Neuroprotection Induced by Energy and Protein-Energy Undernutrition Is Phase-Dependent After Focal Cerebral Ischemia in Mice

Experiments were approved by local government authorities (Bezirksregierung Düsseldorf) in accordance with E.U. guidelines (Directive 2010/63/EU) for the care and use of laboratory animals. Sample size calculations determined that 12 animals per group were required for neurological examinations and histochemical studies, given that the effect size was 1.167, the alpha error was 5%, and the beta error (1 statistical power) was 20%. Experimenters were blinded by a third person not involved in the assessments randomizing the animals, weighing, and providing the food pellets.

Adult male C57BL6/j mice (8 weeks, 26–30 g; Harlan-Netherlands, Rossdorf, Germany) were randomized to three diets: (a) normal nutrition (C1000; 3518 kcal/kg, 20% protein (i.e., casein); Altromin, Lage, Germany), (b) energy-reduced nutrition (C1012 mod.; 1313 kcal/kg, 20% protein (casein); Altromin), and (c) protein-energy-reduced nutrition (C1003 mod.; 1300 kcal/kg, 8% protein (casein); Altromin). Diets were offered ad libitum over 7, 14, or 30 days. Animals were then submitted to 30 min intraluminal MCAO. Throughout the study, animals were housed in single cages in a 12 h:12 h light/dark cycle. Food consumption and calorie intake were measured daily. Body (i.e., nose-anus) length was determined prior to diet modification. Body weight and BMI were determined weekly. Stool changes and behavioral abnormalities, namely, spontaneous motor hypoactivity, were checked daily.

Mice were anesthetized with 1.0–1.5% isoflurane (30% O₂, remainder N₂O). Rectal temperature was maintained between 36.5 and 37.0 °C using a feedback-controlled heating system. Cerebral laser Doppler flow (LDF) was recorded using a flexible probe (Perimed, Järfälla, Sweden) attached to the skull overlying the core of the middle cerebral artery territory. A midline neck incision was made. The left common and external carotid arteries were isolated and ligated, and the internal carotid artery was temporarily clipped. A silicon-coated nylon monofilament (0.21-mm tip diameter; Doccol, Sharon, MA, USA) was introduced through a small incision of the common carotid artery and advanced to the circle of Willis for MCAO [5, 6]. Reperfusion was initiated by monofilament removal after 30 min. Thirty minutes of MCAO was chosen, since this model induced reproducible injury of the striatum and the most lateral cortex with little animal dropouts. After surgery, wounds were carefully sutured and anesthesia was discontinued. Twenty-four hours later, animals were evaluated using the Clark score [18], which captures general and focal neurological deficits. Immediately before sacrifice, plasma samples were obtained by cardiac puncture after 5 h fasting that were used for analysis of total cholesterol, low-density lipoprotein cholesterol (LDL), triglycerides, and glucose levels (ADVIA® 2400; Siemens, Erlangen, Germany). One set of animals (n = 12/group) was transcardially perfused with normal saline followed by 4% paraformaldehyde. The animals’ brains were cut into 20-μm-thick coronal sections for histochemical studies. Another set of animals (n = 6/group) was transcardially perfused with normal saline. From the animals’ brains, tissue samples were collected from the middle cerebral artery territory for Western blots and real-time quantitative polymerase chain reaction (qPCR) studies. For this purpose, a 2-mm-thick coronal brain slice ranging from 1 mm rostral to 1 mm caudal to the bregma was prepared, from which a triangular slice containing the striatum and the most lateral parietal cortex was dissected. This selection strategy was chosen to exclude partial volume effects of infarct reductions or expansions on gene expression results. From the same animals, liver samples were also obtained.

Coronal sections collected at millimeter intervals across the brain were stained with cresyl violet. Infarct volume was determined by subtracting the area of healthy tissue in the ischemic hemisphere from that in the contralesional hemisphere [5, 6].

