PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
 
PLoS One. 2017; 12(9): e0181068.
Published online 2017 September 6. doi:  10.1371/journal.pone.0181068
PMCID: PMC5587114

Automated recording of home cage activity and temperature of individual rats housed in social groups: The Rodent Big Brother project

William S. Redfern, Conceptualization, Funding acquisition, Investigation, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing,1,* Karen Tse, Data curation, Formal analysis, Investigation, Methodology, Writing – review & editing,1 Claire Grant, Data curation, Investigation, Methodology, Writing – review & editing,2 Amy Keerie, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – review & editing,1,¤a David J. Simpson, Investigation, Methodology, Validation, Writing – review & editing,2 John C. Pedersen, Investigation, Methodology, Validation, Writing – review & editing,1 Victoria Rimmer, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – review & editing,2,¤b Lauren Leslie, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – review & editing,2 Stephanie K. Klein, Formal analysis, Investigation, Methodology, Validation, Writing – review & editing,1 Natasha A. Karp, Data curation, Formal analysis, Validation, Writing – original draft, Writing – review & editing,3 Rowland Sillito, Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Writing – review & editing,4 Agis Chartsias, Data curation, Formal analysis, Software, Validation, Writing – review & editing,4 Tim Lukins, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – review & editing,4 James Heward, Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Visualization, Writing – review & editing,4 Catherine Vickers, Conceptualization, Project administration, Writing – review & editing,5 Kathryn Chapman, Conceptualization, Project administration, Writing – review & editing,5,¤c and J. Douglas Armstrong, Conceptualization, Funding acquisition, Project administration, Resources, Software, Supervision, Visualization, Writing – review & editing4,6
Judith Homberg, Editor

Abstract

Measuring the activity and temperature of rats is commonly required in biomedical research. Conventional approaches necessitate single housing, which affects their behavior and wellbeing. We have used a subcutaneous radiofrequency identification (RFID) transponder to measure ambulatory activity and temperature of individual rats when group-housed in conventional, rack-mounted home cages. The transponder location and temperature is detected by a matrix of antennae in a baseplate under the cage. An infrared high-definition camera acquires side-view video of the cage and also enables automated detection of vertical activity. Validation studies showed that baseplate-derived ambulatory activity correlated well with manual tracking and with side-view whole-cage video pixel movement. This technology enables individual behavioral and temperature data to be acquired continuously from group-housed rats in their familiar, home cage environment. We demonstrate its ability to reliably detect naturally occurring behavioral effects, extending beyond the capabilities of routine observational tests and conventional monitoring equipment. It has numerous potential applications including safety pharmacology, toxicology, circadian biology, disease models and drug discovery.

Introduction

Behavioral studies in academic research and drug discovery are commonly conducted on rats, which are also the preferred rodent species for toxicology and safety pharmacology studies [1;2;3]. In drug discovery, such studies are used to inform critical decisions; in preclinical safety assessment they are required by regulatory authorities prior to human exposure [4]. Conventional approaches to recording the behavior of laboratory rats are limited to ‘snapshot’ assessments, whether for academic research or in the pharmaceutical industry. Typical examples include manual observations by a trained observer using multi-parameter assessments of global neurobehavioral effects (e.g., the Irwin test or Functional Observational Battery (FOB) [2;5;6;7;8;9;10], or a scoring scale for abnormal behaviors (e.g., stereotypies [11;12;13]; seizures [14]). For convenience, behavioral assessments are usually performed during the daytime, when these nocturnal animals are generally less active. Ambulatory activity measurements can be automated using photobeam arrays [15;16;17;18;19;20], a passive infrared sensor [21;22], a stabilimeter (‘rocking cage’/’jiggle box’) [15;23] or birds-eye videotracking [24;25;26;27]. A wider set of behaviors can be detected using pressure sensors beneath the cage [28;29;30] or by videoanalysis software [31]. These automated methods necessitate the rats, which are social animals [32;33], to be singly housed, typically in bespoke cages, and generally on bench tops rather than cage racks, requiring dedicated space. In mice, methods have been described using radiofrequency identity (RFID) transponders or differently colored fur dyes for recording ambulatory activity of individuals when group-housed [34;35;36;37;38;39]; with a recent exception [39], these required the use of bespoke cages. This presents a significantly greater challenge for rats, which are ~10 times larger in size, therefore with their RFID transponder potentially at a greater distance from the baseplate RFID reader (located beneath the cage floor) than with mice.

As well as general activity and behavior we are (or should be) interested in what is happening to the physiology of our rats: temperature, for example, is a key indicator of physiological homeostasis. Rats respond to ingestion of toxic agents by lowering their core temperature [40;41;42], which affects the absorption, metabolism and excretion of drugs [43] and their overall toxicity [40;41;42]. Larger decreases in temperature may reflect a more profound toxicity, a pharmacological effect on central thermoregulatory control, impaired metabolic heat production, vasodilatory heat loss, or a physiological response to hypoglycemia [44] or hypoxia [45]. Conversely, increases in temperature in rats may reveal other toxicological mechanisms, including skeletal muscle toxicity or an immunological response [46]. Core temperature can be recorded via a rectal thermistor or by radiotelemetry, both of which have their drawbacks. Measurement of rectal temperature requires manual restraint which itself elevates core temperature [47], whereas intraperitoneal radiotelemetry necessitates surgical laparotomy and is relatively expensive [42]. Other methods, e.g., a subcutaneous RFID transponder read-off through the cage wall via a hand-held transceiver [48;49;50] or infrared imaging [27], only enable manual, snapshot measurements.

In 2011 AstraZeneca set a challenge under the National Centre for the 3Rs (NC3Rs; [51; 52]) inaugural CRACK IT open innovation scheme [53], calling for novel technology to record activity, behavior and temperature of individual rats continuously when group-housed in conventional individually ventilated home cages in a portable cage rack (the ‘Rodent Big Brother’ project). The specifications also included the requirement for a rack-based (rather than bench top) system, an important feature with regard to welfare and husbandry requirements, incorporation into long-term studies, and space constraints. We have developed home cage continuous monitoring technology (Fig 1) that can acquire ambulatory activity and subcutaneous temperature of individual rats when group-housed, for at least 28 days, 24 hours per day. The system avoids invasive surgery and the technology is unobtrusive. In addition it acquires continuous side-view high-definition (HD) video of the entire cage, which can be used to extract additional behaviors from the video. This paper describes the technological development, evaluation, optimisation and validation of the approach.

Fig 1
Schematic overview of the Rodent Big Brother (Home Cage Analyzer; ActualHCA™) system.

Methods

Description of the hardware

The Home Cage Analyser (ActualHCA™) system (Actual Analytics Ltd, UK) fits in a standard IVC cage rack (e.g., SealSafe Blue-line range, Tecniplast S.p.A., Buguggiate, Italy), with infrared lighting strips located above the cage, a baseplate beneath the cage, and a camera, computer and power supplies in a vacant cage slot to the side of the cage (Fig 1). This obviates the need for either bespoke caging or benching, and enables the rats to be group-housed in their normal home cage in an adapted IVC cage rack.

