Protocol to investigate the gradual selection and deployment of goal-oriented search strategies during unsupervised navigation in mice

Summary

The ability of rodents to effectively navigate in an environment is based on trial-and-error learning and flexible decision-making and can be analyzed via navigational trajectories. We present a protocol for studying the deployment of search strategies in mice using the Morris water maze. We describe steps for assigning mice to different maze variations and procedures for post-training tracking and analysis. This protocol represents an effective behavioral readout to probe brain networks involved in strategy deployment and goal-oriented behavior.

For complete details on the use and execution of this protocol, please refer to Parrini et al.¹

Graphical abstract

Highlights

• •Steps for analyzing orientation strategies used by mice learning MWM
• •Objective criteria for classifying swimming trajectories
• •MWM variations to explore how different information supports goal-oriented learning

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

The ability of rodents to effectively navigate in an environment is based on trial-and-error learning and flexible decision-making and can be analyzed via navigational trajectories. We present a protocol for studying the deployment of search strategies in mice using the Morris water maze. We describe steps for assigning mice to different maze variations and procedures for post-training tracking and analysis. This protocol represents an effective behavioral readout to probe brain networks involved in strategy deployment and goal-oriented behavior.

Before you begin

The Morris water maze (MWM) is a well-established unsupervised navigation task reliably and effectively learned by rodents.^2,3,4

The traditional MWM setup consists of a large circular pool filled with opaque water. The pool is surrounded by distinct visual cues placed at fixed positions. A hidden platform is submerged just beneath the water surface and serves as the target location that mice need to find in order to complete the task and be removed from the pool.

Most studies where this paradigm is used primarily focus on the latency to reach the hidden platform as a straightforward readout of mice performance.⁵ Other measures proposed in the literature include the swimming path length, which is suggested to be a better measure than escape latency.^1,6,7

The readouts mentioned above provide no information as to how mice go about solving the MWM task, whereas analyzing deployment and implementation of search strategies can offer a deeper understanding of the cognitive processes involved in navigation.^8,9

We have recently used the procedures described in this protocol together with brain region-specific loss and gain of function experiments to provide insights into how mice learn to navigate mazes and to reveal brain mechanisms of goal-related search strategy selection and deployment throughout the process of unsupervised navigation learning.¹

Institutional permissions

Animal care, housing, and procedures were approved and performed in accordance with the Veterinary Department of the Canton Basel-Stadt.

Preparation of the testing room and behavioral setup

Timing: 2–3 weeks

• 1.The testing room needs to be large enough to accommodate a circular pool with a diameter of 140 cm, vertical stands used to display high contrast (black/white) spatial cues (plastic board on metallic stands, 50 cm from the pool external wall), desk and computer.

CRITICAL: The pool, 100 cm height, is made of chemical resistant black polyethylene with a centrally located drainage tubing and a metal wheeled base. In order to minimize any distraction, the room needs to be quiet and the pool must be surrounded by an opaque curtain. Moreover, any remnants from previous experiments (for example paper towels used to clean cages) must be removed. In this way the only cues that mice can use to navigate are those placed and controlled by the experimenter. It is important that the cues are not moved during the whole length of the experiment. The curtain should allow access to the pool from different cardinal points in order to minimize any spatial bias in animal learning.

• 2.The water in the pool must be maintained at a consistent temperature throughout the experiment. The ideal temperature for mice is 24°.
• 3.The water in the pool needs to be made opaque with the use of liquid nontoxic water-based paint (approximately 1:100).

CRITICAL: Ideally the pool should be directly connected to a water supply system. It is recommended to change the water every week in case of medium occupancy or every few days in case of longer behavioral sessions with many mice. Remember to check water temperature before each daily training session. In case of need hot water can be added.

• 4.The room must be adequately lit to ensure that the animal can see the pool cues and navigate the maze.

CRITICAL: The lighting should not be too bright, as extreme lighting condition can affect animal's behavior. Ideal light levels are 30–40 lux.

• 5.To monitor animal’s behavior a camera (CCTV attached via coaxial cable to a capture card in the PC) needs to be placed above the pool and a specific tracking software (Viewer2 from Bioserve) should be used to follow swimming trajectories online during the experiment. The setup of the tracking system (pool area and quadrants) should be kept constant during the whole length of the experiment.
• 6.Pre-testing cages need to be prepared and immediately available in the room.

CRITICAL: These cages are not covered with lids and do not contain any bedding material apart from paper towel. Moreover the pretesting cages do not include any shelter or any enrichment in general as mice are gently placed in these cages just before starting the experiment and in between trials.

