Neurobiology of Disease 62 (2014) 456–468

Contents lists available at ScienceDirect

Neurobiology of Disease

j ourna l homepage: www.e lsev ie r .com/ locate /ynbd i

Exacerbation of ischemic brain injury in hypercholesterolemicmice is associated with pronounced changes in peripheral andcerebral immune responses

Josephine Herz a,b, Sabine I. Hagen b, Eileen Bergmüller b, Pascal Sabellek b, Joachim R. Göthert c, Jan Buer d,Wiebke Hansen d, Dirk M. Hermann b, Thorsten R. Doeppner b,⁎a Department of Paediatrics I, Neonatology, University Hospital Essen, University Duisburg-Essen, Essen, Germanyb Department of Neurology, University Hospital, Essen, Germanyc Department of Hematology, West German Cancer Center, University Hospital of Essen, Essen, Germanyd Institute of Medical Microbiology, University Hospital Essen, University Duisburg-Essen, Essen, Germany

⁎ Corresponding author at: Department of NeurologyMedical School, Hufelandstr. 55, 45147 Essen, Germany. Fax

E-mail address: thorsten.doeppner@uk-essen.de (T.R.Available online on ScienceDirect (www.sciencedir

0969-9961/$ – see front matter © 2013 Elsevier Inc. All rihttp://dx.doi.org/10.1016/j.nbd.2013.10.022

a b s t r a c t

a r t i c l e i n f o

Article history:Received 3 July 2013Revised 7 October 2013Accepted 24 October 2013Available online 31 October 2013

Keywords:Cerebral ischemiaHypercholesterolemiaNeuroinflammation

Inflammation contributes to ischemic brain injury. However, translation of experimental findings from animalmodels into clinical trials is still ineffective, since the majority of human stroke studies mainly focus on acuteneuroprotection, thereby neglecting inflammatorymechanisms and inflammation-associated co-morbidity factorssuch as hypercholesterolemia.Therefore, bothwildtype andApoE−/−mice that exhibit increased serumplasma cholesterol levels fedwith normalor high cholesterol diet were exposed to transient middle cerebral artery occlusion. Analysis of peripheral immuneresponses revealed an ischemia-induced acute leukocytosis in the blood, which was accompanied by enhancedmyeloid cell and specifically granulocyte cell counts in the spleen and blood of ApoE−/− mice fed with Westerndiet. These cellular immune changes were further associated with increased levels of pro-inflammatory cytokineslike IL-6 and TNF-α. Moreover, endogenous stroke-induced endothelial activation as well as CXCL-1 and CXCL-2expression were increased, thus resulting in accelerated leukocyte, particularly granulocyte accumulation, andenhanced ischemic tissue damage. The latter was revealed by larger infarct volumes and increased local DNA frag-mentation in ischemic brains of ApoE−/− mice onWestern diet. These effects were not observed in wildtype miceonnormal orWestern diet and inApoE−/−mice onnormal diet. Our data demonstrate that the combination of bothApoE knockout and a high cholesterol diet leads to increased ischemia-induced peripheral and cerebral immuneresponses, which go along with enhanced cerebral tissue injury. Thus, clinically predisposing conditions relatedto peripheral inflammation such as hypercholesterolemia should be included in up-coming preclinical strokeresearch.

© 2013 Elsevier Inc. All rights reserved.

Introduction

Inflammation is involved in stroke-induced brain damage (Iadecolaand Anrather, 2011). However, difficulties remain with regard to thetranslation of inflammation-targeting therapeutic approaches frompre-clinical to clinical studies (Endres et al., 2008; Enlimomab AcuteStroke Trial Investigators, 2001; Fisher et al., 2009), which might bedue to a neglect of co-morbidities in clinical trial designs.

Stroke induces inflammatory responses in the ischemic brain.Glial and endothelial cell activation lead to immune cell infiltration,exacerbating ischemic tissue damage (Dirnagl et al., 1999). Nevertheless,stroke also provokes peripheral immune responses, which influence sec-ondary lesion growth and thus modulate long-term outcome (Macrez

, University of Duisburg-Essen: +49 201 723 1660.Doeppner).ect.com).

ghts reserved.

et al., 2011). These systemic immune reactions involve a rapid activationof the immune system with an increased cytokine and chemokine pro-duction in the blood and spleen (Emsley et al., 2003; Offner et al., 2006;Smith et al., 2004), followed by secondary immunosuppression andincreased susceptibility to bacterial infections (Dirnagl et al., 2007;Meisel et al., 2005; Prass et al., 2003).

There is emerging evidence that inflammatory factors outside thebrain markedly influence stroke susceptibility and outcome. Systemicinflammationhas been shown to exacerbate brain damage in experimen-talmodels of cerebral ischemia (Langdon et al., 2010;McColl et al., 2009).As such, peripherally delivered IL-1β is known to increase tissue damagevia mechanisms involving modulation of tight junction proteins andpriming of peripheral innate immune cells (McColl et al., 2008). The clin-ical relevance of the interaction between peripheral inflammation andbrain tissue injury is emphasized by epidemiological and clinical studies,demonstrating a close correlation between increased plasma IL-6 levelsand stroke severity (Smith et al., 2004).

457J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

Nevertheless, the majority of pre-clinical studies do not considerco-morbidities such as hypercholesterolemia, which are frequentlyshown in stroke patients. Hypercholesterolemia does not only promoteatherosclerotic plaque development but also induces local inflamma-tion within the vessel wall of peripheral arteries, which is associatedwith broad systemic immune changes, affecting almost all immunecell subtypes of the innate and adaptive immune system (Drechsleret al., 2010; Smith et al., 2010; Wu et al., 2009).

A fewpre-clinical studies emphasize the role of the immune system inthe combined setting of hyperlipidemia and stroke, but are hampereddue to focusing on single molecules and immune cell subsets (Kimet al., 2008). Until now, there is only limited information about the mod-ulation of peripheral immune responses which cannot be appropriatelyattributed to effects either of hypercholesterolemia or brain ischemiadue to insufficient controls. Finally, most studies do not discriminatehow hypercholesterolemia is induced. Thus, dietary and genetic inter-ventions for induction of hypercholesterolemia are often combinedwithout assessing individual effects of each of these interventions.

In a comprehensive approach we investigated ischemia-inducedperipheral immune responses and their concomitant reactions in theischemic brain with special emphasis on immune cell infiltrates, endo-thelial cell activation and brain tissue injury. Wildtype and ApoE−/−

mice fed with a regular or cholesterol-rich food were used in order tomimic different severities of hypercholesterolemia.

Materials and methods

Mice

Experiments were performed in accordance to National Institutes ofHealth (NIH) Guidelines for the Care and Use of Laboratory Animalswith local government approval. Male wildtype and ApoE−/− micewhich were generated on the same C57BL/6 genetic background wereeither fed with a normal chow or a cholesterol rich chow, so calledWestern diet (TD.88137 Adjusted Calories Diet, Harlan Laboratories) for6 weeks and submitted to left-sided middle cerebral artery occlusion(MCAO) or sham surgery as described below. Animals were randomlyattributed to treatment paradigms, and experimenters were blinded toboth treatment and data analysis. After 24 and 72 h,micewere sacrificedfor quantification of peripheral blood and spleen leukocyte subsets byflow cytometry as well as for analysis of plasma cytokine concentrationand splenic mRNA expression (n = 6–10 per group and time point).Immunohistochemical analysis of tissue injury, endothelial andmicrogliaactivation, and leukocyte infiltration of ischemic brainswas performed at72 h post-ischemia (n = 6–8 per group). Quantification of chemokineabundance (CXCL-1 and CXCL-2) in brain tissue protein lysates was per-formed via enzyme-linked immunosorbent assay (ELISA) measurementsat 24 h post-ischemia (n = 6–8). For quantification of peripherally de-rived leukocyte subsets in ischemic brains, an additional set of micewas generated to perform flow cytometry analysis on tissue isolatedimmune cell suspensions at 72 h post-ischemia (n = 15–20 per group).In total 12–16 mice per group were exposed to MCAO and 6–10 tosham operation for analysis at 24 h post-ischemia. For analysis at 72 hpost-ischemia a total number of 21–28 mice per group were exposed toMCAO and 6–10 to sham operation. A detailed description of the experi-mental setup is presented in Suppl. Fig. S1.

