The role of very long chain fatty acids in yeast physiology and human diseases

Fatty acids (FAs) are a highly diverse class of molecules that can have variable chain length, number of double bonds and hydroxylation sites. FAs with 22 or more carbon atoms are described as very long chain fatty acids (VLCFAs). VLCFAs are synthesized in the endoplasmic reticulum (ER) through a four-step elongation cycle by membrane embedded enzymes. VLCFAs are precursors for the synthesis of sphingolipids (SLs) and glycerophospholipids. Besides their role as lipid constituents, VLCFAs are also found as precursors of lipid mediators. Mis-regulation of VLCFAmetabolism can result in a variety of inherited diseases ranging from ichthyosis, to myopathies and demyelination. The enzymes for VLCFA biosynthesis are evolutionary conserved and many of the pioneering studies were performed in the model organism Saccharomyces cerevisiae. A growing body of evidence suggests that VLCFA metabolism is intricately regulated to maintain lipid homeostasis. In this review we will describe the metabolism of VLCFAs, how they are synthesized, transported and degraded and how these processes are regulated, focusing on budding yeast. We will review how lipid metabolism andmembrane properties are affected by VLCFAs and which impact mutations in the biosynthetic genes have on physiology. We will also briefly describe diseases caused by mis-regulation of VLCFAs in human cells.


Introduction
Fatty acids (FAs) are a highly diverse class of molecules that can have variable chain length, number of double bonds and hydroxylation sites. FAs are building blocks for all membrane lipids and are also used to post-translationally modify proteins. Therefore, FAs are essential molecules in all organisms. FAs with 22 or more carbon atoms are described as very long chain fatty acids (VLCFAs). This special FA sub-class is mainly incorporated into SLs and glycosylphosphatidylinositols (GPI)-anchored proteins but can also be found in other lipid classes. The abundance of VLCFAs in cells has to be tightly regulated to avoid their accumulation which affects organelle structure, function and inheritance (De Carvalho and Caramujo 2018;Schneiter and Kohlwein 1997).
In this review we will focus on VLCFA metabolism and its control mechanisms in the yeast Saccharomyces cerevisiae. As for many other processes, yeast as a model organism has been essential to identify the enzymes necessary for VLCFA biosynthesis, activation, regulation and degradation (Tehlivets et al. 2007;Welch and Burlingame 1973). Subsequently, evolutionary conserved homologs have been found in all higher eukaryotes and their role in human pathophysiology soon became evident (Poulos et al. 1992;Wanders et al. 1987). FA elongation was soon shown to be distinct from de novo FA synthesis, as enzymatic activities for both processes were retrieved in different cellular subfractions (Nugteren 1965).
VLCFAs are synthesized from the products of de novo FA synthesis in the endoplasmic reticulum (ER) through a fourstep elongation cycle by membrane embedded enzymes. VLCFAs are mainly incorporated into SLs, but can also be found in phosphatidylinositol (PI), GPI-anchors and in storage lipids as components of lipid droplets (Olson et al. 2015a;Tehlivets et al. 2007). SLs in yeast consist of a long chain base (LCB) that is synthesized from serine and palmitoyl-CoA by the serine palmitoyltransferase (SPT). Amide linkage of a VLCFA to a LCB results in the formation of a ceramide. Subsequent addition of inositol-phosphate, mannose and a second inositol-phosphate results in the generation of the complex SLs in yeast.
VLCFAs can also be metabolized via β-oxidation in peroxisomes to generate energy in cells ( Van Roermund et al. 2003). In addition, VLCFAs and VLCFA-containing lipids have been linked to endocytic sorting and trafficking and have been implicated in the generation of highly curved membranes (David et al. 1998;Obara et al. 2013;Schneiter et al. 2004). Misregulation of VLCFAs has also been linked to oleate mediated lipotoxicity and activation of the unfolded protein response (UPR; (Micoogullari et al. 2020;Wang et al. 2018).
