HEART FAILURE IMPAIRS MOOD AND MEMORY IN MALE RATS AND DOWN-REGULATES THE EXPRESSION OF NUMEROUS GENES IMPORTANT FOR SYNAPTIC PLASTICITY IN RELATED BRAIN REGIONS

Abstract

Chronic heart failure (HF) is a serious disorder that afflicts more than 26 million patients worldwide. HF is comorbid with depression, anxiety and memory deficits that have serious implications for quality of life and self-care in patients who have HF. Still, there are few studies that have assessed the effects of severely reduced ejection fraction (≤40%) on cognition in non-human animal models. Moreover, limited information is available regarding the effects of HF on genetic markers of synaptic plasticity in brain areas critical for memory and mood regulation. We induced HF in male rats and tested mood and anxiety (sucrose preference and elevated plus maze) and memory (spontaneous alternation and inhibitory avoidance) and measured the simultaneous expression of 84 synaptic plasticity-associated genes in dorsal (DH) and ventral hippocampus (VH), basolateral (BLA) and central amygdala (CeA) and prefrontal cortex (PFC). We also included the hypothalamic paraventricular nucleus (PVN), which is implicated in neurohumoral activation in HF. Our results show that rats with severely reduced ejection fraction recapitulate behavioral symptoms seen in patients with chronic HF including, increased anxiety and impaired memory in both tasks. HF also downregulated several synaptic-plasticity genes in PFC and PVN, moderate decreases in DH and CeA and minimal effects in BLA and VH. Collectively, these findings identify candidate brain areas and molecular mechanisms underlying HF-induced disturbances in mood and memory.

INTRODUCTION

Chronic heart failure (HF) is a serious cardiovascular disorder that afflicts more than 26 million patients worldwide [1]. In the United States, it is estimated that 6.2 million people suffer from HF and the prevalence is projected to increase by 46% by 2030, more than doubling the cost of health care from $30.7 to $68.8 billion [2]. Once established, HF morbidity, prognosis and mortality are largely influenced by the degree of neurohumoral activation, which is a pathophysiological mechanism that involves increased centrally-driven sympathetic activity and elevated circulating levels of neurohormones [3–5]. It is now becoming well-accepted that, in addition to neurohumoral activation and the associated systemic cardiac, renal and vascular complications, HF is comorbid with depression, anxiety and memory deficits, both in human patients [6–10] and experimental animal models [11, 12]. These mood and cognitive disturbances have serious implications for quality of life and self-care in patients who have HF because hospital readmission and mortality rates are significantly higher in HF patients that have depression and/or cognitive impairments compared to HF patients without these comorbidities [9, 13–15].

It is critical to identify the neurobiological mechanisms and substrates that underlie HF-induced mood and cognitive disturbances HF in order to identify treatments that may improve quality of life and decrease mortality in this prevalent disease. In HF patients, neurocognitive impairments, including learning and working memory deficits, delayed recall, depression, and anxiety are correlated with significant cerebral grey matter loss and damage in brain areas relevant to cognitive function, including hippocampus, amygdala and prefrontal cortex (PFC) [16–20]. However, the mechanisms through which HF produces mood and cognitive impairments are not known. This question is difficult to address in human patients due to the influence of genetic, environmental and psychosocial variables that cannot be controlled. As a result, it is critical to develop valid animal models of mood and cognitive disturbances in HF in order to begin to identify underlying mechanisms. Despite evidence that cognitive performance is worse in HF patients with reduced ejection fraction than in HF patients with preserved cardiac function [21, 22], there are limited studies that have assessed the effects of severely-reduced ejection fraction (≤ 40%) on cognition in non-human animal models. Moreover, very limited information is available regarding the effects of HF on genetic and molecular markers of synaptic plasticity in brain areas critical for memory and mood regulation. This information is necessary to begin to understand the mechanisms through which HF impairs mood and memory. As a result, in the present study we aimed to investigate the effects of HF that is characterized by severe reductions in ejection fraction on behavioral measures of mood and memory that are known to be dependent on the hippocampus, amygdala and PFC. To this end, we induced HF using a well-established ischemic HF rat model [23–26] and then tested the rats in a battery of behavioral tests to assess anhedonia and anxiety (sucrose preference test [SPT] and elevated plus maze [EPM], respectively) as well as spatial working memory (spontaneous alternation; SA) and emotional long-term memory (one-trial inhibitory avoidance; IA). Moreover, we used an unbiased quantitative real time polymerase reaction (qRT-PCR) approach to identify the effects of HF on the simultaneous expression of 84 genes critical for synaptic plasticity in brain areas critical for mood and memory, including dorsal and ventral hippocampus (DH and VH, respectively), basolateral and central amygdala (BLA and CeA, respectively), PFC and PVN.

MATERIALS AND METHODS

Subjects

Male Wistar rats (5 wks-old, 180–200 g) were purchased from Envigo Laboratories (Indianapolis, IN) and housed in Optirat cages (Animal Care Systems, Centennial, CO, USA) on a 12 h light-dark cycle (lights on at 8:00 am) with ad libitum access to pelleted food and water in their home cages unless otherwise stated. Rats were housed individually following surgery and during behavioral testing. All procedures were performed in compliance with the NIH guidelines for the care of laboratory animals and approved by the Georgia State University Institutional Animal Care and Use Committee.

Induction of heart failure (HF)

HF was induced by coronary artery ligation as described previously [23–25]. Briefly, animals (4 days after arrival) were anesthetized with isoflurane (2–4%) and intubated for mechanical ventilation. A left thoracotomy was performed and the heart exteriorized. The ligation was placed on the main diagonal branch of the left anterior descending coronary artery. Buprenorphine SR-LAB (0.5 mg/kg sc; Zoo Pharm, Windsor, CO, USA) was given immediately after surgery to minimize postsurgical pain. In our hands, the mortality rate in this model is ~20% and 10–20% of rats than undergo this procedure develop mild or no functional HF (i.e., ejection fraction > 40%). Sham control animals underwent the same procedure with the exception that the coronary artery was not occluded. Transthoracic echocardiography (Vevo 3100 systems, Visual Sonics, Toronto, ON, Canada) was performed 4 weeks after surgery under light isoflurane (3–4 %) anesthesia. Previous studies have shown that the degree of HF following the myocardial infarction remains consistent over time for at least 12 weeks [27–29]. The measurements of the left ventricle internal diameter and its posterior and anterior walls in systole and diastole, and the ejection fraction and fractional shortening were obtained throughout the cardiac cycle from the short-axis motion imaging mode.