Induction of focal cerebral ischemia

Induction of stroke was performed using the intraluminal monofila-ment occlusionmodel as described previously (Herz et al., 2012). Briefly,animals were anesthetized with 1% isoflurane (30% O2, remainder N2O).Rectal temperature was maintained between 36.5 and 37.0 °C using afeedback-controlled heating system. Cerebral blood flow was analyzedby laser Doppler flow (LDF) recordings andwasmonitored during ische-mia for up to 15 min after reperfusion onset. For induction of cerebral

ischemia, amidline neck incisionwasmade and the left common carotidartery (CCA) and the external carotid artery were isolated and ligated. Anylon monofilament coated with silicon resin was introduced through asmall incision into the CCA and advanced to the carotid bifurcation forinduction of MCAO. Twenty minutes later, reperfusion was initiated bymonofilament removal. In sham-operated animals, all procedures wereperformed exactly as for MCAO except for occlusion of the MCA. Assuch, the left CCA and the external carotid arterywere isolated and ligatedand a small incisionwasmade in the CCA followed by immediate ligationwithout introducing a filament. After the surgery, wounds were carefullysutured, anesthesia was discontinued and animals were placed back intotheir cages.

Analysis of post-ischemic tissue injury and immunohistochemistry

For infarct volumemeasurement and immunohistochemical analysis,mice were transcardially perfused with ice-cold 0.9% NaCl at 72 hpost-ischemia. Brains were removed and fresh frozen on dry ice. Todetermine infarct volumes 20 μm cryostat sections (every 400 μm be-tween +1 mm and −3 mm from bregma) were stained with cresylviolet followed by a computer-based analysis of infarct volumes onscanned sections using Image J software by subtracting the area of thenon-lesioned ipsilateral hemisphere from that of the contralateralside. Infarct volume sizes were calculated by integration of the lesionedareas.

For assessment of cell death, endothelial activation and analysis ofinflammatory reaction, 20 μm cryostat sections taken at the level ofbregma of mice that were transcardially perfused with 0.9% salinewere used for immunohistochemistry using published fluorescenceand avidin-peroxidase protocols (Bacigaluppi et al., 2009; Herz et al.,2012). Cell death was analyzed via staining of DNA fragmentationusing terminal transferase dUTP nick end labeling (TUNEL) accordingto the manufacturer's protocol (In situ Cell Death Detection Kit, Roche,Switzerland). For the remaining conventional immunohistochemistries,the following primary antibodies were used: rabbit anti-Iba1 (1:100;Wako, Germany), rat anti-CD45 (1:20, BD Biosciences, Germany),biotinylated goat anti-mouse ICAM-1 (1:50, R&D Systems, USA), ratanti-mouse CD31 (1:500, BD Biosciences, Germany), and goat anti-mouse VCAM-1 (1:100, R&D Systems). For endothelial markers, slideswere fixed with methanol/acetone followed by incubation with primaryantibodies in phosphate buffered saline (PBS) containing 0.2% Tween-100. As secondary antibodies, Cy3-conjugated anti-rat (1:100, JacksonLab., UK) antibody, streptavidin Alexa Fluor 488 (1:100, Invitrogen,Germany) andAlexa Fluor 488 conjugated anti-goat antibody (Invitrogen,Germany) were used for CD31/ICAM-1 and CD31/VCAM-1 co-stainings,respectively. For Iba-1 stainings, slides were fixed with 2% paraformalde-hyde (PFA) followed by incubation with the primary antibody over nightat 4 °C in PBS 0.3% Triton-X. Primary antibody binding was visualized byAlexa Fluor 488 anti-rabbit antibody (1:500, Invitrogen, Germany). Foravidin-peroxidase staining of CD45, endogenous peroxidase was blockedwith 0.3% hydrogen peroxide in 70% methanol in Tris-buffered saline(TBS). Sections were incubated with biotin-conjugated secondary anti-bodies, immersed with Vectastain® AB kit (Vector Laboratories, USA)and revealed with diamino benzidine tetrahydrochloride (DAB) (Sigma,USA).

Sections were evaluated under a bright light microscope (Axioplan;Zeiss, Germany) connected to a CCD camera (Microfire; AVT Horn,Germany) and a fluorescence microscope (BX 41; Olympus, Germany)connected to a CCD camera (CC12; Olympus).

Cellular injury was analyzed by counting TUNEL+ cells in 6 differentregions of interest (ROI, 62.500 μm2) and mean values were calculatedfor individual animals. For analysis of CD31/ICAM-1 and CD31/VCAM-1co-stainings, positively stained CD31 vessels and ICAM-1 or VCAM-1vessels were counted separately on photographs of the same region ofinterest (ROI, 125.00 μm2 each) in 3 defined ROIs within the lesion

458 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

followed by calculation of percent values for ICAM-1 and VCAM-1positive vessels of all CD31 positive vessels.

In order to analyze the inflammatory reaction, Alexa Fluor 488 andDAB-labeled Iba-1 and CD45 stainings, respectively, were evaluated at40× magnification and images were digitalized with a CCD camera.We examined photographs of 6 regions of interest (ROI, 56.800 μm2

each) within the border region of the lesion for both antigens. Imageswere converted into grayscale and the inflammatory area was quantifiedusing Image J software (NIH, USA) as previously described (Herz et al.,2012). Data are expressed as area per ROI.

Flow cytometry

Animals were euthanized by i.p. injections of chloralhydrate(420 mg/kg body weight (BW)). Blood was taken from the vena cavaand collected in ethylenediaminetetraacetate (EDTA) coated collectiontubes followed by transcardial perfusion with ice-cold 0.9% NaCl.Spleens were removed and single cell suspensions were prepared bymechanical dissociation through a 70 μm cell strainer (BD Biosciences,Germany). Blood samples were centrifuged at 1000 g for 10 min andthe supernatantwas frozen at−80 °C until further analysis for cytokines.For the remaining pellet and spleen cell suspensions, red blood cells werelysed and washed twice with fluorescent activated cell sorter (FACS)buffer (0.5% bovine serum albumin (BSA) in phosphate buffered saline(PBS)). Brains were dissected and hemispheres were divided intoipsilesional and contralesional parts. Four to five hemispheres werepooled per experiment (3–4 independent experiments per group) andhomogenized through a 70 μm cell strainer (BD Biosciences) followedby 37% Percoll (in 0.01 N HCl/PBS) centrifugation (20 min, 2800 g).Myelin was removed and the remaining cell pellet was washed twice inFACS buffer.

Isolated CNS, spleen and blood cells were incubated with blockingantibody rat anti-mouse CD16/CD32 (FC fragment, BD Biosciences,Germany) for 15 min at+4 °C followed by incubationwith the followingantibody cocktail for further 30 min: rat anti-mouse CD45 Alexa Fluor700 (BDBiosciences, Germany), rat anti-mouseCD11bFITC (eBiosciences,Germany), hamster anti-mouse CD11c APC (BD Biosciences, Germany),rat anti-mouse CD86 PE (eBiosciences, Germany), and rat anti-mouseMHCII eFluor450® (eBiosciences, Germany). In order to exclude deadcells from analysis, 0.25 μg/ml propidium iodide was added directly be-fore measurement. Cells were analyzed using an LSR II (BD Biosciences)and FACS Diva software (BD Biosciences). Absolute cell numbers forblood and CNS were analyzed using BD TrueCount beads (BD) accordingto the protocol on the basis of CD45 positive events. Spleen cell countswere determined by trypan blue exclusion and a hemocytometer.

Quantification of chemokine expression in brain tissue lysates

For analysis of CXCL-1 and CXCL-2 expression, tissue samples wereharvested from the striatum of mice sacrificed at 24 h post-ischemia bytranscardiac perfusionwith 0.9%NaCl. Tissue sampleswere homogenizedand lysated on ice in NP40 lysis buffer containing phosphatase andprotease inhibitor cocktails. Protein content was measured usingthe bicinchoninic acid assay (BCA assay, Biorad, USA) and chemokineabundance was determined using the mouse CXCL-1 and CXCL-2Quantikine ELISA kit (R&D Systems, USA) according to the manufac-turers' protocol.

Cytokine response

Plasma sampleswith a volumeof 25 μl were used for quantification ofIL-6 and TNF-αusing the Procarta Cytokine assay kit (Panomics, Fremont,CA) according to the manufacturer's recommendations. The assay wasrun with a Luminex200 instrument using Luminex IS software (LuminexCorporation, Austin, TX).