RFID dual identification and temperature transponders (factory calibrated) were supplied by BioMark/Destron (Boise, ID 83702, USA). The Biomark BioTherm13 Passive Integrated Transponder (PIT) is an RFID device that complies with the specifications of ISO Standards 11784 (ID code compatibility) and ISO 11785 (communications protocol). This PIT Tag is packaged in a laser-annealed glass ampoule that is designed specifically for subcutaneous (or intramuscular) implantation. Dimensions are 2.12 ± 0.10 mm in diameter x 13 ± 0.4 mm in length. Temperature recording range: 33.0°C to 43.0°C; accuracy ± 0.5°C (factory pre-calibrated).

For the baseplate RFID reader, Actual Analytics worked with BioMark USA to specify and design a baseplate that would work with the BioTherm 13 RFID transponders. This comprised a 2D arrangement of twelve transceiver coils (in a 3 x 4 array) in a waterproof casing. The dimensions of the casing were 38 x 50 cm, approximately the dimensions of the base of the rat cage.

Each antenna is powered up sequentially and then the strongest RFID tag within the vicinity of the electromagnetic field is read (Fig 2). A complete cycle of the 12 antenna takes 1.1s (0.93 Hz). Note that due to the physics of the system, an RFID tag may be reported on the nearest as well as adjacent antennae, if two tags are present on an antennae, only one will be reported and finally the tag needs to be within the active detection field for sufficient time for it to be charged and read (~60 ms) therefore rapidly moving animals can be missed for a few cycles. Spatial filtering is used to resolve the most likely location given an array of reads and interpolation to estimate mostly likely positions between reads. In all cases both raw data and filtered/interpolated data are retained and marked as such. The automated ambulatory tracking will be truncated compared to the actual path taken by the animal (Fig 3).

Fig 2
Schematic overview of RFID transponder activation and detection by baseplate.
Fig 3
Illustration of difference between actual track of the rats and automated tracking of RFID transponder to nearest of 12 antennae in the baseplate.

To enable continuous collection of HD video, each module incorporates strips of infrared LEDs at 860 nm wavelength, to illuminate the cage from above. A USB 3.0 camera with matched 4.5mm lenses and daylight filters (700 nm cut-off) captures infrared (grayscale) video at 25 fps at HD (720p) resolution.

During the course of this project, the performance of the Actual HCA hardware was improved by increasing the magnetic field strength, achieved by a system upgrade that included tuning the baseplates and additional Faraday-shielding of sources of stray electromagnetic fields (notably, the rack-located power supplies). This is collectively referred to as the ‘shielding upgrade’ within this article.

Software and architecture

Data capture

A small onboard computer runs a package called Actual HCA Capture™ (Actual Analytics Ltd, Edinburgh, UK), to manage the video capture, calibrate the system, and reboot the system in the event of power failures. Chunks of video (user-defined but typically 15–30 minutes) and matched baseplate data are captured and stored to a local hard drive.

Data analysis

A second piece of software, Actual HCA Analyser™ (Actual Analytics Ltd, Edinburgh, UK), which reads and analyses the raw data generated by Actual HCA Analyser™ can be run on any remote computer. It provides the opportunity to view the data, analyse/filter raw data and produce video overlays combining analytics and video. Raw data of movement and temperature from the RFID transponder are recorded via the baseplate as the number of transitions between baseplate antennae, and subcutaneous temperature, both tied to each individual animal. Video footage can be analysed using motion detection for cage level events (e.g., video-based motion detection) or to extract individual behaviors. The software also includes a range of analytical algorithms (e.g. activity statistics; circadian rhythm analysis; location preference; etc.).

Server

A third and optional software component helps manage a multi-cage installation. It automates the copying of the data files from individual units onto a server. It also automates some of the post capture data processing thereby reducing the number of processes that the Analyser in the computer of each enclosure has to do. This speeds up analysis and interaction with the data. It also manages the application of post-hoc analytics for automated behavioral detection that is then available for review on the Analyser.

Assessing RFID transponder efficiency at different heights and orientations above the baseplate

Evaluation of the impact of the spatial orientation and height of the RFID transponders was achieved by placing an RFID transponder on an adjustable non-metallic platform on the surface of a baseplate. The device was placed above each baseplate for 60 s over a height range of 4–13 cm, in 1 cm increments, and the signal read frequency for each recording was analysed. Before and after the ‘shielding upgrade’, we compared the average read rate across 4 baseplates using 2 RFID transponders in the vertical orientation at different heights above each of the 12 antennae of each baseplate. After the ‘shielding upgrade’, an evaluation was undertaken using a single RFID transponder in different positions and orientations with respect to a pair of antennae on one baseplate, namely vertical; horizontal aligned towards the centre of an antenna (X-plane); or at right angles to this orientation (Z-axis). This was done in order to evaluate the effects of spatial positioning of the RFID transponder above and between antennae, within the range expected to occur in vivo.

In vivo studies

Ethical statement

The use of animals was kept to an absolute minimum required to achieve statistical significance for validation purposes; a total of 48 rats were used for the work described in this paper. All procedures were conducted in accordance with the United Kingdom Animal (Scientific Procedures) Act 1986, approved by institutional ethical review committees (Alderley Park Animal Welfare and Ethical Review Board and Babraham Institute Animal Welfare and Ethical Review Board) and conducted under the authority of the Project Licence (40/3729 and 70/8307, respectively). All animal facilities have been approved by the United Kingdom Home Office Licensing Authority and meet all current regulations and standards of the United Kingdom.

Animals, housing and husbandry

Male Han Wistar rats (weight range 200–275 g; age range 7–13 weeks at start of data acquisition) were obtained from Charles River Laboratories in Margate, Kent, UK. Vendor-supplied health reports indicated that the rats were free of known viral, bacterial and parasitic pathogens. They were housed in groups of 3 in Tecniplast IVC cages (model number 1500U), in a 12 h light: dark cycle with room temperature set to 21°C (recorded range: 16.9–24.5°C), in a semi-barrier facility. The cages contained a 1–1.5 cm layer of 4 mm3 Aspen chip bedding together with environmental enrichment (sizzlenest nesting material, Datesand) medium Aspen brick chew sticks (Datesand) and a red plastic play tunnel (transparent in infrared light). The rats had access to food (RM1 (E) IRR 0.25 pelleted diet, Special Diet Services, UK) and water ad libitum. Rats were allocated to cages on arrival and remained in the same social group throughout the study. They were identified by waterproof tail markings. Rats were test-naïve prior to the studies. Animal welfare was assessed throughout by daily monitoring of appearance, behavior and cage environment. Rats were allowed to acclimatise to the animal unit for at least 1 week before implantation of the RFID transponder, with at least a further 2 days of recovery post-implantation before using acquired data.

