“Reinventing the wheel” to advance the development of pain therapeutics

Abstract

Chronic pain affects approximately one third of the population worldwide. The primary goal of animal research is to understand the neural mechanisms underlying pain so better treatments can be developed. Despite an enormous investment in time and money, almost no novel treatments for pain have been developed. There are many factors that contribute to this lack of translation in drug development. The mismatch between the goals of drug development in animals (inhibition of pain-evoked responses) and treatment in humans (restoration of function) is a major problem. To solve this problem, a number of pain-depressed behavioral tests have been developed to assess changes in normal behavior in laboratory animals. The use of home cage wheel running as a pain assessment tool is especially useful in that it is easy to use, provides an objective measurement of the magnitude and duration of pain, and is a clinically relevant method to screen novel drugs. Pain depresses activity in humans and animals, and effective analgesic treatments restore activity. Unlike traditional pain-evoked tests (e.g., hot plate, tail flick, von Frey test), restoration of home cage wheel running evaluates treatments for both antinociceptive efficacy and the absence of disruptive side effects (e.g., sedation, paralysis, nausea). This article reviews the literature using wheel running to assess pain and makes the case for home cage wheel running as an effective and clinically relevant method to screen novel analgesics for therapeutic potential.

Introduction

Two goals of pain research with animals are to understand the mechanisms underlying pain and to develop treatments for pain. An understanding of the molecules, neurons, and circuits underlying pain has progressed dramatically over the past 50 years. In contrast, the development of novel treatments for pain has largely failed. Most “new” pain treatments are merely modifications of existing analgesic drugs (Kissin 2010). This mismatch between our increased knowledge and poor treatment options suggests there is a fundamental problem with the translation of results from animal research to clinical practice. This external validity problem has several potential causes: a) The neurobiology of pain in animals and humans may be sufficiently different to preclude translation; b) Animal models of specific pain conditions may not recapitulate the condition in humans; and c) The methods used to assess pain (nociception in animals) are insufficient.

Although all three of these problems contribute to the failure in drug development, research suggests that the neurobiology of pain in animals and humans is similar enough that translation should be possible. Likewise, animal models of human pain conditions have advanced dramatically and will continue to advance with the use of genetic techniques to mimic human disease. However, these advances alone will not improve the development of new analgesic compounds unless parallel advances in preclinical pain assessment also occur. The objective of this review is to make the case that the use of home cage wheel running to assess pain in rodents is a clinically relevant screening tool for new pain therapeutics.

Alignment of Treatment Goals in Animal Studies and Pain Patients

Identification of effective pain therapeutics in animals is not possible without careful consideration of the treatment goal in pain patients. Drugs that block pain, such as opioids, may relieve suffering from a particular pain problem, but can cause dangerous side effects such as respiratory depression, sedation, and dependence. The current opioid crisis, highlighted by the dramatic increase in opioid overdose deaths in the last decade (Volkow and Blanco 2020), clearly demonstrates the problem with focusing exclusively on pain inhibition. Clinical recommendations for the use of opioids during the 1990s and 2000s emphasized reducing pain intensity, marked by the campaign to make pain the “5th Vital Sign” (Henry et al. 2017; Scher et al. 2018). In contrast, recent guidelines for using opioids published by the Centers for Disease Control and Prevention (Dowell et al. 2016), recommend that reducing pain intensity should not be the primary treatment goal (Ballantyne and Sullivan 2015). Rather, these guidelines recommend prioritizing a reduction in pain-related functional impairment. In practice, this approach includes reducing pain to facilitate recovery from surgery or dampening suffering in a chronic pain patient so they can return to work. In either case, the goal is recovery of function.

A similar shift to function-based pain assessment has begun to occur in animal research. Traditionally, nociception in animals was assessed by measuring evoked responses to application of a noxious stimulus. These are known as pain-evoked tests. Inhibition of this type of pain has limited translational value. This lack of clinical relevance has driven the development of a wide range of new pain assessment tools (Negus et al. 2006; Tappe-Theodor et al. 2019). These tests include the use of pain-depressed behaviors to assess the effect of pain on normal behavior. Examples include a decrease in eating/drinking, nest building, or activity caused by persistent pain. The goal of drug development using pain-depressed behavioral tests is the same goal for treating pain patients — the restoration of function.

