The OPP samples used in this study were prepared according to the methods described in Sambanthamurthi et al. (2008) . OPP contains numerous phenolic acids, with three isomers of caffeoylshikimic acid as major components [20, 21]. Other phenolic acids present include protocatechuic acid and p-hydroxybenzoic acid. The detailed composition of OPP is as described earlier .
Animals, diets and treatments
All male inbred BALB/c mice which were designated for this study (n = 20) were purchased from the Institute of Medical Research, Kuala Lumpur, Malaysia, at around five weeks of age just after weaning. All animal procedures were approved by the Animal Care and Use Committee of the University of Malaya, Kuala Lumpur, Malaysia. The animals were randomly assigned into cages (n = 5 per cage) and acclimatised for one week, during which a standard chow diet purchased from the University of Malaya and distilled water, were given. At the start of the experiment, the diet of the animals was changed to a custom-made normal diet (58.2% kcal/kcal carbohydrate, 27.2% kcal/kcal protein and 14.6% kcal/kcal fat, including cellulose, mineral mix, vitamin mix and DL-methionine).
The control group (n = 10) was supplemented with distilled water while the treatment group (n = 10) was supplemented with OPP, as drinking fluids ad libitum. The antioxidant content of the OPP given was 1500 ppm gallic acid equivalent. Food and fluids were changed daily. During the animal feeding process, body weights were monitored every week, while fluid intake was monitored every day, for six weeks. Food intake and faecal output were monitored for seven consecutive days between week two to week three.
After six weeks, the mice were sacrificed via euthanasia with diethyl ether followed by exsanguination. Blood samples were collected after an overnight food fast (with fluids still provided) via cardiac puncture, while three major organs including livers, spleens and hearts were excised, rinsed in 0.9% w/v sodium chloride solution and weighed. Half of the organs (n = 5) were preserved in 10% formalin diluted with phosphate buffered saline for histology analysis, while the other half (n = 5) were snap-frozen in liquid nitrogen and stored at -80°C until the total RNA extraction process for gene expression analysis.
Haematology analysis on blood samples
About 200 μL from half of each blood sample collected (n = 4) was aliquoted into a blood collection tube containing ethylenediaminetetraacetic acid to prevent clotting. These whole blood samples were sent immediately after dissection of the animals to the Clinical Biochemistry and Haematology Laboratory, Department of Veterinary Pathology and Microbiology, Faculty of Veterinary Medicine, University of Putra Malaysia, Serdang, Selangor, Malaysia, for haematology analysis. The analysis was carried out using the Animal Blood Counter Vet Haematology Analyser (Horiba ABX, France).
Clinical biochemistry analysis on blood samples
In order to obtain sera, blood samples were allowed to clot at room temperature for two hours before being centrifuged at 1000 xg for five minutes, after which the supernatant layers were collected and stored at -20°C. The blood serum samples obtained were then sent for clinical biochemistry analysis using the Roche/Hitachi 902 Chemistry Analyser (Roche/Hitachi, Switzerland) in the Clinical Biochemistry and Haematology Laboratory, Department of Veterinary Pathology and Microbiology, Faculty of Veterinary Medicine, University of Putra Malaysia. Clinical biochemistry parameters which were examined include alanine aminotransferase, aspartate aminotransferase, glucose, serum total protein, albumin, globulin, ratio of albumin to globulin, total cholesterol, triglycerides, low-density lipoproteins and high-density lipoproteins. Of a total of ten serum samples per group obtained for clinical biochemistry analysis, two samples in the control group and three samples in the treatment group were excluded due to blood lysis.
Histology analysis on organs
Histology analysis on dissected organs (n = 5) preserved in 10% buffered formalin was carried out in the Department of Pathology, Faculty of Medicine, University of Malaya, using standard routine procedures with haematoxylin and eosin staining.
Total RNA extraction for gene expression analysis
All precautions in handling RNA were taken in this study. Total RNA isolation from mouse organs (n = 5) was carried out using the RNeasy Mini Kit (Qiagen, Inc., Valencia, CA) and QIAshredder homogenizers (Qiagen, Inc., Valencia, CA), preceded by grinding in liquid nitrogen using mortars and pestles. The total RNA samples obtained were subjected to NanoDrop 1000A Spectrophotometer (Thermo Fisher Scientific, Waltham, MA) measurement for yield and purity assessment. Integrity of the total RNA samples was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and Agilent RNA 6000 Nano Chip Assay Kit (Agilent Technologies, Santa Clara, CA). Four total RNA samples with the highest RNA Integrity Numbers and 28S/18S rRNA ratios within each condition (either control or treatment) were then selected for microarray studies.
Microarray hybridisation, washing and scanning
Amplification of total RNA samples which were of high yield, purity and integrity was carried out using the Illumina TotalPrep RNA Amplification Kit (Ambion, Inc., Austin, TX). The cRNA produced was then hybridised to the Illumina MouseRef-8 Version 1 Expression BeadChip (Illumina, Inc., San Diego, CA), using the Direct Hybridization Kit (Illumina, Inc., San Diego, CA). Microarray hybridisation, washing and scanning were carried out according to the manufacturer's instructions. The raw gene expression data obtained are available at Gene Expression Omnibus  (Accession number: GSE28824).
Microarray data analysis
Quality control of the hybridisation, microarray data extraction and initial analysis were carried out using the Illumina BeadStudio software (Illumina, Inc., San Diego, CA). Outlier samples were removed via hierarchical clustering analysis provided by the Illumina BeadStudio software and also using the TIGR MeV software (Institute for Genomic Research, Rockville, MD) , via different distance metrics. A minimum of three biological replicates per condition (with outliers removed) was then considered for further analysis. For livers, four replicates in the control group and three replicates in the treatment group were analysed. For spleens and hearts respectively, four replicates in the control group and four replicates in the treatment group were analysed.
