LAUSR

Abstract Background Post-mortem studies in schizophrenia patients have repeatedly found abnormalities in the γ-aminobutyric acid (GABA) inhibitory neurotransmission system in the form of parvalbumin (PV)-expressing GABAergic interneuron reduction and aberrant glutamic acid decarboxylase mRNA expression and receptor distribution. However, these findings cannot be separated from medication effects. Evidence from the methylazoxymethanol acetate (MAM) animal model of schizophrenia has suggested a reduction of hippocampal PV interneurons in MAM-treated rats results in hippocampal dysfunction in the form of increased pyramidal neuron activity, which contributes to dysregulation of the stress response and concurrent disinhibition of dopaminergic neurons in the midbrain. Administration of a positive allosteric modulator of the α5 subunit-containing GABA-A receptor (α5 GABAAR) reversed dopamine neuron hyperactivation in the ventral tegmental area and attenuated schizophrenia-like behaviour in MAM-treated rats, suggesting this receptor subtype’s involvement in the development of psychosis. Conflictingly, an in vivo PET imaging study found no difference in α5 GABAAR receptor binding compared to controls but demonstrated a negative correlation between binding in the hippocampus and negative symptoms of schizophrenia patients. However, this study included previously medicated patients. To understand the involvement of α5 GABAAR in hippocampal dysfunction without a confound of antipsychotic medication, we investigated whether these receptors are differentially expressed in MAM-treated compared to control rats using autoradiography. Methods MAM-treated (n=21) and saline (vehicle)-treated (n=22) male rats were assessed for schizophrenia- and anxiety-like phenotypes via amphetamine-induced hyperlocomotion (AIH) and the elevated plus maze assays (EPM), respectively. To quantify the density of α5 GABAAR, we used [3H]Ro15-4513, a selective radioligand binding to the benzodiazepine site of this subunit. [3H]Flumazenil, an antagonist of the benzodiazepine-binding site of α1-3,5 GABAAR, was used on adjacent brain slices of the same animals to investigate more general binding to GABAAR. Brain sections were treated with the radioligands and x-ray sensitive films were exposed to the brain sections and radioactivity standards for 4 (flumazenil) or 8 (Ro15-4513) weeks. Developed films were imaged and normalised to background, radioactive binding in images was calculated based on the standard curve using the robust regression method in Graphpad Prism 8. Subsequently, quantified images were sampled for the CA1 of the ventral hippocampus using MCID software. Independent t-tests were used to investigate group differences in time spent in open arms during the EPM, total distance moved during AIH, and receptor binding. The significance threshold was set to p=0.05. Results MAM-treatment induced a schizophrenia-like phenotype (AIH: t(39)=‒3.022, p=0.004) and an anxiety-like phenotype (EPM: t(41)=2.810, p=0.008) compared to saline. We found reduced α5 GABAAR binding in the ventral hippocampus CA1 region of MAM-treated rats compared to saline-treated rats (t(41)=2.563, p=0.014), but no differences in flumazenil binding to α1-3,5 GABAAR (t(41)=1.333, p=0.190). Discussion MAM treatment induced schizophrenia- and anxiety-like phenotypes, accompanied by an α5 GABAAR density decrease in the CA1 region of the ventral hippocampus. Specific reduced binding to α5 was corroborated by the absence of any binding changes in the flumazenil treated sections. These findings reveal part of the molecular mechanisms behind hippocampal dysfunction and implicate the α5 subunit as a potential novel target for treatment of schizophrenia.