Functional variability in butyrylcholinesterase activity regulates intrathecal cytokine and astroglial biomarker profiles in patients with Alzheimer's disease

at SciVerse ScienceDirect

Neurobiology of Aging xxx (2013) 1e17

Contents lists available

Neurobiology of Aging

journal homepage: www.elsevier .com/locate/neuaging

Functional variability in butyrylcholinesterase activity regulates intrathecalcytokine and astroglial biomarker profiles in patients with Alzheimer’s disease

Taher Darreh-Shori a,*, Swetha Vijayaraghavan a, Shahin Aeinehband b, Fredrik Piehl b,Rickard P.F. Lindblomb, Bo Nilsson c, Kristina N. Ekdahl c,d, Bengt Långströme,f,g, Ove Almkvist a,h,Agneta Nordberg a,i

aDivision of Alzheimer Neurobiology Center, Department of Neurobiology, Care Sciences and Society, Karolinska Institutet, Stockholm, SwedenbDepartment of Clinical Neuroscience, Unit for Neuroimmunology, Karolinska University Hospital Solna, Stockholm, SwedencDepartment of Immunology, Genetics and Pathology, Division of Clinical Immunology, Uppsala University, Uppsala, Swedend Linnæus Center of Biomaterials Chemistry, Linnæus University, Kalmar, SwedeneDepartment of Organic Chemistry and Biochemistry, Uppsala University, Uppsala, SwedenfDivision of Experimental Medicine, Neuropsycho-pharmacology Unit, Centre for Pharmacology and Therapeutics, Imperial College, London, UKg PET and Cyclotron Unit, Department of Nuclear Medicine, Odense University Hospital, University of Southern Denmark, Odense, DenmarkhDepartment of Psychology, Stockholm University, Stockholm, SwedeniDepartment of Geriatric Medicine, Karolinska University Hospital Huddinge, Stockholm, Sweden

a r t i c l e i n f o

Article history:Received 12 October 2012Received in revised form 2 April 2013Accepted 28 April 2013

Keywords:ButyrylcholinesteraseAstrocytesMicrogliaCholinoceptive cellsAlzheimer’s diseaseBCHE genotypeGlial fibrillary acidic proteinS100BComplement systemCholinergic anti-inflammatory pathway

* Corresponding author at: Division of Alzheimer Nment of Neurobiology, Care Sciences and Society, Karol4th, 141 86 Stockholm, Sweden. Tel.: þ46 8 585 863 1

E-mail address: [email protected] (T. Darre

0197-4580/$ e see front matter � 2013 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.neurobiolaging.2013.04.027

a b s t r a c t

Butyrylcholinesterase (BuChE) activity is associated with activated astrocytes in Alzheimer’s disease brain.The BuChE-K variant exhibits 30%e60% reduced acetylcholine (ACh) hydrolyzing capacity. Considering theincreasing evidence of an immuneeregulatory role of ACh,we investigated if genetic heterogeneity in BuChEaffects cerebrospinal fluid (CSF) biomarkers of inflammation and cholinoceptive glial function. Alzheimer’sdisease patients (n ¼ 179) were BCHEeK-genotyped. Proteomic and enzymatic analyses were performed onCSF and/or plasma. BuChE genotype was linked with differential CSF levels of glial fibrillary acidic protein,S100B, interleukin-1b, and tumor necrosis factor (TNF)-a. BCHE-K noncarriers displayed 100%e150% higherglial fibrillary acidic protein and 64%e110% higher S100B than BCHE-K carriers, who, in contrast, had 40%e80% higher interleukin-1b and 21%e27% higher TNF-a compared with noncarriers. A high level of CSFBuChE enzymatic phenotype also significantly correlated with higher CSF levels of astroglial markers andseveral factors of the innate complement system, but lower levels of proinflammatory cytokines. Theseindividuals also displayed beneficial paraclinical and clinical findings, such as high cerebral glucose utili-zation, low b-amyloid load, and less severe progression of clinical symptoms. In vitro analysis on humanastrocytes confirmed the involvement of a regulated BuChE status in the astroglial responses to TNF-a andACh. Histochemical analysis in a rat model of nerve injury-induced neuroinflammation, showed focalassembly of astroglial cells in proximity of BuChE-immunolabeled sites. In conclusion, these results suggestthat BuChE enzymatic activity plays an important role in regulating intrinsic inflammation and activity ofcholinoceptive glial cells and that this might be of clinical relevance. The dissociation between astroglialmarkers and inflammatory cytokines indicates that a proper activation and maintenance of astroglialfunction is a beneficial response, rather than a disease-driving mechanism. Further studies are needed toexplore the therapeutic potential of manipulating BuChE activity or astroglial functional status.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Over 70 years ago the enzyme butyrylcholinesterase (BuChE)was discovered (Stedman and Easson, 1932), yet as of today, its

eurobiology Center, Depart-inska Institutet, Novum Floor2; fax: þ46 8 585 854 70.h-Shori).

ll rights reserved.

biological role remains unclear, particularly the essence of BuChEfunction in the central nervous system (CNS) is largely unknown.However, the distribution pattern and observations in acetylcho-linesterase (AChE) knockout mice point at the involvement ofBuChE in neural function such as coregulation of cholinergicneurotransmission (Darvesh et al., 2003; Mesulam et al., 2002).

A substantial number of genetic variants of BCHE have beenidentified (Darreh-Shori et al., 2012; Darvesh et al., 2003). TheBCHE-K variant is the most common functional point mutation of

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e172

this enzyme (Darreh-Shori et al., 2012; Darvesh et al., 2003). SerumBuChE activity of individuals with the BCHE-K variant is 30%e33%lower compared with subjects with the wild type variant (Darveshet al., 2003). Pharmacogenetic studies suggest that the K variantmight be involved in the pathogenesis of Alzheimer’s disease (AD),particularly in relation to the major known genetic risk factor of AD(i.e., the ε4 allele of apolipoprotein E [ApoE]; Darreh-Shori et al.,2012; Darvesh et al., 2003; Lane et al., 2008). In patients with AD,the difference in the plasma BuChE activity is more pronounced;thus, K heterozygotes show 30% and homozygotes 50% lower BuChEactivity than the wild type carriers (Darreh-Shori et al., 2012). Theexpression of BuChE is also substantially increased in the hippo-campus and temporal cortex of patients with AD, whereasexpression of AChE is reduced (Perry et al., 1978). A recent post-mortem study shows a strong correlation between BuChE activityand L-deprenyl as a measure of activated astroglial cells in the ADbrain (Kadir et al., 2011).

The major source of BuChE in the CNS is attributed to non-excitable cells such as astrocytes and microglia, which also expressvarious nicotinic acetylcholine (ACh) receptors, in particular, the a7subtype (Carnevale et al., 2007; Shytle et al., 2004), indicating thatBuChE might play a regulatory role in the functional status ofcholinoceptive nonexcitable cells, such as astrocytes, oligodendro-cytes, and microglial cells, via its ACh hydrolyzing activity (Darreh-Shori et al., 2011a). Astroglial cells serve multiple functions in theCNS, including maintenance of the extracellular microenvironmentand regulation of neurotransmitter levels (Maragakis and Rothstein,2006).

Deprenyl binding to monoamineoxidase-B is putatively regar-ded as a promising positron emission tomography (PET) markerfor in vivo assessment of activated astroglial cells in the brain(Carter et al., 2012). Recent in vivo studies with 11C-deuterium-L-deprenyl-PET suggest that presence of activated astrocytes andmicroglia at early and at late stages of AD (Carter et al., 2012).

Two commonly used cerebrospinal fluid (CSF) glial markers areglial fibrillary acidic protein (GFAP) and S100B. GFAP is the subunitprotein of intermediate filaments in astroglial cells. S100B isregarded as an astrocytic cytokine and might mediate interactionsbetween glial cells or between glial cells and neurons (Mrak andGriffin, 2001). However, GFAP is more specific for astrocytes thanS100B, because oligodendrocytes, ependymal cells, choroid plexusepithelium, vascular endothelial cells, and some neurons alsoexpress S100B in the brain (Steiner et al., 2007).

Hypertrophy of astrocytes accompanied by elevated expressionof GFAP is observed in postmortem AD brain with increasing Braakstage (Beach and McGeer, 1988; Kashon et al., 2004; Wharton et al.,2009), and at a given Braak stage, Apo ε4 carriers show slightlyhigher GFAP levels than noncarriers (Wharton et al., 2009). CSFlevels of GFAP also show an age-dependent increase (Rosengrenet al., 1994) and an inverse association is reported between cogni-tive function and GFAP levels in the occipital, parietal, and temporallobes (Kashon et al., 2004).

Involvement of the innate immune complement (C) system inAD pathology is well established (Veerhuis, 2011) and CSF levels ofC3 have recently been reported to correlate with cognitive declinein patients with AD (Wang et al., 2011). However, any relationbetween BuChE and CSF complement levels has been unknown.

The aim of the current study was to study the relation betweena genetic variation in BuChE activity and a number of CSFbiomarkers reflecting astroglial activation and innate inflammation.To this end, 179 AD patients were genotyped with regard to BCHE-Kand ApoE ε4 allele variants, and with proteomic and enzymaticanalyses of CSF protein levels of ApoE protein, S100B, and GFAP,interleukin (IL)-1b, and tumor necrosis factor (TNF)-a, and activityand protein levels of BuChE and AChE in CSF and plasma. In

addition, several complement components were measured ina subcohort of the AD patients. Then, we challenged some of thefindings in vitro using human primary astrocytes.

The finding of a strong association between the K variant alleleor ACh-hydrolyzing activity of BuChE and the CSF levels of the glialmarkers suggests that the BuChE genotype and phenotype exertsregulatory effects on the activity of S100B- and/or GFAP-expressingcells and ongoing pathology or symptoms in AD, possibly viaregulation of extracellular ACh levels.

2. Methods

2.1. Patients and CSF and plasma samples

The current study population and the selection criteria are thesame as reported recently (Darreh-Shori et al., 2012). Briefly, the ADpatients were selected based on the following criteria: for geno-typing analysis, all the patients had to have a clinical diagnosis ofprobable AD, from which at least 1 Mini Mental State Examination(MMSE) assessment was available before treatment with anycholinesterase inhibitors (ChEIs). For the enzymatic and proteomicanalysis, in addition, at least 1 CSF or plasma sample (or both) hadto be available before any ChEIs therapy. In total, 179 patients ful-filled these criteria. All CSF and plasma samples used in the currentstudy were taken from baseline (i.e., before the ChEIs therapies).

This study and the primary clinical studies were approved by theEthics Committee of Karolinska University Hospital Huddinge, andthe Faculty of Medicine and Radiation Hazard Ethics Committee ofUppsala University Hospital, Uppsala, Sweden. This study wasconducted according to the Declaration of Helsinki and subsequentrevisions. Informed consent was obtained from each patient or theresponsible caregivers.

2.2. Neuropsychological assessments

Global cognitive function was assessed using the standardMMSE test and was available from all of the patients included in thestudy (Darreh-Shori et al., 2012).

A battery of neuropsychological tests data addressing differentcognitive domains (episodic memory, attention/executive ability,and visuospatial ability; Darreh-Shori et al., 2011a) were availablefrom a subgroup of the patients (n ¼ 66). To reduce the number ofstatistical analyses, these data were Z-transformed and the overallcomposite Z-scores of the 3 cognitive domains were calculated asdescribed in Darreh-Shori et al. (2011a).

2.3. PET assessments using fluorodeoxyglucose and PittsburghCompound-B

Part of the analysis is based on data acquired in CSF of thepatients with probable AD, who had undergone the (18)F-fluorodeoxyglucose (FDG)-PET (n ¼ 50) and/or PittsburghCompound-B (PIB)-PET (n ¼ 29) assessment as described inDarreh-Shori et al., (2011a).

2.4. Single nucleotide polymorphism analysis

The BCHE rs1803274 single-nucleotide polymorphism (SNP)(CATATTTTACAGG AAATATTGATGAA[A/G]CAGAATGGGA GTGGA-AAGCAGGATT) was used, which corresponds to the so-called Kvariant of BuChE. The SNP was investigated in genomic DNA usingpolymerase chain reaction (PCR) using a TaqMan SNP genotypingassay kit according to the manufacturer’s instruction and ona StepOne Plus thermal cycler (Applied Biosystems), as described inDarreh-Shori et al. (2012).

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 3

2.5. BuChE and AChE measurements

The activity and protein levels of BuChE and AChE in CSF and thehuman astrocyte culture medium was measured as described inDarreh-Shori et al. (2006). Briefly, the activities of BuChE and AChEwere measured using a modified version of Ellman’s colorimetricassay, using butyrylthiocholine iodide (5.0 mM final concentration;Sigma, St. Louis, MO) and acetylthiocholine iodide (0.5 mM finalconcentration; Sigma) as a substrate in the presence of the selectiveAChE inhibitor BW280C51 (1.0 mM final concentration; Sigma) orthe selective BuChE inhibitor, ethopropazine (0.1 mM finalconcentration; Sigma) as described in Darreh-Shori et al. (2008).

