Effects of social isolation and environmental enrichment on atherosclerosis in ApoE^−/− mice

Abstract

Social support and a stimulating environment have been suggested to reduce stress reactions and cardiovascular risk. The aim of this study was to assess the role of environmental enrichment and social interaction for development of atherosclerosis in atherosclerosis prone mice. Male ApoE^{− / −} mice were divided into four groups and followed during 20 weeks: (i) enriched environment (E, n = 12), (ii) deprived environment (ED, n = 12), (iii) enriched environment with exercise (E-Ex, n = 12) and (iv) socially deprived by individual housing (SD, n = 10). Plasma lipid and cytokine concentrations were measured. Atherosclerosis was quantified in cross-sections of innominate artery and en face in thoracic aorta. Plaque area was significantly increased in SD mice in the innominate artery (P < 0.05 vs. all other groups), but not in the thoracic aorta. Plasma lipids were increased in SD mice (P < 0.001 vs. all for total cholesterol, P < 0.05 vs. E and P < 0.01 vs. ED for triglycerides). Plasma concentration of granulocyte-colony stimulating factor (G-CSF) was decreased in SD mice compared to E mice (P < 0.05). Thus, social isolation increased atherosclerosis and plasma lipids in ApoE^{− / −} mice. Reduction in plasma G-CSF levels may hamper endothelial regeneration in the atherosclerotic process. While environmental enrichment did not affect atherosclerosis, social isolation accelerated atherosclerosis.

• ApoE^{− / −} mice
• atherosclerosis
• corticosterone
• enrichment
• plasma lipid
• social deprivation

Introduction

Accumulating evidence suggest that psychosocial stress has a large impact on our health and on development of cardiovascular disease (Rozanski et al. 1999). Recently, the results of the INTERHEART study indicated that the presence of psychosocial stressors greatly increase the risk of acute myocardial infarction in man (Rosengren et al. 2004). However, to study psychosocial stress in humans is complex and to gain mechanistic insight is challenging. To this end, animal models have been developed in which stressors can be carefully controlled and evaluated. In higher species, such as cholesterol fed cynomolgus monkeys, chronic stress induced by social interruption and reorganization accelerates coronary artery atherosclerosis (Shively et al. 1989). However, also rabbit, rat and mouse models of psychosocial stress have been developed and the neurohormonal reaction patterns to stress are similar to those seen in humans (Koolhaas et al. 1997). Genetically, engineered atherosclerosis prone mice provide important tools for mechanistic studies of atherosclerosis development. One brief report suggests that repeated exposure to stressors accelerates development of atherosclerosis in ApoE^{− / −} mice (Kumari et al. 2003). However, these models have not yet been exploited in this area of research.

Social interactions and lack of social support are important sources of psychosocial stress in humans (Rozanski et al. 1999) as well as in animal models (Shively et al. 1989; Koolhaas et al. 1997; Marashi et al. 2003). Social isolation is also known to increase cortisol levels in monkeys and to increase atherosclerosis in rabbits (Stanton et al. 1985; McCabe et al. 2002). Further, the level of environmental enrichment in animal cages has been shown to affect both stress related neuroendocrine function as well as immunological function in mice (Marashi et al. 2003). Male mice living in cages environmentally enriched with plastic houses, climbing material and wooden scaffolding display higher plasma cortisol levels most likely explained by increased social interactions and agonistic behavior (Marashi et al. 2003).

The aim of the present study was to investigate the effect of environmental enrichment or social isolation on the development of atherosclerosis in genetically engineered ApoE^{− / −} mice, on a C57BL/6 background, which are prone to atherosclerosis. Apolipoprotein E (ApoE) is a ligand for the light density lipoprotein (LDL) receptor and plays an important role in the clearance of circulating lipoproteins from the plasma. ApoE^{− / −} mice will thus display increased lipoprotein levels in plasma accompanied by increased atherosclerosis (Plump et al. 1992; Breslow 1993). We studied development of atherosclerosis in four different colonies of mice. ApoE^{− / −} mice were either socially deprived by housing alone (isolation) in individual cages or housed in large colonies with three different levels of environmental enrichment.

