One hundred and sixty two male Sprague Dawley rats, aged between 5 and 6 months, bred and maintained in standard Makrolon cages (length: 60 cm, width: 34, height 20 cm; 3 animals per cage) filled with autoclaved woodchips in the Core Unit of Biomedical Research, Division of Laboratory Animal Science and Genetics, Medical University of Vienna, were used. Food (sniff®, Soest; Germany) and tap water was available ad libitum. Facility conditions were: temperature: 22 ± 2°C; humidity: 55 ± 5%; 12 h artificial light/12 h dark cycle: light on at 7:00 a.m.). All procedures were carried out according to the guidelines of the Ethics committee, Medical University of Vienna, and were approved by Federal Ministry of Education, Science and Culture, Austria. The animals are subsequently used in another study, the present tests representing a pretest to that. The follow-up study requires this number of animals and at the same time allows multivariate statistics.

Behavioral tests were performed in the following sequence: Elevated plus maze, open field, rota-rod, forced swim test, holeboard test. The apparatuses (Bilek+Schüll GMBH; Vienna, Austria) were custom made as well as the software (TIBE V.1.0; Vienna, Austria) for analysis. Hole visits in the holeboard were recorded manually from videos.

The between tests interval was 1 week. Because the ratio between the number of analyzed variables and number of individuals is critical for the validity of PCA results we chose one common test only for each feature. Also the number of variables within each test was minimized to those with the highest variance between individual rats.

These tests represent a standard test battery used for the evaluation of side effects in pharmacological studies (Sunyer et al., 2007, 2008).

The elevated plus maze is a common test to assess anxiety (Belzung and Griebel, 2001). It consists of a cross of two closed arms with side walls and two open arms. Rats and mice usually prefer to stay in closed arenas and avoid the exploration of open arenas (Carobrez and Bertoglio, 2005; Arantes et al., 2013). Thus, the longer the animals stay on open arms the lesser they are considered to be anxious. The elevated plus maze consists of two plus shaped plastic lanes (10 cm in width, 1.10 m in length at a height of 62 cm), with one lane (closed arm) surrounded by black plastic walls (40 cm in height), except at the crossing point with a gap of 10 cm, so that each arm had a length of 50 cm. The animals were placed in this center and allowed to explore the maze for 5 min. The movements were recorded by a tracking system and stored on a computer. The ratio of the time (s) the animals spent in the closed and the open arms (EPMRCO), the ratio of distances traveled and the rest in closed and open arms (EPMRDOC), the local movement (EPMlocal, <10 cm/s) and the number of entries into the open arms were considered for the analysis. The last three variables were choosen in order to compare it with corresponding variables in the open field (Carola et al., 2002) with regard to locomotion and the motivation to explore open arenas (the exploration of the central arena in the open field was almost invariant between animals).

The open field test is widely used to test exploratory behavior and general activity of mice and rats (Crusio, 2001). The open field consists of a black plastic board (1.20 m × 1.20 m) surrounded by black plastic walls (50 cm in height). The rats were placed in the arena and allowed to explore it for 10 min. The movements were recorded by a video tracking system and stored on a computer. Considered were the ration between distance traveled (m), and the time the animals did not move (s) OFRDR, local movements (OFlocal, <10 cm/s) and the average velocity of movement (OFveloc; m/s). The usually used variables of the ration of the time spent in the central and peripheral part of the arena in order to assess the motivation to voluntarily explore unsafe areas were not considered because there was minimal variability between animals, thus they do not provide information about individuality and unnecessarily increases the number of variables in the PCA. In order to provide information about the readiness to enter unsafe areas the number of entries in open arms of the elevated plus maze test has been considered (Carola et al., 2002).

The rota-rod is a widely used test to assess motor disabilities of rodents. The time-length the animals are able to stay on a rotating rod is considered to be a correlate of motor abilities (Brooks and Dunnett, 2009). The apparatus consists of four adjacent rods (10 cm in width at a height of 40 cm) separated with plastic barriers. The animals were placed on the rods which rotated with increasing speed. Rats were tested three times. The time until the rats fall off the rods was measured and the mean time (RRtime; s) for the three tests was used for analysis.