BuChE protein levels were measured using a sandwich enzyme-linked immunosorbent assay (ELISA), essentially as described inDarreh-Shori et al. (2008). The protein level of CSF AChE-R andAChE-S variants were measured using an ELISA-like methodessentially as described in Darreh-Shori et al. (2008). The MAB337-Ab (mouse monoclonal antibody; Chemicon Int) was used forassessing the protein level of the CSF AChE-R variant in CSF. TheMA03-42 Ab (HR2 mouse monoclonal antibody; Affinity Bio-Reagents) was used for measuring the protein level of the AChE-Svariant in CSF or in the culture medium, as described in Darreh-Shori et al. (2008). The overall dilution factor for CSF or plasmawas 4. The dilution factor for culture medium was 2.

2.6. CSF S100b and GFAP protein assay

In advance, all CSF samples were diluted 5 times with: TBS (trisbuffer saline [containing 10 mM Tris-HCl, pH7.4 and 0.9% NaCl]);0.05% Triton X-100 (TX); 6 mM Ethylenediaminetetraacetic acid(EDTA) 0.1%bovine serumalbumin (BSA); and 0.01% sodiumazide; allfrom Sigma (TBS-TX-BSA) and kept frozen at �70 �C until the assay.

The capturing antibody for the GFAP ELISAwere the monoclonalmouse antibody, SMI26 (Covance), which was diluted 1/3000 ina coating buffer (sodium carbonate buffer, pH 9.6 containing 0.02%wt/vol sodium azide). The detecting and the secondary antibodieswere the polyclonal rabbit antibody, Z0334 (from Dako, Glostrup,Denmark, diluted 1/3000 in TBS-TX-BSA), and the alkaline phos-phatase (AP)-conjugated bovine anti-rabbit immunoglobulin G (sc-2376 from Santa Cruz Biotech, diluted 1/3000 in TBS-TX-BSA),respectively. Purified normal human brain GFAP protein(A86823H, from Biodesign International) was used as the standardprotein, which was diluted in the TBS-TX-BSA buffer ranging from5.0 to 0.04 ng/mL (by 2 times serial dilution).

The capturing, detecting, and the secondary antibodies for theS100B ELISA were the monoclonal mouse antibody, S2532 (fromSigma, diluted 1/3000 diluted in the coating buffer), the polyclonalrabbit anti-S100 antibody, Z0311 (fromDako, diluted 1/3000 in TBS-TX-BSA), and the AP-conjugated Swine anti-rabbit immunoglobulinG (D0306, from Dako, diluted 1/3000 in TBS-TX-BSA), respectively.The recombinant full-length protein, corresponding to amino acids1e92 of human S100b (ab54050, from Abcam) was used as thestandard protein, which was diluted in the TBS-TX-BSA bufferranging from 25.0 to 0.20 ng/mL (by 2 times serial dilution).

Briefly, Nuncmaxisorb ELISA plates were coated overnight at 4 �Cby adding 100 mL per well of the appropriate capturing antibody. Theplates were thenwashed 3 times for 5 minutes with TBS (300 mL perwell), and incubated for 30e60 minutes at room temperature (RT;22.5 �C) with 250 mL per well of blocking solution (4% BSA wt/vol, inTBS). The plates were washed 3 times for 5 minutes with 300 mL perwell of TBS-TW [TBS containing 0.05% Tween20 (TW)] and incubatedwith 100 mL perwell of the standards and the diluted CSF samples (allin triplicates) at 4 �C overnight. After washing as before, the plateswere incubated for 2 hours at RT with 100 mL per well of the corre-sponding detecting antibodies. The plateswerewashed as before and

incubated at RT for 2 hours with working solution of the appropriateAP-conjugated secondary antibody. The plates were washed 4 timeswith TBS-TW, oncewithdiethanolaminebuffer (1.0M, pH9.8; Sigma)and then incubated with 150 mL per well of the substrate (di-Sodium4-nitrophenyl phosphate, 10 mM; #73724 from Sigma) in thediethanolamine buffer at RT. The end point reaction was monitoredusing a microplate spectrophotometer reader (SpectraMax 250;Molecular Devices Corporation) at a 405-nmwavelength at RT, usingSOFTmax PRO software (version 2.6.1 for PC; Molecular DevicesCorporation).

2.7. Measurements of IL-1b, IL-6, and TNF-a levels in CSF

CSF and plasma levels of IL-1b, IL-6, and TNF-a were measuredusing commercial high-sensitivity ELISA kits (Quantikine HS fromR&D Systems, Abingdon, UK), according to the manufacturer’sinstructions. Because TNF-a analyses required a substantial amountof CSF, we had to limit analyses to a subgroup of the patients (n ¼49), from whom in vivo brain data assessed with PET were alsoavailable, as is described in the following text.

2.8. CSF complement factor measurements

In a recent study using an experimental rat model of CNS injury,we found that the expression of complement factor C3 is linked tothe expression of BuChE (unpublished data).

We hence investigated the interrelationship of the complementfactors in CSF with phenotypic display of BuChE, which might bemore relevant than BCHE genotype (Darreh-Shori et al., 2012), ina small subgroup of the AD patients (n ¼ 10) and 10 patients withother neurological disorders (OND) as control individuals. TheADCSFsampleswere selected for this analysis from the ADpatientswho hadin vivo FDG- and PIB-PET assessment, and were noncarriers of theBCHE-K allele. This is because BCHE genotype was not available forthe OND individuals, and they were most likely noncarriers of the Kallele because of the expected high frequency of the wild type allelein general Caucasian populations (Darreh-Shori et al., 2012).

The following components of the complement system: C3, C3a,C1q, C4, and H, were measured using sandwich ELISA in these CSFsamples, essentially as described in Henningsson et al. (2007).

2.9. Immunohistochemistry

Five adult Dark Agouti rats were exposed to unilateral ventralroot avulsion (VRA) as previously described (Strom et al., 2012). Fivedays postoperatively, the animals were euthanized with CO2 andperfused via the ascending aorta with 60 mL of room temperaturephosphate-buffered saline (PBS) containing heparin (LEO PharmaAB) (10 IU/mL), followed by 150 mL of 4% paraformaldehyde, thespinal cords were then dissected out and kept in 4% para-formaldehyde overnight and then transferred to 10% sucrose in PBSbefore mounting and sectioning of the L3eL5 segments. The spinalcord sections were serially cut (14-mm) on a cryostat (Leica Micro-systems) at the level of the L4 segment. Sections were thawed ontoSuperfrost plus microscope slides (Menzel-Gläser) and storedat �20 �C until further processing for immunohistochemistry.Sections were postfixed for 30 minutes in 4% formaldehyde at RTbefore incubation with the following antibodies; GFAP (rabbit anti-mouse, 1:200; Dako), Iba1 (rabbit anti-rat, 1:200; Wako), BuChE(mouse anti-rat 1:200; the antibody was developed and kindlyprovided by Dr Anna Hrabovska [Hrabovska et al., 2010]), and thenrinsed in PBS, incubated for 60 minutes with appropriateflourophore-conjugated secondary antibody (Alexa Fluor 488donkey-anti rabbit, 1:150 and Alexa Fluor 594, goat anti-mouse,1:300; both from Invitrogen), diluted in PBS and 0.3% Triton

Table 1BCHE-K allele-dependent demographic and biochemical characteristics in plasma and CSF of patients with AD

Variable BCHE genotype

K/K K/Wt Wt/Wt Overall

Demographic characteristicn 13 59 107 179Age 74 � 2 73 � 1 72 � 1 72 � 1MMSE score 23 � 1.3 24 � 0.6 24 � 0.4 24 � 0.3Sex (F/M) 4/8 34/25 59/48 97/81Apo E (ε4�/�/ε4þ/�/ε4þ/þ) 5/5/3 17/25/17 29/58/17e 51/88/37

Glial and inflammatory markersn 8 40 74 122CSF GFAP protein, ng/mL 0.565 � 0.184 1.336 � 0.113b 1.658 � 0.131b,c 1.494 � 0.092CSF S100b protein, ng/mL 46.62 � 9.30 82.46 � 7.14b 106.32 � 5.66b,d 95.31 � 4.44Plasma IL-1b protein, pg/mL 2.25 � 0.77 3.09 � 0.49 2.74 � 0.28 2.81 � 0.24CSF IL-1b protein, pg/mL 2.31 � 0.40 1.78 � 0.18 (37) 1.27 � 0.11 (68)b,d 1.51 � 0.10Plasma IL-6 protein, pg/mL 39.12 � 28.95 39.44 � 14.77 24.95 � 5.65 0.26 � 5.98CSF IL-6 protein, pg/mL 5.04 � 1.91 (7) 4.81 � 0.77 (37) 4.19 � 0.53 (68) 4.45 � 0.42CSF TNF-a, pg/mLa 0.498 � 0.115 (2) 0.716 � 0.047 (12) 0.558 � 0.025 (35)d 0.595 � 0.023 (49)

The digits in the parentheses indicate the number of available samples that was assessed.Key: AD, Alzheimer’s disease; Apo, apolipoprotein; CSF, cerebrospinal fluid; F, female; GFAP, glial fibrillary acidic protein; IL, interleukin; M, male; MMSE, Mini Mental StateExamination; TNF, tumor necrosis factor; Wt, wild type.

a CSF samples were only available for 2 subjects with the K/K genotype for measurement of TNF-a.b p < 0.01 compared with K/K genotype.c p < 0.05 compared with Wt/K genotype.d p < 0.01 compared with Wt/K genotype.e Apo ε4 genotype for 3 patients was not available.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e174

X-100, then rinsed in PBS and mounted in PBS-glycerol (1:3).Sections processed for immunohistochemistry were examined andphotographed using a Leica DM RBE microscope system and pro-cessed using Adobe Photoshop CS5. The animal experiments wereapproved by the local ethical committee for animal experimenta-tion (Stockholms Norra Djurförsöksetiska Nämnd).

2.10. Human astrocyte cultures

Primary human astrocytes were obtained from ScienCellResearch Laboratories (Cat no. 1800). Cells were cultured in astro-cyte medium (Sciencell cat no. 1801) in poly-L-lysine-coated T-175flasks, and handled according to manufacturer’s recommendations.Confluent cells were then disattached, seeded in a 48-well plate(3 � 105 cells per well) and left in medium overnight to recover.After this, the cells were stimulated with ACh or TNF-a alone, AChwith TNF-a, or left unstimulated as control samples. ACh concen-tration varied between 1 mM to 100 mM and TNF-a concentrationwas kept constant at 20 ng/mL.

The astrocytes were stimulated for 24 hours, after which thecells were lysed for RNA extraction and subsequent reverse tran-scription PCR expressional analysis. The cultured media were savedfor measuring AChE and BuChE activity and protein levels. Allconditions were in triplicate. The pH of the mediumwas lowered to7.1 to increase the stability of ACh. Recombinant human TNF-a wasobtained from R&D systems (Cat no. 210-TA-010). Acetylcholinechloride was obtained from Sigma (Cat no. A2661).

2.11. RNA preparation and reverse transcription PCR

Cells were lysed in RLT buffer (Qiagen) for total RNA preparation.Total RNA was extracted, purified, and on column DNase I-treatedusing an RNeasy Mini kit (Qiagen) and RNase-Free DNase Set(Qiagen), according to the manufacturers’ protocols. The RNA wassynthesized to cDNA by incubation with 5x iScript reaction mix(Bio-Rad) for 5 minutes at 25 �C, 30 minutes at 42 �C, and 5minutesat 85 �C. All steps were performed under RNase-free conditions.Real-time PCR was conducted using a 3-step PCR protocol and Bio-Rad CFX manager software (version 2.1). Samples were run intriplicate. All primers and probes were designed with primer-blast

(www.ncbi.nlm.nih.gov/tools/primer-blast), and checked for spec-ificity using melt curve analysis. The housekeeping gene glyceral-dehyde 3-phosphate dehydrogenase was used to normalize themRNA expression levels of the studied transcripts. Normalizedexpression levels were calculated using Bio-Rad CFX manager.

The primer sequences were as follows: GAPDH: forward: 50-AGGGCTGCT TTTAACTCTGGTAAA-30, reverse: 50-CATATTGGAACA-TGTAAACCATGTAGT TG-30; and BCHE: forward: 50-ATGGGCT-CTTCTCCTCCTAC-30, reverse: 50-GTGGAGTCTTTCACGAGGAC-30; andthe complement factor C3: forward: 50-CACAGCCAAAGATAA-GAACC-30, reverse: 50-GATACGGTGGGTGATCTTG-30.

2.12. Statistical analysis

Data are given as mean and standard error of the mean. Factorialanalysis of variance was used for the effect of factors such as Apo ε4and BCHE-K genotype, sex, etc. A significant analysis of varianceresult (p < 0.05) was followed by Bonferroni-Dunn post hoc anal-ysis to test the significance of results between groups. The 2-tailedZ-test and the nonparametric Spearman rank correlations wereused for the correlation analysis, which was then visualizedgraphically using a simple regression plot.

3. Results

3.1. Patients

An overview of the demographic data for the overall studypopulation and the absolute values of the assessed CSF and plasmabiomarkers is summarized in Table 1. For simplicity, furtherpresentations of relative changes in these biomarkers among thegenotype groups are presented as percentages of the mean value ofthe BCHE-K homozygotes (BCHE-Kþ/þ). It should be noted that someanalyses were not conducted in the entire cohort.

