思覺失調症的多基因結構及預測基因調控表達：與早發性思覺失調症及類思覺失調病質性狀之關聯

Abstract

本論文共包括兩部分，欲藉由思覺失調症家庭樣本，透過多基因風險分數(Polygenic risk scores, PRS) 量化思覺失調症的多基因遺傳風險，並探討其與均質性狀，包括：早發性思覺失調症與類思覺失調病質性狀之間的關聯性。此外，本研究亦通過量化基因調控表現來揭示基因變異對基因表達的功能性影響，以理解思覺失調症的生物機轉。 第一部分，利用將過去全基因組關聯分析 (Genome-wide association studies, GWAS) 的結果轉化為基因調控機制並建構出思覺失調症的基因調控表達風險分數 (Predicted genetically regulated expression risk scores for schizophrenia, SZ-GeRS)，藉以探討早發性思覺失調症的遺傳架構並尋找其相關的核心基因。在此研究中，我們使用595位來自314個多發性思覺失調症家庭具有發病資料紀錄的患病手足進行分析，而這些患者中有223位被定義為早發性思覺失調症 (Early-onset Schizophrenia)，372位則為晚發性 (Late-onset Schizophrenia)。 首先，我們針對這些患者進行 SZ-GeRS的計算，SZ-GeRS 是基於思覺失調症轉錄組關聯研究 (Transcriptome-wide association study) 中對應基因的效應大小 (effect size)，加權預測基因調控表達量 (Predicted genetically regulated expression, GeRS) 後計算而得。在本研究中，我們除了評估了每個基因的 SZ-GeRS 與早發性思覺失調症之間的關聯性，還進一步考慮了多基因遺傳風險，評估了幾種基於基因效應累積所產生的基因調控表達風險分數與早發性思覺失調症的關聯性，其中包括：p值閾值化基因調控表達風險分數 (p-value thresholding GeRS)、共表達基因模組之基因調控表達風險分數 (module-based GeRS)、以及核心基因之基因調控表達風險分數 (hub-genes based GeRS)。我們發現，由精神疾病相關的基因共表達模組 (gene co-expression modules) 所建構的 module-based GeRS 當中，有一個模組與早發性思覺失調症顯著相關且能解釋2.3% 的變異。接著，我們在另一個獨立的驗證樣本中驗證這一個 module-based GeRS 區辨早發與晚發的能力。在此模組得到驗證後，我們針對落在這個模組中用於計算 module-based GeRS 的基因進行生物路徑分析 (Ingenuity Pathway Analysis)，進而辨識出六個樞紐基因，包括 RUVBL2、COPS6、TUBA4A、PSMB5、PSMB2 和LRPPRC。而基於這些樞紐基因計算的 hub-genes based GeRS，解釋了0.5%的早發性思覺失調症變異。此外，我們還發現，在控制思覺失調症的 PRS (SZ-PRS) 和潛在共變數後，由精神疾病相關的基因共表達模組所建構的 module-based GeRS 可以在此研究樣本及驗證樣本中各別額外解釋早發性思覺失調症中2.5%和0.7%的變異。 本研究顯示，基因型預測之基因調控表達風險分數提供獨立於多基因風險分數以外的獨立訊息，且具有潛力可辨識出與早發性思覺失調症相關的潛在風險基因，並為進一步理解早發性思覺失調症的生物學機制提供了新的視角。而釐清這些潛在樞紐基因的生物功能性仍然有許多挑戰，值得未來更近一步的探討與研究。 本論文的第二部分，欲透過多基因風險分數探討思覺失調症和類思覺失調病質性狀 (Schizotypal traits) 之間重疊遺傳決定因素的潛在關聯。 本研究樣本為來自315個多發性思覺失調症家庭中具有Schizotypal traits測量資料的538位患者及其543位未罹病的家屬。患者的思覺失調症症狀 (Schizophrenia Symptoms) 使用「活性症狀量表（Scales for the Assessment of Positive Symptoms, SAPS）」以及 「負性症狀量表（Scales for the Assessment of Negative Symptoms, SANS）」評估，而未罹病家屬的類思覺失調病質 (Schizotypy) 則利用「修正版準精神分裂性人格違常結構化問卷 (Modified Structured Interview for Schizotypy, SIS) 」來評估。我們參照過去的因素分析結果，對本研究樣本SAPS、SANS及SIS量表進行驗證性因素分析 ，並確認患者的三因子思覺失調症症狀 (包括：正向症狀、負向症狀、思維紊亂因子) 及未罹病家屬的四因子類思覺失調病質特徵 (包括：正向因子、負向因子、人際敏感因子、及社會退縮/內向性因子)。 首先，我們使用精神病基因組聯盟（Psychiatric Genomics Consortium）提供的針對思覺失調症的大型病例對照研究的統合分析結果作為發現樣本，計算本研究中每個樣本的SZ-PRS。接著，我們分別在患者及未罹病的家屬中比較SZ-PRS與不同因子的 Schizotypal traits 之間的關係。結果顯示，SZ-PRS僅與患者的三個思覺失調症症狀因子中的思維紊亂因子 (Disorganization factor) 相關，並且與未受影響親屬的四個類精神分裂症因素中的社會退縮/內向性因子 (Social isolation/Introversion factor) 相關。 此外，本研究進一步合併患者和其未罹病的親屬樣本，並計算出「對齊性正向思覺失調症狀-類思覺失調病質」因子 (Aligned Positive Schizophrenic-Schizotypy factor) 以及「對齊性負向思覺失調症狀-類思覺失調病質」因子 (Aligned Negative Schizophrenic-Schizotypy factor)。我們發現在合併樣本後，比起 Aligned Positive Schizophrenic-Schizotypy factor，SZ-PRS 可以解釋更多Aligned Negative Schizophrenic-Schizotypy factor 的變異。且與單獨分析患者和未罹病的親屬樣本相比，合併樣本中SZ-PRS與各因子的關聯效應略有增加。 最後，由於思覺失調症與其他精神疾病的基因遺傳風險具有高度重疊性，因此我們額外利用合併後樣本評估雙性情緒障礙症 (Bipolar Disorder, BIP)、重鬱症 (Major Depression Disorder, MDD)、及五個一般人格特質 (General Personality Traits) 的多基因風險分數與兩個「對齊性思覺失調症狀-類思覺失調病質」因子的相關性。結果顯示，「對齊性思覺失調症狀-類思覺失調病質」因子僅與SZ-PRS相關，而與BIP、MDD或一般人格特徵的PRS皆無關。我們的研究結果支持類思覺失調症的負向因子比正向因子更能反應思覺失調症的遺傳易感性的假設。 總結來說，本論文展示SZ-PRS及SZ-GeRS在探索思覺失調症均質亞型性狀的遺傳架構和生物機轉上之價值。

