Bioinformatic and data mining analysis of proteomic and genechip microarray data from an explant model of articular cartilage inflammation

关节软骨退化是人类和伴侣动物骨关节炎（OA）中关节功能障碍和残疾的主要原因，并导致慢性发病率、疼痛和生活质量受损。骨性关节炎是一种复杂的多因素疾病，我们正在研究疾病的早期阶段以及饮食/功能性活动在支持关节健康方面的可能作用。作为这项工作的一部分，我们最近开发并利用了马关节软骨的外植体培养模型，利用蛋白质组学和跨物种肽匹配技术来识别分泌的生物标志物。我们还使用人类Affymetrix基因芯片微阵列，通过将马关节软骨细胞的信使RNA交叉杂交到这个人类微阵列平台上，来研究软骨中的基因表达。这项工作是BBSRC与威豪和诺丁汉大学兽医与科学学院合作的一项正在进行的研究项目的成果，该项目始于2006年10月，刚刚进入第三个年头（BBSRC资助BS/S/M/2006/13141）。这种方法已经取得了丰硕的成果，并产生了大量的蛋白质组学和基因表达数据集，这些数据集需要通过生物信息学和数据挖掘技术进行详细的比较分析。这个CASE项目将应用数据挖掘和网络生物学方法来识别早期OA发展的潜在生物标志物，期望这项工作也将为软骨细胞对炎症扰动的反应提供见解。我们的目标是通过比较暴露于各种疾病相关分解代谢刺激的对照软骨样本和实验样本之间的差异来实现这一目标。我们建议使用的分解代谢刺激包括促炎细胞因子(即白细胞介素1 β (il -1 β)，白细胞介素6 （IL-6）和肿瘤坏死因子α (tnf - α))，活性氧，静态机械负荷和降低的细胞外ph。我们还将研究合成代谢生长因子如胰岛素样生长因子I (IGF-I)，转化生长因子β (tgf - β)，结缔组织生长因子（CTGF）和一组精心挑选的营养来源活性物质，以其在软骨移植系统中的假定抗炎特性而闻名。使用这种方法提取的蛋白质组学和基因表达数据将有助于识别具有区分对照和刺激外植体能力的生物标志物。我们将开发最合适的模型来区分样品，并确定在协议的哪个阶段，数据挖掘过程能够更好地识别手头区域的个体生物标志物，使用来自蛋白质组学和质谱实验的数据以及来自微阵列研究的差异表达基因。此外，生物信息学密集阶段将侧重于序列片段的表征，以确定肽片段来自哪些蛋白质，确定样品中存在的分泌蛋白质的潜在作用，最后推断蛋白质的生物学功能和相互作用网络。

Articular cartilage degradation is the major cause of joint dysfunction and disability in osteoarthritis (OA) in humans and companion animals and leads to chronic morbidity, pain and impaired quality of life. OA is a complex, multifactorial disease and we are investigating the early stages of disease and the possible role of diet/functional actives to support joint health. As part of this work, we have recently developed and exploited an explant culture model of equine articular cartilage to identify secreted biomarkers using proteomics and cross-species peptide matching techniques. We have also used human Affymetrix GeneChip Microarrays to study gene expression in cartilage by cross-hybridizing messenger RNA from equine articular chondrocytes onto this human microarray platform. This work is the result of an ongoing BBSRC CASE studentship in collaboration with WALTHAM and the School of Veterinary Medicine and Science at the University of Nottingham which started in October 2006 and has just entered its third year (BBSRC grant BS/S/M/2006/13141). This approach has been fruitful and is generating large amounts of proteomic and gene expression datasets that require detailed and comparative analysis by bioinformatics and data-mining techniques. This CASE project will apply data mining and network biology approaches to identify potential biomarkers of early OA development, with an expectation that this work will also provide insight into the response of chondrocyte to inflammatory perturbation. We aim to achieve this by comparing differences between control cartilage samples and experimental samples exposed to a variety of disease relevant catabolic stimuli. The catabolic stimuli we propose to use include pro-inflammatory cytokines (i.e. interleukin 1 beta (IL-1beta), interleukin 6 (IL-6) and tumour necrosis factor alpha (TNF-alpha)), reactive oxygen species, static mechanical load and reduced extracellular pH. We will also study the effects of anabolic growth factors such as insulin-like growth factor I (IGF-I), transforming growth factor beta (TGF-beta), connective tissue growth factor (CTGF) and a panel of carefully selected nutrient derived actives known for their putative anti-inflammatory properties in the cartilage explant system. The proteomic and gene expression data extracted using this approach will help identify biomarkers with the capacity to discriminate between control and stimulated explants. We will develop the most appropriate model to discriminate between samples and determine at which stage of the protocol the data mining process is better able to identify individual biomarkers for the domain at hand, using data from proteomic and mass spectroscopy experiments and differentially expressed genes from microarray studies. Additionally a bioinformatics-intensive stage will focus on the characterization of sequence fragments to identify which proteins the peptide fragments originate from, identify potential role(s) for the secreted proteins present in the samples and, finally, infer the biological functions and interaction networks of the proteins.