Brain sections obtained from the rostrocaudal level of the midstriatum were rinsed for 20 min in 0.3% H₂O₂ in 70% methanol/0.1 M phosphate-buffered saline (PBS), immersed in 0.1 M PBS containing 5% bovine serum albumin (BSA) (05470; Sigma-Aldrich, Darmstadt, Germany), and incubated for 1 h in biotinylated anti-mouse IgG (1:100; Santa Cruz, Heidelberg, Germany), followed by diaminobenzidine (DAB) tetrahydrochloride (D5905; Sigma-Aldrich) staining with an avidin-biotin complex peroxidase kit (Vectastain Elite; Vector Labs, Burlingame, CA, USA) [5]. IgG extravasation was analyzed by measuring the area of IgG leakage.

Adjacent brain sections were subjected to terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) using a commercially available In Situ Cell Death Detection Kit (Roche, Mannheim, Germany) [5, 6]. TUNEL+, that is, DNA-fragmented cells were evaluated under a Zeiss AxioObserver.Z1 microscope equipped with Apotome optical sectioning by counting the total number of labeled cells in the striatum.

Adjacent sections were immersed in 0.1 M PBS containing 0.3% Triton X-100 (PBS-T) and 5% normal donkey serum (D9663; Sigma-Aldrich). Sections were incubated overnight at 4 °C in monoclonal rabbit anti-NeuN (1:400; ab177487; Abcam, Cambridge, UK), monoclonal rat anti-CD45 (1:200; 550539; BD Biosciences, Heidelberg, Germany), polyclonal rabbit anti-ionized calcium binding adaptor protein (Iba)-1 (1:500; Wako Chemicals, Neuss, Germany), monoclonal rat anti-glial fibrillary acidic protein (GFAP) (1:200; 130300; Invitrogen, Dublin, Ireland), or polyclonal rabbit anti-inducible nitric oxide synthase (iNOS) (1:100; sc-650; Santa Cruz, CA, USA) antibodies that were detected with Alexa Fluor-488– or Alexa Fluor-594–labeled secondary antibodies (NeuN, Iba-1, GFAP, and iNOS) or biotinylated secondary antibodies followed by DAB staining with an avidin-biotin complex peroxidase kit (Vectastain Elite, Burlingame, CA, USA) (CD45). NeuN, Iba-1, GFAP, and iNOS labelings were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (D9542; Sigma-Aldrich). Sections were evaluated under a motorized Zeiss AxioObserver.Z1 inverted epifluorescence microscope equipped with Apotome optical sectioning (NeuN, Iba-1, GFAP, and iNOS) or an Olympus X52 microscope (CD45) by counting the total number of NeuN+, CD45+, or iNOS+ cells in the striatum, in which ischemic injury is most reproducible, or analyzing the area covered by activated microglia (Iba-1) or reactive astrocytes (GFAP). The latter analysis was preferred to cell counting, since individual cells could not always unequivocally be discriminated. The latter data were shown as percent changes.

From the brain and liver tissue samples, messenger RNA (mRNA) was extracted using an RNeasy Mini Kit (Qiagen, Hilden, Germany). mRNA was converted to cDNA using a high-capacity RNA-to-cDNA kit (Thermo Fisher Scientific, Langenselbold, Germany). Real-time qPCR was performed in a StepOnePlus real-time PCR system using primers selected by the PubMed primer-BLAST tool (https://​blast.​ncbi.​nlm.​nih.​gov/​) (Suppl. Table 1). The efficiency of these primers had been confirmed in melting curves. β-Glucuronidase (β-Gluc) was used as a housekeeping gene; the brain and liver tissue from healthy mice served as control. Results were quantified using the 2^−∆∆Ct method. PCR were performed in triplicate, of which mean values were formed for each mouse.

