Fusion of core pathways reveals a horizontal synergistic mechanism underlyingcombination therapy
Zhong Wang a,⁎, Zhi-Wei Jing a, Cai-Xiu Zhou a, Liang Zhang b, Jing Cheng b, Zhan-Jun Zhang c, Jun Liu a,Cun-Shuan Xu d, Peng-Tao Li e, Yong-Yan Wang a
a Institute of Basic Research in Clinical Medicine, China Academy of Chinese Medical Sciences, 18 Baixincang, Dongzhimennei, Beijing 100700, Chinab Life Science Institute, Tsinghua University, Beijing 100091, Chinac Beijing Normal University, The Key Laboratory of Traditional Chinese Medicine Protection and Utilization, 19 XinJieKouWai Street, HaiDian District, Beijing 100875, Chinad College of Life Science, Henan Normal University, Xinxiang 453002, Henan Province, Chinae Beijing University of Traditional Chinese Medicine, Beijing 100029, China
Combination therapies have recently been shown to be more effective than monotherapies that may providesynergistic effects in the treatment of stroke, but its selective mechanism still remains unclear. Based on themedian-effect method, the combination therapy of jasminoidin and ursodeoxycholic acid had a synergic effecton reducing the infarct volume. The numbers of up- or down-regulated genes by at least 1.5-fold in thevehicle, jasminoidin, ursodeoxycholic acid, and the combination of jasminoidin and ursodeoxycholic acidtreatment groups were 228, 95, 136, and 101, respectively. According to clustering and principal componentanalysis, the pattern of gene expression in the combination group was similar to that of jasminoidin grouprather than ursodeoxycholic acid group. Based on these nine top sequences in the combination groupexcluding four overlapping pathways (MAPK-ERK, Kitlg, Icam1-Ap1, and prolactin), the jasminoidin grouphad four (PRLR-STAT1, AcvR2-AcvR1B, ACVR1/2A-SMAD1, GHR-NF-κB) contributing pathways, and theursodeoxycholic acid group had one (IL-6) contributing pathway. Based on the multiple-pathway-dependentcomparison analysis (MPDCA), it may lead to the conclusion that jasminoidin possibly contributes moreimportant pharmacological effect in the combined treatment as jasminoidin regulated 80% of the pathwaysthat the combination group mediated. The study reveals a horizontal synergistic effect by optimizing thefusion of more pathways from the compounds with more contribution to the combination therapy. Ratherthan selecting compounds only based on experience in the past, this study would give a new insight into thesystematic strategies for designing synergistic combination therapies.
Stroke remains a major cause of death and disability worldwide.The pathophysiology of stroke is highly complex. The molecularevents that mediate ischemic brain damage, including glutamateaccumulation, aberrant calcium fluxes (Stanika et al., 2010), freeradical formation (Allen and Bayraktutan, 2009), and lipid peroxida-tion (Gaur and Kumar, 2009), are logical targets for pharmacologicalintervention. However, disappointing outcomes highlight the limita-tions of the single-target drug paradigm (Hiroaki, 2007). Combinationtherapies have recently been shown to be more effective thanmonotherapies that may provide synergistic effects in the treatmentof cardiovascular diseases (Sleight et al., 2006). Analysis of the 117drug combinations identified general and specific modes of action (Jia
et al., 2009); developing a strategy for determining the mostpromising combinations and prioritizing their evaluation is crucialand remains a major challenge (Black and Sang, 2005).
It seems unlikely that a single pathway is sufficiently critical todefine an outcome on its own. One promising alternative is to usecombination therapy to provide neuroprotection in cerebrovascularsurgery (Zhang et al., 2006). The discovery of new neuroprotectiveagents has spurred efforts to understand the intracellular signalingpathways that mediate the cellular response to stroke and to identifythe mechanism to this response. It is thought that such a mechanismwould fit within the concept that several brain injury pathways mustbe inhibited to optimize therapeutic efficacy (Zanelli et al., 2005). Thesimultaneous or sequential action on targets of different relatedpathways represents horizontal synergistic and additive interactions,such as those of fludioxonil with FK506 (Chen et al., 2006) andondansetron with droperidol (Chan et al., 2006). Vertical synergisticand additive interactions are represented by interactions betweenterbinafine with azoles (Perea et al., 2002) and ampicillin with
279Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
imipenem (Fuda et al., 2004). Although several tools for visualizingthese interactions graphically and for analyzing the biological datahave been developed, the numbers of core pathways that are affectedsimultaneously in a single-compound intervention and the variationsin combination therapy responsible for a global phenotype changeremain unclear.
