In animal models of amyloid pathology, the earliest phenotype is neuronal hyperexcitability caused by suppression of glutamate reuptake. In this scenario, at an early stage of Alzheimer’s disease (AD), glutamate spillover to peri- and extrasynaptic sites causes in conjunction with binding of β-amyloid detrimental activation of extrasynaptic N-Methyl-D-Aspartate-Receptors (NMDAR). The extrasynaptic NMDAR activation induces the translocation of a signalosome to the nucleus that is assembled by the protein messenger Jacob. Jacob docks the signalosome to CREB, leading to its transcriptional inactivation, called CREB shutoff. Specifically, Jacob directly binds to the bZIP domain of CREB and recruits the CREB dephosphorylating protein phosphatase-1 (PP1) to the CREB complex. The mechanism implies an interaction with the nuclear adaptor protein LIM only 4 (LMO4), a transcriptional co-activator of CREB that hinders de-phosphorylation and transcriptional inactivation of CREB. Jacob displaces LMO4 from the CREB complex and renders thereby CREB susceptible to de-phosphorylation. The association with LMO4 strengthens subsequent binding of Jacob to PP1. Accordingly, a Jacob gene knockout in a transgenic mouse model of AD results in strongly reduced CREB shutoff and less prominent neuronal cell loss. Based on a detailed molecular analysis of the binding interface of Jacob, CREB and LMO4 we next performed structural modelling. We then used this information and screened chemical libraries to identify a small chemical compound called Nitarsone. Nitarsone has been in use in veterinary medicine as well as in the treatment of cancer in humans. We found that Nitarsone selectively prevents binding of Jacob, but not of CREB to LMO4 and thereby hinders assembly of this signalosome and restores CREB transcriptional activity in the face of β-amyloid pathology. In addition, Nitarsone treatment in vivo prevents to a remarkable extent impairment of synaptic plasticity as well as cognitive decline in mouse models of AD. These results underscore the significance of macromolecular protein transport from NMDAR to the nucleus for disease progression and related cognitive impairment at an early stage of AD. Furthermore, we have demonstrated that transport from NMDAR to the nucleus is druggable and the intervention holds promise for the treatment of synaptic dysfunction in AD.