Technological developments were crucial to identify and understand the different VLCFA-containing molecules in cells. However, we are only just beginning to understand the cellular pathways regulating VLCFA homeostasis.
Here, we will summarize the control mechanisms of VLCFA metabolism and review new insights. Because of their importance in pathophysiology, we will briefly refer to disorders caused by mis-regulation of VLCFAs in humans.

The VLCFA biosynthesis pathway
De novo synthesis of FAs starts with the conserved initial step of malonyl-CoA synthesis by carboxylation of acetyl-CoA. In yeast, this reaction is catalyzed by the acetyl-CoA carboxylase Acc1 (ACC in mammals; Roggenkamp et al. 1980). Acc1 possesses activity as a carboxyltransferase and as a biotin carboxylase and requires biotin as a co-factor (Sumper and Riepertinger 1972). Malonyl-CoA is the universal FA precursor. In eukaryotes, FA synthesis is carried out by type I fatty acid synthases (FAS) a very large multifunctional protein complex. In yeast, the FAS complex consists of the two catalytic subunits Fas1 and Fas2 and the recently identified regulatory subunit Tma17 (Lomakin et al. 2007;Schweizer and Hofmann 2004;Singh et al. 2020). All five steps of FA biosynthesistransfer, elongation, reduction, dehydration, and another reductionare carried out by the FAS protein complex. In a twostep transfer mechanism, the acetyl-group of acetyl-CoA is first transferred to the sulfhydryl group of a cysteine on the FAS. In parallel, a malonyl group is transferred to the 4′-phosphopantetheine of the acyl carrier protein region of the FAS. In the second step, a 4-carbon β-keto acid derivative is generated by a transfer of two carbons from the acetyl group. Finally, a three-step reaction of reduction, dehydration and reduction yields a saturated acyl chain elongated by a two-carbon unit. Constant repetition of this cycle yields a 16-carbon FA chain. The elongation cycle is terminated by hydrolysis or transesterification of palmitate to an acceptor (for a more detailed review on FA synthesis see Leibundgut et al. 2008;Schweizer and Hofmann 2004;Tehlivets et al. 2007).
VLCFA elongation follows a cycling four step process of condensation, reduction, dehydration and reduction with each step being carried out by distinct enzymes. The first and rate limiting step is the condensation of acyl-CoA with malonyl CoA and a simultaneously occurring decarboxylation to form 3-ketoacyl-CoA. The elongation up to C18 acyl-CoA in yeast is catalyzed by Elo1. The elongation of C18 to C20 or up to C26 acyl-CoA are catalyzed by the two elongases Elo2 (Fen1) and Elo3 (Sur4) while mammals have seven different elongases (ELOVL1-7). In a second step 3-ketoacyl-CoA is reduced to 3-hydroxyacyl-CoA. The reducing enzyme is Ifa38 (Ybr159w), utilizing nicotinamide adenine dinucleotide phosphate (NADPH) as a reductant (Han et al. 2002). In the third step, Phs1 dehydrates 3-hydroxyacyl-CoA to 2,3-trans-enoyl-CoA (Denic and Weissman 2007). Finally, 3-hydroxyacyl-CoA is reduced to an acyl-CoA containing two more carbon chain units than the original acyl-CoA by the 2,3-trans-enoyl-CoA reductase Tsc13 ( Figure 1; Kohlwein et al. 2001).

Characterization of the VLCFA biosynthesis enzymes in yeast
The first and rate-limiting step of VLCFA elongation is the condensation reaction catalyzed by three different  Figure 1: The VLCFA biosynthesis pathway. Overview of the four-step FA elongation cycle in S. cerevisiae. Enzymes and co-factors are illustrated for each step. Details about the four steps -condensation, reduction, dehydration and reduction-and enzymes are described in the text. NADPH, nicotinamide adenine dinucleotide phosphate; CoA, coenzyme A. elongases Elo1-3. The molecular mechanism of VLCFA elongation in yeast is described as an adjustable caliper (Denic and Weissman 2007). Elo1 extends C12-C16 fatty acyl-CoAs to C16-C18 FAs (Toke and Martin 1996). Elo2 utilizes these products to elongate them further up to C22 FAs and Elo3 uses saturated C18 FAs to synthesize C20-26 VLCFAs. Elo2 and Elo3 have partially overlapping activities, however the final step of C24 to C26 elongation is exclusively performed by Elo3 (Oh et al. 1997). The three ELO genes and their products will be described in detail below.
