Impact of Enrichment of Metabolic Cages with Ball-Shaped Shelters in Studies of Protein Metabolism in Rats (Rattus norvegicus)
Housing in metabolic cages for quantitative collection of urine and feces is necessary for nitrogen (N) metabolism studies but may have negative consequences for rat welfare. We hypothesized that providing shelters in metabolic cages would affect the rats’ behavior and reduce the precision of measuring N metabolism. Forty-eight growing male Sprague–Dawley rats were housed in metabolic cages for 9 d, constituting 4 d of adaptation and 5 d for quantitative measurement of feed intake and urine and feces excretion. Using a 2-factorial approach, half of the rats were fed a soybean meal diet, and the other half were fed a diet based on green protein (GP). Half of the rats in each dietary group were provided a ball-shaped shelter in their metabolic cage, and the other half had no shelter (n = 12 per treatment). Video recordings of rat location (cage floor, feed tube, or shelter) were performed on days −3 (adaptation period), 1, and 4 (collection period). Dry matter (DM) and nitrogen (N) digestibility were determined based on DM and N intake and fecal excretion. N metabolism was calculated based on additional measurements of urinary N excretion. The rats used the shelter primarily during the daytime and spent less time in the feed tube during the day than during the night (P = 0.0002). For rats without shelter, there was no difference between day and night in their presence in the feed tube (P = 0.94). The shelters did not interfere with measurements of feed intake, fecal DM, N excretion, or any of the derived N-metabolism-associated responses. There was also no significant effect on the variation associated with the estimated values, and no differences in contamination were detected visually or by rinsing the shelters. We conclude that using ball-shaped shelters to enrich metabolic cages is not a major risk of biocontamination and may improve the rats’ welfare.
Introduction
Metabolic cage housing of laboratory rats is necessary for a wide range of nutritional, metabolic, pharmacokinetic, and pharmacodynamic studies in nutritional and biomedical research.12 This type of housing allows accurate measurements of feed and water intake and quantitative urinary and fecal output. However, the reduced space allowance, grid flooring, absence of bedding and enrichment, and social isolation may impose behavioral and physiologic responses indicative of reduced animal welfare.15,18,25 Previous studies confirmed that rats prefer a cage with a shelter; however, in metabolism studies, such enrichment is challenging due to the risk of interference with the collection of feces and urine.19 Cages cannot include surfaces for male rats to mark with urine, or feces to stick on,19 and the design must not lead to biocontamination due to changes in eating, or defecation and/or urination habits. Furthermore, in nitrogen (N) metabolism studies, an increased risk of N loss via ammonia emission due to microbial urease activity may be present if urea in the collected urine is not trapped and acidified.4
Several studies have investigated the effect of metabolic cage housing of rodents on their behavioral and physiologic conditions.6,8,11,18,19,25,28 In general, the results have shown changed feed intake and feeding behavior, reduced weight gain, and changes in some physiologic indicators, but varying hormonal responses related to the hypothalamic-pituitary-adrenal axis depending on the age of the animals.8 Prototypes of enrichment devices have been tested, demonstrating that enrichment to some extent can improve the welfare of the animals.19 Apart from measures of mouse behavior and energy expenditure in cages modified to alleviate cold stress, improvement of flooring by modification of the wire grid26 or by use of platforms27 to our knowledge, there are no published studies demonstrating that shelter enrichment can be employed without interference with the quality of the quantitative data. Such knowledge is important to improveme the welfare of animals used in nutritional and biomedical studies when metabolic cages are used.
The current study aimed to investigate the effect of enrichment of metabolic cages with ball-shaped shelters in N-metabolism studies using growing Sprague–Dawley rats that were fed 2 different protein sources. The study included evaluations of the rats’ behavioral responses, effect on variability, and detectable differences in quantitative measures of N digestibility and N metabolism.
Materials and Methods
Complying with the Danish laws and regulations regarding the care and use of animals in research (Ministry of Environment and Food of Denmark, Act on Animal Experiments no 474 of May 15, 2014, as stipulated in the executive order number 12 on January 07, 2016), the animal experiment was performed according to a license obtained from the Danish Animal Experimentation Inspectorate, Ministry of Food, Agriculture, and Fisheries.
Forty-eight 4-wk old male standard murine pathogen-free21 Sprague–Dawley rats (NTac:SD; Taconic, Laven, Denmark) in 2 blocks of 24 rats with an initial body weight of 70.5 g (SD 3.85 g, SEM 0.55 g) at arrival were acclimatized to the housing facility for 5 d. During acclimatization, the rats were kept in cages (50 × 35 × 28 cm) of 6 rats and in a room with a 12:12-h artificial light cycle, a room temperature of 25 °C, and relative humidity of 60%. During this period, the rats were provided with Altromin AIN-93G powder diet (Brogaarden, Lynge, Denmark) for ad libitum intake and had ad libitum access to water. The cages were fitted with a 3 cm layer of Tapvei 4H Aspen bedding (Brogaarden, Lynge, Denmark) and a flat-roofed aluminum shelter (20 × 20 × 10 cm).