Methods

Animals

Male ApoE^{− / −} mice were obtained from Taconic M&B Breeding and Research Centre (Bomholtgaard, Denmark) at 4 weeks of age, with an average weight of 10.6 ± 0.2 g. At five weeks of age, the mice were randomly divided into four groups exposed to four different housing conditions and followed during 20 weeks of experiment.

Housing

Environmentally enriched (E)

Mice (n = 12) were kept in a large Perspex cage (800 × 500 × 250 mm) with access to plastic nest-boxes, tunnels and nest building material. Tunnels and nest-boxes were rearranged once a week and replaced with new ones once a week, to promote a stimulating environment. The nest-box that the mice actually chose to use as a nest was not moved or replaced, to create a familiar “home” within the stimulating surroundings. Once a week dirty sawdust was removed and replaced with new sawdust, and once every three weeks the entire cage was cleaned. This was done to avoid disturbance of established hierarchies, but without compromising sanitary status. Mice were handled as little as possible.

Environmentally enriched with access to voluntary running wheels (E-Ex)

Mice (n = 12) were kept under similar conditions as (E), but with access to three running wheels.

Environmentally deprived (ED)

Mice (n = 12) were kept in a similar sized large Perspex cage but with a barren environment. Mice were deprived of all environmental stimuli except for sawdust, food and water. The cage was cleaned and sawdust was changed three times a week to maintain a sterile and unfamiliar environment.

Socially deprived (SD)

Mice (n = 10) were socially deprived by individual housing in smaller cages (300 × 200 × 150 mm). The isolated mice had access to nest building material, to allow maintenance of normal body temperature. Cages and sawdust were changed once a week.

Water and standard chow were available ad libitum to all groups. Mice were housed at 21–25°C in a room with a lights on at 6 am and lights off at 6 pm. All procedures involving mice were approved by the Regional Animal Ethics Committee at Göteborg University, in accordance with the European Communities Council Directives of 24 November 1986 (86/609/ECC).

Body weight measurements

Mice were weighed every second week throughout the study.

Salt appetite as an indicator of stress

Stress has previously been shown to induce an increased appetite for salt in rats (Bourjeili et al. 1995). We measured salt appetite, at week 17 of the experiment, to assess the level of stress in the four mouse colonies. Preference for salt in drinking water was measured for two consecutive days by letting mice choose between tap water and a 1% NaCl solution. A ratio for salt preference was calculated by dividing volume of 1% NaCl solution consumed by volume of tap water consumed.

Measurement of exploratory behavior using activity boxes

Environmental enrichment and social isolation have previously been shown to affect exploratory behavior in rats (Varty et al. 2000). Exploratory behavior was therefore measured in the current study at week 16 of the experiment, as previously described (Svensson et al. 2005). Briefly, exploratory behavior was measured during two hours in a novel environment, using activity boxes (Kungsbacka Mät-och Reglerteknik AB, Fjärås, Sweden). Boxes were supplied with photocell detectors recording three-dimensional activity of the mice. We recorded and analyzed three different variables of exploratory behavior: (i) locomotor activity, as an overall measure of the mouse's activity, (ii) rearing activity, as a measure of exploratory activity and (iii) corner time, as a measure of anxiety, fear or inactivation (Svensson et al. 2005).

Urinary corticosterone concentrations

Urine from week 14 of the experiment was used for corticosterone analysis. Samples were collected between 08:00 and 12:00 noon, by placing mice in separate Perspex cages without sawdust. The majority of mice spontaneously voided urine within two hours. As a step in endogenous metabolism, corticosterone undergoes glucuronidation before it is excreted in urine. Samples were therefore incubated with β-glucuronidase (final concentration in samples was 50 U/mL) in an equal volume of acetate buffer (pH 4.65) at 37°C for three full days to regenerate unmetabolized corticosterone for measurement. After incubation pH was adjusted to pH7 by adding an equal volume of 100 mM sodium phosphate buffer (pH 7.5). Corticosterone was then measured using a correlate-EIA corticosterone kit (Assay Designs, Ann Arbor, MI, USA, inter-/intra-assay variability: 13.1 and 8.0%CV respectively, sensitivity: 26.99 pg/mL), according to the manufacturer's protocol. Urinary corticosterone concentration was corrected by urine creatinine concentration, measured according to the manufacturer's protocol (Sigma Diagnostics, Inc., St Louis, MO, USA).