• 7.A 100 W (Philips) infrared light needs to be placed beside the pre-testing cages in order to ensure animals are warm during the waiting time between two different trials and that they are completely dry before being returned to their home cages.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Experimental models: Organisms/strains
Mouse: C57BL6/J male or female from 3 to 8 months PV-Cre mice (129P2-Pvalbtm1 (Cre)Arbr/J) and TRAP2 mice (Fos2A-iCreER knock-in (also called Fos2A-iCreERT2 )	Charles River Laboratories Jackson Laboratory	N/A
Software and algorithms
GraphPad Prism 9.5.0	GraphPad Software	https://www.graphpad.com; RRID:SCR_002798
Viewer2	Biobserve	http://www.biobserve.com/behavioralresearch/products/viewer/; RRID:SCR_014337
EthoVision XT	Noldus	https://www.noldus.com/ethovision-xt; RRID:SCR_000441
Excel	Microsoft	https://www.microsoft.com/en-us/microsoft-365/excel
Other
Custom-made MWM pool	This paper	N/A
Liquid nontoxic white paint	Stationery shop	Giotto-Schoolpaint ready to use 500 mL - white

Step-by-step method details

Habituation and handling of animals

Timing: 1–2 weeks

The ideal age of mice at the onset of the experiment is 3 months. Different lines were used (as described in the key resources table) and they have showed no statistically significant difference of behavior in terms of relative use of swimming trajectories throughout learning of the MWM (this is also true for different sexes). It is crucial for mice to be habituated to the experimenter who will be the sole person responsible for their handling during the whole length of the protocol.

• 1.If mice are purchased from an external supplier, ensure that they arrive at least one week prior to behavioral testing.
• 2.If mice are tested after a major surgery (for example in chemogenetic gain or loss of function experiments), the minimum waiting duration of one week needs to be respected before testing.
• 3.Spatial learning and memory deficits were reported in single housed mice compared to group housed mice in the Morris water maze.^10,11,12 For this reason, single housing is not recommended in this protocol; the ideal housing condition is of three mice per cage.

Morris water maze: Day1 visible platform

Timing: 1 day

During the first day of training (day1) the setup of the room (included visual cues and lighting) is completed and will stay fixed for the whole duration of the experiment. The escape platform, however, is visible and is positioned just above the water surface. This ensures that mice get used to a completely new condition (swimming in water) and to the general goal of the task (finding and jumping on a platform in order to be removed from water).

• 4.Ensure the water is opaque and at the right temperature. If necessary, add some liquid non-toxic paint and mix properly before introducing the animals.
• 5.Start the camera and the tracking software.
• 6.Place the platform (10 cm diameter) 0.5 cm above the water surface in the center of a fixed quadrant. The position of the platform will not be changed during the whole duration of the behavioral session on day 1.
• 7.Transport mice, preferably using a trolley, from the housing room to the experimental room 10–15 min before starting the experiment.
• 8.Gently remove mice one by one from the home cage by holding the base of their tail and placing them on a gloved hand.
• 9.Transfer mice to pretesting cages and allow them 5 min to get habituated to it.

CRITICAL: The pretesting cages must not present any odor cue. For this reason, they always need to be cleaned with 70% ethanol before and after each session.

• 10.Gently remove one mouse from the pre-testing cage and place it in the pool facing the wall (to minimize acquisition of information about cues before the trial starts).
• 11.Take a step back from the pool and ensure curtains are closed. Tracking should automatically start once the mouse is released.

CRITICAL: Mice are trained to find the platform for 4 trials per day each lasting for maximum 60 s, with an inter-trial interval of 5 min. It is standard procedure to virtually divide the pool surface in 4 quadrants, using two imaginary perpendicular lines crossing at the center of the pool. The ends of each line demarcate four cardinal points (which do not correspond to the magnetic compass direction). For each trial, the entry point of the mice in the pool is chosen among these four points. As the platform is placed at the center of a given quadrant, two of the four cardinal points will be closer to the platform (in the example in Figure 2, the closer points will be S and E). An alternation of entry from one closer point or one farther point is to be maintained (for example, if the pool organization would be as in Figure 2 an example of the four entry points for each trial could be E,W,S,N).

• 12.During the inter-trial interval mice are returned to the pretraining cage and are placed under the infrared lamp.

