Acyl-CoA synthetases as regulators of brain phospholipid acyl-chain diversity

Abstract

Each individual cell-type is defined by its distinct morphology, phenotype, molecular and lipidomic profile. The importance of maintaining cell-specific lipidomic profiles is exemplified by the numerous diseases, disorders, and dysfunctional outcomes that occur as a direct result of altered lipidome. Therefore, the mechanisms regulating cellular lipidome diversity play a role in maintaining essential biological functions. The brain is an organ particularly rich in phospholipids, the main constituents of cellular membranes. The phospholipid acyl-chain profile of membranes in the brain is rather diverse due in part to the high degree of cellular heterogeneity. These membranes and the acyl-chain composition of their phospholipids are highly regulated, but the mechanisms that confer this tight regulation are incompletely understood. A family of enzymes called acyl-CoA synthetases (ACSs) stands at a pinnacle step allowing influence over cellular acyl-chain selection and subsequent metabolic flux. ACSs perform the initial reaction for cellular fatty acid metabolism by ligating a Coenzyme A to a fatty acid which both traps a fatty acid within a cell and activates it for metabolism. The ACS family of enzymes is large and diverse consisting of 25–26 family members that are nonredundant, each with unique distribution across and within cell types, and differential fatty acid substrate preferences. Thus, ACSs confer a critical intracellular fatty acid selecting step in a cell-type dependent manner providing acyl-CoA moieties that serve as essential precursors for phospholipid synthesis and remodeling, and therefore serve as a key regulator of cellular membrane acyl-chain compositional diversity. Here we will discuss how the contribution of individual ACSs towards brain lipid metabolism has only just begun to be elucidated and discuss the possibilities for how ACSs may differentially regulate brain lipidomic diversity.

1. Introduction

Fatty acids are critical substrates for numerous metabolic processes, have high potential for energy production, and serve as the foundation for cellular membrane hydrophobic regions. Across the body, the distribution and metabolic fate of fatty acids are highly differential. Metabolically, fatty acid fate can be placed into three major categories: oxidation for energy, storage as triacylglycerol, and use for phospholipid synthesis and remodeling. While the oxidation and storage of fatty acids are the main metabolic fates for certain cell types or in response to feeding/fasting conditions, the use of fatty acids for incorporation into phospholipids, either new or during remodeling, is a major metabolic fate of all cells across the body and under all metabolic conditions because phospholipids are critical structural components of all membranes.

The brain is very lipid-rich, due to its high phospholipid content, and is considered the second most fatty organ next to adipose tissue. The importance of lipid metabolism in maintaining healthy brain function is exemplified by the numerous models of neurological diseases that stem from primary defects in lipid metabolism. For instance, Niemann Pick, Tay-Sachs, Mabry syndrome, and numerous mental retardations, narcolepsies, and paraplegias are caused by primary inherited defects in lipid metabolism [1]. Many neurological diseases and disorders that do not necessarily result from an inherited defect in a lipid-metabolizing gene are characterized by clear defects and abnormalities in lipid metabolism and lipid content. Yet the role of these lipid-related defects in the etiology and progression of disease remains unclear [1]. Previously, we showed that a lipid-metabolizing enzyme, acyl-CoA thioesterase 7, is critical for preventing neuronal lipotoxicity, ALS-like symptoms, and seizure susceptibility in mice [2, 3]. These outcomes were later confirmed to occur in the first report of a human with an inherited defect in acyl-CoA thioesterase 7 [4]. Thus, it is possible that cases of neurological diseases and disorders with unknown etiologies are yet to be linked to lipid metabolism as the primary defect.

Historically, the brain is not thought to oxidize fatty acids for energy. This concept is challenged with data supporting beta-oxidation of the omega-3 fatty acid eicosapentaenoic acid (EPA) in the brain [5] and the ability of astrocytes to oxidize fatty acids [6–8]. The overall contribution and criticality of fatty acids as an energy source in the brain remained unclear until a recent report demonstrated that mitochondrial fatty acid oxidation is seemingly dispensable for the brain [9]. Namely, the brain-specific loss of carnitine palmitoyltransferase 2, a non-redundant enzyme required for mitochondrial oxidation of long-chain fatty acids, does not result in a major overt phenotype in mice [9]. The opposing metabolic fate of fatty acid oxidation is fatty acid storage in triacylglycerol. Neurons are not known to store lipids as triacylglycerol; however evidence demonstrates that neurons contain triacylglycerol lipase, suggesting that these cells have the ability to metabolically degrade stored lipids [10]. Yet consensus remains that very little fat is stored as triacylglycerol in the brain. Above all, the major metabolic fate for fatty acids in the brain is for incorporation into phospholipids.

Lipids constitute 50% of the brain’s dry weight, predominantly in the form of phospholipids [11]. Phospholipids are the major component of cell membranes encasing all cells, organelles, and vesicles. Membranes and their fatty acid composition are in constant flux through the actions of membrane remodeling, vesicle endocytosis and exocytosis, autophagy, and the formation, expansion, and retraction of cells and organelles [1]. These cellular processes occur at particularly high rates in the brain. At synapses, synaptic vesicles undergo exocytotic fusion with the presynaptic plasma membrane, followed by endocytosis-mediated retrieval to regenerate synaptic vesicles. This vesicle recycling process requires membrane reorganization at each step to provide an adequate lipid environment for sustaining synaptic responses [12–14]. Dendritic spines are constantly changing in number and shape and undergo enlargement or shrinkage upon stimuli that affect total plasma membrane area [15–17]. Other cellular processes that involve lipid remodeling include the axoplasmic transport of vesicles containing proteins and lipids, myelination, and axon regeneration [18–20]. Furthermore, all these processes are influenced by learning, memory formation, age, and pathogenic conditions [21]. Thus, phospholipids are dynamic molecules essential for physiological brain function.

