Docosahexaenoic acid (DHA) is an omega-3 fatty acid that is essential for normal brain growth and cognitive function1,2,3,4. Consistent with its importance in the brain, DHA is highly enriched in brain phospholipids5,6,7. Despite being an abundant fatty acid in brain phospholipids, DHA cannot be de novo synthesized in brain and must be imported across the blood–brain barrier, but mechanisms for DHA uptake in brain have remained enigmatic. Here we identify a member of the major facilitator superfamily—Mfsd2a (previously an orphan transporter)—as the major transporter for DHA uptake into brain. Mfsd2a is found to be expressed exclusively in endothelium of the blood–brain barrier of micro-vessels. Lipidomic analysis indicates that Mfsd2a-deficient (Mfsd2a-knockout) mice show markedly reduced levels of DHA in brain accompanied by neuronal cell loss in hippocampus and cerebellum, as well as cognitive deficits and severe anxiety, and microcephaly. Unexpectedly, cell-based studies indicate that Mfsd2a transports DHA in the form of lysophosphatidylcholine (LPC), but not unesterified fatty acid, in a sodium-dependent manner. Notably, Mfsd2a transports common plasma LPCs carrying long-chain fatty acids such LPC oleate and LPC palmitate, but not LPCs with less than a 14-carbon acyl chain. Moreover, we determine that the phosphor-zwitterionic headgroup of LPC is critical for transport. Importantly, Mfsd2a-knockout mice have markedly reduced uptake of labelled LPC DHA, and other LPCs, from plasma into brain, demonstrating that Mfsd2a is required for brain uptake of DHA. Our findings reveal an unexpected essential physiological role of plasma-derived LPCs in brain growth and function.
This is a preview of subscription content, access via your institution
Open Access articles citing this article.
Inflammation and Regeneration Open Access 17 July 2023
Nature Communications Open Access 09 June 2023
Improvement of retinal function in Alzheimer disease-associated retinopathy by dietary lysophosphatidylcholine-EPA/DHA
Scientific Reports Open Access 06 June 2023
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Rent or buy this article
Prices vary by article type
Prices may be subject to local taxes which are calculated during checkout
Kidd, P. M. Omega-3 DHA and EPA for cognition, behavior, and mood: clinical findings and structural-functional synergies with cell membrane phospholipids. Altern. Med. Rev. 12, 207–227 (2007)
Horrocks, L. A. & Yeo, Y. K. Health benefits of docosahexaenoic acid (DHA). Pharmacol. Res. 40, 211–225 (1999)
Mozaffarian, D. & Wu, J. H. Omega-3 fatty acids and cardiovascular disease: effects on risk factors, molecular pathways, and clinical events. J. Am. Coll. Cardiol. 58, 2047–2067 (2011)
Connor, W. E. Importance of n-3 fatty acids in health and disease. Am. J. Clin. Nutr. 71, 171S–175S (2000)
Breckenridge, W. C., Gombos, G. & Morgan, I. G. The lipid composition of adult rat brain synaptosomal plasma membranes. Biochim. Biophys. Acta 266, 695–707 (1972)
Innis, S. M. Dietary (n-3) fatty acids and brain development. J. Nutr. 137, 855–859 (2007)
Salem, N., Jr, Litman, B., Kim, H. Y. & Gawrisch, K. Mechanisms of action of docosahexaenoic acid in the nervous system. Lipids 36, 945–959 (2001)
Armulik, A. et al. Pericytes regulate the blood–brain barrier. Nature 468, 557–561 (2010)
Bell, R. D. et al. Pericytes control key neurovascular functions and neuronal phenotype in the adult brain and during brain aging. Neuron 68, 409–427 (2010)
Berger, J. H., Charron, M. J. & Silver, D. L. Major facilitator superfamily domain-containing protein 2a (MFSD2A) has roles in body growth, motor function, and lipid metabolism. PLoS ONE 7, e50629 (2012)
Lafourcade, M. et al. Nutritional omega-3 deficiency abolishes endocannabinoid-mediated neuronal functions. Nature Neurosci. 14, 345–350 (2011)
