A lipid neurochemist's siren: docosahexaenoic acid and its elusive function in the central nervous system


  • Eric J. Murphy

    Corresponding author
    1. Department of Basic Sciences, School of Medicine and Health Sciences, University of North Dakota, Grand Forks, North Dakota, USA
    • Address correspondence and reprint requests to Eric J. Murphy, Associate Professor, Department of Basic Sciences, School of Medicine and Health Science, University of North Dakota, 501 N. Columbia Rd., Room 3700, Grand Forks, ND 58202-9037, USA. E-mail: eric.murphy@med.und.edu

    Search for more papers by this author


Read the full articleUnesterified docosahexaenoic acid is protective in neuroinflammation’ on doi: 10.1111/jnc.12392.

Abbreviations used

alpha-linolenic acid


docosahexaenoic acid


eicosapentaenoic acid




neuroprotectin D1

Neurochemists have had a burgeoning love affair with docosahexaenoic acid (DHA, 22:6n-3) since it was discovered to be uniquely enriched in the CNS. As if called by a siren, lipid neurochemists have been drawn to DHA and seduced by its elusive function in brain physiology. Simply, why is this fatty acid enriched in the CNS? What is the reason for this enrichment?

In 1961, DHA was tentatively identified in rat brain and noted to increase during brain maturation, but more importantly it was found to be uniquely enriched in the brain as compared to other organ systems (Biran and Bartley 1961). This discovery initiated a flourish of activity. DHA was found to be predominantly in the phosphatidylserine and phosphatidylethanolamine pools in human gray matter (O'Brien et al. 1964). Others also observed that DHA levels increase during rat brain maturation (Marshall et al. 1966). Later work in mice further substantiated that DHA is highly localized to phosphatidylserine and ethanolamine glycerophospholipids, while it is minimally found in choline glycerophospholipids (Sun and Horrocks 1968).

The efficiency of alpha-linolenic acid (ALA, 18:3n-3) elongation and desaturation to its longer chain derivatives eicosapentaenoic acid (EPA, 20:5n-3) and to DHA is an often debated topic. It is important to note that DHA, but not EPA is found enriched in the CNS. Although we have recently reviewed the ALA conversion debate (Barceló-Coblijn and Murphy 2009), 50 years ago it was observed that dietary ALA increases brain DHA content and that DHA content is protected from variations in dietary linoleic acid (18:2n-6) intake (Mohrhauer and Holman 1963). This groundbreaking study demonstrates that DHA is protected from changes in dietary fatty acids, but is indeed enhanced upon intake of its precursor, ALA. Another study of that era also noted that dietary fatty acids can alter brain fatty acid composition in a time-dependent manner and that brain DHA can be derived from dietary precursor ALA, but noted that high linoleic acid intake can indeed moderately reduce brain DHA levels (Rathbone 1965). Some 40 years later we, as well as others, have shown that dietary ALA can enhance brain DHA content, although not to the same extent as dietary DHA intake (Barceló-Coblijn et al. 2005). This study demonstrates two very important points. First, the mammal has the capacity to make DHA from dietary ALA and enrich brain phospholipid DHA levels. Second, that dietary DHA simply does this more efficiently.

The mystery as to why EPA is not found concentrated in the brain, despite the high levels of DHA was recently illuminated by a study in which the authors observed that unlike DHA, EPA is rapidly lost in the CNS, accounting for the overall paucity of this n-3 family fatty acid in brain (Chen et al. 2011). This study established an important concept that not all n-3 fatty acid family members are treated equally in the brain, again suggesting that the enrichment of DHA into specific brain phospholipid classes has an important function.

In the current issue, Orr et al. (2013) demonstrate through an elegant set of complex experiments that a critical function of DHA is to modulate brain inflammatory response. Although it is known that exogenously added DHA complexed with albumin reduces ischemic infarct volume and brain edema in mice (Marcheselli et al. 2003; Belayev et al. 2005a) and enhances survival following brain intracranial hemorrhage by protecting blood–brain barrier integrity (Belayev et al. 2005b), the mechanisms underlying the role of the brain's endogenous DHA in CNS protection is less known. This exogenously added DHA complexed with albumin imparts protection via formation of 10, 17S-docosatriene (neuroprotectin D1, NPD1) in the ischemic region (Belayev et al. 2005a). Furthermore, NPD-1 formation reduces cyclooxygenase-2 expression following ischemia and reduces interleukin-1β induced nuclear factor-kappa-light chain enhance of activated B cells (NFκB) activation resulting in an upstream suppression of the signal for cyclooxygenase-2 expression (Marcheselli et al. 2003). This suggests an important role for exogenously added DHA is to be converted to NPD-1, which in turn regulates gene expression and results in an attenuation of expression of genes encoding mediators of proinflammatory proteins. What is lacking is a more complete understanding of the role that endogenous DHA and free, unesterified DHA has in this process.

