This article has Open Peer Review reports available.
Optimized animal model to mimic the reality of stress-induced depression in the clinic
© The Author(s). 2017
Received: 2 September 2016
Accepted: 26 April 2017
Published: 6 May 2017
Animal models are useful tools for verifying the relationship between stress and depression; however, an operational criterion for excluding the resilient animals from the analysis has not been established yet, which hinders the model’s ability to more accurately mimic the scenario in humans.
To induce depression-like symptoms, rats received maternal deprivation (MD) during PND1–14, and/or chronic unpredictable stress (CUS) exposure. The latent profile analysis (LPA) was used to determine latent subgroups in treatment naive adult rats. The percentile method was used to distinguish sensitive and non-sensitive behaviors in rats.
The sucrose preference rate of treatment naive adult rats was fit using a Beta distribution, while immobility time was fit using a Gamma distribution. Indexes of behavioral tests revealed the 4-class model as the best fit for treatment naive adult rats. The incidence of stress-resilience in MD rats was significantly higher than that in CUS rats and MD + CUS rats. There was a significantly higher incidence of stress-resilience in CUS rats compared with MD + CUS rats. Recovery rate of anhedonia-like and sub anhedonia-like behaviors in CUS rats was significantly higher than that in MD and MD + CUS rats. There was a significantly higher recovery rate of anhedonia-like behaviors in MD rats compared to MD + CUS rats.
The percentile method is suitable for setting up an operational cutoff to classify depression-like, sub depression-like, and resilient behaviors in rats exposed to MD and CUS.
KeywordsDepression Animal model Latent profile analysis Percentile method Cutoff
Depression is a complex, heterogeneous, and often chronic psychiatric disorder [1, 2]. In the clinic, depression exhibits various clinical phenotypes and inconsistent responses to current antidepressant therapy . Currently, depression is viewed as a multifactorial disease that encompasses environmental, social, psychological, and biological factors. Among these etiologies, psychological stress is one of the most relevant risk factors for psychopathologies, and the exposure to stressful events powerfully triggers depressive episodes . However, ethical limitations prevent the validation of the stress hypothesis and the causality between stress and depression in humans. Animal models are therefore important for investigating the molecular and cellular mechanisms as well as treatment response to antidepressant agents in different subtypes of depression.
It is normally acknowledged that the same stress only causes psychopathology in a small portion of the population, and only the patients with obvious symptoms receive treatments or intervention. Our previous study observed that some animals are resilient and can cope with the applied stressors, while some are much more vulnerable or susceptible to stressors and drop into an anhedonia and/or despair-like statue . However, the animals that are not sensitive to stress are also included in the analysis, which is not the scenario in the clinic. To mimic the scenario of depression development in humans, animal models should also exclude animals not sensitive to stresses. An operational criterion needs to be established for depressive behaviors to distinguish between animals that are sensitive to stress from those that are not sensitive to stress.
The involvement of stress in the development of depression has been widely investigated in animals. For example, animal models established with maternal deprivation and chronic unpredictable stress-exposure are widely used models that mimic stress experienced in early life and during adulthood of humans, respectively [6, 7] and the validity of these depression models have been widely verified [8, 9]. Due to patients with depression who have experienced stresses in both their early life and adulthood, the combination of these two stress models may better reflect the multiple stresses-induced depression in human. However, how an animal model of depression mimics the responses exhibited by humans where only stressor-sensitive individuals can develop depression has not been well addressed although some studies may actually exclude animals not sensitive to stresses [10, 11]. Moreover, different animals, different stresses, and different behavioral test methods may yield different readouts in behavioral tests. A universal approach to establish an operational criterion for depressive behaviors to exclude behaviorally abnormal animals and animals not sensitive to stresses may improve the current animal models of depression.
The percentile method is used in medicine to determine reference limits [12–14]. The primary advantage of this method is that the reliability of sample percentiles can be determined without making any assumptions about the distribution shape (Gaussian or Skewed distribution) of the population. Determining the percentiles of a sample is required for the exclusion of specified population percentages (generally, more than 5%, less than 15%) [12, 14]. However, whether the percentile method could be used to determine the cutoff value of depressive behaviors in animals has not been validated. In this study, we first performed a latent profile analysis (LPA) to see whether latent subgroups can be identified in a large sample of untreated adult rats using the sucrose preference rate and immobility time. If it is successful, we then determined a cutoff value of anhedonia-like and despair-like behaviors using a percentile method depending on readouts of behavioral tests, followed by analysis of the incidence of different behavioral intensity and the effect of escitalopram treatment.