The Problems with Pain-Evoked Tests

The first problem with pain-evoked tests is that many drugs block pain-evoked responses. These drugs are often promoted as analgesics, but anesthetics, paralytics, and poisons can inhibit pain-evoked behaviors (i.e., a paralyzed or dead rat will not respond on the hot plate test). It seems obvious that nociceptive tests that cannot distinguish between a paralytic and an analgesic should not be used to screen drugs for therapeutic potential. Nonetheless, pain-evoked tests have been used extensively for this purpose. Hundreds of published manuscripts describe “novel analgesics” using pain-evoked tests, but none of these drugs have survived translation to clinical use. The one exception is ziconotide (SNX-111; Prialt), an ω-conotoxin peptide that can only be administered intrathecally because of dangerous side effects (McIntosh et al. 1982; McGivern 2007). The second problem is the inability of pain-evoked tests to assess disruptive side effects. Most drugs do not survive clinical trials because of unpleasant and/or dangerous side effects. Even opioids, the most powerful and widely used analgesics for severe pain, are hampered by a wide range of side effects (e.g., sedation, constipation, nausea, respiratory depression, and dependence). None of these side effects are evident when screening drugs using pain-evoked tests. In fact, some side effects may appear to enhance antinociception by limiting movement. The lack of a response on the hot plate test could be caused by sedation, antinociception, or a combination of the two. Of course, side effects can be assessed using other tests, but only if you know what side effects to expect in advance. The therapeutic potential of a drug should be defined by its ability to restore normal activity and that requires reducing pain in the absence of disruptive side effects. A third problem with pain-evoked tests is that the endpoint is the complete inhibition of pain. Acute pain is protective. The treatment goal for chronic pain patients is to dampen abnormal pain without disrupting nociceptive reflexes and behaviors needed for safety. Pain-evoked tests, especially reflexive tests such as the tail flick reflex, do not make this distinction. Analgesic drugs that inhibit nociceptive reflexes put the patient at risk of further injury. The endpoint for a pain test in animals should be the same for a chronic pain patient — the restoration of function — and not the inhibition of nociceptive reflexes.

The Promise of Pain-Depressed Tests

The premise underlying pain-depressed tests is that normal ongoing behavior, whether in humans or other animals, is disrupted by pain. Pain-depressed tests differ from pain-evoked tests in two important ways. The most obvious difference is that the noxious stimulus produces opposite effects on behavior. Pain-evoked tests (e.g., tail-flick, von Frey) increase a behavior (e.g., withdrawal of the tail or paw), whereas pain-depressed tests decrease behavior (e.g, hindpaw inflammation reduces wheel running). This difference is important because it shifts the goal of treatment from inhibition of behavior (e.g., inhibiting paw withdrawal) to restoration of behavior (e.g., restoring wheel running). As described above, anesthetics, paralytics, poisons, or drugs with a bad side effect profile may inhibit pain-evoked responses, but they will not restore a pain-depressed behavior. Restoration of function makes pain-depressed behavioral tests well suited to screen drugs for therapeutic efficacy because a drug must reduce pain in the absence of disruptive side effects. A drug may produce potent analgesia, but if side effects disrupt behavior then clinical application will be severely limited. The power of pain-depressed tests is that they screen drugs for antinociception in the absence of disruptive side effects. The second important difference between pain-evoked and pain-depressed tests is that pain-depressed tests require the induction of persistent pain. Pain-evoked responses can be elicited whether an animal has a persistent pain condition or not (e.g., use of the von Frey test following induction of hindpaw inflammation). The difference is that application of an acute noxious stimulus in an otherwise pain-free animal evokes a withdrawal response, but only transiently disrupts ongoing behavior measured with a pain-depressed test. The induction of a persistent pain condition enhances the clinical relevance of pain-depressed behavioral tests by mimicking chronic pain conditions in humans. A wide range of pain conditions has been shown to depress behavior in animals. The merits and limitations of each test varies depending on the specific test, pain condition, species, and drug treatment (Tappe-Theodor et al., 2019).

Wheel Running as a Pain Assessment Tool

Wheel running is one of many pain-depressed behavioral tests. The assessment can be done in an animal’s home cage or by moving the animal into a test cage. In the laboratory, the wheel is connected to a computer, so each wheel revolution is recorded with a time stamp. The time stamp allows total distance and speed to be assessed for any time frame (e.g., hour, day, or week). Pain is quantified as a decrease in wheel running and antinociception is defined as restoration of activity.

Home cage wheel running is a clinically relevant method to assess nociception because of the many parallels with human behavior. Activity is a voluntary behavior with a clear diurnal rhythm in both rodents and humans (Lockard 1966). The main difference is that humans are primarily active during the light phase of the cycle and rodents are primarily active during the dark phase. Because wheel running data can be collected continuously, the impact of nociception is most evident during the rodent’s active dark phase of the cycle. In contrast, most labs assess nociception during the light phase when rodents would normally be asleep. A decrease in voluntary activity is a common characteristic of chronic pain and a defining symptom for some conditions such as migraine. Likewise, a wide range of pain conditions depress wheel running (see the Analysis of Wheel Running Studies below).

Wheel running is easy to use and is a completely objective and unbiased assessment tool. Although all pain assessment methods aspire to this goal, many tests rely on someone scoring a behavior (e.g., grimace scale; formalin test; pain-depressed nesting; von Frey test) or measuring a latency to respond (e.g., hot plate and tail flick tests). Conscious and unconscious bias is a common threat in that tests that produce unexpected results can be repeated and measurements of the consequence of a behavior can be inaccurate (e.g., weighing burrowing material, measuring food or water consumption). In contrast, home cage wheel running occurs in the absence of anyone in the room while data are collected. The number of wheel revolutions are simply downloaded from the computer for analysis at the end of the day or week.