2. Methods

Unbiased lipidomic profiling across mouse tissues was performed with Purdue’s Bindley Metabolite Profiling Facility as previously described [22]. Data was analyzed by calculating the percent distribution for each ion (ion peak m/z intensity divided by total ion intensity*100).

3. Influence of phospholipid acyl-chain diversity on membrane properties and cellular function

Cellular membranes are highly heterogeneous assemblies comprised of proteins and lipids, predominated by phospholipids. Membrane diversity arises from the structural components of phospholipids, which consist of a headgroup and a hydrophobic tail containing two-fatty acids. Membrane composition influences various membrane properties including membrane curvature, bending, packing defects, and recruitment and stability of lipid-binding proteins and membrane-spanning proteins. While the headgroup can affect these properties [23], so too can the acyl-chain. The length of the phospholipid acyl-chain directly regulates membrane thickness and is important for proper transmembrane protein fit and function, whereby the hydrophobic region of the phospholipid membrane needs to match the length of the hydrophobic segments of the protein. A hydrophobic mismatch can be relieved by membrane lipid remodeling or can result in lateral displacement of the protein and/or alterations in protein folding, which can ultimately affect protein function [24]. In addition to acyl-chain length, its unsaturation also affects membrane properties such as membrane bending. A high percentage of straight-chained saturated fatty acids will render the membrane tight and rigid, whereas the incorporation of kinked unsaturated fatty acids will confer flexibility and introduce packing defects and deformation [25, 26]. Thus, an increase in unsaturated fatty acid content will facilitate processes, such as membrane vesicle cycling that require membrane deformation. In the brain, membrane vesicle cycling at neural synapses is crucial for neurotransmission and highly influenced by the lipid composition of the synapse [27]. Besides influencing the architecture of the membrane, some proteins interact selectively with specific lipids to mediate synaptic vesicle fusion, which highlights the importance of lipids in mediating protein function required for cellular processes [12, 28]. Given that the phospholipid acyl-chain modulates membrane biophysical properties, it is predicted that small changes in acyl-chain composition have the potential to impact numerous essential biological functions.

In addition to the many functions of membrane acyl-chains, membrane fatty acids serve as precursors for the synthesis of signaling molecules. For instance, the omega-6 polyunsaturated fatty acid (PUFA), arachidonic acid (AA, C20:4n6) can be cleaved from membrane phospholipids by phospholipase activity and be subsequently enzymatically processed into pro-inflammatory bioactive lipid-derivatives such as thromboxanes, leukotrienes, and prostaglandins [29] and into anti-inflammatory mediators such as EETs and lipoxins [30, 31]. The omega-3 PUFA docosahexaenoic acid (DHA, C22:6n3) can be similarly cleaved and enzymatically metabolized into a different class of anti-inflammatory and neuroprotective bioactive lipid-derivatives called pro-resolving mediators such as neuroprotectins [32, 33]. The pro-resolving mediators generated from the enzymatic oxidation of DHA elicit biological actions including modulation of apoptosis, neurotransmission, and play key roles in the resolution of inflammation [34, 35]. An additional metabolic fate of fatty acids is their ligation to amines, amino acids, or neurotransmitters forming lipoamines with biological activity [36]. One such lipoamine is N-docosahexaenoyl-ethanolamide (synaptamide) which promotes synaptogenesis and is anti-inflammatory [37, 38]. Given the wide variety of neuroprotective functions that PUFAs play in the brain, it is speculated that modulating the levels of these PUFAs in membrane phospholipids directly impacts brain function, response to injury, and disease susceptibility.

4. Tissue-specific phospholipid acyl-chain diversity

The acyl-chain composition of phospholipids is highly distinct across tissues. To demonstrate this variance, heatmaps were generated from primary data using broad and unbiased phospholipid profiling across several tissues (Fig. 1). Phospholipid profiling was performed with Purdue University’s Bindley Metabolite Profiling Facility across mouse tissues, as previously described [22, 39]. The data are presented as a percentage of total ion counts and include lipid species that are ≥2% in abundance in at least one of the tissues. These data not only demonstrate the diversity of phospholipid composition across tissues but also demonstrate clustering for specific lipid species in either the central nervous system or in the periphery. For instance, the central nervous system had high abundance of phosphatidylcholine (PC) with 0- to 1-unsaturated bonds, whereas peripheral tissues were robustly enriched in PCs with 2-unsaturated bonds (Fig. 1). Among PCs with 3 or more unsaturated bonds, peripheral tissues had high abundance but roughly equal distribution of 3-, 4-, 5-, and 6-unsaturated PCs. However, there was an exceptional enrichment of specific species in isolated tissues, such as enrichment of C34:4 in the heart, C38:5 in the testes, and C38:6 in the soleus muscle. The central nervous system was particularly enriched with 4- and 6-unsaturated PCs, but not unsaturated PCs of 3- (none of which were more than 2% in total phosphatidylcholine) or 5- (which may reflect the low abundance of EPA in the brain) unsaturations. Among brain regions, the hippocampus is particularly enriched with PC36:4 and the cerebellum with PC38:6 and PC40:6 likely reflecting enrichments with the omega-6 AA (C20:4n6) and the omega-3 DHA (C22:6n3) in these regions, respectively.

Fig. 1. Phosphatidylcholine acyl-chain diversity across tissues.

Heatmap of PC by acyl-chain length and degree of saturation across A) mouse tissues and B) brain regions, n = 6. Data represent percent of total ion count. Hippo, hippocampus; cere, cerebellum.

The plasma displayed a highly distinct phospholipid profile when compared to tissues. These results are of interest because plasma lipids are often used as surrogate indicators for whole-body lipid/fatty acid status. The incongruencies between plasma and tissue fatty acid composition challenge the validity of such generalizations. Indeed, the disconnect between plasma and tissue abundance of a particular PUFA, DHA, is well documented by numerous studies that nicely demonstrate the effectiveness of DHA supplementation to increase DHA in plasma and peripheral tissues, but not in the brain [40–45]. This effect is independent of the dietary molecular format of DHA and is demonstrated for supplementation of DHA in the form of a phospholipid, ether ester, free acid, triacylglycerol, and lysophospholipid [40, 44, 45]. Therefore, the notion that plasma fatty acid profiles accurately represents that of the central nervous system remains unvalidated.