Maternal deprivation (MD)
The MD paradigm was designed as previously described . Briefly, pups were separated from their mothers for 6 h daily from PND 1 to PND 14 (the separations occurred at 9:00–15:00). To block communication between littermates, each offspring was placed in a single cell (a cage with a size of 32 cm × 32 cm × 14 cm was divided into four cells of the same size) covered with dry sawdust. At the end of the separation period, pups were returned to their maternal cages. Temperature inside the cell was maintained at 32 °C for the first five deprivations and 30 °C for the other deprivations.
Chronic unpredictable stress (CUS)
The CUS paradigm was modified from a previously established protocol . Briefly, rats at 10 weeks old were exposed to a variety of sequential stressors for 4 weeks: electric foot shock for 20s (0.6 mA, 1-s duration, average 1 shock/10s), an elevated open platform (10 cm × 10 cm, 160 cm in height) for 2 h, crowding for 10 h (5–6 rats cage), wet bedding for 15 h, water deprivation for 24 h, food deprivation for 24 h, and restraint stress for 2 h. In order to establish unpredictability, stressors were distributed randomly every day at different times.
In this study, the widely used antidepressant escitalopram (Esc) (H. Lundeck A/S Copenhagen, Denmark) was applied, and the protocol was conducted as previously described . Escitalopram was dissolved in 0.9% physiological saline (1 mg/ml) immediately before use; then, it was administered intraperitoneally and daily (5 mg/kg) in the morning over a 4-week period of time for depressive rats (n = 20 each group). Naïve rats (n = 20) received saline as a vehicle injection. In order to maintain the depressive-like behavior, CUS was still exposed during the antidepressant treatment.
Sucrose preference test (SPT)
Sucrose preference test was performed as described previously  to assess the hedonia level in rats. This test was carried out before CUS and at the end of each week. Rats were kept individually in separate cages, and were allowed to adapt to two bottles of solution (filled with 1.5% sucrose solution) for 24 h. For the next 24 h, one bottle of sucrose solution was replaced with water. Then, the rats were subjected to 18 h of food and water deprivation, followed by exposure to two pre-weighed bottles of solution (1.5% sucrose solution and plain water, respectively) for 1 h. The position of the bottles was switched. After the test, the weight of sucrose solution and water consumed was recorded. Sucrose preference was calculated as a ratio of the weight of sucrose solution consumption to the weight of total fluid intake.
Forced swimming test (FST)
FST was carried out to measure behavioral despair following a previously established protocol . The test was performed at the beginning and at the end of the CUS procedure, as well as at the end of the antidepressant treatment. Two swimming sessions were conducted: a 15-min pretest on day 1 followed by a 5-min test the next day. Rats were forced to swim in a glass cylinder (21 cm diameter × 46 cm high) containing 30 cm depth of water (25 °C). Behavior was video-recorded using SMART video-tracking system (Panlab, Spain). The immobility time (time a rat spent in floating with no active activity other than those necessary to keep head above the water) was analyzed. Water was changed after each test.
Data analysis, excluding latent profile analysis (LPA), was performed using the statistical software SPSS 17.0. LPA was performed using Mplus version 7.0. First, the distribution of indexes in behavioral tests was analyzed. Then, exploratory LPA was performed to identify latent subgroups based on the behavioral indexes of treatment in native adult rats. The model selection and comparison were conducted by evaluating the fit of a 1-class model; the number of latent classes was then increased until the addition of latent classes was not justified. Five analyses were used to determine the optimal number of class: 1) the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Adjusted BIC (aBIC) values were calculated with lower values indicating a better fit; 2) The P value of Lo-Mendell-Rubin-adjusted likelihood ratio test (A LMR) was used to determine whether more classes (P < 0.05) or fewer classes (P > 0.05) fit better; and 3) A standardized measure of entropy was used to assess classification uncertainty based on model-based posterior probabilities with higher value indicating better classification of individuals. After determining the best class solution, indexes of behavioral tests were compared across the latent classes using the one-way analysis of variance (ANOVA) . The percentile method was used to determine the operational cutoffs for identifying depressive and sub-depressive behaviors. Chi-square test was used to compare the incidence of depressive behaviors. A P < 0.05 was considered statistically significant.