Selection of optimal implantation site for the RFID transponder

Four different subcutaneous injection sites for the RFID transponder were compared: interscapular, flank: vertical orientation, flank: horizontal orientation, and ventral midline (‘vertical’ and ‘horizontal’ refer to the approximate orientation of the RFID transponder with the rat standing on all four limbs; Fig 4). In the initial optimisation study, each rat was implanted with one transponder under brief anaesthesia with isoflurane (induced at 4.5%, maintained at 3.0–4.0%) during implantation. The experiment was undertaken over a time period of several weeks, in 3 cohorts of rats, n = 6 (i.e., 2 cages) per implantation site. As there was no other prior information a group size of 6 males was chosen, which is commonly used in the Irwin test/FOB [7;10]. A follow-up study was undertaken comparing the two best sites evident from the initial comparison study (ventral midline and flank: vertical), with two sets of 6 rats in a direct, head-to-head comparison of these two sites, following improvement of baseplate performance by the ‘shielding upgrade’. For the ventral midline site there is limited loose skin and care has to be taken not to inject the transponder intraperitoneally; creating a small subcutaneous pocket prior to inserting the pre-sterilised trocar needle, ejecting the RFID transponder, and sealing the entry site with surgical adhesive was preferred.

Fig 4
Schematic illustration of the 4 implantation sites/orientations for the RFID transponder evaluated.

The performance/experience of each site was compared across the following pre-set criteria (Table 1): ease of implantation; lack of inadvertent intraperitoneal injection/migration from original injection site/external extrusion; RFID signal strength (read frequency); baseplate individual ambulatory activity tracking vs. birds-eye individual manual tracking (for 60 min, achieved by mounting a video camera pointing vertically downwards through the transparent cage roof; Fig 3); baseplate individual ambulatory activity tracking vs. side view motion detection (2 cages of 3 rats; 24 h and 3–4 weeks); light-dark phase ambulatory activity contrast; light-dark phase temperature contrast; histopathology of implantation site to detect any evidence of inflammation.

Table 1
Criteria used to select optimal subcutaneous RFID implantation site/orientation.

Validation of vertical activity measurements

This was undertaken on video footage from the above studies. Male Han Wistar rats were recorded continuously for 7 consecutive days (frame rate set at 25 frames per second) when housed in groups of 3 in standard individually ventilated cages. Episodes of vertical activity, defined as any movement above a vertical threshold (set at 8 cm at the far wall of the cage), and rearing behaviour (when front paws were elevated and rat was in an upright posture) were manually annotated using ActualTrack™ on two 1-hour samples from the light phase and the dark phase, respectively (Fig 5).

Fig 5
Assessing the precision of the vertical activity measure.

The automated detection of vertical activity per cage was correlated against manual annotation of vertical activity and against the more strictly defined rearing behaviour for each cage from both light and dark phase video footage. The accuracy between the automated detection and manual annotations was determined by frame-by-frame accuracy over each 15-min bin from the light and dark phase video footage. The vertical activity of individual rats in each cage was also identified for the automated detection using the individually implanted subcutaneous RFID transponder. Episodes of rearing behaviour of individual rats in each cage were manually counted from the light and dark phase video footage, to compare with the automated detection.

Initial in vivo observations using the new technology

Side-view motion detection (pixel movement) was undertaken on continuous 7-day HD video from 12 separate cages, each containing 3 rats, tested in the same facility spanning an 8-month period. This comprised all the rats used in the present study. Although they had RFID transponders in different subcutaneous locations, this was not relevant to the video analysis. The 12-hour activities for light phase and for dark phase were averaged separately over a 7-day period.

Two environmental changes were introduced (separately) in order to evaluate whether the ActualHCA™ system could detect perturbations: single-housing rats for 16 h (to assess effects on subcutaneous temperature), and cage changing (to assess effects on ambulatory activity). Cage changing occurs once per week for laboratory rats in our facility, and comprises relocation of the social group of 3 rats into a clean cage with fresh bedding and environmental enrichment. This is known to stimulate activity temporarily as the animals explore their new surroundings. The effects of a dosing procedure (oral gavage with water, 10 mL/kg) was also evaluated.

Terminal procedures and histopathology

All animals were euthanized by overdose of isoflurane in accordance with the Humane Killing of Animals under Schedule 1 to the Animals (Scientific Procedures) Act 1986. The subcutaneous implantation sites were grossly examined for abnormalities and thereafter collected including the RFID microchip in-situ. Tissues around the implantation site were fixed in 10% neutral buffered formalin. All tissues were processed to wax blocks, sectioned, stained with hematoxylin & eosin (H&E), and examined microscopically.

Statistical analysis

Assessing RFID transponder efficiency at different orientations

A linear regression analysis, fitting Eq 1, was used to explore the variation in the read rate from the RFID transponder when a range of elevations and orientations were tested across a number of different baseplates. The model was developed based on domain knowledge, following graphical exploration of data. An interaction between baseplate*shielded was considered but not found to be significant (data not shown). Terms were selected as significant in explaining variation at p<0.05 threshold. Graphically the behaviour of the residuals were explored to assess the quality of the model fit and found the model was a good fit to the data (data not shown).

read rateshieldedheightRFID.IDbaseplate
[Eq 1]

A mixed model regression analysis, fitting Eq 2, was used to explore the variation in the read rate from the RFID transponder fitted in vivo for a range of positions within the animals. Implantation site, baseplate and whether the instrument was shielded were treated as fixed effect and rat was treated as a random factor to account for the repeat readings taken from an individual animal. Model optimisation included a test of the covariance structure, comparing either homogenous or heterogeneous variance across implant site using a likelihood ratio test. Terms were selected as significant in explaining the variation at p<0.05 threshold. Graphically the behaviour of the residuals were explored to assess the quality of the model fit and found the model was a good fit to the data (data not shown).

read rateshieldedbaseplateimplant site + 1
[Eq 2]

Stability of RFID transponder performance

A two-sided paired t-test was used to compare the read rate average readings from week 1 to week 4. To ensure the statistical test was appropriate for the data, the distribution of the difference was explored graphically and found to be normally distributed.

Validation of measure

To evaluate agreement between two approaches to a measure a Bland-Altman plot [54] was used to analyses the distribution of the differences between measures. The intraclass correlation coefficient (ICC)(method: ICC1k average absolute agreement) [55] was calculated as a measure of agreement.

Sources of variation in temperature measure

A mixed model regression analysis, fitting Eq 3, was used to explore the variation in the temperature reading for data collected over four weeks. Terms were selected as significant in explaining the variation at p<0.05 threshold. Graphically the behaviour of the residuals were explored to assess the quality of the model fit and found the model was a good fit to the data (data not shown).

dependent variabledaycageweekbaseplatephase + 1|Rat
[Eq 3]

Sources of variation in number of transitions measure

A mixed model regression analysis, fitting Eq 3, was used to explore the variation in the number of transitions when summed in 15 minute bins for data collected over four weeks. Terms were selected as significant in explaining the variation at p<0.05 threshold. Graphically the behaviour of the residuals was explored to assess the quality of the model fit (data not shown) and the model had issues with the data being bound due to a high presences of zeros where the animal did not move. An alternative Poisson model was not suitable due to over-dispersion. A negative binomial model was fitted and diagnostics were good, providing the same conclusions as the mixed model regression. The standard regression results are presented within the manuscript as they are easier to interpret.