Carrié, I., Clement, M., de Javel, D., Frances, H. & Bourre, J. M. Phospholipid supplementation reverses behavioral and biochemical alterations induced by n-3 polyunsaturated fatty acid deficiency in mice. J. Lipid Res. 41, 473–480 (2000)
Kim, H. Y. Novel metabolism of docosahexaenoic acid in neural cells. J. Biol. Chem. 282, 18661–18665 (2007)
Simopoulos, A. P. The importance of the omega-6/omega-3 fatty acid ratio in cardiovascular disease and other chronic diseases. Exp. Biol. Med. (Maywood) 233, 674–688 (2008)
Croset, M., Brossard, N., Polette, A. & Lagarde, M. Characterization of plasma unsaturated lysophosphatidylcholines in human and rat. Biochem. J. 345, 61–67 (2000)
Granell, M., Leon, X., Leblanc, G., Padros, E. & Lorenz-Fonfria, V. A. Structural insights into the activation mechanism of melibiose permease by sodium binding. Proc. Natl Acad. Sci. USA 107, 22078–22083 (2010)
Paroder-Belenitsky, M. et al. Mechanism of anion selectivity and stoichiometry of the Na+/I- symporter (NIS). Proc. Natl Acad. Sci. USA 108, 17933–17938 (2011)
Shui, G. et al. Comparative plasma lipidome between human and cynomolgus monkey: are plasma polar lipids good biomarkers for diabetic monkeys? PLoS ONE 6, e19731 (2011)
Wong, P. et al. Pregnenolone rescues schizophrenia-like behavior in dopamine transporter knockout mice. PLoS ONE 7, e51455 (2012)
This work was supported in part by grants from the Singapore Ministry of Health’s National Medical Research Council CBRG/0012/2012 (to D.L.S.), the Singapore National Research Foundation Competitive Research Program grants 2008-01 (to E.L.G.) and 2007-04 (to M.R.W.), National University of Singapore’s Life Sciences Institute (to M.R.W.), and Singapore National Medical Research Council Translational and Clinical Research Program NMRC/TCR/003-GMS/2008 (to X.Z.). We would like to thank B. Tan (Duke-NUS) for technical assistance with lipid extractions, and S. Ying (Duke-NUS) for assistance with behavioural phenotyping.
The authors declare no competing financial interests.
Extended data figures and tables
a, Expression of Mfsd2a in endothelium is co-localized with glucose transporter Slc2a1 (Glut1). Arrowheads show endothelial cells in blood brain vessels. b, Mfsd2a is highly expressed in micro-vessels in brain, shown here are sections in dentate gyrus regions. c, Mfsd2a and the pericyte marker PDGFRß co-localize in brain microvasculature. d, Mfsd2a is not expressed in pericytes. Yellow and red arrowheads indicate endothelial cells and a pericyte, respectively. Similar expression pattern of Mfsd2a is found in endothelium of micro-vessels in brain of monkey. e, Mfsd2a is highly expressed in micro-vessels and is co-localized with glucose transporter Slc2a1 (Glut1) in brain, shown here are sections in cerebellum of P4 monkey. f, g, Expression of Mfsd2a in endothelium in of brain micro-vessels. Shown is the hippocampal region. The marker for astrocytes is GFAP. KO, knockout; WT, wild type.
Extended Data Figure 2 Localization of Mfsd2a at the blood–brain barrier of E15.5 fetus and lipid analysis.
a, Mfsd2a (red) is highly expressed in micro-vessels and is co-localized with glucose transporter Slc2a1 (Glut1, green) in fetal brain. Scale bars 100 µm. b, Placental and fetal weights. Placentas and fetuses from two HET pregnant mice (E18.5) crossed with a Mfsd2a-knockout male were collected and weighed. There were no significant differences in placental and fetal weight between HET (n = 7) and knockout (n = 11) mice. c, Tail suspension was used to test for the presence of the paw-clasping phenotype of 10-week-old wild-type and knockout mice. d, Mass spectrometry measurement of phospholipids in the E18.5 fetal brain of wild-type (n = 6) and knockout (n = 5) mice showed that fetal brains of knockout mice had significantly reduced DHA levels, whereas AA levels were increased. ***P < 0.001 (see Source Data accompanying this figure for the full data set).