Gene expression analysis by RT-PCR

For gene expression studies, total RNA was extracted from spleencell lysates using RNA purification kit (Promega). First-strand cDNAwas synthesized using 500 ng of total RNA and TaqMan reverse tran-scription reagents (Applied Biosystems). The PCR amplification wasperformed in 96-well optical reaction plates for 40 cycles with eachcycle at 94 °C for 15 s and 60 °C for 1 min using FAST SYBR® GreenMaster Mix (Applied Biosystems). The PCR was conducted on theStepOne Real-Time PCR system (Applied Biosystems). CT values werenormalized for the housekeeping gene β-actin [ΔCT = CT (targetgene) − CT (β-actin)] and compared with a calibrator sample using theΔΔCT formula [ΔΔCT = ΔCT (sample) − ΔCT (calibrator)]. As calibratorsample we utilized spleen cell lysates of naïve C57BL/6 mice of thesame age, sex and strain. The following primers were used: catalase(cat) fwd: 5′-ACATGGTCTGGGACTTCTGG-3′, catalase (cat) rev: 5′-CAAGTTTTGATGCCCTGGT-3′, NADPH oxidase (nox) fwd: 5′-ACTGCGGAGAGTTTGGAAGA-3′, NADPH oxidase (nox) rev: 5′-GGTGATGACCACCTTTTGCT-3′; and β-actin fwd: 5′-GCTACAGCTTCACCACCACAG-3′,β-actin rev: 5′-GGTCTTTACGGATGTCAACGTC-3′.

Quantification of reactive oxygen species (ROS) producing granulocytes

ROS levels were measured using the fluorescent probe 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester(CM-H2DCFDA) according to the manufacturer's recommendations(Invitrogen, Germany). In order to specifically identify ROS producinggranulocytes, splenocytes of wildtype mice on normal diet and ApoE−/−

mice on Western diet were sorted for CD11b by magnetic activated cellsorting (MACS®) technology. Isolated cells were incubated with 5 μMCM-H2DCFDA for 5 min (37 °C, 5%CO2) andwashed twicewith PBS after-wards followed by activation with 500 ng/ml Phorbol 12-myristate13-acetate (PMA) for 30 min at 37 °C. Cells were harvested on ice andROS producing granulocytes were immediately quantified via flowcytometry.

Data analysis

Statistical analysis was performed using PASW Statistics 18.0. Flowcytometry data were analyzed by three-way ANOVA evaluating effectsof ApoE−/−, Western diet andMCAO. Histological and protein expressiondata from non-ischemic and ischemic brain tissues were analyzed bytwo-way ANOVA. Whenever there was a significant effect by strain, dietor MCAO or a significant interaction between these factors, we furtherdetermined the effects between single groups using one-way ANOVAfollowed by Gabriel or Games Howell post hoc tests depending on vari-ance homogeneity.

Results

ApoE−/− mice fed with Western diet reveal increased stroke-inducedtissue damage

Induction of 20 min ofMCAO resulted in reproducible infarcts of theleft-sided striatum inwildtypemice as revealed by cresyl violet staining.Significantly increased infarct volumeswere noticed on day 3 in ApoE−/−

mice fed with a high cholesterol diet (Fig. 1A). Accordingly, two-wayANOVA showed a significant interaction effect for ApoE−/− × diet(F1,23 = 7.753; p b 0.05). We and others have recently shown thatthese mice exhibit the highest cholesterol levels compared with theother experimental groups examined, i.e. wildtype mice on normal orWestern diet and ApoE−/− mice on normal diet (ElAli et al., 2011; Kimet al., 2008). Likewise, TUNEL staining revealed an increased density ofDNA-fragmented cells in ischemic tissues of ApoE−/− mice on Westerndiet, thus confirming that hypercholesterolemia exacerbates ischemicbrain damage (Fig. 1B).

Wildtype normal diet ApoE Western diet- /-

Infarct volume

Wildtype normal diet ApoE Western diet- /-

DNA fragmentationA) B)

Fig. 1.ApoE−/−mice fedwithWestern diet reveal increased ischemic brain injury. Brains ofwildtype andApoE−/−mice fedwith a normal or cholesterol rich (so calledWestern) chowandexposed to 20 min MCAO followed by 72 h reperfusion were analyzed for infarct volumes via cresyl violet stainings (A) and DNA fragmented cells via TUNEL staining (B). Photographsshow representative photomicrographs of the respective stainings for wildtype mice on normal diet and ApoE−/− mice fed with Western diet. Scale bar is 1 mm in (A) and 50 μm in(B). Bars represent mean values + SD. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

459J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

Cerebral ischemia induces early leukocytosis in ApoE−/− mice fed withWestern diet

There is emerging evidence for an interaction between peripheral in-flammation and post-ischemic inflammatory processes in the brain thatcontributes to development of ischemic brain injury (Macrez et al.,2011). However, hypercholesterolemia is known to also induce broadsystemic immunological changes on its own (Drechsler et al., 2010; Wuet al., 2009). To investigate whether or not ischemia-induced peripheralinflammation is affected by hypercholesterolemia, we first determinedabsolute numbers of circulating leukocytes both in the blood and in thespleen. Since immune responses are regulated in a time-dependentmanner, involving a rapid acute activation (Offner et al., 2006) followedby suppression at sub-acute time points (Dirnagl et al., 2007), we chosetwo different time points for analysis of peripheral immune responses,i.e. 24 and 72 h post-ischemia.

While we did not observe any effect of cerebral ischemia on theabsolute amount of circulating leukocytes in wildtype mice on nor-mal or on Western diet as well as in ApoE−/− mice on normal diet,a significant ischemia-induced acute leukocytosis was detected in theblood of ApoE−/− mice on Western diet at 24 h after stroke (Fig. 2A).Three-way ANOVA demonstrated a significant interaction effect forApoE−/− × diet × ischemia (F1,42 = 9.852; p b 0.01). Contrary to obser-vations in the blood, we did not detect any significant changes of leuko-cyte numbers in the spleen after 24 h. After 72 h significantly reducednumbers of splenocytes were noticed in ischemic mice reflectingischemia-induced immunosuppression that was independent of hyper-cholesterolemia (Fig. 2A), suggesting differentially regulated immune re-actions in the spleen and blood following brain ischemia.

In order to evaluate to which extent lymphocytes might contributeto increased overall leukocyte numbers in the blood and spleen, wealso analyzed CD11b−CD11c− SSClow cell counts. At 24 h post-ischemiathe amount of circulating lymphocytes was increased in ApoE−/− miceon Western diet but decreased again at 72 h (Fig. 2B). On the contrary,lymphocyte cell counts were not regulated in the spleen at this early

time point. However, after 72 h lymphocyte numbers are reduced dueto ischemia, particularly in wildtypemice and ApoE−/−mice on normaldiet. This immunosuppressive effect was less pronounced in groups onWestern diet.

Dendritic cells are one of the first cells accumulating in large numbersin the ischemic brain (Gelderblom et al., 2009) and dendritic cells main-tain peripheral inflammation through the release of cytokines andchemokines during hypercholesterolemia. Therefore, we next quantifiedCD11chigh expressingmononuclear leukocytes (mainly dendritic cells) inthe blood and spleen of hypercholesterolemic ischemic mice. We did notdetect any strong differences for dendritic cell numbers in the blood andspleen at 24 h post-ischemia (Fig. 2C). However, the total number ofCD11chigh cells was exclusively reduced in ischemic wildtype mice onnormal diet but not in groups of hypercholesterolemia, resulting in signif-icantly increased CD11chigh cell numbers in ischemic ApoE−/− mice onWestern diet as compared to ischemic wildtype mice (Fig. 2C). Ingeneral, we detected relatively high percentages of circulating dendriticcells even in sham operated wildtype mice on normal diet with2.64 ± 1.14% whereas na ve wildtype mice without any interventionrevealed 1.32 ± 0.1% (n = 6; data not shown) which might be ex-plained by anesthesia and surgery induced stress responses (Ho et al.,2001).

In order to characterize the activation status of these cells we furtherdetermined the proportion ofMHCII+ and CD86+dendritic cells. Exceptfor an ischemia-induced reduction in MHCII expression in blooddendritic cells of wildtype mice on normal diet after 72 h reflectingpost-ischemic immunosuppression, we did not detect any furtherdifferences in the blood or in the spleen, neither due to ischemianor due to genetically or dietary-induced hypercholesterolemia atthese two time points (Suppl. Fig. S2A). Similarly to MHCII, we did notobserve any strong regulations for the amount of CD86 expressingdendritic cells after 24 h, whereas slightly increased proportions ofCD86 expressing cells were detected in the blood and spleen after72 h in hypercholesterolemic, most strongly in ApoE−/− mice. Thesefrequencies were not further modulated by MCAO (Suppl. Fig. S2B).

