Integrative genetics of behavior with high throughput technologies

功能变异的识别是遗传影响疾病分析的最终阶段，也是理解基因在行为中的作用的起点。公共数据库包含超过 2200 万个序列变异，其中大部分是单核苷酸多态性。然而，大多数罕见和不常见的变体是未知的，并且大多数的功能也是未知的。通过体内和体外功能基因组学研究的结合，我们发现了几个功能位点，并证明了它们在与酗酒和成瘾相关的复杂行为中的作用。 我们发现或帮助定义了一系列改变行为的功能多态性的体外和体内效应。在酗酒中，OPRM1 Asn40Asp 错义变体（Bergen 等人，1997）被其他人证明是有功能的，并且与纳曲酮治疗反应相关（Anton 等人，2008）。血清素转运蛋白 HTTLPR 位点中常见的功能性 LA->LG SNP（Hu 等人）增强了与行为和中间表型的联系。这些包括抑郁症患者的 SSRI 反应（Hu 等，2007）、对情绪的神经影像学反应以及导致自杀的基因 x 应激相互作用（Roy 等，2007）。 检测到常见的 HTR2C Ser23Cys（Lappalainen 等人）和 HTR2A Asn452His 等位基因（Ozaki 等人）并显示其功能，并被其他人与行为（包括氯氮平反应）联系起来。在首发精神分裂症患者中，我们发现功能性 DRD2 启动子多态性影响抗精神病药反应（Lencz 等，2006）。我们追踪了 NPY（Zhu 等人，2008）、GCH1（Tegeder 等人，2006）和 DISC1（以 Hodgkinson 等人，2004 年开始的一系列论文）的功能位点和单倍型与各种行为（包括情绪、精神分裂症和临床疼痛结果）的联系。通过深度测序，我们发现了一个 HTR2B 终止密码子，该密码子在芬兰人中相对常见，在其他人群中不存在，并且与冲动行为相关但不是决定因素。这个终止密码子也与家族中的冲动行为共分离，基因敲除的小鼠更冲动，寻求新奇的能力更高。作为使用本段中列出的多态性进行的分子研究类型的一个例子，HTR2B 终止密码子导致 HTR2B RNA 的可变无义介导的衰变并阻断受体的表达（Bevilacqua 等人，Nature，2010）。 低表达神经肽 Y (NPY) 单倍型会增加焦虑和情绪，但对分子（NPY RNA 和蛋白质）和中间表型（对情绪刺激和疼痛/压力反应的大脑成像测量）的影响比复杂行为更强。 （周等人，《自然》，2008）。 中间表型增强了行为的遗传分析。我们帮助发起的成像遗传学范式（Heinz 等人，2000 年；Hariri 等人，2002 年；Egan 等人，2001 年；2003 年；Zubieta 等人，2003 年）不断产生突破性的发现。我们与 NIDA 的合作者一起发现，CHRNA5 Asn398（尼古丁 GWAS 的主要功能位点）的作用机制涉及背侧前扣带回/腹侧纹状体环路的减弱，该环路的减弱预示着有或没有其他精神疾病诊断的吸烟者对尼古丁的渴望（Hong 等，2010）。临床亚表型分析使 HTR1B 与反社会酒精中毒（Lappalainen 等人）、血清素转运蛋白（SLCA4）与焦虑（Mazzanti 等人；Hariri 等人）、BDNF Val66Met 与情景记忆（Egan 等人）、COMT Val158Met 与焦虑（Enoch 等人）、执行认知（Egan 等人；Lipsky 等人；Malhotra 等人）和疼痛阈值联系起来（Zubieta 等人；Diatchenko 等人）和 GTP 环化水解酶对慢性疼痛和实验性疼痛反应（Tegeder 等人）。在这些研究中，大脑成像和认知测量发挥着重要作用。额叶认知缺陷是精神分裂症、酗酒和其他疾病的危险因素。多巴胺通常会增强前额皮质的效率。 Met158 是一种常见的 COMT 变体，可导致 COMT 活性降低四倍。因此，它是通过影响额叶多巴胺而发挥认知功能的候选等位基因。我们发现 Met158 与额叶认知功能和额叶皮质效率降低之间存在等位基因剂量关系（Egan 等人）。在基线认知功能不同的人群中观察到与认知的关系：精神分裂症、中度至重度头部损伤（Lipsky 等人）和对照组（Malhotra 等人）。我们提出 Val158 有一个反优势：压力恢复能力。在两个人群中，Met158 预测女性焦虑并降低额叶脑电图一致性（Enoch 等人），并且 Met158 与较低的疼痛/压力弹性相关（Zubieta 等人；Diatchenko 等人）。 Met158 预测疼痛/压力后无法激活内吗啡释放（Zubieta 等人）。总体而言，中间表型中基因的影响大小较大（Goldman 和 Ducci，2007），并且如 NPY（Zhou 等人，2008）、COMT（Zubieta 等人，2003）和 CHRNA5（Hong 等人，2010）等基因对中间表型的影响所说明。 Linkage、gwas 和 RNA-seq 是基因组、无假设方法的代表。在我们通过家族连锁暗示的 Chr 4 GABAA 亚基簇中发现了酗酒与 GABRA2 的连锁不平衡（Long 等人）。我们证明这种关联是焦虑调节的（Enoch 等人）。另一个与酒精中毒和酒精反应有关的 GABAA 基因簇位于 Chr 5。在这个基因簇中，我们报道了与酒精中毒的连锁不平衡（Radel 等人），并且 GABRA6 具有与酒精依赖以及对酒精和地西泮的反应相关的错义变异（Iwata 等人、Schuckit 等人）。家族连锁扫描发现 CRH-BP（一种压力相关蛋白）与酗酒相关的脑电图特征在全基因组范围内显着存在关联，随后在两个人群中发现了 CRH-BP 与脑电图的关联。 （以诺等人）。我们的脑电图 gwas（Hodgkinson 等人，PNAS，2010）在相对较小的平原印第安人样本中检测到了三个独立的全基因组范围内的重要基因座，并且我们在第二个人群中复制了两个基因座（包括一个亚阈值基因座）（Hodgkinson 等人，PNAS 2010）。我们最近使用新一代测序来分析恒河猴早期母性剥夺导致的大脑组蛋白甲基化和转录组变化（Barr等人，正在准备中）以及慢性可卡因和酒精对人脑的影响（Zhou等人，PNAS 2010）。后一项研究发现，与动物模型中急性和亚急性暴露中检测到的基因网络相比，人类慢性药物暴露调节的基因网络之间存在明显差异。 创始人群体、极端先证者和暴露群体增强了检测基因效应的能力。我们的芬兰数据集源自创始人人群，并从犯罪酗酒先证者中确定，从而丰富了 II 型早发性酗酒。西南和平原印第安人样本代表了创始人群体。一个具有高逆境暴露的非裔美国人物质依赖样本揭示了童年逆境的 GxE 和自杀倾向的 HTTLPR（Roy 等，2007）。压力和贫困，而非非洲血统，预示着成瘾的高风险（Ducci et al, 2009）。 MAOA VNTR 先前通过压力相互作用与失控有关，但与美国印第安妇女的酒精依赖和 ASPD 的结果有关，其中一半人在儿童时期遭受过性虐待（Ducci 等，2008）。在芬兰犯罪样本中观察到 MAOA VNTR 和睾酮之间存在强烈的相互作用（Sjoberg 等人）。如前所述，芬兰人的深度测序检测到了对冲动行为具有重要意义的群体特异性 HTR2B 终止密码子（Bevilacqua 等，Nature，2010）。