Effects of cage changing on activity and temperature and the effects of single housing on subcutaneous temperature were not evaluated by statistical analysis, as these were casual observations rather than pre-designed experiments.

Data are expressed as mean ± standard error of the mean (SEM) unless otherwise stated.

Data availability

All data generated or analysed during this study are included in this published article or are deposited in Zenodo: 10.5281/zenodo.804041

Results

Assessing RFID transponder efficiency in different spatial orientations

‘Ex vivo’, we explored a range of elevations/orientations of the RFID transponder above and around different baseplates, in situ in the cage rack in the animal housing room. This enabled us to map the RFID signal strengths, which result from the interplay between the strength and direction of the magnetic fields generated by the baseplate antennae, and the orientation and height of the RFID transponder, across the baseplates (Figs (Figs22 and and5).5). Fig 6 indicates that the read frequency (‘read rate’) will vary as an animal moves across the baseplate, or rears/climbs. A regression analysis assessing for sources of variation of read rate found only two variables apart from height that were statistically significant (S1 and S2 Figs), namely the ‘shielding upgrade’ and inter-baseplate variability. These were statistically significant but only had a small effect on read rate: the ‘shielding upgrade’ increased the read rate (by 0.06 ± 0.01 Hz; P = 0.0000125) and one of the baseplates had a tendency to lower read rates (by 0.052 ± 0.02 Hz; P = 0.012).

Fig 6
Impact of height and position on RFID transponder read rate.

In vivo, four different locations/orientations were evaluated for subcutaneous implantation of the RFID transponder, namely interscapular, flank: vertical, flank: horizontal, and ventral midline (6 rats per implantation site) (Fig 4). A regression analysis assessing for sources of variation within the read rate in vivo found that differences between individual baseplates had a statistically significant but minor impact (considered to be within normal manufacturing tolerances) whilst the ‘shielding upgrade’ increased the read rate considerably (by 0.14 ± 0.02 Hz; Fig 7A and S3 Fig). The impact of transponder position was even more substantial with the best read rate seen with the RFID transponder implanted in the ventral midline position (0.74 ± 0.02 Hz, pre-‘shielding upgrade’). Compared to the ventral midline position, the read frequency from the flank: vertical site was on average 0.1 ± 0.02 Hz lower, the flank: horizontal site was 0.3 ± 0.04 Hz lower and the interscapular site was 0.44 ± 0.03 Hz lower. The ‘shielding upgrade’ improved the read rate for ventral midline (and flank: vertical), as illustrated in Fig 7A.

Fig 7
RFID read frequencies from the 4 implantation sites, improvement by baseplate ‘shielding upgrade’, and maintenance of RFID performance over 4 weeks of continuous in vivo use.

Stability of performance of RFID transponders over time

In vivo, read frequency of the RFID transponders remained constant throughout the 4 weeks of continuous testing (Fig 7B). This was reassuring, as they are being activated ~2.5 million times over a 28-day period.

Post-mortem histological evaluation of RFID implantation sites

The skin and surrounding tissues around the implanted microchip showed no gross abnormalities. Subcutaneous empty pouches were indicative of the former localisation of the RFID microchip. No or slightest cellular reactions (minimal focal chronic reactive granulation tissue) were observed microscopically in H&E stained sections collected from the implantation area of all rats. Fig 8 is a typical illustration of this finding. Histopathologically, no clear preferred implant site could be identified.

Fig 8
Evidence that the RFID transponders do not elicit an inflammatory tissue reaction.

Validation of the automated ambulatory activity measure

This was addressed in two ways, for each of the 4 RFID implantation sites (n = 5–6 per site): by comparison to manual tracking via a birds-eye view camera over a 60-minute period, and by comparison to side-view motion detection over 7–28 days. A correlation plot and a Bland-Altman plot of the automated readout against the birds-eye manual tracking found the quality of the correlation depended on the RFID implantation site (S6S9 Figs). The correlation was highest for the ventral midline site (ICC = 0.83); visual inspection of the correlation plot highlighted a drop in the automated measure below the line of equivalence at higher levels of activity (Fig 9). As evident in the time course plot (Fig 9A), the highest levels of activity occurred during the initial part of the 60-minute assessment (immediately following the disturbance to the animals caused by placing the cage in the rack). It is likely there was more thigmotaxis accompanied by rearing during the first few minutes which together with faster locomotion will tend to impact on tracking accuracy.

Fig 9
Validation of baseplate-derived ambulatory activity by comparison to manual tracking.

Similarly, concordance between the baseplate-derived ambulatory activity and the side-view, whole-cage pixel movement was also dependent on the RFID implantation site (S10 Fig). The correlation was highest for the ventral midline and flank: vertical sites. Fig 10 illustrates this concordance for the ventral midline RFID implantation site (post-‘shielding upgrade’) over 7 days). The concordance is good despite the fact that the latter measurement also includes non-ambulatory movements.

Fig 10
Concordance between baseplate-derived ambulatory activity and overall movement within the cage by video analysis.

Exploring sources of variation in the raw data

Regression analysis was used to assess what factors were adding variation to the ambulatory activity measure (a function based on transitions detected) when the data was summed in 15 minute bins in vivo using the data from the ventral midline implantation site following the ‘shielding upgrade’ (S4 Fig). This analysis found we could detect the 15 transition decrease in activity events detected in the light phase compared to the dark phase and that day 1 and week 1 were typically 5 transitions higher than later time periods. The baseplates varied by less than 5 transitions with respect to each other within a 15 minute bin. This evaluation indicates that the technology is capable of detecting changes of this magnitude (> ~5 transitions per 15 minutes).

Regression analysis was used to assess what factors were adding variation to the temperature data obtained when data averaged across 15 minute bins in vivo using the ventral midline implantation site data following the ‘shielding upgrade’ (S5 Fig). The analysis found: temperatures were on average 0.25°C lower during the light phase, Day 1 has a higher temperature by around 0.05°C compared to other days, week 1 by around 0.1°C higher compared to other weeks and the ‘red’ and ‘yellow’ baseplates gave 0.05°C and 0.08°C lower readings respectively. With such a large dataset, the regression has high sensitivity to detect small changes so whilst statistically significant these are below the level of biological interest. This evaluation indicates that the technology is capable of detecting changes of this magnitude (> ~0.1°C).

Selection of optimal implantation site for the RFID transponder

Four implantation sites have been evaluated and Fig 11 provides an overall comparison against the criteria set out in Table 1. The conventional and obvious location for an RFID transponder in the rat would normally be the interscapular region, as there is an abundance of loose skin. However, this location performed least well out of the four implantation sites evaluated, presumably because of the greater average height above the baseplate in this location compared to the other sites. Of four potential implantation sites, the ventral midline position was slightly superior (to flank: vertical) for recording ambulatory activity. Although a vertical orientation of the RFID transponder is optimal when directly above an antenna, where the magnetic field direction is also near-vertical (Figs (Figs22 and and4),4), in between adjacent antennae a horizontal orientation is more favourable, so there is no clear advantage to either site. However, the ventral midline location is closer to the baseplate, where the magnetic field is stronger. Moreover, the flank position is offset from the midline of the animal, so that the baseplate’s position estimate is also offset from its center of mass. The only slight drawback to the ventral midline location was that it required more surgical care in the implantation (refer to Methods).

