Molecular Brain Research 93 (2001) 70–80www.elsevier.com/ locate /bres

Research report

Characterisation of gene expression changes following permanentMCAO in the rat using subtractive hybridisation

a , b b c b*Stewart Bates , Simon J. Read , David C. Harrison , Simon Topp , Rachel Morrow ,a d e b aDavina Gale , Paul Murdock , Frank C. Barone , Andrew A. Parsons , Israel S. Gloger

aDepartment of Molecular Biology, GlaxoSmithKline Pharmaceuticals, New Frontiers Science Park, Harlow, Essex, UKbDepartment of Neuroscience, GlaxoSmithKline Pharmaceuticals, New Frontiers Science Park, Harlow, Essex, UK

cDepartment of Bioinformatics, GlaxoSmithKline Pharmaceuticals, New Frontiers Science Park, Harlow, Essex, UKdDepartment of Gene Expression Sciences, GlaxoSmithKline Pharmaceuticals, New Frontiers Science Park, Harlow, Essex, UK

eDepartment of Cardiovascular Pharmacology, GlaxoSmithKline Pharmaceuticals, King of Prussia, PA, USA

Accepted 26 June 2001

Abstract

Failure of several putative neuroprotectants in large multicentred clinical trials has re-focussed attention on the predictability ofpre-clinical animal models of stroke. Model characterisation and relationship to heterogeneous patient sub-groups remains of paramountimportance. Information gained from magnetic resonance imaging (MRI) signatures indicates that the Zea Longa model of rat middlecerebral artery occlusion may be more representative of slowly evolving infarcts. Understanding the molecular changes over several hoursfollowing cerebral ischaemia will allow detailed characterisation of the adaptive response to brain injury. Using a fully characterisedmodel of Zea Longa middle cerebral artery occlusion we have used the representational difference analysis (RDA) subtractivehybridisation method to identify transcripts that accumulate in the ischaemic cortex. Along with a number of established ischaemia-induced gene products (including MCP-1, TIMP-1, hsp 70) we were also able to identify nine genes which have not previously beenshown to accumulate following focal ischaemia (including SOCS-3, GADD45g, Xin). 2001 Elsevier Science B.V. All rights reserved.

Theme: Disorders of the nervous system

Topic: Ischemia

Keywords: Stroke; Ischemia; Gene; Differential expression

1. Introduction end-points [34], whilst genomic characterisation remainslimited.

Human stroke is a polygenic condition with large unmet Animal models have shown that cerebral ischaemia is amedical need. The failure of several putative neuroprotec- powerful stimulus to gene expression [20], and changes intive agents in recent large multi-centred clinical trials gene expression accompany many of the steps in the[6,27] has led to critical re-examination of the predictabili- development of stroke [2]. The genetic contribution toty of many pre-clinical models of ischemia. Heterogeneity developing cerebral ischemic injury or protection has beenof the human stroke population and a multitude of well extensively studied in genetically heterogenous animaldefined animal models of ischemia, have led to attempts to models. Detection of differential expression has employedrefine model choice dependent on patient subgroups [7,34]. several techniques including differential display [45],Comparisons between preclinical models and human stroke subtractive hybridisation [49] and DNA microarrays [39].subpopulations have focused on imaging and physiological Several models of experimental middle cerebral artery

(MCAO) in rodents [9,11], have been well characterised interms of infarct pathophysiology and neuroprotective*Corresponding author. Tel.: 144-1279-627-963; fax: 144-1279-627-strategies [21,30,44]. Magnetic resonance imaging has266.

E-mail address: stewart a bates@sbphrd.com (S. Bates). documented variable rates of temporal and spatial pro-] ]

0169-328X/01/$ – see front matter 2001 Elsevier Science B.V. All rights reserved.PI I : S0169-328X( 01 )00186-3

S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80 71

gression of the infarct, attributable to the type of model generate dscDNA. Representations for both tester andused [1]. Within models cellular milieu varies across the driver were generated by restriction of the dscDNA withinfarct leading to the development of multiple molecular DpnII, ligation to oligonucleotides and PCR amplification.penumbras where gene expression varies spatially and Clones induced by permanent MCAO were enriched bytemporally (for review see Ref. [37]). Clearly, physiologi- PCR for re-annealed tester hybrids following solutioncal differences do exist between models, and therefore hybridisation of denatured tester representation with anauthors have attempted to align infarct progression of excess of denatured driver representation. After two roundspreclinical models with individual patient sub-groups [34]. of subtraction, the library was DpnII restricted and ligated

In the current study, we use representational difference into the BamHI site of pcDNA3.1. A total of 480 clonesanalysis (RDA), a PCR-based subtractive hybridisation were grown up overnight and DNA extracted using 96-protocol [15,29], to assess differential expression in the well format miniprep analysis. Clones were subjected toLonga et al. [30] model of permanent middle cerebral automated sequence analysis with vector primers, con-artery occlusion. Analysis of the subtracted library iden- catamerised clones separated into individual DpnII de-tified many gene products that have previously been shown lineated fragments and identities confirmed by Blastto accumulate following MCAO, as well as a number of analysis of the Genbank/EMBL and GenseqN databases.gene products that had not previously been linked to

cerebral ischaemia. Using SYBR Green RT-PCR and 2.4. RT-PCR and SYBR Green RT-PCRTaqMan RT-PCR, 17 gene products were confirmed to

accumulate following MCAO, including nine newly de- cDNA templates for both RT-PCR and SYBR Greenscribed MCAO-induced genes. PCR were generated from 1 mg of poly A RNA (sham and

MCAO) using the SMARTE RACE kit (Clontech) accord-ing to manufacturer’s instructions.

