Keywords

14.1 Major Facilitator Superfamily

Lipids are organic compounds that are essential in living cells. Mammalian cell membranes are largely made up of glycerolipids, phospholipids, and cholesterol, organized into a lipid bilayer. The transport of molecules across this hydrophobic membrane is vital for cell growth, metabolism, and signal transduction, and are facilitated by transport proteins such as channels and transporters. Primary active transporters like ATP-binding cassette transporters are fueled by energy released from ATP hydrolysis. Conversely, secondary facilitative transporters do not utilize ATP hydrolysis for transport and can be generally categorized as facilitative or active facilitate. The former transports solutes down their concentration gradients across membranes, while active facilitative transporters transport solutes against their concentration gradients either as symporters or antiporters. Facilitative active transporters derive their energy for transport through coupling solute transport with the transport of ions such as sodium or protons down their concentration gradients across membranes [45, 60, 75].

The major facilitator superfamily (MFS) is one of the largest families of secondary transporters. The vast majority of characterized MFS transporters transport minimally polar hydrophilic substrates such as mono- and disaccharides, amino acids, and nucleosides [45, 53]. However, there are three MFS transporters that are exceptions to this general feature of the MFS family, namely, Spns2, Mfsd2a, and Mfsd2b [39, 52, 73]. Spns2 and Mfsd2b are sphingosine-1-phosphate transporters [39, 73], while Mfsd2a, the subject of this review article, is a lysophosphatidylcholine transporter.

MFS proteins have a highly conserved fold that is composed of 12 transmembrane alpha helices separated into two 6-transmembrane units that exhibit a pseudo twofold symmetry about an axis perpendicular to the membrane plane [45, 53]. Within each of these 6-transmembrane domains, the two 3-transmembrane units are organized as inverted repeats [45, 60, 75].

The majority of crystal structures of MFS proteins are bacterial proteins, with some exceptions being human glucose transporters [26, 27, 75], where the first MFS structures to be elucidated are that of Escherichia coli Lactose:H+ symporter (LacY) [1] and Escherichia coli Glycerol-3-phosphate:Pi antiporter (GlpT) [36]. A common transport mechanism that has been proposed as a result of these structural and biochemical studies is that substrates are transported in a rocker-switch, alternating access mechanism [75]. The N- and C- terminal domains rotate about the central substrate binding site and open exclusively to either the cytoplasm or extracellular space at any one time, rocking between an inward open or outward open state [1, 36, 45]. The four important transmembrane helices that surround the central pocket and essential for transport activity are domains 1, 4, 7, and 10. Transmembrane domains consisting of 2, 5, 8, and 11 or 3, 6, 9, and 12 are positioned just outside the core helices and mediate the interference between the N- and C- domains and support the structural integrity of the transporter, respectively [75].

14.2 Major Facilitator Superfamily Containing 2a

Major facilitator superfamily domain containing 2a (MFSD2A) was first identified by Angers et al. as an orphan transporter to be significantly induced in brown adipose tissue (BAT) of mice lacking both nuclear receptors retinoid-related orphan receptor alpha and gamma (RORα and RORγ) [5]. The Mfsd2a gene is approximately 14.3kb long, with 14 exons and 13 introns. Analysis of the amino acid sequence indicates it is most closely related to the bacterial-sodium melibiose symporter MelB at a 43–37% similarity [29]. Importantly, amino acid sequence of both mouse and human MFSD2A proteins is approximately 85% identical and is highly conserved from fish to human [11].

Mfsd2a is expressed in the brain, spinal cord, BAT, liver, kidney, lung, placenta, testes [11], and eye [74]. mRNA expression of Mfsd2a is greatly induced in murine liver and BAT during fasting and follows an oscillatory expression profile consistent with a circadian rhythm, with peak expression at circadian time 12 [5]. Additionally, Mfsd2a mRNA was also significantly upregulated exclusively in BATs by cold exposure and β-adrenergic receptor signaling pathway [5]. Berger et al. identified Mfsd2a to be induced by fasting and regulated by both peroxisome proliferator-activated receptor alpha (PPARα) and glucagon signaling in the liver, which turns over rapidly in liver upon refeeding [11]. While mRNA can be detected in BAT, Mfsd2a protein level is extremely low [11]. The function of Mfsd2a in BAT has not been determined.