3.2. CSF levels of GFAP and S100B are strongly dependent on thegenotype of BuChE

CSF GFAP levels differed highly between BCHE-K carriers andnoncarriers (108%e150%; F ¼ 6.5; p < 0.002). Post hoc analysis

35

#

***

K++ K+- K-- K++ K+- K-- K++ K+- K-- K+/* K--

BCHE -K genotype

837

68

−150

−50

50

150

−100

0

100

200

300

400

−200

0

200

400

600

840

74

−80

−40

0

40

80

CSF

leve

ls (%

K++)

A GFAP B S100B C IL-1β D TNF-α

14

** # #

# #

*

#

***

**

−100

0

100

200

300

CSF

S10

0B (n

g/m

L)

0 2 4 6 8 10 12 14

CSF BuChE activity (nmol/min/mL)

−1

0

1

2

3

4

5

6

CSF

GFA

P (n

g/m

L)

0 2 4 6 8 10 12 14

CSF BuChE activity (nmol/min/mL)

0

100

200

300

400

CSF

S10

0B (n

g/m

L)

0 1 2 3 4

CSF GFAP (ng/mL )

r = 0.54p < 0.0001n = 122

r = 0.35p < 0.0001n = 122

r = 0.45p < 0.0001n = 122

E F G

Fig. 1. Genotypic and phenotypic displays of butyrylcholinesterase (BuChE) on astroglial markers imply this enzyme as a potential regulator of astroglial function in the centralnervous system. The cerebrospinal fluid (CSF) levels of glial fibrillary acidic protein (GFAP) (A) and S100B (an astrocytic cytokine) (B) shows robust BCHE-K allele dosage-dependency.Similarly, the CSF interleukin (IL)-1b (C) and tumor necrosis factor (TNF)-a (D) levels show a strong K dosage-dependency but in the opposite direction. The CSF level of GFAPcorrelates well with the CSF S100B (E). The phenotypic display of the BuChE positively correlates with the CSF levels of GFAP (F) and S100B (G). Note that CSF was only available for 2subjects with the K/K genotype for the measurement of TNF-a , thereby the comparison was made between all K carriers (Kþþ and Kþ�, denoted Kþ/*) and noncarriers (K��). TheCSF GFAP and S100B was assessed using sandwich enzyme-linked immunosorbent assay. * p < 0.05, ** p < 0.01 and *** p < 0.001 compared with the K– group. # p < 0.05 and## p < 0.01 compared with the Kþ� group. Digits indicate the number of subjects in each group. Data are given as percentages of mean values for the Kþþ group and standard errorof the mean in (AeD). Kþþ and Kþ�, homozygotes and heterozygotes of the BCHE-K allele, respectively. Abbreviation: K��, noncarriers of the K allele.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 5

indicated that CSF GFAP was 150 � 15% higher in BCHE-K�/�

(n ¼ 73) than in BCHE-Kþ/þ AD subjects (n ¼ 8; p < 0.0009; Fig. 1A),and 42 � 15% higher than in BCHE-Kþ/� heterozygotes (n ¼ 40; p <

0.048; Fig. 1A). The difference in CSF GFAP levels between BCHE-Kþ/þ and BCHE-Kþ/� was also significant (108 � 17%; p < 0.0086).These findings indicate that CSF GFAP displays BCHE-K dosage-dependent regulation. In contrast, CSF GFAP levels did not differbetween Apo ε4 carriers and noncarriers (p> 0.93) or betweenmenand women (p > 0.46).

Similar to GFAP, CSF S100B levels also showed a strong K gen-eedosage dependency (F ¼ 7.4; p < 0.0009). Post hoc analysisshowed that CSF S100B was 111 � 11% higher in BCHE-K�/�

compared with BCHE-Kþ/þ AD patients (p < 0.0015; Fig. 1B). Thecorresponding difference between K noncarriers and the Kheterozygotes was 47 � 11% (p < 0.0099; Fig. 1B). The differencesbetween heterozygotes and BCHE-Kþ/þ individuals was alsosignificant (64 � 14%; p < 0.031; Fig. 1B).

As for GFAP, CSF S100B levels did not differ between Apo 34carriers and noncarriers (p > 0.75) or between men and women(p > 0.96).

3.3. CSF levels of GFAP and S100B are strongly dependent on thephenotypic display of BuChE

In a previous study, we have shown that the K variant protein isassociated with 21%e49% reduced BuChE activity in CSF of ADpatients in the presence of the Apo ε4 allele (Darreh-Shori et al.,2012). To investigate if this reduced activity of the BuChE-Kvariant protein is responsible for the differential level of GFAPand S100B among the BCHE-K genotype groups, we performeda correlative analysis. The BuChE activity in CSF correlated posi-tively with the GFAP levels (r ¼ 0.33; n ¼ 121; p < 0.0002; Fig. 1F)and the S100B levels (r ¼ 0.44; p < 0.0001; Fig. 1G). In addition, CSF

CSF

IL-6

leve

l(L

n(p

g/m

LC

SF))

−1

0

1

2

3

4

5

0 1 2 3 4

Br = −0.31p < 0.0007n = 112

CSF GFAP levels (ng/mL CSF)

−2

−1

0

1

2

3

0 1 2 3 4

CSF

IL-1

βle

vel

(Ln

(pg/

mL

CSF

))

Ar = −0.46p < 0.0001n = 113

CSF S100B levels (ng/mL CSF)

−2

−1

0

1

2

3

0 50 100 150 200 250

Cr = −0.32p < 0.0005n = 113

−1

0

1

2

3

4

5

0 50 100 150 200 250

Dr = −0.15p < 0.2n = 112

Fig. 2. The cerebrospinal fluid (CSF) levels of astroglial markers show an unexpected reciprocal association with the levels of cytokine in the CSF of the Alzheimer’s disease (AD)patients. The astroglial maker, glial fibrillary acidic protein (GFAP) in the CSF of AD patients correlates negatively with the levels of the proinflammatory cytokines, interleukin (IL)-1b(A) and IL-6 (B) in the CSF. The CSF levels of the astrocytic cytokine, S100B, also shows a similar pattern with CSF levels of IL-1b (C) and IL-6 (D). These findings might thereforeindicate that the brains of the AD patients, who were relatively successful in regulating astroglial function, produce fewer proinflammatory cytokines. Abbreviation: Ln, naturallogarithm.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e176

GFAP levels were also positively correlated to S100B levels (r¼ 0.54;n ¼ 121; p < 0.0001; Fig. 1E).

In contrast, only a trend of correlation was observed betweenCSF AChE activity and CSF GFAP (r ¼ 0.089; p ¼ 0.33; n ¼ 121;Supplementary Fig. 1A) or CSF S100B (r ¼ 0.17; p ¼ 0.067;Supplementary Fig. 1C). However, the protein level of the secretoryread-through AChE-R, but not the synaptic AChE-S splice variantcorrelated with the CSF GFAP level (r ¼ 0.31; p < 0.0022; n ¼ 97;Supplementary Fig. 1B) and the CSF S100B level (r ¼ 0.37;p < 0.0002; n ¼ 97; Supplementary Fig. 1D).

3.4. CSF levels of IL-1b and TNF-a are strongly dependent on thegenotype of BuChE

Release of the astrocytic cytokine, S100B, is associated withactivation of microglia and expression of proinflammatory cyto-kines such as IL-1b, IL-6, and TNF-a (Gasque, 2004; Giulian et al.,1988; Koppal et al., 2001; Mrak and Griffin, 2001). Because astro-cytes and microglia are cholinoceptive and the BuChE-K variant isassociated with reduced ACh-hydrolyzing capacity of the enzyme(Darreh-Shori et al., 2012), we investigatedwhether the K allelewasrelated to the CSF levels of the proinflammatory cytokines, IL-1b,IL-6, and TNF-a. An overview of the absolute values of the data ispresented in Table 1.

CSF levels of IL-1b significantly differed between the BCHE-Kgenotype groups (F ¼ 5.74; n ¼ 112; p < 0.0043). Post hoc analysisunexpectedly showed that the K noncarriers (n ¼ 68) displayed an81% lower level of IL-1b in CSF than BCHE-Kþ/þ (n ¼ 8; p < 0.0089;Fig. 1C). The corresponding difference between the K noncarriers

and BCHE-Kþ/� heterozygotes was 40 � 8% (n ¼ 37; p < 0.012;Fig. 1D). In contrast, the difference between heterozygotes andBuChE-Kþ/þ did not reach statistical significance (41 � 14%;p < 0.20).

As noted in the “Methods” section, CSF levels of TNF-a wasdetermined only in a subgroup of the patients (n ¼ 49), who hadalso undergone in vivo FDG- (and/or PIB-) PET assessments. In thissubgroup, a total of 14 patients were carriers of the K allele (Fig. 1B).Nonetheless, even among this subgroup, significant differencesbetween K carriers and noncarriers were apparent (F ¼ 5.34;p < 0.0082; n ¼ 49). Consistent with the observation for IL-1b, wefound that the K noncarriers had a 27�4% lower level of TNF-a thanthe K carriers (p < 0.0030; n ¼ 14; Fig. 1D; or just betweenheterozygotes [n ¼ 12] and noncarriers, 21 � 4%; p < 0.013).

3.5. GFAP and S100B inversely correlate with CSF proinflammatorycytokines

GFAP and S100B have been implicated as surrogate markers forastrogliosis and thereby as a measure of activity of these inflam-matory cells in the AD brain. We hence expected to find a positivecorrelation between these 2 glial markers and the levels of proin-flammatory cytokines in CSF, which proved not to be the case.Instead, we found an inverse correlation between GFAP and IL-1b(r ¼ �0.46; p < 0.0001; n ¼ 112; Fig. 2A), and IL-6 (r ¼ �0.25;p < 0.0068; n ¼ 112; Fig. 2B), but not with TNF-a, most likelybecause of the fewer numbers of subjects with available TNF-a data(r ¼ �0.16; p < 0.28; n ¼ 49).

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 7

The CSF levels of the astrocytic cytokine, S100B, also showedreciprocal association but reached statistical significance only withIL-1b levels (r¼�0.32; p< 0.0001; n¼ 112; Fig. 2C) but not with IL-6 (Fig. 2D).

3.6. Cross-talk between peripheral and central compartments mightbe relayed via an IL-1beBuChEeACh regulatory axis of functionalastroglial status

Evidence suggest a close cross-talk between the peripheralinflammatory cascades and the brain responses, most likely medi-ated by transport of proinflammatory cytokines to the brain (Laskeet al., 2013; Ofek and Soreq, 2013), which in turn might affect thefunctional status of astrocytes in the brain. We hence investigatedthe interrelationship of peripheral cholinesterase activity andproinflammatory cytokines.

Consistent with this view, we found a significant positivecorrelation between the plasma BuChE activity and the plasmaIL-1b (Fig. 3A), and between plasma and CSF levels of IL-1b(r ¼ 0.35; n ¼ 73; p < 0.0021; Fig. 3B). Similar to CSF, the plasmaBuChE activity also showed a strong positive correlation with theCSF levels of astroglial markers, S100B (Fig. 3C), and GFAP (Fig. 3D).In contrast, plasma AChE activity did not show any correlation withplasma IL-1b (Fig. 3B), or with CSF S100B (r ¼ �0.08; n ¼ 73;p < 0.51; Fig. 3G) and GFAP (r ¼ 0.19; n ¼ 73; p < 0.1; Fig. 3H).

However, it should be noted that the relationship betweenplasma BuChE and plasma IL-1b was the opposite of the relation-ship between CSF BuChE and IL-1b (compare Fig. 3A with Fig. 3Eand Fig. 1C and Darreh-Shori et al. (2011b)).

It is also important to note that we found no significant differ-ence between the BCHE-K genotype groups with regard to plasmalevels of IL-6 (F ¼ 0.68; p ¼ 0.51; n ¼ 133; Table 1) or plasma levels

−2

2

6

10

14

18

0 1 2 3 4 5 6 7

Plasma BuChE activity (μmol/min/mL)

Plas

ma

IL-1

β le

vels

(pg/

mL)

−2

0

2

4

6

8

−2 2 6 10 14 18

CSF

IL-1

β le

vels

(pg/

mL)

Plasma IL-1β levels(pg/mL)

Plasma (μm

0

50

100

150

200

250

300

0 1 2

CSF

S10

0B (n

g/m

L)

A B C

F G

r = 0.35p < 0.0021n = 73

−2

2

6

10

14

18

0 50 100 150 200 250 300

Plasma AChE activity (nmole/min/mL)

r = −0.18p < 0.13n = 73

r = 0.43p < 0.0002n = 67

r = 0.54p < 0.00n = 73

0

50

100

150

200

250

300

CSF

S10

0B (n

g/m

L)

0 50 10

Plasma(nmo

Plas

ma

IL-1

β le

vels

(pg/

mL)

Fig. 3. Interrelationship of peripheral plasma butyrylcholinesterase (BuChE) activity and inte(AD) patients. Plasma BuChE correlates positively with the levels of IL-1b in the blood circulathe CSF levels of IL-1b in the AD patients (B). Additionally, high plasma BuChE activity is alsolevels of the astroglial markers, S100B (C) and glial fibrillary acidic protein (GFAP) (D). This clthe phenotypic display of this enzyme in CSF and the astroglial markers (illustrated in Fig. 1Ain CSF is associated with low IL-1b levels in the CSF of the AD patients (E), which is in lineassociation with the plasma levels of IL-1b (F), or with the astroglial markers, S100B (G) andwith BuChE (AeE) .

of IL-1b (F ¼ 0.46; p ¼ 0.64; n ¼ 134; Table 1). No correlations wereobserved between IL-6 levels in plasma and CSF (r ¼ 0.12; n ¼ 66;p < 0.33), between plasma IL-6 and the plasma BuChE (p < 0.31) orAChE (p < 0.45), or between plasma levels of IL-1b or IL-6 and theastroglial markers, GFAP or S100B, in CSF.

