For decades, the elevated plus maze (EPM) has been the most popular test for the evaluation of anxiety-like states of rodents following experimental manipulations and for testing pharmacological agents (Griebel and Holmes, 2013; Haller et al., 2013). The test was developed and pharmacologically validated in rats (Pellow et al., 1985) and mice (Lister, 1987), and is comprised of two enclosed arms and two exposed arms joined by a central square. Shepherd et al. (1994) modified the design into an elevated “zero-maze” (EZM) with alternating “open” and “closed” quadrants, which had the benefit of removing the central square region of the EPM, resulting in simpler analysis of closed and open quadrants only. Interpretation of time spent in the central square of the EPM is often difficult, and was particularly a problem in mice as they can spend at least 20%–30% of the test session in this region (Lee and Rodgers, 1990; Rodgers et al., 1992). The EZM also eliminated the “boxed-in” regions of the closed arms of the EPM, allowing more continuous exploration of the apparatus.

The EPM and the EZM (as well as other tests including the open field and light-dark box) are referred to as unconditioned, “approach-avoidance” tasks that rely on rodents’ innate conflict between approaching and exploring (foraging) vs. avoiding potentially dangerous areas (Cryan and Holmes, 2005; Cryan and Sweeney, 2011). The bright and exposed regions of the EPM and EZM represent the “dangerous” areas of the mazes, whereas the darkened and enclosed regions are perceived as safer. Although significant criticisms of these simple tests should be noted, and on their own unconditioned tests of anxiety are insufficient for meeting the needs of all pre-clinical research in anxiety (Haller et al., 2013; Ennaceur and Chazot, 2016), they will likely remain in use due to their extensive history and ease of use, and are thus worth further exploration for understanding as models of anxiety-like behaviors in rodents.

Although the two tests are based on the same concept, designed and carried out similarly, and results are interpreted in the same way, it can be unclear which test is more appropriate to employ in a given experimental design as the output is not necessarily identical. The goal of this research was to compare, under identical testing conditions, the behavior of male and female C57BL/6J mice in the EPM and EZM. As behavior in the EPM is known to change with repeated trials, mice were tested at daily or weekly intervals for a total of five trials.

In this direct comparison of performance of male and female mice in the EPM and EZM, we found that although mice were equally active in both mazes when presented as novel environments, the animals tested in the EZM spent significantly more time in the anxiogenic “open” regions of the maze than the mice that were tested in the EPM. There are few studies in which the EPM and the EZM are both employed in the same laboratory as assays of anxiety-like behaviors. However, Amani et al. (2013) also tested separate groups of mice in the EZM and EPM, and showed that mice tested in the EPM only spent about half the amount of time in the open arms of the apparatus as the mice tested in the EZM spent in the open quadrants, which is very comparable to our current results. Greater amounts of time in the open quadrants of the EZM compared to open arms of the EPM have also been observed when a within-subjects design was employed (Pearson et al., 2015). Braun et al. (2011) directly compared the two mazes in rats, and also reported increased times spent in anxiogenic regions of the EZM compared to the EPM.

The amount of exploration and the percent of time spent in the open arms of the EPM decreased significantly (by approximately 35% and 50%, respectively) in both male and female mice after the first trial in the apparatus regardless of whether they were tested daily or weekly. The “one-trial tolerance” phenomenon (OTTP) is well-documented in the EPM in both rats and mice, in which after a first exposure to the maze, rodents significantly decrease their exploration of the open arms on subsequent exposures, and anxiolytic agents (e.g., benzodiazepines) are no longer effective at increasing the amount of time spent in anxiogenic zones (e.g., File et al., 1990; Rodgers et al., 1992; Rodgers and Shepherd, 1993; Treit et al., 1993; Holmes and Rodgers, 1998, 1999; Zhou et al., 2015). There are extensive studies aimed at describing the conditions under which the OTTP occurs and the potential underlying mechanisms, discussion of which goes beyond the scope of this article (but see Holmes and Rodgers, 1999; Gomes and Nunes-De-Souza, 2009; Roy et al., 2009; Zhou et al., 2015). However, it is widely agreed that the OTTP makes the EPM an unsuitable assay for measuring anxiety-like traits in longitudinal studies.

It is possible that extending the inter-trial interval well beyond 1 week may allow re-testing in the EPM without carry-over effects from previous exposure to the maze. It has been shown that increasing the inter-trial interval to 28 days returns performance of saline-treated rats to that observed in trial 1 (Schneider et al., 2011), although a room change was also required to preserve the anxiolytic-like effects of midazolam during the second trial. Zhou et al. (2015) recently confirmed that a 28-day interval restores behavior to baseline performance in rats, but there are currently no studies that have attempted to extend the testing interval in mice to a degree that would eliminate the OTTP. However, many experimental designs may require sampling anxiety-related behaviors at shorter intervals, thus making the EPM a difficult test to employ for longitudinal studies.