Home cage wheel running provides a continuous measurement of both the magnitude and duration of pain and analgesia. Injection of CFA to induce hindpaw inflammation in a rat causes a near complete inhibition of wheel running for two days and a gradual recovery to baseline levels of running in 10-12 days (Kandasamy et al. 2016). Administration of morphine restores wheel running for 1-2 hours (Kandasamy et al. 2017c). In contrast, all pain-evoked tests and most other pain-depressed behavioral tests (e.g., nest building, burrowing, ICSS) are limited to specific assessment times, thereby making it difficult to obtain a complete time course of pain and antinociception. Home cage wheel running also limits confounds such as stress-induced antinociception caused by moving the animal to a test chamber. Assessment of wheel running also uses the same behavioral endpoint for all pain conditions, allowing the magnitude and duration of different types of pain and the duration and efficacy of different treatments to be compared.

Problems and Solutions

Although there are many advantages to using wheel running to assess pain and analgesia, readers should be aware of potential problems and shortcomings. First, depression of wheel running is not specific to pain. Wheel running is a measure of animal well-being. Anything that disrupts well-being will depress running, including the unpleasant side effects of drugs (Kandasamy et al. 2017c; Kandasamy et al. 2018a; Kandasamy et al. 2018b), illness (Hopwood et al. 2009), and opioid withdrawal (Kandasamy et al. 2017a). The fact that wheel running is sensitive to many different types of stimuli does not preclude its use as a pain assessment tool. Rather, it enhances its value by simultaneously assessing the overall impact of a treatment. A powerful analgesic that produces unpleasant side effects or illness will not restore wheel running and would not be recommended as a treatment.

Wheel running is extremely variable both within and between animals. Within subject variability is caused by active and inactive periods throughout the day, much as human activity varies over the course of a day. This variability makes analysis of changes in activity over short periods of time such as 5-minute bins nearly impossible. Variability is reduced by assessing activity over longer bins such as hour-by-hour or day-by-day. Between-subject variability is also large with some rats not running enough to be included in data analysis (we set a criterion of 400 revolutions on the baseline day to be included in an experiment) and a few rats running in excess of 10,000 revolutions in a day. This is a particular problem when comparing male and female rodents, because females run significantly more than males (Eikelboom and Mills 1988; Rosenfeld 2017; Purohit et al. 2020). Further, older rodents tend to run less than younger adults (Bartling et al. 2017). Between subject variability can be managed by transforming data to a percent of each rat’s baseline level of running.

The cost of enhanced clinical relevance is in resources such as time and wheels. It can take up to 3 weeks for running to reach a stable plateau (Stevenson et al. 2011). Given that most of the increase in running occurs during the first week, we use a 7- to 8-day baseline period before inducing pain and testing medications. Depending on the pain condition, the testing phase of the experiment can be a single day (e.g., migraine pain) or 3 weeks (e.g., neuropathic pain). Providing each rat 24-hour access to a wheel limits the number of animals that can be tested at a time. The high cost of commercially available computerized running wheels will be a limitation for many labs.

Wheel running requires housing animals individually in order to link running to a particular animal. The use of radio telemetry technology provides a method to group house animals, but data collection is challenging because multiple animals can simultaneously share and/or compete for a wheel (Weegh et al. 2020). One solution is to house animals in pairs or groups and move them to individual cages to assess wheel running, but this approach trades isolation stress for handling and novelty stress. We have found that social housing facilitates recovery of wheel running depressed by pain, but the effects are small relative to the drastic changes in wheel running caused by pain (Stickney and Morgan, 2020).

The ability to detect false positives is another potential limitation of wheel running. Although a wide range of pain conditions have been shown to depress wheel running (see below), the decrease in running could be caused by factors secondary to pain such as stress, depression, or general malaise. False positives can also occur with analgesic treatments. Stimulants such as amphetamine increase wheel running in pain-free animals (Ferreira et al. 2006). The data are less clear in animals experiencing pain-depressed running because stimulants that are dopamine agonists have antinociceptive effects (Bustamante et al. 2004; Meyer et al. 2009; Lazenka et al. 2017). An increase in wheel running is likely the result of both stimulant and antinociceptive effects. In contrast, non-dopamine stimulants such as caffeine do not restore running in injured animals (Cobos et al. 2012). Studies using other pain-depressed tests such as responding for a rewarding stimulus (de la Puenta et al., 2017; Stevenson et al., 2009) support this finding. These studies highlight the need to include a pain-free control group to assess stimulant (and depressant) effects of a drug. Although a drug that increases or decreases activity by itself may or may not have analgesic effects, drugs that inhibit pain-evoked responses and restore pain-depressed behavior are good analgesic candidates (Stevenson et al., 2009).

The ability of exercise to facilitate recovery from pain (Bement and Sluka 2005; Stagg et al. 2011; Grace et al. 2016) has been raised as a criticism of home cage wheel running as a pain assessment tool. The main difference between using wheel running as an independent (exercise) and dependent (pain measurement) variable is that 3 weeks of wheel running is required to reduce pain, whereas pain assessment typically occurs within days. This criticism raises an important question about the relationship between activity and pain. The use of activity to facilitate recovery or manage pain is common and widespread in medical practice. Patients are encouraged to get up and move as soon as possible following surgery or as a way to deal with chronic pain. In contrast, small cages limit the ability of animals to move following the induction of pain conditions and probably exacerbates the pain condition. A running wheel in the home cage provides an opportunity for animals to increase their activity in a manner consistent with clinical recommendations. Thus, instead of being a criticism, the ability to move provided by home cage wheel running enhances the clinical relevance of animal research.