Nonetheless, overall observations once more pointed out BuChE(but not AChE) and IL-1b as 2 important determinants of astroglialfunctional status and pathophysiological hallmarks of AD, consis-tent with the results illustrated in Fig. 1 and numerous precedingreports on the association of BCHE and IL-1 genetic polymorphismsand AD (Darreh-Shori et al., 2006; Di Bona et al., 2008; Giulian et al.,1988; Mesulam and Geula, 1994; Payao et al., 2012; Perry et al.,2003; Yuan et al., 2013).

3.7. In vivo FDG- and PIB-PET correlates with the CSF levels of GFAPand S100B

The unexpected reciprocal association of astroglial markers,GFAP and S100B, and the proinflammatory cytokines in CSF, mightindicate that a high CSF level of these 2 markers might reflecta beneficial astroglial functional status and response. We henceinvestigated whether in vivo PET markers could support thisnotion. We found that the level of glucose utilization (FDG) in mostof the brain regions correlated positively with CSF levels of GFAPbut mainly among Apo ε4 carriers (Fig. 4A). In contrast, the corre-lation among the Apo ε4 noncarriers was reciprocal (r ¼ �0.593;p < 0.02; n ¼ 16).

In addition, we also found that CSF levels of GFAP correlatedinversely with the averages of PIB retention in the cortical regions(r ¼ �0.32; p < 0.05) and the whole-brain regions (r ¼ �0.52;p < 0.002; Fig. 4B). However, the strongest correlation was observedmainly in the subcortical brain regions, such as the thalamus

0

1

2

3

4

0 1 2 3 4 5 6 7

CSF

GFA

P (n

g/m

L)

BuChE activity ol/min/mL)

3 4 5 6 7

Plasma BuChE activity (μmol/min/mL)

H

D E

01r = 0.33p < 0.005n = 73

−1

0

1

2

3

4

5

6

2 4 6 8 10 12 14

CSF BuChE activity (nmole/min/mL)

CSF

IL-1

β le

vels

(pg/

mL)

r = −0.29p < 0.005n = 73

0

1

2

3

4

CSF

GFA

P (n

g/m

L)

0 50 100 150 200 250 300

Plasma AChE activity (nmole/min/mL)

0 150 200 250 300

AChE activity le/min/mL)

r = −0.12p < 0.51n = 73

r = 0.19p < 0.1n = 73

rleukin (IL)-1b is related to the functional status of astroglial cells in Alzheimer’s diseasetion of patients with AD (A). The plasma IL-1b in turn shows a positive correlation withassociated with high astroglial functional status as deduced by cerebrospinal fluid (CSF)osely resembles the corresponding positive association between the BCHE genotype andand 1B and 1F and 1G). In contrast to the observation in plasma (A), high BuChE activitywith the BCHE genotype (Fig. 1C). Plasma acetylcholinesterase (AChE) activity show noGFAP (H) in CSF of the patients, contrasting the strong association of these biomarkers

Non-APOE4A B

C D

FDG

-PET

in th

e w

hole

brai

n( %

Mea

n)

r = 0.66p < 0.0002

n = 33

− 40

− 20

0

20

40

60

− 80 − 40 0 40 80 120

CSF

leve

ls o

f GFA

P (%

Mea

n)

r = −0.53p < 0.006

n = 29

− 60 − 40 − 20 0 20 40 60 80

CSF levels of TNF-α (%Mean)

−60

−40

−20

0

20

40

60 r = −0.61p < 0.0001

n = 51

−100

− 75

− 50

− 25

0

25

50

75

100

− 50 − 25 0 25 50

CSF levels of GFAP (%Mean)

FDG

-PET

in th

e w

hole

brai

n( %

Mea

n)

PIB in the whole brain (%Mean)

APOE4/*

PIB in the whole brain (%Mean)

− 100

− 75

− 50

− 25

0

25

50

75

100

− 50 − 25 0 25 50

CSF

leve

ls o

f TN

F-α

(%M

ean)

r = 0.32p < 0.045

n = 29

Fig. 4. A high cerebrospinal fluid (CSF) level of glial fibrillary acidic protein (GFAP) is a good sign whereas a high level of tumor necrosis factor (TNF)-a seems to be a poor sign ofin vivo markers of Alzheimer’s disease (AD). Levels of the astroglial marker, GFAP, in the CSF of AD patients correlates positively with the in vivo cerebral glucose utilization (A),particularly in apolipoprotein (Apo) ε4 (APOE4) carriers. In contrast, GFAP shows a reciprocal association with the in vivo marker of amyloid-b (Ab) deposition in the brain of the ADpatients (B). These findings strongly suggest that the brain of the AD patients who managed better in regulating astroglial function were most successful in either preventing accu-mulation of or getting rid of theAbdeposits andmaintaining themost preserved function. Alternatively, a high level of CSFGFAP is an indication that theADpatients are at amilder stageof the disease compared with those with low CSF GFAP. However, the opposite pattern of association between CSF TNF-a level and cerebral glucose utilization (C) or the in vivo brainlevels of Ab load (D) counter argues the latter alternative, because apparently CSF levels of TNF-a did not follow theGFAP or S100B levels (see Fig.1). In all axes of the graphs, the data areshown as percentages of the overall mean value minus 100%. Abbreviations: FDG, (18)F-fluorodeoxyglucose; PET, positron emission tomography; PIB, Pittsburgh Compound-B.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e178

(r¼�0.60; p< 0.0002; n¼ 29), putamen (r¼�0.51; p< 0.003), andregions of the medial temporal lobe (temporal inferior anterior,r¼ �0.41; p< 0.02; temporal inferior posterior, r ¼ �0.27; p< 0.08;temporal lateral anterior and posterior, r ¼ �0.36; p < 0.03; uncusposterior, r¼�0.42;p<0.01; anduncusanterior, r¼�0.32;p<0.05).

In general, no correlations were observed between CSF S100Band FDG- or PIB-PET data.

3.8. In vivo FDG- and PIB-PET correlates with the CSFproinflammatory cytokines

Then, a similar correlation between IL-1b, IL-6, and TNF-a levelsand in vivo brain PET data was also performed. We found that theCSF levels of TNF-a inversely correlated with glucose utilization inessentially all of the brain regions (Fig. 4C).

With regard to PIB-PET data, CSF TNF-a showed positive corre-lations or trends mainly with PIB retention in the brain regions:thalamus (r ¼ 0.33; p < 0.04; n ¼ 29), putamen (r ¼ 0.29; p < 0.07),medial temporal lobe (temporal inferior anterior, r¼ 0.38; p< 0.03;temporal inferior posterior, r ¼ 0.27; p < 0.08; uncus posterior,r ¼ 0.38; p < 0.02; uncus anterior, r ¼ 0.27; p < 0.08), and thetemporal cortex (r ¼ 0.32; p < 0.05; n ¼ 29); the average corticalregions (r ¼ 0.31; p < 0.06), average whole-brain regions (Fig. 4D;r ¼ �0.53; p < 0.0061; among Apo ε4 carriers; n ¼ 21). Thestrongest TNF-a correlation was with PIB retention in the rightcerebellar cortex (r ¼ 0.53; p < 0.001). The overall pattern washence that a high PIB retention was related to higher expressionlevels of TNF-a in the CSF of the subjects.

3.9. Correlates of cognition support the effect of BCHE genotype andphenotype on astroglial inflammatory response

We found a positive correlation between CSF levels of S100B andthe patients’ performance score on the MMSE (r ¼ 0.26; p < 0.005;n ¼ 122), which was most pronounced among the K carriers(r ¼ 0.46; p < 0.0009; n ¼ 48) or among the Apo ε4 carriers(r ¼ 0.46; p < 0.0001; n ¼ 91).

Similarly, the overall composite Z-scores of the patients corre-lated positively with the CSF levels of S100B (r ¼ 0.34; p < 0.005;n ¼ 66), which again was most pronounced among the K carriers(r ¼ 0.43; p < 0.04; n ¼ 23) or among the Apo ε4 carriers (r ¼ 0.45;p < 0.002; n ¼ 45).

The results of cognitive testing indicated a K allele dosage-dependency, mainly among the Apo ε4 carriers (Fig. 5). In addi-tion to the correlation with S100B (Fig. 1G), the CSF BuChE activitycorrelated also with the overall composite Z-scores (r ¼ 0.38;p < 0.0014; n ¼ 66), especially among Apo ε4 carriers (r ¼ 0.61;p < 0.0001; n ¼ 45).

The overall composite Z-scores of the patients correlated nega-tively with the CSF levels of IL-1b (r ¼ �0.28; p < 0.03; n ¼ 60). Theassociation was strongest in the K carriers (r ¼ �0.43; p < 0.04;n¼ 23) and in patients carrying Apo ε4 (r¼�0.33; p< 0.03; n¼ 42).Similarly, the patients’ MMSE scores showed reciprocal correlationwith the CSF levels of IL-1b (r ¼ �0.31; p < 0.0008; n ¼ 113; Apo ε4carriers, r ¼ �0.40; p < 0.0002; n ¼ 85).

The MMSE scores, but not the overall composite Z-score,also showed reciprocal correlation with the CSF levels of IL-6

Wt/WtWt/KK/K

**

Ove

rall

Cog

nitio

n(c

ompo

site

Z-s

core

s)

−16

−12

−8

−4

0

3 210

8

32 9

APOE ε4/* Non-APOE ε4−40

−32

−24

−16

−8

0

Red

uced

MM

SE S

core

(% M

axim

um 3

0)

9

542 1776 28

−100

−50

0

50

100

150C

SF S

100B

(%K/

K)

62

2713

57 14

***

*

**

*

A

B

C

#

Fig. 5. The interaction between apolipoprotein (Apo)E ε4 (APOE ε4) and BCHE-Kgenotypes on the astrocytic cytokine, S100B, and the patients’ cognition. (A) Thecerebrospinal fluid (CSF) levels of S100B show a robust K allele dosage dependency inthe Apo ε4 carriers but not in the noncarriers. Among the Apo ε4 carriers, the CSF levelsof S100B show 50% lower values per each K allele. In contrast, among the Apo ε4noncarriers, the heterozygotes and noncarriers of the K allele have an essentiallysimilar level of CSF S100B. (B) The pattern of Alzheimer’s disease (AD) patients’composite Z-score of cognition follows closely the opposite pattern of CSF S100B. Inother words, the highest levels of CSF S100B are associated with the best cognitiveperformance among the Apo ε4 carriers AD patients. (C) Confirms the pattern offindings in (B) using the cognitive scores of the AD patients in the Mini Mental StateExamination (MMSE) test. Overall, the regulatory effect of butyrylcholinesterase(BuChE) on the CSF S100B levels and thereby on the relative preservation of thepatients’ cognition is most apparent among the Apo ε4 carriers. The digits in thecolumns indicate the number of subjects. Data are given as percentages of mean valueof the Kþþ group in (A), the calculated composite Z-score in (B) and percentages ofdecline from the maximum attainable score on the MMSE test in (C), and standarderror of the mean. The calculated overall composite Z-score is the average of theZ-score of the following cognitive domains: episodic memory, attention/executivefunction, and visuospatial ability, as described in the Methods section. * p < 0.05,** p < 0.01, and *** p < 0.001 compared with the wild type BCHE homozygotes (theK�� group). # p < 0.082 indicates a strong trend of differences with the K�� group.Abbreviation: Wt, wild type.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 9

(r ¼ �0.24; p < 0.01; n ¼ 113; Apo ε4 carriers: r ¼ �0.26; p < 0.02;n ¼ 86).

No correlation was observed between cognition and the CSFTNF-a most likely because of fewer numbers of CSF samples(n ¼ 49). No correlation was observed between cognition and theCSF levels of GFAP.

Overall pattern of the observations hence suggest a dynamicmodel for the astroglial functional status that follows an inverseU-shape, with regards to proinflammatory cytokines and the in vivoparameters and clinical manifestation of AD. This seems to bestrongly related to the individual functional variability in the BuChEactivity, predicted by either inheritance (BCHE genotype) oracquired phenotypic display of the enzyme by, for instance, highApoE protein (Darreh-Shori et al., 2012), or a combination of both.Thus, to get more insight about the role of BuChE we used 2different experimental paradigms.

3.10. Immunohistochemical analyses support the notion ofa regulatory role of BuChE on function of cholinoceptive astroglialcells

Mechanical nerve injury by VRA is a simple and reproduciblemodel of nerve injury-induced neuroinflammation (Piehl et al.,2007). We used this experimental model and examined whetherBuChE expression shows a colocalization pattern with the focalassembly of GFAP-positive astrocytes and Iba1-positive microglialcells. It should be emphasized that we used this model to excludepossible secondary nonenzymatic involvement of BuChE, whichmight arise through its molecular interaction with specific patho-logical features present in the AD brain (Darreh-Shori et al.,2011a, 2011b), such as Ab and tau deposits and/or AD-derivedastrocytosis (Geula et al., 1994; Guillozet et al., 1997; Mesulamand Geula, 1994).

Triple staining of spinal cord sections 5 days after VRA, revealedfocal colocalization of GFAP-positive glial cells around areas withstrong focal BuChE immunolabeling (Fig. 6AeE).

