Assessing complex movement behaviors in rodent models of neurological disorders

Highlights

• This report describes four specific examples of Force-Plate Actimeter (FPA) analysis of behavior.

• FPA analysis captures and quantifies specific features of behavioral movement phenotypes.

• Examples include nociceptive, circling, place preference, and small complex behaviors.

• FPA analysis is powerful and useful in quantifying minute details of motor behaviors.

• FPA analysis greatly expands the scope and detail of preclinical behavioral phenotyping.

Abstract

Behavioral phenotyping is a crucial step in validating animal models of human disease. Most traditional behavioral analyses rely on investigator observation of animal subjects, which can be confounded by inter-observer variability, scoring consistency, and the ability to observe extremely rapid, small, or repetitive movements. Force-Plate Actimeter (FPA)-based assessments can quantify locomotor activity and detailed motor activity with an incredibly rich data stream that can reveal details of movement unobservable by the naked eye. This report describes four specific examples of FPA analysis of behavior that have been useful in specific rat or mouse models of human neurological disease, which show how FPA analysis can be used to capture and quantify specific features of the complex behavioral phenotypes of these animal models. The first example quantifies nociceptive behavior of the rat following injection of formalin into the footpad as a common model of persistent inflammatory pain. The second uses actimetry to quantify intense, rapid circling behaviors in a transgenic mouse that overexpresses human laminin α5, a basement membrane protein. The third example assesses place preference behaviors in a rat model of migraine headache modeling phonophobia and photophobia. In the fourth example, FPA analysis revealed a unique movement signature emerged with age in a digenic mutant mouse model of Tourette Syndrome. Taken together, these approaches demonstrate the power and usefulness of the FPA in the examination and quantification of minute details of motor behaviors, greatly expanding the scope and detail of behavioral phenotyping of preclinical models of human disease.

Introduction

Behavioral phenotyping comprises the basis of perhaps the most fundamental features of the validation of animal models of human disease. Many of the inroads in the effective use of rodent models of disease and the development of therapeutic interventions have relied upon straightforward observation of changes in the behavioral repertoire of a subject exposed to various stimuli, physiological modifications, or drugs. In rodent models of human disease, if a genetic variant or environmental perturbation (typically neurological) is responsible for behavioral dysfunction, then a rodent model incorporating the variant or perturbation should display corresponding behavioral changes. The use of standardized, robust behavioral analyses of the mutant mouse can assess whether specific alterations lead to disease-relevant phenotypes. Many disorders in humans, particularly those of the nervous system, are associated with deficits in cognitive, sensory, motor, or social aspects of behavior. Behavioral screening using a battery of well-defined tests can help determine whether the rodent model displays behavioral features similar to the human disease. In such tests, the face validity of rodent models has been largely dependent on the similarity of the subject’s movement and actions as reflecting the human condition. While many of the simplest and most hardy behavioral phenotyping approaches have relied on straightforward investigator observation of the animal subjects’ behavior, this approach can be significantly compromised by inter-observer variability, scoring consistency, and the ability to observe extremely rapid, small, or repetitive movements. Accordingly, many useful refinements in behavioral phenotyping have emerged through development of apparatus to objectively and quantitatively measure specific aspects of rodent behavior.

Many common tests of general locomotor activity are based on observation of the subject’s movements when placed in a novel open arena. Rodents have a natural drive to explore new surroundings, and this can be leveraged to assess patterns of movement such as total distance traveled, rearing, circling, bouts of low motility, and stereotypic behaviors (e.g., grooming). Open-field arenas for rodents can be circular or square, and are typically divided into sub-areas to facilitate observation of the patterns of movement around the arena.