CRITICAL: If a given mouse finds the platform before the 60 sec cut-off, it is allowed to stay on the platform for 5 s and is then returned to its pretraining cage. If it did not find the platform, it is manually guided, allowed to stay on the platform for 5 s and then returned to its pretraining cage. Avoid to just place the animals on the platform, they should be gently guided to swim towards it and to jump on it.

CRITICAL: When removing mice from the platform allow them to jump on the experimenter hand. Mice will associate the different handling methods to the different phases of the trial. In particular, holding them at the base of their tail will be associated with the beginning of the task, while jumping on the experimenter hand will be associated with the end of a given trial and/or of the daily learning session.

• 13.Observe the behavior of each mouse online during testing.

Note: An automatic trigger in the tracking software that indicates the end of the trial when the mouse has been on the platform for 5 s is preferable.

• 14.Once all 4 trials of day1 are completed return the animals to their home cage and in the housing room. Ensure they are properly dry before this step.

Figure 2.

Schematic representation of the MWM pool with definition of swimming areas in the tracking system

1, closer wall zone; 2, annulus; 3, target area; 4, platform; 5, goal corridor. The dashed lines represent the virtual division of the pool in 4 quadrants and the creation of 4 fictive cardinal points. These will be used as entry location of mice in the pool.

Morris water maze: Invisible platform (day2–day10) and four different variations with distinct goal-oriented criteria

Timing: 10–25 days

We describe four variations of the classical MWM protocol (Figures 1A–1D). These consist of placing mice in the same water maze setup, including the same salient fixed cues around the pool, but with different platform position schemes reflecting distinct criteria to locate the hidden platform. Each variation can be used to train mice over several days in order to study the relative use of goal-specific swimming trajectories. Moreover, mice trained in one of the variations for up to 10 days can be switched (after 4–5 days of pause) to a different one in order to study flexible adaptation to the new goal-related rule.

Figure 1.

Strategy deployment in 4 different variations of the MWM

Strategy deployment plots for position (A, n = 12), chaining (B, n = 11), radius (C, n = 10), and random (D, n = 6) maze. Bars represent standard error. Figure reprinted and adapted with permission from Parrini et al.¹

The use of this approach together with murine disease models or brain-specific gain or loss of function experiments provides an effective way to test brain network’s function during strategy deployment and goal-oriented behavior.

• 15.Repeat steps 4 and 5 of the previous section.
• 16.
  Define the variation of the MWM to be used (Figure 1) and place the platform, which is now hidden under the water surface, accordingly (if required, change the platform position at each trial):

  • a.
    Position Maze: the platform remains in the same position throughout all trials and days, in the opposite quadrant of the one used during day 1. This corresponds to the classical and most used version of the MWM and allows assessment of mice spatial orientation capabilities based on the use of distal cues.
  • b.
    Chaining Maze: the position of the platform is pseudo-randomly changed at each trial but the distance from the pool wall is kept constant at 20 cm. Location of the platform in this variation of the maze is based on the acquisition and use of intra-maze characteristics, namely the distance of the platform from the wall.
  • c.
    Radius Maze: the position of the platform is pseudo-randomly changed at each trial along a fixed radius in a given quadrant. Similar to the position maze with respect to the use of distal cues to locate the target quadrant, it requires more flexibility in the search of the platform at a non-fixed position.
  • d.
    Random Maze: the position of the platform is pseudo-randomly changed at each trial in the whole area of the pool. This is a “control” variation where no regularities can be identified and used for orientation.
• 17.
  For pseudo-randomly assigned platform positions proceed as follows:

  • a.
    Subdivide each of the four pool quadrants into three wedges of the same size.
  • b.
    Define five possible positions along the radius from the wall to the center of the pool, with platform centers at respectively 5, 20, 35, 50 and 65 cm from the wall.
  • c.
    Run an Excel macro, based on the use of the Excel RAND() formula, at the beginning of each trial, which returns three random numbers.

The first one (from 1 to 4) represents the quadrant, the second one (1–3) represents the wedge inside the quadrant and the third number (1–5) represents the position along the radius. For the chaining protocol the third number is fixed at 2 (corresponding to the platform center at 20 cm from the wall). For the radius protocol the first number representing the quadrant and the second number representing the wedge are also fixed and only the third number is modified. For the random protocol all numbers are assigned and changed pseudo-randomly before each trial.

• 18.Repeat steps from 7 to 12 in the previous section.
• 19.While observing the behavior of each mouse during the trials, classify the swimming trajectory as described in the section below.

Note: The scoring performed during the behavioral session is then compared to the scoring performed offline, based on swimming trajectories extrapolated via Ethovision software.