Iba1 is regarded as a marker of activated microglial cells. Addi-tional histochemical analysis with Iba1 immunolabeling produceda similar focal assembly of Iba1-positive glial cells, which wasevident in areas with considerable BuChE immunolabeling (Fig. 6Fand G).

3.11. CSF levels of complement components are also associated withthe BuChE phenotype

Emerging evidence points at an unexpected role for thecomplement system of the innate immunity in synaptic eliminationand/or remodeling in neurodegenerative diseases (Stephan et al.,2012).

Further independent genetic analyses on the previouslymentioned experimental rat model have suggested a link betweenexpressions of BuChE and the complement factor C3 (unpublisheddata).

We hence performed an analysis of CSF of a small group of theAD patients (n ¼ 10) and a group of OND patients (n ¼ 10), andinvestigated whether CSF levels of complement factors wererelated to phenotypic display of BuChE, which might be morerelevant than BCHE genotype (Darreh-Shori et al., 2012).

Highly significant correlations were observed between levels ofseveral components of the complement system and the BuChEactivity (Fig. 7AeD) and BuChE protein expression in CSF (p < 0.01vs. C3, C1q, C4, and H). Noteworthy, the correlations between CSFAChE activity and the complement components was much weaker,and CSF AChE protein levels did not correlate at all with the CSFlevels of the complement factors (all p values > 0.36), suggestingthat the relationship between BuChE and the various inflammatorymarkers is mainly specific to the activity of this enzyme rather thanto AChE.

The concentrations of the complement factors and BuChEactivity were significantly higher in the CSF of patients withAD compared with patients with OND (Fig. 7EeG). However, itshould be noted that the OND group was composed of muchyounger patients (30.3 � 2.9 years) than the AD group (69.8 � 2.7years).

Interestingly, even if data were available only for a smallsubgroup of the AD patients, there were highly significant corre-lations between the CSF levels of complement factors and thecortical glucose utilization (Supplementary Fig. 2AeC) and thecognitive status of the AD patients (Supplementary Fig. 2EeG). Ingeneral, there was also a significant correlation between CSF S100Band the complement factors, particularly the C3a (SupplementaryFig. 2D), and C4 (Supplementary Fig. 2H).

Fig. 6. Focal butyrylcholinesterase (BuChE) immunolabeling with glial fibrillary acidic protein (GFAP) and Iba1-positive glial cells in a rat model of neuroinflammation. MergedBuChE-GFAP (AeE) and merged BuChE-Iba1 (FeJ) immunofluorescence micrographs of rat spinal cord sections 5 days after the ventral root avulsion experimental paradigm, asa model of nerve injury-induced neuroinflammation (Piehl et al., 2007). Triple staining for BuChE (red), GFAP (green), and nuclear DAPI staining (not shown). (BeE) Enlarged zonesof the micrograph (A), illustrating GFAP-positive astrocytes adjacent to focal BuChE immunolabeling. Similarly, (F) illustrates merged triple staining for BuChE (red), nuclear DAPI,and a marker of activated microglia, Iba1 (green). (GeJ) are expanded zones from the micrograph (F), showing Iba1-positive microglia surrounding the areas and/or cells with strongBuChE immunolabeling. Abbreviations: Iba1, Ionized Calcium-Binding Adapter Molecule 1; DAPI, 40 ,6-diamidino-2-phenylindole.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e1710

3.12. Human primary astrocytes adjust their expression and releaseof BuChE but not AChE in the culture medium in response to TNF-a

Next, we stimulated human brain primary astrocytes withrecombinant human TNF-a alone or with varying concentrations ofACh (Fig. 8AeF). Human brain astrocytes released AChE and BuChEinto the culture medium. The AChE activity was generally higher inthe culture medium (AChE: 1.11 � 0.14 nmol/min/mL) than BuChEactivity (0.376 � 0.07 nmol/min/mL). This is almost the same pro-portionality that we have reported in CSF of AD (CSF AChE,approximately 25 nmol/min/mL vs. CSF BuChE, approximately8 nmol/min/mL) (Darreh-Shori et al., 2012).

TNF-a alone reduced the gene expression of BuChE by 60%(Fig. 8A), which was also accompanied by a similar strong reductionin the BuChE activity and protein levels that were measured in theculture medium of the cells compared with that of the unstimu-lated cells (Fig. 8B and C). In contrast, the cells did not alter theirrelease of AChE in the cell culture medium in response to TNF-a, aswas deduced by the measured AChE activity and protein in theculture medium relative to the basal levels (Fig. 8D and E). Thus,astrocytes most readily changed the expression and release of

BuChE, plausibly as an adaptive response, compatible with CSFobservations.

Furthermore, TNF-a stimulation caused a 12-fold increase in thegene expression of the complement factor, C3, compared with theunstimulated astrocytes (Fig. 8F). The TNF-induced reductions inthe BuChE levels and increases in the C3 gene expression remainedunaltered when ACh was present at micromolar levels (Fig. 8).Intriguingly, stimulation of the astrocytes with TNF-a and 1e10mMof ACh seem to normalize BuChE and C3 levels toward the basallevels of the unstimulated cells (Fig. 8AeD). Again, the combinedTNF-ACh treatment caused no changes in the released levels ofAChE in the culture medium (Fig. 8D and E).

Noteworthy was that at millimolar ACh concentrations (0.1e100mM) in combination with TNF-a, the elevated C3 gene expressionshowed a reverse ACh-dependent reduction to the basal levels. Atthe highest combination of TNF-a with ACh (100 mM), the geneexpression analysis indicated that the expression of C3 went downback to approximately 50% of the levels observed in the unstimu-lated cells (Fig. 8F), and the BuChE gene expression reached back tothat in the unstimulated astrocytes (Fig. 8A). This was accompaniedby more than 50% and 75% increases in the BuChE activity and

Fig. 7. Phenotypic display of butyrylcholinesterase (BuChE) in cerebrospinal fluid (CSF) shows a strong association with the CSF levels of several components of the innatecomplement immune system. Highly significant positive correlations between CSF BuChE activity and the complement factors, C3 (A), C1q (B), C4 (C), and H (D) in a subgroup of theAlzheimer’s disease (AD) patients (n ¼ 10), and patients with other neurological disorders (OND, n ¼ 10). (E) and (F) The highly significant differences in the CSF concentrations ofthe complement factors between patients with AD and OND. Similarly, the CSF BuChE activity status shows highly significant difference between AD and OND (G). The opendiamonds and filled diamonds denote patients with OND and with AD, respectively. Concentration of complement factors were determined using an enzyme-linked immunosorbentassay. BuChE activity was assessed colorimetrically. ** p < 0.01; *** p < 0.001.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 11

protein levels, respectively (Fig. 8B and C). In contrast, the levels ofAChE activity and protein in themediumwere reduced to 50% of theunstimulated cells. This was hence the only assay condition thatresulted in a change in the AChE level (Fig. 8D and E). Noteworthy isthat the direction of change was the opposite of an expectedfeedback response for lowering the ACh level that was observed forBuChE (Fig. 8B and C).

In general, the results of stimulation with only ACh stronglysupported those observed with the combination of ACh and TNF-a(Fig. 8GeL). In addition, the result indicated that low micromolarconcentrations of ACh were as effective as stimulation with TNF-ain reducing the BuChE activity and protein levels in the medium ofthe astrocytes (compare Fig. 8B and C with H and J). Again, nochanges were observed in the activity and protein levels of AChE.

4. Discussion

In this study, we found a strong association between BuChEgenotype and phenotype and biomarkers of glial functions andinflammation. The pattern of the observations are compatible witha model in which the activity of astroglial cells follows an inverseU–shape dynamic in the continuum of AD, in which the individualfunctional adaptably in the overall ACh-hydrolyzing status byBuChE seem to play a crucial role (Fig. 9).

Themajor difference between the K and thewild type variants ofBuChE is related to the subjects’ overall cholinesterase activity

status, with more than 30%e60% higher ACh-hydrolyzing activity inthe CSF of wild type carriers compared with the K homozygotes(Darreh-Shori et al., 2012). Compelling evidence indicates that AChacts as an anti-inflammatory agent exerted via a7 nicotinic recep-tors (Wang et al., 2003). Astroglial cells are cholinoceptive andexpress several cholinergic proteins, including the a7 nicotinicreceptors (Yu et al., 2005), which are ion channels. This implies thatthe putative immunoregulatory action of ACh requires the presenceof certain extrasynaptic concentrations of ACh in the brain paren-chyma that can fine-tune glial cell activation status. In this regard,extracellular ACh concentration is expected to be relatively lower inthe brain parenchyma of individuals carrying the wild type formcompared with carriers of BCHE-K (see Fig. 9) or to be more readilyable to adjust the extracellular ACh concentration when needed bychanging the levels of expression and release of this enzymecompared with the K variant carriers (Darreh-Shori et al., 2012). Inother words, the brain parenchyma of the K and wild type carriersare expected to possess relatively different intrinsic capabilities tocope with events related to cholinergic signaling. Obviously thismight be of special relevance in AD, because of alterations in theneuronal cholinergic networks and a state of chronic inflammatoryactivation of glia.

In a large and well characterized sample of patients with AD, inthis study we found that CSF levels of GFAP and S100B significantlydepend on BCHE-K gene dosage and display a highly significantcorrelationwith BuChE activity, with patients carrying the BuChE-K

C3 expression

BuChEActivity Protein

BuChE expression

AChEActivity Protein

0

40

80

120

qPC

R(%

Us)

0

50

100

150

200Le

vels

in c

ultu

re m

ediu

m

(%U

s) 6000

12000

18000

0

60

120

180

qPC

R(%

Us)

0

50

100

150

200

0

40

80

120

1 10 100

1 10 100qP

CR

(%U

s)

0

50

100

150

200

250

Leve

ls in

cul

ture

med

ium

(%

Us) 6000

12000

18000

0

60

120

180

qPC

R(%

Us)

0

50

100

150

200

Stimulation with TNF-α and ACh

Stimulation with ACh alone

Us

TNF-

α µM mM ACh

Us

TNF-

α µM mM ACh

Us

TNF-

α µM mM ACh

Us

TNF-

α µM mM ACh

Us

TNF-

α µM mM ACh

Us

TNF-

α µM mM ACh

A B C D E F

G H I J K L

Fig. 8. Regulatory changes in the butyrylcholinesterase (BuChE) level induced by tumor necrosis factor (TNF)-a and acetylcholine (ACh) in human primary astrocyte cell cultures.(AeE) Stimulation of TNF-a alone or in combination with micro- to millimolar concentrations of ACh affect the gene expression of BuChE (A) in the astrocytes, BuChE activity (B), andprotein (C) released in the culture medium. The gene expression of complement factor C3 (F) follows an opposite pattern of changes in the astrocytes compared with the BuChEpattern. In contrast, the activity (D) and protein (E) levels of acetylcholinesterase (AChE) in the culture medium was not affected. (GeL) Stimulation with lower micromolarconcentrations of ACh alone exhibit a similar effect on the BuChE activity (H) and protein (I) levels in the culture medium of the cells, and the BuChE gene expression (G) is minimallyaffected in the absence of TNF-a compared with combined stimulation with both TNF-a and low ACh concentrations (AeC). In addition, ACh alone fails to affect the C3 geneexpression (L). Contrary to the observed changes in the BuChE levels, no changes in the levels of AChE activity (J) and protein (K) in the cell culture medium occurs when ACh ispresent at 1 mM to 10 mM. Only at the highest tested ACh concentration (100 mM), alone or together with TNF-a, the AChE levels show a significant reduction (D) or (J) and (E) or (K),contrasting the strong increases in the BuChE levels (B) or (C) and (H) or (I) at this high ACh concentration. Abbreviation: Us, unstimulated.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e1712

variant having relatively lower CSF levels of GFAP and S100B. Inmany pathological conditions the levels of glial biomarkers andcytokines are concomitantly elevated, suggesting a reciprocalpositive correlation. Unexpectedly, however, we found that thehighest levels of the proinflammatory cytokines IL-1b and TNF-a inthe CSF of K carriers.

This finding might indeed seem counterintuitive, becauseactivation of astroglial cells is expected to go hand-in-handwith therelease of these proinflammatory cytokines. However, the pre-sumed strong correlation between glial activation and expression ofproinflammatory mediators mostly derive from experimentalmodels or comparisons between healthy and diseased individuals.The functional consequences of glial activation in chronic diseasessuch as AD are likely to be much more complex. Thus, particularqualities of a fine-tuned and successful activation and/or prolifer-ation of astroglial cells might lead to a relatively more efficientremoval of the noxious stimuli generated by the underlying diseaseprocess, in turn leading to an ameliorated disease course (Fig. 9).

Consistent with this notion, we found that higher CSF GFAPlevels were indeed indicative of relatively lower in vivo brain PIBretention, and higher cerebral glucose utilization in the AD patients.In contrast, high CSF TNF-a displayed a positive correlation toseverity of the disease, as deduced by their reciprocal associationwith cerebral glucose utilization and increased Ab load (PIBretention) in the brain, particularly among the Apo ε4 carriers.Consequently, our findings suggest that the higher levels of GFAPand S100B in CSF, contrary to what is found in postmortem brain,

reflect a beneficial response directed at coping with the ongoingunderlying AD disease process and that BuChE plays an importantregulatory role in this context.

