Validation and standardization of a new radial-arm water maze protocol using a murine model of mild closed head traumatic brain injury

Cognitive impairments can be a significant problem after a traumatic brain injury (TBI), which affects millions worldwide each year. There is a need for establish reproducible cognitive assays in rodents to better understand disease mechanisms and to develop therapeutic interventions towards treating TBI-induced impairments. Our goal was to validate and standardize the radial arm water maze (RAWM) test as an assay to screen for cognitive impairments caused by TBI. RAWM is a visuo-spatial learning test, originally designed for use with rats, and later adapted for mice. The present study investigates whether test procedures, such us the presence of extra-maze cues influences learning and memory performance. C57BL/6 mice were tested in an 8-arm RAWM using a four-day protocol. We demonstrated that two days of training, exposing the mice to extra-maze cues and a visible platform, influenced learning and memory performance. Mice that did not receive training performed poorer compared to mice trained. To further validate our RAWM protocol, we used scopolamine. We, also, demonstrated that a single mild closed head injury (CHI) caused deficits in this task at two weeks post-CHI. Our data supported the use of 7 trials per day and a spaced training protocol as key factor to unmask memory impairment following CHI. Here, we provide a detailed standard operating procedure for RAWM test, which can be applied to a variety of mouse models including neurodegenerative diseases and pathology, as well as when pharmacological approaches are used.

4 84 capability to test more mice in one session. The result of our study, support our hypothesis that few trials per 85 day spaced over 4 days, is sensitive at detecting cognitive deficits following a mild closed head injury (CHI). 86 Our RAWM protocol will likely be widely applicable to detect cognitive deficits in other mouse models of injury 87 or disease.   The study used 141 adult mice (72/69 ♀/♂) 3-4 months old, C57BL/6J mice (Jackson laboratory, Bar 01 Harbor, ME; stock number: 000664). The number of mice used for each experiment is reported in the figure 02 legend. Animals were group-housed (4-5 per cage) in a controlled humidity (43-47%) and temperature (22-23º 03 C) environment and 12/12-h (7 am-7 pm) light/dark cycle with free access to food and water. Behavioral 04 experiments were conducted by the same operator (TM) and performed between 7.30 a.m. and 3.30 p.m.

05
Mice were assigned randomly to groups before the start of each experiment. Each cage had at least 06 one animal from each group/ treatment in a random order and the person performing the experiments 07 scopolamine injections (KNR) and RAWM test (TM) was blinded to treatment. Also, the person performing the 08 RAWM test (TM), three days before the beginning of the experiment transferred the mice to a clean cage, 09 marked the tail for easy identification, and handled them for the following three days. For the handling, a 5 10 mouse was allowed to explore the experimenter's arm and hand for 1-2 min, then returned to its home cage.
11 Mice were never exposed to the maze before the start of the experiment. During the behavioral tests, mice 12 were transferred to the experimental room and relocated in a clean recovery cage without bedding at least 30 13 minutes before the start of the test. The recovery cage had paper towels inside and was placed half on a 14 heating pad to help the mice recovery from the swimming activity and half off the heat. Also, during the test, 15 wet paper towels were promptly replaced, so mice could have a faster recovery in a dry and warm recovery 16 cage. Male and female mice were tested in separate cohorts.

18
Radial arm water maze apparatus 19 The RAWM test was performed in a circular pool (diameter=121 cm, depth 75 cm, Fig.1 A, B) 20 (MazeEngineers, Boston, MA). A base (9 cm high), and an 8-shaped platform were added to the pool and eight 21 identical metal inserts having a V shape (approximately 65 cm high by 42 cm long) were inserted in the pool to 22 make an 8 arm RAWM (Fig.1 A, B). Arms were raised 9 cm above the level of water, to discourage mice from 23 climbing over the inserts and jumping in the dead area of the pool or on the floor. The arms were made of 24 stainless steel to avoid corrosion due to the continuous contact to the water. The water was made opaque by 25 the addition of white liquid non-toxic paint (Sax 2684 Versatemp Non-Toxic Heavy Body Tempera Paint), and 26 the water temperature was held constant at 21 ± 1º C. The escape platform was a circular (diameter 8 cm, 57 27 cm high, Fig.1 C) clear acrylic adjustable platform submerged 1 cm below the water level in one of the eight 28 arms and was defined as the "goal arm". The apparatus was isolated from the rest of the room by double black 29 black-out curtains. Four extra-maze cues (triangle, square, checkerboard pattern and cross made of white 30 corrugated plastic and black vinyl material; Fig.1 E) were hung on the inside of these curtains around the pool 31 ( Fig.1 A) at a height of 12 cm from the top of the pool to the cues. Four dimmable overhead lights were used 32 and light intensity was kept between 4 and 6 lux in each arm and center of the maze. The platform was made 33 visible by a flag (16 cm high, Fig.1 D) when needed.

34
A camera was positioned directly above the center of the pool and all experiments were recorded.
35 EthoVision XT 11.0 (Noldus Information Technology) was used for video recording and scoring behavior. Mice were tested for a total of 4 days and received 7-trial per day. To reduce fatigue, the 7 trials were 43 divided in two blocks: block 1(trials 1-3) and block 2 (trials 4-7). To encourage mice to learn the location of the 44 platform, an alternation of visible and hidden platform was used during block 1 of training days (trial 1 and 3, 45 day 1 and 2) and a hidden platform was used during block 2 ( Fig. 2 A). During testing days, the platform was 46 hidden in both block 1 and block 2 ( Fig. 2 B).