CRITICAL: It is important to use objective criteria for strategy classification. This can be obtained via blinded analysis by at least two independent experimenters and via the definition of separate areas in the online tracking software and Ethovision, together with quantification of the time spent in each area (see section below).

Classification and analysis of search trajectories

Timing: 1 week

A key feature of this protocol is to score mice behavior based on the search strategy deployed. This involves classifying trajectories according to the 8 categories defined in the following point 23. This could be used, for example, when a detailed step-by-step analysis of swimming trajectories or swimming accuracy measurements needs to be performed. However, it is also possible to score the behavior based on search strategies reflecting distinct goal-related criteria as described in point 24 of this section. The latter approach better reflects the gradual deployment of a more efficient searching behavior throughout the task. Moreover, it helps defining general behavioral patterns for the acquisition and use of goal-related information.

• 20.
  Define the swimming areas in the online tracking system and Ethovision as follows (Figure 2).

  • a.
    A 5 cm large circular area close to the pool wall (closer wall zone);
  • b.
    A circular area 5–35 cm from the pool wall (anulus);
  • c.
    A 20 cm large corridor along direct connection between start and platform positions (goal corridor);
  • d.
    an area with the platform as a center and 30 cm radius (target area).
• 21.Derive from Ethovision the swimming trajectories for each trial and each day of training.
• 22.Obtain from Ethovision the time spent in each area.
• 23.
  Classify the trajectories as follows:

  • a.
    Thigmotaxis (T): >60% of swim time within closer wall zone.
  • b.
    Chaining (C): >60% time spent within the anulus zone.
  • c.
    Directed search (S1): >80% of time in goal corridor.
  • d.
    Scanning in quadrant (S2): >80% time spent in the target quadrant excluding the target area.
  • e.
    Focal search (S3): >80% time spent in the target area.
  • f.
    Direct swim (S4): 100% time spent in goal corridor.
  • g.
    Global search (G1, G2): not fitting the definitions for any of the spatial sub-strategies, nor for chaining, and >20% swim time spent within the closer wall zone (G1) or <20% of swim time spent within the closer wall zone (G2) (Figure 3).
• 24.Group sub-strategies S1-S4 into Spatial search (red code; search criterion defined by particular position in space; specifically, >80% of time in goal corridor and/or target quadrant Figures 1, 2, and 3) and G1-G2 into Global search (blue code; no maze-specific search criterion; defined as not fitting definitions of spatial searching or chaining, and <60% of swim time within closer wall zone; Figures 1, 2, and 3) and compare their relative deployment with that of chaining (C-green code).

Figure 3.

Classification and distribution of strategies in the position maze

(A) Schematic of main searching strategies and corresponding substrategies The red lines and circles refer to the classification criteria as described in Figure 1. T, thigmotaxis; G1, global search 1; G2, global search 2; S1, directed search; S2, scanning in quadrant; S3, focal search; S4, direct swim.

(B) Substrategy deployment during position maze training (n = 12). Figure reprinted and adapted with permission from Parrini et al.¹

Expected outcomes

The behavior displayed by mice in each variation of the MWM is highly reproducible among different animals and lines (Figure 1).

Position maze

During position maze training, the initial prevalent use of Global search strategies (G1 and G2) is gradually replaced by a highly biased increase in the deployment of maze-specific search strategies (C and S1-S4; Figure 1A). This pattern is even more evident when maze learning is monitored based on the three main categories of Global, Chaining and Spatial search. Thus, a 20% of spatial searching on days 3 and 4 is accompanied by 15%–20% of chaining from day 4 on, and by a preferential and increasing deployment of spatial searching at the expense of global searching from day 5 on (day 5: about 33% spatial and 55% global searching; day 10: 80%–85% spatial and less than 5% global searching).

Chaining maze

During learning of this MWM variant, since platform position is varied pseudo-randomly at every trial along a constant distance from the wall, the use of spatial strategies does not produce reliable success. Indeed, the analysis reveals early deployment of chaining but not spatial searching (chaining: about 20% on day 3, 80% on day 6; spatial searching: 0% from days 2–6), a gradual replacement of global searching with chaining between days 3 and 8 (global searching: about 70% on day 3, 20% on day 8), and a nearly complete absence of spatial searching throughout chaining maze learning (Figure 1B).

