The role of SLC transporters for brain health and disease

Abstract

The brain exchanges nutrients and small molecules with blood via the blood–brain barrier (BBB). Approximately 20% energy intake for the body is consumed by the brain. Glucose is known for its critical roles for energy production and provides substrates for biogenesis in neurons. The brain takes up glucose via glucose transporters GLUT1 and 3, which are expressed in several neural cell types. The brain is also equipped with various transport systems for acquiring amino acids, lactate, ketone bodies, lipids, and cofactors for neuronal functions. Unraveling the mechanisms by which the brain takes up and metabolizes these nutrients will be key in understanding the nutritional requirements in the brain. This could also offer opportunities for therapeutic interventions in several neurological disorders. For instance, emerging evidence suggests a critical role of lactate as an alternative energy source for neurons. Neuronal cells express monocarboxylic transporters to acquire lactate. As such, treatment of GLUT1-deficient patients with ketogenic diets to provide the brain with alternative sources of energy has been shown to improve the health of the patients. Many transporters are present in the brain, but only a small number has been characterized. In this review, we will discuss about the roles of solute carrier (SLC) transporters at the blood brain barrier (BBB) and neural cells, in transport of nutrients and metabolites in the brain.

Introduction

The brain is the second heaviest organ in the human body, which consumes approximately 300 kcal equivalent to 20% of the total daily energy for functional activities [1, 2]. It is known that glucose is the primary fuel source for the high energy demand of the brain. Additionally, other nutrients, such as short-chain fatty acids, ketone bodies, and amino acids, are also utilized by the brain. The brain is protected from potentially harmful substances in the bloodstream via the blood–brain barrier (BBB). Nutrient delivery across the BBB is granted through a tightly controlled transport system. As such, endothelial cells in the capillaries of the BBB are expressed with numerous solute transporters for the import and export of nutrients. These cells are also equipped with vesicular transport systems to ensure the adequate transports of nutrients.

The BBB is a functionally specialized structure of the central nervous system (CNS)-associated blood vessels. It regulates the interaction of neural cells with blood components. The BBB is a part of the neurovascular unit (NVU), which comprises several cell types. The first layer of NVU is endothelial cells, which forms the inner most layer of blood vessels. The CNS blood vessels are stabilized by the enwrapping of pericytes and astrocyte endfeet (Fig. 1). Neurons are also a part of NVU where they innervate blood vessels to regulate blood flow. This ensures that a sufficient level of oxygen is provided in response to an increase in neuronal activity. Except for the circumventricular organs and parts of the hypothalamus, where blood vessels are permeable to hormones and small peptides, blood vessels in the remaining parts of the CNS are impermeable to water-soluble molecules. Thus, the CNS vasculature is equipped with two major transport systems, the vesicular transcytosis and protein-mediated transport. The vesicular transcytosis pathway comprises clathrin- and caveolae-mediated transcytosis. However, this pathway only occurs in the larger segments of CNS vasculature, such as the arteries and veins, but not in the capillaries, which account for approximately 85% of the CNS vasculature [3]. The protein-mediated transport pathway includes the activity of transporters, such as ATP-binding cassette transporters (ABC transporters) and solute transporters (SLC transporters). The latter facilitates the import and export of small molecules at the BBB and between neural cells. Interestingly, SLC transporters are highly enriched in the capillaries of CNS vessels, but not in the larger blood vessels [4]. As such, the transport of nutrients mediated by these solute transporters is anticipated to play a key role in the development and functions of the brain. In addition, the BBB also acts as a major obstacle for the entry of therapeutic compounds for treatment of neurological disorders. In this review, we will summarize the current knowledge of the solute transporters with a focus on their roles in the transport of nutritional molecules across the BBB and between neuronal cells. We highlight the physiological and molecular roles of glucose transporters (GLUTs), monocarboxylate transporters (MCTs), amino acid transporters, and lipid transporters in the brain.

Fig. 1.

Schematic diagram of the major solute transporters at the blood–brain barrier (BBB) and neuronal cells. GLUT1 (primarily located in endothelial cells and astrocytes) and GLUT3 (located in neurons) facilitate glucose transport from blood to neurons. Glucose is either converted into lactate via glycolysis or directly transported via GLUT1 expression in the abluminal side, where neurons take up via GLUT3. MCT1 is responsible for lactate release in the endothelial cells in the BBB. Lactate sources could be generated from blood or endothelial glycolysis. Lactate is exported through MCT1/4 which is subsequently imported through MCT2 into neurons. Facilitative amino acid transporter is responsible for transport of amino acids that cannot be synthesized in the brain. Active amino acid transporters are mostly located in abluminal side. For lipid transports, Mfsd2a, which is located in the endothelial cells of the BBB, is the major transporter of lipid. Many solute transporters (“others”) are also expressed in the BBB. Future studies are warranted to dissect the molecular roles such as the identification of ligands. BBB blood–brain barrier; GLUT glucose transporter; MCT monocarboxylate transporter; Mfsd2a major facilitator superfamily domain-containing protein 2a; LPC phosphatidylcholines; FFA free fatty acids, PC phosphatidylcholines, AA amino acids

Glucose transporters (GLUTs)

Glucose is transported from the bloodstream, via the BBB to neuronal cells. A vast majority of glucose is directly delivered to neuronal cells for energy production. However, a part of glucose is broken down through glycolysis to generate adenosine triphosphate (ATP). Interestingly, endothelial cells of the CNS blood vessels convert glucose into lactate, which is then transported to neuronal cells via monocarboxylate transporters (MCTs, see below). Eventually, lactate is utilized by neuronal cells for energy production. Recent evidence showing that lactate metabolism is important in memory formation [5, 6]. Lactate and H⁺ are co-transported in the brain where they regulate the intracellular redox state and pH level in neuronal cells [7, 8]. Besides being the primary energy source, several metabolites from glucose are required for brain functions. For example, glucose is used for the pentose phosphate pathway for generation of precursors for nucleotide synthesis. A part of glucose is also converted into glycogen for storage in astrocytes [9]. This energy reserve has a role in learning activities [10, 11]. Glucose is also utilized to produce Acetyl CoA for de novo synthesis of fatty acids and cholesterol in membrane biosynthesis and myelination in neural cells [12, 13].

Glucose transporters in the brain

The physiological concentration of glucose in the bloodstream is maintained at 4–6 mM [14, 15]. After a caloric intake, blood glucose level rises up to 7–9 mM [16] that is higher than in the brain parenchyma, where the glucose level is maintained at 1–2 mM [15]. This difference in concentrations of glucose between blood and the brain facilitates its transport to neuronal cells. Glucose transport is mediated by a family of facilitated glucose transporters (GLUTs). It should be noted that GLUTs are different from the other glucose transporters at enterocytes and proximal tubules of the kidney [17], which belong to a family of active glucose transporters (SGTLs) encoded by SLC5 gene family. GLUTs belong to the solute carrier SLC2 genes of the major facilitator superfamily (MFS) [18, 19]. There are 14 GLUTs. These proteins show 28–65% amino acid sequence identity with GLUT1 [19]. GLUTs are transmembrane proteins with 12 helices containing an N-linked glycosylated position. These transporters possess a cytoplasmic linker domain in which their N and C terminus are facing the cytoplasm [18]. GLUTs are categorized into 3 classes based on their sequence. The main difference between class 1 and 2 to class 3 is the location of their N-linked glycosylation sites [19]. GLUTs are facilitative glucose transporters, meaning that they facilitate glucose transport from high to low concentrations without requiring ATP or sodium to drive the transport activity [1, 18, 20]. GLUTs operate via conformation changes to allow the passage of glucose through the transport cavity. Glucose binding to the outward-facing conformation will trigger a conformational change, which allows the movement of glucose to the cytosolic side. The transport cycle is reversed when glucose is released to the cytoplasm [21]. The recent crystalized structures of GLUTs have provided the mechanistic basis of how glucose is transported across membranes of neural cells [22, 23].

In the brain, GLUT1 and GLUT3 are the main glucose transporters with different kinetic properties. These two transporters play critical roles in neuronal functions. GLUT1 is localized in the endothelium and astrocytes while GLUT3 is found in neurons [24–27]. They are responsible for glucose uptake from blood to neural cells.