During the following 9-d experimental period, the animals were divided into 4 treatment groups (n = 12 per treatment) and housed individually in metabolic cages with grid floor (320 cm2 and diameter of 20.2 cm) for rats <150 g (3700M061; Tecniplast, Varese, Italy) with ad libitum access to water. The rats were allocated to treatment according to their body weight to ensure that all treatment groups had the same average start weight and with as uniform variation between groups as possible. Twenty-four cages were distributed on 4 racks with 3 rows of 4 cage holders so that 2 racks each held 4 metabolic cages at the top and the bottom row, and 2 racks had 4 cages in the middle row (Figure 1). Half of the cages were enriched with a polycarbonate ball-shaped shelter (Crawl Ball™; diameter of 10.2 cm and 3 openings of 5.8 cm in diameter; Datesand, Manchester, UK) (Figure 2). Half of the rats were fed a diet based on Chinese organic soybean meal (SBM), while the other half received a diet in which the protein source was organic grass-protein concentrate (GP) as described by a previous study.20 All 4 treatments were represented in each row. Because we expected higher within-group variation due to the presence of the shelter and could, therefore, detect smaller differences between treatment groups than in our conventional N-balance trials, we increased the number of animals 3 times compared with the recommended 4 rats for digestibility trials7 and twice as many as used in our standard procedure for N-balance studies.5
Citation: Journal of the American Association for Laboratory Animal Science 64, 2; 10.30802/AALAS-JAALAS-24-000018
Citation: Journal of the American Association for Laboratory Animal Science 64, 2; 10.30802/AALAS-JAALAS-24-000018
Each rat received 12 g DM in their feed trough daily between 0730 and 0830. Dietary N was provided by SBM or GP supplemented with 0.55% dl-methionine (M9500; Sigma-Aldrich, St. Louis, MO) and 0.2% l-cystine (C8755; Sigma-Aldrich) to ensure sufficient sulfur-containing amino acids (Table 1). Different amounts of protein sources were added to each diet to ensure that the N contents were equal and corresponded to a calculated intake of 180 mg N/day. Furthermore, the diets were supplemented with mineral Altromin AIN-93G (3.5% of DM) and vitamin AIN-93 mix (1.0% of DM; Brogaarden, Lynge, Denmark). The remaining constituent was an N-free mixture consisting of pregelatinized maize starch (Cargill, Charlottenlund, Denmark), sucrose (Nordic Sugar, Copenhagen, Denmark), microcrystalline cellulose (Vivapur 101; JRS Pharma, Rosenberg, Germany), and rapeseed oil (COOP, Albertslund, Denmark).
| SBM (%) | GP (%) | |
|---|---|---|
| SBM | 21.50 | |
| GP | 19.78 | |
| Vitamin mix AIN-93 | 0.96 | 0.98 |
| Mineral mix AIN-93G | 3.38 | 3.42 |
| Choline bitartrate | 0.19 | 0.20 |
| dl-methionine | 0.5304 | 0.5377 |
| l-cystine | 0.1929 | 0.1955 |
| N-free mixturea | 73.25 | 74.89 |
GP, green protein diet; SBM, soybean meal diet.
Four days were allowed for adaptation of the animals to the metabolic cages and the experimental diets. During this period, feed was provided every morning, while all feces and urine were collected in disposable aluminum dishes (Abena Cater-Line, Aabenraa, Denmark) placed under the cage floor and lined with cottonwool wetted with 0.20 mL citric acid (3.12 N, pH 1.56). In the following 5-d data collection period, the cages were fitted with a collector unit consisting of a separating funnel and 2 tubes for separate collection of urine and feces, which were emptied daily in connection with feeding. Both urine and feces were pooled per rat over the 5 d and stored frozen until analysis. To prevent evaporation of ammonia and contamination with diet or feces particles, the urine tubes were equipped with a small polypropylene funnel (funnel: outer diameter 35 mm/stem: 6.3 mm; VWR International, Søborg, Denmark) covered with 4-layered 30 × 30-mm cotton gauze swaps and 1 mL of 3.12 N citric acid was added every day when collecting 24-h urine was collected.
At the end of the data collection period, cages and ball-shaped shelters were visually inspected for contamination, and the shelters were rinsed with 150 mL water for the determination of N in the rinses.