Fig 11
Comparison of the 4 subcutaneous RFID transponder implantation sites/ orientations.

Validating the automated vertical activity measure

Vertical activity was detected using automated detection of motion above a pre-set height in the cage image (Fig 5), analogous to a conventional photocell beam-break system [15;17;18;19;20]. Evaluation of two 1-hour video samples from the light and dark phases, respectively, revealed that the automated detection of vertical activity correlated well with the manual annotation of vertical activity (Fig 12). This high concordance was evidenced by an inter-class correlation of >0.91 and an overall mean frame-by-frame accuracy of 88.6 (for manual annotation of rearing) and 94.7% (for manual annotation of vertical activity). The discrepancies between automated detection and manually annotated rearing are due to behaviours that cross the vertical threshold without animals rearing (e.g., climbing on the play tunnel, tail flicking and play fighting), and the partly obscured view of the far wall of the cage due to the food hopper. Whilst we use the term ‘vertical activity’ as it is closest in definition to the actual measurements obtained, it is a very good proxy for ‘rearing’. Automated detection was also able to assign individual (rat) identity to the vertical activity events occurring within the social group, from the video footage (S11 Fig).

Fig 12
Validation of vertical activity and rearing measurements by temporal correlation.

Initial observations using the new technology

Light-dark phase differences

Using side-view motion detection on 12 separate cages, each containing 3 rats, tested in the same facility spanning an 8-month period, the 12-hour activity for light phase and for dark phase were averaged separately over a 7-day period. The range of activity levels observed were: light phase, 5.8 to 16.2 x108 pixel movements; dark phase, 8.9 to 19.3 x108 pixel movements (Fig 13), indicating that some groups of 3 co-housed rats were more active during the light phase than others were during their dark phase. However, the light and dark phase activities were inter-related, whereby the ratio of dark phase: light phase activity levels remained within a reasonably narrow range. The mean ratio of dark phase: light phase activity was 1.4-fold (range 1.2 to 1.7-fold). A similar ratio was found for baseplate-derived ambulatory activity data (ventral midline site) derived from 2 cages of 3 rats (i.e., n = 6), the mean ratio of dark phase: light phase activity was 1.8-fold. Fig 1 incorporates a 7-day plot of ambulatory activity, as well as temperature, illustrating a clear circadian rhythm for both ambulatory activity (see above) and subcutaneous temperature (approximately 0.5°C higher during the dark phase).

Fig 13
Relationship between light and dark phase overall activity for 12 cages of 3 rats.

Previous publications on 24 h activity in male Wistar rats have used single-housed animals monitored by PIR sensors [21] or infrared photocell arrays [56;57], and reported that dark phase activity was approximately 4 times higher than that during the light phase. It is possible that single-housing alters behavior, augmenting the contrast in activity between dark and light phases. This would require further evaluation.

Effects of routine perturbations

In studies on two separate groups of 6 rats (each comprising two cages of 3 rats), various perturbations were seen to impact on ambulatory activity and temperature (Fig 14). Following a routine (weekly) cage change, ambulatory activity recorded by the baseplates was increased for approximately 30–60 min. We also observed a brief increase in ambulatory activity coincident with an oral dosing procedure (Fig 14A). When rats were switched from group-housing to single-housing there was an immediate decrease in subcutaneous temperature which remained approximately 0.5°C lower throughout the 16 h period of single-housing, which spanned light and dark phases (Fig 14B).

Fig 14
Illustration of the effects of routine perturbations on ambulatory activity and temperature.

Discussion

Optimal location for RFID transponder

Using both ex vivo and in vivo evaluations we have determined the optimum position–ventral abdominal midline–for an RFID transponder to allow continuous measurement of ambulatory activity and temperature in individual rats in group-housed situations. Validation of the automated readings for this implantation site found good concordance between baseplate-derived ambulatory activity and manual (birds-eye) tracking of ambulatory activity. It should be borne in mind that the automated tracking is truncated compared to the actual track of the animal, which will limit the degree of the concordance. This could also partly explain the drop-off in accuracy of tracking at higher levels of activity; the highest activity bin was the first 15 minutes following the cage change. It is conceivable that a combination of thigmotaxis (commonly seen when rats first enter and explore a ‘novel’ arena; [22]), the initial high speed of locomotion (which may be more difficult to track) and rearing (which elevates the height of the RFID transponder above the baseplate reader, and which is usually higher during initial exposure of rats to a new arena; [58]), will tend to impact on tracking accuracy.

Automated detection of vertical activity (defined by pixel movement above the horizontal cut-off line) was highly accurate when verified by manual assessment. However, vertical activity is not entirely equivalent to rearing, as evident by the presence of outliers when automated vertical activity was compared to manually verified rearing, but it does provide a strong correlate. Its current capability is equivalent to the conventional method of measuring ambulatory and vertical activity using two stacked arrays of photocell beams [15;17;18;19;20], but has the advantages of enabling group-housing in the home cage, 24/7 monitoring, with continuous measurement of subcutaneous temperature, and provision of 24/7 HD video for manual behavioral analysis.

Sources of data variability

There was a small degree of variability in RFID read rate, recorded ambulatory activity and recorded temperature between different baseplates, but these were relatively insignificant in terms of biological significance, and are considered to be small in magnitude compared to the size of effects we would generally be interested in measuring. The system also revealed that baseline ambulatory activity differed between different cages of rats (quantified both via baseplate and video analytics). We have demonstrated the potential ability of the technology to detect effects on ambulatory activity and/or subcutaneous temperature caused by routine procedures commonly performed in rodent housing facilities (oral gavage dosing procedure; cage change; single-housing).

Potential applications of the technology

As the effects of drugs can increase, diminish or remain the same on repeated dosing [27], various authors have encouraged the inclusion of behavioral endpoints in repeat-dose toxicology studies over the last three decades [27;59;60;61;62], and technology to facilitate this is long overdue. This new technology (ActualHCA™) has the potential to transform the way we do repeat-dose toxicity studies in rodents, and one likely deployment is in early repeat-dose toxicology studies in rats, where it would provide valuable additional data on adverse effects of new molecular entities on activity, behavior and temperature. Furthermore, in addition to toxicology and safety pharmacology, the technology has potential applications across the entire spectrum of behavioral neuroscience and drug discovery: behavioral phenotyping of transgenic animals [63;64]; CNS drug discovery [64;65], neurological and other disease models [31]; circadian biology [66]; drug dependence and withdrawal syndromes [67]; evaluation of environmental enrichment preference in rodent cages.

Work is ongoing to validate the current functionality pharmacologically, and to extend the behavioral recognition capability to both common (e.g., eating, drinking, grooming) and uncommon behaviors (e.g., convulsions), as has been achieved to varying extents in mice [68].