Analysis of wheel running studies

History

The first use of wheel running as a measure of pain and analgesia was published by Clark and colleagues in 2004 (Clark et al. 2004). Clark et al. (2004) demonstrated the efficacy of oxymorphone against postoperative pain following splenectomy in male mice assessed for 24 hours/day in the animal’s home cage. Veterinary medicine expanded upon this finding in subsequent years as exemplified by the occasional use of wheel running to assess animal well-being following surgery (Adamson et al. 2010; Tubbs et al. 2011; Zegre Cannon et al. 2011).

Wheel running was also used occasionally to assess the functional consequences of inflammatory conditions such as arthritis and, in some cases, subsequent restoration of function by analgesics (Loram et al. 2007; Krug et al. 2009; Stevenson et al. 2011; Kolstad et al. 2012). A significant increase in the use of wheel running to assess nociception occurred following publication of a paper by Cobos and colleagues in 2012. A primary goal of this paper was to validate the use of wheel running as a clinically relevant method to assess pain. Wheel running was assessed outside the home cage for one hour. Although bilateral hindpaw inflammation was required to depress wheel running, traditional analgesics such as opioids and nonsteroidal anti-inflammatory drugs, but not the stimulant caffeine, restored wheel running (Cobos et al. 2012).

Many studies in rats and mice using a wide range of pain conditions followed (Table 1). These studies assessed pain caused by pancreatitis (Cattaruzza et al. 2013), inflammation (Grace et al. 2014; Kandasamy et al. 2016; Pitzer et al. 2016b; Sheahan et al. 2017; Kandasamy et al. 2017c; Avrampou et al. 2019; Oto et al. 2019), colitis (Häger et al. 2018; Weegh et al. 2020), surgery (Kendall et al. 2016), neuropathy (Pitzer et al. 2016a; Sheahan et al. 2017; Whitehead et al. 2017; Griffiths et al. 2018; Kami et al. 2018), migraine (Christensen et al. 2016; Kandasamy et al. 2017b; Kandasamy et al. 2018a; Kandasamy et al. 2018b), cancer (Tang et al. 2016), bone fracture (Shi et al. 2018), corneal abrasion (Hegarty et al. 2018), and low back pain (La Porta and Tappe-Theodor 2020).

Table 1:

List of pain conditions that depress and treatments that restore wheel running

Type of Pain	Pain Model	Species	Sex	Acquisition Period Prior to Pain	Wheel in home cage?	Duration of assessment	Analgesic	Restored Function?	Reference
Inflammation	CFA (1 paw)	Rat	M/F	3 or 8 days	Yes	23 h/day	-	-	(Kandasamy et al. 2016)
	CFA (1 paw)	Rat	M/F	8 days	Yes	23 h/day	Morphine	Yes; at low doses	(Kandasamy et al. 2017b)
	CFA (1 paw)	Rat	M/F	8 days	Yes	23 h/day	Buprenorphine	No	(Kandasamy et al. 2017b)
	CFA (1 paw)	Mouse	M	-	Yes	24 h/day	-	-	(Pitzer et al. 2016b)
	CFA (2 paw)	Mouse	M	1 hour/day for 3 days	No	1 hour	Naproxen	Yes	(Cobos et al. 2012)
							Celecoxib	Yes
							Ibuprofen	Yes
							Prednisolone	Yes
							Diclofenac	Yes
							Morphine	Yes; at low doses
							Caffeine	No
	CFA (2 paw)	Rat	M	3 days	Yes (locked for 23 hours)	1 hour	-	-	(Grace et al. 2014)
	CFA (2 paw)	Mouse	M	2 h/day for 2 days (locked)	No	2 hours	-	-	(Sheahan et al. 2017)
	CFA (1 knee)	Mouse	F	Yes, length unknown	No	17 hours	Botulinum Toxin Type A	Yes; at low doses	(Krug et al. 2009)
	CFA (leg)	Mouse	M	3 days	Yes	24 h/day	Meloxicam	No	(Kolstad et al. 2012)
							Acetaminophen	Yes
							Buprenorphine	Yes
	CFA (1 paw)	Mouse (RGS4KO)	M/F	2 days	Yes	1 hour	-	-	(Avrampou et al. 2019)
	Carrageenan (1 knee)	Mouse	F	Yes, length unknown	No	17 hours	Morphine	No	(Krug et al. 2009)
							Buprenorphine	Yes; at low doses
							Botulinum Toxin Type A	No
	Carrageenan (1 paw)	Rat	F	4 days	Yes	12 h	-	-	(Loram et al. 2007)
Arthritis	MIA (1 knee)	Rat	M	7 or 21 days	Yes	24 h/day	-	-	(Stevenson et al. 2011)
	Collagen-induced arthritis (tail)	Mouse	M	0 days	Yes	24 h/day	Tofacitinib	Yes	(Oto et al. 2019)
Visceral	Acetic Acid	Mouse	M/F	3 weeks	Yes	30 min	Morphine	Yes	(Miller et al. 2011)
	Pancreatitis	Mouse	-	7 days	Yes	24 h/day	-	-	(Cattaruzza et al. 2013)
	Colitis	Mouse	F	14 days	Yes, mice were group-housed	19 hours	-	-	(Weegh et al. 2020)
	Colitis	Mouse	F	14 days	Yes	20 hours	-	-	(Häger et al. 2018)
Post-surgical	Laparotomy	Mouse	F	-	-	5 minutes	Buprenorphine Intermittent	No	(Kendall et al. 2016)
							Buprenorphine sustained	Yes
	Hepatectomy	Mouse	M	3 days	Yes	24 h/day	Buprenorphine	Yes	(Tubbs et al. 2011)
							Meloxicam	Yes
							Flunixin meglumine	Yes
	Ventral Laparotomy	Rat	M	3 days	Yes	24 h/day	Carprofen	Yes	(Zegre Cannon et al. 2011)
							Tramadol	Yes
							Carprofen + Tramadol	Yes
	Mammary Fat Pad Removal	Mouse	F	1 day	-	24 h/day	Buprenorphine	No	(Adamson et al. 2010)
							Carprofen	No
							Buprenorphine + Carprofen	No
	Splenectomy	Mouse	M	7 days	Yes	24 hours/day	Liposome-encapsulated oxymorphone	Yes	(Clark et al. 2004)
							Buprenorphine	No
Neuropathy	Spared Nerve Injury	Mouse	M	1 day	No	24 h/day	-	-	(Pitzer et al. 2016a)
	Chronic Constriction Injury	Rat	M	1 h/day for 7 days	No	1 hour	-	-	(Whitehead et al. 2017)
	Partial Sciatic Nerve Ligation	Mouse	-	14 days	Yes	24 h/day	-	-	(Kami et al. 2018)
	Paclitaxel	Rat	M	1 hour on day prior to recording	No	~15 hours	-	-	(Griffiths et al. 2018)
	Chronic Constriction Injury	Rat	M	7 days before surgery	Yes	23 h/day	Morphine	No	(Green-Fulgham et al., 2020)
Migraine	Dural AITC	Rat	F	8 days	Yes	23 h/day	THC	Yes; at low doses	(Kandasamy et al. 2018a; Kandasamy et al. 2018b)
	Dural AITC	Rat	F	8 days	Yes	23 h/day	Morphine	Yes; at low doses	(Kandasamy et al. 2018b)
	Dural AITC	Rat	F	8 days	Yes	23 h/day	Sumatriptan	Yes	(Kandasamy et al. 2017a)
Other	Breakthrough cancer pain	Mouse	M	1 hour/day for 3 days	No	20 minutes	Morphine	Yes	(Tang et al. 2016)
	Tibia fracture	Mouse	M	-	Yes	24 h/day	-	-	(Shi et al. 2018)
	Corneal abrasion	Rat	M	8 days	Yes	23/day	-	-	(Hegarty et al. 2018)
	Low back pain	Mouse	M	0 days	Yes	14 h/day	-	-	(La Porta and Tappe-Theodor 2020)

AITC: allyl isothiocyanate

CFA: Complete Freund’s Adjuvant

MIA: monosodium iodoacetate

RGS4KO: Regulator of G protein signaling 4 knock-out

THC: Δ⁹-tetrahydrocannabinol

Pain Conditions

Inflammation is easy to induce and widely used as a persistent pain condition in laboratory animals. Unilateral injection of carrageenan (100 L) into the paw of female rats depresses wheel running for at least 3 days following injection (Loram et al. 2007). Inflammation of the knee from direct injection of carrageenan also depresses wheel running (Krug et al. 2009). In stark contrast, unilateral hindpaw injection of CFA (20 L) did not depress wheel running in rats or mice when running was limited to 1 hour a day (Cobos et al. 2012; Grace et al. 2014). Unilateral CFA administration or injecting CFA into the L1 dorsal dermatome did not depress wheel running (Grace et al. 2014). This finding raised questions about whether depressed wheel running is a behavioral readout of pain or merely a consequence of damage to the hind paw.

Subsequent studies showed that the sensitivity of wheel running as a measure of inflammatory pain could be enhanced by placing the wheel in the home cage. As mentioned earlier, home cage wheel running reduces confounds from stress-induced antinociception and increased motivation when moving animals to a test chamber with a wheel for a limited time. A single hindpaw injection of CFA nearly completely eliminated wheel running when male and female rats were given continuous access to a wheel in their home cage (Kandasamy et al. 2016). This depression of wheel running is persistent and gradually returns to baseline levels approximately 10 days after CFA injection (Kandasamy et al. 2016). Similarly, a unilateral hindpaw injection of CFA or tail injection of collagen reliably depressed wheel running in mice when assessed for either 1 or 24 hours when the wheel is in the animal’s home cage (Pitzer et al. 2016b; Avrampou et al. 2019; Oto et al. 2019). Inflammatory pain has been the most frequently used condition in studies of pain-depressed wheel running for several reasons: 1) It is easy to induce; 2) It has been well-characterized in pain-evoked tests; 3) The magnitude and intensity of the pain is consistent; and 4) It represents a common clinical problem.