RT-PCR was performed using AmpliTaq Gold (Applied2. Materials and methods Biosystems Inc. (ABI)) under standard conditions (10 mM

Tris pH 8.3, 50 mM KCl, 1.5 mM MgCl , 200 mM dNTPs,22.1. Middle cerebral artery occlusion (MCAO) 5 U AmpliTaq gold, 400 nM of each gene specific primerand 5 ng of cDNA) using a 10-min hot start at 958C

Middle cerebral artery occlusion was carried out accord- followed by 20, 25 or 30 cycles of 958C for 1 min, 598Cing to the method of Zea-Longa et al. [30]. Male Sprague– for 1 min, 728C for 1.5 min. PCR reactions were analysedDawley rats weighing 300–350 g were anaesthetised with by 2% agarose gel electrophoresis, and visualised usinghalothane. The left middle cerebral artery (MCA) was ethidium bromide fluorescence.

occluded by an intraluminal filament at the origin from the SYBR Green PCR was performed in accordance withCircle of Willis. The rats (four per group) were euthanased manufacturer’s guidance (Applied Biosystems Inc.). PCR

3, 6, 12 or 24 h after surgery by halothane overdose. reactions were set up in 13 SYBR Green PCR MasterParallel groups of rats received sham surgery in which an mix, 400 nM each gene-specific primer, 5 ng template,identical procedure was followed but omitting the insertion and, following initial incubations at 508C for 2 min andof the filament. A further group of naive rats received no 958C for 10 min, subjected to 40 cycles of PCR (958C for

treatment at all prior to euthanasia. 15 s, 598C for 1 min) using an ABI Prism 7700 sequencedetector. The generation of dsPCR product was monitored

at each cycle by increased SYBR Green fluorescence2.2. RNA extractioncompared to an inert reference control (ROX). The thres-

hold for detection of SYBR Green fluorescence was set toThe ipsilateral cortices from two MCAO and two sham-DRn50.5, and the cycle number at which the thresholdoperated animals were used for RNA extraction. PolyA1was achieved (Ct) used as a measure of relative abun-RNA was isolated directly from the tissues using thedance. For each gene, duplicate sham and MCAO samplesFastTrack 2.0 kit (Invitrogen) according to the manufactur-were run and the mean difference in cycle number requireder’s protocol. The eluted RNA was ethanol precipitatedto achieve the Ct value between MCAO and sham samplesand dissolved in RNase free water for use in the subtrac-used as a measure of change in gene expression intion.response to MCAO. Assuming a 2-fold increase in PCRproduct level with each exponential PCR cycle, the

2.3. Representational difference analysis (RDA) approximate fold difference between samples was calcu-(mean Ct MCAO2mean Ct sham)lated (2 ). For consistency, fold

The subtractive hybridisation protocol was essentially as differences were corrected to GAPDH levels (assumingdescribed previously [15]. Briefly, 5 mg of poly A1 GAPDH to remain unchanged), and variation between

enriched RNA from both tester (rats subjected to perma- duplicate SYBR Green PCRs expressed as standard error.nent MCAO) and driver (sham-operated rats) was used to The sequences of the primer pairs used are:

72 S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80

MCP-1: AAC TAC AGT CTT CTA TAA AAT TGC, 3. ResultsATC TAC ATC TTG CAT TTA AGG; MIP3a: TCC TGGAGC TGA GAA TGG, ATC TTT TCG ACT TCA GGT 3.1. Generation and analysis of the subtracted libraryG; hsp27: GGT GCC TCG GTT TCC C, TGA GAC ACCGCT GCT GAG; hsp70: CCT CGT ACA CCT GGA TCA In order to identify gene products that are up-regulatedGCA, AAC AAG CAG ACG CAG ACC TT; COX-2: following pMCAO, RNA extracted from the cortex ofCAA TGC GGT TCT GAT ACT GGA A, AAC ACC MCAO and sham-treated rats was subjected to subtractiveTGA GCG GTT ACC ACT T; TGF-b1: GGC ACC ATC hybridisation using the representational difference analysisCAT GAC ATG AA, CAG GTG TTG AGC CCT TTC (RDA) method. After two rounds of hybridisation, aCA; TIMP-1: ATG AGA TCA AGA TGA CTA AGA subtracted library was generated and cloned intoTGC, AAG CAA TGA CTG TCA CTC TCC; ARC: TTC pcDNA3.1 for automated sequence analysis. The quality ofTGC AGT ACA GTG AGG G, CGT CCA CAT ACA the subtracted library was assessed by virtual northernGTG TCT GG; LIF: AAA GAA GCC TTC CAA AGG analysis using the subtracted library as a probe, confirmingAAG AA, CCG CAC TAA TGA CTT GCT TGT ATG; that the library was enriched for clones that were moreSOCS-3: CTG GTA CTG AGC CGA CCT CTC T, GTG abundant in the MCAO cDNA than in the sham cDNAACC GTT GAC AGT CTT CCG; VGF: ATC CAA GGC (data not shown). Sequenced clones were grouped intoGTA CCA AAG TCT AAG, GGA ACC GCC CAG GAA contigs and identities discerned by Blast alignment (TableTG; CD44: CGT TAC GCA GGT GTA TTC C, CGT 1).GTC CTT CTA TGA ACC C; CD14: ATC TCA GTT Using this approach, the library was shown to consistACA ACA GGC TGG, CTA GGC AAG GAC GTT GG; primarily of known genes (either identical to previouslyCD81: ACG CTC AAC TGT TGT GG, ACA ATG GCT identified rat genes or orthologues of previously describedGCA ATT CC; osteoactivin: TGG CCA TGT TTG TCA human/mouse genes), although a small number of novelCC, GAG AGT GAA GAC TTA GAG CAT CC; clones were also identified. Although these clones areGADD45g: TTC TGC AGT ACA GTG AGG G, CGT novel in the sense that they show no homology to clones inCCA CAT ACA GTG TCT GG; Xin: AGG AAG GTG any currently available database, we cannot at this pointAGA CTT GTT GG, AAG ATG AGA TTA GAA ATC rule out the possibility that these clones represent distalATG CAG; GAPDH (Clontech): ACC ACA GTC CAT UTR regions of known genes. For this reason, we continueGCC ATC AC, TCC ACC ACC CTG TTG CTG TA; to work on these and now restrict our follow-up analysis tob-actin: AAC CGT GAA AAG ATG ACC, CTG ATC the known genes identified in this work.CAC ATC TGC TGG Encouragingly, the subtracted library contained a num-

ber of genes with well established roles in the ischaemicresponse to MCAO (Table 1 and references therein), whichis again indicative that the subtracted library is indeed