14.3 Lipids and Essential Fatty Acids Are Important for Brain Growth

The brain is made up of glycerophospholipids, cholesterol, and sphingolipids, making it one of the most lipid-rich organs in the body [41]. Prenatal brain development is a complex developmental process that begins with the development of the neural tube, which ultimately differentiates into the brain and spinal cord. This is also the time where hundreds of specialized cell types come together, organizing a network of synaptic connectivity and a functioning blood-brain barrier (BBB) (Fig. 14.1) [6, 24, 61]. The BBB separates the brain from blood and serves to maintain a tightly controlled environment where toxins and pathogens are prevented from freely entering or leaving the brain by diffusion. The BBB is governed by tight junctions of endothelial cells of blood vessels, supported by astrocytes and pericytes [6, 24]. This is followed by postnatal brain growth, which is accompanied by the proliferation of astrocytes and oligodendrocytes [14, 28, 42] and myelination of axons and synaptogenesis [8, 28, 51]. Massive amounts of membrane phospholipids are therefore required for brain growth, where it has been postulated that lipids are derived exclusively from de novo biosynthesis within cells of the brain.

Fig. 14.1
figure 1

Blood-brain barrier. The BBB is governed by tight junctions (TJ) of endothelial cells (EC) of blood vessels, supported by astrocytes (A) and pericytes (P)

Full size image

De novo lipogenic gene expression is controlled by sterol regulatory element-binding proteins (Srebp-1 and Srebp-2). In support of the vital role of de novo lipogenesis in brain development, the genetic deficiency of Scap, an essential chaperone protein for Srebp, in neurons in the developing central nervous system resulted in microcephaly and early postnatal lethality [65]. In addition, deficiency of Scap in mature astrocytes and oligodendrocytes have profound effects on myelination [71].

14.4 LPC-DHA Transport into the Brain

Docosahexaenoic acid (DHA) is an omega-3 fatty acid composed of 22 carbons and 6 double bonds. DHA can be synthesized by the liver through chain extension and desaturation of the essential fatty acid linolenic acid. DHA is highly enriched in brain phospholipids, particularly in the phosphatidylethanolamine (PE), phosphatidylserine (PS), and to a lesser extent, phosphatidylcholine (PC) pools within membranes, and comprises up to 15% or more of the total fatty acid composition of the prefrontal cortex [13, 50]. In humans, DHA is rapidly taken up as early as the end of the second trimester, coinciding with the development of the BBB where considerable amounts of membrane phospholipids are required for the growing brain [17, 38, 61]. DHA is continuously acquired from early postnatal days until approximately 2 years of age [23, 47, 66]. While DHA supplementation studies in term infants or pregnant and lactating women have been inconclusive for enhancing cognitive development [25, 31, 55], DHA supplementation in preterm infants has shown some benefit to cognitive development, presumably because preterm infants might have lower brain DHA levels [7, 18]. Likewise, decreased levels of DHA in the developing brain have been associated with negative effects on cognitive function [33, 49] and neurodevelopmental disorders [19, 35, 48]. Importantly, DHA itself cannot be de novo synthesized and must be transported across the BBB into brain.