Across brain regions, the overall phospholipid profile was similar (Fig. 1B). Like the periphery, brain phospholipids contain large amounts of saturated fatty acids, such as palmitate and stearate. However, the high content of phospholipids in the brain results in a unique enrichment with the polyunsaturated fatty acids: AA and DHA, which are 3–4 fold higher in the central nervous system compared to the periphery [46]. The uniqueness of the PUFAs, AA and DHA, is largely defined by the unsaturated bonds located at the omega-6 and omega-3 position, respectively. Mammals cannot desaturate fatty acids at the omega-3 or −6 position thus, fatty acids with these bonds must be consumed from exogenous dietary sources and are therefore essential dietary nutrients. Omega-6 fatty acids are found in land plants and grains and are thus high in plant-based foods and oils, grains, grain-products, and grain-fed animal products. The omega-3 fatty acid, DHA, is particularly abundant in sea plants, thus predominantly consumed indirectly via marine animal (i.e. fish) consumption. The precursor to DHA is alpha-linolenic acid (ALA), which can be found in certain nuts and vegetables; however, very little ALA is metabolically converted to DHA within land animals, making direct DHA consumption via marine products the ideal dietary source to increase bodily DHA. The importance of omega-3 and omega-6 fatty acids is exemplified by the effects of insufficient dietary intake of these fatty acids, which leads to impaired cognition and increased risk for neurological disorders [47–51]. Omega-3 and omega-6 fatty acid accretion in the brain is particularly high during the brain growth spurt, a period that includes the last trimester of gestation and up to the first two years of life when deficient consumption of omega-3 fatty acids is associated with impaired cognition [52–55]. Cognitive deficits associated with low omega-3 fatty acid intake can be observed beyond development, during aging [56–59]. Numerous age-related neurological diseases and disorders are inversely associated with omega-3 fatty acids [58–64]. While several observational, animal, and randomized control trial studies show positive outcomes, null outcomes between omega-3 fatty acid consumption and brain health are also widely reported [64–68]. The reason that omega-3 fatty acid consumption fails to mediate beneficial effects in some intervention studies remains debated [56, 69]. One plausible explanation relates to unexplained variable rates of absorption and membrane integration of dietary and/or supplemental forms of omega-fatty acids into the brain [40–45]. Here, we will focus on examining a class of enzymes, acyl-CoA synthetases, that likely contribute to the metabolic enrichment of fatty acids into brain phospholipids.

5. Acyl-CoA synthetases as regulators of cellular phospholipid acyl-chain diversity in brain

Phospholipids acquire their acyl-chain composition through two major pathways: de novo synthesis and remodeling of phospholipids. The de novo synthesis of phospholipids occurs predominantly through the Kennedy pathway and cytidine diphosphate (CDP) pathways. The initial step in these pathways is the activation of a fatty acid by Acyl-CoA Synthetase (ACS) followed by the ligation of the acyl-CoA to the sn-1 position of glycerol-3-phosphate by glycerol-3-phosphate acyltransferase (GPAT) to form a lysophosphatidic acid (LPA). A second acyl-CoA, generated by ACS, is then ligated onto the LPA by an LPA-acyltransferase (LPAAT/AGPAT) to form a phosphatidic acid (PA). Thus, the acyl composition of PA, the common precursor of all newly synthesized phospholipids, is influenced by the concerted action of ACS, GPAT, and LPAAT/AGPAT. PA is metabolized to diacylglycerol (DAG) by the action of phosphatidic acid phosphatase (PAPase/lipin) for synthesis of phosphatidylethanolamine, -choline, and -serine. The synthesis of phosphatidylglycerol and cardiolipin in the mitochondria, or phosphatidylinositol at the ER is mediated by the conversion of PA to CDP-DAG by CDP-diacylglycerol synthetase. Each step of this pathway that stands to influence the acyl-chain specificity by substrate preferences for either fatty acids (ACSs) or acyl-CoAs (acyltransferases) is comprised of enzyme families that consist of up to 5 members, except for the ACS family, which consists of 26 family members in humans and 25 in rodents [70–73]. Thus, the selectivity of the acyl-chain in de novo phospholipid synthesis may largely depend on the ACS family.

The second pathway that regulates the acyl-chain composition of phospholipids is by remodeling via the Land’s cycle, which is a highly active process in the brain [74–76]. The Land’s cycle involves the hydrolysis of fatty acids from phospholipids within membranes by phospholipases (PLAs). The resulting lysophospholipid is re-acylated using a new acyl-CoA, generated by ACS. Re-acylation is performed by lysophosphatidylx acyltransferase (LPxAT), ‘x’ referring to the head group of the lysophospholipid substrate. The contribution of LPxATs to acyl-chain diversity is widely investigated [1, 77–79]. LPxATs are members of the LPAAT/AGPAT and the membrane bound-O-acyltransferases (MBOAT) families of enzymes that incorporate acyl-CoAs into the sn-2 position of phospholipids, the primary site of PUFA esterification. Studies on knockout mice demonstrate that LPxATs can regulate PUFA levels in membrane lipids in a substrate- and tissue-specific manner (Reviewed extensively in [73, 80]). For example, LPAAT3 deficient mice exhibit reductions in DHA-containing phospholipids in the retina and testes while LPIAT1/MBOAT7 deficient mice show a defect in the remodeling of AA within phosphatidylinositol [81–84]. While LPxAT activity affects acyl-chain diversity, we highlight that the substrate for LPxATs are acyl-CoAs, the product of ACS enzyme activity. Hence, the addition of a new acyl-chain during the Land’s cycle is driven by the combined preferences of LPxATs and the availability of ACS-generated acyl-CoA substrates. It is important to consider that cellular acyl-CoAs concentrations are rather low and will result in toxicity upon accumulation, therefore acyl-CoAs are rapidly metabolized suggesting that substrate selection may predominately fall to ACS enzyme action rather than acyltransferases. This concept was exemplified by our recent work demonstrating that loss of one particular ACS, Acsl6, greatly diminishes membrane DHA in the brain [39]. In summary, both the Kennedy and Land’s cycle rely on ACS action, performed by a large and diverse family of enzymes, which likely shapes fatty acid selectivity for phospholipid acyl-chain diversity by providing the limiting substrate, acyl-CoA.