2.5. TaqMan RT-PCR enriched for differentially expressed transcripts. In additionto the genes with known MCAO regulated expression, we

TaqMan RT-PCR was performed essentially as de- identified a number of biologically interesting genes thatscribed previously [13,31]. Briefly, total RNA was ex- have not been previously linked with the stroke response.tracted using Trizol reagent from the left and right cortices In order to confirm the differential expression of these

¨of MCAO, sham-operated and naıve rats (four rats in each genes, as well as to put their expression changes in contextgroup). First strand cDNA synthesis was carried out in with the known response to MCAO, a selection of genes,triplicate by oligo(dT) priming from 1 mg of each RNA with both known and newly identified links to MCAO,

pool, and used for TaqMan PCR using an ABI 7700 were chosen for PCR confirmation of their accumulationsequence detector (ABI). PCR conditions were as follows: following MCAO. Seventeen genes were chosen, eight508C for 2 min, 958C for 10 min followed by 40 cycles of with known induction following MCAO (MCP-1, MIP3a,958C for 15 s, 608C for 1 min. PCR primers described hsp27, hsp 70, COX-2, TGF-b1, TIMP-1 and ARC) andabove were employed, with the exception of MCP-1 nine for which increased expression was first identified in(TGCATCAACCCTAAGGACTTCA, GGAAAAGAGA- this study (LIF, SOCS-3, VGF, CD44, CD14, CD81,GTGGATGCATTAGC) and TaqMan probes synthesised: osteoactivin, GADD45g and Xin).(LIF: CCCAGGAGCTGGCAGCCCA; SOCS-3:CAACGTGGCTACCCTCCAGCATCTTT; MCP-1:CACCTTTGAATGAACTTGACCCATAAATCTG; VGF: 3.2. Confirming differential expression using semi-CTTCCCCAAGGTGCGTCGGCT). Standard curves were quantitative PCRplotted for all primer /probe sets using genomic DNA

¨standards, absolute levels of mRNA for naıve, sham- In order to confirm that clones identified by RDA wereoperated and MCAO-treated rats deduced and fold in- indeed differentially expressed in the starting material,

¨creases versus naıve rats plotted. semi-quantitative RT-PCR was employed. Initially three

S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80 73

Table 1Listing of genes identified by RDA as candidates for accumulation following pMCAO

Gene name Accession no. [ of frags Reference

Chemokines MIP3a U90447 22 [45]MCP-1 M57441 9 [50]

Cytokines /growth factors Inhibin b-A M37482 6 [26]TGF-b1 X52498 2 [51]LIF AB10275 2 [42]

Cytokine signalling COX-2 S67722 13 [35]SOCS-3 AJ249240 11GADD45g AB020978 3IL-1 receptor type II Z22812 2 [47]IL-3 receptor AJ000555 1 [54]iNOS D83661 1 [16]

Heat shock proteins hsp27 M86389 23 [55]hsp70 L16764 12 [55]

Cell surface molecules CD44 U46957 20CD14 AF087943 14Osteoactivin AF184983 4CD81 U19894 2

Protease inhibitors TIMP-1 L31883 34 [48]PAI-1 M24067 2 [53]

Structural ARC U19866 1 [24]

Others Xin orthologue (m) AF051945 10VGF M60525 7p21WAF-1 U24174 2 [36]

Novels 20 novels n /a 37

The subtracted library was cloned, sequenced, subjected to Blast analysis and the identities (gene names and accession numbers) of the genes present listedand grouped according to function. Also included are the number of times fragments for individual genes ([ of frags) were found in the course ofsequencing the subtracted library. Where genes have previously been shown to be induced following MCAO this is indicated, (reference) and anappropriate publication cited.

genes were analysed, MCP-1, MIP3a and hsp27, as well as Semi-quantitative RT-PCR is, however, time consumingGAPDH as a control for a gene which should remain and difficult to quantify, so we turned to the use of SYBR

relatively stable. Gene specific primers were synthesised Green dye and an ABI Prism 7700 sequence detector,for the four genes and these were shown to amplify correct which we have shown in detail elsewhere (Gale et al.,size products from the appropriate clones (data not shown). 2001, manuscript in preparation) to be ideal to overcome

RT-PCR was performed with the gene-specific primers these problems. SYBR Green PCR allows the quantifi-using the sham and MCAO cDNAs as templates. To able, real-time amplification of PCR products, so theensure that amplification was being monitored during the differential expression of MIP3a, MCP-1 and hsp27 wereexponential amplification phase, PCR reactions were am- re-analysed using this technique (Fig. 1b). The levels ofplified for 20, 25 and 30 cycles and the presence of two control genes, GAPDH and b-actin, remained roughlyamplified products confirmed by agarose gel electropho- constant between the sham and MCAO samples (1.9- andresis (Fig. 1a). The amount of GAPDH PCR product 5.2-fold differences, respectively), which is consistent withproduced at each monitored cycle was roughly identical the semi-quantitative RT-PCR results seen above; a modestbetween the sham and MCAO samples, as might be increase in b-actin following MCAO is consistent with ourexpected. By contrast, there are clear differences in the previous data [13,23]. Again, all three genes identifiedamount of PCR product amplified from the sham and through RDA were shown to be expressed at a higher levelMCAO cDNAs for the three genes identified by RDA, the following MCAO than in the sham library, but this time aproducts, as expected, being more abundant in the MCAO rough estimate of the differences in expression is possible.cDNAs than in the sham control. While a product for To allow consistency, the fold differences were normalisedhsp27 can be seen from the sham samples at later time- to GAPDH (Fig. 1c), and all further data is similarlypoints, the two chemokines seem to be largely undetect- expressed. While hsp27 showed a 83-fold increase follow-able in the sham samples even after 30 cycles. These data ing MCAO, the two chemokines showed far larger in-are consistent with the premise that these three gene creases (950- and 2150-fold for MCP-1 and MIP3a,products accumulate following pMCAO. respectively, see Fig. 1c), which is again consistent with

