Exercise Training After Isoproterenol in Middle-Aged Female Rats Reversed Effects on the Heart, Without Affecting the Brain


 Introduction Females with cardiovascular disease may be more susceptible to concomitant mental problems, such as depression and cognitive decline. Exercise training has beneficial effects on the cardiovascular system as well as on mental functions. Aim of the present study was to study the effects of exercise training on heart, brain and behavior in the isoproterenol (ISO) model in middle-aged female rats. Methods Twelve months old female Wistar rats were submitted to ISO injections (70 mg/kg s.c., on two consecutive days) or received saline. One week later, rats were assigned to either exercise training (treadmill running) or control handling for five weeks. During the last 7 days, tests were performed regarding depressive-like behavior and cognitive function. Then, rats were sacrificed and heart and brains were dissected for (immuno)histochemistry.Results ISO-induced cardiac effects were eminent from cardiac fibrosis and declined cardiac function. Exercise training reversed cardiac damage and partly restored ISO-induced cardiac dysfunction. However, ISO treatment could not be associated with neuroinflammation, nor impaired hippocampal neurogenesis or neuronal function. Accordingly, no mental problems, such as cognitive decline or depressive-like behavior were observed. Actually, hippocampal microglia hyper-ramification and increased sucrose preference were observed after ISO. Exercise left neuroinflammation and behavior merely unaltered, and even reduced neuronal function. Conclusion Our data indicated that the substantial cardiac damage after ISO in middle aged female rats, and the subsequent beneficial effects of five weeks exercise training on the heart, were not reflected in changes in the brain nor in altered behavior.


Introduction
Cardiovascular disease is often associated with mental problems, such as depression and cognitive decline. Major depression was found in 15-20% of cardiovascular disease patients, while even up to 65% reported symptoms of depression [1,2]. Similarly, cognitive impairment or dementia was increased (odds ratio of 1.45) in patients with coronary artery disease [3]. Moreover, women were indicated at higher risk for heart-failure-associated depression [4,5], but higher prevalence of cognitive decline is reported for both female [6], and male patients [5]. Mental problems are no innocent bystanders in cardiovascular disease since they are associated with increased morbidity and mortality [7,8]. However, the pathophysiology of the heart-brain interaction [4,5] is far from fully understood. We hypothesize that the in ammatory response necessary for infarct healing can become derailed and re ected in the brain as neuroin ammation, associated with mental problems [9]. Although evidence supported a key role for in ammation-neuroin ammation [10], e cacy of anti-in ammatory treatment sofar is poor.
Animal studies have contributed to build up a better understanding of this heart-brain interaction. We [11] and others [12,13] reported depressive-like behavior in rodents after coronary artery ligation-induced heart failure, which was sensitive to cardiovascular- [14] as well as different types of brain-targeted treatment [15][16][17][18][19]. In addition, decline of cognitive performance was observed in this model [20]. However, the coronary artery ligation model requires major thoracic surgery, and we previously showed that the outcome in this model could be attributed to the combined effect of surgery for coronary artery ligation and effect of the ligation itself [21]. Therefore, in the present study, the isoproterenol (ISO)-induced myocardial infarction model was used [22]. Two injections of ISO with 24 hours in between increased cytokine production leading to cardiac brosis [23], left ventricular hypertrophy and dilatation, and ultimately heart failure [24]. Not many behavioral studies have been performed in this model, but reduced exploratory behavior [25] and cognitive decline [26] suggest relevant behavioral consequences. In a recent study, decreased open eld exploration and declined sucrose preference were observed shortly after ISO [27]. In these studies, an imbalanced pro-and anti-oxidant system [25] or mitochondrial function [26], rather than neuroin ammation, were observed as underlying mechanism. Moreover, all three behavioral studies were only performed in male rats.
Exercise training is well-known for its positive impact on physical and mental well-being [28], and this bene cial in uence could be associated with anti-in ammatory mechanisms [29], as well as increased brain derived neurotrophic factor (BDNF) expression [30,31]. The anti-in ammatory properties of exercise could balance the activated (neuro)in ammation and thereby improve cognition and mood [32]. Similarly, exercise may improve cognitive function and mood by counteracting a declined BDNF expression [33]. In the ISO model, bene cial effects of physical exercise were observed [34,35], but studies mainly focused on cardiac aspects. Nevertheless, exercise training could prevent upregulation of 18 cytokines after ISO [35], supporting anti-in ammatory potential of exercise training in this model. Exercise training from 12 weeks before to 7 days after ISO inhibited levels of proin ammatory TNFα and IL6, while IL10 levels increased [36]. However, effects of exercise training on mood and cognition are fairly unexplored in the ISO model.
Aim of the present study was to evaluate the effects of exercise training on heart, brain and behavior in the ISO model in middle-aged female rats.

