Age-related cognitive impairment is associated with long-term neuroinflammation and oxidative stress in a mouse model of episodic systemic inflammation

Adult male Swiss mice (2 months old, 25 to 30 g) and aged male Swiss mice (12 months old, 40 to 45 g) were obtained from the Laboratory Animal Breeding Center (ICTB) of Oswaldo Cruz Foundation (FIOCRUZ). The animals were maintained on a 12:12-h light–dark cycle with free access to food and water. Aged mice were maintained one per cage over the course of experiments. All procedures were performed in accordance with international guidelines for use of Laboratory Animals and approved by the Institutional Animal Welfare Committee of FIOCRUZ (CEUA), Rio de Janeiro, Brazil.

Young and aged mice were randomly divided into two groups (n = 10 per group for each experiment): naive controls or episodic systemic inflammation (SI). Episodic systemic inflammation (SI) was produced by weekly injections of a lipopolysaccharide solution (LPS from Escherichia coli 0111:B4, Sigma-Aldrich, Inc.) prepared in physiological saline (sterile 0.9% NaCl). Intraperitoneal injections of LPS (0.33 mg/kg) were made in 100 μL volume using 1 mL syringes with 30G × 12.7 mm needles; the solution was sonicated for 5 min before injections. All injections were given at 12:00 noon (± 1 h), once a week for 6 weeks to produce an episodic systemic inflammation without inducing tolerance to LPS due to the 1 week interval between injections [23]. Control mice were manipulated the same way as LPS-injected animals, but were not injected with vehicle to avoid any lesion that could induce a local inflammation (naive controls). Sickness behavior and systemic inflammation were assessed in different phases of episodic systemic inflammation in young and aged mice by a researcher blinded to experimental groups (control or SI) (Fig. 1). To monitor systemic inflammation during the evolution of SI, we evaluated white blood cells (WBC) in the peripheral blood at several time points after LPS injections. One drop of blood was collected from the mouse-tail, made free of red blood cells with Türk solution and the total number of WBC was manually counted in a Neubauer chamber. Peripheral blood differential count was made in blood smears, air dried, and stained with panoptic technique. The percentages of individual morphological forms of WBC were established and absolute number of blood polymorphonuclear granulocytes (neutrophils) was calculated by absolute cell count/mL = total WBC/mL × % of the cell on the differential count. Long-term outcomes (behavioral, biochemical, and immunohystochemical studies) were evaluated 1 to 2 weeks after the last LPS injection (Fig. 1c). Aged mice used in this study were middle-aged that presented signs of senescence, such as lipofuscin deposits [24, 25] (Additional file 1: Figure S1). This experimental model overall did not induce severe systemic inflammation or mortality. However, a few aged mice died after SI (less than 10% mortality). Some aged mice presented tumors that were detected after euthanasia, and these animals were excluded from the study. Aged mice that developed severe symptoms were also excluded from the study and euthanized (less than 5%).

All the procedures used in this study were designed to minimize the potential discomfort during behavioral tests. Clinical manifestations of sickness behavior were evaluated 24 h after each weekly LPS injection, when mice were scored as being mildly, moderately, or severely affected as described previously with some modifications [7, 26, 27]. Briefly, the scoring assessed mice appearance (piloerection, bloated abdomen, sunken eyes), alertness (interest in the environment, moving away from experimenter hand, locomotion), grip strength (time holding to a bar 30 cm high with fore paws), and body temperature (measured with an infrared thermometer) (Fig. 1d). After SI, 1 week after the last LPS injection, all groups of mice were submitted to behavior tests. Depressive-like behavior was evaluated with the forced swimming test. Mice were placed in a transparent cylindrical tank filled with temperature-controlled tap water (25 °C). Mice were hold by its tail, gently and slowly placed in the water, and their immobility time was recorded for 3 min. The lack of escape-related mobility behavior is measured by the immobility time and indicates depressive-like behavior.

Episodic-like memory of mice was assessed with the object recognition test. This type of memory is required to the natural exploratory abilities of rodents exposed to a new environment and involves medial temporal lobe neural circuits (hippocampus and entorhinal cortex). The object recognition test was performed according to a methodology previously described [28] in a square wooden open-field apparatus (40 × 40 × 40 cm) and consisted of three phases: habituation, familiarization, and test phase. In the habituation phase each animal was allowed to freely explore the open-field arena in the absence of objects for 10 min. The familiarization phase was performed 24 h after habituation and consisted of a 5 min exploration of identical objects in the arena. The test was made 3 h after familiarization phase and consisted of 5 min exploration of two different objects, one familiar and a substituted novel object. The novel object had different color, shape, and size that allowed recognizing it as novelty. The time each animal explored the familiar and the novel object was filmed, and an observer blinded to experimental conditions analyzed the videos. Exploration was defined as the animal directing its nose within 2 cm of the object while looking at, sniffing, or touching it. The arena and objects were cleaned with a 10% ethanol solution between trials to ensure that behavior of animals was not guided by odor cues. An exclusion criterion based on a minimal level of object exploration was used to exclude animals with naturally low levels of spontaneous exploration (novel + familiar exploration time < 10 s). Animals that explored the objects for less than 10 s were excluded from the study. The object recognition index was calculated as time for the novel object − time for the familiar object/total exploration time.