Behavioral measures of mood and memory

Behavioral testing was started 5–10 weeks after the induction of HF with the measures of mood in one cohort (Cohort 1) and memory in another (Cohort 2). All tests were selected because they assessed mood and cognition in a relatively minimally stressful manner and all behavioral testing and scoring were conducted by an experimenter blind to Sham/HF condition. Approximately 5 weeks after surgery, Cohort 1 was tested in SPT. SPT is a commonly used measure of anhedonia, which is a major symptom of depression [30]. SPT relies on the natural preference of rodents for sweet foods and solutions; a decreased preference for a sweet solution versus tap water is assumed to reflect depression-like behavior [31, 32]. SPT was assessed in home cages with continuous ad libitum access to standard chow. Rats were first habituated to sucrose (1% wt/vol) by removing their water and giving them access to two bottles that both contained 1% sucrose for 24 h. The next day they were adapted to the preference test by giving them access to one bottle containing 1% sucrose and one containing tap water for 18 h (2 pm to 8 am). For the preference test the following day, all fluids were removed for 12 h (8 am to 8 pm) and the rats were then given access to one bottle of tap water and one bottle of 1% sucrose for 12 h. To avoid biases in drinking behavior based on the location of the bottle (i.e., left versus right), the position of the sucrose and water bottles were counterbalanced across the adaptation and preference test days [33]. Sucrose preference was calculated by dividing the amount of sucrose ingested by the total amount of fluid consumed on the sucrose preference testing day. The experiment started with 12 Sham rats and 8 HF rats and then the data from four rats (Sham n = 3; HF n = 1) were excluded due to excessive leaking from bottles.

The 12 Sham rats and 8 HF rats in Cohort 1 were then tested in the elevated plus maze (EPM) approximately 10 days later. They were handled for 3 min/day for 2 days before the EPM test and all testing was conducted during the light phase between 8:00 am and 12:00 pm. The EPM is a well-established behavioral test of anxiety [34, 35] that relies on the natural tendency of rodents to prefer dark enclosed spaces and fear height and open spaces, and their innate motivation to explore novel environments [36, 37]. The plus-shaped stainless-steel apparatus was elevated 50 cm above the floor and composed of four arms (50 cm x 12 cm). Two arms were surrounded by a 40 cm high black acrylic wall (i.e., closed arms) and two arms did not have walls (i.e., open arms). The rats were placed in the central square (12 × 12 cm) facing the closed arm, their behavior was video-recorded for 5 min, and the video-recordings were analyzed by a trained observer. The apparatus was cleaned with 70% ethanol after each rat was tested. An arm entry was defined as entering an arm with all four paws. Measures of anxiety included the percentage of time spent in the open arms [(time spent in open arm/total time spent on the maze) x 100] and the percentage of entries into the open arms [(open arm entries/ total arm entries) x 100]. Risk assessment was also used as a measure of anxiety [38]. Specifically, anxious animals make more stretch attend attempts, which were operationally defined as when a rat stretched its body towards an open arm after having been motionless, and then retracted [38, 39].

IA and SA were assessed in Cohort 2 approximately 10 weeks after surgery. Due to a delay in securing IACUC approval for the IA task, we had more rats that participated in SA then IA and thus at the end we added a cohort in which we measured IA but not SA, resulting in 17 Sham rats and 13 HF rats for SA and 14 HF rats for IA. All rats were handled for 3 min/day for 2 days before SA and IA training, all testing was conducted during the light phase between 8:00 am and 12:00 pm, and the apparatuses were cleaned with 70% ethanol after each rat was tested. Rats were first assessed in SA, which is a test of spatial working memory [40, 41]. The underlying assumption is that rats must remember their visits to previous locations in order to successfully alternate between spatial locations. This is supported by the finding that removing extra-maze cues or increasing the interval between arm choices decreases alternation scores [40, 41]. To assess SA, eat rat was placed in a Y-maze composed of three equally spaced arms (120°; 60 cm x 17.5 cm). The floor of each arm was composed of stainless steel (3.5 cm wide) and the top (14 cm wide) was covered with black Plexiglas. All of the rats were placed in the same starting arm of the Y-maze and allowed to explore the maze freely for 8 min. An alternation was defined as entering three different arms consecutively, and a percent alternation score was computed by dividing the number of alternations made by the number of arms entered minus two and then multiplying the resulting quotient by 100. The number of arms entered was used as a measure of activity.

The rats were then given training in the one-trial IA task approximately 15 days after SA testing. IA is commonly used to measure emotional, long-term memory [42, 43]. Each rat was placed into a trough-shaped apparatus (91 × 15 × 20 × 6.4 cm) that was separated into two compartments by a retractable guillotine door. The dark compartment (60 cm long) had a metal floor through which shock could be delivered. A 15-W light was placed over the lighted compartment (31 cm long) and was the only source of illumination in the room. For training, each rat was placed into the lighted compartment facing away from the retractable door and the door was then opened after the rat turned toward the door or after 12 s had passed. When the rat entered the dark compartment with all four paws, it was given a footshock (0.5 mA, 1 s) and then removed from the apparatus and returned to its home cage. The current level was assessed regularly throughout the course of the experiment with a digital multi-meter (AstroAI AM33D). For the retention test 24 h later, each rat was placed in the lighted compartment, the retractable door was lowered, and the latency (s) to cross over to the dark compartment (maximum 600 s) was recorded and used as a measure of retention. Footshock was not delivered on the retention test. Compared to other emotional memory tasks that depend on the hippocampus, amygdala and PFC, such as classical fear conditioning and spatial water maze, IA is minimally stressful because rats are exposed to only one training trial and the footshock is very brief.