In calculations of feed intake, feed refusals were subtracted. Before analysis, feed particles contaminating feces were manually removed, and the net weight of feces was calculated. Out of 46 animals, 31 (67.4%) had no visual contamination of feces with feed particles, while 7 (15.2%) had residues corresponding to 0% to 1% of allocated feed, and 8 rats (17.4%) had feed contaminations in the range 1.4% to 5.1%. The feed contaminants were not subtracted from net feed intake due to uncertainties in the origin and salivary content of the removed particles. Two rats (one fed SBM with shelter, and one fed GP without shelter) were taken out of the experiment and euthanized at the end of the adaption period due to weight losses >20% (humane endpoint). The remainder 46 rats all showed weight gain in the adaptation period. After the conclusion of the experiment, groups of 5 rats were killed with carbon dioxide by gradual displacement to a CO2 level of >70% in a closed chamber until respiration had ceased and death ensued. Following, the rats were kept in a −20 °C freezer until disposal.
Video recordings of rat location.
For rows containing cages, 2 video cameras (Hikvision Network Camera model DS-2CD2055FWD-I 2.8 mm; Hikvision Digital Technology, Hangzhou, China) were fixed to the rack above the row to record from 2 cages each (Figure 1). The cameras were connected to a computer for data storage. The computer was placed outside the housing room to eliminate noise.
The rat location (shelter, cage floor, and feed tube) was sampled from video recordings by scan sampling1 every 5 min for periods of 30 min (7 times) starting at 1000, 1100, 1200, 1300, and 1400 (daytime) and at 2200, 2300, 2400, 0100, and 0200 (nighttime) on days −3 (adaptation period), 1 and 4 (start and end of data collection period) using the S-Vidia Client (ver. 6.0.12.212; Tacoma, WA). Location was defined as min. 80 % of the body was located in the position. This approach resulted in a total of 35 scans per day and nighttime, respectively, or 70 scans per observation day. One trained observer did all video analyses. Since the presence of a shelter in the cage was easily observable, it was impossible to blind the observer.
Chemical analyses.
The dry matter (DM) content of feces samples was determined by freeze drying. The DM contents of ingredients, diets, and feces after equilibration to air relative humidity were determined in duplicates by drying the samples at 103 °C to constant weight. N in diet and feces were analyzed by the Dumas procedure using thermal conductivity after complete combustion at 1300 °C in pure oxygen (LECO CNS-2000, Carbon, Nitrogen and Sulphur Analyzer, St. Joseph, MI).9 N content was converted into CP content by multiplication with 6.25. N in urine and rinses was analyzed by the modified Kjeldahl method (Method 984.13; AOAC, 2000) using a Kjeltec™ 2400 (Foss, Hillerød, Denmark).
Data handling.
Based on net feed intake and fecal excretion, the DM and N digestibility was calculated according to a previous study.10 The DM digestibility of the N-free mixture and the fecal excretion of N after ingestion of the N-free diet (endogenous N loss) were estimated previously10 and used for calculations of the digestibility of the ingredients SBM and GP.
The feed conversion ratio (FCR) was calculated as DM intake divided by weight gain in the collection period.
The digestibility of DM (ADDM), where only the DM contribution from SBM or GP was included, was calculated as follows:where ADDMN-free mixture is the digestibility of the N-free mixture estimated to be 92.8% based on a large set of rat digestibility trials (unpublished data) and a previous study.10
The apparent fecal digestibility of N (ADN) was calculated as follows:
Furthermore, the apparent N digestibility (ADNaa corrected) of the ingredients (SBM or GP) was calculated by subtracting the N contribution from the supplemented amino acids to the diets (aa corrected) assuming that these were fully digested.
The endogenous fecal N loss (FEN) in grams was calculated using a correction of 1.01 mg N/g diet DM intake as estimated by a previous study10 in a rat feeding trial with an N-free mixture:
The standardized fecal N digestibility (SDN) was then calculated to be corrected for FEN:
SDN was calculated for both total N intake (SDN) and N intake corrected for added amino acids (SDNaa corrected) as described above.
Retained N (in g/d) was calculated as N intake minus N excreted in feces and urine and in percentage of intake as follows:
Biologic value (BV) was calculated as follows:where Nendogenous is the endogenous loss of N in urine using a value of 15.2 mg/d as previously determined for growing rats.
Statistical analysis.
Scans from video recordings of rat location.
Data from the video recordings were visually checked for normality. Due to a nonnormal distribution, the number of observations of each rat being present in a specific area was divided by the total number of observations for an observation day (n = 70) and daytime (n = 35) and nighttime (n = 35) separately.