Identification of functional variants is an end-game of analysis of genetically influenced diseases and a starting point to understand roles of genes in behaviors. Public databases are populated with >22 million sequence variants, mostly single nucleotide polymorphisms. However, most rare and uncommon variants are unknown, and functionality of most is unknown. Through a combination of in-vivo and in vitro functional genomics studies we have discovered several functional loci and demonstrated their roles in complex behaviors relevant to alcoholism and addictions. We discovered or helped define in vitro and in vivo effects of a series of functional polymorphisms altering behavior. In alcoholism, an OPRM1 Asn40Asp missense variant (Bergen et al, 1997) was shown by others to be functional and linked to naltrexone treatment response (Anton et al, 2008). A common, functional LA->LG SNP in the serotonin transporter HTTLPR locus (Hu et al) enhanced linkage to behaviors and intermediate phenotypes. These include SSRI response of depressed patients (Hu et al, 2007), neuroimaging responses to emotion, and gene x stress interactions leading to suicidality (Roy et al, 2007). Common HTR2C Ser23Cys (Lappalainen et al) and HTR2A Asn452His alleles (Ozaki et al) were detected and shown to be functional, and linked by others to behavior, including clozapine response. In first-episode schizophrenics, we found that a functional DRD2 promoter polymorphism influences antipsychotic response (Lencz et al, 2006). We traced linkages of functional loci and haplotypes of NPY (Zhu et al, 2008), GCH1 (Tegeder et al, 2006) and DISC1 (a series of papers beginning with Hodgkinson et al, 2004) to various behaviors including emotionality, schizophrenia and clinical pain outcome. By deep sequencing we found an HTR2B Stop codon that is relatively common in Finns, absent in other populations, and associated with but not determinant for impulsive behavior. This stop codon also cosegregated with impulsive behavior in families, and the mouse gene knockout was more impulsive and higher in novelty seeking. As an example of the type of molecular studies performed with polymorphisms listed in this paragraph, the HTR2B stop codon led to variable nonsense mediated decay of the HTR2B RNA and blocked expression of the receptor (Bevilacqua et al, Nature, 2010). A low expression Neuropeptide Y (NPY) haplotype increased anxiety and emotionality but had stronger effects on molecules (NPY RNA and protein) and intermediate phenotypes (brain imaging measures of responses to emotional stimuli and pain/stress) than complex behavior. (Zhou et al, Nature, 2008). Intermediate phenotypes augment genetic analyses of behavior. The imaging genetics paradigm we helped initiate (Heinz et al, 2000; Hariri et al, 2002; Egan et al, 2001; 2003; Zubieta et al, 2003) has continued to yield groundbreaking findings. With collaborators at NIDA, we helped discover that the mechanism of action of CHRNA5 Asn398, the major functional locus from nicotine GWAS, involves weakening of a Dorsal Anterior Cingulate/Ventral Striatal circuit, the weakness of which predicts nicotine craving in smokers with and without other psychiatric diagnoses (Hong et al, 2010). Clinical subphenotyping enabled linkage of HTR1B to antisocial alcoholism (Lappalainen et al), serotonin transporter (SLCA4) to anxiety (Mazzanti et al; Hariri et al), BDNF Val66Met to episodic memory (Egan et al), COMT Val158Met to anxiety (Enoch et al), executive cognition (Egan et al; Lipsky et al; Malhotra et al), and pain threshold (Zubieta et al; Diatchenko et al), and GTP cyclohydrolase to chronic pain and experimental pain response (Tegeder et al). In these studies brain imaging and cognitive measures play prominent roles. Frontal cognitive deficit is a risk factor in schizophrenia, alcoholism and other diseases. Dopamine generally enhances prefrontal cortical efficiency. Met158, a common COMT variant, leads to four-fold reduction in COMT activity. It is thus a candidate allele for cognitive function via effect on frontal dopamine. We found an