Jasminoidin and ursodeoxycholic acid, the chemical structures ofwhich are shown in Fig. 1, are two active ingredients in Qingkailingwhich is a herbal formula clinically used to treat stroke in China. InPubMed from January 1, 2000 to January 1, 2010, a total of 133 articleson jasminoidin and 1536 articles on ursodeoxycholic acid have beenpublished. Based on thenumber of articles, the top four pharmacologicalactions of jasminoidin are presented as neuroprotective agents,cholagogues, enzyme inhibitors, and anti-inflammatory agents. Andthe top four pharmacological actions of ursodeoxycholic acid arepresented as cholagogues, immunosuppressive agents, gastrointestinalagents, and anti-inflammatory agents. Jasminoidin has been shown tohave a preventive effect against ischemic stroke by promoting theexpression of brain-derived neurotrophic factor (BDNF) and inhibitingthe expression of caspase-3 (Zhang et al., 2006). Jasminoidin is also anewly identified agonist of glucagon-like peptide-1 (GLP-1) receptor,which protects PC12 cells from oxidative damage via the mitogen-activated protein kinase (MAPK) pathway (Liu et al., 2007). Ursodeoxy-cholic acid has been reported to have cytoprotective and antioxidativeproperties (Brito et al., 2008; Yasukawa et al., 2009). In this study, weused the multiple-pathway-dependent comparison analysis (MPDCA)to explore the various potential core pathways in ischemic mousehippocampal cells treated with jasminoidin, ursodeoxycholic acid, orthe combination of both.
2. Material and methods
2.1. Animal model
Animal experiments were performed in accordance with thePrevention of Cruelty to Animals Act 1986 and the National Institutesof Health guidelines for the care and use of laboratory animals forexperimental procedures, and were approved after review by a localcommittee. One hundred seventy adult male mice (3 months old, 38–48 g, Kunming strain, China) were divided into 5 groups of 34 miceeach. Focal cerebral ischemia–reperfusion model was induced in miceanesthetized with 2% pentobarbital (4 mg/kg, i.p.) by occluding theleft middle cerebral artery with an intraluminal filament as described(Hara et al., 1996). The middle cerebral artery was exposed andligated for 1.5 h using an intraluminal filament and then reperfusedfor 24 h. The sham-operated mice underwent identical procedureswithout middle cerebral artery occlusion.
Fig. 1. The chemical structure of jasmi
2.2. Drug administration
Experimental animals were randomly divided into 5 groups:sham-operated mice; ischemic mice receiving one of the three herbalpreparations at a dose of 2 ml/kg [25 mg/ml jasminoidin, 7 mg/mlursodeoxycholic acid, or the combination of jasminoidin andursodeoxycholic acid with a ratio of 1:1)]; or vehicle-treated mice(0.9% NaCl, 2 ml/kg). The herbal preparation or vehicle was given byintravenous injection in the tail vein once a day. The herbalpreparations were a chemically standardized product from ChinaNatural Institute for the Control of Pharmaceutical and BiologicalProduct or Beijing University of Traditional Chinese Medicine, and itscomposition was validated using fingerprint chromatographic meth-odologies. These preparations were dissolved in 0.9% NaCl just beforeuse. Jasminoidin and ursodeoxycholic acid at concentrations of 1, 2, 4,8, and 16 mg/mlwere combined (1:1) in equal volumes to analyze thesynergistic effect using CompuSyn software (ComboSyn, Inc. USA).The combination index (CI), a quantitative measure based on themass-action law of the degree of drug interaction in terms ofsynergism and antagonism for a given endpoint of the effectmeasurement (Chou and Talalay, 1981), was calculated.
2.3. Histological analysis
After 24 h reperfusion, 6 animals from each group were anesthe-tized with chloral hydrate (400 mg/kg). The brain was perfusedimmediately with 37 °C saline through cannulation of the aorta. Theblood was washed out, and the brain was then perfused with cold 4%formaldehyde for 30 min to induce polymerization. The brain wasremoved and post-fixed in 4% formaldehyde for at least 24 h,embedded in mineral wax, sectioned coronally into 5-μm slices, andstained with thionine. And the hippocampal CA1 region was selectedfor observation.
The infarct ratio after the 24 h reperfusion was calculated inanother 13mice from each group. In brief, the cerebrumwas removedand cut into five slices in the coronal plane 1, 3, 5, and 7 mm from theprefrontal cortex. The slices were transferred to 4% TTC solution andincubated for 30 min at 37 °C in darkness and then transferred into10% formalin. Images of the slices were captured using a digitalcamera (Color CCD camera TP-6001A, Topica Inc., Japan). The area ofthe infarct region was calculated using a Pathology Image AnalysisSystem (Topica Inc.), and the ratio of the infarct volume to the totalslice was calculated.
noidin and ursodeoxycholic acid.
280 Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
2.5. Neurological deficits
Behavioral changes in mice were assessed 1.5 h after ischemia(pre-treatment) and 24 h after reperfusion (post-treatment) andscored as described previously (Bederson et al., 1987), with a minormodification as follows: 0, no observable neurological deficits(normal); 1, failure to extend right forepaw (mild); 2, circling to thecontralateral side (moderate); and 3, loss of walking or righting reflex(severe).
2.6. RNA extraction and labeling
Total RNA was extracted from the hippocampuses of 9 mice fromeach group using the RNeasy Micro Kit (Qiagen, Valencia, CA). Thequality of total RNA was assessed using a Bioanalyzer microchip(Agilent, Palo Alto, CA). One hundred nanograms of total RNA wasamplified following the small sample labeling protocol. The protocolinvolved two rounds of reverse transcription and in vitro transcriptionwith the biotin label incorporated during the second round of in vitrotranscription.