a, A representative image of brains of two 8-week-old wild-type and knockout littermates. b, The brain weight of knockout (n = 4) mice is significantly lower than that of wild-type (n = 4) littermates. c, Gross morphology of brains and sagittal sections of brains. Sagittal brain sections of 8-week-old wild-type and knockout mice were stained with NeuN to visualize neuronal cells and Mfsd2a polyclonal antibody to visualize expression of Mfsd2a. Mfsd2a is shown to be widely expressed in brain. Scale bars, 1 mm. d, Haematoxylin and eosin (H&E) staining of hippocampus region of 8-week-old wild-type and knockout mice, indicating a smaller hippocampus in knockout mice. Scale bars, 500 µm. ***P < 0.001. Data are expressed as mean ± s.e.m.
Extended Data Figure 4 Mfsd2a-knockout mice exhibit deficits in learning and memory, and severe anxiety.
a–d, The Y-maze test (a, b) and novel-object recognition test (c, d) were used to assess spatial learning, short-term memory (STM) and long-term memory (LTM) of the wild-type and knockout mice, respectively. Knockout mice exhibited significantly decreased total arm entries in a Y-maze test for spatial working memory. Knockout mice showed significantly reduced preferences for novel objects in novel-object recognition tests, indicative of defects in short-term memory and long-term memory, respectively. Train, the training period. e–l, Zero-maze tests (e–h) and light–dark box tests (i–l) were used to assess anxiety of the wild-type and knockout mice, respectively. Knockout mice showed decreased transitions and head dips into open arms during Zero-maze test for anxiety behaviours. Knockout mice showed decreased entry into light box and increased latency to enter light box during light–dark box tests for anxiety. m–o, Open-field test for activity. Knockout mice showed reduced travel distance in the open-field test for locomotor activity. During the open-field test, knockout mice had no vertical activity indicative of motoric dysfunction, and decreased time spent in the centre, indicative of reduced exploration compared to wild-type mice. The increased time spent in the corners of the open field suggests that knockout mice were more anxious than wild-type mice, and are congruent with our results from the Zero-maze and light–dark box tests. Wild-type mice (n = 11–13) and knockout mice (n = 8–10). ***P < 0.001, **P < 0.01, *P < 0.05. Data are expressed as mean ± s.e.m.
a, Thin-layer chromatography (TLC) analysis of phospholipids and neutral lipids of HEK293 cells transfected with mouse Mfsd2a and human Mfsd2a after overnight incubation with 100 µM [14C]DHA. Std, unesterified [14C]DHA. Experiments were repeated three times with triplicates. b, TLC analysis of phospholipids and neutral lipids of HEK293 cells transfected with mouse Mfsd2a and mutants after overnight incubation with 100 µM [14C]oleate. The TLC protocol used was described in the Methods. CE, cholesteryl ester; PC, phosphatidylcholine; PE, phosphatidylethanolamine; TAG, triglyceride. c, Localization of Mfsd2a, D92A and D96A at the plasma membrane (red). d, Western blot analysis of expression of Mfsd2a, D92A and D96A in HEK293 cells post 24 h transfection. e, Biological incorporation of radiolabelled LPC [14C]DHA into PC. f, Biological incorporation of radiolabelled LPC[14C]oleate into PC. Cells expressing human (e, f) or mouse (f) Mfsd2a were incubated with LPC [14C]DHA or 50 µM LPC [14C]oleate. Lipids were extracted from cells after 30 min incubation with LPC [14C]DHA and 120 min incubation LPC [14C]oleate and analysed using TLC method for resolving PC and LPC. Experiments were repeated two times with triplicate. g, h, Dose-dependent transport of LPC [14C]DHA (g) and LPC [14C]oleate (h) by human Mfsd2a (hMfsd2a) and empty plasmid (mock) expressing HEK293. These experiments were performed in triplicate. i, Time-dependent transport of 50 µM LPC [14C]oleate. Mouse Mfsd2a (wild-type) and mutant constructs D92A and D96A were tested for uptake of radiolabelled LPCs at indicated times. This experiment was performed in triplicate. j, Increased net uptake of LPC ligand in cells expressing mouse Mfsd2a. TLC analysis of phospholipids of HEK293 cells transfected with mouse Mfsd2a and mutants after 1 h post incubation with 100 µM unlabelled LPC oleate. Shown numbers are fold changes of PC levels relative to mock. Experiments were repeated three times with duplicates. k, Transport activity of mouse Mfsd2a (wild-type), D92A, D96A, and mock expressing cells was not significantly different at indicated pHs. This experiment was performed in triplicate. l, Activity of Mfsd2a is sodium- but not lithium-dependent. Data were expressed as fold change of Mfsd2a expressing cells compared to corresponding mock cells treated with the same conditions. This experiment was performed in triplicate. m, Transport activity of Mfsd2a is not BSA-dependent as LPC palmitate solubilized in either ethanol or micellular form was transported by Mfsd2a, albeit to a lower level than with BSA. This experiment was performed in triplicate. Data are expressed as mean ± s.e.m.