Unbiased qRT-PCR array approach to assess the impact of HF on 84 molecular markers of synaptic plasticity

The effect of HF on synaptic plasticity-associated gene expression was assessed in DH, VH, PFC, CeA and BLA because of their roles in mood and memory. Specifically, manipulations that impair DH or PFC function attenuate sucrose preference [44–46], which is consistent with evidence indicating that damage to these areas produces depression in humans [16, 17]. EPM is dependent on the integrity of CeA [47], BLA [48] and PFC [49, 50], which is congruent with a wealth of evidence indicating that these brain areas are involved in regulation of mood [51–55]. SA is a task that depends on the integrity of the DH [56, 57] and PFC [58, 59] and IA requires an intact DH [60–62], VH [63], BLA [64, 65], CeA [66], and PFC [63, 67, 68]. Approximately 10 weeks after surgery, a different group of rats (Sham n = 5; HF n = 5) were anesthetized deeply with pentobarbital (Euthasol, Virbac, ANADA #200–071, Fort Worth, TX, USA, 390 mg/kg, i.p.), decapitated quickly with a guillotine, and their brains dissected, frozen in 2-methylbutane (Fisher Scientific, USA) at −40°C and then stored at −80°C. To collect micropunches, brains were allowed to reach −20°C and then mounted in a cryostat (CM3050S; Leica Microsystems, Wetzlar, Germany) and brain slices (250 μm) were obtained using the following stereotaxic coordinates from the Paxinos and Watson atlas (infralimbic and prelimbic PFC, 3.7 to 2.2 mm; PVN, −1.4 to −2.12 mm; BLA and CeA, −1.8 to −3.14 mm; DH, −2.8 to −4.16 mm and VH, −4.8 to −6.04 mm from bregma) [69]. A 2.0 mm stainless-steel punch was used to collect punches from the PFC (n=4 midline punches/rat obtained from 4 different sections), whereas a 1.0 mm stainless-steel punch was used to collect punches from the PVN (n=3 midline punches/rat obtained from 3 different sections), DH (n=16 punches/rat obtained bilaterally from 4 different slices), VH (n=16 punches/rat obtained bilaterally from 4 different slices), BLA (n=10 punches/rat obtained bilaterally from 5 different slices) and CeA (n=10 punches/rat obtained bilaterally from 5 different slices) (see Figure 6). For each brain region, tissue from rats was pooled together for each treatment condition and kept on dry ice until RNA extraction. RNA extraction and isolation were performed using the miRNAeasy Mini kit (Qiagen, Hilden, Germany) and the QIAzol Lysis Reagent (Qiagen) following manufacturers’ protocols. The pooled RNA concentration was measured using NanoDrop One (Thermo Scientific, Waltham, Massachusetts, USA) and was in the range of 115 − 220 ng/μL prior to cDNA synthesis. cDNA synthesis was performed using the RT2 First Strand Kit (Cat. No. 330401, Qiagen) and the SimpliAmp Thermal Cycler (Applied Biosystems, Thermo Fisher Scientific) following the manufacturers’ protocols. The synthesized cDNA from the six brain areas and two treatment conditions was run in three independent gene array reactions using the RT² Profiler™ Rat Synaptic Plasticity PCR Array (Cat. No. 330231, Qiagen; i.e., 36 arrays in total) on a Roche Lightcycler 96 (Roche Life Science, Indianapolis, IN, USA). This kit measures the expression of 84 genes critical for synaptic alterations implicated in memory and mood, including immediate early genes (IEG)s, genes implicated in long-term potentiation (LTP) and long-term depression (LTD), cell adhesion and extracellular matrix molecules, CREB cofactors, and neuronal receptors (plus five housekeeping genes and two negative controls).

Figure 6. Anatomical location of micro-punches used for gene expression studies.

Schematic representation displaying the location of the punches in the PFC, PVN, BLA, CeA, DH and VH performed for the tissue sample collection for the array and qRT-PCR. Brain coronal sections were obtained from Swanson, L. W. (2004) Brain maps: structure of the rat brain, 3rd edition [143]. The grids represent 1 mm in each vertical and horizontal orientation.

Validation of gene array results

To validate the array findings, qRT-PCR was conducted using specific 10x QuantiTect primers diluted in 1.1 mL TE pH 8.0 (Qiagen). The same cDNA from for the array was used for the validation. The genes that were analyzed were Adcy1 (QT00386421), Arc (QT00373086), Homer1 (QT00391111), Mapk1 (QT00190379), and Nr4a1 (QT00181048). B2m (QT00176295) was used as the reference gene. These genes were selected based on their common involvement in synaptic plasticity, and due to the fact than they showed variable responses in the array study (i.e., widespread or selective changes across brain regions, and up- or down-regulated). All individual qRT-PCR reactions (brain region, primer and condition) were performed in triplicate.

Statistical analyses

Statistical analyses were performed using IBM SPSS Statistics 25 (IBM, Armonk, New York) and GraphPad Prism 7 (GraphPad Software, San Diego, CA). Data points that were two standard deviations from the mean were considered outliers and excluded from further analyses. This resulted in the removal of two retention latency data points from HF rats and consequently the training latencies from these rats were also excluded. Outlier detection programs in Prism and SPSS corroborated that these two data points were outliers and no other data points were identified as outliers. Given that we predicted a priori that HF surgery would impair heart function, mood and memory, unpaired one-tailed t-tests were used to assess differences between Sham and HF rats in the echocardiographic parameters and behavioral tests. Behavioral data are expressed as means ± standard error of the mean (SEM), and an alpha level of 0.05 was used as the criterion for statistical significance.