The distribution of sucrose preference rate and immobility time in treatment naive adult rats
Descriptive statistics of readouts of normal rats in behavioral tests
sucrose preference rate
Immobility time (s)
Mean ± SD
0.70 ± 0.21
86.76 ± 57.28
The number of latent classes for behavior of treatment native adult rats
Fit statistics of competing latent classes
Class assignment probability by class (N = 309)
n = 44 (14.24%)
n = 100 (32.36%)
n = 153 (49.51%)
n = 12 (14.24%)
The characteristic of behavior in different classes of adult rats
Comparison of behavioral test indexes among four classes
n = 44
n = 100
n = 153
n = 12
sucrose preference rate
0.39 ± 0.12
0.74 ± 0.18
0.78 ± 0.14
0.54 ± 0.13
1 < 4 < 2,3
52.18 ± 29.86
136.67 ± 26.61
52.71 ± 23.51
227.77 ± 28.99
4 > 2 > 1,3
The operational cutoffs for sucrose preference rate and immobility time
Due to LPA suggesting the existence of latent subgroups with a range of severity of depression in treatment native rats, the percentile method was further used to determine the operational cutoffs based on the skewed distribution of sucrose preference rate and immobility time. The 95-, 90- and 85-percentile (or 5-, 10- and 15-) are used widely for determining reference limits in medicine [12, 14]. In this study, 5- and 10-percentile cutoffs were used to define the anhedonia-like and sub anhedonia-like behavior; 95- and 85-percentile values were used to define the despair-like and sub despair-like behavior. Results showed that the 5-percentile and 10-percentile values of sucrose preference rate were 0.28 and 0.41, respectively (Table 1). This finding means that a sucrose preference rate lower than 0.28 could be defined as anhedonia-like, while a sucrose preference rate between 0.28 and 0.41 could be defined as sub anhedonia-like, and a sucrose preference rate higher than 0.41 could be defined as anhedonia-resilient. Similarly, in the forced swimming test, the 95-percentile of immobility time was 187.95 s, while the 85-percentile was 149.40s (Table 1). This finding suggests that an immobility time longer than 187.95 s could be defined as despair-like, while an immobility time between 149.40s and 187.95 s could be defined as sub despair-like, and an immobility time shorter than 149.40s could be defined as despair-resilient. After escitalopram treatment, rats recovering from depression-like or sub depression-like status (sucrose preference rate > 0.41, immobility time < 149.40s) were defined as recovering, while others were defined as being escitalopram treatment-resilient.
The sucrose preference rate of stressed rats in sucrose preference test
0.18 ± 0.05*Δ
0.35 ± 0.05*
0.71 ± 0.16
0.20 ± 0.06*Δ
0.34 ± 0.06*
0.71 ± 0.17
0.21 ± 0.05*Δ
0.35 ± 0.06*
0.69 ± 0.18
MD + CUS
0.19 ± 0.05*Δ
0.31 ± 0.06*
0.70 ± 0.16
The immobility time of stressed rats in forced swimming test
245.26 ± 30.78*Δ
176.68 ± 31.28*
87.66 ± 37.94
248.91 ± 30.51*Δ
172.68 ± 10.07*
86.99 ± 39.74
242.30 ± 33.55*Δ
173.54 ± 10.41*
87.51 ± 39.14
MD + CUS
248.82 ± 35.30*Δ
176.01 ± 12.58*
88.58 ± 36.40
The incidence of depression-like and sub depression-like behaviors in CUS-induced rats with and without MD
The incident of anhedonia and sub-anhedonia in stressed rats
MD + CUS
The incident of despair and sub despair in stressed rats
MD + CUS
The effect of escitalopram treatment on depression-like and sub depression-like behaviors
The rate of recovery in depressive behaviors after escitalopram treatment
Sub anhedonia-like n(%)
Sub despair-like n(%)
MD + CUS
Depression is a highly heterogeneous disease and may be caused by different factors. This requires animal models to validate the casualty between stress and depression that should exclude animals resilient to stresses to mimic the real scenario in human. In addition, the treatment naive animals purchased from commercial resources are generally regarded as behavior normal. However, some animals are much more susceptible to stressors and may have already dropped into an anhedonia and/or despair-like statue during growth and delivery , but these animals were not generally excluded from study. It is therefore important to establish a cutoff value to exclude animals resilient to stresses or already having behavioral abnormalities.