To better understand the possible role the large family of ACS enzymes has in regulating acyl-chain diversity, the totality of ACS enzymatic action should be considered. For instance, ACS not only performs the initiation reaction required for cellular fatty acid metabolism, it is also an energy-dependent and therefore committing step for fatty acid cellular retention. Thus, the formation of an acyl-CoA accomplishes two outcomes: First, it traps the fatty acid within the cell, and second, it activates the fatty acid for downstream metabolism [85, 86]. The notion that each ACS uniquely contributes to regulating cellular lipidome diversity is predicated upon several observations. First, the existence of 26 ACSs in humans and 25 in rodents, all encoded by distinct genes and whose phylogenetic expansion increases with higher ordered organisms, suggests non-redundancy and selectivity at a functional level [70, 87, 88]. Second, each ACS enzyme exhibits diverse and unique tissue- and subcellular-specific dispersion throughout the body [89–91]. Third, each ACS enzyme has distinct enzyme kinetics, substrate preferences, and regulatory mechanisms [70]. Fourth, members of the ACS family cannot compensate for one another [92–96]. Fifth, mounting evidence demonstrates that each ACS enzyme directly regulates the channeling of fatty acids towards specific metabolic fates [92–96]. Sixth, ACS action sits at the committed and pinnacle site in cellular lipid metabolism where acyl-chain selectivity ultimately occurs. Acyl-CoA partitioning by ACS enzymes is a long-established theory to explain the regulatory mechanisms that control cellular fatty acid metabolic flux [85, 86, 97, 98]. This theory is founded on the many nuances of the ACS family [70, 89, 90, 98–100]. Data supporting this theory was generated from in vitro models of ACS overexpression or knockdown where distinct ACSs exert distinct effects on fatty acid metabolic flux and cannot compensate for one another [95, 101–103]. This theory was largely confirmed by in vivo data demonstrating that several different ACS enzymes direct fatty acids towards specific downstream metabolic pathways [39, 90, 93, 104]. The ability of ACS enzymes to channel fatty acids towards metabolic fates is predicted to depend, in part, on the subcellular location and the binding partners of a particular ACS, properties which can vary between cell-types and physiological conditions, even for the same ACS enzyme. For example, one such ACS, ACSL1, is predominantly localized to the mitochondria and is shown to preferentially guide fatty acids to mitochondrial beta-oxidation in the heart, muscle, and brown adipose [92, 105]. However, ACSL1 also regulates cardiolipin acyl-chain composition, suggesting a role for ACSL1 in mitochondrial beta-oxidation and phospholipid metabolism simultaneously [106]. The regulation between these dual roles for this enzyme is partly explained by transient protein-interactions of ACSL1 that are mediated by metabolic states. Specifically, ACSL1 binds to lipid droplet proteins only in the nutrient-deprived state when fatty acids are needed for energy production through beta-oxidation [107, 108]. These data suggest that temporary ACS-protein interactomes facilitate fatty acid metabolic flux through demand-driven mechanisms. Here, we will describe the potential roles that ACSs may play in regulating brain lipidome diversity.

6. Acyl-CoA synthetase nomenclature

The ACS family of enzymes is large and thus the nomenclature is an important consideration. The nomenclature of the ACS family is primarily based upon the preferred fatty acid chain-length and is broken into subfamilies termed short (ACSS), medium (ACSM), long (ACSL), and very-long-chain (FATP/ACSVL) ACSs, with several other subfamilies including the family-member (ACSF) and bubblegum (ACSBG) ACSs (Table 1) [70, 109]. Of note, the ACSVL subfamily is often referred to as the fatty acid transport protein (FATP) subfamily. While each of the six members of the FATP/ACSVL family has acyl-CoA synthetase activity, the use of FATP rather than the ACSVL designation results in the interpretation that FATPs are transporter proteins transporting fatty acids across the plasma membrane [110]. Debate remains regarding the ability of FATPs to transport fatty acids, independent of ACS activity [109]. However, it is speculated that the ability of FATPs, as well as all other ACS enzymes, to increase cellular fatty acid uptake is mediated by the activation of fatty acids after passing through the plasma membrane, thereby trapping fatty acids as a direct result of their CoA-ligation via ACS activity, a concept termed ‘vectorial acylation’ [111]. Importantly, endogenous FATP/ACSVLs are not localized to the plasma membrane negating their role as fatty acid transporters for cellular fatty acid uptake [110]. In agreement, overexpression of an ACS activity-dead FATP4 ablates its ability to increase fatty acid uptake [112]. Despite attempts to rename the FATP subfamily of enzymes to ACSVLs [113, 114], the widespread use of FATP nomenclature continues, often resulting in erroneous experimental designs and data interpretations. The ACS nomenclature is further complicated by differences between species (i.e. humans, mice, rats, etc. reviewed elsewhere [70, 110, 114]); thus caution is advised when using nomenclature as a basis for assigning ACS protein function and for interspecies interpretations.

Table 1.

Human ACS proteins in the brain. ‘-‘ denotes lack of alias, enrichment, or detection in brain.