2.7. Microarray
The Mouse Brain Array (Boao Capital Co Ltd., Beijing, China) con-taining 16,463 mice oligo clones (Incyte Genomics, Inc., Santa Clara,CA) was used to conduct gene expression profiling. Each clone wasprinted as duplicate spots on a given chip, generating four technicalreplicates for each clone. A single intensity value for each clone wasgenerated by averaging the quadruplet measurements after smooth-ing spline normalization. Some genes were represented by morethan one clone on the array. All clones had been verified by DNAsequencing. Six to nine biological replicates for each group werehybridized. The RNA of the vehicle group as a pool was labeled withCy3, and the RNA of the other groups with Cy5. The microarrays werehybridized, washed, and scanned according to standard protocols.
2.8. Microarray data analysis
Robust multiarray analysis was used for preprocessing of the rawCEL files (fluorescence intensity files). The data were normalizedusing GeneSpring per chip normalization (normalized to the 50thpercentile) and per gene normalization (normalized to the median)(Silicon Genetics, Redwood City, CA). One-way analysis of variance(ANOVA) and significance analysis of microarrays (SAM)were used tocompare the means of altered genes in the different groups, with allmeans from each group taken from at least three independentmicroarrays. Genes showing significant changes relating to thetreatment (Pb0.05 after adjusting for the false discovery rate (FDR))were identified and imported back into GeneSpring for fold-changefiltering, clustering analysis, principal component analysis (PCA), andgraphic representation. All data were subjected to intensity-dependent(LOWESS) normalization. We set a cutoff of N1.5-fold (compound orsham treatment vs. vehicle treatment) in all three experiments. Anygene whose detection P value was N0.05 under all experimentalconditions was discarded from the analysis as unreliable data. We usedpaired t tests to compare the relative changes in gene expression aftersingle-agent treatment vs. combination treatment.
The experimental analysis was based on theMinimum Informationabout a Microarray Experiment (MIAME) guidelines and the Micro-array Quality Control (MAQC) project. The results have been sub-mitted to the Array Express database. Statistical significance wasascribed to the data when the smoothing spline normalizationPb0.05. The null hypothesis was rejected at the 0.05 level.
Three animals from each group were anesthetized with chloralhydrate (400 mg/kg) and decapitated. Total RNA was extracted fromthe hippocampus and cDNAs were synthesized from 200 ng of totalRNA using TaqMan reverse transcription reagents (Applied Biosys-tems, Foster, CA). PCR was carried out using 2 μl of cDNAs with the aidof an SYBR Green PCR Core Reagent Kit in an ABI Prism 7000 SequenceDetection System. Reactions containing no cDNA or no primers servedas negative controls. The relative expression levels of IL-6 (primer:GAACTCCTTCTCCACAAGCGCCTT, CAAAAGACCAGTGATGATTTTCACCAGG)were analyzed using the comparative Ct method for relative quantifi-cation and the 2–ΔΔCt method was used after normalizing withglyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expres-sion. The data were presented as the percent changes compared withthe matched controls.
2.10. Western blotting
The hippocampus was removed from the brains of the remainingthree mice from each group and homogenized. The proteins (40 μg perlane) were separated by sodium dodecyl sulfate (SDS)-polyacrylamidegel electrophoresis (PAGE) and transferred tonitrocellulosemembranes(Hybond-C, Amersham, UK) by electroblotting. Membranes wereincubated in 5% nonfat milk for 1 h and incubated with antibodies toanti-extracellular signal-regulated kinase-1 (ERK1) (Promega), ERK2(Promega), and anti-FosB (Santa Cruz), and developed using enhancedchemiluminescence (Amersham). Band density was determined with aGS-700 densitometer (Bio-Rad).
2.11. Pathway analysis
Pathway analysis was performed with KEGG and Pathway Studiov5.0 (Ariadne Genomics, Rockville, MD) (Samardzija et al., 2006),according to the manufacturer's instructions provided based on themolecular interaction database ResNet core 4.0. The degree ofsimilarity (DS) was processed as indicated in the Pathway Studiov5.0 user manual. We constructed a correction network with PathwayStudio using three methods: directly linked to the entities selected forexpansion, common targets, and common regulators. Only thoseresults that were consistent with each other in these three methodswere included in the analysis. Significance was assessed based on thenumber of gene hits in a pathway compared with the total number ofgenes assigned to a specific pathway with Pb0.01.
3. Results
3.1. Differences in pharmacological effects among different treatmentgroups
Infarct volume and behavior grade were used to evaluate the effectsof the compound preparations jasminoidin, ursodeoxycholic acid, andthe combination of both. The total infarct volume was significantlysmaller in the compound treatment groups compared with the vehiclegroup (F jasminoidin=12.22, P=0.018; F ursodeoxycholic acid=9.15,P=0.033; F combination=18.43, P=0.0006 vs. vehicle, n=13;ANOVA) (Fig. 2A). The infarct volume was reduced further by the com-bination therapy compared with the effects of jasminoidin orursodeoxycholic acid alone. In the sham-operated group, the hippo-campal neuronswith normal structure or granule-likeNissl bodieswereobserved. In the hippocampuses of the vehicle group, Nissl bodies weremissing completely from the neurons. By contrast, the hippocampalneurons in the jasminoidin, ursodeoxycholic acid, and combinationgroups showedsignificant alleviationof the lossofNissl bodies; theNisslbodies could be identified but were not abundant (Fig. 2B).
