化約論者的瘋狂─以貝氏取向探討AKT1在精神分裂症多巴胺假說中的角色

Abstract

本研究旨在探討精神分裂症候選基因Akt1對增強學習模型中酬賞預測誤差之影響，並進一步探討AKT1在精神分裂症多巴胺假說中所扮演的角色。在近幾年的研究中，AKT被發現是多巴胺D2受體下游細胞內訊息傳遞因子之一，且與抗精神病藥物之藥效作用有密切關連。同時由於多巴胺假說認為紋狀體中過剩的多巴胺僅與精神分裂症中的正性症狀有關，且治療正性症狀的抗精神病藥物主要作用在D2受體之上。因此透過這些機制AKT1極可能參與正性症狀之產生或調節。近期一些關於精神症(即正性症狀之總稱)的研究嘗試採取貝氏推論之觀點，認為精神症之生成與多巴胺系統所產生不正常之酬賞預測誤訊號有關。奠基於這些理論及發現，本研究的假設為：若Akt1涉入精神症的生成，則Akt1缺損之小鼠的酬賞預測誤強度會與正常控制組小鼠有所不同。本研究採用Akt1異型合子之小鼠在動態搜索T形迷津之行為資料推算增強學習模型之參數，由此推估酬賞預測誤訊號之強度。實驗一結果顯示在所有三個測試狀況下Akt1異型合子之小鼠較正常控制組有更強的酬賞預測誤訊號。進一步透過西方墨點法分析腹側及背側紋狀體AKT1的磷酸化程度，異型合子小鼠較無法反應甲基安非他命所引發的磷酸化，顯示AKT1的活性與多巴胺有關。實驗二如同較強的酬賞預測誤所預期的，此小鼠在其他酬賞關聯學習作業的習得階段中有較迅速的學習表現。進一步以線性擬合法分析配合律，也發現Akt1異型合子之小鼠對於酬賞有較高的敏感度。最後，實驗三隱性抑制作業證明由實驗一所得較高酬賞預測誤之結果並無法預測以嫌惡學習為基礎的行為結果。本研究的結果顯示，AKT1可能透過調控腦中多巴胺系統的酬賞預測誤強度參與精神分裂症正性症狀之產生。

The current research aims to investigate how the schizophrenia candidate gene Akt1 (protein kinase Bα) participates in functions of the dopamine system, and further to identify its role in the dopamine hypothesis of schizophrenia. According to recent findings, AKT is found to be a downstream regulator under dopamine D2 receptor, and participates in antipsychotics remedy by acting as an intermediate in the antipsychotics-induced signaling cascade. The dopamine hypothesis of schizophrenia emphasized that psychosis is resulted from excessive dopamine concentration in the striatum, furthermore, antipsychotics mitigates psychosis by acting on dopamine D2 receptors. Recent theory, based on the view that brain is a Bayesian inference machine, regards psychosis is related to disruptions in the reward prediction error (RPE) signal produced by the midbrain dopamine system. Based on these findings, we hypothesized that if AKT1 is involved in the pathogenesis of psychosis, the RPE signal should be different in magnitude between Akt1 deficient and normal mice. In experiment 1, we estimated parameters in the reinforcement learning model by utilizing the behavioral data collected from a dynamic foraging T maze task perform by male Akt1 heterozygous (HET) and wildtype (WT) mice in order to infer the reward prediction error magnitude. The results showed that, compared with WT littermates, Akt1 HET exhibits higher reward prediction error magnitude among all three testing sections. In experiment2, consistent with the prediction of higher RPE magnitude, Akt1 HET mice learned more rapidly than WT mice in reward-related tasks. Revealed by the Western blots analysis, a reduction of methamphetamine-induced phosphorylated AKT1 was found in the ventral and dorsal striatum of Akt1 HET mice but not in WT controls, indicating the activity of AKT1 is indeed related to dopamine. Matching law analysis further revealed that Akt1 HET mice have higher reward sensitivity compared with WT controls. Finally, in experiment 3, revealed by latent inhibition paradigm, we showed that higher RPE signal cannot predict performance in an aversive-based behavioral paradigm. Our study suggests that, AKT1 might participate in the pathogenesis of psychosis by regulating the RPE magnitude in the dopamine system.