We first present the mean values of ratios ± SE at a specific location (cage floor, feed tube, or shelter). We then tested if there was a difference between the 2 housing systems in the use of the feed tube specifically, using a linear mixed model (SAS 9.4; SAS Institute, Cary, NC) with a first-level autoregressive covariance structure:Yijml = μ + αi + βj + δm + (βδ)jm + (δρ)ml + (βδρ)jml + ωl + εijml,where Yijml is the dependent variable (Ratio); μ is the overall mean; αi is diet (i = SBM and GP); βj is shelter (j = YES or NO); δm is time of day (m = Day and Night); ρl is observation day (l = −3, 1 and 4); (βδ)jm is the interaction between shelter and time of day; (δρ)ml is the interaction between time of day and observation day; (βδρ)jml is the interaction between shelter, time of day, and observation day; ωl is the subject of the repeated effect of rat; and εijml is the normal distributed residual error.
Rat performance and nutritional responses.
For the statistical analysis, measures of feed intake, body weight and weight gain, fecal and urinary excretion, digestibility, and N retention, the rat was considered the experimental unit.
Initially, we investigated if there was a difference between SEMs associated with the provision of shelter (overall across diets No Shelter compared with Shelter) or within each diet (SBM No Shelter compared with SBM Shelter and GP No Shelter compared with GP Shelter) by folded F tests in SAS 9.4 for Windows. Differences in SEM caused by having access to the shelter were nonsignificant and there was no consistent trend in the size of SEM.
Data analysis was consequently accomplished using the MIXED procedure of SAS based on the normal mixed model:Yijk = μ + αi + βj + (αβ)ij + tk + εijk,where Yijk is the dependent variable; μ is the overall mean; αi and βj are the fixed effects of diet (i = SBM or GP) and shelter (j = YES or NO); (αβ)ij is the interaction among fixed effects; tk is the random effect of block (k = 1 or 2); and εijk is the normal distributed residual error. If an interaction was detected, a pairwise comparison of groups adjusted for multiple comparisons was conducted using the Tukey–Kramer post hoc test. Values are presented as least square means with standard error of the means (SEM).
For DM intake in the adaptation and collection period, N intake, daily gain in the adaptation period, concentrations of N in feces and urine, and retained N in grams per day and percentage of intake, there was a statistically significant effect of dietary treatment in the folded F tests of variance. Hence, for these responses, we also ran the model including the Satterthwaite approximation to account for inequality of variance between treatment groups. None of these led to changes in significance level compared with the model assuming equal variance except for the N intake, so the model assuming equal variance is presented for simplicity and clarity of presentation, and P values for the model with unequal variance for N intake are presented as footnote.
Results
Rat location: video recordings.
The display of ratios of location at each observation point (Figure 3) shows that for rats without shelter, no diurnal variation in location was found (Figure 3A), while for the rats with access to a shelter, there was a shift in time spent at the different locations between day and night with use of the shelter mainly during daytime at the expense of time used on the cage floor and in the feed tube (Figure 3B). To exclude the interdependence between locations, we tested the effect on the ratio of observations in the feed tube of rats with and without shelter during night- and daytime on the 3 observation days (Figure 4). Across observation days and time of day, rats with access to a shelter were less likely to spend time in the feed tube (proportion of scans: 0.19 ± 0.020) than the rats without shelter (0.27 ± 0.020, P = 0.0041). Across cages with and without shelter, the rats were also more likely to be in the feed tube during the nighttime (proportion of scans: 0.26 ± 0.018) than during the daytime (proportion of scans: 0.19 ± 0.018, P = 0.0062). However, there was an interaction between access to shelter and time of day (P = 0.0006). While no effect of shelter was found regarding the proportion of scans spent in the feed tube during the nighttime (P = 1.000), rats with access to a shelter were less likely to be in the feed tube during the daytime (proportion of scans: 0.11 ± 0.025) than rats without shelter (proportion of scans: 0.28 ± 0.025, P = 0.0002). Consequently, no significant difference was found between day and night in the presence of the feed tube for the rats without shelter (P = 0.94), while rats with shelter spent significantly less time in the feed tube during the day than during nighttime (P = 0.0002).
Citation: Journal of the American Association for Laboratory Animal Science 64, 2; 10.30802/AALAS-JAALAS-24-000018
Citation: Journal of the American Association for Laboratory Animal Science 64, 2; 10.30802/AALAS-JAALAS-24-000018
There was a significant correlation between observation day and time of day (P = 0.0002, Figure 4), with the feed tube being used less during night time on d 4 than on days −3 (P = 0.0009) and 1 (P = 0.0060) for rats both with and without access to shelter, while no significant effect of observation day was seen during daytime (P > 0.74). Finally, we observed, that rats fed GP had a higher ratio of presence in the feed tube (proportion of scans 0.26 ± 0.020) compared with the rats fed SBM (0.20 ± 0.020, P = 0.029).