Fig. 2.Differences inpharmacological effects amongdifferent treatmentgroups. (A) Infarct volumeratioofmice at24 hafter reperfusion,which is shownaspercentage.n=13.Mean±S.D.*Pb0.05, **Pb0.01 vs. vehicle group. (B) Hippocampus in the sham-operated group, neurons with normal structure and plaque-like or granule-like Nissl bodies were observed;hippocampus in the vehicle group,Nissl bodies inneurons almost lost completely; hippocampus in the treatment groups of jasminoidin, ursodeoxycholic acid and combination of both, theloss of Nissl bodies was alleviated significantly, and Missal bodies were identified, but not abundant. ×400. (C) Grade of behavior deficit in mice at 1.5 h after ischemia and 24 h afterreperfusion. n=10. Mean±S.D. *Pb0.05, **Pb0.01 vs. vehicle group, #Pb0.05, ##Pb0.01 vs. pre-treatment. (D) The dose–effect curves of single or combined drug treatment, and thecombination of jasminoidin and ursodeoxycholic acid exerts synergic effects (E) on infarct volume, as reflected by the median-effect method. CI, combination index; Fa, fraction affected.
281Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
In the jasminoidin, ursodeoxycholic acid, and combination groups,the behavior grade was significantly lower after reperfusion than thatat 1.5 h after ischemia (pre-treatment) within a same group, as well asthat at 24 h after reperfusion in the vehicle group (Pb0.05). Theseresults suggested that the combined treatment of jasminoidin andursodeoxycholic acid significantly relieved the behavior disabilitycaused by cerebral ischemia–reperfusion injury in mice (Fig. 2C).
To test whether jasminoidin and ursodeoxycholic acid had asynergic effect on reducing the infarct volume, the dose–effect curvesof the single or combined drug treatment were analyzed by themedian-effect method, where the CI of b, =, and N1 indicatedsynergistic, additive, and antagonistic effects, respectively. The dose–effect curves of jasminoidin, ursodeoxycholic acid, and combinationtherapy (Fig. 2D) at concentrations of 1–16 mg/ml had CI valuesb1(Fig. 2E), indicating synergic effects between the agents.
3.2. Distribution of genes with altered expression over a range ofconditions
In the sham-operated, vehicle control, jasminoidin, ursodeoxycholicacid, and combination groups, 1,436 (8.72%) transcripts of 16,463 totalgeneswere present (Pb0.05). The FDRwas 4.4%. Up- or down-regulationof at least 1.5-fold occurred in 228, 95, 136, and 101 genes in the vehicle,jasminoidin, ursodeoxycholic acid, and combination groups, respectively(Fig. 4A). Fivehundred sixty (76.79%) of these genes could be categorizedinto12groups in accordancewithgeneontology annotationbasedon themain function of their protein products as identified. Of the 177 knownfunction genes in the vehicle control group that were differentiallyexpressed relative to the sham-operatedmice, 80were up-regulated and97 were down-regulated. Forty-five, 48, and 35 genes showed increasedexpression in the jasminoidin, ursodeoxycholic acid, and combination
282 Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
groups, respectively; these up-regulated genes were the reverse of the 97down-regulated genes in the vehicle control group. By contrast, 41, 38,and 45 genes showed decreased expression in the jasminoidin,ursodeoxycholic acid, and combination groups, respectively; thesedown-regulated genes were the reverse of the 80 genes up-regulated inthe vehicle control group (Fig. 3). The number of up-regulated (35–48,36.08–49.48%) or down-regulated (38–45, 47.50–58.75%) genes in thetreatment groups accounted for one-third to one-half of the genesshowing altered expression in the vehicle control group. The set of genesassociatedwith energymetabolism included54, 30, 35, and30geneswithaltered expression in the vehicle, jasminoidin, ursodeoxycholic acid, andcombination groups, respectively. These represented 30.51% (54/177),34.88% (30/86), 40.70% (35/86), and 37.50% (30/80) of the total genes ofknown function with altered expression in these groups, respectively.
3.3. Clustering and independent PCA of overlapping genes
Only 19 genes showed altered expression in all of the three treat-ment groups (see Supplementary Table 1 online), although 21(jasminoidin and ursodeoxycholic acid), 30 (ursodeoxycholic acid andcombination), and 25 (jasminoidin and combination) genes showedalterations that were shared between two groups (Fig. 4A). Recentstudies have shown that grouping geneswith similar expressionprofilesmay reveal the function of a coordinately controlled gene cluster. Wetherefore assessed the patterns of gene expression in the threetreatment groups by unsupervised (unbiased) agglomerative hierar-chical clustering. The patterns of gene expression in the jasminoidin and
Fig. 3. Altered genes based on the Gene Ontology category contribution amo
combination groups were closer to each other than those of theursodeoxycholic acid group (Fig. 4B). The results from the first to thirdprincipal components included N90% of the detected variation in geneexpression (Fig. 4C) and were consistent with those of the clusteringanalysis.