3Rs benefits

The ‘Rodent Big Brother’ project has delivered improvements to the quality and quantity of scientific data acquired from short- and long-term studies in rats, by combining the latest technological advances with 3Rs drivers [51;52;53;69]. Table 2 compares the functionality of Actual HCA™ with conventional technologies used in rats. The benefits include being able to greatly increase the information content and dimension of existing protocols by incorporating automated, continuous measurement of activity and temperature, and continuous capture of behavioral video, without disturbing the animals. The technology also has significant implications for animal welfare. The number of additional, standalone studies can be reduced, along with the number of animals used. Single housing of this social species can be avoided, minimising the potential anxiety caused. The ability to continuously monitor body temperature noninvasively also has significant advantages, removing the need for restraint or surgery and minimizing temperature fluctuations caused by stress.

Table 2
Anticipated advantages and drawbacks of Actual Home Cage Analyzer™ over existing technologies in rats

Importantly, the technology allows subtle changes in behavior, activity and temperature to be detected sooner and combined to improve the assessment of beneficial or adverse effects of compounds. More widely it could deliver global improvements in drug discovery and development through improved data quality, reduced and refined animal use, and increased efficiency.

Supporting information

S1 Fig

Understanding sources of variation in RFID transponder read rate from an ex vivo experiment.

(DOCX)

S2 Fig

Relationship between RFID transponder read rate and height above baseplate from the ex vivo experiment.

(DOCX)

S3 Fig

Understanding sources of variation in RFID transponder read rate in vivo.

(DOCX)

S4 Fig

Understanding sources of variation for ambulatory activity.

(DOCX)

S5 Fig

Understanding sources of variation for subcutaneous temperature.

(DOCX)

S6 Fig

Ambulatory activity validation for ventral midline RFID implantation site (pre-‘shielding upgrade’).

(DOCX)

S7 Fig

Ambulatory activity validation for flank: vertical RFID implantation site (pre-‘shielding upgrade’).

(DOCX)

S8 Fig

Ambulatory activity validation for flank: horizontal RFID implantation site (pre-‘shielding upgrade’).

(DOCX)

S9 Fig

Ambulatory activity validation for interscapular RFID implantation site (pre-‘shielding upgrade’).

(DOCX)

S10 Fig

Correlation plots of ambulatory movement of the rats derived from the baseplate RFID reader versus side-view pixel movement detection, for each of the 4 implantation sites.

(DOCX)

S11 Fig

Automated detection of individual rearing activity within a cage of 3 rats S1 Video Clip.

Video clip illustrating view of cage containing 3 rats from side-view HD camera. The cage is illuminated by infrared lighting strips (visible at the top of the image) and contains a plastic play tunnel as part of the environmental enrichment; this is red in color but appears transparent in infrared lighting.

(DOCX)

S1 Video Clip

60 s video clip of 3 male Han Wistar rats co-housed in a Tecniplast SealSafe Blue-line individually ventilated cage (IVC).

The infrared lighting strips are visible at the top of the image; the plastic play tunnel appears transparent in infrared light. The side-view, whole-cage video enables manual evaluation/quantification of behavior at any time of day or night. From the video, software quantifies whole-cage pixel movement as a measure of overall activity within the cage, and also detects vertical activity of individual rats.

(AVI)

Acknowledgments

This work was funded in part by the NC3Rs CRACK IT Scheme. Emma O’Callaghan, Ben Allison and David Craig of Actual Analytics contributed to the successful development of this technology. We would like to thank Heiden Bliss and Anthony Carson of BioMark for their assistance in developing the system described here. We thank Chris Pollard and Mark Pilling for helpful comments on the manuscript. We would also like to acknowledge the contribution of various colleagues in Laboratory Animal Sciences (LAS) at AstraZeneca. This publication is dedicated to the memory of Liz Beard (LAS, AstraZeneca), who played a small but crucial role during the early phase of this project.

Funding Statement

This work was supported by National Centre for the 3Rs (NC3Rs), London, UK, CRACK IT scheme (2011; 'Rodent Big Brother'), Award number: 15413‐106223, https://www.nc3rs.org.uk, CV and KC from NC3Rs are co-authors. The funders had no role in study design, data collection and analysis, or decision to publish. The NC3Rs authors (CV and KC) did contribute to the preparation of the manuscript particularly in reflecting the 3Rs/animal welfare impact of this technology. The specific roles of these authors are articulated in the author contributions’ section.

Data Availability

Data Availability

All data generated or analysed during this study are included in this published article or are deposited in Zenodo, 10.5281/zenodo.804041.