To this end, Orr et al. show using the fat-1 mouse that increases in brain free, unesterified DHA levels and enhanced levels of DHA found esterified in brain phospholipids result in a profound reduction in expression of genes known to be associated with inflammatory response induced by lipopolysaccharide (LPS) as well as reduction in histochemical markers of inflammatory response such as astrocyte activation glial acidic fibrillary protein (GFAP) and microglial activation (Iba-1). It is important to note that the same level of arachidonic acid (ARA, 20:4n-6) is found in both free form and esterified to phospholipids in these mice. This first set of experiments use the fat-1 mouse that has been engineered to express fat-1, a desaturase that converts n-6 fatty acids to n-3 fatty acids (Kang et al. 2004), thereby enhancing the endogenous level of n-3 fatty acids.

Because mammals normally lack the capacity to convert n-6 fatty acids to n-3 fatty acids, it was necessary to determine if the high level of free DHA in the fat-1 mice, but not the elevated levels of DHA found esterified in the phospholipids, attenuates LPS-induced neuroinflammatory response. To assess this point, two different experiments were done. First, wild-type mice were fed a 2% fish oil diet, which enhances endogenous esterified and free, unesterified DHA to the same level as found in the fat-1 mice. Under these conditions, there are no differences between fat-1 mice and fish oil fed mice in proinflammatory gene expression, suggesting that an elevation in free, unesterified DHA attenuates inflammatory response. Second, wild-type mice were fed diets of either 2% fish oil, or the n-3 deficient diet of safflower oil. In this experiment, mice fed the fish oil diet have slightly elevated DHA in the phospholipids, but free DHA levels are not different between groups and are at a much lower level than in fat-1 mice. Under these conditions, there is no change in LPS-induced inflammatory response between groups, suggesting that the increase in levels of DHA found esterified in phospholipids is without an effect on brain inflammatory response. These very difficult feeding studies are indicative that the elevated free DHA found in the fat-1 mice is critical for the attenuating the LPS-induced neuroinflammatory response, rather than the DHA found in phospholipid pools.

Although these experiments demonstrate that free DHA is necessary to reduce LPS-induced neuroinflammatory response, the mechanism is not illuminated. Although a potential mechanism would be the formation of NPD-1 or other docosanoids, the previous experiments do not address this possibility. However, direct infusion of DHA or 17S-HpDHA, the precursor of NPD-1, into the brains of mice previously infused with LPS does test this potential mechanism. Under these conditions, IL-1β and CCL3 expression is equally reduced under either condition, but GFAP expression is only reduced by 17S-HpDHA. Infusion of DHA enhances brain NPD-1 levels, but not to the same extent as infusion of its precursor. Interestingly, ARA infusion under these same conditions did not exacerbate the neuroinflammatory response.

Collectively, these experiments eloquently approach an important, but difficult question. Does free DHA attenuate inflammatory response? Although it is not definitively demonstrated, the collective evidence supports the point that free, unesterified DHA in the brain rather than the elevated DHA levels associated with the phospholipid pools attenuates inflammatory response in the CNS (Fig. 1). In a lot of studies in which free, unesterified fatty acids are measured in the brain, there is a strong potential that any increase seen in free, unesterified fatty acids reflects unintentional release from phospholipid pools during tissue processing. This is an absolute critical consideration, but fortunately the potential for these changes in free DHA to be associated with a fixation artifact during preparation of the brain and downstream processing is minimal as the tissue was properly fixed using head-focused microwave irradiation [for review on this technique see (Murphy 2010)]. Thus, all of these data point to free DHA attenuating LPS-induced neuroinflammatory response, in part, via its conversion into NPD-1.