Blood Spleen

x106

cells

/ml

x106

cells

/ml

x106

cells

x106

cells

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

x105

cells

/ml

x105

cells

/ml

x106

cells

x106

cells

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

x104

cells

/ml

x104

cells

/ml

x105

cells

x105

cells

A)

B)

C)

leukocytes

lymphocytes

dendritic cells

Fig. 2. Ischemia induces rapid leukocytosis in the blood of ApoE−/− mice fed withWestern diet. Quantification of absolute leukocyte counts in the blood using flow cytometry and viablecell count analysis in the spleen determined by trypan blue exclusion staining at 24 and 72 h post-ischemia in wildtype mice on normal diet and in animals with genetically and dietaryinduced hypercholesterolemia (wildtype mice on Western diet, ApoE−/− mice on normal and on Western diet) (A). Lymphocyte (B) and dendritic cell counts (C) were determined byanalyzing CD11b−CD11−SSClow and CD11chigh frequencies of CD45+ cells, respectively. Absolute cell counts were calculated for individual mice and mean values + SD are summarizedin the graphs. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

460 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

Cerebral ischemiamodulates peripheral myeloid cell counts in ApoE−/−micefed with Western diet

Among the various immune cell subtypes, myeloid cells are the firstthat infiltrate the brain and major contributors to ischemic brain injury(Jin et al., 2010; Lakhan et al., 2009). Besides, hypercholesterolemia in-duces peripheral monocytosis and neutrophilia (Drechsler et al., 2010;Wu et al., 2009). In light of this, we determined the absolute cell countsof CD11b+CD11c− cells in the blood and in the spleen. According toscatter characteristics, we further distinguished granulocytic frommono-nuclear CD11b+ cells mainly containing macrophages and monocytes.

Analysis of acute immune responses after 24 h revealed thatischemia induces a significant increase of myeloid and particularlygranulocyte cell numbers in the blood but not in wildtype mice onWestern diet and ApoE−/− mice on normal diet (Fig. 3A). Three-wayANOVA confirmed a ApoE−/− × diet × MCAO interaction effect forCD11b+ cell counts (F1,42 = 6.572; p b 0.05). Splenic myeloid cellcounts of ischemic ApoE−/− mice on Western diet were also increasedas compared to ischemic wildtype mice on normal diet (Fig. 3B).

After 72 h, circulating myeloid cell numbers declined in all experi-mental groups indicating that surgery and anesthesia associated immuneresponses were generally down-regulated. However, ischemic ApoE−/−

mice on Western diet still revealed significantly increased amounts ofcirculating CD11b+ cells as compared to ischemic wildtype mice onnormal diet (Fig. 3A). In contrast, myeloid cell numbers in the spleendecreased in all ischemic groups as compared to sham controlsreflecting once again immunosuppression. Nevertheless, the most pro-nounced effectwas observed forwildtype andApoE−/−mice on normal

diet, resulting in significantly increased cell counts in ischemic ApoE−/−

mice on Western diet as compared to ischemic wildtype mice on aregular diet (Fig. 3B).

Ischemia-induced cerebral endothelial activation is enhanced in ApoE−/−

mice on Western diet

Stroke induces endothelial activation, which is reflected by increasedadhesion molecule expression thereby enabling the infiltration of pe-ripheral immune cells that contributes to brain tissue damage (Dirnaglet al., 1999). In order to investigate whether or not ischemia-inducedendothelial cell activation is modulated by peripheral hypercholesterol-emia, we quantified the amount of intracellular adhesion molecule 1(ICAM-1) and vascular adhesion molecule 1 (VCAM-1) expressingvessels within ischemic brain tissues. To exclude any effects on vesseldensity, we performed co-stainings for the pan endothelial cell markerCD31 combined with ICAM-1 or VCAM-1 (Figs. 4A, B). Vessel densitywas not affected neither by the ApoE knockout nor by a high cholesteroldiet (Figs. 4A, B, C). However, both absolute and relative numbers ofICAM-1 expressing vessels were significantly increased in ischemicbrains of ApoE−/− mice on Western diet but not in wildtype mice onWestern diet and ApoE−/− mice on normal diet as compared towildtypemice on normal diet (Figs. 4A, D, E). Although less pronouncedcompared to ICAM-1 we observed significantly increased amounts ofVCAM-1 positive vessels in ischemic brains of ApoE−/−mice onWesterndiet as compared to wildtype mice on normal diet (Figs. 4B, F, G). Sinceboth, a high cholesterol diet and the ApoE−/− phenotype might alsoalter the cerebral vascular endothelium independent of brain ischemia

x106

cells

x105

cells

/ml

x105

cells

/ml

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE -/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

x106

cells

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE -/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

A)

B)

Fig. 3. Increased proportion of peripheral myeloid leukocyte subsets in ischemic ApoE−/− mice fed with Western diet. Flow cytometry analysis of isolated white blood cells (A) andsplenocytes (B) for quantification of CD11b+ myeloid cell counts at 24 and 72 h post-ischemia. Myeloid cells were further distinguished based on their scatter properties to differentiatebetween granulocytic cells (CD11b+SSChigh, filled bars) andmacrophages/monocytes (CD11b+SSClow, open bars). Bars represent mean values − SD. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001indicating significances for total CD11b+ cells. CD11b+SSChigh and CD11b+SSClow cell counts were separately analyzed for significance: (A) ##p ≤ 0.01 indicating significance forCD11b+SSChigh of MCAO/ApoE−/−/Western diet vs. sham/ApoE−/−/Western diet; MCAO/wildtype/normal diet, MCAO/wildtype/Western diet and MCAO/ApoE−/−/normal diet at 24 h.#p ≤ 0.05 indicating significance for CD11b+SSClow of MCAO/ApoE−/−/Western diet vs. MCAO/wildtype/normal diet, MCAO/wildtype/Western diet and MCAO/ApoE−/−/normal dietat 72 h. (B) ##p ≤ 0.01 indicating significance for CD11b+SSChigh of MCAO/ApoE−/−/Western diet vs. MCAO/ApoE−/−/normal diet at 24 h; and for CD11b+SSChigh as well asCD11b+SSClow of MCAO/wildtype/normal diet vs. sham/wildtype/normal diet at 72 h.

461J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

(Drake et al., 2011; Jin et al., 2013; Zechariah et al., 2013), we furtherperformed analysis on contralateral hemispheres. Vessel density,ICAM-1 and VCAM-1 expression were not regulated, neither by theApoE knockout nor by the Western diet (Suppl. Figs. S3A, B). Of note,ICAM-1 expression in the contralateral hemisphere was similar to thevalues in the ipsilateral hemisphere in wildtype mice on normal dietsuggesting that endothelial activation has already partially decreasedto basal levels at 72 h reperfusion in thesemice (Suppl. Fig. S3A). There-fore, we further investigated ICAM-1 expression at 24 h reperfusion,when immune cell infiltration is supposed to start (Gelderblom et al.,2009). Indeed, compared to 72 h reperfusion ICAM-1 expression wasapproximately 2 times higher at 24 h post-ischemia in wildtype miceon normal diet (Fig. 4D, Suppl. Fig. S3C). In contrast to 72 h reperfusionwhere exclusively ApoE−/− mice onWestern diet showed significantlyincreased ICAM-1 expression (Fig. 4D) endothelial activation was

Fig. 4. Endothelial cell activation is increased in ischemic brain tissue of ApoE−/− mice fed withendothelial cellmarker CD31 at 72 h post-ischemiawithin ischemic brain tissues are shown as rApoE−/− mice fed with normal or Western diet (A). Scale bar is 50 μm. Absolute numbers of CICAM-1+ vessels (E) and VCAM-1+ vessels (G) were determined. Bars represent mean values

significantly enhanced in all examinedmodels of hypercholesterolemia(wildtypemice onWestern diet, ApoE−/−mice on normal andWesterndiet) after 24 h (Suppl. Fig. S3C).