The Morris water maze test was performed as previously described [18] 1 week after the last injection of LPS, to evaluate mice spatial memory. This type of memory, which also involves the medial temporal lobe, permits the formation of a cognitive map of the external space to which the animal is located and through which allow it better interaction with the environment [29]. The water maze consisted of a circular tank with a diameter of 1.04 m and height of 55 cm, located inside a room with some visual cues fixed in the walls. The tank was filled with water and divided into four quadrants, and in each quadrant, a transparent acrylic platform (13 × 30 × 13 cm) submerged 1 cm from the water surface was placed. The experiment was recorded in a Samsung camera. The animals were subjected to daily training sessions for 4 consecutive days to find the platform. If the animal did not find the platform after 60 s, it would be conducted manually to it after 60 s. During all training sessions, the time each mouse required to find the platform (latency) was registered. In the fifth and final day of experiment, the test session was held, in which the platform was removed and mice were challenge to find the platform quadrant. The time mice spent in the platform quadrant in the test session, velocities of swimming, and latency time in the training sections were analyzed with the ANY-maze behavioral tracking software.

Microglia morphological studies were carried out at 4 h after the first LPS injection (acute SI) and 1 week after the last LPS injection (long-term SI) in the hippocampus of young and aged mice. Initially, microglia activation was evaluated by counting the total number of Iba1-positive cells and the number of Iba1-positive cells with activated morphology per field and expressed as percentage of activated microglia. Microglia were classified as activated when cells presented shortening of processes and increase in cell body size (hypertrophy) [30, 31]. A researcher blinded to experimental conditions performed this counting using the ImageJ Cell Counter plugin; at least four fields were counted per animal. A second study was developed to assess the acute morphological changes of microglia in the entorhinal cortex in response to SI (4 h after the last LPS injection). Microglia were considered dystrophic when abnormalities in cytoplasmic structure were observed, such as formation of cytoplasmic spheroids, gnarling, beading, and fragmentation, and a distinct loss of fine branches (deramification). As it is rare to see all of these dystrophic changes occurring in a single cell, most dystrophic microglia display just one or two of these characteristics [31]. The criteria used to classify dystrophic microglia were the presence of at least one of these features. For immunohistochemistry, the brains were dissected and immediately immersed in 4% paraformaldehyde for 3 days at 4 °C. After fixation brains were crioprotected in 20% sucrose for 2 days, frozen in dry ice, and stored at − 80 °C. Coronal sections (40 μm) were obtained in a cryostat (Leica Microsystems, Germany), washed with phosphate buffer saline at pH 7.4, and blocked for 2 h with 1% BSA, 2% normal goat serum and 0.3% Triton X-100 blocking solution at room temperature. Sections were incubated overnight at 4 °C free floating in primary antibody solution. For microglia studies, a rabbit IgG anti-Iba1 (1:400 dilution; catalog number 019-19741 from Wako Chemicals USA, Inc.) primary antibody in blocking solution was used. For oxidative stress studies, a rabbit IgG anti-HNE (1:400 dilution; catalog number HNE11-S from Alpha Diagnostic International Inc., Texas, USA) primary antibody in blocking solution was used. After primary antibody incubation, the sections were washed with PBS and incubated with the secondary antibody Alexa Fluor 488 goat anti-rabbit IgG (1:1000 dilution; catalog number A-11008 from Molecular Probes, Eugene, OR, USA) for 2 h at room temperature and washed with PBS. Sections were mounted in Vectashield Antifade Mounting Medium with DAPI (for nuclei counterstaining). No reactivity was observed in brain sections when the primary antibody was absent.

