The work presented here provides direct in vivo evidence of anti-inflammatory and protective action of LCω3 PUFA supplementation on age-associated cognitive decline. Mice fed with the control diet, displayed poorer working memory performances with age. This was accompanied by an increase in the expression of CD11b and GFAP as well as mRNA levels of proinflammatory cytokines in the hippocampus. The increase in GFAP mRNA expression was associated with the appearance of shorter astrocytic processes in the DG, but not in the CA1 and CA3 as revealed by GFAP immunofluorescence. A 2-month dietary supplementation with fish oil in aged mice increased brain levels of DHA and significantly reduced mRNA levels of TNFα, IL-6 and CD11b. While GFAP expression was not reduced by the supplementation, the length of astrocytic processes in the DG was restored in the CA1 and CA3. In addition, working memory was improved by the LCω3 PUFA-supplemented diet. Taken together, our results provide new insight into the use of LCω3 PUFA supplementation in the elderly to limit development of neuroinflammation and cognitive impairment.
In the present study, we formulated 2 diets in order to provide different levels of DHA to the brain. In this work, the LCω3 PUFA supplementation was given for 2 months to mice fed throughout life with the control diet. This was of importance that the control and the LCω3 diets contained the same total amount of fat, therefore avoiding the possibility of a general effect of lipids. We previously showed that the long-term consumption of the control diet provided DHA to the brain of both young and aged mice 
. We further found that adult and aged mice fed with the LCω3 PUFAs for 2 months had a higher DHA level in the brain compared to mice fed with the control diet. This is in agreement with previous studies showing that short-term consumption of a DHA-enriched diet in form of fish oil diet potently increased brain DHA level in young 
and aged mice 
. It is of importance that EPA was also increased in the brain of aged mice fed with LCω3 PUFAs, although the total levels remained very low as compared to those found for DHA. EPA increase has been recently reported in the brain of rats fed with a highly enriched EPA diet 
, however whether this increase was linked to preformed EPA incorporation in the brain or to a metabolic conversion of α-LNA remains to be determined.
Increased brain EPA and DHA levels in the aged brain were associated to the decrease in neuroinflammatory processes in the hippocampus. Particularly, age-associated increase of TNFα, IL-6 and CD11b mRNA expression was greatly reversed in the hippocampus. This was accompanied by the restoration of astrocytic phenotype in the DG region of aged mice. Addition of EPA and DHA to the diet could account for the improvement of neuroinflammatory processes linked to aging since these lipids have anti-inflammatory effects in the brain as shown by us and others 
. In our study, (GLA+EPA)/AA ratio was increased in the brain of aged mice fed with the fish oil diet, suggesting a balance in favor of anti-inflammatory prostaglandins production in the brain 
. However, the role of anti-inflammatory prostaglandins remained to be determined. In addition, the delta 5 desaturase index increased suggesting that EPA did not limit synthesis of AA, as previously described in humans 
. Indeed, the efficacy of EPA and DHA supplementation in decreasing proinflammatory cytokines production in the brain of aged mice could also be linked to other mediators such as resolvins or neuroprotectins 
To our knowledge, this is the first evidence of the improvement of astrocyte morphology by a short term dietary LCω3 PUFA supplementation. Reactive astrocytes are normally characterized by the overexpression of GFAP, the major component of astrocytic cytoskeleton. Until recently, neuroinflammation-induced astrogliosis was described as a rise in the amount of GFAP and the development of a hypertrophic phenotype 
. Our data confirmed the increase in GFAP expression already described in the aged human and rodent brain. More importantly, by using a 3D-reconstruction analysis, we showed a decrease in the length of astrocytic processes in the DG, but not in the CA1 and CA3 regions of aged mice. This is in agreement with another recent study, showing that the area occupied by GFAP-positive structures, as measured by an unbiased stereological approach, was reduced in the DG during aging, in spite of a better visibility of these structures due to increased GFAP content 
. Morphological modifications of astrocytes in this context were dependent on proinflammatory cytokines. This change in the area occupied by astrocytes, rather than their number, has thus been suggested to better reflect functional changes in astrocytes 
. While the increase in GFAP mRNA expression was not reversed by LCω3 PUFA supplementation in old mice in our study, the length of astrocytic processes in the DG was restored, accompanied by an increase in the astrocytic processes length in the CA1 and CA3 region of aged mice. In vitro, DHA influences astrocyte morphology and function rather than GFAP expression, corroborating our results 
. In addition, a beneficial role for DHA has been described in astrocyte maturation 
, which generally involves changing from a flat epithelioid morphology to a process-bearing stellate morphology 
. Whether the effects of LCω3 PUFAs on astrocytes morphology in the hippocampus are direct or mediated by cytokines remains to be elucidated. However, knowing that LCω3 PUFAs have a strong impact on proinflammatory cytokines production in the brain 
and that proinflammatory cytokines, in particular IL-6, are produced by and activate astrocytes 
, it is tempting to speculate that the effects of LCω3 PUFAs on astrocytes morphology were linked to their effects on neuroinflammation.