Glucose transporter 1 (GLUT1, Slc2a1)

In the brain, GLUT1 is primarily localized in endothelial cells, the first layer of the BBB (Fig. 1). GLUT1 contains 492 amino acids which is encoded by the SLC2A1 gene [28]. There are 2 distinct bands of GLUT1 protein in Western blots, corresponding to 55 kDa and 45 kDa. The 55-kDa form is associated with the micro-vessels while the 45-kDa isoform is expressed in neuronal cells. The extent of N-glycosylation at Asn45 in the first extracellular loop accounts for the difference in molecular weights of these two GLUT1 isoforms [24]. Importantly, the glycosylation of Asn45 is essential to maintain GLUT1 structure with high affinity for glucose uptake [29]. In brain endothelial cells, approximately 11–31% amount of GLUT1 protein resides in the cytoplasm, while 38–64% amount is found at the abluminal membrane [25]. GLUT1 has a high affinity to glucose with K_m ranging from 1 to 2 mM [20, 30]. The major substrate for GLUT1 is d-glucose, however, it also transports other sugars, such as mannose and galactose [31, 32]. GLUT1 activity is inhibited by cytochalasin B and phloretin [33, 34]. Protein kinase C (PKC) mediates phosphorylation of GLUT1 at Serine 226, which increases glucose uptake in neurons or BBB-associated endothelial cells [35]. In contrast, thioredoxin-interacting protein (TXNIP) negatively regulates glucose uptake by directly binding GLUT1 and triggers GLUT1 internalization. TXNIP also involves in downregulation of GLUT1 mRNA at a transcriptional level [36].

GLUT1 is essential for early development. Homozygous deletion of GLUT1 (GLUT1^−/−) in mice results in embryonic lethality [37–39]. These homozygous GLUT1^−/− embryos were formed between embryonic day 10–18 (E10-18). However, they had visible abnormal phenotypes including small body size, lack of eyes, and overall developmental abnormalities. None of GLUT1^−/− pups was born, and the embryos died in utero between E10 and E14 [37]. Additionally, injection of fertilized mouse eggs with antisense cDNA for GLUT1 (GT1AS) at E11- E18 resulted in GLUT1 deficiency. These GLUT1-deficient embryos exhibited smaller body size and died before birth [38]. In zebrafish, GLUT1 mRNA translation was inhibited by the introduction of antisense morpholino oligonucleotides at 2–4 days post fertilization [40]. This resulted in an 88% reduction of GLUT1 protein expression and impaired the development of the cerebral endothelial cells [40]. The brain of zebrafish morphants also showed a significant reduction of markers for blood vessels, such as vascular endothelial growth factor receptor 2 (VEGFR2), vascular endothelial (VE)-cadherin (VE-Cadherin), and cluster of differentiation 31 (CD31). This study indicates that GLUT1 is essential for BBB formation in zebrafish. These findings highlight the critical role of GLUT1 in early development.

In heterozygous GLUT1 knockout mice, the GLUT1 protein level in the brain was reduced by half. This reduction in GLUT1 expression was correlated to functional disorders of the animals, such as microvascular defects, decreased glucose uptake, motor activity impairment, and abnormally low glucose concentration within the cerebrospinal fluid. Other phenotypes including low birth weight with microcephaly were also observed [39, 41]. When electroencephalography (EEG) was performed on the adult heterozygous mutants, epileptiform discharges (ED) were recorded showing phenotypes related to brain circuitry [39]. In humans, patients with GLUT1 deficiency may exhibit ED without seizures [42]. EDs may disrupt short-term cognition while frequent EDs would impair long-term cognition [43]. EDs and seizures in non-acute settings and EDs in patients with neurological disorders could be a biomarker for abnormal brain function as a result of GLUT1 deficiency [43]. EDs in patients without epilepsy were reported to have no effect on processing of information [44]. In another study, mouse carrying a missense mutation of proline at the 324^th position of the GLUT1 was created under chemical treatment of N-ethyl-N-nitrosourea (ENU) [39]. Multiple pathological phenotypes were exhibited in these animals such as early embryonic death of homozygotes; the heterozygotes had decreased glucose levels in cerebral fluid, learning deficit, abnormal sleep–wake patterns, and spontaneous seizure-like behaviors.

GLUT1 not only plays important roles in embryonic neurogenesis, it is also crucial for postnatal CNS angiogenesis. In vitro, GLUT1 inhibition reduced the proliferation rate of the mouse brain-derived endothelial bEND.3 cell line by 40%, which affected angiogenic behaviors of these cells [45]. In mice, endothelial-specific GLUT1 knock-out (GLUT1EC^−/−) mice under tamoxifen treatment at postnatal (P) day 1 to 3 resulted in specific GLUT1 loss in the CNS EC [45]. This affected retinal and brain angiogenesis by reducing the number of proliferating endothelial cells which led to shortened vascular length. Four days after the first dose of tamoxifen, the mutant mice showed electrocorticographic abnormalities which were the electrical signatures of epileptic seizures. Six days after tamoxifen treatment, phenotypic alterations including reduced spontaneous movement and lack of explorative behaviors were observed. At day 8 and day 10, inflammation and neuronal loss were observed in several hippocampal areas with signs of progressive loss of CNS homeostasis. Mutant mice deceased 3 weeks after the first tamoxifen injection. However, GLUT1 conditional knockout in brain endothelial cells did not alter BBB tight junction structural and functional characteristics in comparison to the wild-type control [45]. This work challenged the view that disruption of BBB is the etiology in GLUT1 deficiency associated disorders.

Glucose transporter 3 (GLUT3, Slc2a3)

A majority of glucose in the bloodstream is transported across the BBB to the brain parenchyma. In neurons, GLUT3 mediates the uptake of glucose for energy production. Human GLUT3 is encoded by the SLC2A3 gene, which shares 64% sequence similarity with SLC2A1 [18, 46]. GLUT3 has 496 amino acids with a predicted molecular weight of approximately 54 kDa. In humans, GLUT3 is found in various organs, but is most highly expressed in the brain [26]. In monkeys, GLUT3 is most abundant in the frontal lobe. GLUT3 is also expressed in the parietal lobe, hippocampus, and cerebellum [47]. Neuronal cells are expressed with high-level expression of GLUT3, especially in the pre- and post-synaptic termini (Fig. 1) [48]. GLUT3 can transport glucose, 2-deoxyglucose, galactose, mannose, xylose, and fucose. It is also reported to be able to transport dehydroascorbic acid in vitro [18, 49]. GLUT3 has a K_m value of 1.8 mM for glucose and the highest turnover rate among the GLUT family [50]. GLUT3 can transport glucose at least 5 times faster than GLUT1 [26, 51]. In the structure of GLUTs, there are special regions of the first extracellular loop, N-terminal of intracellular loop 6, central residues of TM9, and TM11 that facilitate with the binding to glucose. In GLUT3, the intracellular loop 6, the region near to the exit site of the transporter pore contains small amino acids (glycine) and residues like asparagine and serine, which are capable of forming hydrogen bonds with glucose. In other GLUTs, these regions tend to have bulkier amino acids for example alanine, isoleucine, and phenylalanine. These properties in GLUT3 ensure a supply of glucose to neurons where the glucose concentration is considerably lower than in the bloodstream [52].

Homozygous deletion of GLUT3 in mice resulted in early developmental arrest at embryonic day 7–12, despite the presence of other GLUT isoforms [53, 54]. Heterozygous knockout of GLUT3 had an increased rate of apoptosis which was correlated to an increase rate of early abortion by 25% in these embryos. It is commonly observed across several studies that the mutant mice that survived with compromised GLUT3 function have smaller size due to growth restriction [53, 55]. However, once reaching adulthood, these mice developed with normal growth and normal body composition and index. This suggests that GLUT3 mutations cause a gene dose-dependent early lethality or late-gestation fetal growth restriction [53]. Postnatal deletion of GLUT3 was generated in neurons using NestinCreERT2 (SLC2A3f/fNestin-CreERT2) [55]. These knockout mice at P15 showed a 90% reduction of GLUT3 in the brain. Electrophysiological record of the mice indicates that loss of GLUT3 induces cortical hyper-excitability. The animals died beyond day 31. The shortened life span cannot be rescued by a ketogenic diet, implying that glucose uptake is critical for neuronal cells [55]. When knockout mice were restricted to the limbic system in adult male mice, the mutants survived. They had reduced locomotor skills and spatial memory and exhibited less fear consistent with a loss of function of the hippocampus and amygdala, but had an increase in spatial exploration and socialization [55]. It is also reported that heterozygous GLUT3^± mice survived and exhibited autism-like phenotypes including impaired learning abilities, increased cerebrocortical activities, and increased startle reflex [54]. However, GLUT3 haplo-insufficiency which had 50% GLUT3 protein level did not disrupt brain glucose uptake or utilization [56]. These results suggest that a reduction of 50% GLUT3 does not affect neuronal uptake of glucose [56].

GLUT3 functional studies were also conducted in zebrafish. GLUT3 KO fish was generated by injection of antisense morpholino (MO) targeting the GLUT3 translational start site. At the highest dose of 15 ng MO, all neuronal hallmarks were lost. The treated embryos were smaller in size and died within 48 h. With lower doses of MO, the morphants had impaired CNS development and growth restriction due to increased apoptotic events. The phenotypes of the morphants were rescued by the expression of zebrafish or rat GLUT3 mRNA. Interestingly, GLUT1 mRNA overexpression did not rescue the MO induced phenotypes. This confirmed that GLUT3 function is conserved among species and GLUT3 is necessary for embryonic CNS formation.

