involved in memory processing, are the regions first and most severely damaged in AD. Although this devastating disease is characterized by an extensive neuronal loss at late stages, only few neurodegenerative changes have been identified using transgenic models. In this sense, we have previously reported a selective early loss of SOM/NPY interneurons in the hippocampus of a PS1xAPP model (Kozack APP751SLxPS1M146L). The aim of the present study has been to determine early neuropathological changes in the entorhinal cortex of this AD model using RT-PCR and immunohistochemical techniques. Numerical densities of different neuro- nal populations were estimated by stereology. Results showed differential susceptibility of various neuronal populations for neurodegeneration at the early age of 6 months. A marked reduction of SOM GABAergic cells was determined whereas there were no changes in the parvalbumin population. Double labeling using GABAergic markers and NeuN was performed to identify pyramidal neurons. A reduction of NeuN labeling was detected at this early stage. There was a strong correlation between cell loss and the amyloid load. In addition, this AD model developed an extensive inflam- matory response with a substantial microglial activation and induction of IL-1beta, TNF-alpha and iNOS expression. This early neurodegeneration provides the anatomical substrate to explain, at least in part, the memory deficits observed in early phases of AD. Moreover, this AD model might be a valuable tool to assess the efficacy of potential early treatments against this disease. Supported by FIS PI030214 (to A.G.), SAF2002--03448 (to J.V.), FIS PI030177 (to D.R.) and Aventis Grants (to J.V. and A.G.). P1-017 FOCAL LENTIVIRUS-MEDIATED EXPRESSION OF TRANSGENIC TAU AND APP Lary C. Walker 1 , Rebecca Rosen 1 , Howard Rees 1 , Stephanie Stephens 1 , Stephanie Carter 1 , Jason Fritz 1 , Anthony Chan 1 , Alexey Bogush 2 , Sam Gandy 2 , Michelle Ehrlich 2 , Allan Levey 1 , Yoland Smith 1 , James Lah 1 , 1 Emory University, Atlanta, GA, USA; 2 Thomas Jefferson University, Philadelphia, PA, USA. Contact e-mail: lary.walker@emory.edu Background: The germline expression of Alzheimer’s disease (AD)-re- lated transgenes in mice has been invaluable to research on the mechanisms of AD pathogenesis. However, no transgenic model yet fully recapitulates the pathology and cognitive impairment of AD, suggesting species-linked barriers to the full manifestation of the disorder. More biologically prox- imate species, such as nonhuman primates, might generate a more repre- sentative AD phenotype, but their long lifespan limits the utility of germ- line transgenesis. The ability to control gene expression in adult animals would greatly enhance the flexibility of disease models. Objectives: To assess the suitability of lentiviral vectors for delivering AD transgenes to neurons in vivo, we developed and tested lentiviral constructs in adult animals. Methods: Vectors containing the gene for green fluorescent protein (GFP), driven by one of 3 promoters, were injected into the dorsal hippocampus and neocortex of adult Sprague-Dawley rats or an adult squirrel monkey. We next introduced lentivirus-borne, Swedish-mutant (KM670/671NL) APP or wildtype human four-repeat tau into the rat hippocampus and neocortex. Results: The ubiquitin C (Ubi) promoter produced consistently superior GFP expression compared to the cytomeg- alovirus and platelet-derived growth factor promoters in adult rats. Ubi- GFP expression was localized primarily in neuronal somata, axons, and proximal dendrites, and was comparable in intensity at 3 weeks, 3 months and 13 months. Initial experiments showed similar expression of lenti-Ubi- GFP in neocortical neurons of an injected squirrel monkey. Local APP Swe and tau expression were robustly increased in neurons after 3 weeks in lenti-APP Swe and lenti-tau injected rats, respectively. At this timepoint, neurons in the lenti-tau injected rats also were immunoreactive with anti- bodies to abnormally phosphorylated tau. Longer term expression studies of the transgenes, alone and in combination, are in progress. Conclusions: Enduring, focal expression of lentivirus-delivered transgenes is feasible in adult animals, presenting a unique opportunity to efficiently model AD pathogenesis in biologically advantageous species. Supported by NIH RR-00165, the Woodruff Foundation, and ARC. P1-018 BACE1 KNOCKOUT MICE FAIL TO PUT ON WEIGHT AT THE SAME RATE AS WILD TYPE CONTROL MICE Alex J. Harper, Judy Latcham, M. Isabel Gonzalez, GlaxoSmithKline R&D, Harlow, Essex, United Kingdom. Contact e-mail: alex.j.harper@gsk.com Background: Beta-site APP Cleaving Enzyme 1 (BACE1) is a key en- zyme in the generation of beta-amyloid, a component of senile plaques in the brains of Alzheimer’s Disease patients. Inhibition of BACE1 activity in humans is an attractive target for a disease modifying therapy in Alzhei- mer’s Disease, however BACE1 may have also a role unrelated to APP cleavage. Objective(s): Transgenic mice carrying human BACE1 cDNA and knockout mice with endogenous murine BACE1 removed were ex- amined for differences in weight and compared to wild type (wt) controls. Methods: A LacZ reporter gene replaced exon 1 of the BACE1 endoge- nous locus to generate the BACE1 knockout mice and human BACE1 cDNA expression was driven by the Calmodulin kinase 2 promoter in the BACE1 transgenic mice. These mice were weighed as they aged and various tissues were weighed post mortem. Results: BACE1 knockout mice were lighter than wt controls and put on less weight as they aged. This effect was more prevalent in female than male knockout mice. Male but not female BACE1 transgenic mice were heavier and put on more weight than wt controls. No differences were seen in brown adipose tissue weights or body temperature, however peri-genital fat pads were lighter in the knock- out mice compared to wt controls. Conclusions: These results suggest an intriguing link between BACE1 and body weight. Despite this phenotype both lines were viable and fertile, however a trend towards increased mortality was seen in the BACE1 knockout mice. An explanation of this phenotype remains to be discovered but may include modulation of appe- tite or of fat metabolism, perhaps through altered APP processing and beta-amyloid levels although these effects may be unrelated to the role of BACE1 as beta-secretase. Our data might suggest that inhibition of this protease could lead to a change in fat metabolism in treated patients, however one key difference is that we have generated mice which com- pletely lack BACE1 protein or have an excess of BACE1 protein while the administration of a therapeutic BACE1 inhibitor would not change protein levels, at least in the short term, only inhibit its enzyme activity. P1-019 APOE- AND AGE-DEPENDENT CHANGES IN CEREBRAL GLUCOSE METABOLISM AND COGNITION IN MICE Laura Kennedy 1 , Guiquan Chen 1,2 , Richard G. M. Morris 1 , Karen Horsburgh 1 , 1 University of Edinburgh, Edinburgh, United Kingdom; 2 Harvard Medical School, Boston, MA, USA. Contact e-mail: l.kennedy@ed.ac.uk Background: Apolipoprotein E (APOE) 4 is a major risk factor for cognitive decline with aging and for developing Alzheimer’s disease (AD). Neuroim- aging studies show brain region-specific glucose hypometabolism in AD patients and cognitively normal APOE 4 allele carriers. Objective(s): This study sought to define whether age and APOE genotype influences local cerebral regional alterations in glucose utilisation (LCGU) and cognitive abil- ity in young (3-4 months) and middle-aged (12-14 months), wild-type C57Bl/6J (WT) mice, APOE-knockout (APOE KO) mice and mice expressing human APOE 3 or 4. Methods: Learning and memory was assessed using the Atlantis platform watermaze and in the same animals LCGU was measured using 2-deoxyglucose autoradiography. Results: In young mice, APOE 3, APOE 4 and WT mice did not differ (p  0.05) in learning performance, but all groups learned significantly faster than APOE KO mice (F = 7.99, df 1/49, p  0.005). In contrast, in middle-aged mice, there was no significant differ- ence in the ability of APOE 3, APOE 4 and APOE KO mice to learn but these mice were significantly slower at learning the task than the WT mice S99 Poster Presentations P1