探討思覺失調症候選基因Akt1在小鼠酬賞決策過程中的行為與電生理反應之角色

Abstract

過去在人類基因研究或是動物實驗中提出許多的證據支持思覺失調症候選基因AKT1和思覺失調症的發病以及多巴胺調控的行為有關。而近期以Akt1基因缺損小鼠為模式的研究更指出了AKT1在與思覺失調症相關的行為、甲基安非他命引發的致敏化行為、目標導向行為、決策行為中的酬賞預測誤差和紋狀體神經活動中扮演了重要的角色。在本研究當中，為了更加了解Akt1在機率相關的酬賞決策行為中有何影響，我們使用Akt1異型合子與野生型控制組之雄性小鼠來進行帕夫洛夫條件配對作業和不同機率區間的二擇一動態酬賞作業。根據帕夫洛夫條件配對訓練當中的行為結果，我們發現Akt1異型合子與野生型控制組無論在徵兆導向行為或目標導向行為表現皆是相似的。然而在二擇一動態酬賞作業的訓練中，則發現Akt1異型合子小鼠只需較少的試驗數即可以穩定地選擇高酬賞機率的選項。本研究進一步去分析連續的選擇反應和選擇後的結果，並將小鼠的選擇行為套用在強化學習模型上以取得行為參數。套用模型得到的參數結果則發現Akt1異型合子小鼠擁有較高的學習速率，但同時他們的選擇一致性則比野生型控制組來得低。而我們在小鼠進行二擇一動態酬賞作業同時記錄了背內側紋狀體的局部場電位，並且分析小鼠在不同決策階段當中的事件關聯電位。功率譜密度的結果當中，異型合子小鼠在基準活動時表現出比較大的伽瑪功率。而在進行二擇一動態酬賞作業時，異型合子小鼠在不同機率區段當中的不同決策階段也表現出異常的事件關聯電位。根據上述的結果，我們證實了Akt1的缺損會造成異常的神經電生理訊號和酬賞預測誤差，並且這樣異常的表現會更進一步去影響酬賞相關的決策行為。另外我們也證實了AKT1缺損和異常的背內側紋狀體活動可以影響酬賞學習和酬賞預測誤差訊號。我們的研究成果可以使世人對Akt1基因在酬賞決策行為上的角色更加了解，並且對於此基因在思覺失調症的致病上有更深的認識。

Accumulating evidence from human genetic studies and animal studies suggests that AKT1, a schizophrenia candidate gene, is involved in the pathogenesis of schizophrenia and dopamine-dependent behaviors. Recent findings also indicate that AKT1 plays an important role in the regulation of schizophrenia-related behaviors, methamphetamine-induced behavioral sensitization, goal-directed behavior, reward prediction error in decision-making, and striatal neuronal activity, particularly in Akt1-deficient mice. In this study, to further investigate the role of Akt1 in the regulation of probabilistic reward-based decision-making, male Akt1 heterozygous (HET) mutant mice and their wild-type (WT) littermate controls were used to perform Pavlovian-conditioned pairing and a 2-choice dynamic foraging task with different reward rates. Behavioral data of the Pavlovian-conditioned pairing revealed that there was no significant difference in the sign-tracking or goal-tracking behaviors between HET and WT mice. However, in the 2-choice dynamic foraging task, HET mice took fewer trials to consistently choose the high reward option compared to their WT controls. Trial-by-trial choice behaviors of mice were further analyzed to estimate parameters for reinforcement learning. Our model-fitting data revealed that HET mice had a higher learning rate but a lower choice consistency compared to WT controls. Based on these findings, in vivo local field potential recordings were conducted in the dorsomedial striatum of mice during different stages of the decision-making process in the dynamic foraging task. Our results revealed a genotypic difference in the power spectrum density (dB) of local field potentials at the baseline. The event-related potential data further indicated genotypic differences at different stages of the decision-making process in different sections of reward rate of the dynamic foraging task. Our results collectively suggest that Akt1 deficiency can result in aberrant electrophysiological responses as well as abnormal reward prediction error signaling and reward learning during decision-making. Furthermore, our findings implicate the contribution of AKT1 deficiency and consequential neural abnormality in the dorsomedial striatum in the alteration of aberrant reward learning and reward prediction error signaling. Findings from this study shed light on our understanding of the role of Akt1 in reward-based decision-making and the involvement of this gene in the pathogenesis of schizophrenia.