Effect of shelter on variation in metabolic responses.
No significant differences in group SEMs between rats with or without shelter were found in the folded F test neither across diet (n = 24) nor within each diet (GP or SBM; n = 12), and there was no consistent trend toward larger variation associated with the use of a shelter in the metabolism cage.
Feed intake, body weight, daily gain, and feed conversion.
There was no effect of the shelter on BW, DM intake, daily gain, feed FCR, or any interactions with diet (P > 0.12; Table 2). Body weight did not differ between the 4 groups when starting the experiment (P > 0.87), but after the 4-d adaption period, the rats fed SBM were 6.6% heavier than the rats fed GP (P = 0.0002) corresponding to a 2.6 times higher daily gain in the former (P < 0.0001). The difference coincided with a lower DM intake in the GP group (P < 0.0001). However, despite identical DM intakes in the data collection period, the weight difference between the 2 dietary groups increased to 8.2% (P < 0.0001), and daily gain was 31% higher in the SBM group than in the GP group in the data collection period (P = 0.0009).
| Time | SBM | GP | SEM | Pr > Fc | |||
|---|---|---|---|---|---|---|---|
| Diet | Shelter | Diet × shelter | |||||
| BW (g) | Start adaptation | 110.2 | 110.0 | 6.6 | 0.923 | 0.964 | 0.868 |
| BW (g) | Start collection | 120.6a | 112.9b | 5.7 | 0.0002 | 0.403 | 0.797 |
| BW (g) | End collection | 131.5a | 121.2b | 3.5 | <0.0001 | 0.413 | 0.720 |
| Daily gain (g/d) | Adaptation | 2.62a | 0.73b | 0.27 | <0.0001 | 0.119 | 0.429 |
| Daily gain (g/d) | Collection | 2.18a | 1.66b | 0.45 | 0.0009 | 0.795 | 0.140 |
| DM intake (g/d) | Adaptation | 11.53a | 9.81b | 0.30 | <0.0001 | 0.818 | 0.770 |
| DM intake (g/d) | Collection | 11.83 | 11.76 | 0.15 | 0.436 | 0.125 | 0.673 |
| N intake (g/d) | Collection | 0.194a | 0.189b | 0.003 | 0.028 | 0.393 | 0.805 |
| N intakeaa corrected (g/d)d,e | Collection | 0.184a | 0.180b | 0.002 | 0.028 | 0.393 | 0.805 |
| FCR | Collection | 6.1b | 10.7a | 3.5 | 0.032 | 0.998 | 0.622 |
BW, body weight; DM, dry matter; FCR, feed conversion ratio (DM intake/daily gain in collection period); GP, green protein diet; N, nitrogen; SBM, soybean meal diet.
Based on dietary N analyses, the daily N intake in the data collection period was 2.6% higher (P = 0.028) in the SBM than in the GP group (2.2% when corrected for the addition of amino acids).
Fecal output, fecal DM, and N digestibility.
No significant effects of shelter enrichment on fecal characteristics, fecal excretion of DM and N, and fecal digestibility of DM and N were found (P > 0.071; Table 3).
| SBM | GP | SEM | P value | |||
|---|---|---|---|---|---|---|
| Diet | Shelter | Diet × shelter | ||||
| DM content (%) | 47.2b | 61.6a | 2.39 | <0.0001 | 0.416 | 0.838 |
| DM excretion (g/d) | 1.15b | 1.86a | 0.050 | <0.0001 | 0.384 | 0.525 |
| N content (% DM) | 2.75b | 4.00a | 0.038 | <0.0001 | 0.282 | 0.571 |
| N excretion (g/d) | 0.032b | 0.074a | 0.0022 | <0.0001 | 0.071 | 0.154 |
| ADDMdiet (%) | 90.3a | 84.1b | 0.24 | <0.0001 | 0.966 | 0.768 |
| ADDMingredient (%) | 80.7a | 48.9b | 1.20 | <0.0001 | 0.964 | 0.767 |
| ADN (%) | 83.7a | 60.6b | 0.50 | <0.0001 | 0.130 | 0.230 |
| ADNaa corrected (%) | 82.8a | 58.6b | 0.52 | <0.0001 | 0.130 | 0.230 |
| SDN (%) | 89.9a | 66.9b | 0.49 | <0.0001 | 0.121 | 0.206 |
| SDNaa corrected (%) | 89.4a | 65.2b | 0.52 | <0.0001 | 0.121 | 0.206 |
ADDM, apparent dry matter digestibility; ADN, apparent nitrogen digestibility; DM, dry matter; GP, green protein diet; N, nitrogen; SBM, soybean meal diet; SDN, standardized nitrogen digestibility.