Additional support for our proposed inverse U-shape dynamicmodel of astroglial cell activity in AD is provided by the recentin vivo PET study (Carter et al., 2012), that reportedmarked increasein the 11C-L-deprenyl binding in the brain of PIB-positive patientswith mild cognitive impairment (MCI) compared with controlsubjects or clinically diagnosed AD patients (Carter et al., 2012). Incontrast, no significant difference was apparent between the brainsof AD patients and brains of control subjects.

Regional L-deprenyl binding in the postmortem AD braincorrelates positively with BuChE activity (Kadir et al., 2011). In thecurrent study, we further explored the relationship betweenastroglial cell activity and BuChE by first using an AD-unrelatedanimal model to exclude a possible nonenzymatic contribution ofBuChE that might be caused by molecular interactions of BuChEwith AD-specific Ab peptides and tau abnormalities present in theAD brain (Mesulam and Geula, 1994). We found that the histo-chemically observed pattern of focal BuChE immunolabeling withGFAP and Iba1-positive astroglial cells was in line with the associ-ation of BuChE with regional L-deprenyl binding in the postmortemAD brain.

Next we used an in vitro cell culture paradigm using primaryhuman brain astrocytes in inflammatory (TNF-a) and anti-inflammatory (ACh) conditions and a combination of both. Theexpression and release of BuChE by astrocytes was altered by TNF-a

Fig. 9. Hypothetical mechanistic explanation for the regulatory influence of the BCHE genotype and phenotype on the levels of astroglial glial markers. The key feature is thatacetylcholine (ACh) has been shown to exert a suppressive action on glial cell activity. Then 30%e60% higher activity of the wild type (Wt) butyrylcholinesterase (BuChE) variantcompared with the K variant might facilitate efficient fine-tuning of the extracellular levels of ACh, and thereby its suppressive and/or regulatory action on cholinoceptive cells suchas astrocytes and microglia (as shown in Fig. 8). A fine-tuned activity of these cells will then be reflected as a relative release or accumulation in cerebrospinal fluid (CSF) levels ofglial fibrillary acidic protein (GFAP) and S100B between the K carriers and noncarriers, which in turn becomes evident because of higher levels of GFAP and S100B in CSF (asillustrated in Fig. 1A and B). A more efficient handling of the Alzheimer’s disease (AD)-derived noxious stimuli is therefore expected in better reduction and/or prevention ofexcessive amyloid-b (Ab) deposition in the brain in the Wt than in K carriers of BuChE as is shown by the reciprocal association between GFAP and PIB-positron emissiontomography (PET) (Fig. 4). This in turn might lead to readjustment of the release of the proinflammatory cytokines, for example, interleukin (IL)-1b tumor necrosis factor (TNF)-a,and complement (C) system (as shown in Figs. 1, 2, and 8), and a best-preserved brain function which might be appreciated by the higher cerebral glucose utilization and cognitiveperformance of the Alzheimer’s disease (AD) patients (as is observed in Fig. 5). Apolipoprotein (Apo) ε4 (APOE4) genotype, through a high expression level of ApoE protein,negatively modifies the phenotypic display of BuChE protein variants toward a pronounced K variant-like phenotypic ACh-hydrolyzing activity (Darreh-Shori et al., 2011b, 2012).Abbreviations: Ab, amyloid-b; PIB, Pittsburgh Compound-B.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 13

and ACh whereas the levels of AChE remained remarkably unaf-fected. Intriguingly, these specific changes in the BuChE levelsoccurred despite the observations that the basal activity level ofAChE, released by these cells in the culture medium was approxi-mately 3-fold higher than BuChE levels, which is essentially thesame relative AChE/BuChE activity observed in human CSF andperipheral circulation (Darreh-Shori et al., 2008, 2012). Thusalthough the overall extrasynaptic ACh-hydrolyzing status isconceivably governed by the sum of the activities of these 2enzymes, the necessary adaptive changes seem to be most readilyaccomplished by a shift in the expression and release of BuChE.

A clinically important aspect of the current report is that ourfindings provide some mechanistic insight about responders andnonresponders to the current cholinesterase inhibitor therapy in thedementia disorders. For example, our model predicts, in line withclinical observations, that subjects with the BuChE-K variant mighthave a lower rate of cognitive decline (Darreh-Shori et al., 2012;O’Brien et al., 2003) and disease progression (Lane et al., 2008)because of an approximate 30%e50% inherent reduction of BuChE(Darreh-Shori et al., 2012), as if continuing (with some reservations)long-term treatment with a cholinesterase (ChE) inhibitor. However,this group is more likely to accumulate severe AD-like pathologies

and they do not respond well to the ChEI therapy (O’Brien et al.,2003), most likely because, when they reach the point of clinicaldiagnosis, have already acquired toomuch damage and/or atrophy tothe neuronal network in the brain, in particular, cholinergic neurons.In contrast, individuals with high BuChE activity (wild type carriers)are more likely to cope with the underlying pathological events ofAD, but if this fails then they will relatively quickly show the clinicalsymptoms related to the cholinergic dysfunction (O’Brien et al.,2003). Therefore, they are more likely to respond to the ChEItherapy (Ferris et al., 2009) because their brains have acquiredrelatively less amyloid-b (Ab)-derived damage. The presence of Apoε4 worsens the situation because of a high expression of ApoE(Darreh-Shori et al., 2011b), which forces a pronounced K variantphenotype across all BCHE genotypes but particularly among the Kþ/ε4þ (Darreh-Shori et al., 2012).

Indeed, an important observation in this and in our previousstudies is that high BuChE activity in AD patients is associated withmore benign clinical and paraclinical findings regarding diseaseprogression (Darreh-Shori et al., 2006, 2011a, 2011b; Lane et al.,2008). Intriguingly, the main genetic AD risk factor, Apo ε4, cau-ses phenotypic modulation of BuChE variants to such degrees thatthe wild type BuChE variant displays enzymatic activity similar to

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e1714

that seen in subjects with the BuChE-K variant (Darreh-Shori et al.,2006, 2012). This phenotypic modulation of BuChE by Apo ε4 hasbeen linked to high CSFApoE protein, a common condition in CSF ofApo ε4-carrier AD patients. Consequently, high ApoE levels lead toaccumulation of a dormant pool of BuChE in the CSF of AD patients(Darreh-Shori et al., 2006, 2011a, 2011b, 2012), which then candisturb the delicate balance of BuChE-regulated astroglial func-tional status. In addition, low CSF BuChE activity status and highApoE protein show a strong correlation with high Ab load in thebrain, high phosphorylated-tau and low Ab levels in CSF, lowcerebral glucose utilization, and poor cognitive performance(Darreh-Shori et al., 2011a). These observations are in close agree-ment with findings in a recent animal study which indicate thathaploinsufficiency of human ApoE reduces Ab deposition (Kimet al., 2011). In line with the current observations, CSF levels of IL-1b increases with rising levels of ApoE protein (Darreh-Shori et al.,2012) and AD CSF with moderate to low BuChE activity statuscontains approximately a 40%e80% higher level of IL-1b (Darreh-Shori et al., 2011b). This is essentially identical with the currentfinding that the BCHE-K carriers had geneedosage dependently40%e80% higher CSF IL-1b, and 27% higher TNF-a. Hence, the brainsof AD patients with high CSF BuChE activity status seem to be lessaffected by the ongoing pathological processes in the brain, or toelicit less inflammatory activation relative to patients with lowerCSF BuChE levels, in particular in carriers of Apo ε4, which areknown to have the most advanced AD pathology (Martins et al.,2006). Overall, these observations strongly indicate that failure toproperly adjust and maintain astroglial activities and functionsplays a key role for conversion of MCI to AD and further diseaseprogression in AD, and that BuChE might play an important regu-latory role in this context (Fig. 9).

The role of BuChE in AD might be even more complex as thepropensity of BuChE and AChE to attenuate or accelerate Abdepositions in the brain differ (Berson et al., 2008; Diamant et al.,2006; Podoly et al., 2008, 2009). Thus, synaptic AChE-S andBuChE-K variants promote Ab fibril formation, whereas the ‘read-through’ AChE-R and wild type BuChE variants counteract Abfibrillization (Berson et al., 2008; Diamant et al., 2006; Podoly et al.,2008, 2009). We have also shown that BCHE-K homozygosity isassociated with a strong reduction in the levels of AChE-S andBuChE in CSF, particularly among Apo ε4 carriers AD patients(Darreh-Shori et al., 2012). Indeed, in patients with AD or Lewybody dementia, a high AChE R/S ratio shows a highly significantassociation with superior performance in cognitive tests, andin vivo FDG- and PIB-PET findings (Darreh-Shori et al., 2004;Vijayaraghavan et al., 2013). We also found in this study that, likeBuChE, the CSF level of AChE-R but not AChE-S positively correlateswith CSF GFAP and S100B. Thus the propensity of BuChE and AChE-R to attenuate Ab deposition in the brain together with their ACh-hydrolyzing activity in the interstitial fluid might be involved.

In the current study, we also found a positive associationbetween the BuChE (and not AChE) activity and IL-1b in theperipheral circulation of the AD patients. In addition, the plasmaBuChE activity correlated directly with the CSF levels of GFAP andS100B. In contrast, although the plasma IL-1b correlated with theCSF IL-1b level, it showed no direct association with the CSF levelsof astroglial functional status (GFAP and S100B). This might reflecta time-shift in the sequence of events. Evidently, a cross-talk existsbetween peripheral circulation and central compartments (Laskeet al., 2013; Ofek and Soreq, 2013), and inflammatory cascades inAD have most likely the character of a chronic low-grade inflam-mation rather than an acute-phase systemic response. Thus, al-though speculative, the lack of a direct association between plasmaIL-1b and astroglial markers and the opposite relationship of BuChEand IL-1b in plasma compared with in CSF might reflect a time-shift

in the sequence of events. High plasma BuChE activity facilitateslow-grademaintenance of plasma IL-1b, which is transported to thebrain, signaling an early release of BuChE by astroglial cells. This, inturn, augments the initial effect of IL-1b, by hydrolyzing extra-synaptic ACh, which relieves its suppressive tune on the glial cells,and thereby facilitates and/or strengthens the functional status ofastroglial cells. This, in turn, leads to a more effective removal of thenoxious stimuli, leading to reduced levels of CSF IL-1b and TNF-a,and prevention of exaggerated inflammatory cascades.

Intriguingly, an increase in AChE-R and BuChE levels in plasmaseem to be related to fastest recovery of inflammatory responses toendotoxin exposure in healthy human volunteers (Ofek et al.,2007). A common denominator between the AChE-R variant andBuChE is that both enzymes are mainly soluble, though the synapticAChE-S variant (or the red blood cells containing the AChE-Hvariant) are destined to be membrane-anchored. This is an impor-tant distinction. Extrasynaptic ACh regulates the function ofa variety of nonneuronal nonexcitable cholinoceptive cells, such asendothelia, microglia, astrocytes, and lymphocytes in the nervoussystem and secondary lymphoid organs (Wessler et al., 1999). Mostof these cells are located very distant from cholinergic neuronalsynapses. Thus, the solubility and secretory nature of BuChE and theAChE-R variant might be a crucial factor for adjusting extrasynapticlevels of ACh because only the free enzymes are expected to reachthese distant sites (Ofek et al., 2007). For instance, Giaccobini, tostudy the function of BuChE, intracortically perfused rat brain witha selective BuChE inhibitor and found that the extracellularconcentration of ACh was increased 15-fold, from 5 nM to 75 nM, inthe animals (Giacobini, 2001).

To reconcile the overall pattern of the observations, we havehypothesized that a complex interplay exists between BuChE, AChE,and Ab peptides in AD brain, in which high ApoE protein levels playa major pathological influence (Fig. 9). Our previous findingssuggest the formation of a molecular complex, termed BuChE/AChE-Ab-ApoE complexes (BAbACs) (Darreh-Shori et al., 2011a), inwhich Ab regulates, in a concentration-dependent manner, the AChhydrolyzing activity of the enzymes in BAbACs. High levels of ApoEprotein, which occurs in AD patients in an Apo ε4 geneedosage-dependent manner, facilitates the formation and accumulation ofBAbACs (Darreh-Shori et al., 2011a). Because Ab is released into thesynaptic cleft and thereby into interstitial fluids in synchrony withsynaptic action potentials (Cirrito et al., 2005), it can be furtherspeculated if cholinergic neurons use Ab for regulating the activityof nonneuronal cholinoceptive cells, such as astrocytes andmicroglia. Thus, by releasing Ab, the activity of BuChE (in theBAbACs) can be adjusted, in turn leading to lessened extrasynapticsuppressive tuning of ACh-mediated regulation of surroundingcells. High levels of ApoE will, however, disturb this fine-tuning ofACh levels by stabilizing and/or prolonging the physical interactionof Ab and BuChE. This hypothesis provides a possible mechanismfor Apo ε4 being an AD risk factor in conjunction with the well-established selective cholinergic deficit in AD and the presence oflow systemic neuroinflammation in AD, and the robust associationof BuChE with the markers of astroglial immune responses in thecurrent study. However, it is important to acknowledge that addi-tional studies are needed to provide experimental support for thisproposed chain of events.

Nonetheless, the current observations highlight the potentialclinical importance of BuChE inhibition to counteract BuChE-derived depletion of ACh in areas in which the concentration ofAb is expected to be high (Mesulam and Geula, 1994).

