elegans has proven to be a valuable model organism for the study of many fundamental biological phenomena, such as apoptosis (Horvitz, 2003), mechanosensation (Chalfie et al., 1985), and RNA interference (Fire et al., 1998). However, studies of C. elegans behavior have pro- ceeded more slowly than cellular or molecular pursuits. Studying the neurobiology of behavior requires the development of novel instrumentation to carefully control stimuli and quantitative methods to comprehensively characterize the response. These tools, combined with progress in in vivo imaging techniques, have begun to emerge in the field, advancing the potential for studying the neuronal processes underlying complex behaviors. Thermal nociception in C. elegans, however, has not been widely studied, especially at the behavioral level. One explanation for this could be that temperature poses particular challenges for instrumentation. For example, aiming, focusing, and localizing a thermal pulse with a particular ramp rate, and further measuring the actual temperature profile felt by the worm, make it difficult to characterize and implement these experiments. Attempts in the past have used a heated metal pen

14 CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 15 or presented a laser stimulus for a long enough time (10 seconds) so that the spatiotemporal profile of the stimulus is unknown (Wittenburg and Baumeister, 1999; Liu et al., 2012). This section describes the characterization of the thermal stimulus assay used in our first study to transiently heat up the whole worm (V1) (Ghosh et al., 2012), then modified for our second study to locally stimulate regions along the body of the worm (V2) (Mohammadi et al., 2013). Pursuant to this, I describe the machine vision techniques and high-content phenotyping used to quantitatively analyze the response.

2.2 Instrumentation

2.2.1 Worm Zapper V1

The original laser-based thermal stimulus assay, Worm Zapper V1, was designed and built by William Ryu. In this setup, a collimated beam with a 1/e diameter of 1.50 mm from a 1440 nm diode laser was manually positioned to transiently heat the entire body of a worm crawling on an agar plate (Figure 2.1). Custom software was written in LabVIEW (National Instruments, Austin, TX, USA) and used to control the ﬁring, power, and duration of the laser, while simultaneously recording images of the crawling worm. This assay was used by Raj Ghosh at Princeton University to obtain the behavioral responses for 4 thermal pulses on the wild type strain N2 in addition to 47 mutant strains with various neurological defects (Ghosh et al., 2012). My contribution to this work was performing the high-content phenotyping of the data to reveal the features used in the analysis. The details of the high-content phenotyping and an illustration of how it was applied in the Ghosh et al. study are described later in this chapter. I also used a thermal camera (CI 7320, Infrared Camera Inc., Beaumont, TX, USA) to calibrate the stimulus; Figure 2.1 shows the relationship between the power of the laser pulse and the temperature of the agar for the 0.5s pulses used in these earlier experiments.

In order to ensure that our measure of the surface of the agar was a valid surrogate measure for the worm’s temperature, I used the thermal camera to measure the temperature of an CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 16

Figure 2.1: Worm Zapper V1. (A) An infrared laser with a beam width larger than the worm (1/e = 1.5 mm) transiently heats the worm. Shown is time lapsed raw data of the worm’s typical escape response: a reversal (red), a deep reorienting turn (called an omega turn) (green), and a resumption of forward movement (blue). (B) Thermal camera measurements of the maximum temperature increase caused by a 0.5 s laser pulse at various currents. Figure reproduced from (Ghosh et al., 2012). anesthetized worm on an agar plate while ﬁring the laser. Figure 2.2 shows there is no perturbation in the temperature proﬁle with the worm in the laser area, so the measured heating of the worm’s body is the same as the agar itself.

Finally, because we cannot use the thermal camera to image the agar through the lid of the plate, we performed all of our thermal camera measurements with the lid off. However, in all of our experiments we keep the lid on. Therefore, to ensure our thermal camera measurements were accurate representations of our experiments, I cut a hole in the lid of a plate and aimed the beam through the lid but imaged on the surface of the agar (Figure 2.3). We found the lid made no difference to the temperature of the beam at the surface of the agar (Figure 2.3).

While the whole-worm heating experiments revealed interesting features of the thermal sensorimotor response in C. elegans (examples given below in the the high-content phenotyping discussion), we wondered if heating the entire worm was confusing the response by activating multiple nociceptors and sensory pathways at once. In order to spatially dissect the noxious response, we modiﬁed Worm Zapper V1 by reducing the beam size with a 100mm focal length lens, and carefully characterized the modiﬁed stimulus in Worm Zapper V2, as described below. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 17

Figure 2.2: Temperature profile of an anesthetized worm on an agar plate before and during laser pulse. A) Picture of anesthetized worm on agar plate with tape marking its position. B) Closer view of the worm in A) C) Image from thermal camera before laser pulse showing worm in field of view D) Image from thermal camera mid-pulse, showing beam is heating area with worm E) Surface plot showing thermal profile of beam with worm F) Horizontal slice through maximum of E), confirming there is no perturbation to the temperature profile of the laser beam with the worm in the heated region. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 18

Figure 2.3: Effect of the lid on the beam temperature at the surface of the agar. A) Image from thermal camera showing hole cut in lid to allow imaging of the surface of the agar (TOP), and the laser beam, which was aimed through the lid and appears on the surface of the agar (BOTTOM). C) Maximum temperature increase as a function of laser current for the lid and no lid measurements.

2.2.2 Worm Zapper V2

A schematic of our complete experimental setup for the localized stimulus experiments is reproduced from (Mohammadi et al., 2013) in Figure 2.4.

Inserting a 100mm focal length lens in the optical path of our infrared laser stimulus setup reduced the optical size of the 2mm collimated beam to 90 microns at the surface of the agar (Figure 2.5). We measured the thermal size of our laser pulse in two ways: 1) moving the beam around a stationary thermocouple (using mirror galvanometers to steer the beam in a grid pattern), and 2) real-time measurement of the pulse using a thermal camera.

Thermocouple measurement:

For our ﬁrst attempt to measure the maximum temperature and thermal beam size for our 150mA 133ms pulse (our highest thermal dose), we placed the junction of a 0.003 inch diameter copper-constantan thermocouple (OMEGA COCO-003) on the surface of an agar plate, and moved the beam position in a grid around the thermocouple by inputing X and Y voltage steps to the galvanometers. We calibrated the voltage to position relationship for aiming the beam using receipt paper (Figure 2.6). Speciﬁcally, we heated the paper with the laser to make CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 19

Figure 2.4: Schematic of localized thermal stimulus assay. Reproduced from (Mohammadi et al., 2013).

Figure 2.5: Reducing the optical size of the 2mm collimated beam to 90 microns at the surface of the agar plate for Worm Zapper V2, by inserting a 100mm lens in the optical setup. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 20

Figure 2.6: Sample pattern made by locally heating receipt paper using beam-positioning galvanometers. Distance between dot centers is 0.3mm. a mark, then moved the beam a particular voltage step to make another mark, and thus were able to measure the voltage to position conversion by measuring the distance between marks. We found that a 0.1V step moved the beam 0.3mm. We were able to maintain linearity in the beam steps using voltage increments as small 0.01V, allowing us to sample the beam with a spatial resolution of 30 microns.

At each beam position, the laser was fired (150mA 133ms pulse) and the thermocouple simultaneously measured the temperature time series. We then waited a set time to allow the temperature to return to baseline, then repeated the time series measurement for that location. When 10 temperature time series measurements were taken for that given beam position, the beam was moved using the galvanometers and the process was repeated for the new beam position. Through this method we were able to obtain a 2D reconstruction of the thermal beam profile. Figure 2.7a shows our measurement of the thermal profile of the beam using the averaged maximum change in temperatures recorded by this beam scanning method, with the thermocouple resting on top of the agar. The copper and constantan wires are clearly visible in the temperature profile, and given copper’s high thermal conductivity, it is very likely the presence of the thermocouple in the beam is resulting in an overestimation of DT (~6ºC). In order to remove the effects of the thermocouple itself in the measurement, I made a hole in the bottom of the petri dish used for the agar plate, and threaded the thermocouple through it so that the junction was right at the surface of the agar. Figure 2.7b shows the new measurement. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 21

Figure 2.7: Thermocouple measurements of 150mA 133ms pulse being moved in an XY grid with 30 micron steps. (A) The thermocouple is placed on top of the surface of the agar. The copper and constantan wires forming the thermocouple junction are clearly seen in the spatial temperature proﬁle, and likely conducting heat towards the center of the beam causing an erroneous temperature measurement. (B) The thermocouple is threaded through a hole in the bottom of the plate, removing the artifact from the wires, and the measurement of DT is signiﬁcantly reduced (colorbar).

There is still a little distortion, likely due to the thermocouple’s junction not being perfectly perpendicular to the surface of the agar, but the DT is noticeably smaller at ~1.6ºC. We were surprised that our stimulus could have such a small temperature jump, since we were able to elicit a robust noxious response with it (discussed in more detail in Chapter 3). Plus, there is clearly high variability and chance for error in the thermocouple measurements. We therefore veriﬁed the temperature proﬁle of our laser pulse with a thermal imaging camera.

Thermal camera measurement:

Figure 2.8 shows the average thermal proﬁle for three of our laser pulses, carefully calibrated using a thermal camera. The beam diameter for the hottest stimulus in our thermal dose response (150mA) was measured to be 220 mm (FWHM). This localized our noxious stimulus to CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 22

Figure 2.8: Thermal proﬁle of 133 ms IR pulse measured using a thermal camera for three pulse amplitudes (60, 100, and 150 mA), where the largest full width half maximum (FWHM) was measured to be 220 mm for the 150 mA laser pulse. Figure and legend reproduced from (Mohammadi et al., 2013). roughly 1/5 of the worm’s body.

2.3 Characteristics of the noxious thermal pulse on a moving

worm

While I was able to obtain an accurate measure of the maximum change in temperature of the noxious stimulus with the thermal camera (Figure 2.8, 150mA), a better measure of how localized our pulse is on the body of the worm during an actual experiment was necessary to validate the assay’s efficacy at spatially dissecting the thermal noxious response (Chapter 3). We must therefore also consider how the heat diffuses along the body of the worm, and if the worm’s motion in the beam region affects its sensation of the stimulus. To this end, we will first consider the diffusion of our thermal pulse along the body of the worm (we have already shown that our measurement of temperature at the surface at the agar is a good proxy for the worm’s temperature), then calculate the Péclet number of the system in order to evaluate the contribution of the worm’s motion (advective transport) to the spatiotemporal profile of the thermal environment. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 23

Figure 2.9a shows the thermal proﬁles for an IR pulse at different time steps after the laser turns off. These curves are ﬁtted with the Gaussian function of the form

(x µ)2 1 P(x)= e 2s2 (2.1) (2ps2)1/2 where the mean µ =< x >, and the variance s 2

s 2 = 2Dt (2.2) where D is the diffusion coefficient. Figure 2.9b plots s 2/2 for the first four time steps shown in Figure 2.9a, and the diffusion coefficient, D = 0.122 mm2/s, is calculated from the slope. The diffusion coefficient for our pulse is very close to the thermal diffusivity of water, which is not surprising since the agar plate is mostly water.

The time course for the decay from the peak temperature is shown in Figure 2.9c. Fitting this to an exponential, we ﬁnd the decay time to 1/e is t ~ 150ms.

We now have a total time estimate for the thermal pulse: 133ms (laser on) + ~150ms (decay to 1/e) = ~300ms. We also have an experimental measure of the diffusion coefﬁcient for our system, 0.122 mm2/s. As shown in Chapter 3, the average reaction time for a worm to withdraw from the noxious pulse is ~250ms. Therefore, the worm is moving at its pre-stimulus crawling speed for the majority of the time of the total pulse length, and at its withdrawal (accelerated) speed for ~50ms. The average speed for the crawling worm pre-stimulus from our dose response in Chapter 3 is ~0.25mm/s, and the maximum mean speed after the stimulus (for the head response) is ~0.85mm/s. Therefore, before the worm reacts, it moves on average 0.063mm in 250ms. When it withdraws, it moves on average an additional 0.043mm for the remaining 50ms of the total pulse length. In both these cases, the Péclet number is < 1:

LU Pe = (2.3) D where L is the characteristic length scale of the ﬂow (in this case, the worm’s movement), U is CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 24

Figure 2.9: A) Spatial temperature distributions for a 150mA 133ms IR pulse at different time steps. B) Variance of Gaussian functions fit to experimental data in A) (a measure of spreading) as a function of time. The slope is the diffusion coefficient, D. C) Decay from peak temperature; exponential fit gives t~150ms. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 25 the (worm’s) velocity, and D is the diffusion coefficient. The small Péclet number tells us that diffusion dominates in our system. Figure 2.9 demonstrates the small degree of spreading relative to the worm’s length (~1mm). Therefore our stimulus can indeed be considered localized, with a 220micron FWHM as described.

2.4 High-content phenotyping

In order to analyze the behavioral data obtained in our thermal impulse assays, we need a high throughput tool to take raw data (images of the worm’s movements) and extract quantitative features and behavioral events. This next section describes the quantitative analysis performed to extract behavioral features used in both the whole worm heating study (Ghosh et al., 2012) and the localized thermal stimulus study (Mohammadi et al., 2013).

Method

The raw data is obtained as a series of images of the worm’s behavior. As in initial pre- processing step, the images were run though a skeletonization algorithm written in MATLAB that thresholds and skeletonizes each frame in the image capture (Figure 2.10). The skeletonization algorithm used here was originally written by George Leung (Emory University) in Summer 2009, and I made small changes to automate it and adapt it to the relevant data. The curvature of the perimeter was used to differentiate the head and tail, since the maximum curvature is found at the tail. The endpoints and centroid were tracked frame by frame for each dataset, and used for the automatic detection of events in each frame as outlined below.

Once the data was skeletonized, we applied feature extraction and event ﬂagging algorithms to identify the behavioral states of interest (in this case, reversals, omega turns, pauses, and forward movement). I chose features of the worm’s shape and movement that provided enough information to accurately identify behavior and ﬂag events. These features were head to tail distance, centroid position and centroid speed. The software I wrote implements an algorithm CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 26

Figure 2.10: Example of raw data passed through the skeletonization algorithm to reduce the image to 41 points (x,y pixel locations) along the body of the worm. that assigns a behavior (omega turn, etc.) to each frame based on the transformation of those features from frame to frame. This allowed us to automatically perform behavioral phenotyping of the worm. It was assumed that the worm was going forward at the moment the thermal stimulus is applied. Accordingly, the first frame in the series was marked as forward, unless it met the criteria of a pause. Initially, all frames were flagged as forward, and then revised based on the outcome of the behavioral flagging algorithm. The event criteria used to identify the behaviors were as follows:

Criteria for an omega turn:

• When the head to tail distance is less than ½of the maximum head to tail distance, the frame is tentatively ﬂagged as the start of an omega turn.

• If the frame ﬂagged in the previous step also corresponds to a COM speed of < 1 pixel/frame, this frame is ﬂagged as the start of an omega turn

• The end of the omega turn is ﬂagged when the head to tail distance is greater that ½of the maximum head to tail distance

• All frames between the start and end frames are ﬂagged as omega turn

Criteria for a reversal: First case: The dataset contains an omega turn. In this case, CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 27

• For frame number 1 to the start of the omega turn, the centroid trajectory is sampled at every 2 frames. If the turning angle is greater than 120 degrees, the frame is ﬂagged as the start of a reversal

• The end of the reversal is the start of the omega turn

Second case: The dataset does not contain an automatically detected omega turn. In this case,

• For all frames in the data set, the centroid trajectory is sampled at every 2 frames. If the turning angle is greater than 120 degrees, the frame is ﬂagged as the start of a reversal

• The end of the reversal is flagged as when there is another turning event greater than 120 degrees in some frame after the frame flagged previously. All frames between these two frames are marked as the worm is in a reversal. After the last frame of the reversal, the frames are flagged as forward frames. The reversal detection then repeats to see if there are any other reversals after the first detected reversal.

Criteria for a pause:

• If the COM speed is less than 0.4 pixels/frame, and the head to tail distance is concurrently at least ¾of the maximum, and these conditions are both met for three consecutive frames, the worm is said to be in a pause state until the COM speed becomes greater than 0.4 pixels/frame.

This concluded the automatic processing of the datasets for behavioral quantification. Once this step was complete, the flags for all the frames were imported into a GUI created in MATLAB along with the images for that dataset. The user could then quickly verify the results from the automated flagging and change any erroneous/missed flags. The results from this semi- automatic verification process were then saved as the final behavioral quantification for the dataset. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 28

A sample of raster plots of behavioral ﬂags generated from this phenotyping algorithm for data obtained for the whole worm stimulus experiment is shown in Figure 2.11 (reproduced from (Ghosh et al., 2012)).