Animals and experimental design
Forty-seven middle-aged (12 months old) female Wistar rats were obtained from the breeding colony of the University of Physical Education, Hungary. Animals were housed in groups of 2 or 3 in cages of 30*42*20 cm with sawdust as bedding. Rats were kept in the conventional animal facility of University of Physical Education, Hungary in a room with 22±2°C and humidity of 50±10%. Light was provided from 7 am to 7 pm CEST. Standard rodent chow (LT/R, Innovo Ltd., Gödöllő, Hungary) and tap water was provided ad libitum. All methods were performed in accordance with the ARRIVE guidelines. All experiments were performed in accordance with relevant guidelines and regulations/legislations. The experiments were conducted under and approved by the general license for animal experiments of the laboratory of Physical Education, University of Budapest, Hungary.
Rats were randomized to 4 experimental groups; rats were treated with isoproterenol (n=27) to induce heart lesions, or received saline injections (n=20), and after one week of recovery, halve of the survivors were subjected to ve weeks of treadmill running while the other halve received control handling (sedentary) (n=10 per group). Exploratory behavior and cognitive performance were assessed at the last 7 days of the training period. After completion of all tests, animals were anesthetized and echocardiographic measurements of cardiac function were obtained. Subsequently, rats were sacri ced, and heart and brain tissues were collected for further analyses of cardiac damage, neuroin ammation and neuronal function.

Cardiac damage
Cardiac damage was chemically induced by isoproterenol hydrochloride (C 11 H 7 NO 3 ·HCl: ISO). ISO is a non-selective β-adrenoceptor agonist that mimics the histological, physical and endocrinological events of human myocardial infarction presumably by myocardial hyperactivity induced ischemia and energy depletion [37]. Rats were injected subcutaneously with ISO (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) in a dose of 70 mg/kg dissolved in 1 ml/kg saline. Control animals received 1 ml/kg saline. Both groups received 2 injections with 24 hours in between. The ISO protocol was based on previous studies [26,38].

Tread-mill running
Both control and ISO treated groups were randomly assigned to treadmill running and sedentary groups. Running was performed on a six-lane rat treadmill (Tartonik Elektronika, Italy) with individual lanes of 12*54*13 cm. The rats were stimulated to stay on the treadmill by cardboard pushers. No electric shocks were used to motivate the animals. The training program lasted for 5 weeks, 5 times per week on each weekday. On the rst week of the training program rats were habituated to running: on the rst day, rats started with 10 minutes of running with a maximal speed of 10 m/min which was gradually increased to 30 minutes and maximal speed of 18 m/min (moderate intensity; approximately 65% of VO 2 max) by the fth day. For the following four weeks each running session lasted 30 minutes [38].

Behavior
Behavioral tests were carried out during the last 7 days of the intervention. Anxiety/depressive like behavior was assessed with open eld exploration (OF) and anhedonia (sucrose preference test; SP). Regarding cognition, short-term memory was tested in the novel object recognition (NOR) and the novel location recognition test (NLR). All tests were carried out in a quiet and clean test room (between 10 and 12 a.m.) with the same temperature and humidity parameters as the housing room.
All tests were recorded with a digital video camera (Canon Legria HFR106, Canon Inc., Tokyo, Japan) and stored on a memory card for later off-line analyses.

Open eld
Open eld exploration test was performed to assess exploratory and anxiety related behavior [11]. A round shaped arena (diameter of 80 cm) was divided into an inner circle (diameter of 32 cm; center area), and an outer annulus (wall area) by black circular lines, and surrounded by a 45 cm tall wall. Animals were placed in the arena and allowed to explore for 5 minutes. After each animal the arena was cleaned with 70% ethanol to remove smell cues. Time spent in the wall area and the number of visits into the center were obtained using Eline software (University Groningen, the Netherlands). More time in the wall area and less visits to the center were regarded as signs of anxious-depressive like behavior.