In humans, approximately 1 in 130 people has only 1 copy of GLUT3. This demonstrates that heterozygotic loss of function in GLUT3 is tolerable [46]. However, GLUT3 is unstable and prone to have copy number variation which causes alteration of the GLUT3 expression level. The event of GLUT3 copy number variation (CNV) was considered to be the potential pathomechanisms of several neurological diseases and immuno-disorders including attention‐deficit/hyperactivity disorder (ADHD), Huntington’s chorea and rheumatoid arthritis [46]. The above evidence indicates that GLUT3 plays important roles in neuronal activities; and lack of this gene could inhibit neuroplasticity, impair learning abilities, increase neuronal excitability and the risk of epilepsy, ultimately leading to behavioral changes and neuropsychiatric diseases in humans and animal models.

The roles of GLUT proteins in pathological conditions in the brain

GLUT1 deficiency syndrome (GLUT1-DS)

In humans, GLUT1-DS is caused by mutations in the SLC2A1 gene, so far this is the only gene that was identified to be associated with the disease. GLUT1-DS is a rare genetic metabolic disorder that accounts for around 1% of idiopathic generalized epilepsies, and approximately 10% of early-onset absence of epilepsies [57]. There are 22 published GLUT1 mutations, in which two types of mutations are nonsense mutations in exon-10 resulting in a stop codon and a truncation of C terminus. They both disrupt GLUT1 structure and abolish its function [58]. Missense variants are predominant in mild or moderate clinical categories, while patients with inserted mutations exhibit the most severe phenotype of GLUT1-DS [59]. Lange et al. and Graham et al. reported the mutations in exon 10 and exon 7 of the Glut1 gene, respectively. These patients exhibited seizure and developmental delay in late infancy. Heterozygous forms of mutated GLUT1 led to a partial deficiency of GLUT1 associated with intractable infantile epilepsy [59]. These babies were born with normal head size, but the children experienced mild chronic encephalopathy with infrequent seizures, spasticity and ataxia. The lighter forms of GLUT1-DS could lead to hypoglycaemia, a syndrome in which the patients experienced low blood sugar and expressed various symptoms including clumsiness, speech delay, frequent perplexity and loss of consciousness [60]. GLUT1-DS affects all ages, in some cases as early as 6 months from birth. This syndrome leads to neuronal disorders and improper development due to insufficient supply of glucose for neuronal activities. Once diagnosed, dietary therapies, such as ketogenic diets or high fat, are prescribed to patients. Dietary supplementation with tri-heptanoin, a source of short-chain fatty acid has been carried out in adults and infants for as long as 5 years with marked improvement of clinical symptoms [61]. The ketogenic diet effectively could control the seizures and other paroxysmal activities, but it has shown less effect on cognitive functions [59].

Alzheimer’s disease (AD)

Alzheimer’s disease (AD) is the number one cause of dementia in the elderly. Alzheimer’s disease patients are affected by a progressive loss of neuronal connections due to neuronal death. The disease starts in medial temporal structures such as the hippocampus, but spreads to other parts of the brain with time. This leads to memory deficits, cognitive functions, and movements, eventually, loss of the abilities to perform the simplest tasks [62]. Amyloid β-peptide (Aβ) plaques and neurofibrillary tangles (NFTs) are the hallmark features in Alzheimer's brain. Aβ is formed by the sequentially cleaved amyloid precursor protein (APP) by β‐ and γ‐secretase enzymes in the brain. APP is thought to have positive impacts on the CNS but its physiological functions are not yet known. The AD hypothesis suggests the imbalance of Aβ production and clearance leads to the accumulation of the toxic form Aβ in the brain is the primary cause of AD. This process together with NFT formation and downstream inflammation results in neuronal dysfunction and neurodegeneration [63]. Various studies suggested that pre-existing conditions in patients of peripheral insulin resistance and diabetes are factors that increase the risk of acquiring AD [64–66]. This raises the possibility that glucose transport malfunction could lead to brain glucose dysregulation, and contribute to AD development and progression.

Recent studies using mouse models reported that GLUT1 deficiency induced APP overexpression, cerebral microvascular degeneration, increased BBB permeability, and decreased Aβ clearance [41, 67, 68]. When GLUT1 was conditionally knocked out in endothelial cells but not in astrocytes, the mutant mouse had the phenotypes resembling AD’s symptoms, such as diminished neuronal activity, behavioral deficits, and progressive neuronal loss and neurodegeneration [67, 69]. LRP1 is an important transported in the CNS, and is involved in endocytosis of major ligands that are involved in AD e.g., apolipoprotein E, APP and especially Aβ [70].LRP1 is expressed mostly in the abluminal side of the BBB, it mediates Aβ transportation within the brain and clearance from the brain into circulation [71]. Hence, reduction of LRP1 expression greatly impacted Aβ clearance and lead to Aβ accumulation. Previous reports showed that LRP1 expression levels were found to be lesser in the endothelial cells of aging brain, AD patients and AD animal models [71]. This reduction of LRP1 was found to be incidentally correlated to AD development. And this was demonstrated in the transgenic AD mouse model overexpressing human APP Swedish mutant, APP^Sw/o [67]. This GLUT1-deficient mouse model phenotype exhibited accelerated amyloid load and aggravated Aβ accumulation. GLUT1^± heterozygote mice expressed less LRP1 than control mice [67]. It was discussed that GLUT1 deficiency led to acceleration of Aβ pathology via LRP1 reduction in the brain micro-vessels [68]. Notably, these changes were reversible by LRP1 rescue in the hippocampus. This evidence highlighted that GLUT1 could be a therapeutic target for AD treatment; and GLUT1 could be a major regulator in Aβ clearance in the aging brain [67]. Investigations of GLUT3’s role in AD were also conducted. Generally, in AD and “asymptomatic AD subjects” (ASYMAD) brains, GLUT3 protein levels were profoundly lower at the early disease stage than in the control groups, and associated with greater amyloid plaque and NFTs. However, GLUT3 protein levels were independent of neuronal loss [72]. The lower GLUT3 and GLUT1 protein expression in the AD brain led to decreased glucose uptake. Subsequently, the low AD brain glucose level was responsible for neurodegeneration via down-regulation of O-GlcNAcylation and hyper-phosphorylation of NFTs [73].

Brain cancers

GLUT3 is highly expressed in brain tumor-initiating cells (BTICs), and its inhibition in BTICS significantly decreased tumor formation [74]. Furthermore, it was reported that GLUT3 was found to be upregulated in resistant tumors with treatment of VEGF blocking antibody bevacizumab [75]. Thus, it was proposed that targeting GLUT3 to reduced glycolysis could be used in combination with anti-VEGF in resistant tumors [75]. GLUT3 was also found overexpressed in the brain of patients with metastatic breast cancers [76]. This upregulation of GLUT3 was not only important to increase glucose levels for ATP generation for the growing tumor cells, but also was critical to mediate a shift in glucose metabolism in these tissues. Knockdown of the binding protein cAMP-response element (CREB) has been found to alter GLUT3 expression in metastatic brain cancer cells [76]. The tumor-suppressor gene p53 directly regulates GLUT3 expression and its loss during oncogenic transformation resulted in GLUT3 upregulation via NF-κB signaling. In PC12 cells, GLUT3 expression could be upregulated through the activation of PI3K/Akt/mTOR signaling pathway [77]. Both extrinsic growth hormones (e.g., EGF and IGF‐1) and intrinsic factors that interfere with growth hormone receptors could stimulate the PI3K/Akt pathway [78]. Hence, drug targeting GLUT3 might potentially benefit GBM patients. However, there are no available efficacious GLUT3 inhibitors in the market. A current structure-based study had identified potential hits that can inhibit GLUT3 in vitro, although further development should be done before the compounds can enter human trials [79].

Monocarboxylate transporters (MCTs)

MCTs belong to a large group of transmembrane proteins encoded by the gene family of the solute carrier 16 (SLC16). The functions of these proteins are the transport of short-chain monocarboxylates, hormones, nutrients, and amino acids [80]. The amino acid sequence of human MCT proteins ranges from 426 to 613 amino acids. Most of MCT proteins (MCT1, MCT2, MCT 3, MCT7 and MCT8) have a 12-transmembrane structure (TMs) in which both N- and C-terminal termini are located in the cytoplasm [81]. The number of TMs in other MCT members varies from 10 to 12 domains [81]. Moreover, MCT family proteins exhibit homologous sequences in the TMs as well as the short loops between them [81]. In contrast, each MCT protein shows unique sequences in the hydrophilic regions, such as the N-termini, the C-termini, and the loop regions connecting TMs 6 and 7. These less conserved regions are not directly involved in the transport activities of MCT [81].