74 S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80

Fig. 1. Confirmation of differential expression using RT-PCR and SYBR Green RT-PCR. Three RDA-identified genes, MIP3a, MCP-1 and hsp27, wereconfirmed to be up-regulated at 24 h following pMCAO. (a) Gene-specific primers were generated to MIP3a, MCP-1 and hsp27, as well as GAPDH as anegative control, and products amplified for 20, 25 and 30 cycles of PCR from cDNAs derived from both sham-treated (s) and MCAO-treated (m) rats.PCR reactions were analysed by agarose gel electrophoresis and visualised using ethidium bromide fluorescence. (b) Similar PCR reactions were performed

to (a), but this time amplification was visualised using SYBR Green fluorescence and monitored with an ABI Prism 7700 sequence detector in real time.Differences in expression were deduced from the difference in the mean number of PCR cycles required to detect dscDNA/SYBR Green fluorescence

when genes were amplified from MCAO cDNA as compared to the sham-treated cDNA. Fold differences are plotted on the x-axis, using the assumptionthat the amount of PCR products doubles with every exponential PCR cycle, against gene name on the y-axis. Amplifications were performed in duplicateand error bars reflect the standard error. (c) Absolute fold difference from (b) are indicated together with the fold differences corrected for variation inGAPDH.

the inability to detect the two chemokines in the sham wide variation in expression levels ranging from the 9.6-cDNA by RT-PCR. These results also suggest that SYBR fold induction seen with the neuropeptide VGF to the

Green PCR analysis is a fast and efficient method for the 274-fold induction seen with LIF (Fig. 2b). The levels ofanalysis of the remaining RDA clones. CD81 were unchanged following MCAO, highlighting the

need to confirm differential expression of all genes iden-3.3. SYBR Green PCR confirmation of differential tified through subtractive hybridisation.

expression3.4. Time-course of induction

Gene-specific primers were synthesised for the 17 geneswhich were chosen for further analysis. These primers In order to gain a better understanding of the time-were used to amplify products from the sham and MCAO course of induction following MCAO, RNA was isolatedcDNAs and the amplification of products monitored using from independent, but similarly treated, rats from a time-

SYBR Green fluorescence (Fig. 2). As expected, several course up to 24 h. The use of a hybridisation probe inof the genes that were previously described to be induced TaqMan analysis, together with the inclusion of naive as

following MCAO showed higher expression in the MCAO well as sham-treated controls served as additional spe-cDNA than in the sham control. The range of accumulation cificity controls to provide additional validation of the

of transcripts varied dramatically from a modest 9.6-fold SYBR Green PCR results. The extent of the inductionsfor TGF-b1 to the dramatic 200-fold induction seen with seen by TaqMan PCR were consistent with those seen

the metalloprotease inhibitor TIMP-1. Similar results were using SYBR Green PCR, and, as the RNA for TaqManseen with the newly identified MCAO induced genes analysis was extracted from different rats from the RNAtested, which, with the exception of CD81, showed ac- used for RDA analysis, it provides additional evidence thatcumulation in the MCAO-treated rats. Again, there was a these gene products are consistently induced in response to

S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80 75

Fig. 2. SYBR Green RT-PCR analysis of RDA-identified genes. Nine newly described as well as five known MCAO-induced genes were analysed bySYBR Green RT-PCR for expression changes following pMCAO. (a) Mean fold induction, corrected to GAPDH, 24 h following pMCAO is expressed on

the x-axis, against gene names on the y-axis. Amplifications were performed in duplicate and error bars reflect the standard error between reactions. (b)Values for mean fold inductions extracted from (a).

MCAO. The profiles indicated that LIF, SOCS-3 and VGF 4. Discussionhad reached near maximal expression by 12 h, and whilethe levels of LIF and SOCS-3 remained at similarly Here we have used the RDA subtractive hybridisationelevated levels at 24 h, the expression of VGF had started protocol to identify gene products that are accumulated into diminish by this later time-point (Fig. 3). By contrast, the ischaemic cortex following pMCAO in normotensiveMCP-1 levels continued to increase across the time-points, rats. Candidate clones were then confirmed to be differen-

indicating that maximal levels were achieved at 24 h or tially expressed using SYBR Green RT-PCR and inlater. Maximal expression of all four genes was seen in the selected cases the time-course of induction was further

left-hemisphere, where the MCAO was carried out, con- examined using TaqMan RT-PCR methodologies.sistent with the increases in expression being as a result of Given that changes in gene expression have beenthe ischemia and evolving brain injury, although some extensively studied in the MCAO model, it was encourag-minor effects on the right hemisphere was noted with LIF ing to identify a number of genes in the subtracted libraryand MCP-1. that have been previously shown to be up-regulated

76S.

Bates

etal.

/M

olecularB

rainR

esearch93

(2001)70

–80

Fig. 3. Time-course of gene expression changes following pMCAO as monitored by TaqMan RT-PCR. RNA extracted from both left (operated) and right (un-operated) cortices of MCAO (fourindividuals), sham-treated (four individuals) or naive rats (two individuals) at 3, 6, 12 and 24 h was used to generate cDNA which served as template for PCR amplification of MCP-1, LIF, SOCS-3

and VGF. Amplification was performed in triplicate and monitored using a fluorescently labelled gene-specific probes and ABI TaqMan technology. Fold induction versus naive rats for MCAO lefthemisphere (dark circles), right hemisphere (open circles), sham left hemisphere (dark square) and sham right hemisphere (open square) are plotted on the x-axis versus time in hours on the y-axis.Error bars reflect standard error between triplicates.