Statistical analyses were performed using SPSS for Windows. Nutritional data and LDF recordings were analyzed by two-way repeated measurement ANOVA followed by unpaired t tests as post hoc tests. Neurological deficits and histochemical data were analyzed by one-way ANOVA followed by Tukey post hoc tests (for normally distributed results) or Kruskal-Wallis tests (for non-normally distributed results). Real-time qPCR data were compared by pairwise t tests. To explore the relationship of calorie intake with infarct volume and neurological deficits, two-tailed Pearson’s correlations were computed. Nutritional data, LDF recordings, and real-time qPCR data are presented as mean ± S.D. values. Neurological deficits, histochemical data, and Western blots are shown as median ± interquartile range box plots with minimum/maximum data as whiskers. p values < 0.05 were defined to indicate statistical significance.

Murinometric analyses revealed progressive body weight (Suppl. Fig. 1A–C) and body mass index (BMI) (Suppl. Fig. 1D–F) loss (each by ~ 20%) over up to 30 days in mice exposed to energy and protein-energy undernutrition. Although the total amount of food ingested was elevated in mice on modified diets (Suppl. Fig. 1G–I), calorie intake was consistently reduced at all time points at which ischemia was subsequently induced (Suppl. Fig. 1J–L). Upon diet modification, stool samples progressively increased in size and adopted a pale color (Suppl. Fig. 2; Suppl. Table 2), indicative of malabsorption syndrome. With progressive undernutrition, thin blood beddings were sometimes found on stool samples (Suppl. Table 2). Motor hypoactivity was frequently noted shortly before MCAO (Suppl. Table 2).

Plasma LDL, which constitutes a comparably small percentage of total cholesterol in mice, since mice lack cholesterol ester transfer protein (CETP) [19], was decreased in mice exposed to protein-energy undernutrition for 7 days, but not for 14 or 30 days (Suppl. Table 3). Total cholesterol, triglycerides, and glucose were not influenced by energy or protein-energy undernutrition (Suppl. Table 3).

In mice receiving non-modified diet, cerebral LDF decreased to ~ 15–20% of baseline values during MCAO, followed by the restoration of LDF to baseline values within 20 min after monofilament removal (Fig. 1A–C). LDF patterns did not differ in mice exposed to energy or protein-energy undernutrition for 7 or 14 days (Fig. 1A, B), whereas decreased reperfusion indicative of hemodynamic impairment was noted in mice exposed to prolonged energy and protein-energy undernutrition for 30 days (Fig. 1C). Neurological examination 24 h after MCAO revealed that energy undernutrition but not protein-energy undernutrition for 14 days decreased general and focal neurological deficits (Fig. 1E, H), whereas diet modification for 7 or 30 days did not influence neurological performance (Fig. 1D, F, G, I). Both energy and protein-energy undernutrition significantly reduced infarct volume, when imposed for 14 days, but not 7 or 30 days (Fig. 1J–L). IgG extravasation, a measure of disturbed blood-brain barrier integrity, was reduced by energy undernutrition for 14 or 30 days and protein-energy undernutrition for 30 days (Fig. 1M–O).

Our data were indicative of three stages of undernutrition: an initial adaptation stage (that is, 7 days after diet modification), in which ischemic injury is not influenced by energy and protein-energy undernutrition, which is followed by a compensated stage (that is, 14 days after diet modification), in which brain tissue is protected against ischemia, and an exhausted stage (that is, after 30 days), in which neuroprotection is lost as a consequence of post-ischemic hemodynamic impairments. Pearson’s correlations revealed that calorie intake in the adaptation stage was negatively correlated with general neurological deficits (r = − 0.275, p = 0.02), but not with infarct volume or focal deficits (Suppl. Fig. 3). In the compensated stage, calorie intake was positively correlated with infarct volume (r = 0.572, p < 0.001), but not with neurological deficits (Suppl. Fig. 3). In the exhausted stage, no correlations of calorie intake with infarct volume or neurological deficits were found (Suppl. Fig. 3).