MCT family comprises 14 members. Depending on their locations and structures, MCTs are involved in the transport of a variety of substrates. Among the MCT isoforms, MCT1-4 have been shown to transport lactate, pyruvate, ketone bodies as well as other monocarboxylates in a proton-dependent manner [82]. Notably, MCT6 was reported to have a high affinity with bumetanide, which is a component of several pharmacological drugs, in a pH-dependent manner [83]. Moreover, a recent study using a knockout MCT6 mouse model suggests the involvement of MCT6 in glucose and lipid metabolism [84]. MCT7 plays an important role for beta-hydroxybutyrate efflux and its gene SLC16A6 is recognized as a new determinant of human height [85, 86]. MCT8 and MCT10 have been known as thyroid hormone transporters [87, 88], and MCT9 has been reported to act as a carnitine efflux transporter [89, 90]. MCT11 is capable of transporting pyruvate as well as exporting lactate from the liver, and is involved in hepatic metabolism as evidenced by the alterations of fatty acids and lipid concentrations following by SLC16A11 knock-down in hepatocytes [91]. MCT12 is implicated in the diffusion of creatine as well as its precursor guanidinoacetate [92–94]. The roles of orphan receptors MCT5, MCT13, MCT14 remain to be explored.

According to The Human Protein Atlas, although the expression of MCTs at mRNA level is present in almost all human tissues, not all MCT proteins are found ubiquitously in the different tissues. Among them, except for MCT3, MCT11, MCT12, and MCT13, the other 10 remain MCTs are widely found in the brain. MCT4 is abundantly found in astrocytes, which is employed to export lactate to neurons. During early stages of development, MCT4 expression is low in embryonic astrocytes, but its expression is increased in adult astrocytes [95]. MCT2 is expressed in neurons, whereas MCT1 is expressed in the BBB. Moreover, MCT8 is widely found in neurons, astrocytes, and BBB [87]. In the brain, MCTs play a pivotal role in the delivery of solute small molecules including monocarboxylates, hormones, nutrients, and amino acids which are essential for brain development. MCT dysfunctions are involved in pathologies in both the central and peripheral nervous systems which are further discussed in the following sections.

MCT1 is a crucial lactate transporter in the central and peripheral nervous systems

MCT1 encoded gene is SLC16A1 [81]. Human, rat and mouse MCT1 have been successfully cloned in which they share 95% sequence identity [81]. MCT1 harbors 12-transmembrane helices with a large cytosolic loop formed between transmembrane helices 6 and 7. As MCT1 protein is not glycosylated, it requires the attachment of CD147 and embigin which function as high glycosylated accessory proteins for folding activities [96]. MCT1 is involved in the transport of short-chain monocarboxylates, such as pyruvate (K_m = 1 mM), lactate (K_m = 3.5–10 mM), acetoacetate (K_m = 5.5 mM), and D-ß-hydroxybutyrate (K_m = 12.5 mM) [82]. We will focus on the essential role of MCT1 in lactate transport in the brain.

MCT1 is abundantly found in brain endothelial cells which plays an important role in the influx of lactate from blood stream into the brain [97]. It was reported that null mutation of MCT1 (MCT1^−/−) caused the embryonically lethal, while the MCT1^± mice showed a normal development [98]. In addition, MCT1^± mice treated with the high fat diet exhibited resistance obesity-induced diet, which suggested the essential roles of MCT1 in the energy homeostasis [98]. Moreover, human MCT1 mutations with subnormal lactate transport rate were linked with the severe muscle injuries [99]. Interestingly, PTEN and Wnt/ß-catenin signaling pathways are known to regulate the expression of MCT1 in brain endothelial cells. There was a significant accumulation of lactate accompanied with marked reduction of MCT1 expression in the PTEN^fl/fl knockout mice. In contrast, overexpression of MCT1 in the PTEN^fl/fl knockout mice could rescue lactate transport in the brain ECs. Hence, these results demonstrated that PTEN signaling is involved in lactate transport across the brain endothelial cells [100]. Moreover, Wnt/ß-catenin, through suppressing MCT1 protein from ubiquitination and degradation, enhances MCT1 expression in the plasma membrane of immortalized rat brain endothelial RBE4 cells [101]. In summary, these studies emphasized that MCT1, which is under the control of PTEN and Wnt/ß-catenin signaling pathways, contributes to lactate transport from blood to brain.

MCT1 is also highly expressed in glial cells [102]. Treatment of MCT1 inhibitor significantly reduced the number of matured oligodendrocytes, suggesting that MCT1 plays a crucial role in the maturation of oligodendrocytes [103]. Moreover, knock-down MCT1 in oligodendrocyte resulted in energy depletion and subsequently lead to the demyelination, degeneration of axons, neuron loss and pathological features similar to multiple sclerosis (MS) [104]. A recent study suggested that nitric oxide induced downregulation of MCT1 in the oligodendrocytes which resulted in mitochondrial dysfunction, axon damage, and features of MS [102].

MCT1 is also a lactate transporter in the peripheral nervous system [105]. Expression of MCT1 has been found in neuroplastic cells, Schwann cells and dorsal root ganglion neurons and deficiency in MCT1 expression impaired the function of these cells [105]. Notably, a recent publication revealed that Schwann cell-specific MCT1 knockout mouse exhibited impairment of glucose and lipid metabolism machinery, reduced myelin-associated glycoprotein, induced c-Jun and p75-neurotrophin receptor, and suppressed maturation of Schwann cells [106]. Interestingly, heterozygous deletion of MCT1 in mice exhibited delayed regeneration of peripheral nerves after sciatic nerve crush that might be due to defective Schwann cell function [107]. Furthermore, MCT1 expression in Schwann cells, but not MCT4, is necessary for maintenance of motor end-plate junctions [108]. These studies emphasize critical roles of MCT1 in homeostasis maintenance of both central and peripheral nervous systems. However, the mechanisms by which MCT1 regulates peripheral nerves remain to be determined.

MCT2 and MCT4 are associated with the astrocyte-neuron lactate shuttle

In comparison to MCT1, there are a few studies related to MCT2 and MCT4. MCT2 was first cloned from a cDNA library from hamster livers as it shares 60% identity with MCT1 [81]. MCT2 is encoded by the SLC16A7 gene [81]. MCT2 is highly found in neurons where it plays an essential role for lactate uptake into the neurons [109]. Furthermore, MCT2 expression was also found in the luminal membrane of endothelial cells, astrocytes end feet, cerebral cortex, hippocampus, and cerebellum [109]. MCT4 is encoded by SLC16A3 gene which is highly found in glycolytic tissues, such as muscles, white blood cells, chondrocytes, and astrocytes [96]. In the brain, MCT4 expression in astrocytes was increased during postnatal periods [95, 110].

According to the classical theory of neuro-energetics, blood glucose is considered as the major fuel for neurons and astrocytes [111]. However, blood glucose is not oxidized via oxidative phosphorylation to provide ATP in the endothelial cells [111]. Instead, a part of blood glucose is undergone glycolysis to generate lactate. Endothelial cells-derived lactate is then exported via MCT1 for further utilization by astrocytes and possibly by neurons. This phenomenon is referred to the “astrocyte-neuron lactate shuttle” [112]. It indicates that lactate plays an important role in the neuro-energetic homeostasis and it is released into brain parenchyma through MCT1 and MCT4 [111]. While MCT1 is responsible for lactate release from endothelial cells, MCT4 appears to export lactate from astrocytes to neurons [111]. The uptake of lactate to neurons is facilitated by MCT2 (Fig. 1). Lactate could be converted to pyruvate by lactate dehydrogenase B (LDHB) which is then used to generate ATP via oxidative phosphorylation [111]. Mice with low MCT2 and MCT4 expression exhibited a reduction in recognition of novel objects 24 h after training, which suggests a role of these transporters in the acquisition of lactate for long-term memory formation [113]. Consistent with this, there was a significant behavioral change in mice with partial knockout of neuronal MCT2 and astroglial MCT4 [113]. Interestingly, infusing lactate could only rescue long-term learning deficiency in the MCT4 knockout mice, not MCT2 [113]. This suggests the essential roles of lactate for both astrocytes and neurons, but they have distinct functions in learning and memory.

The roles of MCT1, MCT2 and MCT4 in Alzheimer’s disease and glioblastoma

Since MCT1, MCT2, and MCT4 hold important roles in the maintenance of neuro-energetic homeostasis, abnormalities in their expression and activities could contribute to brain disorders including Alzheimer’s disease and glioblastoma. AD patients often undergo significant difficulties in their life due to memory deterioration and cognitive impairment [114]. Deficiency in energy supply such as found in GLUT1 deficiency has been proposed as a major pathological factor leading to AD [115]. Interestingly, lactate has been recently regarded as one of the major energy sources for brain metabolism, and alteration in lactate levels is correlated with AD pathogenesis. A study conducted in 16 mild AD patient samples revealed that there was an elevated expression of MCT4 in the cerebrospinal fluid [116]. To examine whether there is any correlation between the MCT4 expression and AD pathogenesis, Morris water maze (MWM) and Western blot were performed in an early AD mouse model APPswe/PS1dE9 (APP/PS1) [116]. The transgenic mouse model APP/PS1 has been reported to harbor the human-mutated APP with Swedish mutation (APP695Swe) and mutant human presenilin 1 with exon 9 deletion (PSEN1dE9) genes and is used as a model for AD research [117]. It was shown that 2–3 month-old APP/PS1 mice exhibited overexpression of MCT4 and a cognitive decline, in comparison to the control group [116]. These results were similar to clinical symptoms that had been observed in patients with mild AD. Notably, knockdown of MCT4 in the APP/PS1 mouse model resulted in a reduction of neuron’s apoptotic rate which correlated with cognitive improvement [116]. However, the fundamental mechanisms of MCT4 in AD pathogenesis remain unclear. Future research may focus on the proposed mechanisms of MCT4 in AD progression.