In contrast to the EPM, behavior in the EZM remained relatively stable for several trials even when animals are tested daily; amounts of exploration and time spent in anxiogenic zones decreased much less than in the EPM over the duration of the experiment. Between the first and second trials, total amounts of exploration and the time spent in the open quadrants decreased by approximately 7% and 5%. Although the change in the amount of ambulation in the EZM from trial 1 to trial 2 was statistically significant during daily testing, the decrease was much smaller than that measured in the EPM (about 35%). When the testing interval in the EZM was extended 1 week, the amount of exploration remained consistent for three trials. Also unlike the EPM, the decrease in open quadrant activity between the first two trials was non-significant; these behaviors remained stable until at least the fourth trial. Behavior of rats in the EZM remains consistent when the testing interval is 1–2 months (Ajao et al., 2012; Kamper et al., 2013), and Blokland et al. (2012) concluded that reducing the testing interval to 24 h does not affect the stability of anxiolytic-like behavior of rats in the EZM for at least four trials. However, Cook et al. (2002) reported a significant decrease in the time spent in the open quadrants of the EZM when male C57BL/6J mice were tested 24 h following an initial trial. The percent time spent in the open quadrants in the first trial of that study was approximately 15% (compared to 40% in our study), the walls of the closed quadrants were transparent, and the test was performed in dim lighting (44 Lux). These large differences in baseline performance and testing conditions make direct comparisons difficult, and more studies are needed to determine if behaviors in the EZM can remain consistent over many trials as our data suggest, or if this assay, like the EPM, will also show habituation of behaviors over multiple trials in mice.

Significant sex differences in this study were limited to activity levels, with female mice ambulating greater distances in the mazes than male mice. We have previously reported increased activity levels in female mice compared to male mice in the EZM and in the open field test (Tucker et al., 2016a), and there is a long history of research largely concluding that female mice tend to be more active than their male counterparts (e.g., Archer, 1975; Kokras and Dalla, 2014), although not all studies support this conclusion (e.g., Bolivar et al., 2000; An et al., 2011). In the current study, there was a trend toward reduced anxiety-like behaviors in female mice (increased time in the open arms (EPM) and open quadrants (EZM)), and we previously found significantly increased time in the open quadrants of the EZM in female mice (Tucker et al., 2016a). Consistent with these results, higher levels of anxiety in male mice in the EPM have also been reported (Rodgers and Cole, 1993; Võikar et al., 2001; Gioiosa et al., 2007; Painsipp et al., 2007; Walf et al., 2008; Hendershott et al., 2016). Behavior of female mice in anxiety tests may be influenced by the estrus cycle (Galeeva and Tuohimaa, 2001; Gangitano et al., 2009; Walf et al., 2009; Koonce et al., 2012), and some studies have demonstrated an anxiolytic-like effect of ovarian hormones in mice (Frye et al., 2004; Olesen et al., 2011; Koonce and Frye, 2013). Despite these effects, the collective behavior of female mice in the EZM in the current study was just as consistent over multiple days as it was in males despite the estrus cycle. Furthermore, it has been demonstrated that the use of female mice at undetermined stages of the estrus cycle does not contribute to variability in the data of behavioral experiments (Prendergast et al., 2014; Tucker et al., 2016a,b).

It should be noted that dissociating anxiety from locomotion may be difficult in tests that rely on exploratory drive such as the EPM and EZM (O’Leary et al., 2013), and with sex differences in levels of activity, tests that rely less on exploration may have greater construct validity in pre-clinical studies of sex differences in anxiety (Kokras and Dalla, 2014). Furthermore, correlations between levels of activity and amount of time spent in anxiogenic regions of the EZM have been reported (Tarantino et al., 2000), lending further support to concerns that differences in anxiety-like behaviors may be confounded by variability in locomotor activity.

A limitation of this study is a lack of data on changes in the effects of anxiogenic and anxiolytic treatments over multiple trials in the EZM. The reduction in the efficacy of anxiolytic drugs in the EPM after a single trial is well-studied (the OTTP), but there is a paucity of data regarding this phenomenon, if it exists, in the EZM. In addition, future studies should look at changes in ethological or defensive behaviors such as stretch attenuated postures and head dips, as these are also parameters that are altered by repeated trials and pharmaceutical agents in the EPM, and considered important assessments in these mazes that help provide a more complete behavioral profile and are less influenced by levels of motor activity (Rodgers et al., 1997; Cryan and Holmes, 2005). Furthermore, differences in behaviors between mouse strains have been measured in both the EPM (e.g., Rodgers and Cole, 1993; Võikar et al., 2001; Rodgers et al., 2002; O’Leary et al., 2013) and the EZM (Tarantino et al., 2000; Cook et al., 2001; Milner and Crabbe, 2008), but direct comparisons of performance on both mazes in more strains are needed.

In conclusion, the design of the EZM encourages greater exploration of the open and brightened “anxiogenic” regions of the apparatus than that of the EPM, likely due to the absence of the central square and the requirement of passage through the open quadrants for continuous movement and exploration. Behaviors (levels of exploration and time spent in the open arms) significantly decreased after the first trial in the EPM (the OTTP); further studies are needed in mice to determine if extending the inter-trial interval well beyond 1 week will allow longitudinal testing with this apparatus. In contrast, in the EZM the same behaviors remained constant for at least three trials in our laboratory testing conditions (independent of inter-trial interval), suggesting this assay may be much more suitable for experimental designs that require the measurement of anxiety-related behaviors at multiple time points following experimental manipulations. However, the high baseline levels of exploration of anxiogenic regions of the EZM (in our laboratory conditions) may render it less sensitive than the EPM to anxiolytic agents; further pharmacological characterization of this apparatus is warranted.

LBT designed the experiment, collected and analyzed the data and wrote the manuscript. JTM served as principal investigator, contributing to the design of the experiment and analysis of the data, and assisted with the manuscript.

This work was supported by The Center for Neuroscience and Regenerative Medicine, 60855-300600-7.01.

The opinions, interpretations, conclusions and recommendations are those of the authors and are not necessarily endorsed by the U.S. Army, Department of Defense, the U.S. Government or the Uniformed Services University of the Health Sciences. The use of trade names does not constitute an official endorsement or approval of the use of such commercial hardware or software. This document may not be cited for purposes of advertisement.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.