The forced swim test is a common test to assess depressive-like behavior. Therefore the time the animals are immobile and not struggling to escape from the water is considered to be a correlate of depressive-like behavior (Slattery and Cryan, 2012). The test consists of a training and a test session on two consecutive days. The rats were placed in a translucent plastic cylinder (20 cm in diameter and 45 cm in height filled with water up to a 33 cm level at a temperature of 25°C). The rats were placed in the cylinder for 10 min during the training and 5 min during the test session. Movements in the cylinder were recorded by a video tracking system and stored on a computer. The time (s) the animals spent immobile during training (FSIMTR) and during test (FSIMT) were used for the analysis.

The holeboard is a test of spatial learning and memory and, by using the same protocol as in the present study, has been found to be reliable and highly sensitive to pharmacological interventions, induction, and maintenance of neuronal plasticity, and discrimination between physiological states of animals (Uzakov et al., 2005; Korz and Frey, 2007; Schulz and Korz, 2010; Meyer and Korz, 2013). The apparatus consists of a black plastic board (1 m × 1 m) with 16 regularily arranged holes (7 cm in diameter and 7 cm in depth). The board is surrounded by translucent plexiglass walls (30 cm in height). Outside at each wall black and white figures (A4) were mounted as proximal cues (Meyer and Korz, 2013). Distal cues consisted of further figures (A3) at room walls, furniture and equipment. Below the board a second board was placed densely scattered with the reward food pellets (dustless precision pellets of 45 mg; Bioserv®, Somerville, NJ, USA) to avoid possible olfactory orientation. During training four holes were baited. Trials were monitored via a video camera and stored on a computer. During the test procedure the animals were food deprived to reach 85% of their initial body weight (which is the widely used procedure) and stayed in the experimental room throughout the whole testing period. The experimental procedure was as follows: 3 days of weighing and handling (15 min) with providing 15–20 reward pellets at day 3 in the housing cage in order to familiarize the animals with the pellets. Two days of habituation (15 min) during which the animals were allowed to explore the maze with reward pellets in each hole and on the surface of the board. Three days of training and testing with 5 trials at day 1, four trials at day 2, and one retention trial at day 3. The trials ended after 2 min or when rats have found all 4 pellets. After each trial the board was cleaned with 0.1% incidin® solution to remove olfactory marks, feces, and urine. Counted were the reference memory errors (visits of unbaited holes) and visits and revisits of baited holes. A reference memory index (RMI) was calculated as: total visits of baited holes/total visits of all holes; thus a value of 1 indicates a good reference memory without errors) and the values for trial 1, 6, and 10 (HB-trial 1, 6, 10; respectively) were considered for the factor analysis.

Each variable was separately analyzed for minimum and maximum values, median, 25 and 75% percentiles, arithmetic mean, standard deviation, and standard error, the lower and the upper 95% confidence interval. Correlations between the measures of each test were detected by the Pearson product moment correlation (two-tailed). In order to reduce the number of dimensions a Factor analysis (principal component with Varimax rotation and Kaiser normalization, maximally 25 iterations to convergence) based on the correlation matrix was perfomed for all variables. Factor rotations simplify the interpretation of the factors. The coordinate axes rotation allows a better distribution of the loadings on the factors. The varimax rotation has been choosen because it is a orthogonal rotation method so that the different factors do not intercorrelate, thus each factor represents an independent behavioral pattern. Varimax rotation reduces the number of variables with high multiple factor loadings. Extracted components with eigenvalues higher than one were considered for further interpretation. In order to separate the animals according to their individual factor loadings (−1 to 1) on the components, loadings lower than −0.7 and higher than 0.7 were considered as border. Factor and correlation analyses have been done with SPSS Version 20, all other analyses by using GraphPad Prism (Version 5.02).

