Early Detection of Apathetic Phenotypes in Huntington’s Disease Knock-in Mice Using Open Source Tools

Apathy is one of the most prevalent and progressive psychiatric symptoms in Huntington’s disease (HD) patients. However, preclinical work in HD mouse models tends to focus on molecular and motor, rather than affective, phenotypes. Measuring behavior in mice often produces noisy data and requires large cohorts to detect phenotypic rescue with appropriate power. The operant equipment necessary for measuring affective phenotypes is typically expensive, proprietary to commercial entities, and bulky which can render adequately sized mouse cohorts as cost-prohibitive. Thus, we describe here a home-built, open-source alternative to commercial hardware that is reliable, scalable, and reproducible. Using off-the-shelf hardware, we adapted and built several of the rodent operant buckets (ROBucket) to test HttQ111/+ mice for attention deficits in fixed ratio (FR) and progressive ratio (PR) tasks. We find that, despite normal performance in reward attainment in the FR task, HttQ111/+ mice exhibit reduced PR performance at 9–11 months of age, suggesting motivational deficits. We replicated this in two independent cohorts, demonstrating the reliability and utility of both the apathetic phenotype, and these ROBuckets, for preclinical HD studies.

Amongst HD's psychiatric manifestations, apathy has an extremely high point prevalence, coupled with a uniquely consistent relationship between severity and HD progression 7,17 . A recent clinical evaluation study of presymptomatic HD mutation carriers found striking increases in the incidence of apathy, even in subjects predicted to be more than 10 years from clinical onset 18 . Apathy, as a psychiatric symptom distinct from depression, has been operationalized to contain aspects of diminished motivation, reduced goal-directed behavior, lack of interest in new experiences, and diminished emotional responsivity 19 . In HD patients, apathy is correlated with functional capacity and cognition, but not depression, suggesting apathy is a distinctive component of the affective landscape of HD 20 .
Motivated by the importance of apathy to the lived experience of HD mutation carriers, we are interested in bringing analysis of apathy into HD preclinical studies. Traditional rodent experiments to test motivated behavior include the progressive-ratio (PR) operant task 21 , in which subjects are required to perform increasingly large numbers of nose-pokes or lever-presses to receive a reward. Commercial operant chambers used to assay PR responses in rodents can cost thousands of dollars each, limiting the number of animals (and thereby statistical power) of preclinical studies of apathy. To address this problem, we have modified a recently described open-source operant chamber -the "ROBucket" 22 -based on the Arduino computing platform, with a total built cost of approximately $150/chamber. To validate the modified apparatus, we studied motivation in 9-11 month old B6.Htt Q111/+ mice, a knock-in mouse model of the HD mutation 23 . Using commercially available tools we, and others, have previously observed motivational phenotypes in this model that precede motor or cognitive changes [24][25][26][27] . Using the open-source ROBucket, we confirm specific deficits in progressive, but not fixed, ratio tasks in B6.Htt Q111/+ mice at this time point, consistent with relatively intact learning but impaired motivation.