The form by which DHA gets taken up into brain, either as unesterified DHA or DHA esterified as lysophosphatidylcholine-DHA (LPC-DHA), has been a point of debate. LPCs circulate in blood bound to albumin [20, 56, 67] where it was first shown by Illingworth and Portman to be taken up and reacylated readily in brains of squirrel monkeys [37]. As early as 1965, it was hypothesized by Switzer and Eder that plasma LPCs serve as precursors for the renewal of cellular membranes [67]. Importantly, Thiès et al. reported a preference for unsaturated fatty acids esterified as 2-acyl-LPC in young rat brains where LPC-DHA was transported 12-fold more than unesterified DHA, suggesting that LPCs might be an efficient delivery of polyunsaturated fatty acids (PUFAs) into the developing brain [69, 70]. Moreover, Lagarde et al. was the first to propose that LPC is the preferred carrier of PUFAs like DHA or arachidonic acid (AA) to the brain [43]. As will be further discussed below, Mfsd2a is the LPC transporter that explains the LPC transport activity first described by Lagarde and co-workers. More recently, it was demonstrated that supplementing adult mice with dietary LPC-DHA, but not unesterified DHA, were able to increase brain DHA levels twofold [64]. Collectively, these findings support the conclusion that LPC-DHA, and not unesterified DHA, is the primary carrier of DHA delivery to the brain. However, Mfsd2a KO mice have residual phospholipid containing DHA in the brain and eye, indicating the possibility of either compensatory de novo biosynthesis, other transport mechanisms, or acquisition of DHA during embryogenesis in the brain and eye prior to blood-barrier formation. It is important to note that single cell sequencing projects and bulk RNA-seq of the blood-brain barrier in mice [68, 72, 79] have shown that mRNA expression for proteins proposed to be involved in the uptake of unesterified DHA by the BBB endothelium, such as LPL, and its essential chaperone GPIHBP1 [78], CD36, and FATP1-6 (Slc27a1-6), and ACSL6 are not expressed by the endothelium of the BBB.

14.5 Mfsd2a Deficiency in the Brain

Importantly, Nguyen et al. and Ben-Zvi et al. discovered Mfsd2a to be highly expressed at the endothelium of the BBB [10, 52]. Through targeted lipidomic analysis, Mfsd2a was found to be the major pathway for brain DHA accretion, where a significant 60–70% reduction in steady-state levels of total percentage DHA-containing phospholipids was observed in brains of 2aKO mice relative to wild-type controls [15, 52]. Conversely, brains of 2aKO mice had a modest 35% increase in steady-state levels of total percentage AA-containing phospholipids [52], a phenomenon commonly observed in rodent models of DHA deficiency [62].

More recently, using endothelial-specific and inducible endothelial-specific Mfsd2a deletion mouse models, Chan et al. showed that Mfsd2a deficiency results in a unique form of postnatal microcephaly, with DHA deficiency preceding the onset of microcephaly [15]. Only adult 2aKO mice exhibit a minor loss of Purkinje cells in the cerebellum and a decrease in neuronal cell density in the CA1 and CA3 regions of the hippocampus [52]. Because the brains of 2aKO embryos are deficient in DHA but are not microcephalic until postnatal life, these cell loss phenotypes are secondary events. These findings also indicate that DHA deficiency is an unlikely cause underlying microcephaly, but rather the absence of bulk LPC transport, where LPCs are phospholipid membrane building blocks.

Recently, transcriptomic and lipidomic analysis in Mfsd2a deficiency mouse models was used as a tool to understand how the brain adapts to DHA deficiency, thus revealing functions of DHA in the brain [15]. It was discovered that Mfsd2a deficiency resulted in a de-repression of the Srebp1 and Srebp2 pathways leading to an increase in de novo synthesis of unsaturated fatty acids in phospholipids. It was shown that Mfsd2a is expressed in neural stem cells (NSCs) isolated from early postnatal mice and that NSCs treated with LPC-DHA and other LPC-PUFAs can acutely downregulate Srebp1 and Srebp2 target gene expression in an Mfsd2a-dependent fashion and that the mechanism is in part through inhibition of Srebp-1 receptor processing [15]. Moreover, Mfsd2a itself is regulated by Srebp, forming a negative feedback loop on Srebp processing that can balance de novo lipogenesis with exogenous uptake of LPC-DHA. The regulation of brain Srebp function by LPC-DHA transported by Mfsd2a might serve the purpose of fine-tuning membrane phospholipid saturation and hence biophysical properties during brain development [15].