Glioblastoma (GBM) is considered as one of the most aggressive brain-related diseases with a high mortality rate. Although the current knowledge about GBM pathogenesis is limited, hyper activities of MCT1 and MCT4 have been identified to play important roles in GBM [118]. Increased expression of MCT1 and MCT4 is correlated with a worse prognosis in GBM patients [118]. GBM cells are highly glycolytic with the elevated production of lactate which is then transported into the tumor microenvironment through MCT1 and MCT4. Up-regulation of MCT1 and MCT4 was reported to induce glycolysis in GBM under hypoxic conditions as evidenced by increased glucose consumption and lactate production in cell models [119, 120]. Moreover, GBM tumors are reported to exert high migratory behaviors as well as invasiveness capabilities [118, 119]. Mechanistically, it appears that MCT1 mediates the interaction between GBM cells and endothelial cells thus promoting proliferation, migration, and angiogenesis through AMPK and Akt signaling pathways [121]. Notably, a subset of GBM cells named GBM cancer stem cells (GSCs) is responsible for self-renewal, proliferation, tumorigenesis, and recurrence of GBM [122]. These GSC cells are enriched with MCT1 and MCT4 levels and inhibition of these two transporters could deplete stemness characteristics of GSCs [122, 123]. Together, targeting MCT1 and MCT4 may constitute a promising therapeutic therapy for GBM patients.

Functions of MCT8 and MCT10 in thyroid hormone uptake

As a pleiotropic hormone, the thyroid hormone has a strong impact on a range of biological activities, especially in brain development. Changes in thyroid hormone levels have been reported to affect brain development, thereby causing severe neurological dysfunctions. For example, while hypothyroidism facilitates depression, memory impairment, hyporeflexia, and lethargy, hyperthyroidism is related to anxiety, irritability, and hyperreflexia [124, 125]. In the brain, thyroid hormones can be presented as an active form 3,5,3′-triiodothyronine (T₃) and thyroxine (T₄) which is the precursor of T₃. Notably, 80% of the brain T3 fraction is produced locally from T4 through type 2 deiodinase (D2) which is expressed in glial cells [126]. Biological activities of thyroid hormones are initiated when active thyroid hormone T3 is transferred to nearby neurons, which are then translocated into the nucleus, interact with nuclear thyroid hormone receptors (TRs), and regulate a wide range of brain development activities, including neural differentiation and migration [127].

Transport of thyroid hormones across the plasma membrane in the brain requires specific transporters, such as the organic anion transporter polypeptide 1c1 (Oatp1c1), MCT8, and MCT10 [128]. Oatp1c1, which is commonly found in astrocytes, is responsible for T4 entry. MCT8 facilitates T3 and T4 transport across the BBB and T3 uptake in neurons [128]. MCT10 is known to be an aromatic amino acid transporter [129], but it was reported to actively transport iodothyronine [88]. Mutations in MCT8 and MCT10 cause a reduction of T3 and T4 transport in the brain, leading to a severely retarded brain maturation and neurological illness [127].

Due to its high expression in the neurons, astrocytes, and BBB, MCT8 appears to be the most important transporter for the transport of thyroid hormones into the brain [87]. Mutation in MCT8 gene (SLC16A2 gene) has been reported to cause Allan–Herndon–Dudley Syndrome (AHDS) [130]. AHDS is a rare X-linked disorder in humans as characterized by the excess levels of serum T3 but low T4. Moreover, most AHDS patients are associated with hypotonia, bradykinesia, motor disability, and even severe psychomotor retardation [131]. These alterations can be explained by the absence of MCT8 in the BBB resulting in a deficiency of T4 level and finally leading to brain hypothyroidism [132]. Pre-clinical experiments in mice were conducted to clarify the underlying mechanisms in AHDS pathogenesis. Increased levels of serum T3 and reduced T4 levels were observed in MCT8 knockout mice and AHDS patients [133, 134]. However, there was no neurological deficiency or behavioral abnormalities in the MCT8 knockout mice [133, 134]. It might be that expression of Oatp1c1 allows the transport of T4 hormone that compensates for the loss of MCT8 in mice [135]. Consistently, Oatp1c1 is responsible for T4 transport through the BBB in addition to MCT8 [136]. Additionally, double knockout of MCT8 and Oatp1c1 led to several alterations in neurological phenotypes with concomitant reduction of both T3 and T4 hormones [135]. In summary, as MCT8 is responsible for most of thyroid hormone transport in the brain, dysfunction of MCT8 leads to AHDS pathogenesis. Manipulation of MCT8 function could therefore be a possible therapy for AHDS patients.

The roles of MCT10 as a thyroid transporter, especially in the brain is less studied. While knock-down of MCT10 in chondrocytes caused a reduction of T3 levels, overexpression of MCT10 in JEG3 cells could induce T3 accumulation [137]. Consistently, transient transfection of COS1 cells with hMCT10 resulted in an enhanced T3 uptake over T4 uptake [88]. It seems that MCT10 showed better T3 transport activity across the phospholipid membrane than MCT8 [88]. The physiological roles of MCT10 in T3 transport in brain development and pathogenesis are yet determined. Hence, further studies established in knock-out mice are necessary for further elucidation of the roles of thyroid hormone transporters in the brain.

Amino acid transporters in the brain

Neuronal cells acquire amino acids, especially essential amino acids from diets for protein synthesis. In addition, these cells take up certain amino acids, such as glutamate, synthesis of neurotransmitters. Furthermore, several amino acids, such as tryptophan, tyrosine, histidine, and arginine, are needed for synthesis of neuromodulators and cofactors. Deficiency of these amino acids leads to emotional stress and impaired cognitive performance [138]. The levels of amino acids in the brain are kept at lower concentration than in blood, except for the levels of glutamine, which are balanced between the brain parenchyma and blood [139, 140]. Therefore, the brain requires transporters to regulate the uptake and release of certain amino acids. These transporters for amino acids are classified into facilitative transporters and active transporters. Below, we summarize the characterization of each amino acid transport system.

Facilitative amino acid transporters

Facilitative transporters are sodium-independent transporters [141]. There are at least three facilitative transport systems found in the BBB including large neutral amino acids (LAT), cationic amino acids (system CAT), and Cys/Glu exchange (system x_c⁻) [139]. According to Hawkins’s group discovery, the system xc- is found exclusively in the abluminal side, while the transporters of the system LAT and CAT are located on both sides of endothelium [139].

Large neutral amino acid transporters (LAT)

Neutral amino acids, such as phenylalanine, tryptophan, leucine, isoleucine, methionine, histidine, tyrosine, valine, and threonine, are transported from blood to the brain through facilitative L amino acid transporters (LAT) system. These transporters are SLC7A5-13 and SLC7A15. One of the well-characterized LATs is LAT1, which is encoded by SLC7A5 gene. LAT1 is located in both the abluminal and the luminal side of the BBB [142, 143]. In the LAT1 system, SLC7A5 binds to glycoprotein SLC3A2 (CD98 or 4F2hc) via a disulfide bridge to form a complex [144]. The glycoprotein CD98 in the LAT1 system facilitates the localization of SLC7A5 in the plasma membrane [145]. LAT transporters are pH sensitive and sodium-independent [146]. LAT2 exerts its highest activity at pH 6.25 [147]. These LAT transporters have high affinity with Phe, Trp, Leu, and His (K_m = 5–50 μM) [146, 148, 149], but are less efficient in transport of Gln (K_m in the mM range) and do not recognize Ala, Pro, and charged amino acids [148–150].

The mechanistic transport of large neutral amino acids by LAT1 remains unclear since the crystal structure of SLC7A5 is not available. However, employing the bacterial homolog of SLC7A5, the substrate recognition sites of mammalian SLC7A5 was predicted [151, 152]. It was suggested that Phe252 in SLC7A5 is a key residue for gate opening. Ser342 and Cys335 are also important for binding of amino acids on the extracellular side of the protein, while Cys407 is associated with substrate on the cytosolic side [152].