In this study, we first performed sucrose preference test and forced swimming test in a large sample of treatment naive adult SD rats. Second, we used the latent profile analysis (LPA) to verify that latent subgroups can be identified in treatment naive adult rats using the sucrose preference rate and immobility time. We found that the naive treatment rats were identified by four classes, which was a low sucrose preference rate and short immobility time in class 1, a high sucrose preference rate and moderate immobility time in class 2, a high sucrose preference rate and short immobility time in class 3, and a moderate sucrose preference rate and long immobility time in class 4, which suggested that the naive adult rats were a heterogeneous group which may mimic the real scenario of heterogeneity in human populations. Third, using the percentile method, we found the existence of a certain natural incidence of depression and resilient to stresses in treatment naive SD rats. As far as we know, this is the first time the percentile method is introduced into an animal model of depression. Due to the stress (Maternal deprivation and/or Chronic unpredictable stress) induced depression-like behaviors were also tested by forced swimming test and/or sucrose preference test in Wistar rats [16, 17] and mice [18, 19]. Thus, the percentile method can also be used to establish cutoff values of depressive behaviors in other animals. Due to different animals, different stresses, and different behavioral test methods can yield different readouts in behavioral tests, a cutoff value depending on the readouts of any behavioral tests lacks a high universality. In contrast, the percentile method has a high universality and can be used in different animals with different behavioral test methods.
Both the clinical and preclinical evidence indicate the existence of subtypes in depression and differences in stresses that may be the main causes leading to the formation of the subtypes [20, 21]. We further tested whether the percentile method has a universality between different stresses (MD vs. CUS), between single and multiple stresses (MD alone, CUS alone, and MD + CUS), and between different subtypes of behaviors (despair-like vs. anhedonia-like). We revealed that 5- and 10-percentile can be used to determine the anhedonia-like and sub anhedonia-like behaviors. While 95- and 85-percentile can be used to determine the cutoffs for despair-like and sub despair-like behaviors. Although significant differences in the mean behavioral readouts between MD, CUS, MD + CUS and NOR rats were observed, the animals’ behaviors underlying three stresses can be grouped as despair-like/anhedonia-like, sub despair-like/sub anhedonia-like, and resilient behaviors using the percentile method. In particular, stress-resilient rats and treatment naive rats responded similar in behavioral tests, which suggests that stress-resilient rats should be subsequently excluded or included according to the purpose of study.
To further elucidate the significance of establishing a cutoff value, we tested the response of rats with different subtypes of behaviors to the current antidepressant escitalopram treatment. The resilient rats were excluded from analysis of escitalopram treatment. Majority of anhedonia-like behaviors were ameliorated by escitalopram treatment. In contrast, both despair-like and sub anhedonic-like behaviors in three models of rats were not reversed by a 4-week escitalopram treatment. Christensen et al. study using Hierarchical clustering analysis revealed a subgroup segregation pattern in the chronic mild stress rat model and antidepressant treatment refractors cluster with anhedonic-like rats. Conversely, stress-resilient rats cluster with rats undergoing anti-depressant-mediated recovery from anhedonia . Thus, establishing a cutoff value to exclude animals resilient to stresses or high sensitive to stresses is also important for preclinical study of new antidepressants.
In this study, the percentile method is suitable for setting up an operational cutoff for classifying the sensitivity of rats to stresses-induced depression. The significances of the present study included: 1) This study first established a cutoff value in three animal models to study stress-induced depression; 2) Further studies or data analysis with the exclusion of no sensitive animals more accurately reflect the real scenario in humans; 3) Preclinical tests with the exclusion of behavior abnormal animals and resilient animals for therapeutic agents will provide more accurate information, and 4) cutoff value established by the percentile method may have higher universality than if it were established based on the actual readouts of behavioral tests. However, this study only used a single strain of male rats. In the clinic, females are generally more susceptible to depression than males. Moreover, only one model of chronic unpredictable stress and one model of maternal deprivation was used. Currently, there are numerous different experimental models for depression study. Therefore, the generalizability of the results should be considered.
This work was supported by National Natural Science Foundation of China (No. 81671341).
Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.
YZ, YW, and XZ conceived and designed the experiments; YW, LW, LX and XW performed the experiments; YZ, HL, YW, and XZ analyzed and discussed the data. YZ was a major contributor in writing the manuscript; All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
All protocols were approved by the Animal Care and Ethics Committee of Central South University. All efforts were made to reduce suffering and discomfort to the animals at all experimental stages.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Kessler RC, Bromet EJ. The epidemiology of depression across cultures. Annu Rev Public Health. 2013;34:119.View ArticlePubMedPubMed CentralGoogle Scholar
- Worldwide Health Organization. Depression. 2012. http://www.who.int/mediacentre/factsheets/fs369/en. Accessed 11 Sept 2014.