47
48 Figure 2: Step-by-Step procedure of RAWM test during training and testing days. During training days 49 (day 1 and 2), the platform is made visible by a flag during block 1 (trial 1 and 3). During the testing days (day 50 3 and 4), the platform is hidden during all trials. The drop location of the mice varies in a semi-random fashion 51 as shown by the red arrow. To reduce fatigue and learning limitation, a staggered design is used in RAWM 52 test, with one cohort (10-15 mice) tested in a block before and then a second cohort is tested.

54
A staggered design was used with both cohort 1 and 2 completing block 1 before moving on to block 2 55 (Fig. 2 C). Mice that did not find the platform in 60 seconds were gently guided to it. Each mouse was given 56 15 seconds on the platform at the end of each trial to explore and memorize the spatial cues. When the trial 57 was over, the mouse was gently removed from the pool, dried with a towel and returned to a heated drying 58 recovery cage before the next trial. The goal arm was the same during all the trials and between mice, but the 59 drop location varied between trials in a semi-random fashion (Fig. 2 A and B). To evaluate if the animals were 60 using extra-maze cues to locate the platform, a group of mice was tested in the same maze but the extra-maze 61 cues were removed and the platform was never made visible during the 4-day test. To reduce learning 62 limitations due to fatigue and massive training [10-14, 24, 25] mice were tested in cohorts containing 10-15 63 mice each.

64
An error was scored every time the mouse entered an arm that did not contained the platform or when 65 it entered the goal arm without escaping. If the mouse spent more than 15 seconds in the same zone, arm or 66 center, it was also counted as an error. Total number of errors, latency to escape, distance and velocity were 67 recorded.

71
Closed head injury surgery 72 The mild closed Head Injury (CHI) model was used in this study to demonstrate the applicability of this 73 new behavioral protocol. CHI was performed as previously described using a digital stereotactically (Stoelting) 74 guided electromagnetic impactor device to produce a highly reproducible TBI with minimal mortality [27, 28].
75 Mice were anesthetized with 5% isoflurane before the surgery and kept under anesthesia with continuous 76 inhalation of isoflurane (2.5-3%, 1L/min) through a nose cone during surgery. Before the surgery body weight 77 was recorded, the head was shaved, sterilized with 70% ethanol and 4% lidocaine cream was applied. Each 78 mouse was secured in a digital stereotaxic frame (Stoelting; Wood Dale, IL, USA) using ear bars. The skull 79 exposed after a midline sagittal incision was made. A 1 ml latex pipette bulb was placed under the head of the 80 mouse and filled with water, this helped to diffuse the force of the impact away from the ear bars.

97
The median of errors made by the animal each day was considered for analysis. Only hidden trials 98 were used in the analysis. The AUC was calculated using the trapezoid method, dividing the whole AUC into 99 trapezoidal segments and counting the area of each segment separately, this was done for both errors and 00 distance [29]. The sum of the area of all trapezoids is the total AUC, visible and hidden trials were used for the 01 analyses. We used the following formula: 24 and a visible escape platform is essential for the mice to learn the task and has a positive influence on RAWM 25 output, or if the mice were using other extra maze cues or non-visuo-spatial search strategies. A 7-trial 26 protocol was used to address this question, in particular two conditions were considered: 1) mice were 27 exposed to the pool that was equipped with 4 extra-maze cues and a visible platform was used during training 28 days (Fig. 2 A); 2) in this group of mice no extra-maze cues were used and mice were never allowed to use a 29 visible platform, mice were never trained. As shown in Figure 3A, we found that there was a difference 30 between mice trained (cues and visible platform) to locate the goal arm compared to the mice without cues or 31 access to visible platform (p=0.0023). Also, the AUC (errors x trials) indicated that mice used cues to learn 32 RAWM assay quicker compared to mice not exposed to cues (Fig. 3 B, p=0.0003). In fact, number of errors 33 made by the animals using cues and a visible platform was significantly lower compared to mice not using any 34 kind of cues. Finally, both the total distance travelled (Fig. 3 C, p=0.0031) and AUC (distance x trials) (Fig. 3 D, 35 p=0.0007) were significantly higher in mice not exposed to the use of cues.
36 Figure 3: Evaluation of the use of cues and visible platform during RAWM test. Mice were exposed to a 37 7-trial test for a total of 4-day. (A) A learning curve of the median error per day is shown, a significant 38 difference between mice using cues and mice not using cues was found (*p=0.0023). In (B) the area under the 39 error curve confirmed that mice used cues to learn the task (***p=0.0003). Average (mean) distance (C) and Because of the heterogeneity of TBI, RAWM protocols can sometimes not be sensitive enough to 00 detect injury-induced deficits in behavior. The goal of this paper was to provide a revised protocol for an 8-arm 01 4-day RAWM test. Our approach provided a sensitive protocol to detect cognitive impairments related to a 02 single mild CHI at 2-week post injury. We believe that our protocol will be a useful resource for others 03 attempting standardized behavioral assays, and those laboratories interested in detecting subtle differences in 04 cognitive function in mouse models of CNS injury or disease.

05
A major finding of our study was that the reduction of number of trials significantly separates the 06 acquisition curves of CHI and SHAM mice and the sex had no effect on results. Previously, we demonstrated 07 that CHI mice had memory impairment in a 6-arm RAWM at 2 week post-CHI in a 4-day protocol and 15-trial

24
While we believe that this protocol is more sensitive to detecting changes in mouse models,

36
In conclusion, we demonstrated that: 1) the use of cues will improve the test acquisition, 2) reduction in 37 the number of trials improves learning, and 3) a single mild CHI in mice could cause cognitive deficits 38 detectable by the reduction of trials. We also provided a standard operating procedure and methods to validate 39 the RAWM behavior for use in phenotyping other mouse models or when a pharmacological treatment would 40 be considered.

Acknowledgments
42 We would like to thank Collen N. Bondar, James B. Watson, and Henry C. Snider for their technical support 43 and thoughtful comments on the manuscript.