In contrast, the diet had a large effect on these responses. Rats fed SBM had softer feces, lower N concentration in feces, and lower daily fecal DM and N excretion than the rats fed GP (P < 0.0001). The apparent DM digestibility of the whole diet (ADDMdiet) was significantly lower for GP than for SBM (P > 0.0001), which was due to a very low apparent DM digestibility of the ingredient (ADDMingredient) compared with SBM (P < 0.0001) based on estimates of the digestibility of the N-free ingredients of the diet (Table 3). The apparent and standardized fecal N digestibility of the diet was 27.6 and 25.6% lower (P < 0.0001) for GP compared with SBM, respectively (Table 3). Assuming 100% digestibility of the supplemented amino acids, ADNaa corrected and SDNaa corrected were reduced equivalently for GP compared with SBM by 29.2 and 27.0%, respectively.
N retention.
The diet significantly affected both urinary N concentration and output, retained N in milligrams per day and in percentage of N intake, and net protein utilization (Table 4).
| SBM | GP | SEM | P values | |||
|---|---|---|---|---|---|---|
| Diet | Shelter | Diet × shelter | ||||
| N concentration in urine (%) | 0.73a | 0.38b | 1.28 | <0.0001 | 0.545 | 0.952 |
| N excretion (mg/d)c | 43.4a | 20.5b | 2.7 | <0.0001 | 0.291 | 0.017 |
| Retained N (mg/d) | 119a | 94b | 2.8 | <0.0001 | 0.671 | 0.349 |
| Retained N (% of intake) | 61.2a | 49.8b | 1.9 | <0.0001 | 0.202 | 0.243 |
| Retained N (% of digested)c | 73.1b | 82.2a | 2.1 | <0.0001 | 0.553 | 0.040 |
| Biologic valuec | 84.0b | 95.8a | 1.9 | <0.0001 | 0.576 | 0.028 |
| Net protein utilization (%) | 63.6b | 75.7a | 2.06 | <0.0001 | 0.163 | 0.217 |
GP led to a 48% reduction in the concentration of N in urine compared with SBM (P < 0.0001) along with a lower amount of N retained in milligrams per day and expressed as % of N intake (Table 4). Despite a higher proportion of digested N being retained corresponding to higher BV, GP also resulted in a significantly lower (P > 0.0001) net protein utilization than SBM (Table 4).
The statistical model showed a significant diet × shelter interaction for daily urinary N excretion (P = 0.017), retained N in percentage of digested N (P = 0.040), and BV (P = 0.028), due to minor differences between rats fed SBM with and without access to shelter, which in the Tukey-adjusted post hoc analysis, however, was not significant for either daily urinary N excretion (P = 0.073), retained N in percentage of digested N (P = 0.234), or BV (P = 0.199). For the N concentration in urine and retained N in milligrams per day and in percentage of N intake, there was no significant effect of shelter (P > 0.16).
Visual inspection and N contamination of the shelter.
Visually, there were no signs of urination or defecation in or on the shelter. For all shelters, the N contents of the rinses were below the detection limit.
In block 1, 2 rats with a shelter in the cage, and one rat without a shelter showed visual signs of urination in the feed, and for one rat there was a minor spillage of urine at the bottom of the metabolic cage. In block 1, a few feed particles were identified in rinses from one rat. In block 2, one feed particle was identified on one shelter, and in rinses from 4 shelters without giving rise to sufficient high N contamination of the rinses to exceed the detection limit. Gnaw marks were seen on one shelter in block 2.
Discussion
Across rodent studies involving metabolic cages, the main argument for not providing a shelter or other enrichments suggested to improve animal welfare is that the items may interfere with the quantitative collection of feces and urine and thereby possibly invalidate or at least make the estimates of digestible and/or metabolizable components less precise.19
Despite a higher number of observations compared with common practice in N-metabolism studies (n = 12 compared with the more typical n = 5), the current study showed no significant effect on either group variations or group least significant means between rats with or without a shelter in the metabolic cage on the central indicators of feed intake, digestibility, metabolism, growth, or feed conversion ratio. Visual inspection and analysis of the rinses of the shelters did not point toward any major contamination interfering with the quantitative measures used in this type of metabolic study. We saw no effect on feed intake or feces volume, neither in the adaptation period nor the following data collection period between rats with and without shelter. This is in contrast to a study19 that found that rats provided with an enrichment device ate less, drank less, and defecated less than rats without enrichment, and both groups had an immediate drop in feed and water intake on the first day in the metabolism chambers. Another study23 previously showed that rats provided with environmental enrichment in the form of shelters in their cages exhibited more exploratory behavior and when given a choice, preferred cages with shelters as compared with those without. Similarly, the previous study19 concluded that a box-shaped prototype enrichment device positioned between the upper and lower part of a standard metabolic cage similar to the ones used in our study, to some extent improved the welfare of rats albeit housing in a metabolic cage, irrespective of the improvement, still was considered stressful.