ACSS1	ACAS2L, AceCS2L, dJ568C11.3, MGC33843	Placenta	Y
ACSS1	ACAS2L, AceCS2L, dJ568C11.3, MGC33843	Placenta	Y
ACSS2	ACAS2, AceCS, ACS, ACSA, dJ1161H23.1	Skeletal muscle	Y
ACSS3	FLJ21963	–	Y
AACS	ACSF1, FLJ12389, SUR-5	–	Y
ACSF2	ACSMW, FLJ20920	Kidney	Y
ACSF3	–	–	Y
ACSM1	BUCS1, MACS1	Breast	–
ACSM2A	A-923A4.1, ACSM2, MGC150530	Kidney, Liver	–
ACSM2B	ACSM2, HXMA, HYST1046	Kidney, Liver	–
ACSM3	SA, SAH	Liver, ovary, pancreas	Y
ACSM4		Lymphoid tissue, testis	–
ACSM5	FLJ20581	Liver	Y
ACSM6	bA310E22.3, C10orf129	Lymphoid tissue, pancreas, stomach	–
ACSL1	ACS1, FACL1, FACL2, LACS, LACS1, LACS2	Adipose, liver	Y
ACSL3	ACS3, FACL3, PRO2194	Parathyroid gland	Y
ACSL4	ACS4, FACL4, LACS4, MRX63, MRX68	–	Y
ACSL5	ACS2, ACS5, FACL5	Intestine	Y
ACSL6	ACS2, FACL6, KIAA0837, LACS2, LACS5	Brain, testis, seminal vesical, ductus deferens	Y
ACSBG1	BG1, BGM, FLJ30320, hBG1, hsBG, KIAA0631, MGC14352	Adrenal, brain, skin	Y
ACSBG2	BGR, DKFZp434K1635, PRTD-NY3	Testis	–
ACSVL1	SLC27A2 FACVL1, FATP2, hFACVL1, HsT17226, VLACS, VLCS	Liver	Y
ACSVL2	SLC27A6, FACVL2, FATP6, VLCS-H1	Adrenal gland	Y
ACSVL3	SLC27A3, FATP3, MGC4365	–	Y
ACSVL4	SLC27A4, FATP4	Intestine	Y
ACSVL5	SLC27A1FATP, FATP1, FLJ00336, MGC71751	–	Y
ACSVL6	ACSB, SLC27A5, FACVL3, FATP5, FLJ22987, VLACSR, VLCS-H2, VLCSH2	Liver	Y

7. Acyl-CoA synthetases in the brain

Of the 26 ACS enzymes, not all are robustly expressed in the brain. Based on the human protein atlas [www.proteinatlas.org] [115] the following ACSs are detectable in the brain: all 3 short-chain ACSs, 2 of the 7 medium-chain ACSMs (ACSM3 and −5), 1 of the 2 bubblegum ACSs (ACSBG1), all 3 family member ACSFs, all 5 of the long-chain ACSLs, and all 6 of the very-long-chain ACSVLs/SLC27As (Table 1). Of all the ACSs only two are specifically enriched in the brain when compared to peripheral tissues: ACSBG1 and ACSL6 [115]. ACSM6 is found in humans and expressed in the brain, but this family member is not present in rodents, suggesting that it may create a further advantage for cognitively enhanced organisms. Except for ACSM6, human, mouse, and rat tissues have similar expression patterns of the ACSs [89]. The detection of the majority of ACSs in the brain suggests a high degree of fatty acid metabolic control in this organ. The expression of ACSs highly enriched in the brain compared to peripheral tissues suggests specific and potentially unique roles for these ACSs in modulating fatty acid metabolism within the central nervous system.

8. Acyl-CoA synthetases in the brain by cell type

The brain consists of several distinct broad cell type classifications. Neurons are a major class of cells in the central nervous system whose function is to transmit information across the brain and to the periphery [116]. A second major class of cells in the brain is glial cells, which can be subdivided into distinct cell types such as astrocytes, oligodendrocytes, and microglia [116]. Astrocytes help maintain neuronal homeostasis and modulate neurotransmission; oligodendrocytes produce myelin; and microglia are the innate immune cells of the central nervous system [116]. Thus, each cell type performs specific, metabolically distinct functions. To estimate if different ACSs may differentially regulate the lipidome in each brain cell type, cell-type-specific ACS expression needs to be defined. To this end, information on differential expression of ACSs across cell types of the brain can be gleaned from large-scale omics studies on isolated cell-types. The ideal time-point to successfully isolate and obtain a high yield of distinct, purified cell types from the brain is early postnatal development [117–119]. However, age is an important factor when assessing cell-type expression of genes in the brain because neurons, astrocytes, and microglial undergo significant maturation processes during fetal and postnatal development that significantly change expression profiles. In agreement, not all of the ACS genes are expressed in early postnatal development, for instance, Acsl6 which is one of the two ACSs most enriched in the brain, is expressed at very low levels in early postnatal period and increases greatly across postnatal development [39, 120]. Thus, any gene that changes between early postnatal and adult is not well captured in these databases. Advances in technology have overcome these obstacles with the advent of single-cell transcriptomics to define distinct cell type transcripts from adult human brain [121, 122]. Here, we took advantage of the Allen Brain Atlas’s use of these single-cell RNAseq data in the Transcriptomics Explorer application to demonstrate the diversity of cell-type distribution and overall abundance of ACSs in the human brain. The total counts per million from adult human brain data is graphically represented (Fig. 3). Here, we observe that some ACSs are highly enriched in a single cell type. For example, Acsbg1 is nearly exclusively expressed in astrocytes, Acsm5 is exclusive to microglia, and Acss1 is most abundant in astrocytes. Acsl6 is split between two cell types, astrocytes and neurons. Yet most of the ACSs are expressed at variable degrees across all cell types. The neuronal cells are further dissected in the database and are broadly separated into either GABAergic or glutamatergic neurons. Interestingly none of the ACSs showed dominance in either GABAergic or glutamatergic neurons, but rather segregate into more narrow cell populations that are differential for each ACS. The milieu of ACSs expressed in distinct cell types and variation of their relative abundance likely contributes to the highly differential regulation of acyl-chain composition across the various cell types of the central nervous system, as demonstrated by brain lipid imaging [123–125]. The use of mass spectrometry-based approaches to resolve single-cell lipidomic fingerprints and high-resolution imaging in the brain has begun to further define cell-type specific lipid composition [126, 127]. Linking the enzymatic lipid metabolizing machinery to the regulation of the cell-specific lipidome requires further investigation.