Oxidative stress

Isoprostanes are prostaglandin-like compounds produced during lipid peroxidation and have been suggested to be a marker for in vivo lipid oxidation (Kadiiska et al. 2005). Further, measuring urinary isoprostanes is non-invasive and thus convenient for animal studies. Isoprostanes were analyzed in urine samples from week 19 of the experiment using an enzyme immunoassay for isoprostane (OxisResearch, Portland, OR, USA, sensitivity: 50 pg/mL). Urinary isoprostane concentrations were corrected by urine creatinine concentrations.

Blood collection

Blood for analysis of plasma lipids and cytokines was collected at termination from the right ventricle into a standard coated K-EDTA tube. Mice were anaesthetized before blood collection with pentobarbital (Apoteksbolaget, Sweden, 0.9 mg/g BW i.p). After centrifugation, plasma was separated and stored at − 20°C until analysis.

Total plasma cholesterol and triglycerides

Total cholesterol and triglycerides in plasma were analyzed enzymatically and the concentrations were subsequently determined spectrophotometrically (Roche/Hitachi analyzer, Roche Diagnostics, Indianapolis, IN, USA, inter-/intra-assay variability: 1.8 and 1.5%CV, respectively).

Cytokine measurements

Using a premixed Bio-Plex Mouse Cytokine panel (Bio-Rad Laboratories, Hercules, CA, USA, inter-/intra-assay variability: < 30 and < 20%CV, respectively, sensitivity: ∼10 pg/mL) cytokines relevant for the atherosclerotic process (Th1 cytokines: INF-γ, IL-2, IL-12 and TNF-α; Th2 cytokines: IL-4, IL-5 and IL-10; others: MCP-1 and G-CSF) were measured in plasma samples from SD and E mice. Briefly, samples were incubated in a 96-well microplate with antibody-coupled beads followed by incubation with biotinylated secondary antibody. Samples were then incubated with streptavidin-PE and fluorescence was read on the Bio-Plex suspension array system.

Blood pressure measurements

Systolic blood pressure (SBP) was measured during the last week (week 20) of the experiment. A computerized non-invasive tail cuff system was used (RTBP Monitor, Harvard Apparatus, Inc., South Natick, MA, USA). The conscious mouse was kept in a restrainer, with a standard acclimatization time of 10 min and gentle heating of the tail before the recording session. During the recording day, six satisfactory measurements were collected for each mouse, which subsequently were averaged. On the day of termination, at 25 weeks of age (week 20 of the experiment), mice were anaesthetized with isoflurane (0.7–1.5%) and mean arterial pressure (MAP) and heart rate (HR) were measured via the left carotid artery as previously described (Johansson et al. 2005).

Termination and fixation

Following blood pressure measurement, mice were killed with an overdose of pentobarbital (Apoteksbolaget, Sweden, 0.9 mg/g BW i.p.). The chest was opened and blood was collected as above. Thereafter, mice were perfused with 0.9% saline through the heart to clear the vascular system. The adrenal glands were removed and separated from fat and adherent tissue and then weighed separately, before the vascular tree was fixed with 4% paraformaldehyde at 100 mm Hg for 5 min. After fixation the aorta was cleared of surrounding fat and tissue and subsequently divided, between the innominate artery and the left common carotid artery, into two segments: (i) the innominate artery and (ii) the thoracic aorta (from the left common carotid artery to the right renal artery, providing a segment of the aorta containing both the aortic arch and the thoracic aorta). The segments were kept in 4% paraformaldehyde for at least one week before quantification.

