Auditory verbal hallucinations (AVHs) in schizophrenia have been suggested to arise from failure of corollary discharge mechanisms to correctly predict and suppress self-initiated inner speech. However, it is unclear whether such dysfunction is related to motor preparation of inner speech during which sensorimotor predictions are formed. The contingent negative variation (CNV) is a slow-going negative event-related potential that occurs prior to executing an action. A recent meta-analysis has revealed a large effect for CNV blunting in schizophrenia. Given that inner speech, similar to overt speech, has been shown to be preceded by a CNV, the present study tested the notion that AVHs are associated with inner speech-specific motor preparation deficits.

The present study aimed to provide a useful framework for directly testing the long-held idea that AVHs may be related to inner speech-specific CNV blunting in patients with schizophrenia. This may hold promise for a reliable biomarker of AVHs.

Hallucinating (n=52) and non-hallucinating (n=45) patients with schizophrenia, along with matched healthy controls (n=42), participated in a novel electroencephalographic (EEG) paradigm. In the Active condition, they were asked to imagine a single phoneme at a cue moment while, precisely at the same time, being presented with an auditory probe. In the Passive condition, they were asked to passively listen to the auditory probes. The amplitude of the CNV preceding the production of inner speech was examined.

Healthy controls showed a larger CNV amplitude (p = .002, d = .50) in the Active compared to the Passive condition, replicating previous results of a CNV preceding inner speech. However, both patient groups did not show a difference between the two conditions (p > .05). Importantly, a repeated measure ANOVA revealed a significant interaction effect (p = .007, ηp2 = .05). Follow-up contrasts showed that healthy controls exhibited a larger CNV amplitude in the Active condition than both the hallucinating (p = .013, d = .52) and non-hallucinating patients (p < .001, d = .88). No difference was found between the two patient groups (p = .320, d = .20).

The results indicated that motor preparation of inner speech in schizophrenia was disrupted. While the production of inner speech resulted in a larger CNV than passive listening in healthy controls, which was indicative of the involvement of motor planning, patients exhibited markedly blunted motor preparatory activity to inner speech. This may reflect dysfunction in the formation of corollary discharges. Interestingly, the deficits did not differ between hallucinating and non-hallucinating patients. Future work is needed to elucidate the specificity of inner speech-specific motor preparation deficits with AVHs. Overall, this study provides evidence in support of atypical inner speech monitoring in schizophrenia.

None Declared

Spinocerebellar ataxia type one (SCA1) is an autosomal dominant neurodegenerative disease caused by an expanded CAG repeat that encodes glutamine (polyQ) in the affected ATXN1 gene. SCA1 pathology is commonly characterized by the degeneration of the cerebellar Purkinje cells (PC) and brainstem. Symptoms include motor dysfunction, cognitive impairments, bulbar dysfunction, and premature death. Atxn1175Q/2Q knock-in mice were previously developed to model SCA1 by inserting 175 expanded CAG repeats into one allele of the Atxn1 gene, producing mice expressing ATXN1 throughout the brain and displaying SCA1 symptoms. Previous research has indicated the role of localization of the ATXN1 protein to the nucleus in pathology. Therefore, the Atxn1175QK772T/2Q mouse model was created by disrupting the NLS in the expanded Atxn1175Q/2Q mice by replacing lysine with threonine at position 772 in the nuclear localization sequence (NLS). Since this amino acid change previously blocked PC disease in another mouse model, the Atxn1175QK772T/2Q mice were created to examine how the NLS mutation affects neuronal cells. RNA sequencing analysis was previously performed and found differentially expressed genes (DEG) with Atxn1175Q/2Q downregulated compared to Atxn1175QK772T/2Q and Atxn12Q/2Q in the cerebellum, medulla, cortex, hippocampus, and striatum. The aim was to analyze these brain regions to validate the RNAseq differential gene expression at a protein level.

Therefore, western blots were performed on the following mouse models (n=12): wild type mice (Atxn12Q/2Q), mice with the nuclear localization sequence mutation (Atxn12QK772T/2Q), and mice with 175 expanded CAG repeats (Atxn1175/2Q). Based off the RNAseq data, the cerebellum was tested with ion channel genes (Cav3.1, Kcnma1, and Trpc3) and the striatum was tested with a gene found in medium-spiny neurons (DARPP-32).

In the cerebellum, Atxn1175/2Q was significantly downregulated compared to Atxn1175QK772T/2Q in Cav3.1, Trpc3, and Kcnma1. Atxn1175Q/2Q was significantly downregulated compared to Atxn12Q/2Q in Trpc3 and Kcnma1. Atxn1175QK772T/2Q was significantly downregulated compared to Atxn12Q/2Q in Trpc3. In the striatum, there was significantly reduced DARPP-32 expression found between Atxn12Q/2Q and Atxn1175QK772T/2Q, Atxn12Q/2Q and Atxn1175Q/2Q, and Atxn1175Q/2Q and Atxn1175QK772T/2Q.

Therefore, the significantly reduced gene expression at the protein level in the cerebellum and striatum validate RNAseq differentially expressed genes. Additionally, the downregulation of both the Atxn1175Q/2Q and Atxn1175QK772TQ/2Q compared to Atxn12Q/2Q in the striatum supports the lack of learning of those mouse models on the rotarod, suggesting that the nuclear localization mutation does not rescue learning. Interestingly, the downregulation of Atxn1175Q/2Q compared to Atxn1175QK772TQ/2Q likely supports the age-related motor decline rescue in the rotarod seen in Atxn1175QK772T/2Q and not Atxn1175Q/2Q.