S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80 77

following MCAO (see Table 1 for references). These genes The accumulation of SOCS-3 in the ischaemic cortex in afunction both as an internal control for the subtraction as time-course that parallels that of the induction of the IL-6well as a standard with which to compare differences in type cytokine LIF, suggests that there is a complexgene expression for the newly described MCAO-regulated interplay of molecules regulating the inflammatory re-genes. The identification of gene products that have proven sponse. Much remains to be elucidated about the action ofto be effective targets for post-ischaemic neuroprotection these molecules, such as whether they are induced in thein the MCAO model, such as COX-2 and iNOS [8,17,33] same cell types, or expressed simultaneously in differentamongst others, provides us with a degree of confidence cell types. Indeed, the JAK/STAT pathway is not uniquethat at least a proportion of the newly ascribed MCAO to the cytokine signalling cascade and, therefore, there mayregulated gene products described here may prove to be be other functions for SOCS-3 in this context in addition totherapeutic targets for stroke. the inflammatory mediators.

4.1. Inflammatory cytokines 4.2. Cell surface molecules

Following ischaemic insult in the CNS, there is a robust Cell surface adhesion molecules are known to play anand prolonged inflammatory response [2,4]. This was important role in the MCAO response, particularly in thehighlighted in the present study by the dramatic increases binding of inflammatory cells to aid their export from thein expression of the known MCAO-induced chemotactic vasculature. Indeed, antibodies that inhibit leukocyte bind-cytokines MCP-1 and MIP3a [45,50], and likely reflect the ing to ICAM-1 have been shown to inhibit leukocyteabsence or near absence of these molecules from the extravasion and provide some level of neuroprotection inun-injured brain. In addition to these known chemokines, certain animal models of stroke [5]. Here we document thethe confirmation here that two further components of the accumulation of transcripts for three additional surfaceinflammatory response, LIF and SOCS-3, also accumulate molecules, CD14, CD44 and osteoactivin, in the ischaemicin the injured cortex highlights the multi-faceted nature of cortex following MCAO. Both CD14 and CD44 are knownthe ischaemic response. to function in the inflammatory response to bacterial

Leukaemia inhibitory factor (LIF) together with IL-6, lipopolysaccharide and proinflammatory cytokines [25,32],ciliary neutrophic factor (CNTF), oncostatin-M (OSM) and their accumulation in the ischaemic cortex may goand cardiotrophin-1 (CT-1) form a sub-group amongst the some way to show that they also function in the in-cytokine family, often termed the IL-6 type cytokines [10]. flammatory response to stroke.The functional similarity of these cytokines can partly be CD14 is a surface molecule of monocytes and to a lesserattributed to the presence of a common gp130 receptor extent neutrophils, so the presence of CD14 may registersubunit, while specificity is achieved through cytokine the entry of monocytes /neutrophils into the ischaemicspecific alpha receptor subunits. Interestingly, two of the region. In addition, the resident brain macrophages, theIL-6-type cytokines (IL-6 [52], and LIF (present study and microglia, are known to express CD14, and the expression[42]) have now been confirmed to be up-regulated follow- of CD14 has been shown to be induced in microgliaing MCAO in the rat, suggesting a broad involvement for following the administration of inflammatory cytokines orthis family of cytokines in the ischaemic response. LIF has bacterial LPS [25]. The sizeable accumulation of CD14been shown to be induced in response to injury in both the transcript following MCAO may therefore reflect both theCNS and PNS, and has an established inflammatory role in induction of endogenous CD14 as well as the infiltration ofa haemopoietic setting [10]. These data are again con- foreign CD14-bearing leukocytes from the circulation.sistent with inflammation being an essential component of CD44 is a widely expressed cell surface, transmembraneCNS response to ischaemic injury, and a role for LIF in protein that is involved in cell to cell and cell to extracellu-modulating gliosis and the inflammatory invasion seems lar matrix interactions. CD44 has been shown to have anlikely. In addition, the infiltration of leukocytes in response important role in the interaction of leukocytes with theto traumatic CNS injury is significantly delayed in LIF vascular endothelium at sites of inflammation [19]. Hy-knockout mice compared to wild-type littermates [41]. aluronan (HA), the major ligand for CD44, is induced in

Once cytokines have bound to their receptors they vascular endothelia by inflammatory cytokines, and thisinitiate a signalling cascade within the cell that ultimately induced HA expression aids the binding and extravasion ofleads to changes in gene expression. In the case of the IL-6 leukocytes from the vasculature [32]. The available datatype cytokines, the primary response results in the activa- suggest that CD44 transcription is not necessarily modifiedtion of janus kinases (JAKs) and phosphorylation of the during the binding/extravasion process, so it is likely thatgp130 component of the receptor, which, in turn favours the accumulation of CD44 in the ischaemic region is thethe association and activation of STAT transcription result of leukocyte accumulation rather than increasedfactors [14]. This process can be negatively regulated at CD44 transcription per se. Obviously, further roles formany levels, and the suppressor of cytokine signalling CD44 in the development of the lesion cannot be ruled out.(SOCS-3) molecule is one such inhibitory molecule [40]. In contrast to the extensively studied CD14 and CD44,

78 S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80

osteoactivin is a relatively novel membrane-spanning The rat orthologue of the Xin gene was shown toglycoprotein. While it has been published in both Genbank accumulate dramatically following MCAO, with approxi-