The LAT system is considered as the major pathway for uptake of essential amino acids in the brain [153, 154]. Mutations in SLC7A5 resulted in reduced uptake levels of branched-chain amino acids that are linked to autism spectrum disorders (ASD) and motor delay in humans [155]. A low expression level of SLC7A5 was also associated with a significantly decreased level of Leu and Ile in the brain [155]. Furthermore, deletion of SLC7A5 in animal embryos causes the lethality at day E11.5 [156]. Low expression levels of SLC7A5 are associated with the development of Parkinson’s disease [157]. Therefore, these findings highlight an important role of SLC7A5 and the roles of essential amino acids for brain development and functions.

Cationic amino acid transporters

Cationic amino acids (CAT), such as Arg and Lys, and orthinine, are transported to the brain via the CAT system in a sodium-independent manner [158]. These transporters belong to SLC7 family which includes SLC7A1-4 and SLC7A14. Among the CAT system, CAT1 (SLC7a1) is expressed almost ubiquitously in adult tissues except the liver and lacrimal gland [159, 160]. The mRNAs of the CAT1 system are enriched in micro-vessels of the brain [158]. CAT1 forms a complex with CD98 for localization onto the plasma membrane. Interestingly, CAT1 shares a high similarity to the transporters in LAT system. The high structural conservation of CAT1 and LAT1 suggests that they may have arisen from the same origin. The Glu107 residue in transmembrane domain 3 is critical for transport activity of CAT1 [161]. Glu107 is conserved in all CAT isoforms, which indicates that this residue is important for substrate binding.

The affinity of the CAT system with cationic amino acids ranges from 50 to 100 μM [158]. The K_m of CAT1 in the BBB is about 470 $\pm$ 106 μM for Lys [162]. In addition, CAT1 is important for the transport of Arg to provide the precursor for synthesis of nitric oxide in the brain [158]. CAT1 transporter is regarded as the major transport pathway of cationic amino acids in the brain. Homozygous deletion of SLC7A1 resulted in the mouse lethality soon after birth [163]. Furthermore, knockout pups had severe anemia and body size was about one-fourth of the wildtype littermate. These results indicate that amino acids transported by CAT1 are essential for development and survival.

The Cys/Glu exchange

The system is also called x_c⁻, which is responsible for the exchange of L-cysteine (Cys) and L-glutamate (Glu). Cysteine is required for glutathione synthesis, which is important for antioxidant and redox reactions. On the other hand, a high concentration of Glu is toxic to neurons. The system x_c⁻ is known to export Glu and import cysteine [164, 165]. However, this system can be forced to operate in reverse, when there is an excessive level of extracellular glutamate, such as after neuronal injury. In this scenario, the influx of glutamate causes the release of cysteine, leading to the reduced intracellular levels of cysteine for glutathione synthesis. The system x_c⁻ is found ubiquitously in brain, macrophages and cell culture lines [166]. RNA-seq in tissue-specific expression also indicated highly expression of this amino acid transport system in the brain [167]. Particularly, the system x_c⁻ is highly expressed throughout many regions of the brain [168]. The x_c⁻ system consists of two subunits: xCT (as known as SLC7A11) and CD98. This system transports Cys and Glu in one-to-one ratio [169, 170]. Loss of function studies showed that the x_c⁻ system is required for cell growth and survival by regulation of redox state [171]. In addition, mice lacking system x_c⁻ by deleting SLC7A11 gene exhibited reduced working memory, but were protected from epilepsy induced by proconvulsive drugs [172].

Active amino acid transporters

The active amino acid transporters transport amino acids against their gradient concentration in a sodium-dependent manner. Hawkins and his group revealed that these sodium-dependent transporters are located in abluminal side, while other studies showed that SNAT3 as nitrogen-rich amino acid transporters located on both sides of endothelial cells [139, 173]. These transporters are grouped into branched-chain amino acids, small neutral amino acids, nitrogen-rich amino acids, system N for Gln transport, and Glu and Asp transporters.

Branched-chain amino acid transporters (SBAT1)

Branched-chain amino acids as Leu, Val, and Ile are transported by the sodium-coupled branched-chain amino acid transporters 1 (SBAT1) [174]. SBAT1 which is encoded by SLC6A15 is uniquely expressed in the brain [174]. SLC6A15 consists of 730 amino acids with 12 putative transmembrane domains. It is predicted to have two or more glycosylation sites [174]. Transport activity of SLC6A15 is dependent on Na⁺ and Cl⁻. In the absence of Cl⁻, uptake of leucine by SLC6A15 was reduced by about 36%. SLC6A15 could also transport a broad range of hydrophobic zwitterionic amino acids [174]. However, it has a high affinity for Leu (K_m = 80–160 µM) [174]. In addition, its transport activity is dependent on pH. The leucine uptake was optimal at pH around 7.5–8.5, which was reduced by 22% at pH 5.5 suggesting that the allosteric effect of H⁺ on Na⁺ or amino acid binding based on the co-transport of H⁺ [174]. As one of the SLC6 family, SLC6A15 is important for neurotransmission [175]. Leucine has been known importantly as precursors for glutamate, glutamine and neurotransmitter as GABA [176]. SLC6A15 knockout mice reduced the uptake of leucine into brain in sodium-dependent manner, compared to wild-type mice [177]. Mice with deletion of SLC6A15 had reduction of leucine uptake in hypothalamus and decrease in neuron activation in ventromedial hypothalamic nucleus [178].

Small neutral amino acid transporter (Also called system A)

Small non-essential neutral amino acids as Ala, Ser, Leu and Cys are actively transported through the sodium-dependent system. The system is located exclusively in the abluminal membrane, found in bovine BBB [141]. SLC38A2 (Snat2) was described as a transporter in this system. The expression of SLC38A2 is highly found under hypotonic stress in blood vessels in rat [174, 179]. SLC38A2 is expressed at low regional specificity [180]. Transport of Gln through SLC38A2 constitutes about 20% of the total Gln transport activity [181]. Interestingly, SLC38A2 acts as an amino acid sensor upstream of mammalian target of rapamycin (mTOR) in hepatic cells [182]. Deletion of SLC38A2 gene might lead to impaired neuron development. However, the physiological role of this system is less characterized.

Nitrogen-rich amino acid transporters (Also called system N)

Nitrogen-rich amino acids, such as His, Gln, and Asn, are actively transported through Snat1 (SLC38A1) and Snat3 (SLC38A3) [139]. As mentioned above, Gln is important for protein synthesis and nitric oxide pathway as well as being a precursor for neurotransmitters. Astrocytes take up glutamate in the synaptic cleft and convert it to glutamine via the enzyme glutamine synthetase. The glutamine is then delivered to the axon terminal where it is converted to glutamate [139]. SLC38A1 and SLC38A3 were identified be located endothelial membrane using mouse model [183]. SLC38A1 expression is specifically in large micro-vessels in the cortex, while SLC38A3 expression is also found in the capillaries [183]. The activities of SLC38A1 and SLC3A3 are sodium-dependent and voltage-independent. The K_m of these transporters with Gln were observed at 1.3 ± 0.4 mM [184]. SLC38A1 plays an important upstream role of mTORC1 [185]. In mouse model of stroke, neuronal cell death is significantly lower in mice deficiency for SLC38A1, as shown by immunohistochemical staining by 2,3,5-triphenyltetrazolium chloride or MAP2 to measure the volume of infarction [185]. Neurons are more protected under ischemia by inhibition of mTORC1 activity that leads to induction of autophagy [186, 187]. Loss of function of SLC38A1 decreases phosphorylation of mTOR and p70S6K [185] via reduction of Glu-dependent mTOR activation [188]. Furthermore, the level of phosphorylated p62 which is a marker for autophagy is higher in the SLC38A1-deficient neurons than wild-type neurons in ischemic stroke. These results imply that SLC38A1-deficient mice are less sensitive to ischemia explained by reduced mTORC1 activation and increased autophagy mechanism. This finding suggests that SLC38A1 may be a target for protection of mTOR induced neuronal death during ischemic stroke.

Glu and Asp transporters

Accumulation of too much Glu in the brain is detrimental to neurons. The function of the excitatory amino acid transporters (EAAT1, EAAT2, and EAAT3) is to reduce the concentrations of Glu in the extracellular fluid and between the synaptic clefts. These EAAT transporters are encoded by SLC1A1-3, respectively. In the brain, glial glutamate transporters function to reduce glutamate levels in the synaptic clefts after neurotransmission, thereby preventing excessive neuronal excitation and excitotoxicity. EAAT1-3 were identified specifically in the abluminal side of the endothelial cells [189, 190] as well as astrocytes and neurons [191]. The transport activity of Glu transporters is dependent on 3Na⁺ and 1H⁺ with K_m for Glu is 14 ± 4 μM [192]. The affinity of Glu for EAAT1, EAAT2, and EAAT3 are 1:3:6, respectively [189]. EAAT1-3 play a key role in maintaining glutamine concentration in the brain. Dysfunctional regulation of these transporters could lead to imbalance of neurotransmitter levels in the brain, leading to psychiatric disorder [193, 194]. Many attempts have been made to develop drugs to cure the psychiatric illness caused by dysfunction of glutamate transporters [195]. Understanding the mechanism and structure the glutamate transporters would help to develop pharmaceutical strategies for drug development for brain diseases.