Another reported feature of Mfsd2a deficiency in the brain and eye is that Mfsd2a knockout mice have increased transcytosis resulting in increased BBB permeability [4, 10]. It has been suggested that the microcephaly and DHA deficiency in 2aKO mice could be due to a leaky BBB, but it is unclear how a leaky BBB would result in less DHA uptake and not more relative to wild-type (WT) mice. Nonetheless, this issue has been resolved in that BBB permeability, but not microcephaly and DHA deficiency, can be completely rescued in Mfsd2a-deficient mice by genetic deficiency of Cav1 [4]. Andreone et al. generated a transporter-dead Mfsd2a knockin mouse model bearing a D96A aspartate to alanine point mutation, a conserved residue with D97 in the human Mfsd2a constituting the sodium binding site, and showed that consistent with the lack of transport activity, Mfsd2aD96A/D96A mice exhibited microcephaly and DHA deficiency in the brain [4]. These findings indicate that microcephaly and DHA deficiency are primary phenotypes of Mfsd2a deficiency, and not a result of a leaky BBB [4], and that LPC transport via Mfsd2a is essential for DHA accretion and postnatal brain growth. Of note, the increased transcytosis phenotype in the BBB or blood-retina barrier of 2aKO mice reported by the Gu lab [4, 10, 16] has not been observed in other studies [46, 74].

14.6 Mfsd2a Deficiency in the Eye

The retina is a highly organized structure, with photoreceptors (PR), extensive retinal glial network, and retinal pigment epithelium (RPE) organized into distinct layers. Rods and cones are the two types of PR found in mammalian eyes, which make up 70% majority of cells in the retina. DHA, localized with rhodopsin [30], is found primarily in phospholipids of membrane discs that make up rod PR outer segments (OS), making the retina a tissue with the highest concentration of DHA per unit area in the body [58]. With daily daylight exposure, OS discs which are photosensitive, accumulate photo-damaged proteins and lipids [9] and must be synthesized continuously throughout one’s lifetime for the maintenance of healthy vision [63, 77]. The villi-containing apical membrane of the RPE is particularly important for this renewal process, where through its interaction with the distal ends of the OS (Fig. 14.2) facilitates the daily phagocytosis of OS discs that make up one-tenth of the OS length. This process of phagocytosis is balanced with an equal rate of disc regeneration, so that the OS length is maintained [76], thus highlighting the importance of lipids and essential fatty acids for membrane biogenesis and turnover.

Fig. 14.2
figure 2

Blood-retinal barrier. The BRB is made up of the inner BRB, formed by tight junctions (TJ) of the endothelium of retinal capillaries (EC), supported by pericytes (P), astrocytes (A) and Müller cells (MC). The outer BRB is governed by TJ of the retinal pigment epithelium (RPE). DHA is found primarily in phospholipids of the outer segment (OS) discs of rod and cones and interact closely with the apical membrane villi of the RPE. As photo-damaged discs need to undergo a constant renewal process for the maintenance of vision, inner segments that contain metabolic machinery synthesize new membrane discs that move along the length of the OS where they are eventually phagocytosed by the RPE. CB cell body, IS inner segments, OS outer segments, AM apical membrane, RPE retinal pigment epithelium, BI basal infoldings, Ch choroid

Full size image

Similar to the BBB of the brain, the eye contains cellular barriers that prevent the diffusion of blood-borne material or lipids from entering the retina freely. The eye contains two blood-eye barriers (Fig. 14.2), the inner blood-retina barrier (inner BRB) that is established by tight junctions between retinal endothelial cells, supported by the pericytes, astrocytes, and Müller cells [22, 59] and the outer BRB that is governed by tight junctions of the RPE [21].