In addition to the behavioral flags, this high content phenotyping tool can output any number of quantitative features, such as the change in the probability distribution of the different behavioral states with time, the probability of transitions between states, mean speed in each behavioral state, mean start time of events, duration of states, and any other quantitative feature based on the worm’s shape or movement. This illustrates one advantage of C. elegans as a tool to study sensorimotor behavior – its simple body postures and behavioral repertoire make it particularly amenable to machine vision techniques for quantitative phenotyping. In his analysis of the thermal responses of 47 strains with mutations in genes encoding candidate molecules involved in thermosensation, Raj Ghosh used 8 features I generated from his very large data set to characterize the responses to an IR stimulus in the low, intermediate, and high thermal regimes (Ghosh et al., 2012). We chose these features because they vary with stimulus strength, as seen in Figure 2.12, reproduced from the paper (Ghosh et al., 2012). By comparing the behavior of the mutant strains to the wild type over 15 seconds using this 8-dimensional feature vector (a “behavioral barcode”) for 4 stimulus intensities (Figure 2.12), we were able to demonstrate that the sensorimotor behavior at different temperature regimes may rely on distinct sets of molecules (Ghosh et al., 2012). It has been suggested that our multidimensional phenotyping of the large library of thermosensory candidate strains demonstrates the potential of such assays for identifying novel thermal sensors, in addition to whole-genome phenotyping of the animal (Schafer, 2012). Although we examined genetic perturbations to the system in our analysis, the utility of this kind of global, high-throughput behavioral quantification can also be applied to pharmacological screens. Recent work examined the effects of neuroactive small molecules on the photomotor response on embryonic zebrafish in a similar way (Kokel et al., 2010). By using behavioral barcodes that functionally cluster, Kokel et. al discovered novel neuroactive drugs and identified targets (Kokel et al., 2010). In C. elegans, the cur- CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 29

Figure 2.11: Raster plots (ethograms) of behavioral states over time (15s) of wild type N2 animals at whole-worm heating stimulus DT = 0.4ºC, 1.0ºC, 4.8ºC, and 9.1ºC (top to bottom). Each row represents the behavior of one worm over time. Blue = forward, red = reversal, black = pause, green = omega turn. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 30

Figure 2.12: Examples of uses for high-content phenotyping tool from (Ghosh et al., 2012).Features that scale with power (A-H), and corresponding behavioral barcodes for candidate mutant strains (I), generated from the high-content phenotyping tool. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 31 rent state-of-the-art target identification for small molecules requires administering a mutagen and screening for suppression of a previously identified drug-induced phenotype (Burns et al., 2006). Applying our functional clustering to drug-induced behavior (see Figure 4 in (Ghosh et al., 2012) for hierarchical clustering of phenotypes from thermal stimuli as an example), similar to that with the embryonic zebrafish (Kokel et al., 2010), we would in future work be able to remove the need for mutagenesis and re-screening for target identification, and advance the potential of high-throughput pharmacological screens on C. elegans for therapeutic applications.

Application of high-content phenotyping to localized stimulus assay

A detailed discussion of how I employed Worm Zapper V2 to spatially dissect the noxious response is given in Chapter 3. As aforementioned in the General Introduction of this thesis, a major goal to harness the potential of C. elegans as a model organism is to make biology predictive and quantiﬁable by observing its behavior with a systems-level approach. While certain metrics (such as those chosen to characterize the response in Chapter 3) may be useful as a simpliﬁed baseline to compare differential behavior due to perturbations of the system, there exists an underlying description or signature of the worm’s sensorimotor response that can be revealed with a mathematical and statistical framework. Here, I will discuss such additional methods to quantify the worm’s behavior.

We can examine the strategies of an organism maneuvering the spatiotemporal ﬂuctuations of its environment by studying its trajectory. Figure 2.13a) shows a sample center-of-skeleton trajectory for a worm stimulated in the head with a noxious thermal pulse. The behavioral states are as shown in Figure 3.1: the worm reverses, does a reorientation event (omega turn), then continues forward motion. We can take this trajectory and look at the mean-square displacement (MSD) as a measure of “orientational memory”: if the behavior is diffusive (i.e., the motion looks like a random walk), the MSD will be linear with respect to time (Stephens et al., 2010; Berg, 1993). CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 32

MSD(t)=< r2 > =< [r(t + t) r(t)]2 > (2.4) 4 t

As expected for short timescales, we find the MSD for the escape trajectory is ballistic and grows as the square of the time difference (Figure 2.13b)) (Stephens et al., 2010). When plotting the square root of the MSD as a function time (Figure 2.13c), it becomes clear that there are two different regimes in the worm’s modular, deterministic movement, separated by the omega turn. Specifically, the worm maximally displaces (reversal) immediately after the stimulus, then once pointed in the opposite direction (this implies spatial memory of stimulus location seconds after the experience), its displacement resembles that of a worm that wasn’t stimulated (forward motion) (Figure 2.13, dotted lines). It seems that the omega turn “resets” the initial withdrawal behavior to some extent, when switching from reversal to forward and concurrently pointing the worm away from the noxious stimulus. Consider the worm’s options for escape: 1) The worm continues to go forward. This wouldn’t make sense since moving towards the stimulus is not a withdrawal response. 2) The worm turns immediately away from the stimulus then continues going forward. In this case the worm will stay in the vicinity of the stimulus for the duration of the turn, which will be longer than the reaction time to initiate a reversal, so it is also not the best strategy. 3) The worm pauses. Above a certain threshold when the stimulus is intense enough, this is also not an effective withdrawal response; it is in the worm’s best interest to withdraw rapidly. 4) The worm initiates a rapid reversal to withdraw its head from the stimulus, then once deemed “safe”, reorients itself maximally away from the location of the noxious stimulus, then continues going forward. This is what the worm actually does, and the analysis of the mean-squared displacement provides some insight into how an organism with a finite repertoire of modular behaviors can assemble them on a systems-level advantageously.

Furthermore, using data obtained from Worm Zapper V2, I also did a quantitative analysis similar to that in the Ghosh et al. study (Ghosh et al., 2012) where I generated a “behavioral barcode” (feature vector) for each strain, and compared it to N2. Although the analysis pre- CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 33

Figure 2.13: Analysis of center-of-skeleton trajectory of a worm crawling on an agar plate, stimulated with a noxious thermal pulse at the head. A) Example of the escape trajectory tracking the center of the skeleton of the worm after a thermal pulse of 150mA 133ms directed at the head. B) The average mean-square displacement (MSD) for worms stimulated in the head with a noxious thermal pulse. Pulse at 1 second (black arrow). As expected for the short timescale, we see ballistic behavior. C) The root-mean-square displacement as a function of time from the data in B) (blue solid line). Pulse at 1 second (black arrow). Linear fits were obtained for times before and after the reorientation event. Green dashed line fits the time pre- omega turn, Red dashed fits the data post-omega turn. The black dotted line shows the 0mA stimulus data. The reorientation event for the noxious response consists of a distribution of escape angles very close to 180 degrees (inset blue arrow, taken from Figure 3.1). CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 34 sented in Chapter 3 of specific mutant defects provides insight into the noxious response, this global comparative analysis also reveals a few interesting results.

Each mutant has a unique barcode describing a number of features extracted from the dataset. Shown in Figure 2.14 are the z-scores of 11 features for 16 of the strains tested with our 150mA 133ms noxious stimulus. The barcodes differ by strain as well as by laser pulse location. Features included are mean reversal start time, mean reversal duration, mean reversal duration standard deviation, mean forward duration, mean pause start time, mean pause start time standard deviation, mean pause duration, max mean speed, standard deviation max speed, reverse to pause transition probability, and reverse to omega transition probability. Please note, the tail response only shows 9 features because the last 2 features (reversal to pause and reversal to omega) do not apply to the stereotypical accelerated forward response.

Using these feature vectors, we can do a comparison amongst strains by looking at the correlation of each vector across the library of strains (Figure 2.15). Naturally every strain is 100% correlated with itself. I have highlighted a few noteworthy strains compared to N2 in Figure 2.15. These are three npr-1 alleles and akIs11.

The neuropeptide receptor gene npr-1 has been shown to increase the threshold of noxious heat in C. elegans (Glauser et al., 2011). In our quantitative mutant screen, we found that behavioral defects in three loss-of-function mutants, npr-1(n1353), npr-1(ad609) and npr-1(ky13) have a dependence on the location of the stimulus (Figure 2.15). This suggests that although npr-1 is expressed in many cell types in the nervous system, there may be a role of the different npr-1 alleles in the spatial discrimination of the noxious response.

The transgenic strain akIs11 contains an integrated array that expresses the nmr::ICE construct (human caspase) resulting in apoptotic death of the forward command interneurons PVC and AVB, as well as the backward command interneurons, AVA, AVD and AVE ((Zheng et al., 1999)). This strongly affects the head, midbody and tail responses (Figure 2.15), however for this discussion, consider the strong negative correlation of the akIs11 strain’s behavior for the midbody and tail compared to N2 in Figure 2.15. We have shown PVD mediates the mid- CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 35

Figure 2.14: Normalized (z-scores) of behavioral feature vectors for 16 strains for head, midbody, and tail stimulation. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 36

Figure 2.15: Correlation amongst strains using feature vectors from Figure 2.14. Highlighted are noteworthy strains compared to N2, namely three npr-1 alleles and akIs11. CHAPTER 2. THERMAL STIMULUS EXPERIMENT AND BEHAVIORAL PHENOTYPING 37 body and tail noxious responses, and PVD is known to synapse equally to AVA (27 synapses) and PVC (28 synapses). In akIs11 where both these interneurons are killed, the worm’s post- stimulus behavior is dominated by a pause state (see feature 7, Figure 2.14). This result in conjunction with our deg-1(u38) result (-PVC) discussed earlier suggests the PVD->AVA, PVC synaptic circuit is the primary ﬁrst layer in the circuit for the midbody noxious response, where a net excitatory synapse to either interneuron is required for initiation of the desired withdrawal direction. Chapter 3

Behavioral response of C. elegans to localized thermal stimuli

The text of this chapter contains sections reproduced from work as it appears in Mohammadi A, Byrne Rodgers J, Kotera I, Ryu WS (2013) Behavioral response of Caenorhabditis elegans to localized thermal stimuli. BMC Neurosci, 14:66. I am the ﬁrst author of this publication, and performed all the experiments and analysis included from the publication in this chapter unless otherwise noted, under the supervision of William Ryu.

3.1 Abstract

Background: Nociception evokes a rapid withdrawal behavior designed to protect the animal from potential danger. C. elegans performs a reflexive reversal or forward locomotory response when presented with noxious stimuli at the head or tail, respectively. Here, we have developed an assay with precise spatial and temporal control of an infrared laser stimulus that targets one-fifth of the worm’s body and quantifies multiple aspects of the worm’s escape response. Results: When stimulated at the head, we found that the escape response can be elicited by changes in temperature as small as a fraction of a degree Celsius, and that aspects of the escape behavior such as the response latency and the escape direction change advantageously

38 CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI39 as the amplitude of the noxious stimulus increases. We have mapped the behavioral receptive ﬁeld of thermal nociception along the entire body of the worm, and show a midbody avoidance behavior distinct from the head and tail responses. At the midbody, the worm is sensitive to a change in the stimulus location as small as 80 mm. This midbody response is probabilistic, producing either a backward, forward or pause state after the stimulus. The distribution of these states shifts from reverse-biased to forward-biased as the location of the stimulus moves from the middle towards the anterior or posterior of the worm, respectively. We identiﬁed PVD as the thermal nociceptor for the midbody response using calcium imaging, genetic ablation and laser ablation. Analyses of mutants suggest the possibility that TRPV channels and glutamate are involved in facilitating the midbody noxious response.

Conclusion: Through high resolution quantitative behavioral analysis, we have comprehensively characterized the C. elegans escape response to noxious thermal stimuli applied along its body, and found a novel midbody response. We further identiﬁed the nociceptor PVD as required to sense noxious heat at the midbody and can spatially differentiate localized thermal stimuli.

3.2 Background

The ability to sense and react to abrupt, painful changes in the environment is critical for an animal’s survival (Wittenburg and Baumeister, 1999; Glauser et al., 2011; Maroteaux et al., 2012). By evoking reﬂexive escape behaviors in response to potentially harmful stimuli, organisms are able to avoid possible tissue damage and minimize injury (Pirri and Alkema, 2012; Bromm and Treede, 1980). Vertebrates and invertebrates possess sensory neurons called nociceptors that detect noxious stimuli, such as harsh touch or acute heat (Chatzigeorgiou et al., 2010; Caterina et al., 1997; Xu et al., 2006). An animal generally senses these types of stimuli as harmful or potentially damaging, and protects itself with an escape response appropriate to the level of the threat. The nematode Caenorhabditis elegans responds to many types of CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI40 noxious mechanical, osmotic, and chemical stimuli (Kaplan and Horvitz, 1993; Li et al., 2011; Way and Chalﬁe, 1989; Hilliard et al., 2002; Tobin and Bargmann, 2004). Here we focus on the thermal noxious response of C. elegans.

C. elegans reacts to noxious temperatures at the head and tail (Wittenburg and Baumeister, 1999; Liu et al., 2012). At these extremities, the trajectory of the escape response of a crawling worm is deterministic – if stimulated in the head, the worm will reverse, and if stimulated in the tail, the worm will accelerate forward. Substantial work has been done on the molecular mechanisms of the head and tail noxious responses (Chatzigeorgiou et al., 2010; Wittenburg and Baumeister, 1999; Liu et al., 2012). Several neurons have been implicated in the sensation of noxious heat – the FLP and AFD neurons in the head, and the PHC neurons in the tail (Chatzigeorgiou et al., 2010; Liu et al., 2012). However, a midbody thermal nociceptor has not yet been identiﬁed. In light of the broader spatial receptive ﬁeld of mechanosensation (Pirri and Alkema, 2012; Li et al., 2011) the reported head and tail behavioral responses may be an incomplete characterization of the worm’s ability to respond to thermal noxious stimuli. Therefore, we performed high-content phenotyping of the worm’s thermal noxious response comprehensively along the body of the worm to characterize its spatial dependence.