Lipid transporters

Nearly half of the brain dry weight contains lipid [196]. The lipid profile of the brain is unique with high levels of polyunsaturated fatty acids (FAs). Neuronal cells take up these FAs, such as arachidonic acid, eicosapentaenoic acid, and docosahexaenoic acid (DHA), to build cell membranes. In addition, DHA could be metabolized to resolving D1 and neuroprotectin D1 which exert signaling and neuroprotective properties [197]. Derivative forms of arachidonic acid are endocannabinoid, which is involved in an important signaling pathway as release of neurotransmitters [198]. DHA is considered an essential fatty acid that cannot be synthesized by the body, and must be taken in the diet. This is in contrast to nonessential FAs that can be synthesized de novo [199]. It is believed that lipids are delivered to the brain via diffusion and flip-flop mechanisms. The fatty acid transport proteins, including fatty acid transport proteins (FATP), fatty acid translocase (FAT/CD36), fatty acid binding proteins (FABPs) are reported to involve in these mechanisms. However, the physiological evidence to prove the transport of those membrane proteins in the BBB was lacking.

In 2014, Nguyen and his colleagues demonstrated that major facilitator superfamily domain containing 2A (Mfsd2a) is a major transporter of lipids in the blood vessels in the CNS [200]. Mfsd2a is specifically and highly located in the brain endothelial cells of micro-vessels. The Mfsd2a expression level in the lung endothelium was 78.8 times lower than in the brain endothelium [201]. Interestingly, Mfsd2a specifically transports esterified DHA in the complex with lysophosphatidylcholines (LPC), but not un-esterified DHA [200]. Besides LPC-DHA, other LPC containing non-essential and essential FAs with a minimal 14-carbon acyl chain, such as oleate, palmitate, linoleate, linolenate, also are transported by Mfsd2a. Mfsd2a transport activity is dependent on an influx of Na⁺, suggesting that its mechanism transport of LPC is unidirectional from blood to the blood vessels [200]. Interestingly, Mfsd2a also transports acylcarnitines [202]. However, it is unclear if this function of Mfsd2a is physiologically relevant in the brain. Mfsd2a belongs to the major facilitator superfamily (MFS) of proteins with 12-transmembrane domains [202]. The protein is a 530-amino-acid length with two glycosylation sites. Mfsd2a protein shows a high conservation in sequence from fish to humans. The model of Mfsd2a protein structure suggested that there are three unique regions of Mfsd2a, including LPC head group binding sites, LPC tail hydrophobic cleft, and the ionic locks [202]. The binding sites of sodium ion have also been identified to be Asp-92 and Asp-97 [200]. Nevertheless, crystalized structures of the Mfsd2a protein would be required to provide more information of its transport mechanism. Atomic structure of gallus Mfsd2a was solved recently [203]. In this structure, Mfsd2a in inward-facing conformation with linolenate was captured. The structure suggested that Na + binding initiates the conformational changes of a large amphipathic cavity of Mfsd2a which allows LPC head group to pass through the transporter. The hydrophobic fatty acid chain of LPC is moved laterally in a hydrophobic cleft between transmembrane domain 5 and 8 or domain 2 and 11 in a “flip-flop” mechanism. This study provides structural basis to explain how FAs such as omega-3 are transported across the blood–brain barrier, which could be potential therapeutic applications such as drug delivery.

Besides being a transporter for lysolipids, Mfsd2a may also play a role in regulating permeability of the BBB [201]. According to this model, Mfsd2a inhibits transcytosis in endothelial cells to maintain BBB integrity [204]. In contrast, Mfsd2a-knockout mice and Asp96Ala knock-in mice exhibit increased transcytosis in brain endothelial cells [155]. However, lack of Mfsd2a did not result in obvious leakage as defined by increased transcytosis of small molecules, such as Alexa Fluor cadaverine or Sulfo-NHS biotin [205, 206]. The increased leakiness that affects BBB functions in Mfsd2a-knockout mice were not observed in other studies [206]. Therefore, it remains rather elusive for the physiological roles of Mfsd2a in regulation of CNS blood vessel integrity.

Maintaining DHA level by Mfsd2a activity in the brain is important for brain development [194, 204]. Dysfunction of Mfsd2a reduces the DHA level in the brain [200]. Mfsd2a-knockout mice exhibit a loss of neuronal cells in hippocampus and cerebellum, resulting in microcephaly and severe anxiety. These findings suggest that lipids transported by Mfsd2a are physiologically important for neuronal cells [194, 200]. Human mutations of serine 339 to leucine in Mfsd2a significantly reduced Mfsd2a transport activity that led to lower uptake of LPC [207]. Homozygous mutation of Ser339Leu displays severe microcephaly symptoms and motor dysfunction, which might be due to deficiency of DHA in the brain [207]. Other inactive mutations in Mfsd2a, such as Thr159Met, Ser166Leu, Pro402His, exhibited a lethal microcephaly syndrome [208, 209], LPC-DHA is critical for brain development and function [210, 211]. The discovery of Mfsd2a explains the mechanisms by which fatty acids are imported into the brain. However, this pathway only contributes approximately 50% essential fatty acids to the brain. Future studies are needed to reveal other pathways for fatty acid import into the brain. Revealing these mechanisms is not only fundamentally important to understand how the brain takes up fatty acids, but also be exploited for development of strategies for drug delivery in the brain.

Transporters for cofactors and other molecules at the BBB

Vitamins have been widely known as essential co-factors for numerous biological reactions. Some of the vitamins, such as vitamin A and E, are critical for the brain functions [212, 213]. For example, vitamin A has an important function in rods and cones of the retina. Vitamins C and E are antioxidants that have important roles in limiting free radical injury. Thus, their concentrations in the brain are maintained at high levels.

Vitamin A plays an important role in regulation of cell growth and differentiation in different organs throughout the human body during embryogenesis to adulthood. An imbalance in vitamin A homeostasis is associated with multiple pathological conditions, such as infectious diseases, skin diseases, visual diseases, and neurological disorders [214]. Plasma retinol-binding protein (RBP) binds to and facilitates vitamin A transport in the blood stream [214]. STRA6, a 75-kDa multi-transmembrane domain protein, and RBP receptor mediate the uptake of vitamin A from RBP into the target cells [214]. It is interesting to note that STRA6 is responsible for 95% of vitamin A uptake [214]. Reduced levels of STRA6 are associated with various human disorders e.g., in the eyes, lung, heart and brain [214]. Complicated structure of STRA6 with a dimer topology which is constituted from 18 transmembrane helices and two intramembrane helices has been solved [215]. Moreover, monomer STRA6 can interact with a molecule of calmodulin (CaM) at the cytoplasmic side which adopts the unconventional conformation of STRA6 [215]. Lack of STRA6 in animal models caused a reduction in vitamin A uptake [214]. Notably, STRA6 is widely found in epithelial barriers including the retinal pigment epithelium (RPE), choroid plexus (CP), and the BBB [216]. The presence of STRA6 in the RPE is critical as it mediates the delivery of vitamin A to rods and cones, which are located next to the RPE in the retina [216]. Brain isolated from the homozygous Stra6^−/− mice exhibited reduced levels of vitamin A in comparison with heterozygous male Stra6^± indicating the crucial functions of STRA6 in vitamin A transport to the brain [216].

Vitamin C has been widely known as a critical reactive oxygen species (ROS) scavenger, which is responsible for neuronal maturation in the brain [217]. Moreover, vitamin C is proposed as an essential cofactor for enzymes related to a variety of reactions of carnitine, amino acids, catecholamines, cholesterol, and some peptide hormones biosynthesis [217]. Vitamin C concentration in the brain is tenfold higher than in the serum [218]. In mammals, vitamin C exists in two forms, the oxidized form (dehydroascorbic acid) and the reduced form (ascorbic acid) [217]. Dehydroascorbic acid is rapidly converted into ascorbate once it is transported into the cells through GLUT families, such as GLUT1 and GLUT3 [217]. In contrast, ascorbic acid is actively transported into the brain through the sodium-dependent vitamin C transporters including SVCT1 and SVCT2 [217]. While SVCT1 is involved in ascorbic acid transport in the body, SVCT2 is especially responsible for ascorbic acid uptake in the brain [217, 219]. The activities of SVCT2 are Na⁺-dependent and also ATP-dependent [217]. Mice lacking SVCT2 cannot survive before birth, and have severe impairment of lung and brain function [220]. It can be explained that SVCT2 is obligatory for vitamin C maintenance in fetal tissues, and deficiency of SVCT2 resulted in low vitamin C levels, which subsequently leads to oxidative stress and fetal death [221]. Moreover, a recent publication reported that redox imbalance induced by the reduced expression of SVCT2 in the plasma membrane could be linked to Huntington’s disease [222]. These studies reveal the critical role of the SVCT2 in the vitamin C uptake in the brain.