Fig. 3. Brain ACS enzyme abundance and cell-type distribution.

RNA, counts per million (CPM), averaged for the different cell types using human brain RNAseq data from the Allen Brain Atlas Transcriptomics Explorer application for each ACS.

When considering the diversity of membrane acyl-chains across cell types and regions of the brain, it is important to distinguish the regulation of acyl-chain uptake from retention. Retention of a given fatty acid within a cell or organ system is mainly driven by the ACS reaction which traps and commits fatty acids for cellular metabolism. However, two factors contribute to acyl-chain composition: first, ACS-mediated selectivity for committing fatty acids to cellular metabolism; and second, the pool of available fatty acid substrates that a given cell is exposed to. The available fatty acid pool is particularly enigmatic when considering the brain because of the longstanding mystery regarding the mode and key players regulating fatty acid transport across the blood-brain barrier, particularly the transport of PUFAs such as DHA. In 2014, Mfsd2a was reported to act as a transporter of lysophospholipids (lysoPLs), preferring DHA-containing lysoPLs [128]. Mfsd2a is highly expressed in the endothelium of the blood-brain barrier and the loss of Mfsd2a in mice reduced brain DHA content by half [128]. These data suggest that transport of lysoPL through Mfsd2a is a major contributor to brain DHA. However, Mfsd2a is also required for cargo exchange across the blood-brain barrier, thus its deficiency results in neuronal loss, thereby reducing brain DHA as a direct result of lower neuronal abundance [129, 130]. Moreover, kinetic studies demonstrate that free DHA, rather than lysoPL-DHA, is the major contributor to brain DHA content [131]. Taken together, these data cast doubt on the role of Mfsd2a as a major regulator of brain DHA. The transport of free DHA across the blood-brain barrier can occur independent of proteins, as free fatty acids can freely cross membranes. However, retention of free fatty acids such as DHA requires their activation within the parenchyma of the brain, thereby requiring the action of ACS. [132]. Beyond acquisition of new fatty acids within the parenchyma, retention of pre-existing membrane acyl-chains also requires ACS action during the highly active process of brain phospholipid remodeling [74–76]. Thus, the combined need for ACSs in phospholipid remodeling, phospholipid de novo synthesis, and brain fatty acid retention, implies that ACSs have the major role in regulating brain acyl-chain specificity in membranes.

9. Evidence of ACSs in regulating brain lipidome diversity

Because brain ACS isoforms have different substrate preferences, subcellular locations, and are expressed differently across cells of the brain, brain ACSs likely have distinct functions in regulating the brain lipidome. Here, we provide a brief summary of relevant findings related to high brain expressing ACSs that regulate the metabolism of long-chain fatty acids, as these are the predominant fatty acids that populate the phospholipidome.

Acsl1:

Acsl1 is expressed across the body but highly in adipose, liver, and muscle tissues [89]. Acsl1 predominantly channels fatty acids into mitochondrial oxidative metabolism for energy production [92, 105, 133]. Acsl1 prefers saturated fatty acids as substrate but also has activity for PUFAs such as AA and linoleic acid (LA, C18:2n6) [134]. While conditionally deficient Acsl1 mice are reported [92, 105, 106, 133, 135–138], the effects of Acsl1 deficiency on brain metabolism, lipidome, and phenotypic outcomes remain unknown.

Acsl3:

Acsl3 is expressed in adipose, liver, and adult brain, but most abundantly expressed in the brain during fetal development [89, 139]. ACSL3 has high substrate preference for alpha-linolenic acid (ALA, C18:3n3), LA, and AA, but very little activity for DHA [140]. In vitro studies and association-based research suggests that Acsl3 has many roles in regulating lipid and cell function, including regulation of transcription, ferroptosis, inflammation, and in channeling fatty acids towards beta-oxidation and phospholipid synthesis [95, 141–143]. The role that Acsl3 plays in brain remains unknown.

Acsl4:

Acsl4 is expressed in the brain, but also widely expressed throughout the body. Acsl4 is distributed across brain cell types and found in the mitochondria, ER, and peroxisomes [144, 145]. Acsl4 has a high preference for EPA and AA, and also uses eicosanoids and related metabolites as substrate [98, 134, 146, 147]. AA is highly abundant in the brain and is substrate for pro- and anti-inflammatory metabolites, and while conditionally deficient Acsl4 mice are reported, the effects of brain Acsl4 deficiency in mice remain unreported [30, 148]. In humans, ACSL4 is implicated in bipolar disorders because valproate, a common drug used to treat bipolar disorders, is a substrate for ACSL4 and appears to disrupt Acsl4-mediated metabolism [149–151]. ACSL4 is also suspected to be important for neural development, a process when neural membrane and phospholipid synthesis increases the incorporation of AA in the brain [55, 152]. In vitro evidence demonstrates that Acsl4 protein expression increases in response to nerve growth factor and that the loss of Acsl4 impairs neuronal differentiation and dendritic spine formation [152,153]. In agreement, ACSL4 is associated with neural development disorders, such as nonspecific X-linked mental retardation [154–157]. These data suggest an important role for Acsl4 in brain development and eicosanoid metabolism, however, further investigation is warranted.