References

1. Baldrick P. Safety evaluation to support first-in-man investigations II: Toxicology studies. Reg Toxicol Pharmacol. 2008; 51: 237–243. [PubMed]
2. Redfern WS., Wakefield ID. Safety Pharmacology In Toxicological Testing Handbook: Principles, Applications and Data Interpretation, 2nd edn, Keller K & Jacobson-Kram D (eds.), Taylor & Francis Ltd; 2006. pp. 33–78.
3. Redfern WS., Valentin J-P. Trends in safety pharmacology: Posters presented at the annual meetings of the Safety Pharmacology Society 2001–2010. J Pharmacol. Toxicol. Methods. 2011; 64: 102–110. doi: 10.1016/j.vascn.2011.05.003 [PubMed]
4. Anon. Guidance for Industry: ICH S7A Safety Pharmacology Studies for Human Pharmaceuticals, US Food and Drug Administration. (2001)
5. Moser VC. Screening approaches to neurotoxicity: A functional observational battery. J Am Coll Toxicol. 1989; 8: 85–93.
6. Mattson JL, Spencer PJ, Albee RR. A performance standard for clinical and functional observational battery examination of rats. J Am Coll Toxicol. 1996; 15, 239–250.
7. Redfern WS Strang I, Storey S, Heys C, Barnard C, Lawton K et al. Spectrum of effects detected in the rat Functional Observational Battery following oral administration of non-CNS targeted compounds. J Pharmacol Toxicol Methods. 2005; 52: 77–82. doi: 10.1016/j.vascn.2005.04.005 [PubMed]
8. Moscardo E, Maurin A, Dorigatti R, Champeroux P, Richard S. An optimised methodology for the neurobehavioural assessment in rodents. J. Pharmacol. Toxicol. Methods. 2007; 56: 239–255. doi: 10.1016/j.vascn.2007.03.007 [PubMed]
9. Moser VC. Functional assays for neurotoxicity testing. Toxicologic Pathology. 2011; 39: 36–45. doi: 10.1177/0192623310385255 [PubMed]
10. Ewart L, Milne A, Adkins D, Benjamin A, Bialecki R, Chen Y et al. A multi-site comparison of in vivo safety pharmacology studies conducted to support ICH S7A & B regulatory submissions. J Pharmacol. Toxicol. Methods. 2013; 68: 30–43. doi: 10.1016/j.vascn.2013.04.008 [PubMed]
11. Creese I, Iverson SD. Blockage of amphetamine-induced motor stimulation and stereotypy in the adult rat following neonatal treatment with 6-hydroxydopamine. Brain Res. 1973; 55: 369–382. [PubMed]
12. Ushijima I, Yamada K, Inoue T, Tokunaya T, Furukawa T, Noda Y. Muscarinic and nicotinic effects on yawning and tongue protruding in the rat. Pharmacol Biochem Behav. 1984; 21: 297–300. [PubMed]
13. Tricklebank MD. The behavioural response to 5-HT receptor agonists and subtypes of the central 5-HT receptor. Trends Pharmacol Sci. 1985; 6: 403–407.
14. Racine RJ. Modification of seizure activity by electrical stimulation: II. Motor seizure. Electroencephalography Clin Neurophysiol. 1972; 32: 281–294. [PubMed]
15. Sanberg PR, Hagenmeyer SH, Henault MA. Automated measurement of multivariate locomotor behavior in rodents. Neurobehav Toxicol Teratol. 1985; 7: 87–94. [PubMed]
16. Crofton KM, Howard JL, Moser VC, Gill MW, Reiter LW, Tilson HA, MacPhail RC. Interlaboratory comparison of motor activity experiments: Implications for neurotoxicological assessments. Neurotoxicol Teratol. 1991; 13: 599–609. [PubMed]
17. Kelley AE. Locomotor activity and exploration In: van Haaren F, ed. Methods in Behavioral Pharmacology. Amsterdam: Elsevier; 1993. pp. 499–518.
18. Patterson JP, Markgraf CG, Cirino M, Bass A. Validation of a motor activity system by a robotically controlled vehicle and using standard reference compounds. J Pharmacol Toxicol Methods. 2005; 52:159–167. doi: 10.1016/j.vascn.2005.05.003 [PubMed]
19. Himmel HM. Safety pharmacology assessment of central nervous system function in juvenile and adult rats: Effects of pharmacological reference compounds. J Pharmacol Toxicol Methods. 2008; 58: 129–146. doi: 10.1016/j.vascn.2008.06.001 [PubMed]
20. Lynch JJ, Castagné V, Moser PC, Mittelstadt SW. Comparison of methods for the assessment of locomotor activity in rodent safety pharmacology studies. J Pharmacol Toxicol Methods. 2011; 64: 74–80. doi: 10.1016/j.vascn.2011.03.003 [PubMed]
21. McDermott C, Kelly JP. Comparison of the behavioural pharmacology of the Lister-Hooded with 2 commonly utilised albino rat strains. Prog Neuro-Psychopharmacol Biol Psychiat. 2008; 32: 1816–1823. [PubMed]
22. Ouhaz Z, Saadia Ba-M’hamed S, Mitchell AS, Elidrissi A, Bennis M. Behavioral and cognitive changes after early postnatal lesions of the rat mediodorsal thalamus. Behav Brain Res. 2015; 292: 219–232. doi: 10.1016/j.bbr.2015.06.017 [PMC free article] [PubMed]
23. McGowan MK, Andrews KM, Grossman SP. Role of intrahypothalamic insulin in circadian patterns of food intake, activity, and body temperature. Behav Neurosci. 1992; 106, 380–385. [PubMed]
24. Livesey PJ, Leppard K. A TV monitored system for recording open-field activity in the rat. Behav Res Methods & Instrumentation. 1981; 13: 331–333.
25. Vorhees CV, Acuff-Smith KD, Minck DR, Butcher RE. A method for measuring locomotor behavior in rodents: contrast-sensitive computer-controlled video tracking activity assessment in rats. Neurotoxicol Teratol. 1992; 14: 43–49. [PubMed]
26. Dunne F, O'Halloran A, Kelly JP. Development of a home cage locomotor tracking system capable of detecting the stimulant and sedative properties of drugs in rats. Prog Neuro-Psychopharmacol Biol Psychiat. 2007; 31: 1456–1463. [PubMed]
27. Redfern WS, Ewart LC, Lainée P, Pinches M, Robinson S, Valentin J-P. Functional Assessments in Repeat-dose Toxicity Studies: The Art of the Possible. Toxicol Res. 2013; 2: 209–234.
28. Van de Weerd HA, Bulthuis RJA, Bergman AF, Schlingmann F, Tolboom J, Van Loo PLP et al. Validation of a new system for the automatic registration of behaviour in mice and rats. Behav Processes. 2001; 53: 11–20. [PubMed]
29. Quinn LP, Stean TO, Trail B, Duxon MS, Stratton SC, Billinton A, Upton N. LABORASTM: Initial pharmacological validation of a system allowing continuous monitoring of laboratory rodent behaviour. J Neurosci Methods. 2003; 130: 83–92. [PubMed]
30. Castagné V, Wolinsky T, Quinn L, Virley D. Differential behavioral profiling of stimulant substances in the rat using the LABORAS™ system. Pharmacol Biochem Behav. 2012; 101: 553–563. doi: 10.1016/j.pbb.2012.03.001 [PubMed]
31. Urbach YK, Raber KA, Canneva F, Planka A-C, Andreasson T, Ponten H, Kullingsjö J, Nguyen HP, Riess O, von Hörsten S. Automated phenotyping and advanced data mining exemplified in rats transgenic for Huntington’s disease. J Neurosci Methods. 2014; 234: 38–53. doi: 10.1016/j.jneumeth.2014.06.017 [PubMed]
32. Patterson-Kane EG, Hunt M, Harper D. Rats demand social contact. Animal Welfare. 2002; 11: 327–332.
33. Balcombe JP. Laboratory environments and rodents’ behavioural needs: a review. Lab Animals. 2006; 40, 217–235. [PubMed]
34. Onishchenko N, Fischer C, Wan Ibrahim WN, Negri S, Spulber S, Cottica D, Ceccatelli S. Prenatal exposure to PFOS or PFOA alters motor function in mice in a sex-related manner. Neurotox Res. 2011; 19: 452–461. doi: 10.1007/s12640-010-9200-4 [PubMed]