Images of microglia from the hippocampus and entorhinal cortex were obtained using a Leica confocal microscope (Leica Microsystems, Germany) or a Zeiss confocal microscope (Carl Zeiss Microscopy, Germany) with Z stack and 3D-reconstruction. Confocal image stacks were used for microglial morphological studies. For quantification of morphological changes, microglial cells were automatedly segmented and separated from their neighboring ones after identification of their cell bodies. First, contrast was optimized using the local contrast enhancer plugin from Fiji, and an automatic threshold was determined with the Li algorithm. Microglial masks were automatically simplified using “open” and “close” functions. Individual cell bodies were obtained through the Iba1 channel, after local contrast optimization, automatic local threshold was determined with the Phansalkar algorithm set with a radius of 20 pixels. Then, a tridimensional watershed was performed (mcib3d suite). Only individual microglial cells were furthered considered for analysis. Cells were skeletonized using 3D skeletonize plugin and labeled for slab, junctions, or end of branches. Conformational Sholl analysis was performed through concentric envelopes computed with a 1 μm radius step starting from the cell body. Densities of slabs, junctions, or end of branches were then analyzed iteration after iteration to compute the number of branches as a function of the distance to cell body. Skeleton was also analyzed using the skeleton analysis algorithm, and the longest path was considered for each cell. The area under the curve was computed with GraphPad Prism software for each microglial cell and microglial cells (3–7 per samples) were pooled for each condition and compared using a Kruskall Wallis test.

Oxidative stress in the brain tissue was quantified by immunodetection of 4-hydroxynonenal (4-HNE), a product and mediator of oxidative stress that binds covalently to proteins [32]. Mice were deeply anesthetized with isoflurane and transcardially perfused with 0.9% NaCl saline, and their brains were dissected and immediately frozen in dry ice. The tissue was homogenized 50 mM Tris, pH 8.0 containing 150 mM sodium chloride, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitor cocktail (Complete Mini Protease Inhibitor Cocktail; Roche Diagnostics, IN, USA), and phosphatase inhibitor sodium orthovanadate. Homogenates were incubated for 30 min on ice, sonicated for 3 min, and centrifuged at 10.000 × g at 4 °C. The supernatant was collected, and the protein concentration was measured using a BCA Protein Assay (Thermo Fisher Scientific, Waltham, MA, USA). For Western blotting, 50 μg of protein samples were separated on polyacrylamide 4–20% gradient gels (Mini-Protean TGX precast Gels®-BioRad, Hercules, CA, USA), and proteins were transferred to PVDF Immobillon-FL membranes (Millipore, MA, USA). Membranes were blocked with Odyssey Blocking Buffer reagent (Li-Cor Biosciences, Lincoln, NE, USA) and incubated overnight with a primary antibody mix of a mouse IgG anti-4-hidroxi-2-nonenal (1:500, Abcam, Cambridge, MA, USA) and a goat IgG anti-actin (1:5000, Santa Cruz, CA, USA). After washing with PBS, membranes were incubated with IRDye secondary antibodies (Li-Cor Biosciences) for 30 min. Immunoreactivity was visualized with Odyssey Infrared Imaging System (Li-Cor Biosciences, Lincoln, NE, USA). Band intensities were calculated using the Image Studio Software. 4-HNE-positive bands consisted of a smear with stronger bands between 50 and 150 kDa that were selected for quantification. 4-HNE signals from each lane were normalized to corresponding actin band signals.

The brains were collected and directly homogenized in 1 ml of TRIzol® Reagent (Thermo Fisher Scientific Inc.), and total RNA was isolated following the manufacturer’s suggested protocol. RNA concentration was quantified using the NanoDrop 2000 Spectrophotometer (Thermo Scientific), and cDNA was synthesized according to manufacturer’s instructions (SuperScript First-Strand Synthesis System for RT-PCR, Invitrogen) and was stored at − 20 °C. Transcripts obtained from the reverse transcriptase reaction were quantified by quantitative fluorogenic real-time PCR using SYBR® green. The real-time PCR amplification was performed, recorded, and analyzed by a 7500 Real-Time PCR System (Applied Biosystems). The specificity of the SYBR® green assay was confirmed by melting-point analysis. The primers used were Sod1 (left 5′ to 3′-GGA CCT CAT TTT AAT CCT CAC, right 5′ to 3′-TGC CCA GGT CTC CAA CAT G), Cat (left 5′ to 3′-AGA GAG CGG ATT CCT GAG AGA, right 5′ to 3′-ACC TTT CCC TTG GAG TAT CTG), Sod2 (left 5′ to 3′-CAG ACC TGC CTT ACG ACT ATG, right 5′ to 3′-CTC GGT GGC GTT GAG ATT G), Nox2 (left 5′ to 3′-GAC CAT TGC AAG TAG ACA CC, right 5′ to 3′-AAA TGA AGT GGA CTC CAC GC), and Hprt (left 5′ to 3′-AGG ACC TCT CGA AGT GTT GG, right 5′ to 3′-TTG CAG ATT CAA CTT GCG CT) at 300 nM concentration. Target gene expression was normalized by the Hprt gene. The levels of gene expression were calculated by the ΔCt method for quantification, and the results were expressed as 2^ΔCt [33].