We used the Multi Ontology Enrichment Tool (MOET) to explore functional categories of genes that were altered in HF within specific brain regions. MOET provides an unbiased analysis of groups of genes within each brain region that are functionally related, co-occur and are statistically enriched in a given gene population. To identify functional classes of genes that are statistically over-represented amongst the genes that are differentially expressed during HF, MOET uses the Gene Ontology (GO) consortium, which provides detailed annotations of genes with respect to their known or putative molecular function, cellular localization and biological processes [70]. MOET provides a quantitative measurement of enrichment based on well accepted statistical methods including hypergeometric distributions to calculate p-values and correct the p-values with Holm-Bonferroni corrections and presents the number of genes involved in different annotations and the corresponding significance level in disease, pathway, biological, cellular and molecular ontologies. Four numbers are used to calculate each p-value

$$\mathrm{p}=\frac{\left(M\mathrm{choose}k\right)\left(N-M\mathrm{choose}n-k\right)}{N\mathrm{choose}n}$$

where n is the number of genes in our list, N is the number of genes in the reference population, k is the number of genes annotated with this item in our list and M is the number of genes annotated with items in the reference population. The p-value was considered significant when p<0.01.

RESULTS

HF rats had severely reduced ejection fraction

Figure 1 shows representative echocardiographic images from a Sham and HF rat and the mean heart function values obtained are shown in Table 1. Compared to Sham, HF rats showed a significant increase in the left ventricle diameter and volume during systole and diastole (p<0.0001) and a significant reduction in the left ventricle posterior and anterior thickness during systole (p<0.0001). Also, the HF rats exhibited a significant reduction in the percentage of fractional shortening and ejection fraction (p<0.0001), but no significant difference in heart rate (HR, p>0.05).

Figure 1. HF rats displayed severely reduced left ventricular function.

Representative echocardiograph images of the movement-mode, short-axis view from a Sham (a) and a Heart Failure (HF) rat (B). AW, anterior wall; PW, posterior wall; LV, left ventricle; LVID, d/s: left ventricle internal diameter in diastole and systole. Images were taken 4 weeks after the myocardial infarction surgery in 9 week-old rats.

Table 1.

Summary data of echocardiography measurements of left ventricular parameters obtained from Sham and HF rats.

	Sham	HF
LVID,d (mm)	7.98 ± 0.18	10.68 ± 0.20****
LVPW,d (mm)	2.13 ± 0.11	2.02 ± 0.17
LVAW,d (mm)	1.97 ± 0.05	2.02 ± 0.12
LV Vol,d (μL)	347.34 ± 17.05	659.89 ± 27.97****
LVID,s (mm)	3.84 ± 0.17	9.19 ± 0.21****
LVPW,s (mm)	3.41 ± 0.13	2.18 ± 0.16****
LVAW,s (mm)	3.49 ± 0.08	2.62 ± 0.15****
LV Vol,s (μL)	68.35 ± 6.69	476.95 ± 24.40****
EF %	81.02 ± 1.39	28.08 ± 1.55****
FS %	52.19 ± 1.50	14.08 ± 0.84****
HR (bpm)	288.88 ± 7.45	276.87 ± 6.07

Values (collected 4 weeks after the myocardial infarction surgery) are expressed as means ± SEM; Sham n = 34; HF n = 27. LVID, d and s: left ventricle internal dimension during diastole and systole; LVAW, d and s: left ventricle anterior wall thickness during diastole and systole; LVPW, d and s: left ventricle posterior wall thickness during diastole and systole; LV Vol, d and s: left ventricle volume during diastole and systole; EF, ejection fraction; FS, fractional shortening; HR (bpm), heart rate (beats per minute).

^****p<0.0001 vs. Sham, unpaired t-test.

HF rats displayed signs of increased anxiety

Our results show that HF rats did not have a statistically significant lower preference ratio for sucrose (Sham: 85.32±3.70% vs HF: 67.02±9.16%, p = 0.05 vs Sham; Fig. 2A) than Sham rats on the preference test and did not differ in the total amount of liquid consumed (Sham: 55.99±12.33g vs HF: 68.57±11.08g, p > 0.18 vs Sham; Figure 2B). HF rats displayed signs of anxiety in the EPM test. Specifically, HF rats spent less per cent time in the open arms of the maze (Sham: 11.13±2.47% vs HF: 5.33±1.56%, p = 0.048 vs Sham; Fig. 3A), made a comparable per cent number of entries into the open arms (Sham: 23.02±3.97% vs HF: 13.70±3.42%, p = 0.057 vs Sham; Fig. 3B) and made more stretch attend attempts than Sham rats (Sham: 3.00±0.45% vs HF: 4.00±0.51%, p = 0.041 vs Sham; Fig. 3C). HF rats entered fewer total arms (i.e. open and closed arms) than SHAM rats (Sham: 14±0.827 vs HF: 11±1.264, p = 0.013).

Figure 2. HF did not produce anhedonia.

(A) In the sucrose preference test, HF rats (n=7) did not have a significantly decreased preference for sucrose on the preference testing day relative to water, (B) and the two groups did not differ in the total amount of liquids consumed. P≥0.05 vs. Sham, unpaired t-tests.

Figure 3. HF increases anxiety and risk assessment.

(A) In the EPM, HF rats (n=8) spent significantly less per cent time in the open arms, (B) did not make fewer per cent entries into the open arms and (C) made more stretch attend attempts compared to Sham rats (n=12). *p<0.05 vs. Sham, unpaired t-tests.

HF rats had impaired spatial working memory and emotional, long-term memory

In the SA task, HF rats made fewer alternations between arms (Sham: 64.53±2.54% vs HF: 55.35±3.97%, p = 0.026 vs Sham; Fig. 4A), indicative of impaired spatial working memory. Although HF significantly disrupted the sequence of arm entries (i.e., percent alternation), it did not affect the number of arms the rats explored in the maze (Sham: 17.00±0.87 entries vs HF: 17.00±0.78 entries, p= 0.4 vs Sham; Fig. 4B), suggesting that impaired alternation was not due to changes in activity levels.