Sucrose preference
This test was performed in a subgroup of rats, randomly chosen from each experimental group. Animals remained group-housed during sucrose preference test, to avoid stress of individual housing. In the ve days habituation phase animals were given two bottles of water. At the start of the sucrose preference test, animals were supplied one bottle of tap water along with a bottle containing 1% sucrose solution. The weight of both bottles was measured. After 24 hours the position of the bottles was reversed. After again 24 hours, bottles were removed and weighed again. The difference in weight indicated uid intake.
Preference was calculated as sucrose solution intake /total intake. As rats shared experimental procedures per cage, average preference was calculated per cage, and assigned to the individual rats.
Lower preference for sucrose was regarded a sign of anhedionia.

Novel object and novel location recognition
The novel object recognition test was performed to assess short term visual memory, which depends primarily on prefrontal cortex function, while the novel location recognition test determined short term spatial memory, associated with hippocampal activity [39]. The two memory tests were combined in one protocol [39] and took place in a black box of 45*55*50 cm. The combined test consisted of 4 phases, each of 3 minutes, with 1 minute in between: 1) in the habituation phase the animal was placed in the test box and allowed 3 minutes to get accustomed to the settings, 2) in the exploration phase the rat was presented two identical objects, 3) in the novel location phase the two identical objects were presented again but one of them on a different location than in the previous phase, 4) in the novel object phase one of the two identical objects were replaced by a different object and put to the same location as in the preceding phase. Between the phases, the objects were removed and cleaned with 70% ethanol to remove smell cues. After each animal the test box and objects were also cleaned with 70% ethanol. All phases were recorded by a digital video camera (Canon Legria HFR106, Canon Inc., Tokyo, Japan). Time spent with exploring the objects were measured using Eline software (University of Groningen, the Netherlands).
Preference for the novel location or the novel object was calculated by dividing the time spent exploring the novel location or novel object by the time spent exploring both objects, while 50% indicated chance level = no recognition. Animals who did not explore the objects or only one of them were excluded at the nal statistical analysis.

Cardiac function
Cardiac function was estimated with transthoracic echocardiography in a subgroup of rats, randomly chosen from each experimental group, because of limited access to the ECHO machine. Rats were anesthetized with pentobarbital (60 mg/kg, ip), and placed in supine position on heating pads (37°C core temperature). Standard two-dimensional and M-mode long-and short-axis images at the midpapillary level were acquired using a 13 MHz linear transducer (12L-RS; GE Healthcare, Horten, Norway) connected to a commercially available system (Vivid i; GE Healthcare). Images were analyzed using dedicated software (EchoPac v113; GE Healthcare). Heart rate was obtained. From the images acquired, stroke index (stoke volume indexed to body weight), cardiac index (cardiac output indexed to body weight = stroke index times heart rate) and left ventricular ejection fraction were calculated.

Tissue collection and processing
At the end of the experiment rats were terminally anaesthetized with 6% sodium pentobarbital solution injected intraperitoneally (2 ml/kg) and perfused transcardially with heparinized (1 ml/l) 0.9% saline. Heart and brain tissues were dissected. Brain and heart tissue was immersion xated in 4% buffered formaldehyde freshly depolymerized from paraformaldehyde. After 4 days, tissue was washed in 0.01M phosphate buffered saline (PBS), dehydrated using a 30% sucrose solution, and subsequently quickly frozen in liquid nitrogen and stored at -80°C until further processing. Heart sections were paced on glass immediately after cutting, and processed for histochemical staining of collagen. For brain tissue, free oating sections were stored in 0.01M (PBS) containing 0.1% sodium azide at 4ºC till further processing for immunohistochemistry. In a subgroup of rats (n=5-6 per group), randomly chosen from each experimental group, immunohistochemistry staining was performed to visualize microglia, immature neurons (double cortin positive cells) and brain derived neurotrophic factor expression, as has been described previously [39]. (Immuno)histochemistry