To perform a systematic quantitative analysis of C. elegans’ response to localized thermal stimuli, we have developed an assay that allows for the precise spatial and temporal application of an infrared (IR) laser beam to the body of C. elegans, and captures its pre- and post-stimulus behavior. We identified key metrics that quantify the response and comprehensively mapped the behavioral receptive field of nociception, revealing a midbody response that is sensitive to very small changes in the stimulus location. Using a multi-dimensional measure of the midbody thermal response behavior, we identified a neuronal candidate (PVD) and a number of molecules (TRPV channels, glutamate) that are involved in the transduction of the nociceptive signals. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI41

3.3 Results

3.3.1 Novel assay for quantifying the noxious response and mapping the

behavioral receptive ﬁeld

To date, thermal avoidance assays have been useful in studying the molecular mechanisms of thermal nociception, but have lacked the ability to carefully control variable doses of heat along the body of the worm (Glauser et al., 2011; Chatzigeorgiou et al., 2010; Liu et al., 2012). In previous studies where regions along the body of the worm were targeted (Wittenburg and Baumeister, 1999; Liu et al., 2012), the laser focus was presented for a time long enough (10 seconds) for the heat to diffuse well beyond the worm’s body. Therefore the spatial extent of the stimulus in these experiments is uncertain because the temperature profile in time and space was not clearly shown. In other experiments of thermal nociception, the main drawbacks are that either the whole worm is heated or a thermal barrier selects for sensory neurons in the head (Glauser et al., 2011; Ghosh et al., 2012; Chatzigeorgiou et al., 2010); either case cannot spatially dissect the noxious response. We have addressed these limitations by designing a new thermal avoidance assay that localizes the heat from an IR laser pulse to small regions (approximately 1/5 of the worm’s body length) along the entire body of the worm, and records the behavioral responses to the noxious stimulus. The thermal profile of the beam was carefully calibrated using a thermal camera. The details of the assay are discussed in Chapter 2. The centroid worm speed and changes in the worm body shape were used to determine the behavioral states of the worm (reverse, forward, omega turn, or pause) before and after the thermal stimuli (Materials and Methods). Figure 3.1 is modified from Figure 1 in the Mohammadi et al. publication (Mohammadi et al., 2013) for inclusion in this chapter without redundancy to Chapter 2. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI42

Figure 3.1: Assay for the spatial dissection of the thermal noxious response. (A) An infrared laser pulse with a 220mm beam diameter (full width half maximum) was used to locally stimulate the entire worm from head to tail, N = 442. The head (1-10), midbody (16-26), and tail (31-41) regions are demarcated using the 41 points along the entire “skeleton” of the worm body. An example of the selective targeting of the head is shown. The probability of the ﬁrst behavioral state after the laser pulse is shown for head, midbody, and tail regions. The head, midbody, and tail responses are statistically different, p < 0.001, Fisher’s exact test. (B) Raw video data of the worm after a head-applied laser stimulus is shown as a time-lapse sequence. After the laser pulse is applied, the worm enters a reversal (red), followed by an omega turn (green), and then resumes its forward motion (blue). CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI43

3.3.2 Multi-parameter, high-content phenotyping of N2 noxious response

for the head, midbody, and tail

We characterized the reaction of N2 (wild type strain) to a range of temperature ramps along the entire body of the worm to measure the spatial nociceptive dose response. The change in the worm’s centroid speed over time is an informative measure of the thermal response because aspects of this time series change with the stimulus strength. We used this metric to quantify behavioral differences in response to variations in both the power and location of the stimulus (Figure 3.2a), similar to a previous study (Ghosh et al., 2012). Several features in the dose response scale with power, most notably the maximum mean speed, and the deceleration from the maximum mean speed. The general shape of the mean speed versus time curves also changes in response to the position of the thermal stimulus along the worm body; this is because these speed curves are a product of the underlying locomotory states, some of which change with the position and power of the IR laser. In order to examine these behaviors and further characterize the wild type noxious response, we generated ethograms for the different stimulus laser powers and locations (Figure 3.2b) (Zariwala et al., 2003; Albrecht and Bargmann, 2011). At lower laser powers, the behavior is stochastic; as the power increases, the worm’s response becomes more deterministic. From the ethogram, we identified another metric that discriminates the stimuli both by its location and intensity, namely the first behavioral state the worm enters after the stimulus: forward, reverse, and pause. We calculated the probabilities of the first response states (first behavioral states after the stimulus), and measured changes in these probabilities in reaction to changes in the position and power of the stimulus (Figure 3.1a).

The stereotypical withdrawal response for a crawling C. elegans thermally stimulated at the head is a reversal, followed by an omega turn, then a recommencement of forward motion (Figure 3.1b). The likely purpose of this behavioral series is to make a three-point turn to reorient the worm away from the noxious stimulus. Arguably the worm’s chance to escape danger improves if it is able to respond more quickly to the threat, and reorient itself so that instead of moving towards the hazard it is moving in the opposite direction (180°). We investigated if CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI44

Figure 3.2: Spatial high-content phenotyping of N2 thermal noxious response. (A) The mean centroid speed as a function of time quantiﬁes the dose-dependent differences for the head, midbody and tail responses. Laser is ﬁred at 1s (arrow). The control speed (CTRL) is offset by 0.25 mm/s for clarity. Shaded regions are SEMs. N > 50 for all DT > 0°C, N = 15 for CTRL. Axes for head and middle are the same as tail. (B) Ethograms separated by stimulation region (head, body, tail) for selected DT demonstrate the evolution of the behavioral states over time, and show the spatial and thermal variation of the response. (C) The reaction time of the worm in response to a head-applied laser stimulus as a function of laser power. (D) The probability distribution of the escape angle as the worm responds to a head-applied laser stimulus. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI45 the escape response improves as a function of the laser power, indicating that these avoidance behaviors changed appropriately for the noxious level of the stimulus. Our results show that the animal’s reaction time does in fact vary inversely with stimulus amplitude (Figure 3.2c) and that the escape angle increases towards 180° with increasing stimulus power (Figure 3.2d).

3.3.3 The noxious response is elicited by a temporal temperature gradi-

ent rather than a temperature threshold.

Previous studies have used high temperatures in the range of 30ºC-35ºC to study the noxious response in C. elegans (Wittenburg and Baumeister, 1999; Glauser et al., 2011; Liu et al., 2012; Chatzigeorgiou et al., 2010). In the context of studying the noxious response, the requirement for high temperature is expected since previous work on mammalian transient receptor potential (TRP) channels in sensory neurons show that a subset of TRPs – the TRP vanilloid group in particular – are gated by high temperatures generally >43ºC and have a steep temperature dependence (Caterina et al., 1997; Caterina et al., 1999; Voets et al., 2004; Basbaum et al., 2009). Remarkably, our dose response and temperature measurements show that the worm’s robust, stereotypical avoidance response to noxious stimulus at the head can be elicited by relatively small changes in temperature ( 1.4ºC) (Figure 2.8, Figure 3.2a). It appears that the  temperature ramp rate as opposed to the temperature change above a threshold induces the avoidance response. The ramp rate for the highest temperature stimulus in our dose response is ~9.4ºC/s, which is in the noxious range of previous experiments with higher absolute temperatures (Ghosh et al., 2012).

We tested our hypothesis that the ramp rate and not the absolute temperature jump is what produces the thermal nociceptive response by using thermal stimuli with a constant ramp rate but with different DTs (~5.9°C/s, DT = 0.22, 0.41, and 0.67°C) (Figure 3.3). When we stimulated the worm with these short duration, small amplitude thermal pulses, we were able to robustly elicit nearly identical noxious responses (Figure 3.3). When the ramp rate is lowered and the animal is stimulated with a similar temperature jump (DT/t ~1.5°C/s, DT = 0.2°C), CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI46

Figure 3.3: Withdrawal behavior is dependent on the ramp rate of the thermal stimulus. The mean speed proﬁle of wild type animals for three head-applied stimuli with the same ramp rate but different absolute temperatures (N = 10 for each DT), showing a noxious response for each. Animals stimulated with a lower ramp rate but similar DT do not exhibit withdrawal behavior, and are statistically different than those stimulated with a higher ramp rate. p < 0.05, Kruskal-Wallis test, Dunn’s multiple comparison.

the response is noticeably lower and statistically different compared to the noxious response elicited by the higher ramp rate but same DT (Figure 3.3; Kruskal-Wallis test, Dunn’s multiple comparison p < 0.05). This indicates that the avoidance response is dependent on a rate of change in temperature, rather than a crossing of a thermal threshold. Previously reported experiments also stimulated worms with an abrupt change in temperature (Wittenburg and Baumeister, 1999; Glauser et al., 2011; Chatzigeorgiou et al., 2010; Liu et al., 2012), but our results show that extreme heat is not required to initiate a noxious response if the DT/t is above some threshold. For our experiments we stimulated the worm for a ﬁxed duration (133ms) at different laser powers to produce a range of DTs and ramp rates. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI47

3.3.4 Spatial sensitivity of the midbody response

The midbody behavior is distinct from the head and tail responses (Figure 3.1a; p < 0.001, Fisher’s exact test). At the two extremities of a forward moving worm, the transition to the first state after the stimulus is deterministic—the worm will reverse if stimulated in the head, and will accelerate its forward motion if stimulated in the tail. At the midbody, however, the response is probabilistic as the worm enters a reversal, a forward, or a pause state (Figure 3.1a). The forward or reverse bias of this behavioral response is strongly correlated with the anterior/posterior position of the stimulus. For example, a laser pulse directed to the anterior middle region closer to the head of the worm will cause a reversal the majority of the time, whereas a laser pulse targeted at the posterior middle has a higher probability to elicit a forward response (Figure 3.4). We uncovered the “sensory middle” of the worm–a region where the worm may move forward, move backward, or enter a pause state, roughly with equal probability. The brief midbody pause state could arise as a behavioral strategy when there is insufficient asymmetry in the signal, in order to give the worm another opportunity to accrue additional anterior/posterior information about the stimulus before initiating its escape. As the stimulating beam is moved a small distance around this “sensory middle” we can measure a change in the probability of the behavioral response. A statistically significant change in behavior suggests that the worm perceives a difference in the stimuli location. Using this measure we found that statistically the worm has the ability to spatially differentiate the location of our constrained thermal stimuli by as little as 80 microns (Figure 3.4; p < 0.05, Fisher’s exact test).

3.3.5 Mutant behavioral analyses identify neurons involved in the mid-

body and tail responses

Mutations in the gene mec-3 affect the development of the bilaterally symmetric pair of nociceptors PVD, such that the neurons lack all but the primary dendritic branching (Way and Chalﬁe, 1989; Tsalik et al., 2003; Smith et al., 2010; Albeg et al., 2011). We found that mec- CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI48

Figure 3.4: Spatial sensitivity of the midbody thermal noxious response. The probability of the ﬁrst behavioral state after the laser pulse as a function of stimulus location along the worm body is shown. At the midbody the worm is spatially sensitive to changes in beam location as small as 80 microns, and modulates its behavior accordingly. * p < 0.05, Fisher’s exact test. N = 442 total animals, N > 20 each bin. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI49

3(gk1126) had a pronounced defect in the midbody and tail response compared to N2 (Figure 3.5a; p < 0.01, Fisher’s exact test), but the head response showed only a very minor defect (3.5a; p > 0.05, Fisher’s exact test). PVD has been shown to be the nociceptor for harsh midbody touch (Chatzigeorgiou et al., 2010; Way and Chalﬁe, 1989; Li et al., 2011), and these results strongly suggest that PVD is also the nociceptor for the midbody thermal avoidance response. Since a mutation in mec-3 also affects the touch receptor neurons (ALM, AVM, PLM, PVM), we tested the touch resistant mec-4(e1339) mutant strain (O’Hagan et al., 2004) to ensure that the touch neurons were not involved. Our behavioral and speed data show that the mec-4 mutant response is statistically similar to wild type (Figure 3.5a; p > 0.05, Fisher’s exact test). Since the touch neurons are not involved in transducing the response this leaves PVD as the primary candidate for thermal nociception at the midbody.

Furthermore, PVC has been identiﬁed as a command interneuron for the forward tail noxious heat response, being a main synaptic output to the PHC neuron (Liu et al., 2012). PVC is also postsynaptic to PVD (White et al., 1986; Husson et al., 2012). Our deﬁnition of the tail region (Materials and Methods) includes the posterior branching of PVD. Our results show a severe defect in deg-1(u38)–a mutant where PVC is degenerated along with four other cell types–in the tail response (3.5b; Kruskal-Wallis test, Dunn’s multiple comparison p < 0.0001). The tail defect seen in the mec-3(gk1126) result (Figure 3.5a) implicates PVD as a possible nociceptor for the tail response, suggesting that PVC is acting as the command interneuron in the thermal avoidance circuit in the tail as a postsynaptic target to both PVD and PHC.

PVD chemically synapses equally to AVA (27 synapses) and PVC (28 synapses) (White et al., 1986). Recent optogenetic analysis of PVD has shown that the probability of backward versus forward movement is determined by the relative synaptic input to the command interneurons as a result of the location of the stimulus along the body of the worm (Husson et al., 2012). We investigated this by further analyzing deg-1(u38) (-PVC) upon noxious stimulation along the body, compared to the wild type response. The loss of functionality of the forward command interneuron effectively shifts the wild type midbody response to the posterior of the CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI50

Figure 3.5: Strains exhibiting spatially defective thermal avoidance behaviors. (A-B) The probability of the first behavioral state after the stimulus and the mean speed profiles showing defective behavior compared to N2 for (A) mec-3(gk1126), ** p < 0.01, Fisher’s exact test; behavior compared to N2 for (A) mec-4(e1339) at the midbody and tail, p > 0.05, Fisher’s exact test; defective behavior compared to N2 for (B) deg-1(u38) at the tail, Kruskal-Wallis test, Dunn’s multiple comparison p < 0.0001. N = 16 and N = 17 for mec-3(gk1126) noxious stimulus at midbody and tail respectively, N = 13 mec-3(gk1126) CTRL. N = 8 and N = 15 for mec-4(e1339) at midbody and tail respectively. N = 36 for deg-1(u38) noxious stimulus at tail, N = 15 deg-1(u38) CTRL. (C) The probability of the first state (Reverse, Forward or Pause) after the stimulus as a function of stimulus location along the body of the worm for N2 and deg-1(u38). N = 442 total animals for N2, N = 91 total animals for deg-1(u38). CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI51 worm, and increases the probability of entering a pause state (3.5c). This result suggests that the worm’s nociceptive “sensory middle” may be determined by the balance of the synaptic inputs to the command interneurons, and elicits spatially sensitive behavior accordingly (reversal for anterior stimulation, forward for posterior stimulation, and an increased pause state when the signal is symmetric). When the forward command interneuron PVC is not functional, the anterior response (reversal) extends to the posterior, since the PVD-AVA activity dominates.

3.3.6 PVD is required for the midbody and tail thermal noxious response

Our mec-3 and mec-4 results suggest that PVD is the sensory neuron underlying the midbody noxious response. In order to confirm its involvement, we eliminated the pair of PVD neurons using laser ablation microsurgery (Bargmann and Avery, 1995). We ablated both PVDL and PVDR in the late L2 stage. We then tested the response of PVD-ablated young adults to localized thermal stimuli. Our results show the head response is consistent with the mock- ablated response, but the midbody and tail responses are severely reduced (Figure 3.6, left column; Kruskal-Wallis test, Dunn’s multiple comparison p < 0.001). We also tested a transgenic strain, ser-2prom3:DEG-3-N293I, where PVD is specifically eliminated (Albeg et al., 2011). These genetic ablation results confirm our laser ablation results – the head response remains unaffected, but we found a clear decrease in the midbody and tail compared to the wild type response (Figure 3.6, right column; Kruskal-Wallis test, Dunn’s multiple comparison p < 0.0001). This demonstrates that the PVD neurons are required for the sensation of noxious heat at the midbody and tail.

C. elegans possesses more motor neuron commissures on the right side compared to the left side and accordingly there are more fasciculations with motor neuron commissures from the secondary branches of the PVDR neuron compared to the PVDL neuron (Smith et al., 2010; White et al., 1986). This left/right asymmetry led us to investigate the single neuron contribution to the midbody thermal noxious response. Using the same method as the double neuron ablation, we ablated either PVDL or PVDR and tested the head, midbody, and tail CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI52

Figure 3.6: The PVD sensory neurons mediate the thermal noxious response in the midbody and tail. Worms with laser ablated PVD neurons (left column) show statistically signiﬁcant defective thermal avoidance behavior compared to mock ablated worms when stimulated in the midbody and tail ( Kruskal-Wallis test, Dunn’s multiple comparison p < 0.001); the head response is similar to the mock ablation result. N = 9 ablated worms, N = 62 total experiments. ser-2prom3:DEG-3-N293I worms have genetically ablated PVD neurons (right column), and show severe defects in the noxious thermal response for the midbody and tail compared to N2 (Kruskal-Wallis test, Dunn’s multiple comparison p < 0.0001). The head response is unaffected. N > 12 for each region. Top row: head, middle row: midbody, bottom row: tail. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI53 responses to our noxious thermal stimulus. The speed versus time analysis suggests that either PVDL or PVDR alone is sufﬁcient since the single neuron ablation data at the midbody and tail respond to the noxious stimulus – although slightly less robustly – compared with mock ablation data (Figure 3.6 left column).

3.3.7 PVD responds differently to spatially localized heat pulses targeted

at different locations near the midbody

Calcium transients in a cell are often used as a proxy of neuronal activation in C. elegans. Thus by monitoring intracellular calcium dynamics, we can monitor neuronal activity in living worms. To show PVD senses localized noxious heat at the midbody, and that it can differentiate the location of the stimuli, we used a G-GECO 1.2 calcium indicator coexpressed with a reference DsRed2 chromophore in the nuclei (strain generated by Ippei Kotera, see Materials and Methods in (Mohammadi et al., 2013) for details) to measure the inﬂux of calcium into PVD when the midbody is heated with an IR laser pulse at two distinct locations, namely anterior to the cell body and posterior to the cell body (Figure 3.7a, 3.7b). The IR stimulus used for calcium imaging was nearly identical to the one used in the behavior measurements (Materials and Methods). Previous studies applied heat to the whole worm (Chatzigeorgiou et al., 2010; Liu et al., 2012), which may have selected for a head response and a calcium inﬂux in PVD may not be seen as a result. We measured calcium transients in PVD when stimulated with a 133ms pulse of heat at a location anterior to the cell body and a location posterior to the cell body (Figure 3.7b). The thermal maxima of the anterior and posterior pulses were approximately

200mm apart, and fall within the spatially distinct behavioral responses at the midbody that we found in Figure 3.4. Our calcium transients conﬁrm PVD’s role in transducing the midbody avoidance response. Furthermore, the difference in the signal due to changing the location of the stimulus demonstrates the neuron’s ability to detect a difference between anterior versus posterior stimulation (Figure 3.7b). CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI54

Figure 3.7: The PVD neurons show a spatially dependent response to localized noxious heat pulses at the midbody. A) Sample raw data of G-GECO 1.2 calcium indicator in PVD (green line) and DsRed2 reference chromophore in PVD (red line) during data acquisition, IR pulse at arrow. B) Calcium response to IR laser pulse at the midbody for anterior (red line) and posterior (blue line) stimulation. Solid lines represent the average normalized ratio of G- GECO 1.2 emission to DsRed2 emission in PVD for N = 8 anterior midbody recordings and N =7 posterior midbody recordings. The arrow is where the IR pulse is applied. Shaded region is SEM.