Figure 4. HF impairs spatial working memory without affecting activity levels.

(A) In the SA test, HF rats (n=13) made significantly fewer alternations between arms compared to Sham rats (n=17), but (B) entered a comparable number of arms. *p<0.05 vs. Sham, unpaired t-tests.

On the IA retention test, HF rats displayed significantly shorter latencies to enter the compartment in which they had previously received shock than Sham rats (Sham: 189.60±58.03% vs HF: 65.41±10.52%, p = 0.025 vs Sham; Fig. 5A), a result that is indicative of impaired emotional, long-term memory. Of note, rats with HF did not have shorter latencies than Sham rats on the training day (Sham: 6.58±0.89 s vs HF: 6.05±1.17 s, p>0.05 vs Sham; Fig. 5B), suggesting that the HF-induced decreases in the retention latencies were not due to hyperactivity.

Figure 5. HF impairs emotional, long-term memory.

(A) In the IA task, HF rats (n=12) entered the dark/shock compartment more quickly than Sham rats (n=17) on the retention test, (B) but their latencies did not differ during training. *p<0.05 vs. Sham, unpaired t-tests.

HF downregulates the expression of several genes involved in synaptic plasticity in the PFC, PVN, CeA and DH

To begin to identify potential molecular mechanisms underlying the HF-induced changes in mood and memory that we observed, we determined whether HF impacted expression of genes involved in synaptic plasticity in brain regions known to be critical neuronal substrates for mood and memory (i.e., PFC, DH, VH, BLA, and CeA). We also included the PVN, a region that is well-established to play a major role in neurohumoral activation in this disease state, in which altered synaptic function and plasticity were described previously [26, 71–74]. Figure 6 shows the size and specific location in the brain where the punches were obtained.

Given the extensive number of genes sampled in the array, and in order to prioritize for further validation and interpretation those genes that were more robustly affected, we set a qualitative and arbitrary threshold of a down- (≤ 40% change) or upregulation (≥ 40% change) in expression. Using this criterion, we generated a heatmap graph (Fig. 7A) that exhibits the profile of genes whose expression were robustly upregulated (≥ 40% change, in blue range color), downregulated (≤ 40% change, in red range color), or showed no changes (white rectangles) in expression when comparing HF to sham rats across the six different brain regions. Figure 7B provides a summary of the main changes observed by brain region. Overall, the results showed that HF downregulated several synaptic plasticity-related genes, with ~ 87% of genes showing a downregulation (≤ 40% in expression) in at least one brain region. Conversely, increased expression (≥ 40% in expression) was observed only in ~5% of genes tested. The PFC and PVN exhibited the largest number of down regulated genes during HF, with ~80% and ~54% genes decreased, respectively. A milder change was observed in the DH and CeA, where HF downregulated ~20% of genes tested compared to Sham rats. Lastly, HF had minimal effects on synaptic plasticity genes in the BLA and VH, with less than 10% of genes being altered (Fig. 7B). Interestingly, the few genes that showed increases were mostly (3/4 genes, 75%) restricted to the VH.

Figure 7. HF significantly alters synaptic plasticity-related gene expression.

(A) Heatmap graph showing the level of expression of different genes regulated by HF in relation to sham rats’ gene expression. Downregulation below 40% change is shown in the red range, upregulation above 40% change is shown in the blue range and the expression of genes that did not change below or above 40% is shown in white. Adcy1, Arc, Homer1, Mapk1 and Nr4a1, the genes used for the qPCR validation, are in bold. (B) Bar graph showing the percentage of genes that were upregulated (blue), downregulated (red), or did not change (white) during HF in PFC, PVN, DH, CeA, BLA and VH. (C1-C4) Heatmap graphs depicting clusters of genes that were upregulated (blue), downregulated (red) or not affected (white) due to heart failure. Clusters are organized according to genes that changed with HF vs. Sham in four to five brain areas (C1); three brain areas (C2); two brain areas (C3) or only one brain area (C4). Categories written in: red represents Egr family; green represents Gria family; purple represents Grm family; orange represents kinases and phosphatases related genes.

HF downregulated (≤ 40% in expression) several important groups of genes, including IEGs (activity-regulated cytoskeletal -associated protein [Arc; aka Arg3.1], the Egr family of zinc fingered transcription factors, and Jun), genes encoding for glutamate receptors, predominantly metabotropic receptors (Grm1, −2, −3, −4, −5, −7), but also ionotropic NMDA receptor subunits (Grin2a and Grin2b). HF also downregulated (≤ 40% in expression) various types of protein phosphatases (Ppp1cc and Ppp2ca) and protein kinases (Prkca, Prkcg and Prkg1) in the PFC and PVN. Given the large number of brain regions sampled coupled with the large number of genes present on the synaptic plasticity array, we clustered genes that exhibited HF-induced changes in gene expression by the number of brain regions in which those changes were observed. The rationale for this clustering was to highlight synaptic plasticity genes that were under coordinate regulation across brain regions implicated in mood, memory, and neurohumoral activation. Figure 7C depicts genes grouped according to the number of brain areas in which they were impacted, with genes changing in four to five brain areas grouped in Cluster 1 (C1), three in Cluster 2 (C2), two in Cluster 3 (C3) and only one brain area in Cluster 4 (C4). These clusters also potentially indicate synaptic plasticity machinery that operate in concert within, and across, different brain regions, or exhibit brain region-specific changes. The genes whose expression changed in most of the brain regions tested during HF included Arc, Adcy1 (type 1 adenylyl cyclase), Grm2 and Grm5, and the proto-oncogene serine/threonine-protein kinase Pim-1 (Pim1; see C1). Interestingly, another gene directly associated to metabotropic glutamate receptor (mGluR) function and synaptic plasticity, Homer1 (Fig. 7A, 7C2), was also downregulated (≤ 40% in expression) during HF in the PFC, PVN and CeA. Cyclic adenosine monophosphate (cAMP) responsive element binding protein 1 gene (Creb1), a gene critically linked to vascular dementia, learning, memory and cognitive abilities [75–77], was also reduced by HF in the PFC, PVN and DH (Fig. 7A, 7C2).

