The impact of copy number variation on local gene expression in mouse hematopoietic stem and progenitor cells

The extent to which differences in germ line DNA copy number contribute to natural phenotypic variation is unknown. We analyzed the copy number content of the mouse genome to a sub-10 kb resolution. We identified over 1,300 copy number variant regions (CNVRs), most of which are < 10 kb in length, are found in more than one strain, and, in total, span 3.2% (85 Mb) of the genome. To assess the potential functional impact of copy number variation, we mapped expression profiles of purified hematopoietic stem and progenitor cells, adipose tissue and hypothalamus to CNVRs in cis. Of the more than 600 significant associations between CNVRs and expression profiles, most map to CNVRs outside of the transcribed regions of genes. In hematopoietic stem/progenitor cells, up to 28% of strain-dependent expression variation is associated with copy number variation, supporting the role of germ line CNVs as major contributors to natural phenotypic variation in the laboratory mouse. To map the CNV content of the mouse genome, we selected 17 Tier 1-3 Mouse Phenome Project strains and three additional strains of biomedical interest, representing all major inbred lineages. We performed comparative genomic hybridization using a long-oligonucleotide array containing 2,149,887 probes evenly spaced across the reference genome with a median inter-probe spacing of 1,015 bases. Labeling, hybridization, washing and array imaging were performed as previously described (PMID:16075461). We performed segmentation using wuHMM, a Hidden Markov Model algorithm that utilizes sequence-level information and can detect CNVs less than 5 kb in length (fewer than five probes) at a low false positive rate (PMID:18334530). To estimate the overall impact of CNV on gene expression in vivo, we performed expression profiling of hematopoietic stem/progenitors cells using the Illumina Mouse Beadchip-6v1 platform. See manuscript for further details.种系DNA拷贝数差异对自然表型变异的贡献程度目前尚不明确。我们以低于10 kb的分辨率分析了小鼠基因组的拷贝数特征，共鉴定出1300余个拷贝数变异区域（copy number variant regions, CNVRs），其中多数长度小于10 kb，且在多个品系中均有分布，这些CNVRs总跨度占小鼠基因组的3.2%（85 Mb）。为评估拷贝数变异的潜在功能影响，我们将纯化的造血干/祖细胞、脂肪组织及下丘脑的表达谱与顺式作用的CNVRs进行了关联定位。在CNVR与表达谱之间的600余个显著关联中，多数关联的CNVRs位于基因转录区域之外。在造血干/祖细胞中，高达28%的品系依赖性表达变异与拷贝数变异相关，这证实了种系CNVs是实验室小鼠自然表型变异的主要贡献因素之一。为定位小鼠基因组的CNV特征，我们选取了17个Tier 1-3级小鼠表型组计划（Mouse Phenome Project）品系，以及3个具有重要生物医学研究价值的额外品系，这些品系涵盖了所有主要的近交系谱系。我们采用包含2,149,887个探针的长寡核苷酸芯片开展比较基因组杂交实验，这些探针在参考基因组上均匀分布，探针间的中位间距为1015个碱基。标记、杂交、洗涤及芯片成像流程均参照已发表文献（PMID:16075461）完成。我们采用wuHMM算法进行片段划分：该算法是一种隐马尔可夫模型（Hidden Markov Model, HMM），可利用序列层面信息，能够以较低的假阳性率检测到长度小于5 kb（探针数少于5个）的CNVs，相关方法已在文献PMID:18334530中详述。为评估CNV在体内对基因表达的整体影响，我们采用Illumina Mouse Beadchip-6v1平台对造血干/祖细胞开展了表达谱分析。更多细节请参见研究论文正文。