3.3.8 Mutant strains show defective noxious behavior suggesting molecules

involved in sensing heat at the midbody

Our quantitative analysis of 21 mutant strains revealed previously reported results for molecules involved in thermal avoidance for the head and tail (Wittenburg and Baumeister, 1999; Liu et al., 2012). The results are summarized in Table 3.1. Two features of the behavior are included in the table, namely the max mean speed and standard deviation. Statistical differences from the wildtype strain are identified for the head, middle, and tail. Please note that this is only a summary, and all the features from the behavioral profiling are shown and discussed in the behavioral barcode discussion in Chapter 2. Statistical tests were done on multiple features and a strain was considered statistically significantly different from N2 if any of those features were statistically significantly different. Here, we focus on those that affect the midbody thermal avoidance response. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI55

Table 3.1: Mutant strains used for thermal nociception assay. Note: Only two metrics from the behavioral quantiﬁcation – max mean speed and standard deviation (MMS ± STD) – are included in this summary. All the features are discussed in Chapter 2. Units are pixels/frame. pN2 : p-value compared to N2, Kruskal-Wallis test or Fisher’s exact test performed for all extracted features. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI56

Figure 3.8: Potential molecules spatially mediating the thermal avoidance response at the midbody. (A-B) The probability of the ﬁrst state after the laser stimulus at the midbody is statistically different compared to N2 for the (A) glr-1(n2466) mutant strain (** p < 0.01, Fisher’s exact test), and (B) the ocr-2(vs29) mutant strain (* p < 0.05, Fisher’s exact test). The osm- 9(ky10) strain shows the same probabilistic midbody response as N2.

The GLR-1 glutamate receptor has been shown to mediate mechanosensory signaling in interneurons postsynaptic to the polymodal nociceptor ASH, and discriminate between sensory inputs (Kaplan and Horvitz, 1993; Maricq et al., 1995; Hart et al., 1995). In particular, glr-1 mutants are defective to nose touch but not to osmotic shock, even though both modalities are primarily sensed by ASH. GLR-1 is expressed in interneurons controlling locomotion (Zheng et al., 1999), including PVC and AVA which are both direct postsynaptic outputs to PVD. Our results show that a mutation in glr-1(n2466) produces a strong midbody behavioral defect (Fig- ure 3.8a). In particular, the probability of forward locomotion is reduced from 0.18 to 0, the probability of backward motion is reduced from 0.65 to 0.46, and the probability of the pause state dramatically increases from 0.17 to 0.54, relative to the wildtype response. This defective behavior suggests that glutamate could be the transmitter for PVD in the midbody thermal noxious response. This is consistent with the ﬁnding that PVD expresses the vesicular glutamate transporter EAT-4, which is required for glutamatergic transmission (Lee et al., 1999).

The TRPV1 subfamily channels are involved in noxious heat perception in humans and mice (Caterina et al., 1999; Basbaum et al., 2009; Caterina et al., 2000). Recently TRPV channels have been found to contribute to the thermal avoidance response in the head and the tail CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI57 of C. elegans (Glauser et al., 2011; Liu et al., 2012), and so we investigated their involvement in the midbody thermal noxious response. OCR-2 and OSM-9 are homologues of the mammalian TRPV channel genes in C. elegans, and are coexpressed in sensory neurons (Tobin et al., 2002). Both are expressed in PVD (Jose et al., 2007; Colbert et al., 1997; Kindt et al., 2007). Our results suggest that ocr-2 is required for noxious heat sensation at the midbody, but osm-9 is not (Figure 3.8b). Therefore, it is possible that the OCR-2/OSM-9 heteromer does not function in PVD to control the noxious heat response, as only ocr-2 produces a behavioral defect.

3.4 Discussion

We developed a high-content thermal nociception assay with precise control of the location of the stimulus in order to spatially dissect the thermal noxious response in C. elegans. By reading the “body language” of C. elegans as a function of stimulus position, we uncovered a number of new features of thermal nociception, including a midbody response distinct from the known head and tail responses (Wittenburg and Baumeister, 1999; Liu et al., 2012). Previous studies (Wittenburg and Baumeister, 1999; Glauser et al., 2011; Chatzigeorgiou et al., 2010; Liu et al., 2012) used a large change in temperature (>10 °C) to elicit escape response, but here we showed that stimuli as small as a fraction of a degree can elicit a response if the change in temperature is fast. In addition the behavioral features of the worm’s response such as the reaction time and the escape angle changes in a way that might be favorable to the worm as the level of the noxious stimuli increases.

Our results also show that the worm can respond to thermal stimuli localized to the midbody. The pair of polymodal nociceptors PVD possess a dendritic arbor that covers most of the worm’s body (Smith et al., 2010; Albeg et al., 2011), and have been shown to sense aversive stimuli such as harsh touch and cold shock (Chatzigeorgiou et al., 2010; Li et al., 2011; Way and Chalﬁe, 1989). Through genetic tools, laser ablation, and calcium imaging, we have CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI58 conﬁrmed PVD’s involvement in sensing an abrupt increase of heat at the midbody and tail.

Since PVD covers the majority of the worm’s body, it is expected that it would have a large receptive field. Using genetic and neuronal ablation, we were able to delineate the thermally- stimulated receptive field of PVD to the middle and tail regions of the worm. We also generated a behavioral receptive field map for PVD by analyzing the worm’s response to the thermal stimulus as a function of stimulus location along the worm’s body. In doing so, we discovered that C. elegans is able to discriminate stimuli at the midbody with a spatial sensitivity of at least 80 microns. This result implies that PVD could be used as a model nociceptor for the study of spatial differentiation of noxious stimuli by a single neuron.

In addition to PVD, we investigated the spatial sensitivity of the midbody response related to the differential synaptic outputs to command interneurons AVA and PVC. It has been recently suggested that relative synaptic inputs to command interneurons due to the position of the stimulus modulate the forward/backward locomotion of the worm (Husson et al., 2012). Our measure of the -PVC worm’s behavioral receptive ﬁeld revealed an extension of AVA initiated reversals into the posterior region of the worm, as well as in increase of pausing in the posterior. In effect, the worm’s “sensory middle” is shifted towards its tail. With PVC not functioning there is no differential excitation of the command interneurons to initiate a spatially biased withdrawal behavior, and the increased pause state in the far posterior may be due to the dominating bias of AVA on the response.

We also discovered a defect in the midbody response of the mutant glr-1. Glutamate receptors play an important role in polymodal nociceptors, as they may serve to select for different stimulation modalities (Kaplan and Horvitz, 1993; Maricq et al., 1995; Hart et al., 1995). GLR-1 has also been implicated in long term memory formation (Rose et al., 2003), as well as the control of locomotion in foraging (Zheng et al., 1999). Our assay found a defect in the midbody response for the mutant glr-1(n2466) (Figure 8a), and this allele is expressed in the command interneurons AVA and PVC (Hart et al., 1995). This suggests a role for glr-1 and glutamate in thermal nociception through PVD. CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI59

The utility in quantifying the noxious response goes beyond investigating the midbody thermal avoidance behavior in C. elegans. The establishment of C. elegans as a model organism for nociception requires a comprehensive quantitative analysis of its wild type behaviors to serve as a benchmark in screening for defects caused by genetic, neuronal and pharmacological factors. We generated a dose response and identiﬁed several features that scaled with stimulus amplitude and can be used as a measure of nociception. Of note, the maximum mean speed of the response, the probability distribution of the ﬁrst behavioral state after the stimulus, the reaction time, and the escape angle are all correlated to the strength of the stimulus. Interestingly, we observed that the reaction time is proportional to the logarithm of the stimuli strength, which suggests so-called logarithmic sensing (Kalinin et al., 2009) consistent with Weber-Fechner (Thompson, 1967) in the sensorimotor transformation for thermal stimuli in C. elegans. Regarding the escape angle of the animal, the articulation of the omega turn in the head noxious response modulates the escape trajectory. On average, the omega turn happens several seconds after the stimulus is presented (Figure 2b). Yet, our results show that the worm’s measurement of the strength of the stimulus is incorporated in the omega turn (Figure 2d). In fact, no worms reorient themselves more than 170° when presented with a less harmful stimulus of 30mA, while 76% of worms do so when the stimulus is more intense (150mA). In addition the response time decreases and the reversal duration increases as the noxious level of the stimulus increases (Ghosh et al., 2012). Further, the wild type midbody pause state could possibly be explained as a behavioral strategy to allow the worm more time to accrue additional information about the location of the stimulus before breaking the symmetry, and to increase the chances of choosing the trajectory that effectively directs the worm from the danger. C. elegans has evolved a complex, multi-faceted behavioral response to a noxious response with multiple features that change in a coordinated way to produce an adaptive protective behavior.

A conceptually similar attempt to quantify pain in animal models is the recent generation of the Mouse Grimace Scale – a catalog of laboratory mouse facial expressions that are meant to quantify the amount of pain felt by acute stimuli (Langford et al., 2010). Studies such as CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI60 these, while promising, have inherent limitations because mammalian behaviors are very complex and difﬁcult to quantify. Furthermore, these animals have a long pre-stimulus history that integrates many environmental stimuli that may confound the pain response. C. elegans can be quickly grown in identical conditions and environmental stimuli controlled, making it a desirable model organism for the study of nociception. Furthermore, there is a molecular similarity in thermal nociception or pain among vertebrates an invertebrates (Tobin and Bargmann, 2004). An example of this overlap is the TRPV channel OCR-2 expressed in sensory neurons, which we suggest is required for thermal nociception in PVD. Even though there will be differences in vertebrate and invertebrate nociception, this work adds to the growing evidence that investigating thermal nociception in C. elegans may help in understanding this sensorimotor transformation in higher organisms.

3.5 Conclusion

We have developed a novel assay to spatially dissect and quantify the C. elegans thermal noxious response with high resolution. The C. elegans avoidance response is a multi-faceted behavior where several features change in a way to improve the animal’s escape. Our analysis revealed a spatially sensitive midbody response distinct from the head and tail responses. The nociceptor PVD is required for the sensing of heat at the midbody, and has the ability to spatially differentiate localized stimuli.

3.6 Materials and Methods

C. elegans strains were grown following standard procedures. Worms were prepared following a previous protocol used for thermal sensory behavioral measurement (Ghosh et al., 2012). Behavior was quantiﬁed using custom programs written in LabVIEW and MATLAB as previously published (Ghosh et al., 2012). The thermal stimulus was measured using a thermal IR camera (ICI 7320, Infrared Cameras Inc, TX) and conﬁrmed by ratio dye imaging. Calcium CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI61 imaging was done using standard techniques with a dual EMCCD Nikon TI (Nikon, USA).

Strains

Strains were cultivated at 20°C on NGM plates with E. coli OP50 according to standard protocols (Brenner, 1974). The strains used in this work were: Wild type N2, deg-1(u38), ﬂp- 21(ok889), glr-1(n2466), mec-3(1338), mec-3(gk1126), mec-4(e1339), mec-10(e1515), mec- 10(tm1552), npr-1(ad609), npr-1(ky13), npr-1(n1353), ocr-2(vs29), osm-6(p811), osm-9(ky10), sem-4(n1378), tax-2(p671), tax-4(p678), trpa-1(ok999), ttx-1(p767), unc-86(n846), obtained from the Caenorhabditis Genomics Center. We also used ser-2prom3:DEG-3-N293I,ser-2prom3:DEG-3-N293I;mec-4(e1611), and mec- 10p:DEG-3-N293I, obtained from the Treinin Lab. The transgenic strain akIs11 was obtained from Rajarshi Ghosh, Kruglyak Lab, Princeton University.

Thermal stimulus assay

Worms were assayed in a temperature-controlled room (22.5°C±1°C). Images were obtained using a Leica MZ7.5 stereomicroscope and a Basler firewire CMOS camera (A6021-2; Basler, Ahrensburg, Germany). A 2mm diameter collimated beam through a 100mm focal length lens from a 1440 nm diode laser (FOL1404QQM; Fitel, Peachtree City, GA) was focused at the surface of the agar, near the center of the camera’s field of view (Figure 1a). The diode laser was driven a Thorlabs controller (LDC 210B and TED 200C; Thorlabs, Newton, NJ). A custom program written in LabVIEW (National Instruments, Austin TX) was used to control the IR laser firing, power, and duration, while simultaneously recording images of the crawling worm at 30 fps for 1 second of pre-stimulus behavior, followed by 15 seconds of post-stimulus behavior. The plate was moved at least 1 second prior to the laser firing so that a random location along the worm’s body was targeted by the laser. The laser and worm positions were simultaneously recorded so the precise location of the pulse when fired is known. The con- CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI62 trol data for all datasets (examples 0mA in Figures 2a and 5a,b) show no discernable effect from the careful movement of the plate. Only forward moving worms were assayed, and each worm was stimulated only once. Images were processed offline using custom programs written in MATLAB (Mathworks, Natick, MA). A thermal camera (ICI 7320, Infrared Camera Inc., Texas, USA) was used to measure the temperature of the agar when heated by the IR laser (Figure 1d). The temperature change caused by the 10mA dose was below the resolution of the thermal camera, so the reported value is from the ratiometric temperature measurement described in (Mohammadi et al., 2013). We also measured the thermal profile of the beam with an anesthetized worm at the center, and confirmed that the measurement of the temperature of the agar is the same as the measurement of the worm’s temperature (see Chapter 2). This indicates that the thermal capacity and conductivity of the worm is nearly identical to that of agar. The noxious stimulus used for the wild type, genetic ablated and mutant analyses was 150mA. The cameleon strain required a higher dose to elicit the noxious response in the tail, therefore 300mA was used to assay the ablated and mock ablated animals.

Neuronal ablation

Laser ablation of PVD was performed essentially as previously described, by focusing a 440nm <4ns pulsed dye laser (Duo-220; Laser Science Inc., Franklin, MA) pumped by a nitrogen laser (337205-00, Spectra-Physics, Santa Clara, CA) onto the imaging plane of an inverted microscope (TE-2000E; Nikon) with a NA1.4 100x oil objective (Plan Apo 100X/1.40; Nikon) (Fang-Yen et al., 2012). Mid-L2 stage C. elegans expressing cameleon in PVD were used to identify the target neurons and align them with the position of the laser beam. The success of each ablation was conﬁrmed using disappearance of YFP ﬂuorescence and the elimination of harsh touch behavioral response.Worms were immobilized for imaging on pads made from 5% or 10% agarose dissolved in M9 buffer with 0.25 µl of 0.1 µm diameter PolyStyrene mi- crospheres (PS02N; Bangs Laboratories, Fishers, IN) 2.5% w/v suspension in M9 buffer. The worms were recovered by transferring the entire pad to an NGM plate with ample food, and CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI63 releasing the worms from the beads with M9 buffer.

Calcium imaging

For calcium imaging of PVD’s response to heat at the midbody, we used the same IR thermal stimulus as for the ratiometric Rhodamine B/110 temperature measurement (described in (Mohammadi et al., 2013)), mounted on top of a Nikon Eclipse Ti microscope with dual EMC- CDs (DU-897E-CSO-BV; Andor, Belfast, UK). Optical recordings were obtained from worms expressing a pan-neuronal G-GECO 1.2-based calcium sensor and reference chromophore DsRed2 in the nuclei of the cells. Recordings were done at the cell body, while the IR pulse was focused on the dendrites anterior and posterior to the cell body, the same distance away from the cell body. Neutral density filters were used to limit photobleaching. The following filter sets were used: (i) 495nm dichroic (T495lpxr; Chroma) with 470nm bandpass (40nm) filter (ET470/40x; Chroma), and (ii) 538nm dichroic (FF538-FDi01, Semrock) with 531nm long pass filter (FF02-531/22, Semrock) and 593nm long pass filter (FF01-593/LP-25; Semrock). Worms were immobilized as they were for neuronal ablation.