All competition assays were performed using 25 µM LPC [3H]palmitate as ligand with or without tenfold molar excess (250 µM) of the indicated competitors. a, The structures of the lipid competitors used in b and c. b, Competition assays with indicated acyl chain LPCs. c, Competition assay with indicated headgroups. Assays were stopped after 30 min of incubation. Competitive activity was expressed as percent to control (activity of Mfsd2a without competitor). 6:0, hexanoate; 8:0, octanoate; 10:0, docanoate; 12:0, laurate; 14:0, myristate; 16:0, palmitate; 18:0, stearate; 18:1, oleate; GPC, alpha-glycerylphosphocholine; LPA, lysophosphatidic acid; LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; LPS, lysophosphatidylserine. d, Representative structures of bioactive lipid competitors used in e and f. e, Competition assay with lysophospholipid forms of plasmalogens and platelet activating factor (PAF). This experiment was performed together with b, so that the control and mock shown in b can be used for reference. f, PAF and lysosphingomyelin (lysoSM) also showed strong competition, whereas sphingosine 1-phosphate (S1P) did not compete for LPC [3H]16:0 uptake. Competitive activity was expressed as percent to control (activity of Mfsd2a without competitor). g, Representative structure of non-biological lysophospholipid analogues foscholine-16 (Miltefosine) with an alkyl chain of 16 carbons. h, Competition assays of indicated foscholines with LPC [3H]palmitate. Assays were stopped after 15 min of incubation. Competitive activity was expressed as percent to control (activity of Mfsd2a without competitor). Foscholine with alkyl chain length of 8 (Fos-8), 10 (Fos-10) and 12 (Fos-12) carbons did not compete, whereas foscholine with an alkyl chain length of 16 carbons (Miltefosine) showed strong competition with LPC [3H]16:0. These experiments were repeated two times with triplicates. ***P < 0.001. Data are expressed as mean ± s.e.m.
Thin-layer chromatography (TLC) analysis of phospholipids of HEK293 cells transfected with mouse Mfsd2a and mutants after 30 min post incubation with 25 µM TopFluor LPE. a, TLC analysis of phospholipids. b, Quantification of intensity of PE band from TLC plate. PE, phosphatidylethanolamine; LPE, lysophosphatidylethanolamine. Experiments were repeated two times with duplicates. Data are expressed as mean ± s.e.m.
Extended Data Figure 8 Brain uptake of unesterified [14C]DHA was not reduced in Mfsd2a-knockout mice.
Male mice aged 7 weeks old were intravenously injected with 1 mmol of [14C]DHA-BSA complex. Brain, liver and heart were collected 2 h post injection for lipid extraction and DPM quantified using scintillation counting. a–c, Uptake of unesterified [14C]DHA in the wild-type and knockout brain, heart and liver were expressed as DPM per g. d–f, Level of uptake of DHA in a–c was converted into nmol per g. g–i, A comparison between the absolute amount of DHA uptake in the form of LPC DHA (converted from Fig. 4 into nmol per g in 2 h) and unesterified DHA (taken from d–f above) in brain, heart and liver of wild-type mice. The same amounts of LPC DHA and DHA were injected in mice. The amount of LPC DHA uptake was far greater than unesterified DHA uptake by wild-type brain. ***P < 0.0001, *P < 0.05. Data are expressed as mean ± s.e.m. (wild-type, n = 5; knockout, n = 5).