Our study showed that the rats used the shelters primarily during daytime. This is concordant with the diurnal rhythm of rats which, as nocturnal animals, are most active in the early dark phase2 and with a higher fecal and urinary excretion during nighttime than during daytime.6 However, the results regarding the effects of enrichment and housing on rat behavior are not univocal, which may relate to differences in experimental settings (type of housing, type of enrichment, methods, and timing of behavioral observations, etc.). One study25 found that metabolic cage housing had an anxiogenic effect with a shift toward inactivity compared with individual housing in standard open-top cages. In a home-cage test, other investigators23 found that single-housed Wistar rats that were provided a shelter spent more time exploring the shelter and surroundings, and less time eating/drinking, rearing, or walking than rats without a shelter, whereas the time spent resting did not differ (mostly using the shelter, if they had one). Consistent with this, an additional study2 found that particular tubing that could be used as shelter reduced activity, while both chew stick and tubing independently reduced the frequency of rearing and alleviated the stress-induced reduction in fecal IgA. Reports on the effect of isolation are conflicting, although most studies show that it leads to hyperactivity.24 Of note, measurable physiologic responses to isolation stress such as urinary excretion of norepinephrine and monoamine oxidase inhibitory activity are greater in young than older rats, which together with behavioral tests (open-field and hot-plate tests) have shown that isolation is more aversive for young rats.8
Interestingly, we did not see any difference between day and nighttime with respect to the prevalence of rats without a shelter locating in the feed tube. This contrasts with the rats that had access to shelter, as they used the feed tube more during nighttime than daytime. This might indicate that the shelter-less rats not only used the feed tube for eating but also as a place for hiding and/or sleeping. This, again, points toward the rats’ preference for having access to a shelter.19
In the current study, we included 2 diets expected to have different palatability based on previous observations, and, thus, potentially could have induced differences in the degree of contamination of the cage and the shelter. We did observe that the rats fed GP had a higher ratio of observations in the feed tube than the rats fed SBM, which might indicate that these rats spent more time consuming the allocated feed. However, we did not observe any difference in spillage (feed particles removed from feces during the collection period) or any consistent difference between the diets, neither with nor without the shelter, in the SEM associated with the estimates. This was supported by the absence of visible signs of urination or defecation in or on the shelters and that the N contents of the rinses of shelters were below the detection limit. This supports our finding that using the ball-shaped shelter is not a major source of error when using meal feeding in metabolic cages. We thus consider observed differences in digestibility and metabolism of protein in the 2 dietary protein sources as a true reflection of the quality of the protein versus being the result of either housing conditions or differences in palatability. Hence, the results also indicate that ball-shaped shelters are superior to other modifications such as platforms, since the latter interferes with quantitative collections of urine.27 However, it should be noted that the 2 types of enrichment were installed to serve different purposes. While the platforms in a previous study27 were installed to provide a solid surface, the ball-shaped shelters used in our study did not provide solid flooring due to the 3 large openings, unless the rats were able to turn the solid, but curved, sides of the shelters downward.
Irrespective of the use of shelters in the cage, feed spillage in the feces and possible contamination of urine are of concern when using metabolic cages for rats. In 15 out of 46 fecal samples, we had to remove feed particles (including saliva) ranging from 0.5% to 5.7% of ‘consumed’ feed, measured as the provided amount minus feed residues, and in most of the cages, feed spillage was observed on the sides of the funnel separating feces and urine. The main reason for this is the construction of the feeding chamber, which encourages rodents to nest (mice) or hide (rats) there and allows them to turn around or bring back feed to the cage area. The spillage reservoir behind the feeding trough does not sufficiently account for this. Occasionally, we observed rats turning backward, having their rear end and tail in or above the feed trough. In a whitepaper, The cage manufacturer addressed the problem of spillage for mice by redesigning the cage (not the feeding unit) in one experiment and fitting the standard cage with a cone-shaped enrichment device to prevent the mice from nesting in the feeder chamber.22 Our previous locally made metabolic cages used until 2011, which were fitted with a clear tubular acrylic tunnel with an open bottom and inner lining of metal wire,5 led to less waste and did not allow for turning around in the chamber. A redesign of the feeding chamber might be a solution to solve this issue.