Statistical analyses

Statistical analyses was performed in Matlab. For ANOVA comparing mutant strains to N2, I performed the Kruskal-Wallis test (Siegel, 1956) followed by Dunn’s multiple comparison test (if necessary) for each of the behavioral metrics described in Chapter 2. If any of these features were signiﬁcantly different (p < 0.05) from N2, that strain was considered statistically signiﬁcantly different. For comparing the probabilities of multiple states within a region at a particular time (for example, comparing the probabilities for two strains of reverse, forward, or pause states (concurrently) 2 seconds after the zap for a head stimulus), I transformed the probabilities to counts and performed Fisher’s exact test to compare the distributions (Siegel, 1956). Fisher’s exact test allows for low and zero counts. My contigency table used here was 2x3: rows were strains CHAPTER 3. BEHAVIORAL RESPONSE OF C. elegans TO LOCALIZED THERMAL STIMULI64 or region (two at a time), and columns were counts for reverse, forward, and pause. Chapter 4

C. elegans as a model system for pain

4.1 Introduction

As frustration grows with failures to promote basic research to clinical applications in the field of pain research (Mogil, 2009; Hill, 2000; Wallace et al., 2002a; Wallace et al., 2002b), it may seem counterintuitive to suggest an invertebrate model, C. elegans, as a model system for pain. Could the field really benefit from a model organism that has even less translational relevance to humans than the commonly used rat, mouse, and other mammalian models? In fact, in the face of clinical endpoint failures from research in current animal models, some have suggested the abandonment of animal models and the reliance on human volunteers for increased efficacy in pain research (Langley et al., 2008). This, however, clearly has its own set of ethical and practical complications (Mogil, 2009). I do not wish to overstate the translational potential of C. elegans as a model organism for pain (at the very least, at the molecular level there are some similarities between the worm and vertebrate models (Tobin and Bargmann, 2004)), or in any way minimize the need for higher animal models in pain research. However, I propose that, taken together, (i) the established genetic and neuronal tools that C. elegans brings to the table (ii) the novel instrumentation and careful behavioral quantification already discussed in this thesis, and (iii) the demonstration in this chapter that the nociceptive response in C. elegans

65 CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 66 is modulated by clinically relevant pain drugs, suggests C. elegans is a beneﬁcial addition to address some current limitations of animal models in this challenging ﬁeld.

To put this work in a greater context, one compelling example supporting the use of C. elegans to elucidate mechanisms underlying human conditions and potential clinical applications is the lung cancer work done in Ravi Salgia’s lab at the University of Chicago (Siddiqui et al., 2008; Loganathan et al., 2011). A great challenge in developing targeted cancer therapeutics is working with the speciﬁc genetic mutations of a given tumor. The usual choice for preclin- ical studies targeting genetic mutations in vivo is the mouse model, but producing transgenic strains takes a relatively long time, and there are numerous pathways and mutations that make the analysis challenging (Siddiqui et al., 2008). C. elegans however allows for the rapid pro- duction of transgenic lines to study mutations and possible inhibitors in vivo, making it a very powerful tool. The Salgia lab was the ﬁrst to implant a human lung cancer mutation, c-Met, into the worm, and observed a strong phenotype of abnormal vulvar development and fecun- dity (Siddiqui et al., 2008), demonstrating the human mutation can in fact be studied in the worm. Furthermore, exposure to nicotine enhances this phenotype (Siddiqui et al., 2008). c- Met mutations are present in lung cancers found in smokers, although not all smokers develop lung cancer (Siddiqui et al., 2008). These results strongly support the utility of C. elegans as a model organism to study molecular mechanisms underlying targeted gene therapy in cancer, and the commonality between environmental toxins (such as nicotine) and enhanced oncogene activity between C. elegans and mammals make the worm an exciting tool for rapid screening of targeted inhibitory drugs (Siddiqui et al., 2008). In this chapter, I hope to demonstrate the utility of C. elegans as a model organism for the study of pain by identifying shared molecular and pharmacological characteristics between the worm and mammalian systems using opioids as well as a mutant strain with defective receptor NMUR-2. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 67

4.1.1 Animal models in thermal pain research

In the classical acute thermal pain experiments for rodent models (Mogil et al., 2006; Le Bars et al., 2001; Eaton, 2003; Mogil, 2009), the heat is applied to an accessible location on the animal, and the latency of the withdrawal response is measured. The two most common thermal assays are the hot plate and tail withdrawal tests (Eaton, 2003; Mogil et al., 2006). In the hot plate test, the animal is placed on to a metal surface heated to a constant temperature, and the time to the ﬁrst nociceptive behavior (paw licking or jumping) is recorded (Le Bars et al., 2001; Mogil et al., 2006). For the tail withdrawal tests, the animal is constrained (a necessity, but a disadvantage for a behavioral assay), and a portion of the animal’s tail is heated either by immersion in a water bath or radiant heat source (Le Bars et al., 2001; Eaton, 2003; Mogil et al., 2006). In both these assays, several factors confound the spatiotemporal proﬁle of the thermal stimulus felt by the animal. These include skin pigment affecting heat absorption, variability in the area heated, and variability in the contact time between the animal and the heat source (affecting peak stimulus intensity and heat localization) (Le Bars et al., 2001). This raises an important question when examining results of such studies – is variability in behavior due to differences in nociceptive sensitivity, or to variation in the stimulus? This lack of experimental precision is addressed in our laser assay described in Chapter 2, and aided by the worm’s small size amenable to freely crawling behavioral experiments.

Furthermore, the difficulty to precisely control the locus of heating in these assays has lim- ited research of the spatial sensitivity of a painful response. For example, studies published over 20 years ago reported spatial sensitivity in the rat tail when administering the tail withdrawal test, both with and without opiates (Yoburn et al., 1984; Ness et al., 1987). Interestingly, this finding is paradoxical because the latency of the response decreases as the stimulus is moved distally along the tail, even though the afferent path is longer (Ness et al., 1987; Le Bars et al., 2001). This is reminiscent of the study by Mancini and colleagues using human subjects to examine the spatial resolution of the fingertips versus the hand, discussed in Chapter 1 (Mancini et al., 2013). However, without adequate instrumentation to investigate this further in current CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 68 animal models, the spatial sensitivity of the tail response has remained unexplained, and is still merely acknowledged as a drawback of the tail withdrawal experiment (Le Bars et al., 2001; Wilson and Mogil, 2001). Our spatially localized thermal assay for C. elegans would be a nice model to further study the spatial sensitivity of the nociceptive response.

Another limitation common to pain assays is the difficulty in assessing nonverbal communication of pain in animals (Keefe et al., 1991; Langford et al., 2010; Le Bars et al., 2001). We cannot ask an animal how much pain it feels, and in human studies that rely on personal assessment, the reported level of pain is a subjective measure. Therefore, the ability to read the model organism’s “body language” in order to objectively measure the perceived level of pain under the effect of a noxious stimulus, analgesics, or hyperalgesics is desirable (Keefe et al., 1991; Langford et al., 2010). Mammalian behaviors are very complex and difficult to quantify. The need for better quantification in behavioral phenotyping has been addressed to some extent in the recent development of the Mouse Grimace Scale, a three point rating of five facial features of the mouse in an effort to increase the accuracy of pain measurements (Langford et al., 2010). As demonstrated in chapters 2 and 3, C. elegans is highly amenable to machine vision techniques that allow objective, quantitative behavioral phenotyping, thus simplifying and removing subjectivity from the evaluation of the pain response.

Finally, control of environmental factors is something that must be considered in all animal experiments. As mentioned in the discussion in Chapter 3, these animals have a very long pre- stimulus history that integrates many environmental stimuli (handling, housing, diet, social interaction, etc.) which may confuse the response. Concerns to this end have been raised in the context of validity of comparisons in transgenic experiments (Wilson and Mogil, 2001). C. elegans can be grown in identical conditions and environmental stimuli carefully controlled, addressing this limitation. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 69

4.1.2 Opioid pharmacology

Bateson’s criteria (Bateson, 1991) for assessment of pain in animals highlights certain consid- erations that, taken together, should be evaluated when judging an animal’s ability to feel pain. Summarizing, these criteria are:

1. Possession of suitable receptors and pathways that process noxious stimuli

2. Avoidance learning

3. Protective motor reaction (avoidance response)

4. Opioid receptors

5. Reduction of nociceptive response by analgesics

6. Higher cortical function

In bold are the relevant criteria for this chapter, namely, evidence for the existence of an opioid receptor in C. elegans, and the modulation of the nociceptive response due to analgesics. As an aside, Chapter 3 demonstrates criteria 1 and 3, and work in our lab is ongoing at this time to evaluate the second criterion (avoidance learning) for our thermal stimulus.

Soon after Pert and Snyder’s discovery of opioid binding proteins in the rat brain in 1973 (Pert and Snyder, 1973), the search for opioid receptors in the mammalian brain began. There are putatively three distinct receptor subtypes (all G-protein coupled receptors) activated by opioids: the delta opioid receptor (DOR), the mu opioid receptor (MOR) and the kappa opioid receptor (KOR) (Goldstein and Naidu, 1989). Analgesic mechanisms of agonist-receptor interactions are thought to be through presynaptic inhibition of neurotransmitter release at the nociceptor axon terminal, or postsynaptic activation of potassium channels to reduce excitabil- ity (Lipp, 1991; Haigler, 1987; Hamon et al., 1988). Opioids are currently the analgesics of of choice in clinical settings, but there are still ambiguities and unknowns throughout the lit- erature (Scherrer et al., 2009; Wang et al., 2010; Powell et al., 2002); for reviews see (Connor CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 70 et al., 2004; Corbett et al., 2006). That being said, morphine (the prototypical MOR agonist) remains the gold standard in clinical pain management (Ruiz-Garcia and Lopez-Briz, 2008; Gilson et al., 2013), because of its effective treatment of moderate to severe pain.

4.2 Evidence of a mu opioid receptor in C. elegans

As aforementioned, current models of pain in animals conventionally use latency of the first nociceptive response to measure pain and drug-modulated analgesia (for example, (Matthes et al., 1996; Mogil et al., 2006)). We have shown in previous work the utility of our multiparameter, high-content phenotyping (Ghosh et al., 2012; Mohammadi et al., 2013). However, in our discussion here of pain and analgesics, for clarity and to keep with conventional data representation, we will choose one metric, namely the maximum change from baseline of the mean centroid speed after the stimulus, to rank the pain response. We have found that this measure concisely and effectively captures statistically significant differences in the behavior. There has been only one previous study to our knowledge characterizing the effects of opioids in C. elegans (Nieto-Fernandez et al., 2009), but the unknown spatiotemporal profile of the thermal stimulus and the qualitative scoring of behavior limit the study, particularly the spatial dissection of the response. We have addressed this with our localized stimulus assay (Chapters 2 and 3) and objective, quantitative response metric, allowing us to better study the effects of opioids on C. elegans.

4.2.1 The thermal avoidance response is modulated by morphine and

naloxone at the midbody

We applied the drugs to young adult worms for very precise times to control the absorbed dose (Materials and Methods). We then analyzed the thermal nociceptive response as in our previous behavioral assays (see Chapter 3). When administered morphine, the head and tail responses were similar to the control data (Figure 4.1). However, morphine notably reduces the thermal CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 71 nociceptive response in C. elegans at the midbody (Figure 4.1). Since analgesia is the main agonist effect of the morphine-MOR interaction, this suggests that opioid receptors could be located at the midbody of C. elegans.

The drug naloxone is used clinically to prevent and treat opioid overdose, particularly heroin overdose (Dixon, 2007; Sporer, 2003). In mammalian systems, naloxone has been shown to act dose-dependently as an agonist (low doses) or inverse agonist (high doses) of the opioid receptor inducing analgesia or hyperalgesia, respectively (Levine et al., 1979). Further- more, at high doses, it has also been shown to block the analgesic effect of morphine when co-administered (Sirohi et al., 2009). This antagonism of the MOR is why the drug is used for acute heroin overdose. The drug concentrations used in our experiments (100mM, unless otherwise indicated) would be considered a relatively high dosage for the small worm. Figure 4.1 shows that pre-treatment with naloxone effectively blocks the analgesic effects of morphine at the midbody (NAL+MOR, p < 0.05 compared to morphine). To assess the inverse agonist effect of naloxone without morphine in C. elegans, we administered it alone (NAL in Figure 4.1), and our results show statistically signiﬁcant hyperalgesia (p<0.05 compared to CTRL) when presenting the worm with or noxious stimulus at the midbody. This suggests that our high dose of naloxone could be acting as an inverse agonist on a MOR subtype speciﬁc to C. elegans at the midbody.

There is no known opioid receptor in C. elegans, but these results suggest that one does exist, and the structure is similar to the mammalian system where naloxone and morphine share high afﬁnity for the receptor. The fact that the magnitude of analgesia depends on the location of the stimulus is intriguing. One possible reason why the head and tail responses appear less sensitive to opioids could be because the head and tail regions possess more than one thermal nociceptor (FLP and AFD in the head, PVD and PHC in the tail) mediating the behavior (Liu et al., 2012; Mohammadi et al., 2013; Chatzigeorgiou et al., 2010). However, to our knowledge, PVD is the only thermal nociceptor at the midbody (Mohammadi et al., 2013). If the putative C. elegans MOR is acting in the typical dendritic nociceptive structures PVD CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 72

Figure 4.1: Localized effect of morphine and naloxone on wild type C. elegans. The change in maximum speed from baseline when stimulated with a noxious thermal pulse in the A) Head, B) Midbody and C) Tail, with various treatments: morphine alone, naloxone as pre-treatment to morphine, and naloxone alone. The analgesic effect of morphine is only seen at the midbody. Similarily, the hyperalgesic effect of naloxone is only seen at the midbody. Naloxone blocks the effects of morphine at the midbody. * p<0.05 compared to CTRL, p<0.05 compared to MORPHINE. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 73 and FLP, the opioid effects could be masked by AFD and PHC activity in the head and tail, respectively. This hypothesis requires further investigation.

Furthermore, a similar phenotype has been observed in rodent models as well (Le Bars et al., 2001; Yoburn et al., 1984; Prentice et al., 1996). In the rat tail ﬂick test, for example, the analgesic effect of morphine is seen primarily at the distal end of the tail compared to the proximal end, and the middle part of the tail gives an intermediate effect (Yoburn et al., 1984; Prentice et al., 1996). While there are many factors involved in the activation of a speciﬁc behavior, this spatial aspect of opioid response modulation could be due to receptor localization (Jordan and Devi, 1998; Wilding et al., 1995). On one level, the response clearly depends on protein availability. However, Wilding and colleagues (Wilding et al., 1995) showed the inhibition of Ca2+ channels by the opioid receptor also depends on the proximity of the channels (i.e., the opioid receptor inhibits only nearby channels). This was a novel insight since up until then, electrophysiology on DRG neurons for opioid pharmacology was done mostly in cultured cells, so the effects of in vivo protein localization could not be studied. Our observed spatial sensitivity in opioid pharmacology in C. elegans could also be due to protein availability and receptor/channel distribution. Further investigation, including determining the distribution of the putative MOR in C. elegans, is vital to begin to understand the molecular basis of this phenomenon.

4.2.2 Preliminary mutant screen identiﬁes MOR candidates in C. elegans

We performed a BLASTP search using the human sequence of the MOR to identify proteins in C. elegans with similar primary sequences, and tested the effect of morphine at the midbody on mutant strains corresponding to the top matches of the search (Table 4.1). We performed a control experiment (same preparation but with M9 instead of morphine) for each mutant strain, and saw no statistically signiﬁcant differences amongst the strains, so for clarity of presentation only the control data for npr-15 is shown in Figure 4.2. The entire dataset for the controls are shown separately in section 4.4 CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 74

Identity match to MOR C. elegans protein 29% NPR-22 28% NPR-15 28% NPR-6 27% NPR-9 26% NPR-11 26% NPR-16 26% NPR-32 Table 4.1: Results from BLASTP search identifying top matches between C. elegans receptors and the human MOR

The NPR-15 and NPR-6 strains appear to be the only ones that do not exhibit analgesia from morphine at the midbody (Figure 4.2, * p < 0.01 compared to N2 morphine response, Kruskal Wallis test). These preliminary ﬁndings suggest that one of these could possibly function as an opioid receptor subtype in C. elegans.