The B vitamins (B1, B2) are important not only for proper functions of the peripheral nervous system, but also the CNS. Vitamin B1 (Thiamine) is a sulfur-containing vitamin serving as a cofactor for a variety of enzymatic processes that involve in brain development [223]. At a physiological pH, thiamine exists in cationic form and acts as the nerve signals through synapses [224]. Cationic thiamine can be actively transported across BBB by specific transporters [225]. One of the transporters is the extra-neuronal monoamine transporter protein (EMT, SLC22A3) which is highly expressed in the brain and nervous tissues [224]. EMTs belong to a family of transmembrane-bound channel proteins that play a key role in the bi-directional transport of organic cations through the cytosol [224].

Vitamin B2 (riboflavin) is an essential water-soluble vitamin in the brain that is required for a diversity of metabolic reactions [226]. Riboflavin acts as a precursor for flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) [226, 227]. Since mammalian neural cells cannot produce riboflavin, it is imported through a specific transporter [227]. There are three known riboflavin transporters that belong to the solute carrier family 52 (SLC52) [227]. RFVT1-3 is encoded by SLC52A1-3, respectively [227]. While RFVT1 and RFVT2 are highly expressed in the small intestine, RFVT3 is widely found in the brain [227, 228]. However, the role of RFVT3 for importing riboflavin into the brain remains largely uncharacterized.

Carnitine (4-trimethylammonio-3-hydroxybutyrate) is a critical co-factor for the ß-oxidation of long-chain FAs [229]. Carnitine, which is not synthesized in the brain, can cross BBB through two families of transporters: the transporter B^0,+ and the organic cation transporters (OCTN) [229]. The B^0,+ transporter (encoded by the family of solute carrier SLC6A19 facilitates the uptake of neutral and positively charged amino acids and carnitine [230, 231]. OCTN transporters, which belong to solute carrier 22 (SLC22), comprise three members, including OCTN1, OCTN2, and OCTN3 [229]. Among them, OCTN2 and OCTN3 were reported to have a high affinity to carnitine [229]. OCTN2 facilitates carnitine transport in a Na⁺-dependent manner. Transport of carnitine by OCTN3 is not dependent on [Na +] gradient [229]. A model of the in vitro BBB performed in human brain endothelial cells (hCMEC/D3) revealed that there was a reduction of carnitine uptake levels in hCMEC/D3 cells transfected with siOCTN2, implying the role of OCTN2 in carnitine delivery at the human BBB [232]. In addition, up-regulation of OCTN2 caused accumulation of carnitine and increased ß-oxidation of long-chain FAs in the glioma cells [233]. These observations suggest that silencing of OCTN2 expression could be employed to induce glioma apoptosis, making OCTN2 a promising target for treatment of glioblastoma [233]. Hence, both carnitine transporters B^0,+ and OCTN are essential for the uptake of carnitine into the brain.

Besides carnitine, OCTN1 has also been found to be important in transport of a phytochemical, ergothioneine across the blood–brain barrier [234]. Inhibition of OCTN1 by VHCL reduced the protective effect of ergothioneine on 7-ketocholesterol induced damage of brain endothelial cells, suggesting an important role of this transporter in the anti-inflammatory and protective effect of ergothioneine [235].

Conclusion and future perspectives

The human brain consumes a disproportionately large amount of energy for its size and weight. It uses approximately 20% of available energy in the body, mostly from glucose. Indeed, deficiency for glucose uptake in Glut1-deficient patients or hypoglycemia results in insufficient energy required for neuronal activity. The brain also ketone bodies and to lesser extent, lactose for energy production mainly in non-neuronal cells. This occurs when blood glucose is reduced during fasting. Leveraging the utilization of ketone bodies is reported to be beneficial in conditions with reduced glucose uptake such as Glut1-DS syndromes. The brain also takes up amino acids and other molecules for neuronal development and functions. Interestingly, the brain is built up with a large part of lipids. Recent evidences show that lipids, such as free fatty acids and LPC, are taken up by the brain. Lack of LPC uptake results in severe brain development and functions. Many other transporters with unknown molecular functions are expressed in the BBB. For example, Mfsd7c (also called Flvcr2) is specifically expressed in CNS blood vessels [236]. Loss of Mfsd7c functions results in mal-development of the brain. Unraveling the molecular functions of Mfsd7c would provide new knowledge about the requirement of nutrients for brain development and functions. Many solute transporters are found in the BBB and neurons and their molecular functions are yet characterized. To determine what types of nutrients and molecules that are required for physiological functions of the brain, it is important to study the molecular functions of these transporters. Therefore, we propose that future research on characterizations of molecular functions of these BBB transporters and neuronal transporters will be key to understand brain health. These fundamental understandings could also provide therapeutic options for treatment of neurological disorders.

Abbreviations

ABC transporter

ATP-binding cassette transporter

AD

Alzheimer’s disease

ADHD

Attention-deficit/hyperactivity disorder

AHDS

Allan–Herndon–Dudley syndrome

Akt

Serine/threonine protein kinase

APP

Amyloid precursor protein

APP/PS1

APPswe/PS1dE9

Aß peptide

ß-amyloid peptide

ASD

Autism spectrum disorders

ASYMAD

Asymptomatic AD

ATP

Adenosine triphosphate

Aβ

Amyloid β-peptide

BBB

Blood–brain barrier

CD31

Cluster of differentiation 31

CD36

Cluster of differentiation 36

CD4

Cluster of differentiation 4

CNS

Central nervous system

CNV

Copy number variation

CREB

CAMP-response element

D2

Type 2 deiodinase

DHA

Docosahexaenoic acid

EAAT

Excitatory amino acid transporters

ECs

Endothelial cells

ED

Epileptiform discharges

EEG

Electroencephalography

EGF

Epidermal growth factor

EMT

Extra-neuronal monoamine transporter

ENU

N-Ethyl-N-nitrosourea

FAD

Flavin adenine dinucleotide

FAs

Fatty acids

FAT

Fatty acid translocase

FATP

Fatty acid transport proteins

FMN

Flavin mononucleotide

GAA

Guanidinoacetate

GBM

Glioblastoma

GLUT

Glucose transporter

GLUT1-DS

GLUT1 deficiency

GLUT1-DS

GLUT1 deficiency syndrome

GSCs

GBM cancer stem cells

GT1AS

Antisense-GLUT1 cDNA

hCMEC/D3

Human brain endothelial cells

His

Histidine

HMGA1

The high mobility Group A1

HRE

Hypoxia-responsive element

HTLV

Human T-lymphotropic virus

IGF-1

Insulin-like growth factor 1

Ile

Isoleucine

KO

Knock-out

LAT

L amino acid transporter

LDHB

Lactate dehydrogenase B

Leu

Leucine

LRP1

Low-density lipoprotein receptor-related protein 1

MCTs

Monocarboxylate transporters

Met

Methionine

MFS

Major facilitator superfamily

Mfsd2a

Major facilitator superfamily domain-containing protein 2a

mTOR

Mammalian target of rapamycin

N- and C-termini

Amino and carboxyl termini

NF-kB

Nuclear factor kappa-light-chain-enhancer of activated B cells

NFTs

Neurofibrillary tangles

Oatp1c1

Organic anion transporter polypeptide 1c1

OCTN

Organic cation novel transporter

PEST domains

Domains enriched of Pro, Glu, Ser, Thr

Phe

Phenylalanine

PI3K

Phosphatidylinositol 3-kinase

PKC

Protein kinase C

RA

Rheumatoid arthritis

RFVT1

Riboflavin transporter 1

RFVT2

Riboflavin transporter 2

RFVT3

Riboflavin transporter 3

ROS

Reactive oxygen species

SLC16

Solute carrier 16

SLC52

Solute carrier family 52

SVCT

Sodium-dependent vitamin C transporter

T₃

3,5,3′-Triiodothyronine

T₄

Thyroxine

Thr

Threonine

TLS

Translation start site

Trp

Tryptophan

TRs

Thyroid hormone receptors

TSC1/2

Tuberous sclerosis proteins 1/2

TXNIP

Identification of thioredoxin-interacting protein

Tyr

Tyrosine

UDP-glucose

Uracil-diphosphate glucose

Val

Valine

VE-Cadherin

Vascular endothelial (VE)-cadherin

VEGFR2

Vascular endothelial growth factor receptor 2

Wnt

Wingless-related integration site

Author contributions

YTKN, HTTH, THN drafted the manuscript. LNN edited and proofreading of the manuscript.

Funding

Government funding sources. Singapore Ministry of Health’s National Research Council NMRC/OFIRG/0066/20, Ministry of Education MOE2018-T2-1–126, MOE-Tier-1, and NUSMED-FOS Joint Research Programme grant on Healthy Brain Aging grants.

Availability of data and material

Not applicable.

Declarations

Conflict of interest

The authors have no conflict of interest.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Applicable. The authors approve for publication of this work.