(Accession number AF184983) and the patents database mate 200-fold induction seen by SYBR Green PCR. Xin(WO9744460-A1), there is no further biological charac- was identified as a transcript that accumulates duringterisation. The patented identification of osteoactivin (or cardiac development [46]. Structurally it has a putativeKIM kidney injury molecule) as a gene product induced DNA binding motif, along with proline rich domains, afollowing hypoxic kidney injury would imply that osteoac- number of nuclear localisation signals and a putative SH3tivin may function in generalised response to inflammatory regulatory domain. These motifs would be consistent withinjury. a transcription factor that shuttles between the cytoplasm

and the nucleus, although none of these motifs have been4.3. Cellular stress genes shown to function experimentally. Interestingly, tissue

distribution data would suggest that Xin expression isThe primary cellular stress genes are the heat shock restricted to cardiac and skeletal muscle, with no detectable

proteins, and well established roles for neuronal hsp70 and expression in uninjured brain [46], or indeed in the spleenastrocytic hsp27 have been described [38]. The beneficial where leukocytes may reasonably to be expected to beeffect that these genes have on stressed cells is highlighted seen. These data would imply that Xin is likely to beby the ameliorating effect on cell death seen when hsp70 is induced in the brain directly in response to the ischaemicoverexpressed in ischaemic neurones. The GADD45 fami- stimulus, with the potential regulation of further genely of stress regulated proteins, by contrast, have been less expression.extensively studied in models of ischaemia, although The complex interplay of neurones, glial cells, endo-GADD45a has been reported to accumulate following thelial cells and the infiltrating leukocytes makes it difficultMCAO [18]. There are currently three members of the to interpret the many processes that occur in response tofamily, a, b and g, and we were able to show here that the experimental stroke. Further work is required to identifymRNA encoding GADD45g accumulates in the ischaemic the individual cell types in which the molecules identifiedcortex following MCAO. Previous studies have shown that here are being expressed, and therefore the response thatGADD45g is the primary GADD45 family member to be may be initiated by intervention. However, it is encourag-induced during death of the neuronal cell line PC12, and, ing to note that we identified many of the gene productsindeed, overexpression of GADD45g is associated with already known to play a role in the response to focalincreased apoptotic rate in cervical ganglion neurones [22]. ischaemia, in addition to the nine newly identified is-The mechanism of action of these molecules is hinted at chaemic targets.from yeast two hybrid studies which identified members of We conclude that subtractive hybridisation methods,the GADD45 family as stimulators of p38/JNK MAP such as RDA, can be used to gain some understanding ofkinase cascade [43]. This is potentially interesting from the the molecular events that characterise experimental strokestroke perspective, because the p38 kinase cascade is models and allow the identification of new genes associ-known to be induced following ischaemic insult, and small ated with brain injury. Future studies are now required tomolecule inhibitors of this class of kinases (CSAIDS) have further establish the spatial localisation of differentiallybeen shown to be neuroprotective in rodent MCAO models expressed genes, but this technique could provide a useful[3,4]. tool to compare a variety of MCAO models.

4.4. Other genesAcknowledgements

The two remaining genes confirmed to accumulate in theThe authors would like to thank the genetic technologies

ischaemic cortex following MCAO currently have onlysequencing group led by Matt Sims for support with

poorly defined functions. VGF was originally identified assequencing, Dave Michalovich for help with bioinfor-

a transcript that accumulates in PC12 cells followingmatics and Karen Philpott, Owen Jenkins, Jeff Aaronson

treatment with neurotrophic growth factors [28]. Theand members of the stroke genomics group for helpful

translated product is cleaved into a number of peptidediscussions.

products which are expressed widely in the CNS, althougha function or a receptor for these peptides have yet to bedefined. A possible role in feeding is implicated from the ReferencesVGF knock-out mice, which are considerably leaner thantheir wild-type littermates, although essentially normal in [1] A.E. Baird, S. Warach, Magnetic resonance imaging of acute stroke

[published erratum appears in J. Cereb. Blood Flow Metab.all other respects [12]. While the functional significance of1998;18(10):preceding 1047], J. Cereb. Blood Flow Metab. 18VGF induction in the ischaemic cortex is not clear at(1998) 583–609.

present, the accumulation of neurotrophic factors in re- [2] F.C. Barone, G.Z. Feuerstein, Inflammatory mediators and stroke:sponse to MCAO may, at least in part, account for the new opportunities for novel therapeutics, J. Cereb. Blood Flowincreased VGF expression [23]. Metab. 19 (1999) 819–834.

S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80 79

[3] F.C. Barone, E.A. Irving, A.M. Ray, J.C. Lee, S. Kassis, S. Kumar, cloning of rat GADD45gamma, gene induction and its role duringA.M. Badger, R.F. White, A.H. Nelson, J.J. Legos, J.A. Erhardt, neuronal cell death, FEBS Lett. 446 (1999) 313–317.E.H. Ohlstein, A.J. Hunter, D.C. Harrison, K. Philpott, K. Ward, [23] Z. Kokaia, Q. Zhao, M. Kokaia, E. Elmer, M. Metsis, M.L. Smith,B.R. Smith, J.L. Adams, A.A. Parsons, Inhibition of p38 mitogen- B.K. Siesjo, O. Lindvall, Regulation of brain-derived neurotrophicactivated protein kinase provides neuroprotection in cerebral focal factor gene expression after transient middle cerebral artery occlu-ischemia., Med. Res. Rev. (2000) in press. sion with and without brain damage, Exp. Neurol. 136 (1995)

[4] F.C. Barone, A.A. Parsons, Therapeutic potential of anti-inflamma- 73–88.tory drugs in focal stroke [In Process Citation], Exp. Opin. Invest. [24] H. Kunizuka, H. Kinouchi, S. Arai, K. Izaki, S. Mikawa, H. Kamii,Drugs 9 (2000) 2281–2306. T. Sugawara, A. Suzuki, K. Mizoi, T. Yoshimoto, Activation of Arc

[5] H. Chen, M. Chopp, R.L. Zhang, G. Bodzin, Q. Chen, J.R. Rusche, gene, a dendritic immediate early gene, by middle cerebral arteryR.F. Todd III, Anti-CD11b monoclonal antibody reduces ischemic occlusion in rat brain, Neuroreport 10 (1999) 1717–1722.cell damage after transient focal cerebral ischemia in rat, Ann. [25] S. Lacroix, D. Feinstein, S. Rivest, The bacterial endotoxinNeurol. 35 (1994) 458–463. lipopolysaccharide has the ability to target the brain in upregulating

[6] W.M. Clark, E.C. Raps, D.C. Tong, R.E. Kelly, Cervene (Nal- its membrane CD14 receptor within specific cellular populations,mefene) in acute ischemic stroke: final results of a phase III efficacy Brain Pathol. 8 (1998) 625–640.study: The Cervene Stroke Study Investigators, Stroke 31 (2000) [26] M. Lai, E. Sirimanne, C.E. Williams, P.D. Gluckman, Sequential1234–1239. patterns of inhibin subunit gene expression following hypoxic–

[7] T.J. DeGraba, L.C. Pettigrew, Why do neuroprotective drugs work ischemic injury in the rat brain, Neuroscience 70 (1996) 1013–1024.in animals but not humans?, Neurol. Clin. 18 (2000) 475–493. [27] K.R. Lees, K. Asplund, A. Carolei, S.M. Davis, H.C. Diener, M.