Our results suggest that npr-6 mutants do not experience morphine-induced analgesia at the midbody. This observation could have a couple of explanations. The ﬁrst is that morphine agonizes NPR-6 directly, therefore NPR-6 could be a MOR subtype in C. elegans. How- ever, a second explanation involving peptidergic signaling is more likely. Although NPR-6 has not been functionally characterized, it has been shown to be activated by (Phe-Met-Arg-Phe) FMRFamide-related neuropeptides (FARPs) (Frooninckx et al., 2012; Keating et al., 2003; Cardoso et al., 2012). Further, initial work on the thermal noxious response in C. elegans demonstrated the avoidance behavior is mediated by in part by endogenous FARPs (Wittenburg and Baumeister, 1999). Also, FARPs are known to inhibit the effects of opioids (Tang et al., 1984; Yang et al., 1985), but the mechanisms are not completely understood. Nonetheless, this could point to a role of FMRFamide receptor NPR-6 in our C. elegans opioid pharmacology model. The interaction of the FMRFamide and opioid systems in C. elegans has been proposed in the previous study of opioids in the worm (Nieto-Fernandez et al., 2009), and this could also potentially explain our observed inhibition of morphine analgesia in npr-6 mutants.

There is not much currently known about the putative receptor NPR-15, but the story is interesting thus far. The function of NPR-15 has not yet been determined in C. elegans, and the CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 75

Figure 4.2: Preliminary measurement of morphine analgesia on midbody thermal noxious response for candidate MOR strains. CTRL is the midbody response for NPR-15 without morphine. All responses other than CTRL are for strains treated with morphine. NPR-15 and NPR-6 appear to be the only strains that do not exhibit morphine-induced analgesia at the midbody. * p < 0.01 compared to N2 morphine response. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 76 ligand specificity has not been confirmed (Mills et al., 2012; Komuniecki et al., 2012). How- ever, NPR-15 has been implicated in inhibiting aversive (chemorepulsive) behavior, which is mediated by the polymodal nociceptor in the head ASH (Mills et al., 2012). What is intriguing is that NPR-15 does not seem to be expressed in ASH itself, but rather in other sensory neurons mediating chemotaxis, AWC and ASER (Mills et al., 2012). The complete expression pattern of NPR-15 in C. elegans is not known; however, this highlights the complexity of the nociceptive response as it requires the integration of many sensory inputs from several neurons to produce the observed behavior. It also appears that nlp-8 (neuropeptide-like protein) may encode ligands that activate NPR-15 (Mills et al., 2012). If NPR-15 is indeed acting as a MOR subtype specific to C. elegans, it may be possible that the ligands encoded by nlp-8 could be acting as endogenous opioids (endorphins). This of course is speculative, and further investigation is necessary to define the role NPR-15 in modulating the pain response in C. elegans.

4.3 C. elegans with defective receptor NMUR-2 display spa-

tially sensitive impaired nociception

Neuromedin U (NMU) is a neuropeptide isolated in the mid-1980s from porcine spinal cord (Minamino et al., 1985). Although the progress identifying functions of NMU has been relatively slow, studies have suggested many important physiological roles of NMU, such as the regulation of feeding and energy balance, smooth muscle contraction, stress response, cancer, maintenance of the biological clock, control of blood ﬂow and pressure, and of particular relevance here, pain (for review, see (Brighton et al., 2004)). Because NMU is so multifunctional, its pharmacological modulation is of great interest for therapeutic applications (Budhiraja and Chugh, 2009; Liu et al., 2009).

NMU has two cognate receptors, NMUR1 (peripherally distributed) and NMUR2 (cen- trally distributed) (Brighton et al., 2004). The behavioral phenotyping of mice deﬁcient in the CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 77

NMUR2 receptor has demonstrated its significant role in the pain response (Torres et al., 2007; Zeng et al., 2006) . Interestingly, in the study done by Torres and colleagues (Torres et al., 2007), mice lacking NMUR2 showed impaired nociception in the hot plate test, but not the tail flick test. The authors suggest that this distinction could be due to supraspinal mediation of nociception through NMUR2 (Torres et al., 2007). This, however, presumes that the tail flick response is purely a spinal reflex, which is inaccurate since electrophysiological recordings show involvement of supraspinal structures (Mitchell and Hellon, 1977). The authors conclude that further studies are required to investigate the spatial specificity of the antinociceptive effect for the NMUR2 knockout mice.

In C. elegans, there is not much known about NMUR-1 or NMUR-2. Recently, a study demonstrated NMUR-1 mutants possess a phenotype similar to that in mammalian systems, namely regulation of feeding (Maier et al., 2010). Currently, there is no phenotype for NMUR- 2 mutants in C. elegans (Frooninckx et al., 2012). Here, we show NMUR-2 mutants are statistically significantly defective in the thermal noxious response at the midbody and tail (Figure 4.3; Kruskal-Wallis test, p < 0.05 for midbody, p < 0.01 at tail), and moderately, but not statistically significantly, defective when stimulated in the head (Figure 4.3; Kruskal-Wallis test, p = 0.08). Please note that these are preliminary results (N=19), therefore the variance on the measurements is quite large. Further investigation in the nociceptive response of NMUR-2 will require a larger dataset to definitely confirm these findings, however, I am certain the midbody and tail responses are reproducible.

Our results suggest that NMUR-2 is involved in the signaling pathway for pain in C. elegans, and its role may be spatially dependent, as seen in rodent models. The fact that NMUR-2 is preferentially distributed in higher CNS regions involved in the processing of pain in the mammalian system (Brighton et al., 2004; Torres et al., 2007), and C. elegans appears to maintain the mammalian antinociceptive phenotype modulated by NMUR-2, further evidentiates that C. elegans could be a useful organism to study pain even though it lacks the conventionally required higher cortical regions and functions. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 78

Figure 4.3: NMUR-2 exhibits defective nociceptive behavior moderately in the head and sig- niﬁcantly in the midbody and tail. Kruskal-Wallis test, *p< 0.05, ** p < 0.01, error bars are SEMs. CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 79

4.4 Materials and Methods

The laser-based localized thermal stimulus experiments and statistical analyses were performed as in previous work (Chapter 3).

The application of drugs to the worms was done as follows:

1. Drug concentrations are 100mM, unless otherwise stated

2. Well fed worms (raised at 20ºC) are washed twice then gently spun down for 1 minute. Aspirate supernatant and discard.

3. Transfer 100ml worms to 1.5ml eppendorf, add 100ml drug. 4. Put worms back at 20 degrees for 30 minutes. NOTE: If pre-treatment drugs are used (for example, naloxone prior to morphine), worms are kept in pre-treatment for 30 minutes, then the second drug is added to same liquid for additional 30 minutes. For morphine alone, M9 was used as a pre-treatment to ensure worms across all datasets were in the liquid media and off of food for the same time.

5. Pour worms on to seeded NGM plate.

6. Transfer to assay plates.

7. Let worms acclimatize for 30 minutes at 20 degrees again.

7. Perform experiments, stop at 30 minutes.

A control (M9 treatment only) was done for each strain for the candidate MOR strains, and the results for the midbody stimulus are summarized in Supplementary Figure 4.1. No statistically signiﬁcant difference was seen among the strains (Kruskal-Wallis test, Dunn’s multiple comparison, p > 0.05), so only one strain was shown in Figure 4.2 for clarity of presentation.

4.5 Future directions

The work presented here has made several advances towards establishing C. elegans as a model system for pain research. Firstly, the morphine and naloxone results in Figure 4.1 strongly suggest the existence of a mu opioid receptor in C. elegans with the same analgesic agonist and hyperalgesic inverse agonist properties as in mammals. This result would not have been appar- CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 80

Supplementary Figure 4.1: Control midbody thermal noxious responses for candidate MOR strains (ie, no drugs and M9 treatment only)

ent without the localized stimulus experiment discussed in Chapter 2, since the effect is only seen at the midbody. Furthermore, a preliminary screen of candidate opioid receptors identified possible MORs in C. elegans. While this is a promising start, there is more work to be done. The next logical step would be to identify the MOR in the worm. This can be done in two ways. First, we can attempt to confirm NPR-15 and NPR-6’s putative roles in the opioid system by doing a genetic rescue of these mutants to see if we can recover the wild type (morphine analgesia) behavior. Secondly, because we already know the ligand of the receptor we are searching for, we can look for receptor expression in several ways. Firstly, we can employ a technique similar to that used in the Pert and Snyder study originally identifying binding sites in the rat brain, namely, labeling the ligand with a (fluorescent) marker to visualize the localization. This technique is often not used to find receptors, however, since it has several technical and strategic drawbacks (Ippei Kotera, personal communication). The first difficulty is the potential lack of specificity of the ligand to the binding sites. Secondly, under a fluorescent microscope, it is not known whether the ligand is physiologically bound to the receptor CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 81 or simply sitting randomly on the membrane. In reality, different approaches are usually implemented to find a receptor for a ligand. To this end, several other better techniques using mass spectrometry and immunoprecipitation assays exist to identify ligand-receptor interactions (Frei et al., 2012; Bantscheff and Drewes, 2012). The specific protocols implementing these procedures is beyond the scope of this thesis, and finding the receptor for a ligand is a lot of work, but the molecular biology expertise exists in the Ryu lab to perform these experiments. The mechanisms underlying the spatial sensitivity of the response can be investigated further once the receptor expression pattern is known.

The NMUR-2 results are also exciting and deserve further exploration. First, we must confirm that the nmur-2 mutant is in fact a pain mutant and not a thermosensory defective mutant by testing it on a thermotaxis assay. If it can perform thermotaxis like wild-type worms, we know that it is able to sense temperature normally and the phenotype is due to a defect in its nociceptive pathway. Furthermore, to confirm the role of nmur-2 in the pain response, and potentially characterize it as a pain gene, we need to do a genetic rescue of the nmur-2 mutation in the worm and see if we can recover wild-type nociceptive behavior. Again, receptor localization must be determined as a first step towards investigating the spatial sensitivity of the response. Since in this case we know the gene and want to determine the expression pattern of the protein, we can do this in a couple of ways. First, one can create a fused construct consisting of a reporter gene fused to the whole gene cassette. This construct is then injected into the worms, and depending on the reporter gene, the expression pattern is visualized by fluorescence or staining (Hunt-Newbury et al., 2007). Another technique utilizes immunohistochemistry to identify gene expression (Duerr, 2006). Briefly, the first step is to generate a specific antibody protein against target protein. The antigen for the target protein is then used to immunize animals and the antiserum is collected and should be purified. The diluted antiserum is applied to the fixed worms, and then the worms are washed very well. A secondary fluorescent antibody is commonly used to visualize the localization pattern of the gene product (Duerr, 2006). Once we have determined the localization of NMUR-2, we then can proceed CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 82 in a number of directions, including high-content phenotype screens using small molecules libraries to identify potential targets to NMUR-2 as novel modulators of pain.

4.6 Concluding remarks

One advantage of C. elegans as a model organism is that its “parts list” (that is, genes and neurons) is known. Although the neural connectivity of the 302 neurons is known and extremely useful, the complex behaviors generated by the interaction of internal states of the worm with environmental cues from its world cannot be predicted by the anatomy alone. Behavioral studies of the worm have lagged behind more traditional cellular and molecular pursuits. This motivates the study of behavior on a quantitative systems-level, since it allows us to holis- tically assess the sensorimotor output from a carefully controlled input. Multidimensional, high-content phenotyping can expose features of the response that then may reveal underlying genetic or neural mediators of behavior. More generally, studying the movement of an organism offers insight into its strategies of maneuvering its environment. To this end, this work has implemented novel instrumentation and comprehensively characterized the C. elegans behavioral receptive ﬁeld to noxious thermal stimuli. We have used calcium imaging, genetic ablation and neuronal ablation to identify PVD as a thermal nociceptor mediating the midbody response, and further demonstrated it is able to spatially discriminate stimuli as close as 80 microns. This provides the basis for further investigation of spatial discrimination of stimuli at the single neuron level. Furthermore, intensity-graded quantiﬁcation of nociceptive behavior is desirable in the study of animal models of pain, since animals cannot verbally communicate their pain level. Therefore, we must read their body language to assess their sensations, as we have done through our high-content phenotyping. We demonstrated the feasibility of the C. elegans nociceptive response being characterized as a pain response by quantifying the effects of morphine and naloxone on its behavior when locally stimulated with noxious heat from head to tail. We demonstrated that morphine’s analgesic and naloxone’s hyperalgesic properties in CHAPTER 4. C. elegans AS A MODEL SYSTEM FOR PAIN 83 mammalian systems also hold in C. elegans at the midbody, strongly suggesting the presence of a mu opioid receptor in the worm. This provides the foundation to build on the techniques presented in this dissertation to use C. elegans as a model organism for pain research. References

Albeg, A., Smith, C. J., Chatzigeorgiou, M., Feitelson, D. G., Hall, D. H., Schafer, W. R., Miller, D. M., and Treinin, M. (2011). C. elegans multi-dendritic sensory neurons: mor- phology and function. Mol Cell Neurosci, 46(1):308–17.

Albrecht, D. R. and Bargmann, C. I. (2011). High-content behavioral analysis of caenorhabditis elegans in precise spatiotemporal chemical environments. Nat Methods, 8(7):599–605.

Apkarian, A. V., Darbar, A., Krauss, B. R., Gelnar, P. A., and Szeverenyi, N. M. (1999). Differ- entiating cortical areas related to pain perception from stimulus identiﬁcation: temporal analysis of fmri activity. Journal of neurophysiology, 81(6):2956–2963.

Arthur, R. P. and Shelley, W. B. (1959). The innervation of human epidermis. J Invest Derma- tol, 32(3):397–411.

Bantscheff, M. and Drewes, G. (2012). Chemoproteomic approaches to drug target identiﬁca- tion and drug proﬁling. Bioorg Med Chem, 20(6):1973–8.

Bargmann, C. I. (1998). Neurobiology of the caenorhabditis elegans genome. Science, 282(5396):2028–2033.

Bargmann, C. I. and Avery, L. (1995). Laser killing of cells in caenorhabditis elegans. Methods Cell Biol, 48:225–50.

Bargmann, C. I., Thomas, J. H., and Horvitz, H. R. (1990). Chemosensory cell function in the

84 REFERENCES 85

behavior and development of caenorhabditis elegans. In Cold Spring Harbor symposia on quantitative biology, volume 55, pages 529–538. Cold Spring Harbor Laboratory Press.

Basbaum, A. I., Bautista, D. M., Scherrer, G., and Julius, D. (2009). Cellular and molecular mechanisms of pain. Cell, 139(2):267–284.

Bateson, P. (1991). Assessment of pain in animals. Animal Behaviour, 42(5):827–839.

Baumgärtner, U., Iannetti, G. D., Zambreanu, L., Stoeter, P., Treede, R.-D. . D., and Tracey, I. (2010). Multiple somatotopic representations of heat and mechanical pain in the operculo- insular cortex: a high-resolution fmri study. Journal of neurophysiology, 104(5):2863– 2872.

Bautista, D. M., Movahed, P., Hinman, A., Axelsson, H. E., Sterner, O., Högestätt, E. D., Julius, D., Jordt, S.-E. E., and Zygmunt, P. M. (2005). Pungent products from garlic activate the sensory ion channel trpa1. Proc Natl Acad Sci U S A, 102(34):12248–52.

Berg, H. C. (1993). Random walks in biology. Princeton University Press.

Berg, H. C. (2004). E. coli in Motion. Springer.

Berg, H. C. and Brown, D. A. (1972). Chemotaxis in escherichia coli analysed by three- dimensional tracking. Nature, 239(5374):500–4.

Bessou, P. and Perl, E. R. (1969). Response of cutaneous sensory units with unmyelinated ﬁbers to noxious stimuli. Journal of neurophysiology, 32(6):1025–1043.

Bingel, U., Gläscher, J., Weiller, C., and Büchel, C. (2004). Somatotopic representation of nociceptive information in the putamen: an event-related fmri study. Cerebral Cortex, 14(12):1340–1345.

Bonica, J. J. (1979). The need of a taxonomy. Pain, 6(3):247–248. REFERENCES 86

Bornhövd, K., Quante, M., Glauche, V., Bromm, B., Weiller, C., and Büchel, C. (2002). Painful stimuli evoke different stimulus–response functions in the amygdala, prefrontal, insula and somatosensory cortex: a single-trial fmri study. Brain, 125(6):1326–1336.