Automated recording and measurement of locomotor activity can be accomplished using apparatus that record rodent movement through breakage of photocell beams, electromagnetic detection, or video tracking software (Drai et al., 2001, Grusser and Grusser-Cornehls, 1998, Thifault et al., 2001). Some of these technologies can be adapted to home cages instead of novel test arenas; this avoids anxiety-evoked thigmotaxis, but this loses the advantage of evoking locomotion by exposure to a novel environment. A distinct advantage of automated systems includes labor-reducing standardization of objective, unbiased quantification and recording of locomotion, and multiple options regarding data analysis. Apparatus such as the force-plate actimeter can also be enclosed in a sound-attenuating, opaque chamber, which removes the impact of observer presence altogether (Chesler, Wilson, Lariviere, Rodriguez-Zas, & Mogil, 2002).

Detailed analysis of locomotor activity, exploratory behaviors, stereotypy, tremor and ataxia are facilitated by apparatus such as the BASi Force-Plate Actimeter (FPA) (Bioanalytical Systems, Inc., Mount Vernon, IN), which provides a very rich data stream that records and quantifies minute details of movement. The actimeter arena consists of a square, stiff horizontal plate (42 cm × 42 cm) attached at the corners to force transducers that record the center of mass (X, Y position) of the animal subject. A Plexiglas enclosure rests a few millimeters above the plate to create a transparent enclosure (Fig. 1). Animals are placed into the testing arena and allowed to move freely for the testing period (as little as 10 min for initial screening purposes, but potentially for hours-long assessment). When the test subject moves on the plate, its movements are sensed by the transducers, whose signals are processed by a computer to detect and score a wide range of behaviors or behavioral attributes, including behavioral events occurring at user-defined frequencies or amplitudes. These include, but are not limited to, stereotypies, tremor, startle, or ataxia. The actimeter simultaneously records larger movements such as locomotion (with mm scale resolution) and fine movements (with frequency resolution up to 100 Hz). Analysis of the data gathered during this time reveals several parameters such as total distance traveled (a sum of X, Y deflections of the center of mass over time; includes both large and small movements), area traveled (a two-dimensional integration of X, Y deflections over time), bouts of low mobility (periods when the X, Y position remains within a defined small area over a period of time, typically 10 s), focused stereotypies (an area-partitioned variance of total force deflections over time), left and right turns, and other aspects of both general locomotion and small body movements, including very small movements like whisking (active sweeping movements of the vibrissae that provide sensory input), respiration, and even heart beat (Crosland et al., 2005, Fowler et al., 2003, Fowler et al., 2010, McKerchar et al., 2006, Smittkamp et al., 2008). This apparatus has been particularly useful in analyzing rotational behaviors, inherent or drug-evoked tremor, and seizure-like activity in rodents (Fowler, Miller, Gaither, Johnson, & Rebec, 2009). This is accomplished via decomposition of the force-time data gathered by the FPA using a mathematical Fourier transformation into a quantification of the power of movements as a function of frequency (Fowler et al., 2009), which conceptually would be similar to evaluating the volume of individual notes in a musical chord. Typical frequencies of normal rat behavior include licking (4–7 Hz), respiration (1–2 Hz), heartbeat (6–8 Hz), and whisking (8–25 Hz; (Deschenes, Moore, & Kleinfeld, 2012)). Locomotion is typically large amplitude and very low frequency (<2 Hz).

The first example uses a common model of persistent inflammatory pain: injection of formalin into the rat footpad. In this experiment, the behavioral responses of rats were quantified using the FPA, resulting in much greater detail than possible counting hind limb flinches. The formalin test is an established, widely used model of persistent inflammatory pain (Porro and Cavazzuti, 1993, Tjolsen et al., 1992). This model is usually used in mice and rats, and may provide a better approximation of clinical chronic pain than traditional phasic nociceptive tests, such as acute radiant thermal, hot-plate, or mechanical paw withdrawal tests. Formalin induces a biphasic pain-related behavioral response, separated by a quiescent interphase. The first (early) phase is due to C-fiber activation, while the second (late) phase is due to peripheral inflammation and central sensitization. The behavioral responses that can be evaluated include lifting (flexing), licking, biting, and flinching of the injected paw. The flinching response seems to be a consistent component of the response, and is easy to observe and quantify. The other behavioral responses have been more problematic, and it has been suggested that flinching is less influenced by conditions that affect non-nociceptive behavior, and, thus, is a better end-point (Tjolsen et al., 1992). In the experiments used in this report as an example of actimetry, intensity of nociceptive behaviors evoked by formalin were measured to provide a quantitative means for subsequently assessing their modification by acute, activational effects of estrogen (Ralya & McCarson, 2014).