[8] G. del Zoppo, I. Ginis, J.M. Hallenbeck, C. Iadecola, X. Wang, G.Z. Kaste, J.M. Orgogozo, J. Whitehead, Glycine antagonist (gavestinel)Feuerstein, Inflammation and stroke: putative role for cytokines, in neuroprotection (GAIN International) in patients with acuteadhesion molecules and iNOS in brain response to ischemia, Brain stroke: a randomised controlled trial. GAIN International Inves-Pathol. 10 (2000) 95–112. tigators [see comments], Lancet 355 (2000) 1949–1954.

[9] G.Z. Feuerstein, X. Wang, Animal models of stroke, Mol. Med. [28] A. Levi, J.D. Eldridge, B.M. Paterson, Molecular cloning of a geneToday 6 (2000) 133–135. sequence regulated by nerve growth factor, Science 229 (1985)

393–395.[10] R.A. Gadient, P.H. Patterson, Leukemia inhibitory factor, Interleukin6, and other cytokines using the GP130 transducing receptor: roles [29] N. Lisitsyn, N. Lisitsyn, M. Wigler, Cloning the differences betweenin inflammation and injury, Stem Cells 17 (1999) 127–137. two complex genomes, Science 259 (1993) 946–951.

[11] M.D. Ginsberg, R. Busto, Rodent models of cerebral ischemia, [30] E.Z. Longa, P.R. Weinstein, S. Carlson, R. Cummins, ReversibleStroke 20 (1989) 1627–1642. middle cerebral artery occlusion without craniectomy in rats, Stroke

20 (1989) 84–91.[12] S. Hahm, T.M. Mizuno, T.J. Wu, J.P. Wisor, C.A. Priest, C.A.Kozak, C.N. Boozer, B. Peng, R.C. McEvoy, P. Good, K.A. Kelley, [31] A.D. Medhurst, D.C. Harrison, S.J. Read, C.A. Campbell, M.J.J.S. Takahashi, J.E. Pintar, J.L. Roberts, C.V. Mobbs, S.R. Salton, Robbins, M.N. Pangalos, The use of TaqMan RT-PCR assays forTargeted deletion of the Vgf gene indicates that the encoded semiquantitative analysis of gene expression in CNS tissues andsecretory peptide precursor plays a novel role in the regulation of disease models, J. Neurosci. Methods 98 (2000) 9–20.energy balance [see comments], Neuron 23 (1999) 537–548. [32] M. Mohamadzadeh, H. DeGrendele, H. Arizpe, P. Estess, M.

[13] D.C. Harrison, A.D. Medhurst, B.C. Bond, C.A. Campbell, R.P. Siegelman, Proinflammatory stimuli regulate endothelial hyaluronanDavis, K.L. Philpott, The use of quantitative RT-PCR to measure expression and CD44/HA-dependent primary adhesion, J. Clin.mRNA expression in a rat model of focal ischemia — caspase-3 as a Invest. 101 (1998) 97–108.case study, Mol. Brain Res. 75 (2000) 143–149. [33] S. Nogawa, F. Zhang, M.E. Ross, C. Iadecola, Cyclo-oxygenase-2

[14] P.C. Heinrich, I. Behrmann, G. Muller-Newen, F. Schaper, L. gene expression in neurons contributes to ischemic brain damage, J.Graeve, Interleukin-6-type cytokine signalling through the gp130/ Neurosci. 17 (1997) 2746–2755.Jak /STAT pathway, Biochem. J. 334 (2) (1998) 297–314. [34] A.A. Parsons, C.A. Campbell, E.A. Irving, S.J. Hadingham, J.

[15] M. Hubank, D.G. Schatz, Identifying differences in mRNA expres- Roberts, J.J. Legos, F.C. Barone, A.J. Hunter, Evolution of brainsion by representational difference analysis of cDNA, Nucleic Acids injury following cerebral ishemia: Relating animal models to clinicalRes. 22 (1994) 5640–5648. stroke populations, Adv. Clin. Neurosci. (2000) in press.

[16] C. Iadecola, F. Zhang, S. Xu, R. Casey, M.E. Ross, Inducible nitric [35] A.M. Planas, M.A. Soriano, E. Rodriguez-Farre, I. Ferrer, Inductionoxide synthase gene expression in brain following cerebral ischemia, of cyclooxygenase-2 mRNA and protein following transient focalJ. Cereb. Blood Flow Metab. 15 (1995) 378–384. ischemia in the rat brain, Neurosci. Lett. 200 (1995) 187–190.

[17] C. Iadecola, F. Zhang, X. Xu, Inhibition of inducible nitric oxide [36] R. Schmidt-Kastner, J. Truettner, W. Zhao, L. Belayev, C. Krieger,synthase ameliorates cerebral ischemic damage, Am. J. Physiol 268 R. Busto, M.D. Ginsberg, Differential changes of bax, caspase-3 and(1995) R286–R292. p21 mRNA expression after transient focal brain ischemia in the rat,

Mol. Brain Res. 79 (2000) 88–101.[18] K. Jin, J. Chen, K. Kawaguchi, R.L. Zhu, R.A. Stetler, R.P. Simon,S.H. Graham, Focal ischemia induces expression of the DNA [37] F.R. Sharp, A. Lu, Y. Tang, D.E. Millhorn, Multiple moleculardamage-inducible gene GADD45 in the rat brain, Neuroreport 7 penumbras after focal cerebral ischemia, J. Cereb. Blood Flow(1996) 1797–1802. Metab. 20 (2000) 1011–1032.