This dissertation comprises two main parts. It aims to quantify the polygenic risk of schizophrenia (SZ) using polygenic risk scores (PRS) in family samples with SZ and to investigate its association with homogeneous traits, including early-onset SZ and schizotypal traits. Additionally, this dissertation seeks to uncover the functional impact of genetic variation on gene expression by quantifying gene regulatory expression, thereby advancing our understanding of the biological mechanisms underlying SZ. Study I aimed to investigate the underlying biological mechanisms of early-onset SZ and identify associated hub genes. This was achieved by translating Genome-wide Association Studies (GWAS) results into regulatory mechanisms and constructing genetically regulated gene-expression risk scores for SZ (SZ-GeRS). In this study, we analyzed 595 individuals from 314 multiplex schizophrenia families, with 223 classified as early-onset SZ and 372 as late-onset SZ. First, we calculated the SZ-GeRS for these patients. The SZ-GeRS is derived by weighting the predicted genetically regulated expression based on the effect sizes of corresponding genes from transcriptome-wide association studies (TWAS) on SZ. Then, we evaluated the association between individual genes' SZ-GeRS and early-onset SZ. Given the polygenic architecture of SZ, we also assessed several types of gene-summing GeRS, including p-value thresholding GeRS, module-based GeRS, and hub-genes-based GeRS. We found that a module-based GeRS from psychiatric disorder-related co-expression modules is significantly associated with early-onset SZ, explaining 2.3% of the variance. Also, we validated the ability of this module-based GeRS to distinguish between early-onset and late-onset SZ in an independent validation sample. Upon validation of this module, we conducted an Ingenuity Pathway Analysis on the genes within this module used to calculate the module-based GeRS, identifying six hub genes: RUVBL2, COPS6, TUBA4A, PSMB5, PSMB2, and LRPPRC. The hub-genes-based GeRS, calculated from these hub genes, accounted for 0.5% of the variance in early-onset SZ. Additionally, we found that, after controlling for PRS for SZ and potential covariates, this module-based GeRS independently explained an additional 2.5% and 0.7% of the variance in early-onset SZ in the study sample and validation sample, respectively. This study demonstrates that SZ-GeRS provide an independent effect over SZ-PRS on early-onset SZ and have the potential to identify putative risk genes associated with early-onset SZ. However, elucidating the biological functions of these potential hub genes remains a significant challenge, warranting further investigation and research in the future. Study II aimed to investigate the associations of overlapping genetic determinants between schizophrenia and schizotypal traits through PRS. This study included 538 patients with SZ and 543 unaffected relatives from 315 multiplex schizophrenia families, all of whom have data on schizotypal traits. We assessed schizotypy in unaffected relatives using a modified Structured Interview for Schizotypy (SIS) and schizophrenic symptoms in patients using the SAPS/SANS scales. Confirmatory factor analysis confirmed three symptom factors in patients (Psychotic, Negative, Disorganized) and four schizotypal traits in unaffected relatives (Positive Schizotypy, Negative Schizotypy, Social Isolation/Introversion, Interpersonal Sensitivity). First, we calculated the SZ-PRS for each sample, and then separately compared the relationship between SZ-PRS and different factors of schizotypal traits in patients and their unaffected relatives. We found that the SZ-PRS was correlated only with Disorganization factor from three Schizophrenic symptom factors in the patients as well as with the Social isolation/Introversion factor from four Schizotypy factors in their unaffected relatives. Additionally, this study further pooled the patient and unaffected relative samples and derived the 2-factor schizotypal trait in the pooled sample (i.e., the Aligned Positive/Negative Schizophrenic-Schizotypy). After pooling patients and their unaffected relatives, SZ-PRS explained more variance in the Aligned Negative Schizophrenic-Schizotypy factor than in the Aligned Positive Schizophrenic-Schizotypy factor. Also, the association effects were slightly increased in the pooled sample compared to those of analyzing patients and unaffected relatives separately. Finally, due to the high genetic overlap between schizophrenia and other psychiatric disorders, we assessed the polygenic risk scores for bipolar disorder (BIP), major depressive disorder (MDD), and five general personality traits in relation to the 2-factor schizotypal trait in the pooled sample. The results indicated that the Aligned Schizophrenic-Schizotypy factors were only associated with SZ-PRS but not with PRS for BIP, MDD, or general personality traits. Our findings support the hypothesis that the negative factor of schizotypy more accurately reflects the genetic susceptibility to SZ than the positive factor. Taken together, this dissertation demonstrates the utility of SZ-PRS and GeRS in exploring the genetic architecture and elucidating the biological mechanisms underlying SZ in its homogeneous subtypes.