High levels of S100B are observed in AD brains, mainly in thevicinity of the neurotic Ab plaques and tau pathologies (Mrak andGriffin, 2001). In this context, the positive correlation betweenS100B and cognitive performance is somewhat surprising because

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 15

the carriers of both the Apo ε4 and BCHE-K alleles had the lowestlevels of CSF S100B and worse cognitive status. However, ourfindings are consistent with the observations in a previous studythat reported higher S100B levels in the CSF of patients withmild tomoderate AD compared with controls or patients with advancedAD, and a positive correlation between CSF S100B and the MMSEscores (Peskind et al., 2001). Notably, we have also observed aninverse U-shape pattern for the longitudinal changes in CSF levelsof S100B and GFAP in a group of MCI patients during 3 years offollow-up (unpublished observations). It might thus be thatdecreased S100B in later stages of AD reflects dysfunctional astro-glial activation, with concomitant microglial activation andincreased release of the proinflammatory cytokines (Giulian et al.,1988; Koppal et al., 2001; Mrak and Griffin, 2001).

Another important observation in the current study was that themeasured CSF levels of complement factors, similar to the astroglialmarkers, showed positive correlation with the CSF BuChE activity,contrasting to the inverse association between BuChE and theproinflammatory cytokines in the CSF. Several complement factorsare produced locally in the brain, where they might act as signalingmolecules mediating uptake of Ab by glial cells (Veerhuis, 2011),and as a bridge between innate and adaptive immune responses(Carroll, 2004; Dunkelberger and Song, 2010; Gasque, 2004).Emerging evidence, however, suggests another unexpected role forthe complement system, encompassing synaptic pruning, elimi-nation, and/or remodeling in the CNS (Stephan et al., 2012), plau-sibly by tagging certain synapses and facilitating their phagocytosisby microglia (Stephan et al., 2012). Interestingly, CSF levels of thecomplement factors, C3 and H, showed positive correlation withcognition and FDG-PET. Thus the release of the complement factorsseems similar to GFAP and S100B, occur at an earlier stage than therelease of the proinflammatory cytokines, and might thereby bepart of an early protective mechanism against aberrant Ab accu-mulation in the brain. Once again we found that alteration of theexpression and release of BuChE by astrocytes might be crucial forpreventing an exaggerated response because we found that highmillimolar ACh-TNF-a depressed C3 expression but led to increasedexpression and release of BuChE in the culture medium by astro-cytes. In contrast, TNF-a alone or in combination with lowermicromolar concentrations of ACh resulted in reduced BuChEexpression but a strong induction of C3. Thus, a well-functioningextrasynaptic ACh-BuChE axis, and thereby a well-tuned comple-ment response might clear and/or reduce the noxious stimuli andAb deposits in the brain microenvironment (Alexander et al., 2008;Ramaglia and Baas, 2009), whereas a badly orchestrated systemmight lead to a late, and exaggerated response by excess microglialactivation. This, in turn, might damage the delicate neuronalnetworks, especially because neurons have proven to be extremelysensitive to complement-mediated damage (Singhrao et al., 2000).

This study has several limitations. There are many geneticvariants of BuChE that were not included in the genetic analyses inthe current study (La Du et al., 1990). For instance, dibucaine-(atypical variant) and fluoride-sensitive variants are named sobecause of the changes in the serum activity of the enzymes’ variantwith addition of sodium fluoride or dibucaine to samples (Bartelset al., 1992; Whittaker and Britten, 1981, 1988). Another J-variantof BuChE is also associated with approximately a 60% loss of theenzyme activity (La Du et al., 1990). We did not determine theseother BCHE genotypes among the current study population,therefore their possible effect on the observations could not beexcluded or specifically addressed. Nonetheless, as a reassuringmeasure, we included the actual ACh-hydrolyzing activity of BuChEin the analyses because Apo ε4 genotype and protein levels canmodulate the phenotypic display of BuChE variants (Darreh-Shoriet al., 2012). Thus, further genetic studies focused on these quite

common BCHE genotypes and their specific effect on the astroglialand proinflammatory markers are warranted.

An additional cautionary point is related to the ACh concentra-tions used in the experiments on cultured human brain astrocytes.Because of uncertainty regarding the actual in vivo ACh concen-tration, we used a wide range of ACh concentrations, up to 100mM.The larger concentration might be an exceedingly high concentra-tion of ACh, although McCaman et al. have reported a minimalestimate of 0.35 mM ACh in the cytoplasm of several cholinergicneurons (McCaman et al., 1973). It was also important to freshlyprepare the ACh solutions before the experiments.

Furthermore, although the in vitro results from astrocyteculture, in agreement with the CSF and plasma findings, did notindicate any changes in the levels of the AChE-S variant afterstimulation with TNF-a and/or ACh; this finding should be inter-preted with caution. First, because cell culture studies have theirown obvious limitation of representing an isolated closed system, incontrast to the in vivo multilevel cellular interactions. Second,because of discontinuation of commercial availability of the anti-AChE-R antibody we were not able to examine whether a shift inthe release of AChE-R had occurred. This is because the CSF analysesindicated that levels of the AChE-R, but not the AChE-S variant wererelated to the CSF astroglial markers. Third, like CSF and plasma, theastrocyte’s culture media analyses indicated that the activity ofAChE dominates. The current findings thus do not argue against thefact that the overall individual ACh-hydrolyzing status, and therebythe extra synaptic equilibrium levels, of ACh are controlled by thesummated activities of either cholinesterases in the peripheral orthe central compartments. Instead, our findings simply suggest thatmodulation in the activity status of astroglial cells might achievepreferentially by alteration in the levels of BuChE levels, rather thanAChE.

In conclusion, in this study, we provide evidence suggesting thatfunctional variability in BuChE activity, depending on allelic varia-tion in the BCHE gene, regulates the intrathecal astroglial biomarkerprofile and cytokines. Thus, reduced BuChE enzymatic activity,either as a result of genetic K variant protein or phenotypicmodulation by the ApoE protein, is associated with worse cognitiveperformance and in vivo pathological signs. The CSF pattern ofastroglial and proinflammatory biomarkers is suggested to reflectan inverse U-shape dynamic in the pathophysiological continuumof AD, inwhich activation of astroglial cells is initiated early, reachesa maximum of activity at the MCI stage, and thereafter declines asthe disease progresses with increased proinflammatory cytokinerelease (Fig. 9). Although difficult to mechanistically prove, ourfindings suggest that proper activation and maintenance of astro-glial function is an important counteracting force against evolutionof AD, and not necessarily one of its driving forces, at least notinitially.

Disclosure statement

The authors declare that they have no potential conflicts ofinterest.

This study and the primary clinical studies were approved by theEthics Committee of Karolinska University Hospital Huddinge, andthe Faculty of Medicine and Radiation Hazard Ethics Committee ofUppsala University Hospital, Uppsala, Sweden. This study wasconducted according to the Declaration of Helsinki and subsequentrevisions. Informed consent was obtained from each patient or theresponsible caregivers.

The animal experiments were approved by the local ethicalcommittee for animal experimentation (Stockholms Norra Djur-försöksetiska Nämnd).

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e1716

Acknowledgements

This research was supported by The Swedish Medical ResearchCouncil (project no. 05817), the regional agreement on medicaltraining and clinical research (ALF) between Stockholm CountyCouncil and the Karolinska Institute, The Strategic ResearchProgram in Neuroscience at Karolinska Institutet, Karolinska Insti-tute Foundations, Loo & Hans Osterman Foundation, DementiaFoundation (Demensfonden), Magnus Bergvalls Foundation, OlleEngkvist Byggmästare Foundation, Åke Wibergs Foundation, OldServants Foundation, The Swedish Alzheimer Foundation, BrainFoundation, Gun and Bertil Stohnes Foundation, Åhlen Foundation,Foundation for Ragnhild and Einar Lundströms Memory, LSHB-CT-2005-512146, and the Swedish Brain Power.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.neurobiolaging.2013.04.027.

References

Alexander, J.J., Anderson, A.J., Barnum, S.R., Stevens, B., Tenner, A.J., 2008. Thecomplement cascade: Yin-Yang in neuroinflammationeneuro-protection and-degeneration. J. Neurochem. 107, 1169e1187.

Bartels, C.F., Jensen, F.S., Lockridge, O., van der Spek, A.F., Rubinstein, H.M.,Lubrano, T., La Du, B.N., 1992. DNA mutation associated with the humanbutyrylcholinesterase K-variant and its linkage to the atypical variant mutationand other polymorphic sites. Am. J. Hum. Genet. 50, 1086e1103.

Beach, T.G., McGeer, E.G., 1988. Lamina-specific arrangement of astrocytic gliosisand senile plaques in Alzheimer’s disease visual cortex. Brain Res. 463, 357e361.

Berson, A., Knobloch, M., Hanan, M., Diamant, S., Sharoni, M., Schuppli, D.,Geyer, B.C., Ravid, R., Mor, T.S., Nitsch, R.M., Soreq, H., 2008. Changes in read-through acetylcholinesterase expression modulate amyloid-beta pathology.Brain 131, 109e119.

Carnevale, D., De Simone, R., Minghetti, L., 2007. Microglia-neuron interaction ininflammatory and degenerative diseases: role of cholinergic and noradrenergicsystems. CNS Neurol. Disord. Drug Targets 6, 388e397.

Carroll, M.C., 2004. The complement system in regulation of adaptive immunity.Nat. Immunol. 5, 981e986.

Carter, S.F., Scholl, M., Almkvist, O., Wall, A., Engler, H., Langstrom, B., Nordberg, A.,2012. Evidence for astrocytosis in prodromal Alzheimer disease provided by11C-deuterium-l-deprenyl: a multitracer PET paradigm combining 11C-Pitts-burgh Compound B and 18F-FDG. J. Nucl. Med. 53, 37e46.

Cirrito, J.R., Yamada, K.A., Finn, M.B., Sloviter, R.S., Bales, K.R., May, P.C.,Schoepp, D.D., Paul, S.M., Mennerick, S., Holtzman, D.M., 2005. Synaptic activityregulates interstitial fluid amyloid-beta levels in vivo. Neuron 48, 913e922.

Darreh-Shori, T., Brimijoin, S., Kadir, A., Almkvist, O., Nordberg, A., 2006. DifferentialCSF butyrylcholinesterase levels in Alzheimer’s disease patients with the ApoEepsilon4 allele, in relation to cognitive function and cerebral glucose metabo-lism. Neurobiol. Dis. 24, 326e333.

Darreh-Shori, T., Forsberg, A., Modiri, N., Andreasen, N., Blennow, K., Kamil, C.,Ahmed, H., Almkvist, O., Langstrom, B., Nordberg, A., 2011a. Differential levels ofapolipoprotein E and butyrylcholinesterase show strong association withpathological signs of Alzheimer’s disease in the brain in vivo. Neurobiol. Aging32, 2320.e15e2320.e32. http://dx.doi.org/10.1016/j.neurobiolaging.2010.04.028.

Darreh-Shori, T.,Hellstrom-Lindahl, E., Flores-Flores,C.,Guan,Z.Z., Soreq,H.,Nordberg,A.,2004. Long-lasting acetylcholinesterase splice variations in anticholinesterase-treated Alzheimer’s disease patients. J. Neurochem. 88, 1102e1113.

Darreh-Shori, T., Kadir, A., Almkvist, O., Grut, M., Wall, A., Blomquist, G., Eriksson, B.,Langstrom, B., Nordberg, A., 2008. Inhibition of acetylcholinesterase in CSFversus brain assessed by 11C-PMP PET in AD patients treated with galantamine.Neurobiol. Aging 29, 168e184.

Darreh-Shori, T., Modiri, N., Blennow, K., Baza, S., Kamil, C., Ahmed, H.,Andreasen, N., Nordberg, A., 2011b. The apolipoprotein E epsilon4 allele playspathological roles in AD through high protein expression and interaction withbutyrylcholinesterase. Neurobiol. Aging 32, 1236e1248.

Darreh-Shori, T., Siawesh, M., Mousavi, M., Andreasen, N., Nordberg, A., 2012.Apolipoprotein epsilon4 modulates phenotype of butyrylcholinesterase in CSFof patients with Alzheimer’s disease. J. Alzheimers Dis. 28, 443e458.

Darvesh, S., Hopkins, D.A., Geula, C., 2003. Neurobiology of butyrylcholinesterase.Nat. Rev. Neurosci. 4, 131e138.

Diamant, S., Podoly, E., Friedler, A., Ligumsky, H., Livnah, O., Soreq, H., 2006.Butyrylcholinesterase attenuates amyloid fibril formation in vitro. Proc. Natl.Acad. Sci. U.S.A. 103, 8628e8633.

Di Bona, D., Plaia, A., Vasto, S., Cavallone, L., Lescai, F., Franceschi, C., Licastro, F.,Colonna-Romano, G., Lio, D., Candore, G., Caruso, C., 2008. Association betweenthe interleukin-1beta polymorphisms and Alzheimer’s disease: a systematicreview and meta-analysis. Brain Res. Rev. 59, 155e163.

Dunkelberger, J.R., Song, W.C., 2010. Complement and its role in innate and adaptiveimmune responses. Cell Res. 20, 34e50.

Ferris, S., Nordberg, A., Soininen, H., Darreh-Shori, T., Lane, R., 2009. Progressionfrom mild cognitive impairment to Alzheimer’s disease: effects of sex, butyr-ylcholinesterase genotype, and rivastigmine treatment. Pharmacogenet. Geno-mics 19, 635e646.