Quantification of aortic plaque area and histology

Plaque area in the thoracic aorta was quantified as previously described (Johansson et al. 2005). Briefly, the thoracic aorta was cut open longitudinally and the outline of the intima was traced to calculate the total area of the vessel. Lesions were outlined in the same manner and plaque area was calculated as the percentage of the total vessel area covered with lesions. The innominate artery was analyzed regarding plaque burden as previously described (Johansson et al. 2005) with some minor modifications. Briefly, the innominate artery was paraffin embedded and serially sectioned followed by staining with Miller's (1971) elastin stain. From the contours of the internal elastic lamina, cross-sectional area of the vessel lumen was measured. Lesions were measured in the cross-sections of the vessel by a blinded observer and plaque burden was calculated as percent plaque area of total lumen area.

Statistics

All data were analyzed regarding skewness and kurtosis (SPSS version 12.0.1, Chicago, IL, USA) to determine if data were parametric or non-parametric. Data with a skewness value close to zero (less than twice its standard error) and a kurtosis value also close to zero were considered parametric. The following data were normally distributed: body weight (BW), total plasma cholesterol and triglycerides, SBP, MAP and HR. These data were analyzed using one-way ANOVA (SPSS) followed by post hoc testing using Tukeys HSD. The following data were not normally distributed: salt appetite, exploratory behavior, urinary corticosterone and isoprostane, plasma G-CSF, aortic plaque area and plaque area in the innominate artery. Non-parametric data were analyzed using Kruskal–Wallis ANOVA followed by Mann–Whitney U-test (SPSS). Exploratory behavior was analyzed as previously described (Svensson et al. 2005) by calculating area under curve, AUC (PRISM 4, GraphPad, San Diego, CA, USA), and subsequently analyzed as non-parametric data (SPSS). All data are expressed as group mean ± SEM. A value of P < 0.05 was considered to be statistically significant.

Results

BW gain was similar among the groups during the last four weeks of the observation period (Table I). Table I shows BWs from the last week of the study and no difference was found between groups (ANOVA P = 0.624).

Table I.  Effects of different housing conditions on body weight, heart rate and blood pressure.

	n	BW (g)	HR (bpm)	SBP (mm Hg)	MAP (mm Hg)
Enriched (E)	12	33 ± 0.4	574 ± 11	139 ± 4*	100 ± 3
Enriched-exercise (E-Ex)	11	32 ± 0.6	571 ± 19	136 ± 5	98 ± 3
Environmentally deprived (ED)	12	33 ± 0.7	564 ± 14	124 ± 2	99 ± 3
Socially deprived (SD)	10	33 ± 0.8	550 ± 17	124 ± 2	95 ± 2

Body weight (BW), heart rate (HR), systolic blood pressure (SBP) and mean arterial pressure (MAP). All measurements were made in week 20 of exposure to the housing condition. SBP was measured in conscious mice, HR and MAP were measured terminally in anaesthetized mice. Data are expressed as mean ± SEM. ^*P < 0.05 vs. SD and ED.

One mouse in the E-Ex group displayed very large plaque area in the innominate artery (27.7% compared to a group average of 0.47%). The same mouse had a BW 15% (5 g) below the group average. This mouse was considered an outlier and was thus excluded from the study.

Salt appetite and exploratory behavior

We found a significantly increased salt appetite (measured in week 17) among SD mice compared to E mice (Kruskal–Wallis P = 0.043; P < 0.05, SD vs. E, Table II).

Table II.  Effects of different housing conditions on urinary corticosterone concentrations, adrenal gland weight, salt appetite and oxidative stress in ApoE^−/− mice.

	Urinary CORT (ng/mg Cre)	Adr wt (μg/g BW)	Salt drinking preference (ratio saline/water)	Isoprostane (ng/mg Cre)
Enriched (E)	109 ± 17	70.5 ± 7.1	1.1 ± 0.4	17.1 ± 2.6
Enriched-Exercise (E-Ex)	81 ± 15	88.4 ± 6.9	1.6 ± 0.4	14.5 ± 2.0
Environmentally deprived (ED)	87 ± 14	86.3 ± 6.9	2.7 ± 0.6	19.2 ± 4.2
Socially deprived (SD)	121 ± 21	76.7 ± 4.5	3.3 ± 0.6*	17.9 ± 4.3

Urinary corticosterone (CORT; measured in week 14), adrenal gland weight (Adr wt; measured in week 20), salt drinking preference (measured in week 17) and urinary isoprostane (measured in week 19). Urinary data were corrected by urine creatinine levels (Cre). Data are expressed as group mean ± SEM; n = 10–12 per group. ^*P < 0.05 vs. E.