Gene expression values were normalised using the rank invariant method and genes which had Detection Levels of more than 0.99 in either the control or treatment samples were considered significantly detected. To filter the data for genes which changed significantly in terms of statistics, the Illumina Custom error model was used and genes were considered significantly changed at a |Differential Score| of more than 20, which was equivalent to a P Value of less than 0.01 .
The genes and their corresponding data were then exported into the Microsoft Excel software (Microsoft Corporation, Richmond, WA) for further analysis. To calculate fold changes, an arbitrary value of 10 was given to expression values which were less than 10. Fold changes were then calculated by dividing the mean values of Signal Y (treatment) with those of Signal × (control) if the genes were up-regulated and vice versa if the genes were down-regulated. Two-way (gene and sample) hierarchical clustering of the significant genes was then performed using the TIGR MeV software to ensure that the replicates of each condition were clustered to each other. The Euclidean distance metric and average linkage method were used to carry out the hierarchical clustering analysis.
Changes in biological pathways and gene ontologies were assessed via functional enrichment analysis, using the GenMAPP  and MAPPFinder  softwares (University of California at San Francisco, San Francisco, CA). The MAPPFinder software ranks GenMAPPs (pathways) and gene ontologies based on the hypergeometric distribution. Readers are referred to Doniger et al. (2003)  for further explanations of the terms used in the MAPPFinder software. GenMAPPs and gene ontologies which had Permuted P Values of less than 0.01, Numbers of Genes Changed of more than or equal to 2 and Z Scores of more than 2 were considered significant.
It should be noted that the MAPPFinder software clusters multiple probes for a distinct gene into a single gene grouping in order to calculate the number of distinct genes which meet the user-defined criteria, not the probes. In this study, up-regulated and down-regulated genes were analysed separately in the functional enrichment analysis but were viewed together in each GenMAPP. Boxes coloured yellow indicate genes which were up-regulated while those coloured blue indicate genes which were down-regulated. The fold changes are indicated next to the boxes. Individual boxes which have different shadings within them indicate the presence of multiple probes (splice transcripts) within a single gene.
Changes in regulatory networks were also analysed through the use of the Ingenuity Pathways Analysis software (Ingenuity® Systems, Redwood City, CA) . For each organ, a dataset containing differentially expressed genes and their corresponding fold changes was uploaded into the application. Analyses of up-regulated and down-regulated genes were carried out separately. Each gene identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base. These genes were overlaid onto a global molecular network developed from information contained in the Ingenuity Pathways Knowledge Base. Networks of these focus genes were then algorithmically generated based on their connectivity.
A network is a graphical representation of the molecular relationships between genes or gene products. Genes or gene products were represented as nodes, and the biological relationship between two nodes was represented as an edge (line). The intensity of the node colour indicates the degree of up-regulation (red) or down-regulation (green). Nodes were displayed using various shapes that represented the functional class of the gene product. Edges were displayed with various labels that described the nature of the relationship between the nodes.
Real-time qRT-PCR validation
Two-step real-time qRT-PCR studies were carried out on six target genes selected to represent the different organs used in the microarray experiments, using TaqMan Gene Expression Assays (Applied Biosystems, Foster City, CA). These genes were selected based on their presence in significant functions, their differential scores, their detection levels, their presence as single splice transcripts in microarrays and their availability as Taqman assays designed across splice junctions. The same aliquots of total RNA samples used in the microarray experiments were utilised for this analysis. Primer and probe sets for the selected genes were obtained from the Applied Biosystems Inventoried Assays-On-Demand (Applied Biosystems, Foster City, CA).
Briefly, reverse transcription to generate first-strand cDNA from total RNA was carried out using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Real-time PCR was then carried out on the first-strand cDNA generated using a 25 μL reaction volume in an Applied Biosystems 7000 Real-Time PCR System (Applied Biosystems, Foster City, CA), with the following conditions: 50°C, 2 minutes, 1 cycle; 95°C, 10 minutes, 1 cycle; 95°C, 15 seconds and 60°C, 1 minute, 40 cycles. For gene expression experiments, reactions for each biological replicate and non-template control (NTC) were carried out in duplicates. For amplification efficiency determination, reactions were carried out in triplicates.
Quality control of the replicates used, real-time qRT-PCR data extraction and initial analysis were carried out using the 7000 Sequence Detection System software (Applied Biosystems, Foster City, CA). A manual threshold of 0.6000 and an auto baseline were applied in order to obtain the threshold cycle (Ct) for each measurement taken. The threshold was chosen as it intersected the exponential phase of the amplification plots . The criteria for quality control of the data obtained include ΔCt of less than 0.5 between technical replicates and ΔCt of more than 5.0 between samples and NTCs .
Relative quantification of the target genes of interest was carried out using the qBase 1.3.5 software (Center for Medical Genetics, Ghent University Hospital, Ghent, Belgium) , which takes into account the calculations of amplification efficiencies and multiple housekeeping genes. Expression levels of target genes were normalised to the geometric mean of three housekeeping genes, arginine/serine rich splicing factor 9 (Sfrs9), guanylate kinase 1 (Guk1) and heterogeneous nuclear ribonucleoprotein A/B (Hnrpab). These genes were chosen as they were shown to be stable across the previously obtained microarray data, and confirmed to be more stable than 18S ribosomal RNA (18SrRNA), via the usage of the geNorm 3.5 software (Center for Medical Genetics, Ghent University Hospital, Ghent, Belgium) .
Statistical analysis was carried out by using the two-tailed unpaired Student's t-test available in the Microsoft Excel software (Microsoft Corporation, Redmond, WA) unless otherwise stated. Differences with P Values of less than 0.05 were considered statistically significant.