Acsl6:

Acsl6 is particularly enriched in the brain, has a high preference for DHA as its substrate, and its manipulation in vitro is directly correlated with neurite extensions [89, 101, 134, 140, 158]. In 2018, we reported the first Acsl6 deficient mouse model and demonstrated that Acsl6 regulates the incorporation of DHA into neural membranes, resulting in ∼50% depletion of brain DHA in Acsl6 deficient mice [39, 132]. While we showed that Acsl6 is required for brain DHA enrichment, the substrate preferences for Acsl6 are a complicated matter due to the identification of at least 5 variants of Acsl6, each with variable enzyme kinetics [140, 159–161]. All variants are detectable in the brain and several of these variants differ in their fatty acid-binding domain, thereby altering fatty acid substrate preference [86, 104, 140, 159, 162, 163]. Thus, the role of each variant in regulating the brain lipidome is unclear. In humans, mutations in ACSL6 are linked to schizophrenia and addictive behaviors such as tobacco smoking [164–169]. Interestingly, people with schizophrenia tend to be heavy smokers. These data strongly suggest a critical role for Acsl6 in enriching the brain with DHA for neuroprotection. Research to determine the nuances of Acsl6-mediated metabolic control, its relation to brain function and disease, and mechanisms therein is ongoing.

Acsbg1:

Acsbg1 is highly enriched in the brain and nearly exclusively in astrocytes. Acsbg1 has broad substrate preference for long-chain saturated and unsaturated fatty acids [170, 171]. The knockdown of Acsbg1 in vitro reduces ACS activity and fatty acid oxidation [171]. Thus, while these data suggest a role for Acsbg1 in long-chain fatty acid oxidation in astrocytes, its role in vivo in brain metabolism and function remain unknown.

Acsvl1/Fatp2:

Fatp2 is a very-long-chain fatty acid preferring ACS detectable in the brain [89]. The overexpression of Fatp2 in vitro increases uptake of omega-3 fatty acids, ALA and DHA [172]. Fatp2 is the center of controversy regarding the role of FATP proteins in mediating fatty acid uptake in a manner independent or dependent of its ACS activity. This is because two variants of Fatp2 exist, Fatp2a and Fatp2b, the latter of which is void of ACS activity due to the absence of an ATP binding site [173]. Because Fatp2b expression in vitro increases fatty acid uptake, it is speculated that ACS activity is not necessary for FA uptake by FATPs [173]. However, ACSs are predicted to dimerize, thus the ACS activity of Fatp2b’s dimer partner may facilitate FA uptake in Fatp2b overexpressing cells [174]. The brain expresses Fatp2a, not Fatp2b, yet the global deletion of Fatp2 does not result in observed brain-related phenotypes, reductions in total brain ACS activity, or changes in brain very-long-chain fatty acid content [175]. Thus, the role that Fatp2 plays in brain metabolism appears to be negligible.

Acsvl4/Fatp4:

Fatp4 is highly expressed in the brain, intestine, oxidative skeletal muscle, skin, sperm, and retina [89, 90]. Fatp4 has preference for very-long-chain fatty acids which are abundantly found in the skin, sperm, and retina, where Fatp4 is enriched [176–178]. Global FATP4 deficient mice present with degeneration of rod photoreceptors and neonatally lethal restrictive dermopathy characterized by wrinkle-free skin and severe defects in the skin barrier that cause death within hours of birth [178–182]. As expected, the loss of FATP4 reduces ≥26-carbon fatty acids in phosphoglycerolipids in the epidermis by more than 50%. FATP4 may also activate hydroxy fatty acids, intermediates of the acyl-ceramide pathway, which may further contribute to skin barrier dysfunction in FATP4 null mice [183]. Within the brain, FATP4 is found in microvessel endothelial cells, but also diffusely across the brain [120, 184]. FATP4 null mice show reductions in very-long-chain fatty acids in the brain, yet neurological phenotypes are not reported.

Acsvl5/Fatp1:

Fatp1 is expressed in the brain, retina, muscles, adipose, kidney, lung, skin and endothelium, and has high ACS activity for the very-long chain fatty acid lignocerate (C24:0) relative to palmitic acid (C18:0) [89, 185]. The global loss of FATP1 in mice results in protection from high-fat diet-induced insulin resistance and is required for thermogenic regulation by brown adipose tissue [186–188]. FATP1 is expressed in the retina and null mice present with reduced response to light by electroretinogram and enhanced age-related morphological deterioration of the retina [189]. Brain lipid composition and the neurological phenotype of the FATP1 null mice remain unknown.

10. Outstanding questions on the role of ACSs in the brain

What are the true subcellular localizations of ACSs?

A property of ACS enzymes that is predicted to play a large role in how these enzymes dictate fatty acid metabolic fate is their differential localization to distinct organelles. For instance, an ACS at the mitochondria is likely to facilitate fatty acid oxidation or cardiolipin remodeling, whereas an ACS at the endoplasmic reticulum may channel fatty acids into phospholipid or triacylglycerol synthesis. ACSs found at the plasma membrane are likely indirectly localized there due to interactions with plasma membrane proteins and/or due to membrane-membrane contact sites where they likely facilitate fatty acid uptake and/or membrane remodeling. The expression of ACSs at the peroxisome and their possible related roles are reviewed elsewhere [190]. However, the precise intracellular spatial distribution of ACSs remains unclear, in part due to the limited availability of immunochemistry-quality antibodies targeting endogenous ACS enzymes. Therefore, a majority of the subcellular location data is generated using models of overexpressed tagged ACSs in cell culture, a method that artifactually results in mislocalization of membrane proteins such as ACSs [110]. Another concern of using overexpression of ACSs in cells is the resulting overproduction of toxic acyl-CoAs [98]. As a consequence, acyl-CoAs are redirected towards alternative metabolic pathways, such as synthesis of triacylglycerols or phospholipids resulting in excess lipid droplet and membrane accumulation and thereby creating a disease-like state for the cell [98]. A secondary method to determine intracellular location is subcellular fractionation, which is often complicated by a high degree of fractional impurity. Thus the advent of highly specific and robust antibodies for detection of endogenous ACSs remains a roadblock for understanding ACS subcellular distribution.

How to relate in vitro findings to in vivo brain metabolism?