SD mice exhibited increased locomotor activity (measured in week 16) compared to the other three groups (Kruskal–Wallis P = 0.000; group means ± SEM: SD 6000 ± 500 counts vs. E 2800 ± 400 counts and E-Ex 2400 ± 300 counts, P < 0.001; SD vs. ED 4400 ± 400 counts, P < 0.01). ED mice also displayed significantly increased locomotor activity compared to both E and E-Ex groups (P < 0.05 vs. E, P < 0.01 vs. E-Ex). Furthermore, SD mice exhibited increased rearing activity compared to the other three groups (Kruskal–Wallis P = 0.000; group means ± SEM: SD 3400 ± 300 counts vs. E 1100 ± 200 counts and ED 1600 ± 200 counts, P < 0.001; SD vs. E-Ex 1500 ± 300 counts, P < 0.01). There were no significant differences between groups regarding corner time (Kruskal–Wallis P = 0.064; group means ± SEM: E 2700 ± 700 s, E-Ex 3600 ± 500 s, ED 2100 ± 500 s and SD 1000 ± 200 s).

Urinary corticosterone concentration

Urinary corticosterone concentrations (measured in week 14) were not significantly different between groups (Kruskal–Wallis P = 0.376; Table II), neither were adrenal gland weights (Table II). Furthermore, no significant correlation was found between corticosterone levels and plaque area (data not shown).

Plasma lipids

Total plasma cholesterol was significantly higher in SD mice compared to the other three groups (ANOVA P = 0.000; P < 0.001 SD vs. E, E-Ex and ED, Figure 1A). Cholesterol concentrations did not differ significantly among the three groups housed in colonies. Plasma triglyceride concentrations were increased in SD mice compared to E and ED mice (ANOVA P = 0.002; P < 0.01 SD vs. E, P < 0.001 SD vs. ED, Figure 1B).

Figure 1 (A) Total plasma cholesterol and (B) plasma triglyceride concentrations in male ApoE^{− / −} mice. Mice were housed in conditions that were: environmentally enriched (E; n = 12), environmentally enriched with access to running wheels (E-Ex; n = 11), environmentally deprived (ED; n = 12) and socially deprived (SD; n = 10). E, E-Ex and ED mice were housed in groups of 12, and SD mice were housed singly. Measurements were made in terminal samples in week 20 of the housing condition. Cholesterol concentration in SD mice was increased compared to all the other groups. Data are expressed as mean ± SEM. **P < 0.01, ***P < 0.001.

Urinary isoprostane concentration

Urinary isoprostanes were measured (in week 14) as an index of oxidative stress but did not differ significantly among groups (Kruskal–Wallis P = 0.827; Table II).

Plasma G-CSF concentrations

Among the cytokines that were analyzed in terminal blood samples a significant difference between groups was found for granulocyte-colony stimulating factor (G-CSF) concentrations. SD mice displayed significantly lower levels of G-CSF than E mice (group means ± SEM: SD 41.0 ± 7.8 vs. E 208 ± 50 pg/mL, Mann–Whitney P < 0.05. No significant differences between groups were found for the other cytokines measured (see Methods, data not shown).

Blood pressure and heart rate

SBP measured in conscious mice (week 20) was significantly higher in E mice compared to SD and ED mice (ANOVA P = 0.004; P < 0.05 E vs. SD and ED, Table I). HR and MAP measured in anaesthetized mice at termination under anaesthesia (week 20) did not differ significantly among groups (ANOVA P = 0.694 and P = 0.733, respectively, Table I).

Atherosclerotic plaques

No significant differences for en face aortic plaque area were found between groups (Kruskal–Wallis P = 0.212; Figure 2A). However, SD mice exhibited a significantly greater plaque area when cross-sections of the innominate artery were compared to the three other groups (Kruskal–Wallis P = 0.044; P < 0.05 SD vs. E, E-Ex and ED, Figure 2B).