35. Howerton CL, Garner JP, Mench JA. A system utilizing radio frequency identification (RFID) technology to monitor individual rodent behavior in complex social settings. J Neuroscience Methods. 2012; 209: 74–78. [PubMed]
36. Shemesh Y, Sztainberg Y, Forkosh O, Shlapobersky T, Chen A, Schneidman E. High-order social interactions in groups of mice. eLife 2013; 2: e00759 doi: 10.7554/eLife.00759 [PMC free article] [PubMed]
37. Weissbrod A, Shapiro A, Vasserman G, Edry L, Dayan M, Yitzhaky A, Hertzberg L, Feinerman O, Kimchi T. Automated long-term tracking and social behavioural phenotyping of animal colonies within a semi-natural environment. Nature Communications 2013. 4:2018 doi: 10.1038/ncomms3018 [PubMed]
38. Hong W, Kennedy A, Burgos-Artizzu XP, Zelikowsky M, Navonne SG, Perona P, Anderson DJ. Automated measurement of mouse social behaviors using depth sensing, video tracking, and machine learning. Proc Nat Acad Sci USA 2015; 112: E5351–E5360. doi: 10.1073/pnas.1515982112 [PubMed]
39. Bains RS, Cater HL, Sillito RL, Chartsias A, Sneddon D, Concas D, Keskivali-Bond P, Lukins TC, Wells S, Acevedo Arozena A, Nolan PM, Armstrong JD. Analysis of individual mouse activity in group housed animals of different inbred strains using a novel automated home cage analysis system. Frontiers Behav Neurosci. 2016; 10: Article 106. [PMC free article] [PubMed]
40. Gordon CJ. Toxic-induced hypothermia and hypometabolism: Do they increase uncertainty in the extrapolation of toxicological data from experimental animals to humans? Neurosci Biobehav Rev. 1991; 15: 95–98. [PubMed]
41. Gordon CJ, Spencer PJ, Hotchkiss J, Miller DB, Hinderliter PM, Pauluhn J. et al. Thermoregulation and its influence on toxicity assessment. Toxicology. 2008; 244: 87–97. doi: 10.1016/j.tox.2007.10.030 [PubMed]
42. Gordon CJ. Response of the thermoregulatory system to toxic insults. Frontiers in Bioscience 2010; 2: 293–311. [PubMed]
43. Anderson KB, Poloyac SM, Kochanek PM, Empey PE. Effect of hypothermia and targeted temperature management on drug disposition and response following cardiac arrest: A comprehensive review of preclinical and clinical investigations. Therapeutic Hypothermia & Temperature Management. 2016; 6: 169–179. [PMC free article] [PubMed]
44. Buchanan TA, Cane P, Eng CC, Sipos GF, Lee C. Hypothermia is critical for survival during prolonged insulin-induced hypoglycemia in rats. Metabolism. 1991; 40: 330–334. [PubMed]
45. Steiner AA, Branco LG. Hypoxia-induced anapyrexia: implications and putative mediators. Ann Rev Physiol. 2004; 64: 263–288. [PubMed]
46. Rusyniak DE, Sprague JE. Toxin-induced hyperthermic syndromes. Medical Clinics of North America. 2005; 89: 1277–1296. doi: 10.1016/j.mcna.2005.06.002 [PubMed]
47. Poole S, Stephenson J. Core temperature: some shortcomings of rectal temperature measurements. Physiol Behav. 1977; 18: 203–205. [PubMed]
48. Kort WJ, Hekking-Weijma JM, TenKate MT, Sorm V, VanStrik R. A microchip implant system as a method to determine body temperature of terminally ill rats and mice. Lab Animals. 1998; 32: 260–269. [PubMed]
49. Legos J., Mangoni A., Read S., Campbell C., Irving E., Roberts J., Barone F. and Parsons A. Programmable microchip monitoring of post-stroke pyrexia: effects of aspirin and paracetamol on temperature and infarct size in the rat. J Neurosci Methods. 2002; 113: 159–166. [PubMed]
50. Prior H, Motyer V, Cobey D, Marks L, Ewart L, Redfern WS et al. Inclusion of microchip transponder body temperature measurements in safety pharmacology and toxicology studies. J Pharmacol Toxicol Methods. 2012; 66: 192.
51. Russel WMS, Burch RL. (1959) The Principles of Humane Experimental Technique. Methuen, London, reprinted by Universities Federation for Animal Welfare, Potters Bar, UK, 1992.
54. Bland JM, Altman DG. Measuring agreement in method comparison studies. Statistical Methods in Medical Research. 1999; 8: 135–160. doi: 10.1177/096228029900800204 [PubMed]
55. Zaki R, Bulgiba A, Ismail R, Ismail NA. Statistical methods used to test for agreement of medical instruments measuring continuous variables in method comparison studies: A systematic review. PLoS ONE. 2012; 7: e37908 doi: 10.1371/journal.pone.0037908 [PMC free article] [PubMed]
56. Terrón MP, Delgado-Adámez J, Pariente JA, Barriga C, Paredes SD, Rodrígues AB. Melatonin reduces body weight gain and increases nocturnal activity in male Wistar rats. Physiol Behav. 2013; 118: 8–13. doi: 10.1016/j.physbeh.2013.04.006 [PubMed]
57. Clemens LE, Jansson EKH, Portal E, Riess O, Nguyen HP. A behavioral comparison of the common laboratory rat strains Lister Hooded, Lewis, Fischer 344 and Wistar in an automated homecage system. Genes, Brain & Behav. 2014; 13: 305–321. [PubMed]
58. Wilson CA, Vazdarjanova A, Terry AV. Exposure to variable prenatal stress in rats: Effects on anxiety-related behaviors, innate and contextual fear, and fear extinction. Behav Brain Res. 2013; 238: 279–288. doi: 10.1016/j.bbr.2012.10.003 [PMC free article] [PubMed]
59. Zbinden G. Neglect of function and obsession with structure in toxicity testing. In Proceedings of 9th International Congress of Pharmacology, Macmillan Press, New York; 1984, vol. 1, pp. 43–49.
60. Evans HL. Behaviors in the home cage reveal toxicity: recent findings and proposals for the future. J Am Coll Toxicol. 1989; 8: 35–52.
61. Luft J, Bode G. Integration of safety pharmacology endpoints into toxicology studies. Fund Clin Pharmacol. 2002; 16: 91–103. [PubMed]
62. Redfern WS. (2015) Inclusion of safety pharmacology endpoints in repeat-dose toxicity studies. In MK Pugsley, M Curtis (Eds.), Principles of Safety Pharmacology, Series: Handbook of Experimental Pharmacology, Vol. 229, Springer.com; 2015. pp. 353–381. [PubMed]
63. Gerlai R. Phenomics: fiction or the future? Trends Neurosci. 2002; 25: 506–509. [PubMed]
64. Spruijt BM, DeVisser L. Advanced behavioral screening: automated home cage ethology. Drug Discovery Today. 2006; 3: 231–237. doi: 10.1016/j.ddtec.2006.06.010 [PubMed]
65. Brunner D, Nestler E, Leahy E. In need of high-throughput behavioral systems. Drug Discovery Today. 2002; 7 (Suppl.): S107–S112. [PubMed]
66. Silver R, Kriegsfeld LJ. Circadian rhythms have broad implications for understanding brain and behavior. Eur J Neurosci. 2014; 39: 1866–1880. doi: 10.1111/ejn.12593 [PMC free article] [PubMed]
67. Celikyurt IK, Kayir H, Ulak G, Erden FB, Ulusoy GK, Uzbay TI. (2011) Effects of risperidone, quetiapine and ziprasidone on ethanol withdrawal syndrome in rats. Prog Neuro-Psychopharmacol Biol Psychiat. 2001; 35: 528–536. [PubMed]
68. Anderson DJ, Perona P. Toward a science of computational ethology. Neuron 2014; 84:18–31. doi: 10.1016/j.neuron.2014.09.005 [PubMed]
69. EMA (2014). In: Guideline on Regulatory Acceptance of 3R (Replacement, Reduction, Refinement) Testing Approaches. European Medicines Agency; http://www.ema.europa.eu/

Articles from PLoS ONE are provided here courtesy of Public Library of Science