Radius maze

In the Radius maze, chaining does not reliably lead to success, and successful spatial search strategies are directed toward an entire radial sector (scan in quadrant S2). The strategy analysis revealed the expected complete absence of chaining and a gradual replacement of global with spatial searching between days 3 and 5 (global searching: about 80% on day 3, 55% on day 5; spatial searching: 20% on day 3, 45% on day 5), with a specific increase in radially directed spatial searching. A comparable deployment of global and spatial search strategies can be observed beyond day 5 (Figure 1C).

Random maze

No reliable decrease in the latency to locate the hidden platform can be observed in this version of the MWM. The strategy plot reveals the expected absence of chaining or spatial strategies and the exclusive presence of global searching, consistent with the notion that global searching represents a default strategy in the absence of perceived goal-related rules (Figure 1D).

Quantification and statistical analysis

For each trial we assigned a score of 1 to the strategy used and we determined the relative percentage of its recurrence over the four trials for each day and mouse. We then calculated the average use of each strategy by all animals during each training day. When mice deployed a chaining strategy they tended to swim in the same direction until they found the platform; we observed no obvious prevalence of a particular swimming direction for chaining in any of the experiments (clockwise or counterclockwise).

All statistical analyses can be performed using GraphPad prism 9.5.0 (GraphPad software, Inc.). Comparisons of data among different MWM variations are possible. Moreover, mice in different treatment groups can be compared in their learning performance of the same maze variation. According to the specific dataset, search strategy patterns are compared using two-way RM ANOVA followed by Tukey or Fisher’s multiple comparison test.

Limitations

As an unsupervised behavioral task, the Morris Water Maze is a particularly valuable tool as it includes characteristics and challenges routinely encountered by rodents under naturalistic circumstances. Compared to other protocols it requires fewer training sessions. Progression (or failure to progress) can be assessed easily and in an accurate way.

The variations described in this protocol and the detailed analysis of swimming trajectories offer the possibility for a deeper insight into cognitive functions and decision making processes.

Despite these advantages, the task presents a few limitations. First of all, one experimental room needs to be dedicated to the task and the presence of more than one experimenter is not recommended. Moreover, while mice are able to perform the task after a major brain surgery, the use of head implants can be very limiting, or impossible if an electrical component is included.

Troubleshooting

Problem 1

Mice refuse to jump on the platform even when they find it and might continue to incessantly randomly swim throughout the entire pool area (Morris water maze: Day1 visible platform).

Potential solution

This problem can occur when mice are not properly habituated to the new housing conditions after being transferred from an external facility or if the housing or experimental conditions are stressful (e.g., loud and sudden noises, unsettling odors, presence of unknown personnel). The habituation period and minimizing the source of stress are both essential. In the case of mouse lines more prone to be affected by stress, longer sessions of handling before the experiment are advised.^13,14,15

Problem 2

Mice do not move in the water and float (Morris Water Maze: invisible platform (day2–day10) and four different variations with distinct goal-oriented criteria).

Potential solution

Pay particular attention to training during the first day of visible platform. Gently guide the mice to the platform and let them jump on it. Aged mice might tend to display this behavior more often and to use non-spatial strategies to solve the maze.^16,17 Do not use mice older than 8 months.

Problem 3

Mice do not show improvement in the deployment of strategies appropriate for the specific goal-related rules in a MWM variation (Morris Water Maze: invisible platform (day2–day10) and four different variations with distinct goal-oriented criteria).

Potential solution

Ensure visual cues are well visible and easily discriminated. Ensure the room does not present any acoustic, odor-related or visual distraction.

Problem 4

The tracking software fails to track the swimming trajectories (Classification and analysis of search trajectories).

Potential solution

This problem can be due to a poor contrast between the mouse and the water surface and becomes particularly evident for mice with a lighter fur. In order to avoid this problem, ensure proper lighting of the pool area and avoid shadows.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Pico Caroni (caroni@fmi.ch).

Technical contact

Questions about the technical specifics related to performing the protocol should be directed to and will be answered by the technical contact, Maria Spolidoro (mariaspo@gmail.com).

Materials availability

This study did not generate new unique reagents. Requests for viral constructs produced and obtained from FMI vector core should be directed to the lead contact.

Data and code availability

• •All data including microscopy data reported in this paper will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank members of the Caroni group for discussions about the project. M.P. and M.S. were supported by funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (MemoryDynamics) and by the Novartis Research Foundation (FMI).

Author contributions

M.S., M.P., and P.C. devised the protocol. M.P. and M.S. conducted all experiments and analyses. P.C., M.P., and M.S. wrote the manuscript.

Declaration of interests

The authors declare no competing interests.