Immunohistochemical studies showed that energy and protein-energy undernutrition for 14 days increased the density of surviving NeuN+ neurons in the ischemic striatum (Fig. 2B) and that protein-energy undernutrition reduced the density of DNA-fragmented, that is, irreversibly injured TUNEL+ cells (Fig. 2E). Undernutrition did not influence neuronal survival (Fig. 2A, C) or cell injury (Fig. 2D, F), when imposed for 7 or 30 days. Protein-energy undernutrition decreased brain leukocyte infiltration, as assessed by CD45 immunohistochemistry (Fig. 3A), and microglial activation, as evaluated by Iba-1 immunohistochemistry (Fig. 3D), when imposed for 7 days. Energy undernutrition for 7 days reduced the density of iNOS+ cells (Fig. 3J), which had the size and shape of microglia. Interestingly, diet modification for 14 days did not alter brain leukocyte infiltration, microglial activation, or iNOS formation (Fig. 3B, E, K). Protein-energy undernutrition for 30 days reduced the brain infiltration of CD45+ leukocytes (Fig. 3C) and decreased astrocytic GFAP immunoreactivity (Fig. 3I).

Real-time qPCR showed that undernutrition regulated metabolism-related, inflammatory, and anti-oxidant genes in the ischemic brain. Sirtuin-1 (Sirt-1) mRNA, which encodes for an NAD-dependent deacetylase that stabilizes mitochondrial function and metabolism partly by deacetylating the transcription regulator peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) [20], and glucose transporter-1 (Glut-1) mRNA, which encodes for a membrane transporter promoting glucose uptake in brain cells [21], were upregulated in the ischemic brain of mice exposed to 7 days energy undernutrition (Table 1), as was Sirt-1 protein, as shown in the Western blots (Fig. 4A). Il-1β mRNA was downregulated by 7 days protein-energy undernutrition (Table 1). Only subtle gene expression changes were noted in the ischemic brain of mice exposed to 14 days undernutrition, that is, an elevation of superoxide dismutase (Sod)-1 mRNA, which encodes for a dismutase degrading superoxide anions, in mice exposed to protein-energy-reduced diet (Table 1). Sirt-1 and Glut-1 mRNAs (Table 1) as well as Sirt-1 protein (Fig. 4B) did not differ from mice receiving non-modified diet, indicating that the metabolic needs of the tissue were adapted to the reduced energy supply. In the ischemic brain of mice exposed to prolonged energy or protein-energy undernutrition for 30 days, Sirt-1 mRNA (Table 1) and Sirt-1 protein (Fig. 4C) were increased, and in the ischemic brain of mice exposed to prolonged protein-energy undernutrition for 30 days, Glut-1 and Sod-1 mRNA (Table 1) were elevated. Conversely, insulin-like growth factor-1 (Igf-1) mRNA, which encodes for a growth factor with insulin-like properties, Il-1β mRNA, and nuclear factor-κb (Nf-κb) mRNA, which encodes for a transcription factor, were reduced by energy or protein-energy undernutrition (Table 1). The responses of Sirt-1, Glut-1, and Igf-1 were interpreted as effort to maintain brain tissue glucose supply and confine metabolic needs in face of energy reserves which faded.

Real-time qPCR showed that protein-energy undernutrition over 7 days downregulated catalase (Cat) mRNA, which encodes for a protein that degrades peroxides formed by dismutases, in the liver (Table 2). After 14 days protein-energy undernutrition, Sod-1 mRNA was increased, whereas glucose transporter-2 (Glut-2) mRNA, Sod-2 mRNA, which encodes for another dismutase, and Cat mRNA were reduced, as was Sod-2 mRNA after 14 days energy undernutrition (Table 2). Prolonged energy and protein-energy undernutrition over 30 days downregulated NADPH oxidase-4 (Nox-4) mRNA, which encodes for a protein that catalyzes the production of superoxide free radical by transferring one electron to oxygen from NADP, and anti-oxidant Sod-2 mRNA (Table 2). Prolonged energy undernutrition over 30 days upregulated glutathione peroxidase-3 (Gpx-3) mRNA (Table 2), which encodes for another peroxidase.