Increased cerebral inflammatory reactions in ischemic ApoE−/− mice onWestern diet

Besides endothelial activation, chemokines play a major role inleukocyte recruitment to the injured brain. Due to our observations ofincreased leukocytosis and particularly granulocytosis in the peripheryof ApoE−/− mice onWestern diet, we were interested in the regulationof chemoattractants for this leukocyte subset in ischemic brains of hy-percholesterolemic mice. Therefore, we analyzed protein expressionof CXCL-1 and CXCL-2 in non-ischemic and ischemic brain tissues at24 h post-ischemia for all experimental groups. In order to exclude

Western diet. ICAM-1 (A) and VCAM-1 (B) stainings combined with stainings for the panepresentative photomicrographs comparingwildtypemice on normal orWestern diet andD31+ (C), ICAM-1+ (D) and VCAM-1+ (F) positive vessels as well as percentage values of+ SD. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

ICAM-1 CD31 ICAM-1 / CD31

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

VCAM-1 CD31 VCAM-1 / CD31

norm

al d

iet

Wes

tern

die

tno

rmal

die

tW

este

rn d

iet

Apo

E-/

-W

ildty

pe

norm

al d

iet

Wes

tern

die

tno

rmal

die

tW

este

rn d

iet

Apo

E-/

-W

ildty

pe

A)

B)

C)

D)

E)

F)

G)

462 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

463J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

any potential differences as a consequence of increased infarct volumes,ELISA measurements were performed on protein lysates isolated fromischemicMCA territories and the respective non-ischemic tissue volumes.As shown in Fig. 5, both chemokines were upregulated in ischemic braintissues of all groups of hypercholesterolemia. Nevertheless, the highestincrease was observed in brain tissues of ApoE−/− mice on Westerndiet. Interestingly, we also observed an increase in non-ischemic tissues(Figs. 5A, B), suggesting that an inflammatory environment exists in thebrain of these mice with regard to specific molecular regulations.

In addition to molecular inflammatory reactions, we were also in-terested how endogenous inflammatory cellular responses might beaffected. Therefore, we analyzed activation of immune reactive cells ofthe brain, i.e. the amount of activated microglia was determined via

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

norm

A)

B)

C)

Fig. 5. Increased cerebral inflammatory reactions in ischemic ApoE−/− mice onWestern diet. Clysates isolated from ischemicMCA territories and the respective non-ischemic tissue volumesdiet at 24 h post-ischemia. Measured chemokine amounts were related to protein amounts. Micof infracted tissues and respective non-ischemic tissue areas inwildtypemice on normal diet anWestern diet, ApoE−/− mice on normal and onWestern diet) 72 h after MCAO (C). Bars repre

immunohistochemistry for Iba-1 in ischemic and non-ischemic braintissues of all experimental groups at 72 h post-ischemia. Whereas wedetected no difference in non-ischemic tissues, we observed significantlyincreased amounts of activated microglia in ischemic tissues of ApoE−/−

mice on Western diet (Fig. 5C). Despite a slight increase due to mereWestern diet and ApoE knockout, the most pronounced effect was de-tected in ischemic ApoE−/− mice on Western diet (Fig. 5C).

Granulocyte infiltration into the ischemic brain tissue is accelerated inApoE−/− mice on Western diet

To check whether the increased adhesion molecule and chemokineexpression correlates with peripheral immune cell infiltration we

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

normal diet Western diet normal diet Western diet

Wildtype ApoE

-/-

al diet Western diet normal diet Western diet

Wildtype ApoE

Wildtype normal diet

ApoE Western diet-/-

XCL-1 (A) and CXCL-2 (B) abundance was determined by ELISA measurements of proteinof wildtype mice on normal diet orWestern diet and ApoE−/−mice on normal orWesternroglia activationwas quantified by immunohistochemistry for Iba-1 within the penumbrad in animalswith genetically and dietary-induced hypercholesterolemia (wildtypemice onsent mean values + SD. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

464 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

performed immunohistochemical and flow cytometry analysis ofCD45+ cells in ischemic brain tissues. In order to avoid false interpreta-tion of flow cytometry data due to varying amounts of ischemic tissuesper hemispheres of the different experimental groups, we first deter-mined leukocyte infiltration locally in 6 defined regions of interest inthe penumbra region of the infarct via immunohistochemistry. Here,we detected an increased infiltration of CD45+ cells in ApoE−/− micefed with Western diet, but not in wildtype mice on Western diet andApoE−/− mice on normal diet as compared to wildtype mice on normaldiet (Figs. 6A, C). Since hypercholesterolemia by its own might induceincreased leukocyte infiltration independent of ischemia (Drake et al.,2011), we also included analysis of contralateral hemispheres. Similarly

non-ischem

Macrophages / monocytes

Lymphocytes

Wildtype normal diet

ApoE Western diet- /-

A) B)

D)

F)

Fig. 6. Granulocyte infiltration is increased in ischemic brains of ApoE−/− mice fed with WestCD45+ on cryostat sections of ischemic (A, C) and non-ischemic (B) brain tissue at 72 h reperfudiet with ApoE−/−mice onWestern diet in ischemic tissues is shown (A). The amount of differen(CD45high, SSClow, CD11b−, CD11c−) (D), dendritic cells (CD45high, SSClow, CD11c+) (E), monocyte(G). Bars represent mean values of 3–4 independent experiments + SD. For each experiment 5 h

to adhesion molecule expression and microglia activation, we did notdetect any differences in CD45 immunoreactivity in non-ischemicbrain tissues between the different experimental groups (Fig. 6B).

To quantify the composition of immune cell infiltrateswe performedflow cytometry using CD45, CD11b, and CD11c co-stainings and scatterproperties to distinguish granulocytes, macrophages/monocytes, den-dritic cells and lymphocytes as previously described (Gelderblom et al.,2009). Absolute cell counts per hemisphere were calculated and inter-experimental variations due to isolation procedures were corrected byrelating values of ipsilateral to values of contralateral hemispheres forindividual experiments. Three to four independent experiments (eachcontaining 5 pooled brain hemispheres) were performed per group.

ischemicic

Granulocytes

Dendritic cells

C)

E)

G)

ern diet. The inflammatory reaction was analyzed via immunohistochemical staining forsion. A representative microphotograph of a staining comparing wildtype mice on normalt immune cell subsets was quantified via flow cytometry (D–G) to distinguish: lymphocytess/macrophages (CD45high, SSClow, CD11b+, CD11c−) (F) and granulocytes (CD45high, SSChigh)emispheres were pooled. *p ≤ 0.05.

465J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

While the amount of lymphocytes, macrophages/monocytes andCD11chigh cells (mainly dendritic cells) as well as the activation (MHCII,CD86) of these cells did not significantly differ between the experimentalgroups (Figs. 6D, E, F; Suppl. Fig. S4), we detected significantly increasedamounts of granulocytes in ischemic ApoE−/− mice onWestern diet butnot in wildtype mice onWestern diet and ApoE−/− mice on normal diet(Fig. 6G). Accordingly, two-way ANOVA showed a significant interactioneffect for ApoE−/− × diet (F1,11 = 5.022; p b 0.05).

Myeloid cells of ApoE−/− mice on Western diet potentially contribute tocerebral tissue injury by release of pro-inflammatory cytokines and increasedROS production

With regard to the above presented modulation of myeloid andspecifically granulocytic cell frequencies both in the periphery and inthe brain of hypercholesterolemic ischemic mice, we were further in-terested to identify potential mechanisms via which these cell typesmight contribute to ischemic tissue injury. Therefore, we measuredconcentrations of IL-6 and TNF-α and quantified the expression of thereactive oxygen species (ROS) generating enzyme NADPH oxidase(NOX) and the anti-oxidative enzyme catalase (CAT) in the blood andin the spleen, respectively. Since the most striking effects regardingbrain tissue damage, peripheral immune response, endothelial andmicroglia activation aswell as CNS immune cell infiltrationwere detectedin ApoE−/− mice on Western diet, we compared only these mice withwildtype mice on normal diet.

We observed increased circulating IL-6 and TNF-α levels in ApoE−/−

mice on Western diet (Figs. 7A, B). Focal cerebral ischemia furtherenhanced IL-6 plasma levels of these mice but not of wildtype mice onnormal diet after 72 h (Fig. 7A). Two-way ANOVA demonstrated asignificant diet/ApoE−/− × ischemia interaction for IL-6 levels at 72 h

A)

C)

Fig. 7. Increased circulating IL-6 and TNF-α levels and elevated expression of oxidative stres(B) concentrations in the blood and mRNA expression of oxidative stress related enzymes, i.e.oxidative enzyme catalase (CAT, (D)) in splenocytes were determined in wildtype mice fed a nBars represent mean values + SD. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

(F1,18 = 8.930; p b 0.01). While NOX expression is increased inApoE−/−mice on high cholesterol diet we detected no further regulationof NOX expression by ischemia at both time points (Fig. 7C). Neither theappliedmodel of hypercholesterolemia nor focal cerebral ischemia influ-enced the expression of the anti-oxidative enzyme catalase (Fig. 7D). Toverify, whether the observed increased NOX expression at mRNA levelsis relevant for ROS production in ApoE−/−mice onWestern diet, we per-formed an H2DCFDA assay on activated MACS® sorted CD11b+ cells andanalyzed the percentage of SSChigh H2CFDA+ cells via flow cytometry. Inaccordance to increased NOX expression in splenocytes of ApoE−/−miceonWestern diet the frequency of ROS producing granulocyteswas signif-icantly increased as compared to wildtype mice on normal diet (Suppl.Fig. S5).