allele-dosage relationship of Met158 to frontal cognitive function and diminished frontal cortical efficiency (Egan et al). The relationship to cognition is observed in populations differing in baseline cognitive function: schizophrenia, moderate-severe head injury (Lipsky et al), & controls (Malhotra et al). We proposed that Val158 has a counter-advantage: stress resiliency. In two populations Met158 predicted anxiety in women and decreased frontal EEG coherence (Enoch et al), and Met158 was associated with lower resiliency to pain/stress (Zubieta et al; Diatchenko et al). Met158 predicted inability to activate endomorphin release after pain/stress (Zubieta et al). Overall, effect sizes of genes in intermediate phenotypes is larger (Goldman and Ducci, 2007) and as illustrated by effects of genes such as NPY (Zhou et al, 2008), COMT (Zubieta et al, 2003) and CHRNA5 (Hong et al, 2010) on intermediate phenotypes. Linkage, gwas and RNA-seq are representative of genomic, hypothesis-free approaches. Linkage disequilibrium of alcoholism to GABRA2 was found at the Chr 4 GABAA subunit cluster we implicated by family linkage (Long et al). We showed the association was anxiety- modulated (Enoch et al). Another GABAA gene cluster implicated in alcoholism and alcohol response is located on Chr 5. Within this cluster we reported linkage disequilibrium to alcoholism (Radel et al) and that GABRA6 has a missense variant associated with alcohol dependence and response to alcohol and diazepam (Iwata et al, Schuckit et al). A family linkage scan yielded genome-wide significant linkage of CRH-BP (a stress-related protein) to an alcoholism-associated EEG trait and this was followed by association of CRH-BP to EEG in two populations. (Enoch et al). Our gwas of EEG (Hodgkinson et al, PNAS, 2010) in a relatively small sample of Plains Indians detected three independent genome-wide significant loci, and we replicated two loci (including one locus that was just sub-threshold) in a second population (Hodgkinson et al, PNAS 2010). We recently used next-generation sequencing to analyze brain histone methylation and transcriptome changes resulting from early maternal deprivation in Rhesus macaques (Barr et al, in preparation) and effects of chronic cocaine and alcohol in human brain (Zhou et al, PNAS 2010). The latter study found distinct differences between the gene networks that are regulated by chronic drug exposure in people, as compared to those detected in acute and subacute exposures in animal models. Founder populations, extreme probands, and exposed populations enhance power to detect gene effects. Our Finnish dataset is derived from a founder population and ascertained from criminal alcoholic probands & thus enriched for Type II early onset alcoholism. SW and Plains Indian samples represent founder populations. An African American substance dependence sample with high adversity exposure revealed GxE of childhood adversity and HTTLPR in suicidality (Roy et al, 2007). Stress and poverty, but not African ancestry, predicted high risk of addictions (Ducci et al, 2009). An MAOA VNTR previously linked to dyscontrol via stress interaction was linked to outcomes of alcohol dependence and ASPD in American Indian women, of whom half had been sexually abused as children (Ducci et al, 2008). A strong interaction between the MAOA VNTR and testosterone was observed in the Finnish criminal sample (Sjoberg et al). As mentioned, deep sequencing in Finns detected a population-specific HTR2B Stop codon significant for impulsive behavior (Bevilacqua et al, Nature, 2010).