- Bauer M, Bschor T, Pfennig A, Whybrow PC, Angst J, Versinani M, et al. WFSBP Task Force on Unipolar Depressive Disorders, World Federation of societies of biological psychiatry (WFSBP) guidelines for biological treatment of unipolar depressive disorders in primary care. World J Biol Psychiatry. 2007;8:67–104.View ArticlePubMedGoogle Scholar
- Heim C, Bradley B, Mletzko TC, Deveau TC, Musselman DL, Nemeroff CB, et al. Effect of childhood trauma on adult depression and neuroendocrine function: sex-specific moderation by CRH receptor 1 gene. Front Behav Neurosci. 2009;3:41.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhang Y, Wang Y, Wang L, Bai M, Zhang X, Zhu X. Dopamine Receptor D2 and Associated microRNAs Are Involved in Stress Susceptibility and Resistance to Escitalopram Treatment. Int J Neuropsychopharmacol. 2015;18.
- Wiborg O. Chronic mild stress for modeling anhedonia. Cell Tissue Res. 2013;354:155–69.View ArticlePubMedGoogle Scholar
- Bogdanova OV, Kanekar S, D'Anci KE, Renshaw PF. Factors influencing behavior in the forced swim test. Physiol Behav. 2013;118:227–39.View ArticlePubMedGoogle Scholar
- Binder E, Malki K, Paya-Cano JL, Fernandes C, Aitchison KJ, Mathé AA, et al. Antidepressants and the resilience to early-life stress in inbred mouse strains. Pharmacogenet Genomics. 2011;21:779–89.View ArticlePubMedGoogle Scholar
- Mrdalj J, Lundegaard Mattson A, Murison R, Konow Jellestad F, Milde AM, Pallesen S, et al. Hypothermia after chronic mild stress exposure in rats with a history of postnatal maternal separations. Chronobiol Int. 2014;31:252–64.View ArticlePubMedGoogle Scholar
- Christensen T, Bisgaard CF, Wiborg O. Biomarkers of anhedonic-like behavior, antidepressant drug refraction, and stress resilience in a rat model of depression. Neuroscience. 2011;196:66–79.View ArticlePubMedGoogle Scholar
- Brennan K, Roberts D, Anisman H, Merali Z. Individual differences in sucrose consumption in the rat: motivational and neurochemical correlates of hedonia. Psychopharmacology. 2001;157:269–76.View ArticlePubMedGoogle Scholar
- Wellek S, Lackner KJ, Jennen-Steinmetz C, Reinhard I, Hoffmann I, Blettner M. Determination of reference limits: statistical concepts and tools for sample size calculation. Clin Chem Lab Med. 2014;52:1685–94.View ArticlePubMedGoogle Scholar
- Goel VV, Poole SF, Longhurst CA, Platchek TS, Pageler NM, Sharek PJ, et al. Safety analysis of proposed data-driven physiologic alarm parameters for hospitalized children. J Hospital Med. 2016;11:817–23.View ArticleGoogle Scholar
- Herrera L. The precision of percentiles in establishing normal limits in medicine. J Lab Clin Med. 1958;52:34–42.PubMedGoogle Scholar
- Wang HH, Wang MC, Wu SQ. [Mobile Phone Addiction symptom Profiles Related to Interpersonal Relationship and Loneliness for College Students: A Latent Profile Analysis]. Chin J Clin Psychol (Chinese), 2015, 23(5):881-885.Google Scholar
- Réus GZ, Carlessi AS, Titus SE, Abelaira HM, Ignácio ZM, da Luz JR, et al. A single dose of S-ketamine induces long-term antidepressant effects and decreases oxidative stress in adulthood rats following maternal deprivation. Dev Neurobiol. 2015;75:1268–81.View ArticlePubMedGoogle Scholar
- Li Y, Ji YJ, Jiang H, Liu DX, Zhang Q, Fan SJ, et al. Effects of unpredictable chronic stress on behavior and brain-derived neurotrophic factor expression in CA3 subfield and dentate gyrus of the hippocampus in different aged rats. Chin Med J. 2009;122:1564–9.PubMedGoogle Scholar
- Kurhe Y, Mahesh R. Pioglitazone, a PPARγ agonist rescues depression associated with obesity using chronic unpredictable mild stress model in experimental mice. Neurobiol Stress. 2016;3:114–21.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang L, Jiao J, Dulawa SC. Infant maternal separation impairs adult cognitive performance in BALB/cJ mice. Psychopharmacology. 2011;216:207–18.View ArticlePubMedGoogle Scholar
- Bai M, Zhang L, Zhu X, Zhang Y, Zhang S, Xue L. Comparison of depressive behaviors induced by three stress paradigms in rats. Physiol Behav. 2014;131:81–6.View ArticlePubMedGoogle Scholar
- Czéh B, Fuchs E, Wiborg O, Simon M. Animal models of major depression and their clinical implications. Prog Neuro-Psychopharmacol Biol Psychiatry. 2016;64:293–310.View ArticleGoogle Scholar