Results
Modification of existing operant chamber design. We first precisely recreated a recently reported open source operant chamber ("ROBucket") 22 . On conducting pilot experiments, we found mice tended to interact with the housing and reward tubing fed through the bucket in the original design. To overcome this distraction, we redesigned the 3D-housing apparatus to be outside the mouse chamber ( Fig. 1) -updated plans are available online at https://zenodo.org/record/101136028. Studies performed with these modifications revealed that mice quickly learned to interact with the active well to receive 10 µl of 20% sucrose, and that the modified ROBuckets (mROBucket) accurately counted nose pokes, compared to direct observations. Normal fixed ratio performance in Htt Q111/+ mice. 9-month-old male B6.Htt +/+ (n = 8) and B6.Htt Q111/+ (n = 12) mice (hereafter Htt +/+ and Htt Q111/+ ) were single-housed and food restricted over 2 weeks (target weight loss of 2%/day, final body weight ~85% free-feeding weight) before operant testing. We observed no impact of genotype on baseline body weight or the rate at which Htt +/+ and Htt Q111/+ mice lost weight during food restriction (Supplemental Fig. 1). After body weight stabilized, mice were placed into mROBucket chambers for 1 hour sessions each day on a fixed ratio 1 (FR1) reinforcement schedule -i.e. one nose poke in the active well resulted in sucrose delivery in the reward well. There was a 1 second timeout after each active well response. Two criteria were required for progression to the next phase: a 3:1 preference for the active well versus the inactive well and 20 or more reinforcements for 3 consecutive days. All mice, with the exception of a single Htt +/+ mouse (Fig. 2, grey panel), quickly learned the task (Fig. 3, average time to FR1 criteria 7.5 ± 2.4 days).
We observed no effect of genotype on days to meet criteria, the active/inactive nose poke ratio, or total nose pokes per session on the FR1 task (Fig. 3). This suggests 9-month-old male Htt Q111/+ mice are able to normally acquire this simple discrimination task.
After meeting both criteria for FR1, we next trained the mice on a fixed ratio 5 (FR5) task for 3 days to familiarize them with tasks requiring multiple nose pokes to achieve reward. There was a 1 second timeout after each active well response. Consistent with the FR1 task, 9-month-old male Htt Q111/+ mice are able to normally acquire this simple discrimination task, with no differences observed between the ratio of active/inactive well responses or the total number of nose pokes per session (Fig. 4). This suggests fatigue and motor dysfunction do not prevent Htt Q111/+ mice from making large numbers of accurate nose pokes in an hour-long session (average of 273.1 ± 11.7 nose pokes/session).
Reduced progressive ratio performance in Htt Q111/+ mice. Regardless of performance, mice were advanced to a progressive ratio (PR) task after 3 days of FR5 training. The PR task requires the animals to respond with an exponentially increasing number of sequential nose pokes in the active well to receive the reward (1, 2, 4, 6, 9, 12, etc., according to the formula R = ||5e (N*0.2) ||−5) 29 . The final number of reinforcements achieved is referred to as the "breakpoint" 21 . We established the criterion for the PR task based on breakpoint stabilization, or less than 10% variation in the breakpoint in trials over 3 consecutive days. Both Htt +/+ and Htt Q111/+ mice learned the PR task, with breakpoints stabilizing between 4-17 days.

Discussion
Apathy is a core feature of affective dysfunction in HD 17 , but analysis of apathy and other affective disturbances is rarely included in preclinical studies of HD 33 . Here, we demonstrate that an inexpensive open source apparatus can robustly and reproducibly detect motivational deficits in 9-11 month old Htt Q111/+ mice. These motivational deficits precede overt motor, cognitive, or neurodegenerative changes in this model 27 , suggesting they may occur amongst the earliest changes associated with mutant huntingtin expression in vivo. A recent study from  the Yhnell et al. tested Htt Q111/+ mice in commercial operant boxes at 6-, 12-, and 18-months of age 24 , finding that Htt Q111/+ mice showed no deficits in fixed ratio testing at 6-or 12-months, but had reduced breakpoint in progressive ratio at 12 months when delivering similar volumes of liquid reward to those used in our study (7.5 μL vs. 10 μL used in our study). While differences in the methodology for the training, progressive ratio task, and breakpoint determination between our study and the Yhnell et al. study limit direct comparison, our results confirm those found in the commercial apparatus: that there are no deficits in fixed ratio testing at a time point between the 6-and 12-month timepoints (9-and 10-mo for our independent cohorts, respectively), while we also confirm the significant progressive ratio deficits found in the commercial apparatus at 12-months in our slightly younger cohorts tested using open-source equipment.
The basal ganglia, and particularly the striatum, are the most strikingly impacted brain regions in Huntington's disease, showing robust volume declines many years before clinical disease onset 7,34 . While no full-length HD mouse models experience similar pronounced neurodegeneration, a number of imaging 30,35,36 and molecular analyses 37 confirm that the caudoputamen is the most strikingly impacted brain region in mice expressing mutant Htt. In humans, focal ischemic basal ganglia damage is associated with a range of motivational deficits 38 , including notable deficits in incentive motivation -the process of activating specific behavioral responses based on predicted reward 39 . In stroke patients with basal ganglia lesions, deficits in incentive motivation occur in the absence of deficits in hedonic responses, consistent with reduced progressive ratio performance in Htt Q111/+ mice (here, and 24,27 ) at a time when sucrose preference tasks suggest no alterations in hedonic drive in these mice 27 . Changes in instrumental motivation may therefore serve as a translatable readout of basal ganglia dysfunction in mice. The deficits in instrumental motivation seen here may provide a behavioral tool for quantifying this dysfunction, however we caution that mouse behavioral phenotypes likely map imperfectly onto complex symptoms -such as apathy -in human HD patients.
Statistical power in neuroscience studies, and especially behavioral studies, is generally low 16 , leading to widespread calls for improvements in conducting and reporting of preclinical work 40 . While the use of operant chambers can improve preclinical studies by automating data collection of complex behavioral tasks, these benefits can be negated by the cost of the apparatus. Traditional operant chambers from commercial suppliers cost thousands of dollars, placing a practical limit on the number of mice that can be screened in parallel in a single lab and thereby limiting the power of these studies. Recent technological developments -including the rapid development of inexpensive additive manufacturing (i.e. "3D printing") and low-cost open source computing platforms (e.g. Raspberry Pi and Arduino) -enable open source alternatives to commercial products. Further, they allow for rapid design and software modifications to assay behavior through multiple paradigms. These tools have been applied to video tracking of behavior 41 , integrated microscopy, temperature control, and optogenetics in small animal experiments 42 , as well as the rodent operant chambers used here 22 . These technologies render practical the fabrication of a large number of chambers (12 were used in the current study), so dozens of mice can be tested in parallel in a single lab, with a limited number of handlers.
These experiments confirm previous findings that motivational deficits occur before pronounced neurodegeneration in the Htt Q111/+ model of HD. They also show that these deficits are assayable using inexpensive hardware and open source software tools, which should enable their more widespread utilization in preclinical studies in HD.