Cardiac collagen
Since ISO was anticipated to cause focal myocardial infarcts, percentage collagen was used to measure cardiac damage. For that, 25 µm thick transverse slices at mid-ventricular level of the heart were stained with Sirius red (Sigma, Aldrich) and fast green as counterstaining [21]. Colour pictures were taken. Image analysis (Image Pro plus, USA) was used to measure the collagen positive (red) area and was expressed as percentage of total left ventricular tissue area.
Microglia activity was used as measure for neuroin ammation. To visualize microglia, immunohistochemical staining of ionized calcium binding adaptor molecule 1 (IBA-1) was performed, as described in detail previously [40]. Brie y, after pretreated with 0.3% H2O2 for 20 min., sections were incubated for 3 days with 1:2,500 rabbit-anti IBA-1 (Wako, Neuss, Germany) in 2% bovine serum albumin, 0.1% triton X-100 at 4°C, followed by a 1 h incubation with 1:500 goat-anti rabbit secondary antibody (Jackson, Wet Grove, USA) at room temperature. The sections were then incubated for 2 h with avidinbiotin peroxidase complex (Vectastain ABC kit, Vector, Burlingame, USA) at room temperature. Labeling was visualized by using a 0.075 mg/mL diaminobenzidine (DAB) solution activated with 0.1% H2O2. All dilutions were made in 0.01 mol/L PBS. All sections were thoroughly rinsed 4 times with 0.01 mol/L PBS between staining steps. Sections were mounted onto glass slides in a 1% gelatin solution and dehydrated through gradients of ethanol and xylol solutions. Photographs were taken from the prefrontal cortex (Pfc), the paraventricular nucleus of the hypothalamus (PVN), and the dorsal hippocampus (hippocampus; CA1, CA3, Dentate Gyrus and Hilus areas) at 200 times magni cation. Microglia morphology was analyzed (Image Pro Plus, USA) according to our previous publication [40], regarding coverage, density, cell size, cell body area and processes area. Microglia activity was calculated as cell body area/total cell size [40].

Neurogenesis
Double Cortin (DCX) staining was used to obtain a measure for hippocampal neurogenesis, as described in detail elsewhere [39]. Images of DCX stained sections of the dentate gyrus of the hippocampus were taken at 50x magni cation. The number of labeled neuronal cell bodies was counted manually by two independent researchers blinded for the experimental groups, and corrected for the length of the DG (Image Pro Plus, USA).
Brain-derived neurotrophic factor (BDNF) For brain function, brain slices were stained with Brain Derived Neurotrophic Factor (BDNF) antibody (Alomone Labs, Israel). In the different areas of the dorsal hippocampus, CA1, CA3, Dendate Gyrus and Hilus, BDNF expression was obtained as corrected optical density (Image-J) compared to an underlying reference area, as described previously [39].

Data analyses
The study has been reported in accordance with ARRIVE guidelines. All reports were performed in accordance with relevant guidelines and regulations/legislations. Data are presented as mean and standard error of mean (SEM), unless indicated otherwise. Results outside twice the standard deviation of its group were considered outliers and were excluded before analyses (maximally 1 per experimental group). Results were compared using two-way analysis of variance (ANOVA) with least square difference (LSD) post-hoc tests, with saline/ISO and sedentary/runner as factors. Association between selected parameters were measured with Pearson linear correlation. For the Novel Object /Novel Location Recognition tests, outcomes were also tested against change level (=50%), using a single sample t-test. A p-value of <0.05 was considered statistically signi cant and presented as *. Potentially relevant tendencies (p<0.1) were mentioned as well.