We then used the Multi Ontology Enrichment Tool (MOET), which provides an unbiased analysis of groups of genes within each brain region that are functionally related, co-occur and are statistically enriched in a given gene population. MOET uses the Gene Ontology (GO) consortium, in order to identify functional classes of genes that are statistically over-represented amongst the genes that are differentially expressed during HF [70]. For this analysis, we focused on the PFC, DH and CeA, which are the three brain regions that showed the most significant changes in gene expression and are implicated in the measures of mood and memory used in our experiments (see Fig. 8; results from the PVN and VH are shown in Supplementary Figure 1. Note that MOET was not applied to BLA given that only 1 gene changed in this region). We highlight two major groups of GOs related to (1) biological, cellular and/or molecular pathways/functions (left graphs, 8A, C and E) and (2) disease-related processes (right graphs, 8B, D and F). For simplicity and relevance, only the GO subcategories in which the gene counts represented 50% or more of the total genes associated with that GO that were down- or upregulated during HF, and that were considered statistically significant by MOET, are displayed.

Figure 8. Quantitative ontology analysis of gene expression in the PFC, DH and CeA of HF rats.

(A,C,E), Summary bar graphs showing the analysis of biological, cellular or molecular gene ontologies (GOs) terms that are significantly enriched based on the number of genes that were up- or downregulated in HF rats vs. Sham in PFC (A), DH (C) and CeA (E). (B,D,F) Summary bar graphs showing the analysis of disease ontology terms highlighting important disease categories that involve the up- or downregulated genes by HF in PFC (B), DH (D) and CeA (F). Categories written in red represent those enriched terms that show up in PFC, DH and CeA; green represents those that show up in PFC and DH; blue represents those that show up in PFC and CeA; and orange represents those that show up in DH and CeA. Hypergeometric distributions with Holm-Bonferroni corrections. All the categories included in the graphs reached statistical significance p<0.00001.

For the first major group of GOs (Fig. 8A, C and E), the MOET analysis revealed that multiple genes that were robustly downregulated during HF are involved in cell communication, cell-cell signaling, cellular response to stimulus and synaptic structure/function/modulation, particularly for glutamatergic synapses (Fig. 8A and C, written in green). In addition, there was a notable involvement of genes involved in cellular metabolic processes and intracellular organelles. Interestingly, an involvement for genes related to learning and memory and cognition was observed to occur only in the DH (Fig. 8C). For the GOs related to disease processes (Fig. 8B, D and F), there was an enriched involvement of multiple genes across the three brain regions, which are associated with CNs diseases, diseases of mental health, as well as cognitive disorders. Importantly, the three regions also displayed involvement of genes involved in cardiovascular/vascular diseases. Finally, and perhaps somewhat surprising, an involvement of genes related to cancer/neoplasms was also observed.

In order to independently validate our PCR array data, we performed qRT-PCR using individual primers for five genes, Adcy1, Arc, Homer1, Mapk1 and Nr4a1, with B2m as the housekeeping gene (see Fig. 9 and Table 2). In all cases, the qRT-PCR replicated the direction of the PCR array results (i.e., increase, decrease or no effects), although the magnitude did not match precisely. For example, the results with the individual primers showed that HF significantly downregulated Adcy1 in the PFC (−1.295±0.008), PVN (−1.541±0.022), CeA (−1.499±0.038), DH (−1.481±0.019), and BLA (−1.125±0.020), but upregulated it in the VH (1.523±0.028), (Fig. 9A), which is the same pattern that we observed in the PCR array (Fig. 9B). A similar profile of results was observed for the individual primers and array for Arc, Homer1, Mapk1 and Nr4a1 (Fig. 9A and B).

Figure 9. qRT-PCR validation of results obtained in the gene array study.

(A), Summary bar graphs showing the mean fold change in mRNA levels for Adcy1, Arc, Homer1, Mapk1 and Nr4a1 genes in HF rats in relation to Sham rats (theoretical mean = +/−1) in the PFC (black bars), PVN (red bars), CeA (blue bars), BLA (green bars), DH (orange bars) and VH (purple bars). Dashed lines indicate the theoretical means. (B), heatmap graph showing the magnitude of change in gene expression in the array and validation shown as fold change in gene expression from −2.39 to +2.39 induced by HF in relation to sham rats’ gene expression. *p<0.05, **p<0.01 and ***p<0.001 vs. theoretical mean, one sample paired t-test.

Table 2.

Summary statistical data for the qRT-PCR validation study.

Brain Region	Gene	Mean	SEM	P value (two tailed)
	Adcy1	−1.295	0.008	0.0007
	Arc	−1.448	0.022	0.0024
PFC	Homer1	−1.074	0.009	0.0141
	Mapk1	−1.067	0.004	0.0041
	Nr4a1	−1.775	0.065	0.0070
	Adcy1	−1.541	0.023	0.0018
	Arc	−1.237	0.010	0.0018
PVN	Homer1	−1.143	0.009	0.0043
	Mapk1	−1.057	0.007	0.0129
	Nr4a1	−2.125	0.020	0.0003
	Adcy1	−1.499	0.038	0.0057
	Arc	−1.956	0.040	0.0017
CeA	Homer1	−1.437	0.017	0.0016
	Mapk1	−1.107	0.005	0.0025
	Nr4a1	−2.296	0.092	0.0051
	Adcy1	−1.125	0.020	0.0253
	Arc	−1.357	0.029	0.0065
BLA	Homer1	1.030	0.012	0.1217
	Mapk1	1.147	0.012	0.0066
	Nr4a1	−1.281	0.064	0.0484
	Adcy1	−1.481	0.019	0.0016
	Arc	−1.370	0.031	0.0069
DH	Homer1	−1.192	0.019	0.0100
	Mapk1	−1.127	0.003	0.0004
	NAa1	−1.332	0.014	0.0018
	Adcy1	1.523	0.028	0.0029
	Arc	−1.220	0.028	0.0162
VH	Homer1	−1.038	0.024	0.2484
	Mapk1	−1.081	0.072	0.3772
	Nr4a1	−1.409	0.026	0.0040

Summary mean fold change (in relation to Sham rats; theoretical mean = +/−1) in mRNA levels for Adcy1, Arc, Homer1, Mapk1 and Nr4a1 genes in HF rats in the PFC, PVN, CeA, BLA, DH and VH.The mean, SEM and p values corresponding to each of the five genes validated in each of the six brain regions are provided. Brains from five rats were pooled per treatment condition and three independent replicates were performed.