Figure 2 (A) Plaque area in thoracic aorta expressed as the percentage of total vessel surface area and (B) plaque area in the innominate artery expressed as the percentage of total lumen area in cross-sections of the vessel in ApoE^{− / −} mice. Measurements were made after 20 weeks in the housing condition. Environmentally enriched mice (E; n = 12), environmentally enriched mice with access to running wheels (E-Ex; n = 11), environmentally deprived mice (ED; n = 12) and socially deprived mice (SD; n = 10). Data are expressed as mean ± SEM. ^#P < 0.05 vs. all.

Discussion

The results of the present study showed increased atherosclerosis in socially isolated ApoE^{− / −} mice compared to mice housed in groups of twelve. Different levels of environmental enrichment did not affect atherosclerosis. Social isolation also increased plasma concentrations of cholesterol and triglycerides, suggesting that social isolation has effects on lipid metabolism which may be causative in the atherosclerotic process. Social isolation possibly increased psychosocial stress in the mice (Matsumoto et al. 2007), in turn causing increased atherosclerosis. It is, however, also possible that altered physical activity and behavior may play a role in the atherogenic process in the present study. Social isolation is known to result in increased aggressiveness which may affect the level of activity (Matsumoto et al. 2005). Interestingly, levels of G-CSF were reduced in socially isolated mice; this may hamper endothelial regeneration in atherosclerosis. Further, living in a barren environment did not affect the extent of atherosclerosis, suggesting that social interaction is more important than environmental enrichment in the development of atherosclerosis in mice.

There were differences in the impact of social deprivation on the extent of atherosclerosis in the thoracic aorta and innominate artery, with a significant effect only on the innominate artery. Such a relationship has previously been described, with only a weak correlation between plaque size in the thoracic aorta and innominate artery (R² = 0.32) (Meir and Leitersdorf 2004). The discrepancy between these two sites may be due to the lesions first developing close to the heart and then progressing through the vascular tree.

SD mice had markedly increased plasma levels of both triglycerides and cholesterol, by 40 and 80%, respectively, at the end of the study compared to the other groups. This is likely a major driving force for the observed increase in atherosclerosis. Psychosocial stress has been shown to increase plasma lipid levels in both animal studies and in humans (Brindley et al. 1993), an effect which has been associated with activation of the sympathetic nervous system (Brennan et al. 1996; Stoney et al. 1999). In an attempt to assess the level of psychosocial stress in our mice, we measured salt appetite. Psychosocial stress has previously been shown to induce a clonidine dependant increase in appetite for salt in rats (Ely et al. 1987; Bourjeili et al. 1995). In the current study, SD mice voluntarily consumed more salt suggesting increased psychosocial stress, perhaps linked to an increased sympathetic drive in these mice. However, previous evidence for a close relation between salt appetite and psychosocial stress were from rat studies (Ely et al. 1987; Bourjeili et al. 1995), and the present findings need further confirmation in mice. Sympathetic activity was not directly assessed in our mice and is of interest for future studies.

Social isolation may also affect physical activity of mice which may in turn affect lipid levels. However, physical activity is difficult to measure without affecting normal behavior of the mice, particularly for individuals in large groups, and was not assessed in this study. Nevertheless, social isolation did not affect BW during the study. We cannot, however, fully account for the possibility that alterations in body composition could have affected plasma lipids in the current study.

Increased oxidative stress and inflammation are believed to be central in several aspects of both initiation and acceleration of atherogenesis (Ross 1999; Harrison et al. 2003; Hansson 2005). Social and emotional stress have been suggested to increase markers for oxidative stress in mice (Miyashita et al. 2006) as well as in humans (Cernak et al. 2000). As a marker to assess oxidative stress we analyzed urinary isoprostane, an end product of lipid peroxidation (Kadiiska et al. 2005). All groups excreted similar amounts of isoprostane and it thus appears as though oxidative stress was not affected by psychosocial stress in the present study. However, measurement of urinary isoprostane only reveals changes in systemic oxidative stress and does not assess local changes of oxidative stress possibly present in the blood vessel wall. Similarly, no significant difference was found for plasma levels of a series of important cytokines all considered to be involved in different aspects of the atherosclerotic process (Hansson and Libby 2006). However, G-CSF, which is a haemotopoietic cytokine that induces release of haematopoietic stem cells and endothelial progenitor cells (EPCs) from the bone marrow, was decreased in SD mice. EPCs are associated with an enhanced re-endothelialization of injured blood vessels, improved endothelial function and reduced atherosclerosis (Kong et al. 2004; Hasegawa et al. 2006; Werner and Nickenig 2006). Possibly, social isolation aggravates atherosclerosis via a reduction in stem cell recruitment and a reduction in the regenerative capacity of the endothelium. However, the role of G-CSF in atherosclerosis is complex; a recent study shows that exogenous treatment with G-CSF accelerates atherosclerosis in ApoE^{− / −} mice (Haghighat et al. 2007).