Results
General ISO-induced mortality was 26%, and occurred within 3 days after ISO treatment; that is before separating the groups according to sedentary/exercise intervention. Body weight at the end of the protocol was slightly (ns) higher in rats that had performed exercise, independent of saline or ISO treatment (saline sedentary: 253±7g; saline exercise: 266±9g; ISO sedentary: 243±7g and ISO exercise: 264±7g; n=10 per group).
Effects on the heart Heart weight to body weight ratio did not differ between groups (saline sedentary: 0.33±0.03%; saline exercise: 0.38±0.01%; ISO sedentary: 0.38±0.02% and ISO exercise: 0.35±0.02%). Figure 1 presents effects on cardiac damage, measured as percentage of collagen at mid-ventricular level. ISO signi cantly increased cardiac collagen, which was signi cantly reversed by exercise.
Effects on cardiac function were measured by echocardiography. Figure 2 presents the main results. Twoway ANOVA revealed a signi cant higher heart rate in exercise versus sedentary rats, which appeared most prominent in the ISO treated rats. Although analysis of stroke index did not reach statistical signi cance in the two-way ANOVA (p=0.089), the indicated reduced stroke index after exercise training (post-hoc analysis; p=0.035) may be of physiological relevance. The product of heart rate and stroke index, cardiac index, was preserved. Left ventricular ejection fraction, as measure for left ventricular function, was signi cantly declined by exercise training as well as ISO, but may be partly reversed by combined ISO plus exercise; the latter being not signi cantly lower than sedentary saline controls anymore. None of the cardiac function parameters correlated with collagen percentage.
Effects on the brain Neuroin ammation was obtained from morphological changes of microglia, indicated as microglia activity; cell body to cell size ratio. Prefrontal cortex microglia activity was not affected by ISO, nor by exercise training. Similarly, no signi cant effects were observed in the paraventricular nucleus of the hypothalamus microglia activity. Moreover, none of the parameters of microglia in these areas appeared affected by either ISO, or exercise, or the combination. Nevertheless, microglia activity in the PVN seemed reduced by about 50% after ISO. Similarly, ISO substantially reduced microglia activity in the hippocampus; an effect that was not affected by exercise (Figure 3). Overall hippocampal parameters for microglia morphology were not different between groups, as ISO increased coverage by only 5 %, but microglia cell size increased by 44%, which was mainly attributable to increased processes (47%), as cell bodies increased only 15%. Exercise training did not affect these observations.
Since the hippocampus plays a role in mood as well as cognition, effects on the different hippocampal areas were analyzed (see Table 1). What was indicated for mean values of the hippocampus, appeared merely re ected in the CA1 area. ISO decreased microglia density, and increased the size of the microglia, hence preserving coverage. Increased cell size could be attributed to increased processes, as cell body size remained unaltered; microglia hyper-rami cation. In the other hippocampal areas, coverage increased in exercise after ISO, compared to exercise after saline treatment, without signi cantly altering morphology of the individual microglia. No signi cant effects were observed on neurogenesis, nor on overall hippocampal or DG BDNF expression (Table 2). However, in the CA1 area, but not in other areas, BDNF was signi cantly decreased by exercise in ISO rats (Figure 4).

Effects on behavior
Effects on short-term memory were obtained from performance in the NOR and NLR tests ( Figure 5). For the NOR test, data of six rats were excluded because they did not meet the degree of exploration criteria (four with no exploration and 2 with exploration of only one object), resulting in 8-9 rats per experimental group. For the NLR test, four rats were excluded for only exploring one object, resulting in 8-10 rats per group. Twelve months old female rats were well capable of recognizing the novel object in the NOR test, irrespective of saline/ISO treatment or sedentary/exercise training. In the NLR test, control saline sedentary rats seemed unable to recognize the relocated object. Exercise training may improve that, as saline-treated exercise rats performed above random level. However, exercise in ISO-treated rats did not improve NLR performance.
Effects on mood were indicated by spatial exploration in the OF and anhedonia by SP ( Figure 6). Spatial exploration in the OF revealed no signi cant effects of ISO, nor of exercise. The SP test revealed a signi cant effect of ISO versus saline, a signi cant effect of exercise versus sedentary, as well as a signi cant interaction. Post-hoc analyses showed higher preference in ISO treated rats. However, whereas exercise declined preference in saline treated rats, it had no effect in ISO treated rats.

Discussion
General Cardiovascular disease is often associated with mental problems, such as depression and cognitive decline. Females seemed more susceptible to develop mental problems than males. Exercise training has bene cial effects on the cardiovascular system as well as on mental functions. Inhibition of (neuro)in ammatory processes may provide a shared mechanism. Aim of the present study was to evaluate the effects of exercise training on heart, brain and behavior in the ISO model in middle-aged female rats. As expected, ISO induced cardiac damage and declined cardiac function. However, these effects could not be associated with neuroin ammation, reduced hippocampal neurogenesis or impaired neuronal function. Accordingly, no signs of cognitive decline or depressive-like behavior were observed after ISO. Exercise training reversed cardiac damage and partly restored ISO-induced cardiac dysfunction, but left neuroin ammation and behavior merely unaltered. Hence, our data indicated that the substantial cardiac damage after ISO in middle aged female rats, and the subsequent bene cial effects of ve weeks exercise training on the heart, were not re ected in changes in the brain nor in altered behavior.

