Hippocampal–prefrontal coherence mediates working memory and selective attention at distinct frequency bands and provides a causal link between schizophrenia and its risk gene GRIA1

Increased fronto-temporal theta coherence and failure of its stimulus-specific modulation have been reported in schizophrenia, but the psychological correlates and underlying neural mechanisms remain elusive. Mice lacking the putative schizophrenia risk gene GRIA1 (Gria1 ^–/– ), which encodes GLUA1, show strongly impaired spatial working memory and elevated selective attention owing to a deficit in stimulus-specific short-term habituation. A failure of short-term habituation has been suggested to cause an aberrant assignment of salience and thereby psychosis in schizophrenia. We recorded hippocampal–prefrontal coherence while assessing spatial working memory and short-term habituation in these animals, wildtype (WT) controls, and Gria1 ^–/– mice in which GLUA1 expression was restored in hippocampal subfields CA2 and CA3. We found that beta (20–30 Hz) and low-gamma (30–48 Hz) frequency coherence could predict working memory performance, whereas—surprisingly—theta (6–12 Hz) coherence was unrelated to performance and largely unaffected by genotype in this task. In contrast, in novel environments, theta coherence specifically tracked exploration-related attention in WT mice, but was strongly elevated and unmodulated in Gria1-knockouts, thereby correlating with impaired short-term habituation. Strikingly, reintroduction of GLUA1 selectively into CA2/CA3 restored abnormal short-term habituation, theta coherence, and hippocampal and prefrontal theta oscillations. Although local oscillations and coherence in other frequency bands (beta, gamma), and theta-gamma cross-frequency coupling also showed dependence on GLUA1, none of them correlated with short-term habituation. Therefore, sustained elevation of hippocampal–prefrontal theta coherence may underlie a failure in regulating novelty-related selective attention leading to aberrant salience, and thereby represents a mechanistic link between GRIA1 and schizophrenia.