[19] P. Johnson, A. Maiti, K.L. Brown, R. Li, A role for the cell adhesion [38] F.R. Sharp, S.M. Massa, R.A. Swanson, Heat-shock protein protec-molecule CD44 and sulfation in leukocyte–endothelial cell adhesion tion, Trends Neurosci. 22 (1999) 97–99.during an inflammatory response?, Biochem. Pharmacol. 59 (2000) [39] M.A. Soriano, M. Tessier, U. Certa, R. Gill, Parallel gene expression455–465. monitoring using oligonucleotide probe arrays of multiple transcripts

[20] J. Koistinaho, T. Hokfelt, Altered gene expression in brain ischemia, with an animal model of focal ischemia, J. Cereb. Blood FlowNeuroreport 8 (1997) 1–8. Metab. 20 (2000) 1045–1055.

[21] J. Koizumi, Y. Yoshida, T. Nakaazawa, G. Ooneda, Experimental [40] R. Starr, T.A. Willson, E.M. Viney, L.J. Murray, J.R. Rayner, B.J.studies of ischemic brain edema. 1. A new experimental model of Jenkins, T.J. Gonda, W.S. Alexander, D. Metcalf, N.A. Nicola, D.J.cerebral embolism in rats in which recirculation can be introduced in Hilton, A family of cytokine-inducible inhibitors of signalling,the ischemic area, Jpn. J. Stroke 8 (1986) 1–8. Nature 387 (1997) 917–921.

[22] S. Kojima, K. Mayumi-Matsuda, H. Suzuki, T. Sakata, Molecular [41] S. Sugiura, R. Lahav, J. Han, S.Y. Kou, L.R. Banner, F. de Pablo,

80 S. Bates et al. / Molecular Brain Research 93 (2001) 70 –80

P.H. Patterson, Leukaemia inhibitory factor is required for normal cloning identifies tissue inhibitor of matrix metalloproteinase-1inflammatory responses to injury in the peripheral and central (TIMP-1) increased gene expression following focal stroke, Strokenervous systems in vivo and is chemotactic for macrophages in 29 (1998) 516–520.vitro, Eur. J. Neurosci. 12 (2000) 457–466. [49] X. Wang, S. Yaish-Ohad, X. Li, F.C. Barone, G.Z. Feuerstein, Use of

[42] S. Suzuki, K. Tanaka, S. Nogawa, D. Ito, T. Dembo, A. Kosakai, Y. suppression subtractive hybridization strategy for discovery ofFukuuchi, Immunohistochemical detection of leukemia inhibitory increased tissue inhibitor of matrix metalloproteinase-1 gene expres-factor after focal cerebral ischemia in rats, J. Cereb. Blood Flow sion in brain ischemic tolerance, J. Cereb. Blood Flow Metab. 18Metab. 20 (2000) 661–668. (1998) 1173–1177.

[43] M. Takekawa, H. Saito, A family of stress-inducible GADD45-like [50] X. Wang, T.L. Yue, F.C. Barone, G.Z. Feuerstein, Monocyteproteins mediate activation of the stress-responsive MTK1/MEKK4 chemoattractant protein-1 messenger RNA expression in rat is-MAPKKK, Cell 95 (1998) 521–530. chemic cortex, Stroke 26 (1995) 661–665.

[44] A. Tamura, D.I. Graham, J. McCulloch, G.M. Teasdale, Focal [51] X. Wang, T.L. Yue, R.F. White, F.C. Barone, G.Z. Feuerstein,cerebral ischaemia in the rat: 1. Description of technique and early Transforming growth factor-beta 1 exhibits delayed gene expressionneuropathological consequences following middle cerebral artery following focal cerebral ischemia, Brain Res. Bull. 36 (1995)occlusion, J. Cereb. Blood Flow Metab. 1 (1981) 53–60. 607–609.

[45] U. Utans-Schneitz, H. Lorez, W.E. Klinkert, J. da Silva, W. Les- [52] X. Wang, T.L. Yue, P.R. Young, F.C. Barone, G.Z. Feuerstein,slauer, A novel rat CC chemokine, identified by targeted differential Expression of interleukin-6, c-fos, and zif268 mRNAs in ratdisplay, is upregulated in brain inflammation [published erratum ischemic cortex, J. Cereb. Blood Flow Metab. 15 (1995) 166–171.appears in J. Neuroimmunol. 1999;94(1–2):222], J. Neuroimmunol. [53] M.Y. Ahn, Z.G. Zhang, W. Tsang, M. Chopp, Endogenous plas-92 (1998) 179–190. minogen activator expression after embolic focal cerebral ischemia

[46] D.Z. Wang, R.S. Reiter, J.L. Lin, Q. Wang, H.S. Williams, S.L. in mice, Brain Res. 837 (1999) 169–176.Krob, T.M. Schultheiss, S. Evans, J.J. Lin, Requirement of a novel [54] K. Appel, M. Buttini, A. Sauter, P.J. Gebicke-Haerter, Cloning of ratgene, Xin, in cardiac morphogenesis, Development 126 (1999) interleukin-3 receptor beta-subunit from cultured microglia and its1281–1294. mRNA expression in vivo, J. Neurosci. 15 (1995) 5800–5809.

[47] X. Wang, F.C. Barone, N.V. Aiyar, G.Z. Feuerstein, Interleukin-1 [55] T. Higashi, H. Takechi, Y. Uemura, H. Kikuchi, K. Nagata,receptor and receptor antagonist gene expression after focal stroke in Differential induction of mRNA species encoding several classes ofrats, Stroke 28 (1997) 155–161. shess proteins following focal cerebral ischemia in rats, Brain Res.

[48] X. Wang, F.C. Barone, R.F. White, G.Z. Feuerstein, Subtractive 650 (1994) 239–248.