Discussion

In the present study, we demonstrate that hyperlipidemia as wellas stroke induces marked changes in peripheral immune responses,which are differentially regulated in a combined setting. The systemicimmunomodulation is accompanied by an increased endogenousstroke-induced endothelial cell and microglia activation, resulting inenhanced leukocyte, particularly granulocyte, accumulation in ischemicbrains and increased tissue injury.

Hypercholesterolemia can be induced by genetic intervention,e.g. usingApoE−/−mice, or by feeding the animalswith a high cholesterolchow. Both settings have been shown to induce leukocytosis and specifi-cally neutrophilia via differentmechanisms. Dietary-induced neutrophiliais caused by an increased release of neutrophils from the bone marrow(Drechsler et al., 2010), whereas ApoE−/− by its own leads to an elevatedhematopoietic stem cell proliferation in the bone marrow, thus resultingin peripheral monocytosis and neutrophilia (Murphy et al., 2011).

B)

D)

s related enzymes in ApoE−/−mice fedwith a high cholesterol diet. IL-6 (A) and TNF-αthe reactive oxygen species generating enzyme NADPH oxidase (NOX, (C)) and the anti-ormal chow and in ApoE−/− mice fed with a Western diet at 24 and 72 h post-ischemia.

466 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

Accordingly,we observedmost prominent effects in the combined settingwith regard to the amount of peripheral myeloid cells in the bloodand spleen. Moreover, exclusively the combination of geneticallyand dietary-induced hypercholesterolemia led to increased cerebralischemic tissue injury. Similarly, Kim et al. (2008) also observed signif-icantly increased infarct volumes in the combined model of hypercho-lesterolemia only.

However, regarding the contribution of peripheral immune cells toischemic brain damage in hyperlipidemic mice, research mainly focusedon the role of macrophages until now (Kim et al., 2008). Therefore, weanalyzed different immune cell subsets, i.e. lymphocytes, dendritic cells,macrophages/monocytes and granulocytes, which are known to partici-pate in secondary brain damage after stroke (Macrez et al., 2011). Forthe first time to our knowledge, we show that the combination of theApoE knockout and a high cholesterol diet leads to an ischemia-inducedrapid leukocytosis in the blood, which is primarily caused by increasedmyeloid cell counts and partially increased lymphocyte numbers. Evenmore, these regulations in the bloodwere also accompanied by increasedmyeloid cell counts in the spleen of ischemic ApoE−/− mice on Westerndiet. Acute peripheral immune activation following stroke is supposed todevelop from the sympathetic nervous system, as previously discussed(Offner et al., 2006). Taken into account that hypercholesterolemia-induced systemic immune responses aremediated via increasedhemato-poietic stem cell proliferation in the bone marrow and enhanced releaseof myeloid cells from the bone marrow (Drechsler et al., 2010; Murphyet al., 2011) our data suggest that both, stroke-induced activation of sym-pathetic neurons and hypercholesterolemia-induced changes in the bonemarrow synergize in the combined model resulting in a net increase ofmyeloid granulocytic peripheral immune cells both in the blood.Although we did not perform ex vivo cell death analysis we do not as-sume, that cell death is involved in the regulation of this acute immuneresponses, because we did not detect a significant reduction in absolutenumbers for any kind of leukocyte subset due to ischemia or hypercho-lesterolemia in the blood at 24 h post-ischemia.

Of note, we did not observe these stroke-induced regulations inwildtype mice fed with normal chow diet indicating that hypercholes-terolemia decreases the threshold for induction of stroke-induced pe-ripheral immune activation. Instead of activation, we rather observedsigns of immunosuppression after 72 h as reflected by reduced amountsofmyeloid cells in the spleen and reducedMHCII expression on dendrit-ic cells of normo-cholesterolemic mice. Despite reduced splenocytenumbers which were independent of hypercholesterolemia, we foundno other significant indications for immunosuppression in ApoE−/−

mice onWestern diet, i.e. myeloid and lymphocyte counts are only par-tially reduced as compared to wildtype mice on normal diet, suggestingthat immunosuppression associated leukocyte apoptosis is less pro-nounced in these mice, e.g. wildtype mice on normal diet display a re-duction of myeloid cells by 60% as compared to 20% in ApE−/− miceon Western diet. This may be of particular importance with regard tocurrently tested stroke therapies for the prevention of bacterial infec-tions using antibiotics (Harms et al., 2008). These treatment optionsmight not help hypercholesterolemic patients but rather potentiate hy-percholesterolemia due to side effects as described for some antibiotics(Vijayalekshmi and Leelamma, 1991), which needs to be consideredsince about half of all stroke patients suffer from hypercholesterolemia(Röther et al., 2008).

Recent data point to a specific role of dendritic cells in cerebralischemia since this subset is one of the first infiltrating the ischemicbrain. Even more, particularly activated dendritic cells are found inhigh numbers (Gelderblom et al., 2009). In spite of increased amountsof splenic dendritic cells in ischemic ApoE−/− mice on Western diet ascompared to wildtype mice on normal diet, which is mainly caused byimmunosuppression-associated dendritic cell loss in the latter group,we did not detect any regulation of the amount of CD11chigh cellnumbers in the blood for both time points. Further, except for a slightincrease in CD86+ but not MHCII+ cell numbers in the periphery the

number and activation of brain infiltrated dendritic cells were notmodulated by the hypercholesterolemia, suggesting a less prominentrole of dendritic cells as compared to myeloid granulocytic cells in thissetting.

Dyslipidemia activates both cells of myeloid origin and endothelialcells. As such, it has been shown that hypercholesterolemia disruptsblood brain barrier integrity and collateral blood flow in ischemic brains(Ayata et al., 2013; ElAli et al., 2011). In addition to thesemolecular andhemodynamic alterations, we have demonstrated that all appliedmodels of hypercholesterolemia, i.e. by a high cholesterol diet andby the ApoE knockout, enhance early ischemia-induced expressionof the neuroinflammation-associated intracellular adhesion molecule 1(ICAM-1) that persisted up to 72 h post-ischemia in ApoE−/− mice onWestern diet only. Although less prominent, the vascular adhesionmol-ecule 1 (VCAM-1) was also upregulated in ischemic brain tissues ofApoE−/− mice on Western diet as compared to normo-cholesterolemicmice. This persistent endothelial cell activation correlated with an accel-erated leukocyte and particularly granulocyte infiltration in ischemicbrain tissue of ApoE−/−mice on high cholesterol diet but not inwildtypemice onWestern diet and ApoE−/− mice on normal diet suggesting thatsustained endothelial activation might extend the time window of peakimmune cell infiltration resulting in overall increased immune cell infil-trates at 72 hpost-ischemia. Considering that ICAM-1 andVCAM-1medi-ate infiltration of all immune cell subsets enhanced cerebral granulocyteinfiltration in ApoE−/− mice on Western diet might reflect ischemia-induced peripheral leuko- and specifically granulocytosis.

Despite previous descriptions of a generally increased endothelialactivation in hyperlipidemic mice (Drake et al., 2011), we did not de-tect any differences in non-ischemic tissues for ICAM-1 and VCAM-1indicating that differences for endothelial activation in ischemicbrain tissues can be rather attributed to stroke-induced responsesthan to hypercholesterolemia. Similarly, we did not detect increasedCD45 and Iba-1 immunoreactivity in non-ischemic brain tissues.However, increased abundance of CXCL-1 and CXCL-2 in non-ischemictissues of all groups of hypercholesterolemia indicates that hyperlipid-emia might partially induce an inflammatory microenvironment inthe unaffected brain as previously suggested (Drake et al., 2011). Similarregulations of CXCL-1 and CXCL-2 expression have been observed in is-chemic tissues with the most prominent increase in ApoE−/− mice onWestern diet, which might partially explain increased granulocyte infil-tration in ischemic brains. Our data indicate that increased endothelialactivation, which is accompanied by increased expression of granulocytespecific chemoattractants combined with peripheral granulocytosismight explain the significantly increased granulocyte infiltration intoischemic brains of ApoE−/− mice on Western diet.