3.4. Reconstruction of the network of pathways among the threetreatment groups
Analysis of the reconstruction network of pathways of the 19overlapping genes among the jasminoidin, ursodeoxycholic acid, andcombination groups using the three methods (directly linked to theentities selected for expansion, common targets, and commonregulators) showed that some genes, such as proopiomelanocortin(Pomc), FBJ murine osteosarcoma viral oncogene homolog (Fos), earlygrowth response factor 1 (Egr1), prolactin (Prl), prostaglandin-endoperoxide synthase 2 (Ptgs2), and Jun-B oncogene (JunB), were“central” in linking the activity of peripheral genes to form thebiologic networks (Fig. 5). An association network was reconstructedusing the common target approach, and 25 overlapping genesbetween the jasminoidin and combination groups were obtainedusing PathwayStudio5.0 (see Supplementary Fig. 1 online). DSanalysis suggested that the KIT ligand (KITLG), interleukin-3 (IL-3),interleukin-6 (IL-6), insulin, and platelet-derived growth factor(PDGF) signaling pathways were the sequences with the highest DSin this comparison (Table 1).
ng vehicle, jasminoidin, ursodeoxycholic acid and combination groups.
Fig. 4. Clustering and PCA of altered genes among jasminoidin, ursodeoxycholic acid and combination groups. (A) The contribution of overlapping and non-overlapping genes amongdifferent treatment groups. (B) Principal component analysis (shown as PCA1–3 in the scatter plot) and (C) clustering analysis gave consistent results that the profiles of geneexpression in the jasminoidin and combination groups showed the greatest similarity among the three groups as shown on the left, and some of the genes included in clustering islisted on the right.
283Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
3.5. Variations in multiple pathways modulated in the three treatmentgroups
We developed the MPDCA approach that could identify and rankthe candidate pathways based on the coexpression with groups ofpathway genes and the relative strength (P and/or r2 values) of thesecoexpression relationships. We listed all significant pathways exam-ined in this study and ranked the results from the MPDCA usingdifferent P values as coexpression cutoffs. Although the top corepathways modulated in the jasminoidin, ursodeoxycholic acid, andcombination groups differed (Fig. 6A), four overlapping pathways(MAPK-ERK, Kitlg, Icam1-Ap1, and prolactin) were identified amongall of the three treatment groups. Of the 5 remaining top sequences inthe combination group excluding the 4 overlapping pathways, the
jasminoidin group had four contributing pathways (PRLR-STAT1,AcvR2-AcvR1B, AcvR1/2A-SMAD1, GHR-NF-κB), and the ursodeoxy-cholic acid group had one contributing pathway (IL-6). This suggestedthat the pathways to which jasminoidin contributed accounted for80% (4/5) of the pathways excluding the overlapping pathways in thecombination group. Interestingly, the core pathways in the combina-tion group were not simply equally fusion pathways from thejasminoidin and ursodeoxycholic acid groups. The results of thereal-time PCR and western blot analysis showed that the p-ERK2pathway level decreased significantly in the jasminoidin, ursodeoxy-cholic acid, and combination groups, which was similar to the patternof the IL-6 pathway in the jasminoidin and combination groups(Fig. 6B). This finding was consistent with the results of primal con-dition analysis shown in Fig. 4B, C, and Fig. 6A.
Fig. 5. Association network based on common regulators from19 overlapping genes among jasminoidin, ursodeoxycholic acid and combination groups obtained by PathwayStudio. In thisfigure, theblue linesdenote regulationof expression; thepurple linesdenote binding; thegray linesdenote squares regulation; thegray circles illustrate direct regulation; theyellowcirclesindicate enhancedmodification; the green circlesmeanproteinmodification; the light green circles illustratemolecular transfer; the light blue circles displaymolecular synthesis; and theblack circles represent chemical reaction.
284 Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
Table 1DS analysis using PathwayStudio by common regulators approach on 19 overlappinggenes among the jasminoidin, ursodeoxycholic acid and combination groups (vs.vehicle control). The P-value was calculated by adding the number of genes in commonregulators between the known pathway and the comparison data set divided by thetotal number of genes. A Fisher right-tailed t-test was used to define the significance ofeach canonical pathway.
285Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
4. Discussion
The present study indicated that: 1) the synergistic effect of thecombination group may be attributable to more signaling pathwaysrelated to ischemic brain injury that the combined preparation couldmediate, compared with the two single preparation and 2) when wecombined the result of their pharmacological effects with the MPDCAresult that the jasminoidin group regulates more pathways in commonwith the combination group than theursodeoxycholic acid group, itmaylead to a conclusion that jasminoidin possibly contributes moreimportant pharmacological effect in the combined treatment.
In previous studies, it was reported that MAPK–ERK pathway, oneof the overlapping pathways among the three treatment groups,played a crucial role in ischemic brain injury (Ferrer et al., 2003; Jiaet al., 2008; Park et al., 2004). After administration of a specificMAPK–ERK pathway inhibitor, hippocampal damage would be reduced insome cerebral ischemic models, showing that the activation of MAPK-ERK pathway is, at least in part, involved in the neuronal deathassociated with these models of excitotoxic injury (Murray et al.,1998; Rundén et al., 1998). Although few studiesare reported aboutthe relationship between the Icam1-Ap1 or prolactin pathway andstroke, there still are some indications showing that ICAM1-AP1 andprolactin have been obviously increased after cerebral ischemic injury(Montecucco et al., 2010; Nijboer et al., 2009; Wallaschofski et al.,2006). Moreover, in one of our previous studies, we also found the
Fig. 6. Active patterns of core pathways under different treatment conditions. (A) Eight, seursodeoxycholic acid and combination groups, respectively. White and black vertical bars rerespectively.Oblique bars representoverlappingpathways. Only significantpathways (Pb0.01)this figure. (B) Western blot analysis and real-time PCR were employed to identify ERK2 and IL
level of ICAM-1 decreased after the administration of combinationtherapy (Zhang et al., 2003). As to KITLG pathways, it was reportedthat the KITLG signaling pathway had pleiotropic roles in adult andfetal life, because it was involved in human germ cell development(West et al., 2010), lymphocyte development (Chappaz et al., 2010),folliculogenesis of ovary (Yao and Ge, 2010), hematopoiesis, sper-matogenesis (Deshpande et al., 2010), melanogenesis (Kobi et al.,2010) and digestive tract motility (Abonia et al., 2010), and in anervous development experiment, it was suggested that as the Kitexpression was increased, the pigmented, dendritic cells developed(Opdecamp et al., 1997), indicating that KITLG might be involved inneurodevelopment. And in this study, after the administration ofjasminoidin, ursodeoxycholic acid and combination of both, theMAPK-ERK, KITLG, ICAM1-AP1 and prolactin pathways all have beenobservably regulated, suggesting that these four pathways might beinvolved in focal cerebral ischemia and the three preparations couldexert effects on stroke via these four pathways.