Brenner, S. (1974). The genetics of caenorhabditis elegans. Genetics, 77(1):71–94.

Brenner, S. (2002). Natures gift to science. Les Prix Nobel, The Nobel Prizes, pages 274–82.

Brighton, P. J., Szekeres, P. G., and Willars, G. B. (2004). Neuromedin u and its receptors: structure, function, and physiological roles. Pharmacol Rev, 56(2):231–48.

Brockmann, D., Hufnagel, L., and Geisel, T. (2006). The scaling laws of human travel. Nature, 439(7075):462–5.

Bromm, B. and Treede, R. D. (1980). Withdrawal reﬂex, skin resistance reaction and pain ratings due to electrical stimuli in man. Pain, 9(3):339–54.

Budhiraja, S. and Chugh, A. (2009). Neuromedin u: physiology, pharmacology and therapeutic potential. Fundam Clin Pharmacol, 23(2):149–57.

Burgess, P. R. and Perl, E. R. (1967). Myelinated afferent ﬁbres responding speciﬁcally to noxious stimulation of the skin. J Physiol, 190(3):541–62.

Burns, A. R., Kwok, T. C. Y., Howard, A., Houston, E., Johanson, K., Chan, A., Cutler, S. R., McCourt, P., and Roy, P. J. (2006). High-throughput screening of small molecules for bioactivity and target identiﬁcation in caenorhabditis elegans. Nat Protoc, 1(4):1906–14.

Cardoso, J. a. C. R., Félix, R. C., Fonseca, V. G., and Power, D. M. (2012). Feeding and the rhodopsin family g-protein coupled receptors in nematodes and arthropods. Front Endocrinol (Lausanne), 3:157.

Caterina, M. J., Lefﬂer, A., Malmberg, A. B., Martin, W. J., Trafton, J., Petersen-Zeitz, K. R., Koltzenburg, M., Basbaum, A. I., and Julius, D. (2000). Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science, 288(5464):306–13. REFERENCES 87

Caterina, M. J., Rosen, T. A., Tominaga, M., Brake, A. J., and Julius, D. (1999). A capsaicin- receptor homologue with a high threshold for noxious heat. Nature, 398(6726):436–41.

Caterina, M. J., Schumacher, M. A., Tominaga, M., Rosen, T. A., Levine, J. D., and Julius, D. (1997). The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature, 389(6653):816–24.

Chalﬁe, M., Sulston, J. E., White, J. G., Southgate, E., Thomson, J. N., and Brenner, S. (1985). The neural circuit for touch sensitivity in caenorhabditis elegans. J Neurosci, 5(4):956–64.

Chao, M. Y., Komatsu, H., Fukuto, H. S., Dionne, H. M., and Hart, A. C. (2004). Feeding status and serotonin rapidly and reversibly modulate a caenorhabditis elegans chemosensory circuit. Proc Natl Acad Sci U S A, 101(43):15512–7.

Chatzigeorgiou, M., Yoo, S., Watson, J. D., Lee, W.-H. H., Spencer, W. C., Kindt, K. S., Hwang, S. W., Miller, D. M., Treinin, M., Driscoll, M., and Schafer, W. R. (2010). Speciﬁc roles for deg/enac and trp channels in touch and thermosensation in c. elegans nociceptors. Nat Neurosci, 13(7):861–8.

Chen, L. M., Dillenburger, B. C., Wang, F., Friedman, R. M., and Avison, M. J. (2011). High- resolution functional magnetic resonance imaging mapping of noxious heat and tactile activations along the central sulcus in new world monkeys. Pain, 152(3):522–532.

Coghill, R. C., Sang, C. N., Maisog, J. M., and Iadarola, M. J. (1999). Pain intensity processing within the human brain: a bilateral, distributed mechanism. Journal of Neurophysiology, 82(4):1934–1943.

Colbert, H. A., Smith, T. L., and Bargmann, C. I. (1997). Osm-9, a novel protein with struc- tural similarity to channels, is required for olfaction, mechanosensation, and olfactory adaptation in caenorhabditis elegans. J Neurosci, 17(21):8259–69. REFERENCES 88

Connor, M., Osborne, P. B., and Christie, M. J. (2004). µ-opioid receptor desensitization: Is morphine different? British journal of pharmacology, 143(6):685–696.

Cooper, B. Y., Vierck, C. J., and Yeomans, D. C. (1986). Selective reduction of second pain sensations by systemic morphine in humans. Pain, 24(1):93–116.

Corbett, A. D., Henderson, G., McKnight, A. T., and Paterson, S. J. (2006). 75 years of opioid research: the exciting but vain quest for the holy grail. British journal of pharmacology, 147(S1):S153–S162.

Cuellar, J. M., Manering, N. A., Klukinov, M., Nemenov, M. I., and Yeomans, D. C. (2010). Thermal nociceptive properties of trigeminal afferent neurons in rats. Molecular pain, 6(1):39.

Dixon, P. (2007). Managing acute heroin overdose. Emerg Nurse, 15(2):30–5.

Dubin, A. E. and Patapoutian, A. (2010). Nociceptors: the sensors of the pain pathway. J Clin Invest, 120(11):3760–72.

Duerr, J. S. (2006). Immunohistochemistry.

Eaton, M. (2003). Common animal models for spasticity and pain. Journal of Rehabilitation Research and development, 40(4; SUPP/1):41–54.

Fang-Yen, C., Gabel, C. V., Samuel, A. D. T., Bargmann, C. I., and Avery, L. (2012). Laser microsurgery in caenorhabditis elegans. Methods Cell Biol, 107:177–206.

Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., Driver, S. E., and Mello, C. C. (1998). Potent and speciﬁc genetic interference by double-stranded rna in caenorhabditis elegans. nature, 391(6669):806–811.

Frei, A. P., Jeon, O.-Y. Y., Kilcher, S., Moest, H., Henning, L. M., Jost, C., Plückthun, A., Mercer, J., Aebersold, R., Carreira, E. M., and Wollscheid, B. (2012). Direct identiﬁcation REFERENCES 89

of ligand-receptor interactions on living cells and tissues. Nat Biotechnol, 30(10):997– 1001.

Frooninckx, L., Van Rompay, L., Temmerman, L., Van Sinay, E., Beets, I., Janssen, T., Husson, S. J., and Schoofs, L. (2012). Neuropeptide gpcrs in c. elegans. Frontiers in endocrinol- ogy, 3.

Ghosh, R., Mohammadi, A., Kruglyak, L., and Ryu, W. S. (2012). Multiparameter behavioral proﬁling reveals distinct thermal response regimes in caenorhabditis elegans. BMC Biol, 10(1):85.

Gilson, A. M., Maurer, M. A., Ryan, K. M., Rathouz, P. J., and Cleary, J. F. (2013). Using a morphine equivalence metric to quantify opioid consumption: examining the capacity to provide effective treatment of debilitating pain at the global, regional, and country levels. J Pain Symptom Manage, 45(4):681–700.

Glauser, D. A., Chen, W. C., Agin, R., Macinnis, B. L., Hellman, A. B., Garrity, P. A., Tan, M.-W. W., and Goodman, M. B. (2011). Heat avoidance is regulated by transient receptor potential (trp) channels and a neuropeptide signaling pathway in caenorhabditis elegans. Genetics, 188(1):91–103.

Goldstein, A. and Naidu, A. (1989). Multiple opioid receptors: ligand selectivity proﬁles and binding site signatures. Mol Pharmacol, 36(2):265–72.

Haigler, H. J. (1987). Neurophysiological effects of opiates in the cns. Monogr Neural Sci, 13:132–60.

Hamon, M., Bourgoin, S., Bars, D. L., and Cesselin, F. (1988). < i> in vivo and< i> in vitro release of central neurotransmitters in relation to pain and analgesia. Progress in brain research, 77:431–444. REFERENCES 90

Hart, A. C., Kass, J., Shapiro, J. E., and Kaplan, J. M. (1999). Distinct signaling pathways mediate touch and osmosensory responses in a polymodal sensory neuron. The Journal of neuroscience, 19(6):1952–1958.

Hart, A. C., Sims, S., and Kaplan, J. M. (1995). Synaptic code for sensory modalities revealed by c. elegans glr-1 glutamate receptor. Nature, 378(6552):82–85.

Hedgecock, E. M. and Russell, R. L. (1975). Normal and mutant thermotaxis in the nematode caenorhabditis elegans. Proc Natl Acad Sci U S A, 72(10):4061–5.

Hill, R. (2000). Nk1 (substance p) receptor antagonists–why are they not analgesic in humans? Trends Pharmacol Sci, 21(7):244–6.

Hilliard, M. A., Bargmann, C. I., and Bazzicalupo, P. (2002). C. elegans responds to chemical repellents by integrating sensory inputs from the head and the tail. Curr Biol, 12(9):730–4.

Horvitz, H. R. (2003). Worms, life, and death (nobel lecture). Chembiochem, 4(8):697–711.

Hunt-Newbury, R., Viveiros, R., Johnsen, R., Mah, A., Anastas, D., Fang, L., Halfnight, E., Lee, D., Lin, J., and Lorch, A. (2007). High-throughput in vivo analysis of gene expression in caenorhabditis elegans. PLoS biology, 5(9):e237.

Husson, S. J., Costa, W. S., Wabnig, S., Stirman, J. N., Watson, J. D., Spencer, W. C., Aker- boom, J., Looger, L. L., Treinin, M., Miller, D. M., Lu, H., and Gottschalk, A. (2012). Op- togenetic analysis of a nociceptor neuron and network reveals ion channels acting down- stream of primary sensors. Curr Biol, 22(9):743–52.

Johnson, K. O. and Phillips, J. R. (1981). Tactile spatial resolution. i. two-point discrimination, gap detection, grating resolution, and letter recognition. Journal of Neurophysiology, 46(6):1177–1192.

Jones, T. L., Sweitzer, S. M., Wilson, S. P., and Yeomans, D. C. (2003). Afferent ﬁber-selective REFERENCES 91

shift in opiate potency following targeted opioid receptor knockdown. Pain, 106(3):365– 71.

Jordan, B. and Devi, L. A. (1998). Molecular mechanisms of opioid receptor signal transduction. Br J Anaesth, 81(1):12–9.

Jordt, S.-E. E., Bautista, D. M., Chuang, H.-H. H., McKemy, D. D., Zygmunt, P. M., Högestätt, E. D., Meng, I. D., and Julius, D. (2004). Mustard oils and cannabinoids excite sensory nerve ﬁbres through the trp channel anktm1. Nature, 427(6971):260–5.

Jose, A. M., Bany, I. A., Chase, D. L., and Koelle, M. R. (2007). A speciﬁc subset of transient receptor potential vanilloid-type channel subunits in caenorhabditis elegans en- docrine cells function as mixed heteromers to promote neurotransmitter release. Genetics, 175(1):93–105.

Julius, D. and Basbaum, A. I. (2001). Molecular mechanisms of nociception. Nature, 413(6852):203–210.

Kalinin, Y. V., Jiang, L., Tu, Y., and Wu, M. (2009). Logarithmic sensing in escherichia coli bacterial chemotaxis. Biophysical journal, 96(6):2439–2448.

Kandel, E. R., Schwartz, J. H., and Jessell, T. M. (2000). Principles of neural science, volume 4. McGraw-Hill New York.

Kaplan, J. M. and Horvitz, H. R. (1993). A dual mechanosensory and chemosensory neuron in caenorhabditis elegans. Proc Natl Acad Sci U S A, 90(6):2227–31.

Keating, C. D., Kriek, N., Daniels, M., Ashcroft, N. R., Hopper, N. A., Siney, E. J., Holden- Dye, L., and Burke, J. F. (2003). Whole-genome analysis of 60 g protein-coupled receptors in caenorhabditis elegans by gene knockout with rnai. Curr Biol, 13(19):1715–20.

Keefe, F. J., Fillingim, R. B., and Williams, D. A. (1991). Behavioral assessment of pain: nonverbal measures in animals and humans. ILAR Journal, 33(1-2):3–13. REFERENCES 92

Kindt, K. S., Viswanath, V., Macpherson, L., Quast, K., Hu, H., Patapoutian, A., and Schafer, W. R. (2007). Caenorhabditis elegans trpa-1 functions in mechanosensation. Nat Neurosci, 10(5):568–77.

Kokel, D., Bryan, J., Laggner, C., White, R., Cheung, C. Y. J., Mateus, R., Healey, D., Kim, S., Werdich, A. A., Haggarty, S. J., Macrae, C. A., Shoichet, B., and Peterson, R. T. (2010). Rapid behavior-based identiﬁcation of neuroactive small molecules in the zebraﬁsh. Nat Chem Biol, 6(3):231–237.

Komuniecki, R., Harris, G., Hapiak, V., Wragg, R., and Bamber, B. (2012). Monoamines activate neuropeptide signaling cascades to modulate nociception in c. elegans: a useful model for the modulation of chronic pain? Invertebrate Neuroscience, 12(1):53–61.

Kwan, K. Y., Allchorne, A. J., Vollrath, M. A., Christensen, A. P., Zhang, D.-S. . S., Woolf, C. J., and Corey, D. P. (2006). Trpa1 contributes to cold, mechanical, and chemical nociception but is not essential for hair-cell transduction. Neuron, 50(2):277–289.

Langford, D. J., Bailey, A. L., Chanda, M. L., Clarke, S. E., Drummond, T. E., Echols, S., Glick, S., Ingrao, J., Klassen-Ross, T., Lacroix-Fralish, M. L., Matsumiya, L., Sorge, R. E., Sotocinal, S. G., Tabaka, J. M., Wong, D., van den Maagdenberg, A. M. J. M., Ferrari, M. D., Craig, K. D., and Mogil, J. S. (2010). Coding of facial expressions of pain in the laboratory mouse. Nat Methods, 7(6):447–9.

Langley, C. K., Aziz, Q., Bountra, C., Gordon, N., Hawkins, P., Jones, A., Langley, G., Nur- mikko, T., and Tracey, I. (2008). Volunteer studies in pain researchopportunities and challenges to replace animal experiments: the report and recommendations of a focus on alternatives workshop. Neuroimage, 42(2):467–473.

Lawson, S. N. (2002). Phenotype and function of somatic primary afferent nociceptive neu- rones with c-, adelta- or aalpha/beta-ﬁbres. Exp Physiol, 87(2):239–44. REFERENCES 93

Le Bars, D., Gozariu, M., and Cadden, S. W. (2001). Animal models of nociception. Pharma- cological Reviews, 53(4):597–652.

Lee, R. Y., Sawin, E. R., Chalﬁe, M., Horvitz, H. R., and Avery, L. (1999). Eat-4, a homolog of a mammalian sodium-dependent inorganic phosphate cotransporter, is necessary for glutamatergic neurotransmission in caenorhabditis elegans. The Journal of neuroscience, 19(1):159–167.

Levine, J. D., Gordon, N. C., and Fields, H. L. (1979). Naloxone dose dependently produces analgesia and hyperalgesia in postoperative pain.

Li, W., Kang, L., Piggott, B. J., Feng, Z., and Xu, X. Z. S. (2011). The neural circuits and sensory channels mediating harsh touch sensation in caenorhabditis elegans. Nat Commun, 2:315.

Lipp, J. (1991). Possible mechanisms of morphine analgesia. Clinical neuropharmacology, 14(2):131–147.

Liu, J. J., Payza, K., Huang, J., Liu, R., Chen, T., Coupal, M., Laird, J. M., Cao, C.-Q. . Q., Butterworth, J., and Lapointe, S. (2009). Discovery and pharmacological characterization of a small-molecule antagonist at neuromedin u receptor nmur2. Journal of Pharmacology and Experimental Therapeutics, 330(1):268–275.

Liu, S., Schulze, E., and Baumeister, R. (2012). Temperature- and touch-sensitive neurons couple cng and trpv channel activities to control heat avoidance in caenorhabditis elegans. PLoS One, 7(3):e32360.

Loganathan, S., Kanteti, R., Siddiqui, S. S., El-Hashani, E., Tretiakova, M., Vigneswaran, H., Cervantes, G., Natarajan, V., Husain, A. N., Vokes, E. E., Kindler, H. L., and Salgia, R.

