Intracerebroventricular administration of streptozotocin in rodents is used to create an animal model of dementia. The rationale for this model is that STZ induces alterations of insulin/IGF pathway and glucose metabolism 
as well as oxidative stress 
leading to cerebral alterations and behavioral impairments 
. Such model is used to evaluate the effect of various drugs 
. However, a wide range of doses are used in the literature in icv-STZ rodents without specifying the effects of doses and the location of the lesions induced. In the literature, the extension of the lesions induced by STZ is merely described, and it is often assumed that the toxic induces diffuse lesions that affect most of the brain regions. Here, we used 3D-MRI and histology to screen the whole brain of rats after bilateral intracerebroventricular administration of low (1 mg/kg) and high doses (3 mg/kg) of STZ. We present evidence that icv-STZ induces focal lesions that involve septal and corpus callosum regions and are associated with a neuronal loss and an inflammation process. A good consistency was detected between results from in vivo
evaluation of brain volume by MRI and quantification of neuronal loss, except in the septum in the 3 mg/kg group, one week after the STZ injection. At this time point, the neuronal loss in the septum was significant while no alteration was seen by MRI at this stage. This can be explained by the lower sensitivity of MRI method as compared to neuronal counting and also by a swelling of the septal tissue because of the severe inflammation at this early stage. A similar lack of perfect relationship between measures of cerebral atrophy detected by MRI and histological evaluations is commonly reported in the literature 
The lesions of the corpus callosum have never been described before, although white matter alterations at the level of the fornix were already reported in icv-STZ models 
. During dementia such as Alzheimer's disease in humans, there is also a severe atrophy of the corpus callosum that is reported and this latter is correlated to cognitive impairments 
. The model can thus mimic these alterations. On the contrary, the lesions of the septum that we highlighted are consistent with a previous report showing a 40% reduction in the weight of the septum after icv-STZ administration 
, The septum has cholinergic projections on several cortical regions including the entorhinal and cingulate cortex 
and on the hippocampus 
. These connections form the septohippocampal system which is implicated in spatial learning, short-term memory and attention 
. The septal atrophy highlighted in our study suggests that icv-STZ rats are a model of cholinergic denervation. Such denervation, septal alterations, but also white matter alterations could be responsible for behavioral alterations reported in icv-STZ models 
. Other models of septohippocampal denervations have been reported in the past. For example the use of 192 IgG-saporin, which selectively destroys cholinergic neurons, can lead to septohippocampal lesions 
. Icv-STZ might be less specific to cholinergic neurons but might also provide a wider range of persistent lesions including expanded and dose-dependant neuroinflammation, a process that is currently associated to neuronal loss in neurodegenerative diseases 
. In other cholinergic depletion models, inflammation was limited to the site of injection, namely the medium septum and the diagonal band of Broca in most cases 
We also highlighted the difference of lesions induced by low and high doses of STZ. At low dose, the corpus callosum was not atrophied whatever the time point, the neuronal loss and atrophy of the septum was not detected one week after Icv-STZ administration and was moderate at three months post administration, and inflammation was moderate. These processes were very severe at 3 mg/kg. The mechanisms induced at the two doses might thus be different with a chronic effect for the 1 mg/kg dose and an acute toxic effect leading to acute neuronal death for the high dose. The lower dose should thus be more interesting to evaluate neuroprotective drugs. Doses of 3 mg/kg and higher should be regarded as a way to model very aggressive neurotoxic lesions rather than subtle alterations due to small mechanistic alterations as would be expected during slowly evolving dementia processes or during aging. An intermediate dose, e.g. 2 mg/kg as tested by Shoham S. and collaborators 
, would perhaps be more relevant to evaluate some neuroprotective drugs.
Two mechanisms have been proposed to explain cerebral alterations induced by icv-STZ administration. First, icv-STZ is known to induce a severe oxidative stress 
. We speculate that the non-selective neurotoxicity and neuroinflammation induced by high doses of STZ close to the site of injection is consistent with an action of STZ via such mechanism. Oxidative stress would be less involved in the effects of STZ at low doses. Second, icv-STZ also induces an insulin resistance brain state 
leading to a widespread effect of STZ on energy metabolism. STZ enters the cells via the glucose transporter 2 (GLUT2). This transporter is present all over the brain 
. Thus one can expect that, if it diffuses all over the brain, STZ leads to a reduced brain metabolism involving most brain regions and not only the septum or the corpus callosum. The location of the lesions close to the injection sites would thus be consistent with a lack of diffusion of STZ all over the brain. We can however not rule out that the STZ diffused all over the brain but that hypometabolism induced by STZ was not severe enough to induce a severe atrophy process.
As a conclusion, we characterized the lesions induced by icv-STZ in rodents. We showed that at high doses, icv-STZ induces severe neurodegenerative lesions in the septum and corpus callosum. The lesions are associated with an inflammation process that might be caused by oxidative stress. High dose icv-STZ models can thus be used to evaluate compounds that modulate oxidative stress and inflammation leading to neuronal loss. The lesions are less severe at low doses and low dose icv-STZ models might thus be more relevant to study mechanisms associated to slowly evolving dementia and neuroprotective treatments.