Methods
Mice. B6.Htt Q111 mice, which have been previously described 43 , were originally obtained from JAX (Research Resource Identifier: IMSR\_JAX:003456) and bred and maintained at the Western Washington University vivarium. Mice were group housed until 9-10 months of age and given access to food and water ad libitum, until two weeks before testing was to begin. For genotyping, presence or absence of the mutant allele was determined by polymerase chain reaction of DNA using the primers CAG1 (5′-ATGAAGGCCTTCGAGTCCCTCAAGTCCTTC-3′) 44 and HU3 (5′-GGCGGCTGAGGAAGCTGAGGA-3′) 45 . All experiments were conducted in accordance to the NIH Guide for the Care and Use of Laboratory Animals and approved by the Western Washington University animal care and use committee (protocol 16-007).
Apparatus. 2-choice operant chambers were constructed according to the design from Devarakonda 22 with modifications. Briefly, Arduino-controlled operant boxes deliver a liquid sucrose reward in response to nose-pokes with various reinforcement schedules. The apparatus here was modified to place the nose-poke photo-beam housing on the outer, rather than inner, wall of the operant chamber (see Fig. 1b). This eliminated time spent exploring the nose-poke housing and associated tubing, and resulted in cleaner acquisition of the FR/ PR tasks. Isolation housing was designed and constructed as a grid of 3 × 4 chambers (35 × 42 × 42 cm each) with view ports in the front to visualize the response readings and vent fans in the back to produce white noise (Fig. 1d).
Behavioral Testing. Mice were single-housed and fasted over two weeks to reduce body weight to 85% of free feeding weight prior to testing and maintained at this weight for the duration of testing. For FR1 testing, mice learned to nose-poke on a fixed-ratio reinforcement schedule where a single nose-poke in the active well elicits delivery of a sucrose reward (10 μL, 20% sucrose), with a 1-second timeout after each active well press. Trial duration was 60 minutes or until the subject received 50 reinforcements, at which point mice were promptly removed. Acquisition criteria for the FR1 schedule were met when mice exhibited discrimination criteria of ≥3:1 for the active:inactive well and received ≥20 reinforcements for 3 consecutive days. Mice not meeting the FR1 acquisition criteria by 17 days were excluded from further testing (this occurred in 1 mouse from each cohort, or approximately 5% of mice). After meeting FR1 criteria, mice were moved to FR5 testing, where the reinforcement schedule requires five pokes to earn 1 reward, for three consecutive days. For PR testing, mice were tested on a progressive ratio schedule of reinforcement where the number of nose pokes required to elicit reinforcement is calculated using the equation Reinforcements = ||5e (N*0.2) ||−5, where N is equal to the number of sucrose solution reinforcements already earned plus 1. When mice earned the same number of rewards (±10%, or within 1 if less than 10 rewards are earned) for 3 consecutive days, they were considered stabilized at their "breakpoint. " For full methods and instructions, see published online methods at https://zenodo.org/record/101136028. Statistical analysis. All data were processed using R statistical software 46 . Welch's t-tests were used to correct for unequal variances, linear mixed effects ANOVAs were run using the 'nlme' package 47 , effect size was calculated with the 'compute.es' package 48 , and power was calculated using the 'pwr' package 49 . Graphics were produced using 'ggplot2' 50 and Illustrator (Adobe).