Wheel running is an especially useful method to assess pain for conditions in which pain-evoked tests are difficult to use such as visceral and migraine pain. Wheel running is depressed by visceral conditions such as pancreatitis- and colitis-related pain (Cattaruzza et al. 2013; Häger et al. 2018; Weegh et al. 2020). Induction of colitis in female mice depressed wheel running for 6 days (Häger et al. 2018). Interestingly, colitis also depresses wheel running for 4 days in group-housed female mice – the first study to demonstrate pain-depressed wheel running in group-housed animals (Weegh et al. 2020). Wheel running is useful to model colitis- and pancreatitis- related pain for several reasons. First, wheel running effectively captures the development of acute pancreatitis pain, which occurs in the few days following induction of pain. Second, given that wheel running is not a labor-intensive test, the animal can remain in its home cage and wheel running can be measured for several days beyond the acute phase to capture chronic pancreatitis pain, which can occur 17 days after induction (Cattaruzza et al. 2013). Given the limited treatment options and the prolonged depression of wheel running in animal models of colitis and pancreatitis, home cage wheel running seems like an ideal method to screen novel therapeutics for these conditions.

Migraine pain is difficult to mimic in rodents because of its spontaneous nature and lack of apparent tissue injury. Instead of spontaneous migraine, scientists activate primary afferents in the dura mater to generate migraine-like pain in rats (Oshinsky and Gomonchareonsiri 2007; Edelmayer et al. 2012). Microinjection of allyl isothiocyanate, a TRPA1 agonist, onto the dura mater of female rats depressed wheel running for 3 hours when assessed inside the rat’s home cage (Kandasamy et al. 2017b). The length of this depression of behavior parallels human data in which migraine attacks last at least 4 hours. However, systemic injection of nitroglycerin, a vasodilator that induces migraine headaches in migraineurs, did not depress wheel running in mice when tested for 30 min outside of the mouse’s home cage (Christensen et al. 2016). Whether the failure of nitroglycerin to depress wheel running was caused by testing outside the home cage or lack of direct activation of dural afferents is not known.

A wide range of surgical procedures have been shown to depress wheel running: Splenectomy (Clark et al. 2004), mammary fat pad removal (Adamson et al. 2010), laparotomy (Zegre Cannon et al. 2011; Kendall et al. 2016), and hepatectomy (Tubbs et al. 2011). As in humans (Gamme et al. 2013), surgical pain produces a transient depression in activity in rodents. Home cage wheel running was depressed for only a day or two following ventral laparotomy, hepatectomy or splenectomy (Clark et al. 2004; Tubbs et al. 2011; Zegre Cannon et al. 2011). One aspect of post-surgical pain that has not been assessed is the transition from acute to chronic pain. Wheel running studies examining post-surgical pain should continue beyond the 24 - 48 hour period to determine potential long-term consequences of post-surgical pain. A model of acute surgical pain that transitions to chronic pain would be especially useful.

Neuropathic pain causes significant disruption of function in human patients (Brell 2014). Similarly, models of neuropathic pain, whether from chronic constriction injury, partial nerve ligation, spared nerve injury, or paclitaxel treatment, produce a prolonged depression of wheel running (Pitzer et al. 2016a; Whitehead et al. 2017; Griffiths et al. 2018; Kami et al. 2018). The one exception was when the effects of spared nerve injury were tested for a short period of time outside the animal’s home cage (Sheahan et al. 2017). One problem with chemotherapy-induced neuropathy (e.g., paclitaxel, cisplatin) is that changes in wheel running can be caused by neurotoxicity (Li et al. 2019) or nephrotoxicity (Volarevic et al. 2019) and not pain.

As revealed by the studies presented in Table 1, wheel running can be used to assess nearly any pain condition. Of course, a number of studies have failed to see depression of wheel running (Table 2). Although it is hard to know why wheel running was not depressed, testing outside of the home cage or the use of a transient pain stimulus are likely causes. Some acute stimuli such as intraperitoneal injection of acetic acid are prolonged enough to depress wheel running (Miller et al. 2011), whereas other stimuli such as administration of formalin into the hindpaw do not appear to depress running (Grace et al. 2014; Ulker et al. 2020).

Table 2:

List of studies that do not demonstrate depression of wheel running following pain

Type of Pain	Pain condition	Species	Sex	Acquisition Period Prior to Pain	Wheel in home cage?	Duration of assessment	Reference
Inflammation	Formalin (2 paw)	Rat	M	24 h/day for 3 days; 1 h/day for 4 days prior to injection	Yes, but locked for 23 h/day for 4 days prior to injection	1 h	(Grace et al. 2014)
	CFA (L1 dorsal dermatome)
	Formalin (L1 dorsal dermatome)
	Carrageenan (tail)	Rat	F	4 days	Yes	12 h	(Loram et al. 2007)
	CFA (1 paw)	Mouse	M	1 hour/day for 3 days	No	1 hour	(Cobos et al. 2012)
	CFA (tail)	Mouse	M	0 days	Yes	24 h/day	(Oto et al. 2019)
	Formalin (1 paw)	Mouse	M/F	-	No	-	(Ulker et al., 2020)
Neuropathy	Spared Nerve Injury	Mouse	F	2 h/day for 2 days	Yes, but locked for 22 h/day	2 h	(Sheahan et al. 2017)
Migraine	Nitroglycerin	Rat	F	2-3 days in group; 30 min sessions individually	No	30 min	(Christensen et al. 2016)
Other	Antiretroviral-induced pain	Rat	F	12 days	Yes	24 h	(Weber et al. 2007)