The second example uses actimetry to quantify intense circling behaviors in a transgenic mouse that overexpresses human laminin α5, which is a basement membrane protein. In this model, a bacterial artificial chromosome containing a portion of human chromosome 20 including the entire laminin α5 gene was phenotyped. Laminin α5 is required for kidney glomerular basement membrane assembly, but is also found in vascular basement membranes located throughout much of the body, including the CNS. The hemizygous transgenic mice display normal kidney function with little or no urinary albumin as assessed by ELISAs, normal blood urea nitrogen levels, and normal ultrastructure of the kidney GBM (Steenhard et al., 2011). However, the homozygous mice also displayed a robust, hyperlocomotive circling behavior far too rapid to quantify (or even observe clearly) with traditional scoring approaches. The behavioral phenotype of the transgenic mice was assessed through a battery of common tests as well as analyses of locomotor activity using a FPA.

The third example shows the utility of modification of the force-plate actimeter to assess place preference behaviors in a rat model of migraine headache. In this case, the Lucite cover over the FPA was redesigned to allow the rat to choose between dark/bright or loud/quieter zones of the plate, allowing quantification of time spent in each zone while allowing the simultaneous assessment of other movement phenotypes (Vermeer, Gregory, Winter, McCarson, & Berman, 2014). Studies of migraine pathogenesis have fallen behind other areas of pain research because of the lack of an adequate rodent behavioral model. Animal models of migraine have been hampered by limitations that reduce their face validity, such as electrophysiological end points that do not encompass nociceptive behaviors (Burstein and Jakubowski, 2004, Oshinsky and Luo, 2006), or behavioral models that rely on orofacial testing of restrained subjects (Oshinsky and Gomonchareonsiri, 2007, Stucky et al., 2011, Wieseler et al., 2010). Most animal studies of migraine have used electrophysiological techniques to assess sensitization of neurons, but none have specifically tested behaviors associated with the clinical criteria for diagnosing migraine in humans (2004). In clinical assessment, migraine headache must be accompanied by phonophobia or photophobia (2004). The rodent model in these studies uses application of “inflammatory soup” to the dura of untethered, conscious, behaving rats as a model of sterile inflammation caused by neurogenic activation of mast cell degranulation (Burstein et al., 1998, Burstein and Jakubowski, 2004, Stucky et al., 2011, Vermeer et al., 2014, Vermeer et al., 2015). The previously unparalleled face validity of this preclinical rodent migraine model may provide a means for more detailed and predictive testing of novel therapeutic interventions for migraine.

In the fourth example, FPA analysis was used to identify a band of repetitive movements that emerge with age in a digenic mutant mouse model of Tourette Syndrome. This experiment leveraged human genetic information from Dr. Sarah Soden’s efforts at Children’s Mercy Hospital (Kansas City) (Soden et al., 2014), that described a family with a father and three affected sons. Through whole exome sequencing a mutation was found in the coding region of the nerve growth factor-related receptor tyrosine kinase-like WNT receptor 1 (ROR1) gene. An additional variation in these patients was identified in the nerve growth factor (NGF) gene. NGF is a neurotrophic factor essential for the survival and maintenance of numerous types of neurons in the peripheral and central nervous systems (Levi-Montalcini, 1987, Pezet et al., 2006, Reichardt, 2006). We hypothesized that impaired NGF expression or loss of signaling by NGF in conjunction with altered ROR1 function could provide a basis for the behavioral phenotypes observed in the affected patients.

These four experimental examples of the assessment of movement have been useful in helping define the specific rat or mouse models of human disease. The focus of this report does not concern the underlying pathophysiological details of the rodent models used, but rather the many ways that FPA analysis can be used to capture and quantify specific features of the complex behavioral phenotypes of these animal models.