Mfsd2a is expressed at the endothelium of the BRB and RPE. The RPE is the major site of Mfsd2a expression and is quantitatively important for DHA accretion into the retina via LPC-DHA transport [74] (Fig. 14.3). Whole-body Mfsd2a-deficient (2aKO) mice displayed a unique form of a slow, progressive retina degeneration [46, 74]. However, a 40% deficiency in phospholipids containing DHA in eyes of 2aKO mice did not result in the expected severe and rapid retinal degeneration nor significant visual dysfunction [46, 74]. Like the brain, upregulation of de novo lipogenesis pathways was observed in eyes of 2aKO mice which might serve, as a compensatory mechanism to synthesize new OS discs in the absence of Mfsd2a [46, 74]. In addition, the BRB was found to be intact in 2aKO mice [46, 74], which is inconsistent with a report that Mfsd2a is required to suppresses transcytosis for the development and maintenance of a functional BRB [16]. This discrepancy is not due to strain-specific differences as the strain used in the Lobanova study was the same as reported by Chow et al. [16]. The most remarkable finding from studying 2aKO retinas is that phototransduction tested by electroretinography [74] or light evoked potential recordings of single rods [46] indicated that phototransduction in 2aKO and WT was indistinguishable. These findings might suggest that the compensatory changes in lipid composition in 2aKO retinas of increased monounsaturated fatty acids and arachidonic acid in phospholipid pools might compensate for the severe reduction in DHA.

Fig. 14.3
figure 3

MFSD2A transports LPC-DHA across the BBB and BRB. DHA can come preformed from the diet or its precursor linolenic acid, conjugated to LPC in the liver, and transported in blood plasma bound to albumin. At the BBB, Mfsd2a translocates LPC-DHA across the endothelial plasma membrane into the brain. Mfsd2a is expressed at both the inner and outer BRB, but Mfsd2a at the RPE is the major route by which LPC-DHA gets into the eye. IS inner segments, OS outer segments, RPE retinal pigment epithelium, Ch choroid

Full size image

14.7 Inactivating Mutations of MFSD2A in Humans

To date, four unrelated consanguineous families with homozygous non-synonymous inactivating mutations in MFSD2A have been identified that presented with severe microcephaly and intellectual impairments [2, 32, 34]. The first two families, one from Libya and the other from Egypt, harbored a p.Thr159Met or p.Ser166Leu protein change [32]. Mutant Mfsd2a proteins were stably expressed and localized to the plasma membrane when expressed in HEK 293 cells, comparable to WT Mfsd2a, but had complete inactivation of transport activity [32]. The third family from Pakistan was a large pedigree, with ten affected family members harboring a p.Ser339Leu protein change that presented with severe non-lethal microcephaly [2]. Again, mutant Mfsd2a proteins were stably expressed and had proper membrane localization when expressed in HEK 293 cells but exhibited a partial inactivation of transport activity relative to WT protein [2]. A fourth family was identified in Israel that harbored a p.Pro402His protein change, with complete inactivation of transport activity, and presented with severe non-lethal microcephaly [34]. Consistent with reduced or complete inactivation of transport activity that would be expected to reduce brain and eye LPC uptake, increased plasma LPC levels have been observed in all affected family members [2, 32, 34]. In further support of this explanation for increased plasma LPC in patients with inactivating mutations in Mfsd2a, plasma LPC levels were also found to be increased by 40% in 2aKO mice, consistent with 85–90% reduction in LPC transport in the brain and eye using tracer studies [32, 52, 74].

Both p.Thr159Met and p.Ser166Leu mutations were found on transmembrane domain 4 of Mfsd2a, p.Ser339Leu was found on transmembrane domain 8, while p.Pro402His was found on the extracellular loop between transmembrane 10 and 11. A molecular explanation for the loss-of-function caused by Ser166Leu and Pro402His is not known. However, Thr159met is homologous to Thr121 in MelB, which is essential for establishing a hydrogen bond with conserved aspartate residues at the sodium binding pocket. Therefore, it can be predicted that Thr159Met inactivity is a consequence of absence of sodium binding [32].