CFA: Complete Freund’s Adjuvant

Effects of analgesics

All of the analgesics tested in studies of pain-depressed wheel running have already been approved for use in humans (e.g., morphine, buprenorphine, ibuprofen), are reformulations of existing medications (e.g., liposome-encapsulated oxymorphone), or are drugs that are exclusively used for veterinary purposes (e.g., flunixin meglumine, carprofen). These studies are necessary to validate the use of pain-depressed wheel running for specific pain conditions. No published studies have used wheel running to screen novel compounds for analgesic activity. The lack of data on novel analgesics is caused in part by the need for validation studies using standard analgesic compounds and the high threshold for success (i.e., antinociception in the absence of disruptive side effects).

The high threshold set by wheel running experiments is important for several reasons: First, the goal of restoring function, is the same for wheel running and pain patients. Doses of morphine and buprenorphine required to restore pain depressed wheel running are much lower than doses required to inhibit pain-evoked behaviors (Krug et al. 2009; Kandasamy et al. 2017c). The ED₅₀ doses for morphine inhibition of pain-evoked responses in rats range from 3 - 5 mg/kg (Morgan et al. 2006), whereas morphine doses of 0.32 - 1 mg/kg restore pain depressed wheel running (Kandasamy et al. 2017c). Likewise, a low dose of ∆⁹-tetrahydrocannabinol (THC) (0.32 mg/kg) restores wheel running depressed by migraine pain, but does not completely inhibit pain-evoked behavior (Craft et al. 2013). Pain-evoked behavior can be inhibited by a higher dose of THC (3.2 mg/kg), but this dose depresses wheel running as do high doses of buprenorphine and botulinum toxin (Krug et al. 2009; Kandasamy et al. 2017c). Despite the antinociceptive effect of high drug doses, side effects prevent restoration of pain-depressed behavior. These results demonstrate the power of pain-depressed tests to screen drugs for clinically relevant outcomes.

Second, wheel running reveals that the efficacy of a particular drug depends on the pain condition. Botulinum toxin A restores wheel running in mice treated with CFA, but not mice treated with carrageenan (Krug et al. 2009). Similarly, buprenorphine restores function after hepatectomy in mice (Tubbs et al. 2011), but not after splenectomy (Clark et al. 2004) or removal of the mammary fat pad (Adamson et al. 2010). We have found that THC restores wheel running depressed by migraine (Kandasamy et al. 2018a; Kandasamy et al. 2018b), but exacerbates depression of wheel running caused by inflammatory bowel disease (personal observation). Wheel running is even sensitive to differences in dosing procedure in that sustained, but not intermittent buprenorphine restores function after laparotomy in mice (Kendall et al. 2016).

Third, the goal of restoration of wheel running will improve translation. Pain-evoked behaviors have been used for many years and have revealed a tremendous amount about the mechanisms underlying pain processing and modulation. Although this information has revealed the complexities of pain, pain-evoked measures have contributed little to the development of new analgesic therapies. Many drugs have been promoted as new treatments for pain on the basis of inhibiting pain-evoked behaviors only to fail in clinical trials. Had restoration of wheel running been used to screen these drugs, few drugs would have been promoted to clinical trials. This more rigid screening process would have frustrated the scientists promoting these drugs, but it would have saved an incredible amount of time and money — time and money that could have been devoted to more promising analgesic drug development.

Recommendations for using pain-depressed wheel running

A wide range of pain conditions have been modeled in animals with a wide range of effects on wheel running. This variability is enhanced by differences in species, sex, location of the wheel (home cage vs. test chamber), duration of wheel running exposure prior to pain, and length of assessment. As seen in Table 1, wheel running methodology varies considerably. Developing a standard methodology will increase the reliability and sensitivity of wheel running and allow comparison of results across studies. Most importantly, a standard running wheel procedure will improve the translation of drugs from the laboratory to the clinic.

Guidelines for the use of wheel running include the use of mice and rats as the most appropriate species. Wheel running is a behavior that rats and mice readily engage in and do so at relatively high rates. Further, various pain models are optimized for both mice and rats so there are numerous opportunities for expanding the use of wheel running beyond what has already been tested (see Table 1). Consistent with recent developments in pain research, both male and female animals should be used.