Mechanisms underlying granulocyte mediated tissue damage in-clude excessive production of reactive oxygen species (ROS) and releaseof pro-inflammatory cytokines targeting endothelial cells but also paren-chymal neural cells (Jin et al., 2010). Indeed, we show that immune cellsof ApoE−/− mice on Western diet produce higher amounts of pro-inflammatory cytokines, such as IL-6 and reveal increased levels ofNADPH-oxidase expression, one of the major enzymes involved in ROSproduction. Even more, the percentage of ROS producing granulocytesis generally increased in ApoE−/− mice fed with Western diet. Thesedata indicate that peripheral immune cell derived inflammatory media-tors contribute to ischemic damage in hypercholesterolemic conditions.Nevertheless, the pathogenic role of granulocytes in ischemic stroke re-mains uncertain since several pre-clinical and clinical studies fail to dem-onstrate a clear correlation between neutrophil infiltration and ischemictissue damage (Beray-Berthat et al., 2003; Emerich et al., 2002).

However, the majority of translational studies did not discriminatebetween hitherto healthy subjects and subjects suffering from hy-percholesterolemia. Thus, it is conceivable that treatments targetinggranulocytes exert beneficial effects exclusively in a hypercholesterolemicinflammation prone condition. Similarly, differential roles of immune cellsubsets have been shown previously for CCR2+ monocytes which are

467J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

important for vascular stabilization in normo-cholesterolemic ischemicmice (Gliem et al., 2012) but harmful in an acute hyperlipidemic setting(Kim et al., 2008). Although patients with metabolic syndrome are oftenseparately analyzed within clinical trials, the results of these studiesmight be considered carefully since these patient populations are oftenheterogeneous and include a complex mixture of interrelated diseaseentities such as hypertension, hyperglycemia, obesity and hypercholes-terolemia (Callahan et al., 2011; Giugliano et al., 2010; Grundy et al.,2005). Moreover, evaluations only based on correlation analysis betweenstroke outcome and blood cholesterol levels might be misleading sinceelevated numbers of circulating leukocyte subsets are predictive for car-diovascular events independent of blood cholesterol levels suggestingthat cellular immunity might be as important as metabolic disturbances(Tuttolomondo et al., 2012).

Our data suggest a bidirectional regulation between peripheral andcerebral responses to brain ischemia which determine final strokeoutcome in hypercholesterolemia. Thus, pre-clinical and clinical trialsshould pay more attention to inflammation associated co-morbiditiesand particularly emphasize the role of granulocytes in this multifactorassociated disease setting.

Conflict of interest

We declare no conflict of interest.

Acknowledgments

We thank Britta Kaltwasser and Beate Karow for technical assis-tance and Janine Gronewold for discussion of statistical analysis.This workwas supported by grants of the German Research Foundation(HE3173/2-1 and HE3173/3-1 to D.M.H.), the Dr. Werner JackstättFoundation (Az. S 134-10.071 to J.H.) and Mercator Research CenterRuhr (An-2011-0081 to J.H.).

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.nbd.2013.10.022.

References

Ayata, C., Shin, H.K., Dileköz, E., Atochin, D.N., Kashiwagi, S., Eikermann-Haerter, K.,Huang, P.L., 2013. Hyperlipidemia disrupts cerebrovascular reflexes and worsensischemic perfusion defect. J. Cereb. Blood FlowMetab.. http://dx.doi.org/10.1038/jcbfm.2013.38 (Epub ahead of print).

Bacigaluppi, M., Pluchino, S., Peruzzotti-Jametti, L., Kilic, E., Kilic, U., Salani, G., Brambilla,E., West, M.J., Comi, G., Martino, G., Hermann, D.M., 2009. Delayed post-ischaemicneuroprotection following systemic neural stem cell transplantation involves multiplemechanisms. Brain 132, 2239–2251.

Beray-Berthat, V., Palmier, B., Plotkine,M., Margaill, I., 2003. Neutrophils do not contributeto infarction, oxidative stress, and no synthase activity in severe brain ischemia. Exp.Neurol. 182, 446–454.

Callahan, A., Amarenco, P., Goldstein, L.B., Sillesen, H., Messig, M., Samsa, G.P., Altfullah, I.,Ledbetter, L.Y., MacLeod, M.J., Scott, R., Hennerici, M., Zivin, J.A., Welch, K.M., SPARCLInvestigators, 2011. Risk of stroke and cardiovascular events after ischemic stroke ortransient ischemic attack in patients with type 2 diabetes or metabolic syndrome:secondary analysis of the Stroke Prevention by Aggressive Reduction in CholesterolLevels (SPARCL) trial. Arch. Neurol. 68, 1245–1251.

Dirnagl, U., Iadecola, C., Moskowitz, M.A., 1999. Pathobiology of ischaemic stroke: an inte-grated view. Trends Neurosci. 22, 391–397.

Dirnagl, U., Klehmet, J., Braun, J.S., Harms, H., Meisel, C., Ziemssen, T., Prass, K., Meisel, A.,2007. Stroke-induced immunodepression: experimental evidence and clinical rele-vance. Stroke 38, 770–773.

Drake, C., Boutin, H., Jones, M.S., Denes, A., McColl, B.W., Selvarajah, J.R., Hulme, S., Georgiou,R.F., Hinz, R., Gerhard, A., Vail, A., Prenant, C., Julyan, P., Maroy, R., Brown, G., Smigova, A.,Herholz, K., Kassiou, M., Crossman, D., Francis, S., Proctor, S.D., Russell, J.C., Hopkins, S.J.,Tyrrell, P.J., Rothwell, N.J., Allan, S.M., 2011. Brain inflammation is induced by co-morbidities and risk factors for stroke. Brain Behav. Immun. 25, 1113–1122.

Drechsler, M., Megens, R.T., van Zandvoort, M., Weber, C., Soehnlein, O., 2010.Hyperlipidemia-triggered neutrophilia promotes early atherosclerosis. Circulation122, 1837–1845.

ElAli, A., Doeppner, T.R., Zechariah, A., Hermann, D.M., 2011. Increased blood–brain barrierpermeability and brain edema after focal cerebral ischemia induced by hyperlipidemia:

role of lipid peroxidation and calpain-1/2, matrix metalloproteinase-2/9, and RhoAoveractivation. Stroke 42, 3238–3244.

Emerich, D.F., Dean, R.L., Bartus, R.T., 2002. The role of leukocytes following cerebralischemia: pathogenic variable or bystander reaction to emerging infarct? Exp.Neurol. 173, 168–181.

Emsley, H.C., Smith, C.J., Gavin, C.M., Georgiou, R.F., Vail, A., Barberan, E.M., Hallenbeck,J.M., del Zoppo, G.J., Rothwell, N.J., Tyrrell, P.J., Hopkins, S.J., 2003. An early andsustained peripheral inflammatory response in acute ischaemic stroke: relationshipswith infection and atherosclerosis. J. Neuroimmunol. 139, 93–101.

Endres, M., Engelhardt, B., Koistinaho, J., Lindvall, O., Meairs, S., Mohr, J.P., Planas, A.,Rothwell, N., Schwanninger, M., Schwab, M.E., Vivien, D., Wieloch, T., Drinagl, U., 2008.Improving outcome after stroke: overcoming the translational roadblock. Cerebrovasc.Dis. 25, 268–278.

Enlimomab Acute Stroke Trial Investigators, 2001. Use of anti-ICAM-1 therapy in ischemicstroke. Results of the Enlimomab Acute Stroke Trial. Neurology 57, 1428–1434.

Fisher, M., Feuerstein, G., Howells, D.W., Hurn, P.D., Kent, T.A., Savitz, S.I., Lo, E.H., for theSTAIR Group, 2009. Update of the stroke therapy academic industry roundtable pre-clinical recommendations. Stroke 40, 2244–2250.

Gelderblom, M., Leypoldt, F., Steinbach, K., Behrens, D., Choe, C.U., Siler, D.A., Arumugam,T.V., Orthey, E., Gerloff, C., Tolosa, E., Magnus, T., 2009. Temporal and spatial dynamicsof cerebral immune cell accumulation in stroke. Stroke 40, 1849–1857.