However, what are the other potential functions for DHA in the brain? While DHA is known to attenuate neuronal apoptosis (Kim et al. 2000), NPD-1 is also known to attenuate apoptosis as well (Marcheselli et al. 2003), suggesting that DHA itself or its metabolism to NPD-1 may directly attenuate apoptosis. However, others have observed that DHA, in a definitive U-shaped concentration-dependent manner, reduces apoptosis via reduced caspase-3 activity and that this protective effect also involves elevated phosphatidylserine levels and enhanced PI3Kinase activity (Akbar and Kim 2002). This is important as DHA can enhance phosphatidylserine synthesis in rat brain microsomes and in C6 glioma cells (Garcia et al. 1998) and is a phospholipid that has high levels of DHA (Sun and Horrocks 1968). But DHA is also associated with a role in brain glucose uptake and cytochrome C activity (Ximenes da Silva et al. 2002), again suggesting a much broader role than merely influencing brain inflammatory response.

One long held hypothesis is that DHA alters membrane biophysical properties that permit it to alter function of membrane-associated proteins. The unique structure of DHA requires substantial space to accommodate it in the membrane, thus it increases the intermolecular space between acyl chains in the membrane. This of course very well may account for the enhanced glucose uptake (Ximenes da Silva et al. 2002) and PI3kinase activity (Akbar and Kim 2002), observed to be modulated by DHA. Perhaps the most well documented biophysical change is in G protein-coupled signaling response in retinal rod outer segment exposed to light (Litman et al. 2001; Niu et al. 2004). Over the years, this body of work has demonstrated that this system in the rod outer segments can distinguish between fatty acid chain length, fatty acid family and degree of unsaturation, with a high DHA containing membrane having optimal activity in this system.

Figure 1.

Docosahexaenoic acid (DHA, 22:6n-3) enters from the plasma and enters the brain going into the free, unesterified pool of DHA in the brain. This pool provides DHA to become activated by long-chain acyl-CoA synthetases to the DHA-CoA pool that is then esterified onto the phospholipids. Phospholipase activity releases esterified DHA and it recycles back to the free, unesterified DHA pool. Evidence from Orr et al. suggest two separate metabolic pools for free, unesterified DHA with a specific pool of DHA being used to form neuroprotectin-1 (NPD-1) that reduces proinflammatory gene expression in the brain. The green pool represents the DHA that goes onto form NPD-1 whereas the bulk of the pool is used to provide DHA for the brain phospholipid pools.

In this issue, does the recent addition of work by Orr and colleagues (2013) add a new dimension to how we think about DHA in the CNS? Yes, in that we now have to think more carefully about the role of basal free, unesterified DHA in brain function, not merely focusing on increasing the amount of DHA in the phospholipids. We know in the human brain that DHA has a very long half-life, on the order of 773 days and that the brain requires only about 5 mg/day of DHA in the adult (Umhau et al. 2009), whereas the half-life of ARA is only 147 days and the brain requires about 18 mg/day (Rapoport et al. 2007). Thus, DHA is a long-lived molecule in the brain and as such we expect a much slower turnover. How this changes the pool size of free, unesterified DHA is an important point to understand in more detail. As DHA has been proposed to be a treatment for Alzheimer's disease and other neurological diseases (Calon and Cole 2007), it is an important consideration to understand which form DHA must have in the CNS to be the most effective in altering brain pathophysiology. With regard to neuroinflammatory response, the work of Orr and colleagues points us in the direction to carefully consider the role of free, unesterified DHA in this process. This is a departure from the status quo of enhancing DHA levels in the phospholipid pools and then working on mechanisms to enhance its release from these pools to minimize the pathophysiology of various brain diseases. Now we must think about where the free, unesterified DHA pool that impacts neuroinflammatory response is localized and how to maximize its ability to limit neuroinflammation. Are increasing the levels of DHA in the brain phospholipids a place to start? Or alternatively working to increase the level of free, unesterified DHA in the correct metabolic pool in a manner that will reduce neuroinflammatory response via formation of NPD-1 the place to start? The work presented by Orr and colleagues suggests that the most important element is the free form of DHA, but indeed, more work must be done to further understand the implications of their finding.

Conflicts of interest

The author declares no conflict of interest.