This experiment was carried out as described for NBD LPC in Fig. 4. a, structure of TopFluor LPC. b, HEK293 cells expressing wild-type Mfsd2a showed significantly enhanced uptake activity to TopFluor LPC compared with mock (empty plasmid), D92A and D96A mutant expressing cells. c, TLC analysis showed that TopFluor LPC was bio-incorporated into PE. Experiments were repeated two times with duplicates. d, Quantification of PE band from the TLC plate shown in c. e, Brain uptake of TopFluor LPC was decreased in knockout mice. Male mice (wild-type, n = 3; knockout, n = 3) aged 7 weeks old were intravenously injected with 300 µg TopFluor LPC–BSA complex. f, Fluorescence from 10 brain sections of wild-type and knockout mice was quantified and expressed as fluorescence intensity per pixel. ***P < 0.001. Data are expressed as mean ± s.e.m.
Heterozygous female mice were gavaged with 100 µl DHA oil (containing 26% DHA triglyceride and 6% EPA, total omega-3 is 35%) every 2 days for 2 weeks before conception in crosses with heterozygous males. During gestation, pregnant mice were continued on gavages of DHA every 2 days. Gavages of mothers continued during breastfeeding and pups were weaned onto normal diet at 3 weeks of age and gavaged every 2 days with DHA for 8 weeks. a, Brain weight of adult wild-type (n = 4) and knockout (n = 4) mice aged 8 weeks after treatment with dietary DHA oil. Knockout mice brains were still significantly smaller. b–e, DHA treatment of knockout mice did not reduce the strong anxiety phenotype as determined using the light–dark box test. To investigate why dietary DHA failed to rescue knockout phenotypes, we tested the hypothesis that uptake of maternally derived DHA (in this case the DHA delivered to the mother via gavage) might not get into the brain of knockout mice during brain development. To test this possibility, pregnant HET mothers intercrossed with knockout fathers were gavaged at E17.5–E19.5 with [14C]DHA and uptake into fetal brains was quantified. Note that heterozygous mice do not exhibit detectable phenotypes, have similar DHA to wild type (not shown) and are thus similar to wild type. The rationale for using HET and knockout intercrosses was to increase the yield of knockout mice in this study. The data shown in f indicate that brains of knockout mice exhibited an 80% reduction in the uptake of [14C]DHA relative to HET mice with the same mothers (n = 4 wild-type, n = 6 knockout). Therefore, Mfsd2a expressed during fetal development is important for DHA transport into brain. *P < 0.05, ***P < 0.001. Data are expressed as mean ± s.e.m.
This file contains a Supplementary Discussion and additional references. (PDF 143 kb)
Mass spectrometry analysis of free fatty acid uptake in HEK239 cells. Mfsd2a expressing HEK293 cells were incubated with 100µM of indicated fatty acid/BSA complex overnight. Lipids extraction and phospholipid analysis by MS were performed as described in Methods section. Amount of each lipid species was normalized to internal standard and expressed as mol percent in total phospholipid analysed. (XLSX 145 kb)
About this article
Cite this article
Nguyen, L., Ma, D., Shui, G. et al. Mfsd2a is a transporter for the essential omega-3 fatty acid docosahexaenoic acid. Nature 509, 503–506 (2014). https://doi.org/10.1038/nature13241
This article is cited by
Inflammation and Regeneration (2023)
Nature Communications (2023)
Improvement of retinal function in Alzheimer disease-associated retinopathy by dietary lysophosphatidylcholine-EPA/DHA
Scientific Reports (2023)
Transmembrane protein 135 regulates lipid homeostasis through its role in peroxisomal DHA metabolism
Communications Biology (2023)
Nature Communications (2023)