Effects of ISO
The ISO model has a long history as a method to induce focal cardiac damage and consequently cardiac dysfunction [41]. Similar to our set-up, two injections of ISO given 24 hours apart mimicked acute sympathetic stress, activating the innate immune system, resulting in focal cardiac necrosis shown by collagen deposition [23,35], that over a period of weeks developed into heart failure [24]. In contrast to Grant et al [42], other researchers indicated sex difference in the ISO model, as male rats had a lower survival rate, developed more massive necrosis and displayed slower repair compared to females [37,43]. Accordingly, when compared to our previous study in male rats of the same age [38], in female rats cardiac brosis and mortality were relatively mild. Still, cardiac brosis was eminent and caused longterm cardiac dysfunction, suggesting activation of in ammation as repair mechanism. However, these effects could not be associated with neuroin ammation (microglia activation) or associated behavioral changes, such as cognitive decline or depressive-like behavior. In fact, in contrast to our previous study in male rats [38], in female rats microglia activity in the hippocampus indicated reduced, rather than elevated neuroin ammation after ISO. Usually, injury evokes "classical" pro-in ammatory activation of microglia, with morphological changes that include shortened dendrites with increased cell body size [44].
However, alternatively activated microglia show opposite morphological changes, and are associated with non-pathological stimuli [45,46]. In our study, we observed a signi cant increase in total microglial cell size, that could be attributed to increased dendrite area of microglia cells, as cell body area remained unchanged. This nding indicated that the microglia in female rats rather converted into the alternatively activated phenotype, associated with microglia priming [44]. This microglia sensitization includes preconditioning to become hyperactive once (re)activated, and could be caused amongst others by chronic mild in ammation.
We observed higher sucrose preference, without altering OF behavior in our female rats. Comparing these observations to the referred behavioral studies in the ISO model [25][26][27] that were all performed in male rats, it may point to sex differences in behavioral responses to ISO as well. Accordingly, in contrast to the female rats in the present study, male rats showed reduced time at the wall in the OF [38]. To our knowledge, no other female rat studies on behavior are available in that regard. Nevertheless, the older studies of Wexler and coworkers [47,48] clearly indicated sex differences in the response of peripheral parameters that could be linked to behavioral changes, such as depression and cognition [49]. For instance, the corticosterone response during the rst week after ISO appeared much higher in female rats than in male rats [48], suggesting different hypothalamic-pituitary-adrenal (HPA) axis activity.
Compared to males, female rats show a more robust HPA axis response, as a result of circulating estradiol levels which elevate stress hormone levels during non-threatening situations as well as after stressors [50], with distinct behavioral consequences (reviewed by [51].

Effects of exercise
Exercise training is well recognized for it is bene cial effect on physical as well as mental health [28]. Indeed, exercise training reversed the ISO-induced scar formation in the heart, thereby partly restoring left ventricular function in our female rats. However, in our previous study in male rats, no cardiac effects of exercise were observed [38]. Actually, in the only study we are aware of that started endurance training after-, instead of, before ISO, exacerbated cardiac damage was reported after four weeks of training [52].
Differences with the present study may be attributed to the age; 8-10 weeks versus 12 months, and sex of the rats, male versus female rats, as well as the point that the exercise training started two days after ISO, compared to the one week delay in our study. Especially the latter aspect could provide an explanation.
Our study was set-up to intervene when cardiac damage was eminent, at one week after ISO, whereas the study of Jazi et al [52] started exercise during the in ammatory healing phase, with no cardiac brosis present yet [35]. In the present study exercise intervention could then have been started when the acute healing process was merely complete, hence interfering with the more chronic in ammatory state; timing is crucial.
Exercise training is indicated to exert it bene cial effects, amongst others, by its anti-in ammatory action [29]. Since we hypothesized that depressive-like behavior [9] as well as cognitive decline [21] in association with myocardial infarction, are at least in part attributable to a derailed peripheralneuroin ammatory response, we anticipated increased neuroin ammation after ISO, which would be reversed by exercise training. Since we did not observe increased neuroin ammation, measured by microglia activity, nor depressive-like behavior or cognitive decline after ISO, accordingly, we were unable to determine potential reversal by exercise training. Although neuronal function, as indicated by altered BDNF levels, was not declined after ISO, results seemed to indicate a decline after exercise in ISO rats.
These observations are in contrast to our expectations; that were a decline after ISO, that could be reversed by exercise. Brain-derived neurotrophic factor (BDNF) has an important role in regulating maintenance, growth and survival of neurons. It is long known that exercise training increases BDNF expression in the hippocampus [53], playing a bene cial role in learning and memory [54]. However, circulating BDNF also increases after exercise, which could be mainly attributed to the increased brain expression (hippocampus and cortex) [55]. Moreover, exercise training after myocardial infarction in male rats also induced BDNF in skeletal muscle and the non-infarct area of the left ventricle, which may contribute to improvement of muscle dysfunction as well as cardiac function [56]. The declined BDNF expression in the hippocampus (CA1 area) in the present study, in contrast to the increase seen in male rats in our previous study [38], would then be contradictory to the improved cardiac aspects after exercise in ISO rats, and may well related to the female sex of the rats.