In relation to connecting ACS function in vitro to its function in the brain, concerns are raised regarding the lack of cell culture models to recapitulate mature adult cells, the complex and dynamic interactions between brain cells, and fatty acid metabolism of cells in vivo. Cells in culture, either primary or established cell lines, are well known to not fully recapitulate cells in vivo. This is particularly true for brain cells types such as neurons that are cultured as precursor/undifferentiated cells and even after differentiation, do not fully depict a mature adult cell. Furthermore, astrocytes are recognized as crucial players in the formation, function, and plasticity of synapses [191, 192]. Thus, highlighting the essentiality of neural-astrocyte interactions. The ability to recapitulate these interactions in a dish using mature astrocytes or neurons, or to study neuron-astrocyte metabolic interdependence is challenging. Additionally, cells in culture are grown in high levels of glucose with limited or no fatty acids available, thus the ability to translate in vitro findings to in vivo is particularly difficult in relation to fatty acid metabolism. Thus, the combination of in vivo and in vitro research approaches is critical to determine the metabolic role of ACSs in regulating brain metabolism, yet relatively few in vivo models of brain ACS manipulation are reported.

How to define comprehensive substrate preferences and kinetics for ACSs?

Comprehensive substrate preferences and kinetics are not established for a vast majority of the ACS enzyme family, in part due to the complex nature of optimizing assay conditions. Traditional methods to characterize ACS enzyme kinetics require optimization for numerous assay components, such as duration of assay and concentrations of ATP, triton-X, magnesium chloride, and Coenzyme A. Moreover, each fatty acid substrate can yield distinct optimized conditions which may also depend on the fatty acid vehicle (albumin vs. cyclodextrin). Enzyme kinetics are traditionally determined for individual ACSs upon generation of recombinant protein in bacteria, followed by purification using biochemical tags. The assay is performed in test-tube based conditions, often outside of native membrane-associated conditions; is run with individual substrates rather than the mixture of fatty acids that are present within cells from which ACS will normally competitively select; and uses excess amounts of substrate, which does not reflect fatty acid intracellular levels [134, 193]. Thus, establishing assay conditions for a single ACS for a single substrate is a complex endeavor performed outside of the enzyme’s natural membrane-associated environment. Even within such controlled assays, different labs yield inconsistent enzyme kinetics [86, 134, 159, 162, 177, 194]. Moreover, ACS enzymes are predicted to dimerize homo- and heterodynamically [86, 107], therefore assay conditions optimized for a single ACS in isolation may be largely changed upon heterodimerization in the natural environment. Total ACS activity assays within tissues or cell types can provide an indicator of total activity but cannot indicate contribution from individual ACS enzymes. Thus, equating kinetic data pertaining to an ACS in isolation to total ACS activity in tissue-based assays is not feasible.

While several groups have performed enzyme kinetics on recombinant ACSs using up to five fatty acid substrates, the immense pool of possible acyl substrates is highly difficult to assess. Specifically, ACS activity for less abundant fatty acids, such as oxylipids, neuroprotectins, eicosanoids, and very-long-chain fatty acids is limited [134]. The lack of data for these substrates is due to the combined effects of cost, limited availability, and instability of such lipids. With the advent of methodologies relying on mass spectrometry, the limitations in substrates for kinetic analysis can be overcome, but has yet to be widely implemented [195]. Additional complications arise because nearly every ACS enzyme has at least one splice variant which are regulated differently by numerous environmental/hormonal factors and in cell-type specific manners. The existence of these variants ultimately increases the diversity of ACS localizations at the subcellular level, binding partners, and enzyme kinetics, thus comprehensive characterizations of each variant across all ACSs is a monumental endeavor [159, 162].

Remaining unknowns for the role of ACSs in vivo.

Understanding the role of each enzyme alone on the regulation of cellular and intercellular fatty acid metabolic flux in a substrate dependent manner, and in concert with binding partners, will reveal major regulatory nodes in lipid metabolism. Since ACS enzymes display non-redundant functions and do not compensate for each other’s loss, the use of knockout mouse models can be a powerful tool to uncover ACS biochemical characteristics and their biological roles in vivo. It is of particular interest to understand how lipid metabolism is controlled across the blood-brain barrier and after subsequent incorporation and recycling within the brain parenchyma. The differential expression of the ACSs across the cells of the brain and within the blood-brain-barrier strongly suggests a degree of regulation conferred by these enzymes but much work remains to further understand this regulation and its importance for neurological health.

In summary, the phospholipid composition of membranes is highly regulated with considerable contribution from ACS enzymes. ACSs are a family of enzymes that confer the first and committed step in cellular metabolism of fatty acids and whom we predict are critical regulators of phospholipid diversity. Our recent work demonstrated a critical role for one such ACS, Acsl6, in regulating brain membrane enrichment with the neuroprotective omega-3 fatty acid DHA [39, 132]. Continued investigations of ACSs in the central nervous system will improve our understanding of normal brain physiology as well as the link between lipid metabolism and pathophysiology of neurological disorders. The substrate preferences, effects on fatty acid metabolic flux, and subcellular localizations of the ACSs and their variants in the central nervous system remain largely unresolved. Determining how fatty acid metabolism is regulated within each cell type, and by which ACS will greatly increase our understanding of how lipid metabolism within and between different cell types leads to proper brain function and disease prevention to inform personalized nutrition and therapeutic strategies.

Fig. 2. Greatest genetic diversity for ACSs within phospholipid metabolism.

Schematic representation of the major enzymes involved in de novo phospholipid synthesis (top) and phospholipid remodeling (bottom). Colored boxes represent the number of genes in the enzyme family.

Abbreviations

LA

linoleic acid

ALA

alpha-linolenic acid

PUFA

polyunsaturated fatty acids

AA

arachidonic acid

EPA

eicosapentaenoic acid

DHA

docosahexaenoic acid

ACS

acyl-CoA synthetase

CoA

coenzyme A