One of the major challenges of combination therapy and drug dis-covery is the poor understanding of the detailedmolecularmechanismsunderlying both disease progression and drug action. Differentialactivation of Akt suggests alternative pathways for cell death and/orsurvival in differentmodels (Papanicolaou et al., 1998) and implies thatdiverse pharmacological mechanisms and pathways contribute to theaction of different compounds on multiple pathways, such as the IL-6(Ortaldo et al., 2001), Src kinase (Yadav et al., 2005), JAK-STAT (France,2006), PI3K-Akt (Hetman et al., 1999), andMAPK-ERK (Nakazawa et al.,2002; Yona et al., 2006) pathways. A target buried in a network can bedescribed as a set of nodes and edges (Kioussi et al., 2006), so system-based approaches to future drug design should consider networkrobustness as the central framework (Bright and Mochly-Rosen, 2005).Our results suggest that any pathway alone is not sufficiently critical todefine the outcomes of brain injury and that several pathways should betargeted to optimize the therapeutic efficacy and to understand theintracellular signaling pathways that mediate the cellular responses tostroke. Recent advances in systems biology and pathway analysis canhelp make true rational design a reality by integrating experimentalobservations with underlying cellular regulation and metabolic net-works (Sivachenko and Yuryev, 2007). A practical application ofsystematic high-throughput screening is the identification of drug
ven and nine active pathways were obtained using PathwayStudio in the jasminoidin,present the contributed pathways in the jasminoidin and ursodeoxycholic acid groups,were listed.Note: thosenewpathwaysnot verified in theResNetdatabasewerenot listed in-6 pathways among different treatment groups. Ln (P value), natural logarithm of P value.
286 Z. Wang et al. / European Journal of Pharmacology 667 (2011) 278–286
combinations based on the interactions between the pathways onwhich they act in disease-relevant phenotypic assays. Mixtures ofinteracting compounds produced by plants may provide importantcombination therapies that simultaneously affect multiple pharmaco-logical targets and provide clinical efficacy beyond that of a singlecompound drug (Nikitin et al., 2003; Schmidt et al., 2007).
Acknowledgments
This work was supported by Hi-Tech Research and DevelopmentProgram of China (863), the National Natural Science Foundation ofChina (90209015) and the foundation of “Eleventh Five” National KeyTechnologies R&D Programmer (2006BAI08B04-06).
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.1016/j.ejphar.2011.05.046.
References
Abonia, J.P., Blanchard, C., Butz, B.B., Rainey, H.F., Collins, M.H., Stringer, K., Putnam, P.E.,Rothenberg, M.E., 2010. Involvement of mast cells in eosinophilic esophagitis.J. Allergy Clin. Immunol. 126, 140–149.
Allen, C.L., Bayraktutan, U., 2009. Oxidative stress and its role in the pathogenesis ofischaemic stroke. Int. J. Stroke 4, 461–470.
Bederson, J.B., Pitts, L.H., Tsuji, M., Nishimura, M.C., Davis, R.L., 1987. Rat middle cerebralartery occlusion: evaluation of the model and development of a neurologicexamination. Stroke 17, 472–476.
Black, D.R., Sang, C.N., 2005. Advances and limitations in the evaluation of analgesiccombination therapy. Neurology 65, S3–S6.
Bright, R., Mochly-Rosen, D., 2005. The role of protein kinase C in cerebral ischemic andreperfusion Injury. Stroke 36, 2781–2790.
Brito, M.A., Lima, S., Fernandes, A., Falcão, A.S., Silva, R.F.M., Butterfield, D.A., Brites, D.,2008. Bilirubin injury to neurons: contribution of oxidative stress and rescue byglycoursodeoxycholic acid. Neuro. Toxico. 29, 259–269.
Chan, M.T.V., Choi, K.C., Gin, T., Chui, P.T., Short, T.G., Yuen, P.M., PoonA, H.Y., Apfel, C.C.,Gan, T.J., 2006. The additive interactions between ondansetron and droperidol forpreventing postoperative nausea and vomiting. Anesth. Analg. 103, 1155–1162.
Chappaz, S., Gärtner, C., Rodewald, H.R., Finke, D., 2010. Kit ligand and I17 differentiallyregulate Peyer's patch and lymph node development. J. Immunol. 185, 3514–3519.