Gasque, P., 2004. Complement: a unique innate immune sensor for danger signals.Mol. Immunol. 41, 1089e1098.

Geula, C., Greenberg, B.D., Mesulam, M.M., 1994. Cholinesterase activity in theplaques, tangles and angiopathy of Alzheimer’s disease does not emanate fromamyloid. Brain Res. 644, 327e330.

Giacobini, E., 2001. Selective inhibitors of butyrylcholinesterase: a valid alternativefor therapy of Alzheimer’s disease? Drugs Aging 18, 891e898.

Giulian, D., Young, D.G., Woodward, J., Brown, D.C., Lachman, L.B., 1988. Inter-leukin-1 is an astroglial growth factor in the developing brain. J. Neurosci. 8,709e714.

Guillozet, A.L., Smiley, J.F., Mash, D.C., Mesulam, M.M., 1997. Butyrylcholinesterase inthe life cycle of amyloid plaques. Ann. Neurol. 42, 909e918.

Henningsson, A.J., Ernerudh, J., Sandholm, K., Carlsson, S.A., Granlund, H.,Jansson, C., Nyman, D., Forsberg, P., Nilsson Ekdahl, K., 2007. Complementactivation in Lyme neuroborreliosiseincreased levels of C1q and C3a in cere-brospinal fluid indicate complement activation in the CNS. J. Neuroimmunol.183, 200e207.

Hrabovska, A., Bernard, V., Krejci, E., 2010. A novel system for the efficient gener-ation of antibodies following immunization of unique knockout mouse strains.PloS One 5, e12892.

Kadir, A., Marutle, A., Gonzalez, D., Scholl, M., Almkvist, O., Mousavi, M., Mustafiz, T.,Darreh-Shori, T., Nennesmo, I., Nordberg, A., 2011. Positron emission tomog-raphy imaging and clinical progression in relation to molecular pathology in thefirst Pittsburgh Compound B positron emission tomography patient with Alz-heimer’s disease. Brain 134, 301e317.

Kashon, M.L., Ross, G.W., O’Callaghan, J.P., Miller, D.B., Petrovitch, H., Burchfiel, C.M.,Sharp, D.S., Markesbery, W.R., Davis, D.G., Hardman, J., Nelson, J., White, L.R.,2004. Associations of cortical astrogliosis with cognitive performance anddementia status. J. Alzheimers Dis. 6, 595e604. discussion, 73e81.

Kim, J., Jiang, H., Park, S., Eltorai, A.E., Stewart, F.R., Yoon, H., Basak, J.M., Finn, M.B.,Holtzman, D.M., 2011. Haploinsufficiency of human APOE reduces amyloiddeposition in a mouse model of amyloid-beta amyloidosis. J. Neurosci. 31,18007e18012.

Koppal, T., Lam, A.G., Guo, L., Van Eldik, L.J., 2001. S100B proteins that lack one orboth cysteine residues can induce inflammatory responses in astrocytes andmicroglia. Neurochem. Int. 39, 401e407.

La Du, B.N., Bartels, C.F., Nogueira, C.P., Hajra, A., Lightstone, H., Van der Spek, A.,Lockridge, O., 1990. Phenotypic and molecular biological analysis of humanbutyrylcholinesterase variants. Clin. Biochem. 23, 423e431.

Lane, R., Feldman, H.H., Meyer, J., He, Y., Ferris, S.H., Nordberg, A., Darreh-Shori, T.,Soininen, H., Pirttila, T., Farlow, M.R., Sfikas, N., Ballard, C., Greig, N.H., 2008.Synergistic effect of apolipoprotein E epsilon4 and butyrylcholinesterase K-variant on progression from mild cognitive impairment to Alzheimer’s disease.Pharmacogenet. Genomics 18, 289e298.

Laske, C., Schmohl, M., Leyhe, T., Stransky, E., Maetzler, W., Berg, D., Fallgatter, A.J.,Joos, T., Dietzsch, J., 2013. Immune profiling in blood identifies sTNF-R1 per-forming comparably well as biomarker panels for classification of Alzheimer’sdisease patients. J. Alzheimers Dis. 34, 367e375.

Maragakis, N.J., Rothstein, J.D., 2006. Mechanisms of disease: astrocytes in neuro-degenerative disease. Nat. Clin. Pract. Neurol. 2, 679e689.

Martins, I.J., Hone, E., Foster, J.K., Sunram-Lea, S.I., Gnjec, A., Fuller, S.J., Nolan, D.,Gandy, S.E., Martins, R.N., 2006. Apolipoprotein E, cholesterol metabolism,diabetes, and the convergence of risk factors for Alzheimer’s disease andcardiovascular disease. Mol. Psychiatry 11, 721e736.

McCaman, R.E., Weinreich, D., Borys, H., 1973. Endogenous levels of acetylcholineand choline in individual neurons of Aplysia. J. Neurochem. 21, 473e476.

Mesulam, M.M., Geula, C., 1994. Butyrylcholinesterase reactivity differentiates theamyloid plaques of aging from those of dementia. Ann. Neurol. 36, 722e727.

Mesulam, M.M., Guillozet, A., Shaw, P., Levey, A., Duysen, E.G., Lockridge, O., 2002.Acetylcholinesterase knockouts establish central cholinergic pathways and canuse butyrylcholinesterase to hydrolyze acetylcholine. Neuroscience 110,627e639.

Mrak, R.E., Griffin, W.S., 2001. The role of activated astrocytes and of the neuro-trophic cytokine S100B in the pathogenesis of Alzheimer’s disease. Neurobiol.Aging 22, 915e922.

O’Brien, K.K., Saxby, B.K., Ballard, C.G., Grace, J., Harrington, F., Ford, G.A., O’Brien, J.T.,Swan, A.G., Fairbairn, A.F., Wesnes, K., del Ser, T., Edwardson, J.A., Morris, C.M.,McKeith, I.G., 2003. Regulation of attention and response to therapy indementia by butyrylcholinesterase. Pharmacogenetics 13, 231e239.

Ofek, K., Krabbe, K.S., Evron, T., Debecco, M., Nielsen, A.R., Brunnsgaad, H.,Yirmiya, R., Soreq, H., Pedersen, B.K., 2007. Cholinergic status modulations inhuman volunteers under acute inflammation. J. Mol. Med. (Berl.) 85, 1239e1251.

Ofek, K., Soreq, H., 2013. Cholinergic involvement and manipulation approaches inmultiple system disorders. Chem. Biol. Interact. 203, 113e119.

T. Darreh-Shori et al. / Neurobiology of Aging xxx (2013) 1e17 17

Payao, S.L., Goncalves, G.M., de Labio, R.W., Horiguchi, L., Mizumoto, I.,Rasmussen, L.T., de Souza Pinhel, M.A., Silva Souza, D.R., Bechara, M.D., Chen, E.,Mazzotti, D.R., Ferreira Bertolucci, P.H., Cardoso Smith Mde, A., 2012. Associationof interleukin 1beta polymorphisms and haplotypes with Alzheimer’s disease.J. Neuroimmunol. 247, 59e62.

Perry, E., McKeith, I., Ballard, C., 2003. Butyrylcholinesterase and progression ofcognitive deficits in dementia with Lewy bodies. Neurology 60, 1852e1853.

Perry, E.K., Tomlinson, B.E., Blessed, G., Bergmann, K., Gibson, P.H., Perry, R.H., 1978.Correlation of cholinergic abnormalities with senile plaques and mental testscores in senile dementia. Br. Med. J. 2, 1457e1459.

Peskind, E.R., Griffin, W.S., Akama, K.T., Raskind, M.A., Van Eldik, L.J., 2001. Cere-brospinal fluid S100B is elevated in the earlier stages of Alzheimer’s disease.Neurochem. Int. 39, 409e413.

Piehl, F., Swanberg, M., Lidman, O., 2007. The axon reaction: identifying the genesthat make a difference. Physiol. Behav. 92, 67e74.

Podoly, E., Bruck, T., Diamant, S., Melamed-Book, N., Weiss, A., Huang, Y., Livnah, O.,Langermann, S., Wilgus, H., Soreq, H., 2008. Human recombinant butyr-ylcholinesterase purified from the milk of transgenic goats interacts with beta-amyloid fibrils and suppresses their formation in vitro. Neurodegener. Dis. 5,232e236.

Podoly, E., Shalev, D.E., Shenhar-Tsarfaty, S., Bennett, E.R., Ben Assayag, E., Wilgus, H.,Livnah, O., Soreq, H., 2009. The butyrylcholinesterase K variant confers struc-turally derived risks for Alzheimer pathology. J. Biol. Chem. 284, 17170e17179.

Ramaglia, V., Baas, F., 2009. Innate immunity in the nervous system. Prog. Brain Res.175, 95e123.

Rosengren, L.E., Wikkelso, C., Hagberg, L., 1994. A sensitive ELISA for glial fibrillaryacidic protein: application in CSF of adults. J. Neurosci. Methods 51, 197e204.

Shytle, R.D., Mori, T., Townsend, K., Vendrame, M., Sun, N., Zeng, J., Ehrhart, J.,Silver, A.A., Sanberg, P.R., Tan, J., 2004. Cholinergic modulation of microglialactivation by alpha 7 nicotinic receptors. J. Neurochem. 89, 337e343.

Singhrao, S.K., Neal, J.W., Rushmere, N.K., Morgan, B.P., Gasque, P., 2000. Sponta-neous classical pathway activation and deficiency of membrane regulatorsrender human neurons susceptible to complement lysis. Am. J. Pathol. 157,905e918.

Stedman, E., Easson, L.H., 1932. Choline-esterase. An enzyme present in the blood-serum of the horse. Biochem. J. 26, 2056e2066.

Steiner, J., Bernstein, H.G., Bielau, H., Berndt, A., Brisch, R., Mawrin, C., Keilhoff, G.,Bogerts, B., 2007. Evidence for a wide extra-astrocytic distribution of S100B inhuman brain. BMC Neurosci. 8, 2.

Stephan, A.H., Barres, B.A., Stevens, B., 2012. The complement system: an unex-pected role in synaptic pruning during development and disease. Annu. Rev.Neurosci. 35, 369e389.

Strom, M., Al Nimer, F., Lindblom, R., Nyengaard, J.R., Piehl, F., 2012. Naturallyoccurring genetic variability in expression of Gsta4 is associated with differ-ential survival of axotomized rat motoneurons. Neuromolecular Med. 14, 15e29.

Veerhuis, R., 2011. Histological and direct evidence for the role of complement inthe neuroinflammation of AD. Curr. Alzheimer Res. 8, 34e58.

Vijayaraghavan, S., Maetzler,W., Reimold,M., Lithner, C.U., Liepelt, I., Berg, D., Darreh-Shori, T., 2013. High apolipoprotein E in cerebrospinal fluid of patients with Lewybody disorders is associated with dementia. Alzheimers Dement., in press.

Wang, H., Yu, M., Ochani, M., Amella, C.A., Tanovic, M., Susarla, S., Li, J.H., Yang, H.,Ulloa, L., Al-Abed, Y., Czura, C.J., Tracey, K.J., 2003. Nicotinic acetylcholinereceptor alpha7 subunit is an essential regulator of inflammation. Nature 421,384e388.

Wang, Y., Hancock, A.M., Bradner, J., Chung, K.A., Quinn, J.F., Peskind, E.R., Galasko, D.,Jankovic, J., Zabetian, C.P., Kim, H.M., Leverenz, J.B., Montine, T.J., Ginghina, C.,Edwards, K.L., Snapinn, K.W., Goldstein, D.S., Shi, M., Zhang, J., 2011. Complement3 and factor h in human cerebrospinal fluid in Parkinson’s disease, Alzheimer’sdisease, and multiple-system atrophy. Am. J. Pathol. 178, 1509e1516.

Wessler, I., Kirkpatrick, C.J., Racke, K., 1999. The cholinergic ‘pitfall’: acetylcholine,a universal cell molecule in biological systems, including humans. Clin. Exp.Pharmacol. Physiol. 26, 198e205.

Wharton, S.B., O’Callaghan, J.P., Savva, G.M., Nicoll, J.A., Matthews, F., Simpson, J.E.,Forster, G., Shaw, P.J., Brayne, C., Ince, P.G., 2009. Population variation in glialfibrillary acidic protein levels in brain ageing: relationship to Alzheimer-typepathology and dementia. Dement. Geriatr. Cogn. Disord. 27, 465e473.

Whittaker, M., Britten, J.J., 1981. A comparison of some methods of phenotyping theplasma cholinesterase variants using benzoylcholine as substrate. Ann. Clin.Biochem. 18, 9e14.

Whittaker, M., Britten, J.J., 1988. Recognition of two new phenotypes segregating theE1k allele for plasma cholinesterase. Hum. Hered. 38, 233e239.

Yu, W.F., Guan, Z.Z., Bogdanovic, N., Nordberg, A., 2005. High selective expression ofalpha7 nicotinic receptors on astrocytes in the brains of patients with sporadicAlzheimer’s disease and patients carrying Swedish APP 670/671 mutation:a possible association with neuritic plaques. Exp. Neurol. 192, 215e225.

Yuan, H., Xia, Q., Ge, P., Wu, S., 2013. Genetic polymorphism of interleukin 1beta-511C/T and susceptibility to sporadic Alzheimer’s disease: a meta-analysis. Mol.Biol. Rep. 40, 1827e1834.