Kemper K E, Littlejohn M D, Lopdell T, Hayes B J, Bennett L E, Williams R P, Xu X Q, Visscher P M, Carrick M J, Goddard M E
Faculty of Veterinary and Agricultural Sciences, University of Melbourne, Royal Parade, Parkville, Victoria, 3052, Australia.
Livestock Improvement Corporation, Cnr Ruakura and Morrinsville Roads, Newstead, Hamilton, 3240, New Zealand.
BMC Genomics. 2016 Nov 3;17(1):858. doi: 10.1186/s12864-016-3175-3.
Polymorphisms underlying complex traits often explain a small part (less than 1 %) of the phenotypic variance (σ). This makes identification of mutations underling complex traits difficult and usually only a subset of large-effect loci are identified. One approach to identify more loci is to increase sample size of experiments but here we propose an alternative. The aim of this paper is to use secondary phenotypes for genetically simple traits during the QTL discovery phase for complex traits. We demonstrate this approach in a dairy cattle data set where the complex traits were milk production phenotypes (fat, milk and protein yield; fat and protein percentage in milk) measured on thousands of individuals while secondary (potentially genetically simpler) traits are detailed milk composition traits (measurements of individual protein abundance, mineral and sugar concentrations; and gene expression).
Quantitative trait loci (QTL) were identified using 11,527 Holstein cattle with milk production records and up to 444 cows with milk composition traits. There were eight regions that contained QTL for both milk production and a composition trait, including four novel regions. One region on BTAU1 affected both milk yield and phosphorous concentration in milk. The QTL interval included the gene SLC37A1, a phosphorous antiporter. The most significant imputed sequence variants in this region explained 0.001 σ for milk yield, and 0.11 σ for phosphorus concentration. Since the polymorphisms were non-coding, association mapping for SLC37A1 gene expression was performed using high depth mammary RNAseq data from a separate group of 371 lactating cows. This confirmed a strong eQTL for SLC37A1, with peak association at the same imputed sequence variants that were most significant for phosphorus concentration. Fitting any of these variants as covariables in the association analysis removed the QTL signal for milk production traits. Plausible causative mutations in the casein complex region were also identified using a similar strategy.
Milk production traits in dairy cows are typical complex traits where polymorphisms explain only a small portion of the phenotypic variance. However, here we show that these mutations can have larger effects on secondary traits, such as concentrations of minerals, proteins and sugars in the milk, and expression levels of genes in mammary tissue. These larger effects were used to successfully map variants for milk production traits. Genetically simple traits also provide a direct biological link between possible causal mutations and the effect of these mutations on milk production.
复杂性状背后的多态性通常只能解释一小部分（不到1%）的表型变异（σ）。这使得鉴定复杂性状背后的突变变得困难，通常只能鉴定出一小部分具有大效应的基因座。一种鉴定更多基因座的方法是增加实验样本量，但在这里我们提出了另一种方法。本文的目的是在复杂性状的QTL发现阶段，将二级表型用于遗传上简单的性状。我们在一个奶牛数据集中展示了这种方法，其中复杂性状是对数千头牛测量的产奶表型（脂肪、牛奶和蛋白质产量；牛奶中的脂肪和蛋白质百分比），而二级（潜在遗传上更简单）性状是详细的牛奶成分性状（个体蛋白质丰度、矿物质和糖浓度的测量；以及基因表达）。
利用11527头有产奶记录的荷斯坦奶牛和多达444头有牛奶成分性状的奶牛鉴定了数量性状基因座（QTL）。有八个区域同时包含产奶和一种成分性状的QTL，包括四个新区域。BTAU1上的一个区域同时影响牛奶产量和牛奶中的磷浓度。QTL区间包含基因SLC37A1，一种磷转运体。该区域最显著的推测序列变异对牛奶产量解释了0.001σ，对磷浓度解释了0.11σ。由于这些多态性是非编码的，因此使用来自另一组371头泌乳奶牛的深度乳腺RNAseq数据对SLC37A1基因表达进行了关联作图。这证实了SLC37A1有一个很强的表达数量性状基因座（eQTL），其峰值关联位于对磷浓度最显著的相同推测序列变异处。在关联分析中将这些变异中的任何一个作为协变量进行拟合，都消除了产奶性状的QTL信号。还使用类似策略在酪蛋白复合区域鉴定了可能的致病突变。
奶牛的产奶性状是典型的复杂性状，其中多态性仅解释了一小部分表型变异。然而，在这里我们表明，这些突变对二级性状可能有更大的影响，如牛奶中矿物质、蛋白质和糖的浓度，以及乳腺组织中基因的表达水平。这些更大的影响被成功用于定位产奶性状的变异。遗传上简单的性状还在可能的因果突变与这些突变对产奶的影响之间提供了直接的生物学联系。