DISCUSSION

Collectively, the present findings show that a rat model of HF characterized by severely reduced ejection fraction recapitulates behavioral symptoms seen in patients with chronic HF, including disruption of several behavioral measures of mood and memory. Moreover, we found that this rat model of HF altered the pattern of synaptic plasticity-associated gene expression in brain areas critical for mood and memory. Specifically, we found that HF increased measures of anxiety in the EPM and impaired SA and IA retention performance, the latter two findings indicating that HF disrupts spatial working memory and emotional, long-term memory, respectively. In order to gain insights into possible molecular underpinnings of cognitive and mood disorders in HF, we conducted a PCR array study that assessed the simultaneous expression of 84 genes critical for synaptic plasticity. Importantly, our findings showed that HF downregulated numerous genes implicated in synaptic plasticity in a brain region-dependent manner, with the most profound changes observed in PVN and PFC, moderate changes in DH and CeA, and minimal alterations in VH and BLA.

To the best of our knowledge, the present results are the first to show that HF impairs SA and IA. The finding that HF affected the sequence of arms that the rats entered in the maze in the SA task without affecting the number of arm entries suggests that the deficit reflects a spatial working memory deficit rather than an effect on another variable that might influence performance in the task, such as activity levels or motivation. Both the SA and IA memory tasks are dependent on the integrity of DH [13, 57, 60, 62] and PFC [58, 59, 63, 67], which are interconnected brain regions critical for memory. The gene array results point to altered synaptic plasticity in several brain areas critical for memory, namely DH, PFC and CeA. Collectively, the findings show that our animal model of HF impacts both working memory (i.e., SA) and long-term memory (i.e., IA), which is consistent with findings showing that rats with severely reduced ejection fraction have impaired long-term memory in the spatial water maze [78] and with results showing that HF patients have deficits in both immediate and delayed recall [21] and working memory [79]. Short-term memory deficits appear to be one of the most pervasive cognitive issues in HF, which is significant because these problems are associated with impaired ability for HF patients to manage their medical condition [21, 80].

The present findings are consistent with previous research showing that HF increases anxiety in the EPM [12, 81–84] but did not replicate the finding that HF decreases sucrose preference (p = .05). The present results extend the previous findings of increased anxiety by showing that these effects are observed in a rat model of HF characterized by severely reduced ejection fraction. It is interesting to note that HF rats entered less arms than Sham rats in the EPM but not in the y-maze used to assess SA. This decrease observed in the EPM may reflect increased anxiety in the presence of the open arms of the EPM whereas all of the arms in the y-maze are enclosed. The increased anxiety observed in HF rats was paralleled by changes in gene expression in several interconnected areas involved in anxiety, specifically the PFC, PVN and CeA, but not in other components of the anxiety circuit, namely VH [85–88]. Future studies may want to examine the impact of HF on other brain areas involved in anxiety, such as the bed nucleus of the stria terminalis and periaqueductal gray area.

Our findings show that HF downregulated multiple genes implicated in synaptic plasticity in a brain region-dependent manner. For instance, changes in genes involved in learning and memory were observed only in DH and not in CeA or PFC. This may reflect the different role of these brain areas in cognition. DH involvement in SA and IA is assumed to reflect an episodic (i.e., autobiographical) memory of where the animal has been (i.e., SA) and the context in which it received footshock (i.e., IA). In contrast, the amygdala’s role in IA may be more related to the emotional component of the task [89–91] and PFC may be necessary for SA and IA due to its critical role in executive function and decision-making [92]. Although spatial abilities and episodic memory appear to be the most compromised cognitive functions in HF [93], HF patients do also show impaired executive function [94–96].

The present results significantly advance our understanding of the consequences of HF by showing that HF characterized by a severe reduction in ejection fraction is associated with the downregulation of several genes in CeA and a relative sparing of BLA. To date, there is very limited evidence implicating the amygdala in any effects of HF in human patients or animal models. In patients, HF is associated with reduced blood flow to amygdala [18] and increased amygdala responses to sympathetic activation [97, 98]. In animal models, HF increases glial density (GFAP immunoreactivity) in the medial amygdala, which is critical for cardiovascular reflexes and central control of sympathetic responses [99].

Our findings show that HF impacts DH expression but has minimal effect on VH, which is consistent with extensive evidence indicating that the hippocampus is not a unitary structure and that there are clear distinctions between DH and VH. For example, DH and VH have different anatomical connections, cellular and circuit properties and patterns of gene expression that likely contribute to the different functions that they serve [100–105]. DH appears to be primarily responsible for the cognitive functions of the hippocampus [100, 103, 104, 106], which is consistent with our finding that that genes associated with learning and memory were only impacted in DH. Our findings also add to evidence suggesting that DH is more vulnerable to neuroinflammation and other pathology than VH [107–111].