Activation of the HPA-axis, leading to the release of corticosterone, is usually a response to stress (Tsigos and Chrousos 2002). Effects on the HPA-axis have previously been shown in socially isolated rats (Serra et al. 2005). However, in the present study all groups showed similar levels of urinary corticosterone. Similarly, no difference in adrenal gland weights among groups was found. Measurements of urinary corticosterone and adrenal gland weight may not be sufficient to exclude a role of the HPA-axis in the atherogenic process in the present study. Nevertheless, the lack of correlation between corticosterone levels and plaque burden makes it unlikely that corticosterone has a major pathophysiological role in the observed increase in atherosclerosis.

SBP was measured with the tail cuff technique and was found to be higher in E mice compared to ED and SD mice. It is noteworthy that mice in this group were rarely handled, due to the experimental design. It is thus likely that these mice were more stressed by the procedure than the other groups showing lower SBP. Unlike the SBP measurements in conscious mice, MAP measurements in terminally anaesthetized mice showed no significant differences among groups. Hence, we found no support for increased blood pressure as a causative factor in the observed increase in atherosclerosis of SD mice.

Social isolation is a strong stimulus, evident from the impact on behavior of the mice. In the current study, SD mice exhibited increased locomotor and rearing activity in a novel environment. This partly corroborates earlier studies in which both environmental enrichment and social isolation have been shown to affect exploratory behavior in rats (Hall et al. 1998; Varty et al. 2000). Furthermore, socially isolated rats display increased anxiety (Hall et al. 1998). In the present study, corner time was assessed in the activity boxes as a measure of anxiety (Svensson et al. 2005). However, no significant differences between groups regarding corner time could be seen.

The present study has several limitations. The design of the study meant that mice were handled differently in the different groups. E and E-Ex mice were handled as little as possible, SD mice were handled once a week and ED mice were handled three times a week. These differences in handling may have influenced the outcome of this study. The group-housed mice exposed to one of the three environments were all in a single cage, so any other factors within the three group cages were not controlled for; this was not an issue for the ten singly housed mice. E-Ex mice were provided with running wheels, but the use of these wheels was not measured. Any role of physical activity in the development of atherosclerosis can thus not be excluded. Handling of mice causes a rapid increase in plasma corticosterone levels that can be seen in urine within less than an hour (Meijer et al. 2005). Thus, urine samples collected within two hours of transferring mice to the collection cage, as in the present study, may not reflect baseline levels of corticosterone.

In conclusion, the results of the present study showed increased atherosclerosis in socially isolated ApoE^{− / −} mice. The concomitant increase in plasma levels of cholesterol and triglycerides suggests that dyslipidemia is the driving force for acceleration of atherosclerosis. The reduced plasma level of G-CSF in socially isolated mice may hamper endothelial regeneration in the atherosclerotic process. No significant alterations in circulating levels of other cytokines were observed. The results of the present study are in line with human population studies where social isolation, as well as lack of social support, has been shown to increase the risk of developing coronary artery disease (Rozanski et al. 1999). Thus, the current study provides a model in which detailed mechanistic studies can be performed. The results have important implications for the conditions in which mice are housed in studies of the development of atherosclerosis.

Acknowledgements

This study was supported by the Swedish Medical Research Council, the Swedish Heart Lung foundation, funds at the Sahlgrenska University hospital (LUA/ALF) and the Lundberg foundation. This study was performed with the assistance of the SWEGENE Center for Mouse Physiology.