High Throughput Functional Dissection Of Adiposity GWAS Loci Using Model Systems

我们认为，人类肥胖遗传学领域已经停滞不前。大规模全基因组关联研究 数十万名受试者成功地识别了一千多部小说的统计基因座 毫无疑问，给肥胖风险增加的变量贴上标签。因为这些基因座几乎都在不是 在任何人的候选基因清单上，都有希望这些新的基因座指向可能导致 新的治疗方法和预防策略。但这些基因座到底做了什么，它们是如何做的，仍然是 对于几乎所有的地点来说，这仍然是一个谜。功能科学家在跟进这些发现时面临着挑战， 因为一场困难的“完美风暴”。基因组中的LD模式使精细定位和 即使在对多个区域进行深度测序后，也能从乘客的变种中统计出司机 因此，确切的因果变异很少为人所知。即使它们是已知的，它们的影响大小 每个变异都是临床上微不足道的，尽管它们具有压倒性的统计学意义。最糟糕的是，据估计 超过90%的人似乎在标记非编码区。我们认为，利用这些资源是一个巨大的限速步骤 新的发现继续从“统计基因”转移到基因(S)，而这些基因正是通过这个基因起作用的。许多 研究人员使用“最接近统计轨迹的基因”的默认注释作为起点，但如果我们的 已发表的实验(Baranski等人，2018年)是可以推广的，这可能有一半是错误的。 参考数据库中的Regulome注释正在成为一种关键资源(例如ENCODE和GTEx)，但 现在还为时尚早。非编码基因组是巨大的，而且许多注释要么缺乏，要么不足， 或者不够具体，不足以仅用这些资源从一个座位到另一个基因进行明确的定位。大队列 研究和TOPMed等财团已经开始在同一个人身上进行各种Omics扫描 发现了哪些基因座，从而为阐明潜在的机制提供了可能性 在这些统计基因的背后，特别是哪些基因，可能是关键基因。但几乎所有这些大型的 人类的努力实际上是必要的，仅限于全血和方便的组织，而很少有人这样做。 在假定的行动组织中是可能的。相比之下，肥胖症最重要的组织，如 大脑，可以在模型系统中访问和操纵，以阐明机制。很可能是这样的 基因座将有不同的生物学解释，但为每个基因座逐一进行功能实验 时间是具有挑战性的，既耗时又昂贵。我们需要的是一种高吞吐量战略，这将 同时审问多个地点。我们计划用我们成功的、发表的高吞吐量果蝇 在一组有苍蝇的人类候选者中有效筛选许多脂肪储存基因的系统 直系同源法(认识到这是一种不适用于所有基因座的“低挂果”方法)，然后 在小鼠身上验证那些从人类到昆虫保守的基因。

We believe that the field of human obesity genetics has stalled. Massive Genome-Wide Association studies of hundreds of thousands of subjects succeeded in identifying over a thousand novel “statistical loci” that unquestionably tag increased obesity risk variants. Because these loci are almost all in regions that were not on anyone’s candidate gene list, the promise has been that these novel loci point to new biology that could lead to new therapies and prevention strategies. But exactly what these loci do and how they do it continues to remain a mystery for almost all loci. Functional scientists have faced challenges following up on these findings, because of a “perfect storm” of difficulties. LD patterns in the genome complicate efforts to fine map and statistically dissect driver from passenger variants, even after deep sequencing of the regions in multiple ethnicities, so the exact causal variants are rarely known. Even if they were known, the effect sizes of these variants are each clinically trivial despite their overwhelming statistical significance. Worst of all, it is estimated that over 90% appear to be tagging non-coding regions. We believe a huge rate-limiting step to exploiting these new discoveries continues to be moving from “statistical loci” to the gene(s) through which they are acting. Many researchers use a default annotation of the “nearest gene” to the statistical locus as a starting point, but if our published experiments (Baranski et al., 2018) are generalizable, this may be wrong about half the time. Regulome annotation in reference databases is emerging as a critical resource (e.g. ENCODE and GTEX), but it is still early days. The non-coding genome is huge, and much of the annotation is either lacking, insufficient, or not specific enough to allow definitive mapping from locus to gene with these resources alone. Large cohort studies and consortia such as TOPMed have begun conducting various Omics scans in the same individuals in which locus discoveries were made, thus providing the potential to shed light on the underlying mechanisms behind the statistical loci, and in particular, suggest which genes, might be critical. But almost all of these large human efforts are of practical necessity limited to whole blood and tissues of convenience, and much less has been possible in the presumed tissues of action. By contrast, the most important tissues for obesity, like the brain, can be accessed and manipulated in model systems to shed light on mechanisms. Likely every such locus will have a different biological explanation, but pursuing functional experiments for each locus one at a time is challenging, time consuming and expensive. What is needed is a high throughput strategy, that will interrogate many loci simultaneously. We propose to use our successful, published high throughput Drosophila system to efficiently screen for many fat storage genes among the set of human candidates that have fly orthologs (recognizing that this is a “low hanging fruit” approach that will not work for every locus), and then validate in the mouse those that are conserved from human to insect.