The location of the wheel, whether inside the home cage or in a separate test chamber or restricting access to the wheel impacts an animal’s stress and drives its motivation to run. Severe pain manipulations (e.g., inflammation of both hindpaws) are required to depress wheel running in both rats and mice when the running wheel is located outside the home cage or access is restricted (Cobos et al. 2012; Grace et al. 2014) because the motivation for the animal to run is much higher if the wheel is only accessible for one hour (Basso and Morrell 2017). In contrast, inflammation of one hindpaw is sufficient to significantly depress wheel running in male and female rats with continuous access to the wheel in their home cages (Kandasamy et al. 2016). The same holds true for neuropathic pain. Spared nerve injury depresses wheel running when mice have continuous access to the wheel (Pitzer et al. 2016a), but failed to depress wheel running when mice had limited access to the wheel (Sheahan et al. 2017).

The amount of time the animal has to obtain stable levels of running has significant implications on the presentation of pain. We have found that unilateral hindpaw inflammation only depressed wheel running for 3 days in animals that received 3 days of continuous wheel access for habituation, but wheel running was depressed for 10 days in animals that received 8 days of habituation (Kandasamy et al. 2016). This difference appears to be caused by a low baseline (it takes at least 7 or 8 days to reach a more stable level of running). A lower baseline would allow rats to recover to baseline faster. This finding is also supported by other studies in which baseline wheel running stabilizes in approximately one week (Bowen et al. 2016; Wolff et al. 2018).

Lastly, the length of the assessment after induction of pain will vary depending on the type of pain. For example, migraine pain is relatively transient in rodents in that chemical activation of dural afferents depresses running for only 3 hours (Kandasamy et al. 2017b). In contrast, induction of visceral pain can depress running for as long as 17 days (Cattaruzza et al. 2013). Thus, the length of the assessment can be as short as 3 hours or long as 4 weeks. The beauty of using home cage wheel running is that wheel running can be continuously monitored to determine when animals return to baseline levels of running.

The Future of Wheel Running: Advances in Technology and Other Parameters

The primary dependent measure in most wheel running studies is the number of wheel revolutions in a designated period of time. Some studies represent this as distance traveled by multiplying the circumference of the wheel by the number of revolutions. Although these parameters are easy to measure and analyze, other dependent variables such as the number of running bouts, speed, duration of running, or quality of running may have important therapeutic implications. Technological advances have been developed to capture wheel rotation fluidity, directionality, velocity, and acceleration (Chomiak et al. 2016) and these can be combined with video monitoring to assess running quality. The fluidity, velocity, and acceleration of wheel running may have important implications for assessing an animal’s ability to function. For example, an injured animal’s velocity, acceleration or quality of running may take longer to recover than the total number of wheel revolutions. Rats with hind paw inflammation will run even though video capture shows a significant limp (Kandasamy et al. 2016).

Technology also provides a way to solve the problem of housing animals individually. Group-housing animals is the obvious solution, but identifying which rat is running on which wheel at which time is challenging. Radio frequency identification (RFID) or telemetry systems can help identify the running performance of individual animals (Mayr et al. 2020; Weegh et al. 2020), although surgical implantation of the necessary hardware creates another confound.

Conclusions

An enormous investment of time, money, and laboratory animals have been devoted to the development of novel analgesics. There is little to show for this effort other than novel formulations of current medications. Although there are many factors that contribute to this failure (Mao 2009; Percie du Sert and Rice 2014; Mogil 2019), the overreliance on pain-evoked tests to screen drugs in animals has been a major problem. The development of pain-depressed behavioral tests and home cage wheel running in particular addresses this problem by providing a clinically relevant method to screen potential pain therapeutics. The endpoint, restoration of function, is the same for wheel running in animals and pain patients. Restoration of function requires inhibition of pain in the absence of disruptive side effects. The ability to simultaneously screen for antinociception and side effects is what distinguishes pain-depressed and pain-evoked behavioral tests. Disruptive side effects depress wheel running but confound interpretation of pain-evoked responses.

We have validated the use of home cage wheel running to screen drugs using standard analgesic treatments in a back-translation process. These studies show that low doses of morphine (0.32 and 1.0 mg/kg) restore wheel running depressed by hindpaw inflammation in rats (Kandasamy et al. 2017c). Although the analgesic effects of morphine do not disappear when the dose increases to 3 mg/kg, side effects (e.g., sedation, nausea) increase causing wheel running to decrease. This decrease in running is evident whether rats are in pain or not. We have found similar effects with administration of low and medium doses of THC to treat migraine-like pain in rats (Kandasamy et al. 2018a).

For each pain condition and potential treatment, a series of questions must be answered such as whether the pain condition depresses wheel running and the treatment restores running. If the treatment does not restore running, three possibilities exist: the drug has no antinociceptive properties, the dose was too small, or the dose was too large. Rigorous analysis of dose-response curves is needed to ensure a drug is either a good or bad candidate for a novel analgesic. We have screened novel drugs that showed promise by inhibiting pain-evoked responses, but none of these drugs were effective at producing recovery of wheel running. Although these findings are disappointing, this outcome is preferable to investing time and money promoting drugs that will ultimately fail to translate to human use. Restoration of wheel running sets a high bar for a drug, but the bar set by pain-evoked tests for the past 50 years has been much too low, wasting billions of dollars and years of research and leaving millions of people in pain. Fewer drugs will be advanced to clinical trials when screened by home cage wheel running, but the drugs that make it through screening will have the greatest chance for success.