Chen, C., Gandhi, V., Plunkett, W., 2006. Sequential blockade strategy for the design ofcombination therapies to overcome oncogene addiction in chronic myelogenousleukemia. Cancer Res. 66, 10959–10966.
Chou, T.C., Talalay, P., 1981. Generalized equations for the analysis of inhibitions ofMichaelis–Menten and higher-order kinetic systems with two or more mutuallyexclusive and nonexclusive inhibitors. Eur. J. Biochem. 115, 207–216.
Deshpande, S., Agosti, V., Manova, K., Moore, M.A., Hardy, M.P., Besmer, P., 2010. Kitligand cytoplasmic domain is essential for basolateral sorting in vivo and has rolesin spermatogenesis and hematopoiesis. Dev. Biol. 337, 199–210.
Ferrer, I., Friguls, B., Dalfó, E., Planas, A.M., 2003. Early modifications in the expression ofmitogen-activated protein kinase (MAPK/ERK), stress-activated kinases SAPK/JNKand p38, and their phosphorylated substrates following focal cerebral ischemia.Acta Neuropathol. 105, 425–437.
France, D.S., 2006. The evolution of high-throughput screening in early oncology drugdiscovery. Am. Assoc. Cancer Res. Educ. Book 97–101.
Fuda, C., Suvorov, M., Vakulenko, S.B., Mobashery, S., 2004. The basis for resistance to β-lactam antibiotics by penicillin-binding protein 2a of methicillin-resistantStaphylococcus aureus. J. Biol. Chem. 279, 40802–40806.
Gaur, V.A., Kumar, A., 2009. Protective effect of naringin against ischemic reperfusioncerebral injury: possible neurobehavioral, biochemical and cellular alterations inrat brain. Eur. J. Pharmacol. 616, 147–154.
Hara, H., Huang, P.L., Panahian, N., Fishman, M.C., Moskowitz, M.A., 1996. Reduced brainedema and infarction volume in mice lacking the neuronal isoform of nitric oxidesynthase after transient MCA occlusion. J. Cereb. Blood Flow. Metabol. 16, 605–611.
Hetman, M., Kanning, K., Cavanaugh, J.E., Xia, Z., 1999. Neuroprotection by brain-derived neurotrophic factor is mediated by extracellular signal-regulated kinaseand phosphatidylinositol 3-kinase. J. Biol. Chem. 274, 22569–22580.
Hiroaki, K., 2007. A robustness-based approach to systems-oriented drug design. Nat.Rev. Drug Discov. 6, 202–210.
Jia, J., Chen, X., Zhu, W., Luo, Y., Hua, Z., Xu, Y., 2008. CART protects brain from damagethrough ERK activation in ischemic stroke. Neuropeptides 42, 653–661.
Jia, J., Zhu, F., Ma, X., Cao, Z.W., 2009. Mechanisms of drug combinations: interaction andnetwork perspectives. Nat. Rev. Drug Discov. 8, 111–128.
Kioussi, C., Shih, H., Loflin, J., Gross, M.K., 2006. Prediction of active nodes in thetranscriptional network of neural tube patterning. Proc. Natl. Acad. Sci. USA 103,18621–18626.
Kobi, D., Steunou, A.L., Dembélé, D., Legras, S., Larue, L., Nieto, L., Davidson, I., 2010.Genome-wide analysis of POU3F2/BRN2 promoter occupancy in human melanomacells reveals Kitl as a novel regulated target gene. Pigment Cell Melanoma Res. 23,404–418.
Liu, J., Yin, F., Zheng, X., Jing, J., Hu, Y., 2007. Geniposide, a novel agonist for GLP-1receptor, prevents PC12 cells from oxidative damage via MAP kinase pathway.Neurochem. Int. 51, 361–369.
Montecucco, F., Lenglet, S., Gayet-Ageron, A., Bertolotto, M., Pelli, G., Palombo, D., Pane,B., Spinella, G., Steffens, S., Raffaghello, L., Pistoia, V., Ottonello, L., Pende, A.,Dallegri, F., Mach, F., 2010. Systemic and intraplaque mediators of inflammation areincreased in patients symptomatic for ischemic stroke. Stroke 41, 1394–1404.
Murray, B., Alessandrini, A., Cole, A.J., Yee, A.G., Furshpan, E.J., 1998. Inhibition of thep44/42 MAP kinase pathway protects hippocampal neurons in a cell-culture modelof seizure activity. Proc. Natl. Acad. Sci. USA 95, 11975–11980.
Nakazawa, T., Tamai, M., Mori, N., 2002. Brain-derived neurotrophic factor preventsaxotomized retinal ganglion cell death through MAPK and PI3K signalingpathways. Invest. Ophthalmol. Vis. Sci. 43, 3319–3326.
Nijboer, C.H., Heijnen, C.J., Groenendaal, F., van Bel, F., Kavelaars, A., 2009. Alternatepathways preserve tumor necrosis factor-alpha production after nuclear factor-kappaB inhibition in neonatal cerebral hypoxia–ischemia. Stroke 40, 3362–3368.
Nikitin, A., Egorov, S., Daraselia, N., Mazo, I., 2003. Pathway studio—the analysis andnavigation of molecular networks. Bioinformatics 19, 2155–2157.