14.8 Proposed Transport Mechanism of Mfsd2a

Mfsd2a does not transport unesterified PUFAs, but PUFAs esterified as a LPC [52]. It was determined through structure-activity relationship studies that lysophospholipid with a minimal acyl chain length of 14 carbons and a zwitterionic headgroup (e.g., PC, PE, and PS) is essential for transport by Mfsd2a [52]. More recently, Quek et al. showed that the acyl-carnitines can also be transported by Mfsd2a, again underscoring the importance of a zwitterionic headgroup and not strictly a phosphorylcholine headgroup as a necessary feature for lysolipid transport [57]. Notably, Mfsd2a has a higher transport capacity for LPCs having unsaturated fatty acids like DHA relative to LPCs with saturated fatty acids like palmitate [52]. This latter finding is important, because it indicates that LPC transport capacity is inversely correlated with the physiological levels of LPCs in human plasma, where LPC-palmitate is the most abundant [2, 32, 56]. Presumably, this preference for LPC-PUFA by Mfsd2a would allow the brain to obtain the lower abundant essential fatty acids diluted in a larger milieu of LPCs containing non-essential fatty acids.

Using homology modeling based upon crystal structures of MelB and LacY, and further refinement by site-directed mutagenesis and biochemical transport analysis, Quek et al. identified the following four important structural features of human Mfsd2a: a sodium binding site, a hydrophobic cleft, a lipid phosphate headgroup binding residue (Lys436), and ionic locks [57]. The hydrophobic cleft is likely involved in LPC acyl chain binding, while the Lys436 is involved in coordinating the LPC phosphate headgroup interaction. The ionic locks are presumably involved in stabilizing the outward open conformation during the transport cycle as previously proposed for similar ionic locks identified on MelB [29]. This proposed model of transport co-opts the standard rocker-switch model, with the exception that LPCs bound to albumin would first bind to the outer leaflet of the plasma membrane and diffuse laterally into Mfsd2a facing the outward open conformation until hydrophobic forces position the acyl chain of the LPC into the hydrophobic cleft and headgroup binding to Lys436 (Fig. 14.4). Sodium binding to its binding site comprising residues Asp93, Asp97, and Thr159 would drive a conformational change to an inward-open conformation that would push the LPC-DHA down along the hydrophobic cleft and flip over to the inner lipid leaflet, where it exits the transporter by diffusing laterally along the inner membrane [57]. This “flipping” activity would in theory allow LPCs to bypass the tight junctions of the BBB endothelium [12]. Once LPCs reach other cells at the BBB such as astrocytes, it could be converted to PC-DHA through activity of the LPCATs [44].

Fig. 14.4
figure 4

Proposed mechanism of LPC transport. (a) Mfsd2a in the outward-open conformation facing the extracellular surface. (b) Sodium binds to sodium binding site, while LPC diffuses laterally into the central cavity and binds to Lys436. (c) In the presence of sodium, Mfsd2a undergoes a conformational change, and LPC “flips” along hydrophobic cleft. (d) LPC diffuses laterally into the inner leaflet and converts to PC-DHA

Full size image

14.9 Concluding Remarks

Mfsd2a is a sodium-dependent lysophosphatidylcholine co-transporter highly expressed at the blood-brain barrier and blood-eye barriers that is essential for normal human brain development. Mfsd2a shows high specificity for the transport of LPCs with long chain and unsaturated fatty acyl chains. LPC-DHA in particular negatively regulates Srebp activity during brain development, and this function is likely important to maintain proper membrane phospholipid saturation. An important question that remains to be answered is a determination of the transport mechanism of LPCs by Mfsd2a. This determination awaits the development of new biochemical assays to reconstitute transport on purified Mfsd2a and the determination of atomic resolution structures. Interestingly, Mfsd2a is expressed by other cell types and tissues such as liver and a determination of the function of Mfsd2a outside of the brain and eye will likely reveal new biology into the function of LPCs. For example, Piccirillo and others have shown that Mfsd2a is required for the maintenance of memory T cells [54], perhaps in part through TOX, which might regulate Mfsd2a [3, 40]. Lastly, a word of caution, many recent papers have been published using non-validated Mfsd2a antibodies that are likely leading to erroneous conclusions on the regulation of, site of, expression of, and involvement of Mfsd2a in particular biological and pathophysiological processes. It is critical that Mfsd2a antibodies be validated using both cell-based overexpression and Mfsd2a deficiency cell or mouse models.