The French press: a repeatable and high-throughput approach to exercising zebrafish (Danio rerio)

Zebrafish and maintenance

We used 40 (male = 20; female = 20) wildtype zebrafish (Danio rerio) that were approximately 6 months old (see below for the rationale for our sample size). Sex was determined from morphology and confirmed with individual egg laying records. Zebrafish were housed in two 3.5 L plastic aquaria supplied with system water (pH = 7.0 to 8.0; water temperature = 27.0 to 29.0 °C) via a filtration system, and placed in a room with a 14.5 h light: 9.5 h dark cycle. Fish were fed once daily in the evening with commercial pellet food (O.range INVE Aquaculture). All experiments were performed with the approval of the Garvan Institute of Medical Research Animal Ethics Committee (Approval number 15/15).

French press system setup

We set up French press exercise units as illustrated in Fig. 1. Zebrafish swam in the top compartment of the coffee plunger (IKEA Upphetta Coffee/tea maker; 400 mL volume; 8 cm diameter) against circular water currents created by a rotating magnetic stir bar (45 × 7 mm; Fig. 1A). We placed coffee plungers on top of single-plate (Labtek Magnetic Stirrer Hotplate; Fig. 1B) and multi-plate (Labtek 10 Place Stirrer; Fig. 1C) magnetic stirrers for maximal swim performance trials and exercise training, respectively. Since swim performance is shown to be sensitive to water temperature (Wardle, 1980), we recorded water temperature for each trial and replaced water as necessary between trials to keep water temperature constant (average water temperature across trials = 25.5 ± 0.1 °C). Full protocols and a video illustration of the system are available in the Supplementary Information.

Preliminary trials indicated that the water speed in the zebrafish swimming compartment was substantially slower than the set speed of the magnetic stirrer. We thus determined actual water speeds experienced by the zebrafish by tethering a freely rotating plastic leaf above the plunger and determined its rpm for a range of stirrer speeds (200–1,400 rpm). Stirrer speed and water speed exhibited a linear relationship (Fig. S3).

Maximal swim performance

We measured individual maximum swimming speeds of 40 adult zebrafish. We gradually increased the stirrer speed at 10 rpm steps every 5 s until maximum swimming speed, defined as the speed at which zebrafish lost their ability to maintain their position in the water column. Importantly, our step duration was long enough to observe the recovery of zebrafish that were temporarily swept up into the water vortex at sub-maximal speeds, and thus avoid underestimation of maximum swimming speeds. To assess individual repeatability and the effect of exercise training, maximum swimming speed was measured twice for each zebrafish (i.e., pre- and post-exercise training; an outline of the experimental design is shown in Fig. S4). One day after the first round of maximal swim performance trials, each zebrafish (n = 40) was randomly assigned to a control (n = 20) or experimental group (n = 20) that were matched for pre-intervention maximum swimming speeds. Both our first and second round of maximal swim performance trials were conducted blind to the non-exercised (control) and exercised (experimental) groups. All maximal swim performance trials were conducted between 9 am and 3 pm.

Body mass and length of each zebrafish were also measured before each of the two maximal swim performance trials. We weighed individual zebrafish by placing each fish into a small, transparent holding container filled with system water and placed atop a measuring scale (OHAUS Adventure Pro Precision Electronic Balance). To measure length, the holding container was then placed on top of graphing paper (1 mm² grid squares) and photographs of each fish were taken directly from above with a digital camera (Canon PowerShot SX710HS). The body length of each zebrafish was estimated by calibrating individual photographs using ImageJ (version 1.51 k; Schneider, Rasband & Eliceiri, 2012).

Exercise training

To investigate the effects of exercise training, the experimental group received a 40-minute swimming regime at a sub-maximal exercise speed (stirrer speed = ∼370 rpm) for five consecutive days using multi-plate magnetic stirrers. The control group was placed in identical French press units for the same duration without any forced exercise (stirrer speed = 0 rpm). Exercise training aimed to improve swimming ability (Brett, 1964; LeMoine et al., 2010) and the sub-maximal swimming speed was determined during preliminary trials as the highest tolerated continuous swimming speed that did not induce exhaustion within 40 min. All exercise training trials were conducted at 9 am.

Statistical analysis

Our sample size of 40 fish (male = 20; female = 20) was determined by balancing logistical considerations (e.g., the use of 10-plate magnetic stirrer, meaning 10 fish can be exercised at one time) and with reasonable power. Our design only allowed us to detect medium to large effects (Cohen’s d = 0.45; note d = 0.3, 0.5 and 0.8 represents small, medium and large effects; sensu Cohen, 1988) setting the power at 0.8 and with the alpha level of 0.05. All power calculations were conducted using the pwr package (Champely, 2017).

We estimated repeatability (ICC estimates) in maximum swimming speeds using the rpt function in the package rptR, which estimates repeatability using a linear mixed-effects model (LMM) framework (Stoffel, Nakagawa & Schielzeth, 2017). Repeatability (R) describes the proportion of among-group (${\sigma }_{\alpha }^2$) variance out of the total variance (Sokal & Rohlf, 1995): ${\displaystyle R={\sigma }_{\alpha }^2∕\left({\sigma }_{\alpha }^2+{\sigma }_{\epsilon }^2\right).}$ We obtained three different estimates of repeatability, namely unadjusted, adjusted, and enhanced agreement repeatabilities, using separate models for each. Unadjusted repeatability (also referred to as agreement or broad sense repeatability; Shrout & Fleiss, 1979) does not control for variance explained by fixed effects in the repeatability estimate, and in our case, only included individual fish ID as a random effect. Moreover, adjusted repeatability accounts for fixed effects by excluding the variance explained by fixed effects from the repeatability estimate, whilst enhanced agreement repeatability accounts for fixed effects by including their variance in the estimate (Nakagawa & Schielzeth, 2010; Stoffel, Nakagawa & Schielzeth, 2017). We included sex and length of zebrafish, as well as an interaction between treatment and measure as our fixed effects (see Table S1 for coding of our variables). Uncertainty in repeatability estimates was obtained using parametric bootstrapping, which was set to 10,000 for all models.

To analyse the effect of exercise training on maximum swimming speed, we constructed linear mixed-effects models using the lmer function in the lme4 package (Bates et al., 2015) fitting individual fish IDs as a random effect. For fixed effects, we fitted an interaction between treatment and measure to assess whether exercise training increased swim speed in the second round of the maximum swim performance trial. We further included as fixed effects sex and length of individual fish (to investigate differences in maximum swim speed with zebrafish sex and length) and water temperature (to control for any differences due to water temperature). Length and temperature variables were mean-centered prior to analysis. To assess change in mass after exercise training, in a separate model we fitted sex of zebrafish (to account for sex differences in mass), and an interaction between treatment and measure as fixed effects. We assessed statistical significance (at the alpha level = 0.05) using Satterthwaite’s approximation for degrees of freedom obtained from the package lmerTest. Confidence intervals for variables were obtained using the confint function provided in the lme4 package. All statistical analyses were conducted in the R programming language (v 3.2.2; R Development Core Team, 2016; see Supplementary Information for data and analyses).