(2011). Role of protein kinase c b and vascular endothelial growth factor receptor in ma- lignant pleural mesothelioma: Therapeutic implications and the usefulness of caenorhabditis elegans model organism. J Carcinog, 10:4. REFERENCES 94

Lu, Y., Pirec, V., and Yeomans, D. C. (1997). Differential antinociceptive effects of spinal opioids on foot withdrawal responses evoked by c ﬁbre or a delta nociceptor activation. Br J Pharmacol, 121(6):1210–6.

Macosko, E. Z., Pokala, N., Feinberg, E. H., Chalasani, S. H., Butcher, R. A., Clardy, J., and Bargmann, C. I. (2009). A hub-and-spoke circuit drives pheromone attraction and social behaviour in c. elegans. Nature, 458(7242):1171–1175.

Maier, W., Adilov, B., Regenass, M., and Alcedo, J. (2010). A neuromedin u receptor acts with the sensory system to modulate food type-dependent effects on c. elegans lifespan. PLoS biology, 8(5):e1000376.

Mancini, F., Haggard, P., Iannetti, G. D., Longo, M. R., and Sereno, M. I. (2012). Fine- grained nociceptive maps in primary somatosensory cortex. The Journal of Neuroscience, 32(48):17155–17162.

Mancini, F., Sambo, C. F., Ramirez, J. D., Bennett, D. L. H., Haggard, P., and Iannetti, G. D. (2013). A fovea for pain at the ﬁngertips. Curr Biol, 23(6):496–500.

Maricq, A. V., Peckol, E., Driscoll, M., and Bargmann, C. I. (1995). Mechanosensory sig- nalling in c. elegans mediated by the glr-1 glutamate receptor.

Maroteaux, G., Loos, M., van der Sluis, S., Koopmans, B., Aarts, E., van Gassen, K., Geurts, A., Largaespada, D. A., Spruijt, B. M., Stiedl, O., Smit, A. B., and Verhage, M. (2012). High-throughput phenotyping of avoidance learning inmice discriminates different geno- types andidentiﬁes a novel gene. Genes Brain Behav, 11(7):772–784.

Matthes, H. W., Maldonado, R., Simonin, F., Valverde, O., Slowe, S., Kitchen, I., Befort, K., Dierich, A., Le Meur, M., Dollé, P., Tzavara, E., Hanoune, J., Roques, B. P., and Kieffer, B. L. (1996). Loss of morphine-induced analgesia, reward effect and withdrawal symptoms in mice lacking the mu-opioid-receptor gene. Nature, 383(6603):819–23. REFERENCES 95

Mills, H., Wragg, R., Hapiak, V., Castelletto, M., Zahratka, J., Harris, G., Summers, P., Korch- nak, A., Law, W., Bamber, B., and Komuniecki, R. (2012). Monoamines and neuropeptides interact to inhibit aversive behaviour in caenorhabditis elegans. EMBO J, 31(3):667– 78.

Minamino, N., Kangawa, K., and Matsuo, H. (1985). Neuromedin u-8 and u-25: novel uterus stimulating and hypertensive peptides identiﬁed in porcine spinal cord. Biochem Biophys Res Commun, 130(3):1078–85.

Mitchell, D. and Hellon, R. F. (1977). Neuronal and behavioural responses in rats during noxious stimulation of the tail. Proc R Soc Lond B Biol Sci, 197(1127):169–94.

Mogil, J. S. (2009). Animal models of pain: progress and challenges. Nature Reviews Neuro- science, 10(4):283–294.

Mogil, J. S., Ritchie, J., Sotocinal, S. G., Smith, S. B., Croteau, S., Levitin, D. J., and Naumova, A. K. (2006). Screening for pain phenotypes: analysis of three congenic mouse strains on a battery of nine nociceptive assays. Pain, 126(1-3):24–34.

Mohammadi, A., Byrne Rodgers, J., Kotera, I., and Ryu, W. S. (2013). Behavioral response of caenorhabditis elegans to localized thermal stimuli. BMC Neurosci, 14(1):66.

Montell, C. (2005). The trp superfamily of cation channels. Science Signaling, 2005(272):re3.

Montell, C. and Rubin, G. M. (1989). Molecular characterization of the drosophila trp locus: a putative integral membrane protein required for phototransduction. Neuron, 2(4):1313– 23.

Ness, T. J., Jones, S. L., and Gebhart, G. F. (1987). Contribution of the site of heating to variability in the latency of the rat tail ﬂick reﬂex. Brain Res, 426(1):169–72.

Nieto-Fernandez, F., Andrieux, S., Idrees, S., Bagnall, C., Pryor, S. C., and Sood, R. (2009). REFERENCES 96

The effect of opioids and their antagonists on the nocifensive response of caenorhabditis elegans to noxious thermal stimuli. Invert Neurosci, 9(3-4):195–200.

O’Hagan, R., Chalﬁe, M., and Goodman, M. B. (2004). The mec-4 deg/enac channel of caenorhabditis elegans touch receptor neurons transduces mechanical signals. Nature neuroscience, 8(1):43–50.

Patapoutian, A., Tate, S., and Woolf, C. J. (2009). Transient receptor potential channels: targeting pain at the source. Nat Rev Drug Discov, 8(1):55–68.

Pert, C. B. and Snyder, S. H. (1973). Opiate receptor: demonstration in nervous tissue. Science, 179(4077):1011–4.

Pirri, J. K. and Alkema, M. J. (2012). The neuroethology of c. elegans escape. Curr Opin Neurobiol, 22(2):187–93.

Powell, K. J., Abul-Husn, N. S., Jhamandas, A., Olmstead, M. C., Beninger, R. J., and Jhaman- das, K. (2002). Paradoxical effects of the opioid antagonist naltrexone on morphine analgesia, tolerance, and reward in rats. J Pharmacol Exp Ther, 300(2):588–96.

Prentice, T. W., Joynes, R. L., Meagher, M. W., and Grau, J. W. (1996). Impact of shock on pain reactivity: Iii. the magnitude of hypoalgesia observed depends on test location. Behav Neurosci, 110(3):528–41.

Price, D. D. and Dubner, R. (1977). Mechanisms of ﬁrst and second pain in the peripheral and central nervous systems. Journal of Investigative Dermatology, 69(1):167–171.

Rose, J. K., Kaun, K. R., Chen, S. H., and Rankin, C. H. (2003). Glr-1, a non-nmda glutamate receptor homolog, is critical for long-term memory in caenorhabditis elegans. J Neurosci, 23(29):9595–9.

Ruiz-Garcia, V. and Lopez-Briz, E. (2008). Morphine remains gold standard in breakthrough cancer pain. BMJ, 337. REFERENCES 97

Ryu, W. S. and Samuel, A. D. T. (2002). Thermotaxis in caenorhabditis elegans analyzed by measuring responses to deﬁned thermal stimuli. J Neurosci, 22(13):5727–33.

Sawin, E. R., Ranganathan, R., and Horvitz, H. R. (2000). C. elegans locomotory rate is modulated by the environment through a dopaminergic pathway and by experience through a serotonergic pathway. Neuron, 26(3):619–31.

Schafer, W. R. (2012). Tackling thermosensation with multidimensional phenotyping. BMC Biology, 10(1):91.

Scherrer, G., Imamachi, N., Cao, Y.-Q. Q., Contet, C., Mennicken, F., O’Donnell, D., Kieffer, B. L., and Basbaum, A. I. (2009). Dissociation of the opioid receptor mechanisms that control mechanical and heat pain. Cell, 137(6):1148–59.

Segall, J. E., Block, S. M., and Berg, H. C. (1986). Temporal comparisons in bacterial chemotaxis. Proceedings of the National Academy of Sciences, 83(23):8987–8991.

Sherrington, C. S. S. (1903). Qualitative differences of spinal reﬂex corresponding with qualitative difference of cutaneous stimulus. J. Physiol., 30:39–46.

Siddiqui, S. S., Loganathan, S., Krishnaswamy, S., Faoro, L., Jagadeeswaran, R., and Salgia, R. (2008). C. elegans as a model organism for in vivo screening in cancer: effects of human c-met in lung cancer affect c. elegans vulva phenotypes. Cancer biology & therapy, 7(6):856–863.

Siegel, S. (1956). Nonparametric statistics for the behavioral sciences.

Sirohi, S., Dighe, S. V., Madia, P. A., and Yoburn, B. C. (2009). The relative potency of inverse opioid agonists and a neutral opioid antagonist in precipitated withdrawal and antagonism of analgesia and toxicity. J Pharmacol Exp Ther, 330(2):513–9.

Smith, C. J., Watson, J. D., Spencer, W. C., O’Brien, T., Cha, B., Albeg, A., Treinin, M., and Miller, D. M. (2010). Time-lapse imaging and cell-speciﬁc expression proﬁling reveal dy- REFERENCES 98

namic branching and molecular determinants of a multi-dendritic nociceptor in c. elegans. Dev Biol, 345(1):18–33.

Sneddon, L. U., Braithwaite, V. A., and Gentle, M. J. (2003). Do ﬁshes have nociceptors? evidence for the evolution of a vertebrate sensory system. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270(1520):1115–1121.

Sporer, K. A. (2003). Strategies for preventing heroin overdose. BMJ, 326(7386):442–4.

Stephens, G. J., Johnson-Kerner, B., Bialek, W., and Ryu, W. S. (2010). From modes to movement in the behavior of caenorhabditis elegans. PLoS One, 5(11):e13914.

Story, G. M., Peier, A. M., Reeve, A. J., Eid, S. R., Mosbacher, J., Hricik, T. R., Earley, T. J., Hergarden, A. C., Andersson, D. A., and Hwang, S. W. (2003). Anktm1, a trp-like channel expressed in nociceptive neurons, is activated by cold temperatures. Cell, 112(6):819–830.

Tang, J., Yang, H. Y., and Costa, E. (1984). Inhibition of spontaneous and opiate-modiﬁed nociception by an endogenous neuropeptide with phe-met-arg-phe-nh2-like immunoreac- tivity. Proc Natl Acad Sci U S A, 81(15):5002–5.

Thompson, R. F. (1967). Foundations of physiological psychology.

Tobin, D. M. and Bargmann, C. I. (2004). Invertebrate nociception: behaviors, neurons and molecules. J Neurobiol, 61(1):161–74.

Tobin, D. M., Madsen, D. M., Kahn-Kirby, A., Peckol, E. L., Moulder, G., Barstead, R., Mar- icq, A. V., and Bargmann, C. I. (2002). Combinatorial expression of trpv channel proteins deﬁnes their sensory functions and subcellular localization in< i> c. elegans neurons. Neu- ron, 35(2):307–318.

Torres, R., Croll, S. D., Vercollone, J., Reinhardt, J., Grifﬁths, J., Zabski, S., Anderson, K. D., Adams, N. C., Gowen, L., and Sleeman, M. W. (2007). Mice genetically deﬁcient in REFERENCES 99

neuromedin u receptor 2, but not neuromedin u receptor 1, have impaired nociceptive responses. Pain, 130(3):267–278.

Tsalik, E. L., Niacaris, T., Wenick, A. S., Pau, K., Avery, L., and Hobert, O. (2003). Lim homeobox gene-dependent expression of biogenic amine receptors in restricted regions of the c. elegans nervous system. Developmental biology, 263(1):81–102.

Voets, T., Droogmans, G., Wissenbach, U., Janssens, A., Flockerzi, V., and Nilius, B. (2004). The principle of temperature-dependent gating in cold- and heat-sensitive trp channels. Nature, 430(7001):748–54.

Wallace, M. S., Rowbotham, M., Bennett, G. J., Jensen, T. S., Pladna, R., and Quessy, S. (2002a). A multicenter, double-blind, randomized, placebo-controlled crossover evaluation of a short course of 4030w92 in patients with chronic neuropathic pain. The Journal of Pain, 3(3):227–233.

Wallace, M. S., Rowbotham, M. C., Katz, N. P., Dworkin, R. H., Dotson, R. M., Galer, B. S., Rauck, R. L., Backonja, M. M., Quessy, S. N., and Meisner, P. D. (2002b). A randomized, double-blind, placebo-controlled trial of a glycine antagonist in neuropathic pain. Neurology, 59(11):1694–1700.

Wang, H.-B. . B., Zhao, B., Zhong, Y.-Q. . Q., Li, K.-C. . C., Li, Z.-Y. . Y., Wang, Q., Lu, Y.-J. . J., Zhang, Z.-N. . N., He, S.-Q. . Q., and Zheng, H.-C. . C. (2010). Coexpression

of d-and µ-opioid receptors in nociceptive sensory neurons. Proceedings of the National Academy of Sciences, 107(29):13117–13122.

Way, J. C. and Chalﬁe, M. (1989). The mec-3 gene of caenorhabditis elegans requires its own product for maintained expression and is expressed in three neuronal cell types. Genes Dev, 3(12A):1823–33.

Weissman-Fogel, I., Brayer-Zwi, N., and Defrin, R. (2012). Spatial resolution of the pain REFERENCES 100

system: a proximal-to-distal gradient of sensitivity revealed with psychophysical testing. Experimental brain research, 216(2):181–190.

White, J. G., Southgate, E., Thomson, J. N., and Brenner, S. (1986). The structure of the nervous system of the nematode caenorhabditis elegans. Philosophical Transactions of the Royal Society of London. B, Biological Sciences, 314(1165):1.

Wilding, T. J., Womack, M. D., and McCleskey, E. W. (1995). Fast, local signal transduction between the mu opioid receptor and ca2+ channels. The Journal of neuroscience, 15(5):4124–4132.

Wilson, S. G. and Mogil, J. S. (2001). Measuring pain in the (knockout) mouse: big challenges in a small mammal. Behavioural brain research, 125(1):65–73.

Wittenburg, N. and Baumeister, R. (1999). Thermal avoidance in caenorhabditis elegans: an approach to the study of nociception. Proc Natl Acad Sci U S A, 96(18):10477–82.

Woodbury, C. J., Zwick, M., Wang, S., Lawson, J. J., Caterina, M. J., Koltzenburg, M., Albers, K. M., Koerber, H. R., and Davis, B. M. (2004). Nociceptors lacking trpv1 and trpv2 have normal heat responses. J Neurosci, 24(28):6410–5.

Xu, S. Y., Cang, C. L., Liu, X. F., Peng, Y. Q., Ye, Y. Z., Zhao, Z. Q., and Guo, A. K. (2006). Thermal nociception in adult drosophila: behavioral characterization and the role of the painless gene. Genes Brain Behav, 5(8):602–13.

Yang, H. Y., Fratta, W., Majane, E. A., and Costa, E. (1985). Isolation, sequencing, synthesis, and pharmacological characterization of two brain neuropeptides that modulate the action of morphine. Proc Natl Acad Sci U S A, 82(22):7757–61.

Yeomans, D. C., Pirec, V., and Proudﬁt, H. K. (1996). Nociceptive responses to high and low rates of noxious cutaneous heating are mediated by different nociceptors in the rat: behavioral evidence. Pain, 68(1):133–140. REFERENCES 101

Yeomans, D. C. and Proudﬁt, H. K. (1996). Nociceptive responses to high and low rates of noxious cutaneous heating are mediated by different nociceptors in the rat: electrophysiological evidence. Pain, 68(1):141.

Yoburn, B. C., Morales, R., Kelly, D. D., and Inturrisi, C. E. (1984). Constraints on the tailﬂick assay: morphine analgesia and tolerance are dependent upon locus of tail stimulation. Life Sci, 34(18):1755–62.

Zariwala, H. A., Miller, A. C., Faumont, S., and Lockery, S. R. (2003). Step response analysis of thermotaxis in caenorhabditis elegans. J Neurosci, 23(10):4369–77.

Zeilhofer, H. U. (2005). Synaptic modulation in pain pathways. In Reviews of physiology, biochemistry and pharmacology, pages 73–100. Springer.

Zeng, H., Gragerov, A., Hohmann, J. G., Pavlova, M. N., Schimpf, B. A., Xu, H., Wu, L.- J. J., Toyoda, H., Zhao, M.-G. G., Rohde, A. D., Gragerova, G., Onrust, R., Bergmann, J. E., Zhuo, M., and Gaitanaris, G. A. (2006). Neuromedin u receptor 2-deﬁcient mice display differential responses in sensory perception, stress, and feeding. Mol Cell Biol, 26(24):9352–63.

Zheng, Y., Brockie, P. J., Mellem, J. E., Madsen, D. M., and Maricq, A. V. (1999). Neu- ronal control of locomotion in c. elegans is modiﬁed by a dominant mutation in the glr-1 ionotropic glutamate receptor. Neuron, 24(2):347–361.