Of importance, aged mice fed with the control diet displayed a decrease in spatial memory performances, as previously described 
. In addition, c-Fos was activated in the CA1 and DG but not in the CA3 region of the hippocampus of young mice after the Y-maze task, as the hippocampus processes spatial information 
. Interestingly, we found a decrease in c-Fos activation in the CA1 and DG regions of the hippocampus of old mice together with impaired Y-maze performances. The decreased expression of the activity-dependent immediate early genes (IEGs) c-Fos and Arc with age has already been reported, especially in the DG 
. Given the role attributed to hippocampus and IEGs in spatial memory 
and memory consolidation 
, it is likely that decreased c-Fos expression in the DG was a contributing factor in the age-related decline in hippocampal functions 
. This was reinforced by the observation that alteration of the intrinsic properties or connectivity of the DG (e.g. insults or genetic variability) has previously been related to learning and memory deficits 
Several papers have reported the benefits of DHA and/or EPA on behavior, in particular in aged rodents 
. However, in most of these studies control mice were submitted to an ω3 PUFA deficient diet 
. In our study, control mice were fed with a control diet with a α-LNA/LA ratio of ¼. Very importantly, short-term LCω3 PUFA supplementation did not provide any cognitive benefits to young mice, but improved spatial memory with a short retention time in old mice. These results suggested that short-term DHA and EPA supplementation provided cognitive protection against spatial memory deficits linked to age in mice fed with a diet containing EFA α-LNA and LA throughout life. This was in accordance with recent studies showing a positive effect of short-term dietary supplementation with DHA and/or EPA on memory performances in aged mice 
. Conversely, other studies reported a lack of protective effects of DHA/EPA-enriched diets on aged-associated cognitive decline 
. Such discrepancies could be linked to the length of the supplementation time and the efficacy of LC PUFA increase in brain cellular membranes. Few studies have addressed the question of the duration of DHA dietary supplementation necessary to achieve a beneficial effect on memory in aged rodents. This would be of high interest in the context of age-associated cognitive decline in humans, in which handling of dietary LCω3 PUFAs could offer an efficient strategy for sustaining cognitive functions.
Indeed, increased level of EPA and DHA in the brain of aged mice could account for the effects of fish oil supplementation on spatial memory improvement. However, the mechanisms by which LCω3 PUFAs reduced cognitive impairment in aged rodents is not fully understood. According to our results, dietary LC PUFAs decreased neuroinflammatory processes in the hippocampus of old rodents. This was in accordance with previous studies showing that dietary ω3 PUFAs increased the recovery of LTP while reducing microglial activation in old rodents 
. Of related interest was the prominent role played by proinflammatory cytokines in inducing cognitive deficits in a model of cerebral microvascular amyloid protein deposition, emphasized by a study using minocycline, a tetracycline that inhibits microglial activation 
. This anti-inflammatory drug significantly improved cognitive performances, along with a reduction in the number of activated microglia and in IL-6 levels. In addition, sulindac, an anti-inflammatory compound 
, as well as minocycline improved learning and memory in a mouse model of Alzheimer's disease 
. Importantly, minocycline partially restores LTP and proinflammatory cytokines expression in the hippocampus of middle-aged rats 
. Very recently, neuroinflammatory processes have been identified as key early events (occurring in midlife, together with age-related cognitive changes) strongly implicated in cognitive dysfunction linked to age 
. This is in accordance with results obtained in the lab (unpublished data) and by others showing that inflammatory events first induce production of proinflammatory cytokines in the hippocampus, followed by impairments in spatial memory/learning 
In this context, one can wonder what would be the amount of LCω3 PUFAs a human would have to consume to approximate the supplementation given to mice in our study and its protective effects. Epidemiological and clinical studies have used plasma and erythrocyte membrane levels of LCω3 PUFAs as a surrogate marker linking LC PUFAs status in the brain to cognitive disease risks, but results remain contradictory 
. To our knowledge, the precise incorporation into the aged brain of LC PUFAs from diet is not fully described. Recently, a study using positron emission tomography reported that the brain incorporation rate of DHA in form of free fatty acids is of 4.6 mg/day/1,500 g in healthy human volunteers with no changes in patients suffering from Alzheimer's disease 
. Further studies are therefore needed to relate human brain rates of DHA incorporation to dietary PUFA intake in aged subjects.
In conclusion, short-term LCω3 PUFA supplementation in aged mice shut down low-grade neuroinflammation, restored astrocytes morphology specifically in the DG region of the hippocampus and improved hippocampus-dependent cognitive performance, strongly supporting the view that dietary supplementation with LCω3 PUFAs could be used as a new therapeutic approach to prevent or reduce age-linked cognitive impairment.