Several critical families of genes were considerably affected in these brain regions in HF rats. These include various IEGs, such as Arc, Egr, Jun (but not Fos), which would suggest an overall reduction in neuronal activation in these areas, which could be a result of reduced inputs from upstream brain regions. Another major group of genes downregulated during HF were genes encoding glutamate receptors. Given the importance of glutamate receptor signaling in maintaining appropriate basal network activity levels, as well as a playing a critical role in both up- and downregulation of the strength of synaptic efficacy [112–114], the reported changes in GLURs may also contribute to lower levels of neuronal activation and compromised synaptic plasticity in HF rats. Other major gene families downregulated during HF included neurotrophic factors (e.g., BDNF, Ntrk2 and Ngf) and kinases and phosphatases (e.g., Ppp1cc, Ppp2ca, Prkca, Prkcg and Prkg1). We independently validated our PCR array data by performing qRT-PCR using individual primers for five genes, and future studies should determine whether these changes in gene expression are accompanied by concomitant changes in protein products.

The MOET analysis revealed that the majority of genes downregulated during HF were functionally related and are involved in cell communication, cellular response to stimulus, dendritic signaling, and synaptic structure/function/modulation, particularly for glutamatergic synapses. More importantly, MOET also revealed that multiple genes that changed across brain regions were associated with diseases of mental health, cognitive disorders as well as cardiovascular/vascular diseases. Several of the same genes that were impacted by HF have been linked to cognitive dysfunction and neurodegenerative diseases such as Alzheimer’s Disease [113–121].

Numerous putative mechanisms, not mutually exclusive, could contribute to the disruption of the molecular underpinnings of synaptic plasticity during HF. One obvious candidate is neuroinflammation, a hallmark pathophysiological phenomenon in HF [122]. Activation of neuroinflammatory pathways within the hypothalamic PVN has been directly implicated as an underlying mechanism contributing to sympathohumoral activation in this disease [123–125]. Importantly, findings in rodents have implicated neuroinflammation and the expression of inflammatory and signal transduction genes in HF-induced cognitive impairment, depression and anxiety-like behavior [11, 12, 27, 126]. This possibility is supported by evidence showing that HF impacts the expression of genes related to inflammation in the hippocampus and PFC of mice [126].

Another critical pathophysiological mechanism in HF is cerebral hypoperfusion associated with low cardiac output that chronically leads to brain ischemia and microinfarcts [79]. In fact, several clinical studies implicate compromised cerebral blood flow as a mechanism contributing to cognitive impairment and dementia in patients with HF [127–129]. Cerebral hypoperfusion has also been reported in rodent models of HF [130, 131]. It is noteworthy that hypoxia reduces the expression of genes that are required for the maintenance of synaptic structure and function [132], and several of the key genes found to be disrupted in HF in this study, including Arc, Egr and mGLURs were previously reported to be affected by hypoxia [133–140]. Of particular interest to this study, Arc expression in the hippocampus was shown to be significantly downregulated following ischemic conditions, both in vitro and in vivo [136]. Moreover, manipulations that increase Arc expression protected against hypoxia-related cognitive impairment and hippocampal damage [141]. Whether chronic brain hypoperfusion and/or neuroinflammation contribute to the disruption of molecular underpinnings of synaptic plasticity and cognitive and mood disorders in HF remains to be determined.

Collectively, the present findings provide support for the growing consensus that HF is not only a neurohumoral cardiovascular problem but is also a disorder of mood and memory. Our study shows that a rat model of HF characterized by severely reduced ejection fraction recapitulates behavioral symptoms seen in patients with HF, which is significant because impaired mood and cognition in HF patients is associated with higher hospital readmission and mortality rates [9, 13–15]. There are few studies of the impact of severely reduced ejection fraction on cognition and the present results appear to be the first to show that HF disrupts SA and IA, providing an animal model of the working memory and long-term memory deficits observed in HF patients. Our study provides the first comprehensive analysis of HF on synaptic plasticity-associated processes and suggests that pathological alterations in plasticity mechanisms in DH and PFC may contribute to mood and memory deficits, which is in agreement with the finding that learning and working memory deficits, depression, and anxiety are correlated with significant cerebral grey matter loss in hippocampus and PFC [16, 17]. Our findings advance studies conducted in humans by suggesting that the brain damage observed in patients with HF is likely accompanied by a decrease in the expression of many genes necessary for the molecular mechanisms involved in synaptic plasticity. Moreover, our findings suggest that structural damage and gene expression may not necessarily go hand in hand because HF impacts VH structure in rats [142], whereas our results show that HF has a negligible effect on synaptic plasticity-associated gene expression in this brain region. Our findings appear to be the first to suggest that the amygdala may also participate in the mood and memory deficits as HF downregulates several genes in CeA with a relative sparing of BLA. Finally, our findings also show that the majority of genes that were downregulated are involved in cell communication, cellular response to stimulus, dendritic signaling, and synaptic structure/function/modulation, particularly for glutamatergic synapse and are associated with diseases of mental health, cognitive disorders and cardiovascular/vascular diseases.

HIGHLIGHTS.

• We induced heart failure (HF) in rats and tested mood, anxiety and memory function.

• HF rats displayed signs of increased anxiety, and impaired memory.

• Downregulation of synaptic plasticity genes was observed in related brain areas.

• We identified candidate brain areas and molecular mechanisms related to mood and memory deficits in HF.

ABBREVIATIONS

Adcyl

Adenylate Cyclase 1

Arc

Activity-regulated cytoskeleton-associated protein

B2m

Beta-2-Microglobulin

BLA

Basolateral Amygdala

cDNA

complementary DNA

CeA

Central Amygdala

CNS

Central nervous system

DH

Dorsal Hippocampus

EPM

Elevated Plus Maze

GO

Gene Ontology

HF

Heart Failure

Homerl

Homer Scaffold Protein 1

IA

Inhibitory Avoidance

LTD

Long-term depression

LTP

Long-term potentiation

Mapkl

Mitogen-activated protein kinase 1

MOET

Multi Ontology Enrichment Tool

Nr4a1

Nuclear Receptor Subfamily 4 Group A Member 1

PFC

Prefrontal Cortex

PVN

Paraventricular Nucleus

qRT-PCR

quantitative Reverse Transcriptase - Polimerase Chain Reaction

SA

Spontaneous Alternation

SPT

Sucrose Preference Test

VH

Ventral Hippocampus