Housing in metabolic cages and use of enrichment may have both anorectic and orexigenic effects2,19 but unwanted anorectic effects of the diet are also important to consider. Besides aversion to taste, structure, or other characteristics of the diet, rats are known to recognize amino acid imbalance and reject diets deficient in essential amino acids.13 Therefore, we added the sulfur-containing amino acids dl-methionine and l-cystine to the diets, as both protein sources (SBM and GP) are deficient in those, and if not, we could expect weight loss instead of weight gain along with invalid measures of digestibility and biologic value of the protein sources. Accounting for this, and assuming that the added crystalline amino acids were fully absorbable, it was possible to compare the nutritional profile of the protein sources, showing that the quality of SBM was superior to the green protein.
Limitations of the study.
Had the rats been habituated to the shelters before entering the metabolic cage, they might have had a higher frequency and duration of use during their stay in the metabolic cages. As the study was designed, the shelters were unfamiliar to the rats along with the unfamiliar environment of the metabolic cage. In laboratory mice, latency to access an enrichment (playpen) decreased over time.17 In our study, we also saw that the presence in the shelter increased along the course of the study from a proportion of scans of 0.23 ± 0.042 on day −3 to 0.41 ± 0.042 on day 4 (data not shown).
We chose to use these shelters because, due to the shape, would have likley have a minimal risk of being used as a platform for the rats with no horizontal surfaces to defecate and urinate on. Although the shelters, available in red and amber clear polycarbonate, are designed to reduce perceived light and thus may be perceived as nontransparent by the rats, they still have 3 relatively large open holes from which light can enter. Of note, the common assumption that red light is invisible to rodents because they are dichromats was documented by a previous study16 to be a misconception.
Two other concerns regarding the choice of shelter were considered; first, the ball-shaped shelter was chosen because it would leave an open floor area over the grid floor, allowing urine and feces to pass through and be collected if the rat urinated or defecated while in the shelter.14 Second, we were concerned that the ball-shaped shelter would occupy too much space in the cage, leaving too little for the rat to move. Furthermore, if the rat relocated the shelter, it could accidentally cover the feed tube and obstruct access to feed. However, we decided not to fix the shelters to the floor of the cages to allow the rats to manipulate and explore them as cage enrichment. Fortunately, we did not observe that the ball-shaped shelters had blocked the entrance to the feed tube. Finally, similar to other studies,4,28 quantification of water intake would have complemented our data on feed intake as a measure of the rats’ behavior in response to the diets and supplementation of shelter.
We used weanling male rats in the current study, as this is a standard rodent model used for determining the nutritional quality of protein for human foodstuffs,7 and our aim of the study was to improve the current setup without compromising the quality of the obtained data. Therefore, we cannot extrapolate the data to older male or female rats. An earlier study has shown that the frequency of urination as scent marking is dependent on sexual maturity and, for females, whether they are estrous or diestrous.3 Therefore, the risk of contamination might be higher when using adult male and female rats.
Finally, although the study demonstrated that the rats used the shelters when available, the study did not include measures allowing us to conclude that the animals’ welfare was improved.
In conclusion, the rats used the ball-shaped shelters when available and mostly during the daytime. Cage enrichment with a ball-shaped shelter was not a significant source of error in our N-balance study. According to our results, they can be implemented in standard metabolism studies. Changes in the design of the metabolic cages to minimize feed spillage and avoid urination in the feed are probably more important for achieving reliable quantitative data of feed, feces, and urine than any minor loss of N caused by using these ball-shaped shelters.
Outline of the experimental setup for metabolic cages and the treatments in blocks 1 and 2 of the experiment. In each block, 2 racks contained 4 cages at the top and the bottom of the rack, and 2 racks contained 4 cages in the middle, all representing all 4 treatments. Above each row containing cages, 2 cameras were fixed to the rack to record activity in 2 cages simultaneously. Dietary treatments were switched in block 2 compared with block 1 to exclude confounding effects of location. GP, green protein diet; SBM, soybean meal diet.
(A) A row of 4 metabolism cages seen from the end of the rack. (B) A metabolism cage equipped with a red ball-shaped shelter as seen from above.
Distribution of how rats used the environment (cage floor, feed tube, or shelter), shown as the mean ratio per observation period ± SE for rats (A) without shelter and rats (B) with shelter based on video recordings.
The relative use of the feed tube for rats without (No) or with (Yes) access to shelter during nighttime and daytime on observation days −3, 1, and 4. Values are least significant means with SEM as error bars (n = 24).
Contributor Notes