Timing
An important aspect of intervention is timing. As we aimed to intervene with the ISO-induced cardiac disease when cardiac brosis was eminent, we started exercise training one week after ISO. According to Alemasi et al. cardiac in ammation is progressing 3 days after ISO, while cardiac brosis seemed not present before seven days after ISO. All referred studies on exercise intervention that had started exercise training before ISO, showed improvement of cardiac parameters, potentially interfering with the in ammatory and/or metabolic responses early after ISO. Our previous study on exercise intervention starting one week after ISO in male rats, showed no effect on cardiac collagen percentage [38]. However, the one study that started exercise training two days after ISO [52], that is within the in ammation phase [35], even showed deterioration of cardiac outcome. These results may indicate that optimal timing of exercise intervention in relation to the time course of the in ammatory/metabolic response to ISO seemed crucial. Since in the present study, we showed a positive effect of exercise on the heart, without clear consequences for brain function, this may add to sex differences in the ISO-induced pathology.
Striking nding was the indicated microglia hyper-rami cation after ISO in female rats. Acute ISO administration produced tachycardia associated with relative ischemia due to imbalance between increased myocardial oxygen demand and reduced coronary blood supply [24]. Another study, evaluating microglia activity four days after an acute imbalance between oxygen supply and demand due to anesthesia, also showed microglia hyper-rami cation, that was explained as a prolonged state of priming [57]. Hence, it could well be that one week after ISO, the acute response to ISO is merely over, leaving only limited scope for effects of exercise training. The microglia hyper-rami cation six weeks after ISO may then represent long-term alertness, that is still present after exercise intervention. Therefore, in our opinion, in order to obtain bene cial effects of exercise intervention after ISO, timing of intervention with respect to the time course of cardiac pathology, is crucial.

Limitations
Although the set-up was chosen carefully, each study has its limitations. Since we used in ammationneuroin ammation-depression/cognitive decline as our working hypothesis, in hindsight it would have been wise to have collected timed blood samples, to further elucidate on the underlying mechanism of our ndings based on plasma markers, TNFα and IL6 [36], or lipocalin [58]. Although the initial event, the cardiac damage, as well as the ultimate behavioral consequences were measured in all rats, the in between factors were only obtained in subgroup of rats, limiting power of the conclusions concerning mechanisms.
The dose of ISO was selected based on effect (mortality) in our previous study [38]. Actually, this dose appeared in between the medium-dose and high-dose of ISO as reviewed by Nichtova et al. [24], with potentially mixed effects described for these dose ranges.
Since individual housing is regarded a stress factor for the rats, we decided to keep the rats group-housed for the sucrose preference test, and take the average value per cage (containing rats belonging to the same experimental group) to assign that to the individual rat. From a methodological point of view this is suboptimal, as it limits obtaining actual individual values.
Finally, although the ISO model is mainly used to induce cardiac damage, one has to keep in mind that ISO-effects may not be limited to the heart, as indicated by changes in adrenal gland weight and corticosterone levels as well [48].

Conclusion
Our data indicated that ve weeks of exercise training in middle-aged female rats, starting when ISOinduced cardiac damage was eminent, decreased cardiac damage and partly restored cardiac dysfunction. However, since ISO could not be associated with neuroin ammation, nor with depressive-like behavior and cognitive decline, anticipated reversal by exercise could not be observed. Hence, our data indicated that the substantial cardiac damage after ISO in middle aged female rats, and the subsequent bene cial effects of ve weeks exercise training on the heart, were not re ected in changes in the brain nor in altered behavior.   Behavior in the open eld test (n=10 per group), as measured time spent at the wall area and the number visits into the center area (upper 2 panels, respectively), and sucrose preference (lower panel; n=4-6 per group), in saline or isoproterenol (ISO) treated rats, under sedentary conditions or after ve weeks of exercise training. *: signi cant difference between groups (p<0.05).