Almost all behavioral variables were not Gaussian-distributed (Shapiro-Wilk normality test): HB-trial 1, W = 0.82, p < 0.001; HB-trial 6, W = 0.67, p < 0.001; HB-trial 10, W = 0.70, p < 0.001; OFRDR, W = 0.86, p < 0.001; OFveloc, W = 0.81, p < 0.001; RR-time, W = 0.97, p < 0.001; FSITR, W = 0.92, p < 0.001; FSIT, W = 0.56, p < 0.001; EPMRCO, W = 0.33, p < 0.001; EPMlocal, W = 0.81, p < 0.001; EPMRDOC, W = 0.61, p < 0.001. Normally distributed are the variables OFlocal, W = 0.98, p = 0.066 and EPMOE, W = 0.98, p = 0.071. Minimum and maximum values, percentiles, arithmetic means, standard deviation and error as well as confidence intervals are given in Supplementary Table 1. The different numbers of values results from some missing values for the different tests. The relative distribution of the raw data are given in Supplementary Figure 1.

The results of the linear regression analysis by using the Pearson coefficient and the resulting p-values are given in Table 1. All of the open field variables are positively correlated with those of the late (trial 6 and trial 10) holeboard memory indices. Local movements in the open field are also positively correlated with the RMI of holeboard trial 1. EPMlocal positively correlates with FSIMT and there is a larger number of correlations of the number of open arm entries, positively with HBtrial6 and with movement parameters in the open field. Different measures within the tests were necessarily correlated and given here for reasons of completeness.

The main results of the principal component analysis are given in Table 2. Considered were only components with an eigenvalue higher than 1. The first three dimensions and the euclidean distances of the variables are given in Figure 1. The first component explains 22.3% of the total variance (after rotation) and is positively highly positively loaded by the open-field variables OFRDR, OFlocal, and OFveloc. All of these variables general acticity this component was considered as representing the trait behavioral pattern to be active. The second component, explaining 16.8% of the variance is loaded by the holeboard reference memory indices and therefore considered as the spatial cognition component. The third component explaining 16.7% represents anxiety as it is characterized by EPM measures EPMROC and EPMRDCO the classic test for anxiety-like behavior. The number of entries in open arms is negatively loaded on this component, however does not reach the criterion of 0.7 factor score. Component 4 reflects depression-like behavior since the results of the test for depression, the forced swim immobility during test loads positively on these factor. Interestingly, also the immobility time during training is highly loaded it does not reach the criterion, although this variable is highly correlated with FSIMT. For that reason a high positive loading of individal rat scores on this component reveals high depression-like behavior. The rotarod measure is highly loaded on factor 5 which is therefore classified as reflecting motor abilities. Factor scores of −0.7 and 0.7 are considered as significant, wheras all values between these borders are considered as intermediate, thus may represent “normal” behavioral patterns. Because the animals are naive and not treated (except the tests), these behaviors are considered as trait patterns (Crusio, 2001). Similar high positive (>0.7) or negative (< −0.7) loadings of individual rat scores on component 1 reveal high vs. low general activity, with the rest representing 'normal' activity within the population. By doing this also for the other components, a pattern of the distribution of spatial cognitive abilities, activity, anxiety-, and depression-like behavior and motor abilities for this population of rats can be calculated and is given in Figure 2. Thus, 23.6% show a high, 18.9% a low, and 57.4% a intermediate activity. The largest “low” (39.9%) and the highest “high” portions are given for the factor spatial cognition and the largest “intermediate” portion (87.8%) for the intrinsic anxiety. Activity, depression and motor ability are close together with regard to their distributions over the three categories, whereas spatial cognition and anxiety differ. Because these results are based on a large cohort of rats we can assume that this is a characteric feature for male Sprague-Dawley rats of this age and that an experimenter recruiting experimental samples can expect these probabilities of indiviudal behavior in experimental samples.

In order to reveal different high or low factor loadings in individual rats we counted double appearances in the different high and low portions of the factors. This overlaps (in %) are indicated in Figure 3. High percent values of individual rats are given for the factors activity and spatial cognition (ac-sc) and depression and motor abilitiy (d-ma) with little difference between high and low performers. The combinations ac-ma, sc-d, and a-d show the highest differences between high and low performers. There is no overlap at all between anxiety and depression in high performers in contrast to low performers, whereas a big overlap between spatial cognition and depression appears in the first.

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.