Statistical analyses and graphs were generated using GraphPad Prism version 5.0 (GraphPad Software). All results are presented as mean and standard errors of means (SEM). The data were analyzed using two-way ANOVA to determine how the two factors, age and SI, affected responses, with Bonferroni posttest for multiple comparisons, and Kruskal-Wallis with Dunn’s multiple comparisons test for non-parametric analysis, unless otherwise noted. Two-tailed univariate t test was also used, when appropriate. Statistical significance is shown on the graphs (*p < 0.05, **p < 0.01, ***p < 0.001). Statistical tests used for each data set are indicated in the figure legends.

The long-term consequences of episodic systemic inflammation in the aging brain are poorly understood. Here, we developed a relevant pre-clinical model to investigate the molecular mechanisms of cognitive impairments associated with aging and episodic systemic inflammation (SI). The SI model consisted of weekly injections of LPS that induced mild symptoms of sickness behavior and higher systemic inflammatory response mainly in aged mice (Fig. 1). The behavior studies showed that SI effect differs depending on age and the cognitive domain tested. Depressive-like behavior was observed after SI in both groups of mice, young and aged (Fig. 2a). Cognitive domains such as memory and learning were affected only in aged mice. Young mice did not present deficits in the memory tests even after SI. However, aged mice were more susceptible to short-term memory impairments after SI, as shown by their poor performance in the object recognition test (Fig. 2b). All groups of mice were capable to swim and learn where the hidden platform was to perform the water maze test (Fig. 2c). Velocity of swimming was also similar between groups (Fig. 2d). Young controls and young SI had similar performances in the water maze test. However, aged SI mice spent less time in the target quadrant indicating they had compromised spatial memory compared to aged controls (Fig. 2e).

Immunohistochemistry analysis of Iba-1-positive cells allowed us to observe microglial morphology and the amount of activated microglia in the hippocampus of young and aged mice at different stages of SI to evaluate changes that were related to age or induced by SI. The percentage of activated microglia after SI was higher in young mice acutely and returned to basal levels long term after SI, while aged mice presented high amounts of activated microglia at the basal level that remained elevated long term after SI (Fig. 3a). We then studied the morphological features of microglia in higher magnification and in a quantitative way in the entorhinal cortex, a brain region that works as a major hub in a widespread network for memory and navigation of mice [34]. Young mice showed morphological features of surveillant microglia, long processes full of branches, whereas aged microglia were morphologically compatible with dystrophic microglia (Fig. 3b). Quantitative analysis of microglia morphological changes showed interesting differences between groups. Total number of microglia skeleton segments was lower in the aged mice (young 415 ± 57 vs aged 118 ± 13, p < 0.0001) whereas young SI showed a trend towards a diminution of microglial cell ramification number (Fig. 3c). Acute microglia response to episodic SI in maximal number of branches was higher in young mice (13.1 naive and 8.4 SI) than in the aged (5.6 naive and 4.9 SI), and SI had few effects on already poorly ramified microglial cell in aged animals (Fig. 3d). Sholl analysis showed a decrease in the number of branches along the microglial cell in response to SI in the young mice (young naïve 207 ± 34 vs young SI 115 ± 13, ***p < 0.001) but not between aged naive and aged SI (57 ± 7 vs 60 ± 8, p = 0.89) (Fig. 3e). Young microglia responded to SI by retracting their processes and decreasing the number of branches from their cell bodies. However, aged microglia did not present overt morphological changes upon SI challenge, being apparently resistant to modulation by SI (Fig. 3c–e).

Microglia are major regulators of neuroinflammation and produce most of the inflammatory mediators in the brain, such as cytokines and reactive oxygen species. The levels of cytokines IL-1β, IL-6, IL-4, and IL-10 were measured in the brains of young and aged mice after SI. Aged mice presented higher levels of both pro-inflammatory IL-1β and IL-6 and pro-resolution IL-4 and IL-10 cytokines in the brain compared to young long term after SI (Fig. 4).

The hippocampus is a critical region for learning and memory processes in the brain, and it is especially susceptible to oxidative stress. We observed increased immunostaining for 4-HNE in the hippocampal CA1 region of aged mice after SI compared to the other experimental groups (Fig. 5a). Accordingly, higher concentrations of 4-HNE adducts in brain protein extracts from aged mice after SI, compared with the naïve controls (Fig. 5b, c). The oxidative stress observed in the aged brains long term after SI was temporally associated with an increased expression of Nox2 in the brains of these mice (Fig. 5c, d). We then measured the gene expression of major antioxidant enzymes superoxide dismutase 1 and 2 (Sod 1 and Sod2) and catalase (Cat) in the brains of young and aged mice, naïve, and long term after SI. Aged mice presented lower expression of Sod1 after SI (Fig. 6a) and no differences in the expression of Sod2 (Fig. 6b). Catalase expression was not affected by SI (Fig. 6c).