ARTICLES Glucocorticoids, the adrenal steroid hormones secreted during stress, are essential to neuronal function 1 ; however, their excess after brain injury is detrimental to neuronal survival 2 . These effects are most pro- nounced in neurons of the hippocampus, a brain region important for learning and memory and enriched with glucocorticoid receptors. Glucocorticoids compromise the ability of hippocampal neurons to survive neurological damage by exacerbating various steps in the excito- toxic glutamate–calcium–reactive oxygen species cascade 3 . This ability of glucocorticoids to compromise neuronal survival is of particular clinical interest because necrotic neurological injuries, such as stroke, seizure or hypoglycemia, provoke considerable secretion of glucocorti- coids. In contrast with the action of glucocorticoids, estrogen is neuro- protective in the aged or injured brain 4 . On the basis of these observations, we designed three genetic strategies to counter this hormonal enhancement of injury and investigate the mechanisms involved: two genes were selected and engineered to directly modulate or interfere with the regulatory pathway of glucocorticoids, and a third gene was designed to con- vert the deleterious effects of glucocorticoid into beneficial estro- genic ones. Our first two approaches blocked the hormonal signal by suppressing glucocorticoid-induced translocation of the gluco- corticoid receptor (GR) to minimize the glucocorticoid-induced regulation of gene expression and to prevent the consequent exac- erbation of excitotoxicity. Our third approach both blocked glucocorticoid-induced GR translocation and converted the glu- cocorticoid signal into a genomic estrogenic effect. This combina- tion provided the most neuroprotection, which is mediated partially through upregulation of brain-induced neurotrophic factor (BDNF). RESULTS Generation and characterization of therapeutic vectors To identify actions of glucocorticoids that may be useful targets for therapeutic intervention, we evaluated three different modes of manipulating the endocrine modulation of neuron death. The first intervention involved neuronal overexpression of 11β-hydroxysteroid dehydrogenase (11βHSD) type 2 (Fig. 1a). This renal cytosolic enzyme unidirectionally hydrolyzes active corticosterone (the pre- dominant rodent glucocorticoid) to its inactive metabolite corti- sone 5 , thereby reducing glucocorticoid signaling. The second intervention was more selective. Although high-stress levels of glucocorticoids are disruptive, moderate increases in gluco- corticoid concentrations facilitate synaptic plasticity in the hip- pocampus and hippocampus-dependent cognition. These opposing effects of glucocorticoids are mediated by two different receptors 6 . The high-affinity mineralocorticoid receptor (MR) is nearly saturated at basal glucocorticoid concentrations, and mediates the beneficial effects of moderate glucocorticoid elevation 7 . In contrast, the lower- affinity GR is heavily occupied only at stress-level concentrations of glucocorticoids and mediates the deleterious effects of glucocorti- coids. Thus, it is desirable to target GR- but not MR-mediated gluco- corticoid actions. To do so, we constructed a rat dominant-negative GR (dnGR; Fig. 1a). No dnGR native to the rat is known; however, an orthologous dominant-negative splice variant of the human GR has been characterized, and served as our template. The β isoform of the human GR 8,9 (GRβ) is identical to the common GR (GRα) for the first 727 amino acids. However, it lacks the 50 C-terminal residues of the α isoform, terminating instead with 15 nonhomologous amino acids. Thus, GRβ cannot bind with ligands or interact strongly with Departments of 1 Biological Sciences, 2 Neurology and Neurological Sciences, 3 Neurosurgery and 4 Biomedical Informatics, Stanford University, Stanford, California, USA. 5 Leibniz Institute for Neurobiology, Magdeburg, Germany. Departments of 6 Psychiatry and 7 Infectious Diseases, Stanford University, Stanford, California, USA. D.K., W.O.O. and Z.S.P. contributed equally to this work. Correspondence should be addressed to D.K. (danielak@stanford.edu). Published online 8 August 2004; doi:10.1038/nn1296 Restructuring the neuronal stress response with anti- glucocorticoid gene delivery D Kaufer 1,3 , W O Ogle 1 , Z S Pincus 1,4 , K L Clark 1 , A C Nicholas 1 , K M Dinkel 1,5 , T C Dumas 1 , D Ferguson 6 , A L Lee 1 , M A Winters 7 & R M Sapolsky 1,2 Glucocorticoids, the adrenal steroids released during stress, compromise the ability of neurons to survive neurological injury. In contrast, estrogen protects neurons against such injuries. We designed three genetic interventions to manipulate the actions of glucocorticoids, which reduced their deleterious effects in both in vitro and in vivo rat models. The most effective of these interventions created a chimeric receptor combining the ligand-binding domain of the glucocorticoid receptor and the DNA- binding domain of the estrogen receptor. Expression of this chimeric receptor reduced hippocampal lesion size after neurological damage by 63% and reversed the outcome of the stress response by rendering glucocorticoids protective rather than destructive. Our findings elucidate three principal steps in the neuronal stress-response pathway, all of which are amenable to therapeutic intervention. NATURE NEUROSCIENCE VOLUME 7 | NUMBER 9 | SEPTEMBER 2004 947 © 2004 Nature Publishing Group http://www.nature.com/natureneuroscience