Opdecamp, K., Nakayama, A., Nguyen, M.T., Hodgkinson, C.A., Pavan, W.J., Arnheiter, H.,1997. Melanocyte development in vivo and in neural crest cell cultures: crucialdependence on the Mitf basic-helix–loop-helix-zipper transcription factor. Devel-opment 124, 2377–2386.
Ortaldo, J.R., Bere, E.W., Hodge, D., Young, H.A., 2001. Activating Ly-49 NK receptors:central role in cytokine and chemokine production. J. Immunol. 166, 4994–4999.
Papanicolaou, D.A., Wilder, R.L., Manolagas, S.C., Chrousos, G.P., 1998. The pathophys-iologic roles of interleukin-6 in human disease. Ann. Intern. Med. 128, 127–137.
Park, E.M., Joh, T.H., Volpe, B.T., Chu, C.K., Song, G., Cho, S., 2004. A neuroprotective roleof extracellular signal-regulated kinase in N-acetyl-O-methyldopamine-treatedhippocampal neurons after exposure to in vitro and in vivo ischemia. Neuroscience123, 147–154.
Perea, S., Gonzalez, G., Fothergill, A.W., Sutton, D.A., Rinaldi, M.G., 2002. In vitroactivities of terbinafine in combination with fluconazole, itraconazole, voricona-zole, and posaconazole against clinical isolates of Candida glabrata with decreasedsusceptibility to azoles. J. Clin. Microbiol. 40, 1831–1833.
Rundén, E., Seglen, P.O., Haug, F.M., Ottersen, O.P., Wieloch, T., Shamloo, M., Laake, J.H.,1998. Regional selective neuronal degeneration after protein phosphataseinhibition in hippocampal slice cultures: evidence for MAP kinase-dependentmechanism. J. Neurosci. 18, 7296–7305.
Samardzija, M., Wenzel, A., Aufenberg, S., Thiersch, M., Remé, C., 2006. Differential roleof Jak-STAT signaling in retinal degenerations. FASEB J. 20, 2411–2413.
Schmidt, B.M., Ribnicky, D.M., Lipsky, P.E., Raskin, I., 2007. Revisiting the ancientconcept of botanical therapeutics. Nat. Chem. Biol. 3, 360–366.
Sivachenko, A.Y., Yuryev, A., 2007. Pathway analysis software as a tool for drug targetselection, prioritization and validation of drug mechanism. Expert Opin. Therapeu.Targets 11, 411–421.
Sleight, P., Pouleur, H., Zannad, F., 2006. Benefits, challenges, and registerability of thepolypill. Eur. Heart J. 27, 1651–1656.
Stanika, R.I., Winters, C.A., Pivovarova, N.B., Andrews, S.B., 2010. Differential NMDAreceptor-dependent calcium loading and mitochondrial dysfunction in CA1 vs. CA3hippocampal neurons. Neurobiol. Dis. 37, 403–411.
Wallaschofski, H., Lohmann, T., Hild, E., Kobsar, A., Siegemund, A., Spilcke-Liss, E.,Hentschel, B., Stumpf, C., Daniel, W.G., Garlichs, C.D., Eigenthaler, M., 2006.Enhanced platelet activation by prolactin in patients with ischemic stroke. Thromb.Haemost. 96, 38–44.
West, F.D., Roche-Rios, M.I., Abraham, S., Rao, R.R., Natrajan, M.S., Bacanamwo, M., Stice,S.L., 2010. KIT ligand and bone morphogenetic protein signaling enhances humanembryonic stem cell to germ-like cell differentiation. Hum. Reprod. 25, 168–178.
Yadav, A., Kalita, A., Dhillon, S., Banerjee, K., 2005. JAK/STAT3 pathway is involved insurvival of neurons in response to insulin-like growth factor and negativelyregulated by suppressor of cytokine signaling. J. Biol. Chem. 280, 31830–31840.
Yao, K., Ge, W., 2010. Kit system in the zebrafish ovary: evidence for functionaldivergence of two isoforms of kit (kita and kitb) and kit ligand (kitlga and kitlgb)during folliculogenesis. Biol. Reprod. 82, 1216–1226.
Yasukawa, K., Iida, T., Fujimoto, Y., 2009. Relative inhibitory activity of bile acids against12-O-tetradecanoylphorbol-13-acetate-induced inflammation, and chenodeoxy-cholic acid inhibition of tumour promotion in mouse skin two-stage carcinogen-esis. J. Pharm. Pharmacol. 61, 1051–1056.
Zanelli, S.A., Solenski, N.J., Rosenthal, R.E., Fiskum, G., 2005. Mechanisms of ischemicneuroprotection by acetyl-L-carnitine. Ann. N. Y. Acad. Sci. 1053, 153–161.
Zhang, N., Zhu, X.L., Li, P.T., Jia, X., Jiang, Y.F., 2003. Effect of cholic acid and jasminoidinon contents of TNF-α, IL-1β and ICAM-1 in rat brain with injury of cerebralischemia and reperfusion. China Journal of Traditional Chinese Medicine andPharmacy 18, 463–465.
Zhang, Z.J., Li, P., Wang, Z., Li, P.T., Zhang, W.S., 2006. A comparative study on theindividual and combined effects of baicalin and jasminoidin on focal cerebralischemia–reperfusion injury. Brain Res. 1123, 188–195.