Giugliano, G., Brevetti, G., Lanero, S., Schiano, V., Laurenzano, E., Chiariello, M., 2010.Leukocyte count in peripheral arterial disease: a simple, reliable, inexpensive ap-proach to cardiovascular risk prediction. Atherosclerosis 210, 288–293.

Gliem, M., Mausberg, A.K., Lee, J.I., Simiantonakis, I., van Rooijen, N., Hartung, H.P., Jander,S., 2012. Macrophages prevent hemorrhagic infarct transformation in murine strokemodels. Ann. Neurol. 71, 743–752.

Grundy, S.M., Cleeman, J.I., Daniels, S.R., Donato, K.A., Eckel, R.H., Franklin, B.A., Gordon,D.J., Krauss, R.M., Savage, P.J., Smith Jr., S.C., Spertus, J.A., Costa, F., American HeartAssociation, National Heart, Lung, and Blood Institute, 2005. Diagnosis and manage-ment of the metabolic syndrome: an American Heart Association/National Heart,Lung, and Blood Institute Scientific Statement. Circulation 112, 2735–2752.

Harms, H., Prass, K., Meisel, C., Klehmet, J., Rogge, W., Drenckhahn, C., Göhler, J., Bereswill, S.,Göbel, U., Wernecke, K.D., Wolf, T., Arnold, G., Halle, E., Volk, H.D., Dirnagl, U., Meisel, A.,2008. Preventive antibacterial therapy in acute ischemic stroke: a randomized con-trolled trial. Plos One. http://dx.doi.org/10.1371/journal.pone.0002158.

Herz, J., Reitmeir, R., Hagen, S.I., Reinboth, B.S., Guo, Z., Zechariah, A., ElAli, A.,Doeppner, T.R., Bacigaluppi, M., Pluchino, S., Kilic, U., Kilic, E., Hermann, D.M., 2012.Intracerebroventricularly delivered VEGF promotes contralesional corticorubral plas-ticity afer focal cerebral ischemia via mechanisms involving anti-inflammatory ac-tions. Neurobiol. Dis. 45, 1077–7085.

Ho, C.S., López, J.A., Vuckovic, S., Pyke, C.M., Hockey, R.L., Hart, D.N., 2001. Surgical andphysical stress increases circulating blood dendritic cell counts independently ofmonocyte counts. Blood 98 (1), 140–145.

Iadecola and Anrather, 2011. The immunology of stroke: frommechanisms to translation.Nat. Med. 17, 796–808.

Jin, R., Yang, G., Li, G., 2010. Inflammatory mechanisms in ischemic stroke: role of inflam-matory cells. J. Leukoc. Biol. 87, 779–789.

Jin, F., Hagemann, N., Brockmeier, U., Schäfer, S.T., Zechariah, A., Hermann, D.M., 2013. LDLattenuates VEGF-induced angiogenesis via mechanisms involving VEGFR2 internaliza-tion and degradation following endosome-trans-Golgi network trafficking. Angiogene-sis. http://dx.doi.org/10.1093/cvr/cvt209.

Kim, E., Tolhurst, A.T., Qin, L.Y., Chen, X.Y., Febbraio, M., Cho, S., 2008. CD36/fatty acidtranslocase, an inflammatory mediator, is involved in hyperlipidemia-induced exac-erbation in ischemic brain injury. J. Neurosci. 28, 4661–4670.

Lakhan, S.E., Kirchgessner, A., Hofer, M., 2009. Inflammatory mechanisms in ischemicstroke: therapeutic approaches. J. Transl. Med. 7, 97.

Langdon, K.D., Maclellan, C.L., Corbett, D., 2010. Prolonged, 24-h delayed peripheral in-flammation increases short- and long-term functional impairment and histopathologi-cal damage after focal ischemia in the rat. J. Cereb. Blood Flow Metab. 30, 1450–1459.

Macrez, R., Ali, C., Toutirais, O., LeMauff, B., Defer, G., Dirnagl, U., Vivien, D., 2011. Strokeand the immune system: from pathophysiology to new therapeutic strategies. LancetNeurol. 10, 471–480.

McColl, B.W., Rothwell, N.J., Allan, S.M., 2008. Systemic inflammation alters the kinetic ofcerebrovascular tight junction disruption after experimental stroke in mice.J. Neurosci. 28, 9451–9462.

McColl, B.W., Allan, S.M., Rothwell, N.J., 2009. Systemic infection, inflammation and acuteischemic stroke. Neuroscience 158, 1049–1061.

Meisel, C., Schwab, J.M., Prass, K., Meisel, A., Dirnagl, U., 2005. Central nervous systeminjury-induced immune deficiency syndrome. Nat. Rev. Neurosci. 6, 775–786.

Murphy, A.J., Akhtari, M., Tolani, S., Pagler, T., Bijl, N., Kuo, C.L., Wang, M., Sanson, M.,Abramowicz, S., Welch, C., Bochem, A.E., Kuivenhoven, J.A., Yvan-Charvet, L., Tall, A.R.,2011. ApoE regulates hematopoietic stem cell proliferation, monocytosis andmonocyteaccumulation in atherosclerotic lesions in mice. J. Clin. Invest. 121, 4138–4149.

Offner, H., Subramanian, S., Parker, S.M., Afentoulis, M.E., Vandenbark, A.A., Hurd, P.D.,2006. Experimental stroke induces massive, rapid activation of the peripheral im-mune system. J. Cereb. Blood Flow Metab. 26, 654–665.

Prass, K., Meisel, C., Höflich, C., Braun, J., Halle, E., Wolf, T., Ruscher, K., Victorov, I.V., Priller,J., Dirnagl, U., Volk, H.D., Meisel, A., 2003. Stroke-induced immunodeficiency promotesspontaneous bacterial infections and is mediated by sympathetic activation reversalby poststroke T helper cell type 1-like immunostimulation. J. Exp. Med. 198, 725–736.

Röther, J., Alberts, M.J., Touzé, E., Mas, J.L., Hill, M.D., Michel, P., Bhatt, D.L., Aichner, F.T.,Goto, S., Matsumoto, M., Ohman, E.M., Okada, Y., Uchiyama, S., D'Agostino, R.,Hirsch, A.T., Wilson, P.W., Steg, P.G., REACH Registry Investigators, 2008. Risk factorprofile and management of cerebrovascular patients in the REACH Registry.Cerebrovasc. Dis. 25, 366–374.

468 J. Herz et al. / Neurobiology of Disease 62 (2014) 456–468

Smith, C.J., Emsley, H.C., Gavin, C.M., Georgiou, R.F., Vail, A., Barberan, E.M., del Zoppo, G.J.,Hallenbeck, J.M., Rothwell, N.J., Hopkins, S.J., Tyrell, P.J., 2004. Peak plasmainterleukin-6 and other peripheral markers of inflammation in the first week of isch-aemic stroke correlate with brain infarct volume, stroke severity and long-term out-come. BMC Neurol. 4, 2.

Smith, E., Prasad, K.M., Burcher, M., Dobrian, A., Kolls, J.K., Ley, K., Galkina, E., 2010. Blockadeof interleukin-17A results in reduced atherosclerosis in apolipoprotein E-deficient mice.Circulation 121, 1746–1755.

Tuttolomondo, A., Di Raimondo, D., Di Sciacca, R., Pecoraro, R., Arnao, V., Buttà, C., Licata,G., Pinto, A., 2012. Arterial stiffness and ischemic stroke in subjects with and withoutmetabolic syndrome. Atherosclerosis 225, 216–219.

Vijayalekshmi, K.S., Leelamma, S., 1991. Mechanisms of hypercholesterolemia producedby some antibiotics. Indian J. Clin. Biochem. 6, 31–38.

Wu, H., Gower, R.M., Wang, H., Perrard, X.Y., Ma, R., Bullard, D.C., Burns, A.R., Paul, A.,Smith, C.W., Simon, S.I., Ballantyne, C.M., 2009. Functional role of CD11c+mono-cytes in atherogenesis associated with hypercholesterolemia. Circulation 119,2708–2717.

Zechariah, A., Jin, F., Doeppner, T.R., Elali, A., Helfrich, I., Mies, G., Hermnann, D.M.,2013. Hyperlipidemia attenuates vascular endothelial growth factor inducedangiogenesis, impairs cerebral blood flow, and disturbs stroke recovery viadecreased pericyte coverage of brain endothelial cells. Arterioscler. Thromb.Vasc. Biol. 33, 1561–1567.