ELO1 was originally identified in a genetic screen for essential genes in FAS defective cells supplemented with tetradecanoic acid (C14:0) Screening for mutants that lose the ability to elongate this FA led to the identification of ELO1 (Toke and Martin 1996). The ELO2 and ELO3 genes were identified based on sequence similarity to ELO1 (Oh et al. 1997). Further characterization of the gene products showed that every elongase has a characteristic product pattern and substrate usage with little overlap between Elo2 and Elo3. Substrate specificity for each elongase was further characterized by heterologous expression of ketoacyl synthases from C. elegans allowing yeast to use otherwise incompatible substrates for elongation (Beaudoin et al. 2000).
Yeast elongases are membrane spanning proteins that form the elongation complex at the ER membrane, necessary for the evolutionary conserved process of VLCFA biosynthesis (Denic and Weissman 2007;Revardel et al. 1995). While single null mutants of an elongases are viable, ELO2/ELO3 double mutants are lethal. In contrast to FAS mutants, elongation deficient phenotypes cannot be rescued by supplementation with exogenous VLCFAs (Rössler et al. 2003).
Single elongase mutants show distinct phenotypes (Ejsing et al. 2009). ELO1 mutants are defective in medium to long chain elongation with more glycerophospholipid species containing 14:0 and 14:1 FAs. Interestingly, the levels mannosyl-inositol-phosphorylceramide (IPC), inositol-phosphorylceramide (MIPC) and mannosyldiinositol-phosphorylceramide (M(IP) 2 C), the complex SL in yeast, are increased in ELO1 mutants. This is likely the consequence of the still active Elo2 and Elo3 enzymes. In line with that, ELO2 or ELO3 mutants show reduced M(IP) 2 C and MIPC levels and subsequent accumulation of the lipid precursors with shortened FA moieties. ELO3 knockout cells accumulate IPC with a C22:0 FA while elo2Δ cells have shorter chain C16:0 and C18:0 FAs in the major IPC species (Ejsing et al. 2009). Reduced SL levels are likely a result of reduced ceramide synthesis. When VLCFA levels are low, ceramide synthesis might become rate limiting. Alternatively the ceramide synthase has reduced activity due to the presence of a shorter, less preferred substrate, when C26 is not available (Oh et al. 1997). Defects in the ELO2-or ELO3 genes have also been correlated with decreased plasma membrane (PM) ATPase activity (Garcia-Arranz et al. 1994), reduced β-glucan synthase activity (El-Sherbeini and Clemas 1995) as well as resistance to sterol synthesis inhibitors and defects in bud localization (Revardel et al. 1995). These effects are all attributed to membrane proteins that are affected due to altered membrane properties resulting from changes in the VLCFA composition (Oh et al. 1997). Over all, this highlights the importance of VLCFAs for cellular function and the complexity of the cellular lipid environment (Ejsing et al. 2009).
The molecular mechanism of how Elo proteins determine the final length of a VLCFA still remains unclear. Since an external template for VLCFA synthesis is lacking and the same group of active sites can synthesize specific products, the molecular mechanism must be included within the biosynthetic enzymes. Purification of all membrane components of the elongation complex and reconstitution into proteo-liposomes led to the identification of a key lumen-facing lysine residue and a cytosol-facing active site for Elo2 and Elo3 that allows the Elo machinery to produce VLCFAs of a defined length. With the active site as a fixed point, a caliper-like mechanism can measure the distance to the lysine residue. Presumably, the lysine residue marks the end of the hydrophobic pocket, being located at the luminal end of a transmembrane domain. Acyl chains are potentially pushed into the transmembrane region of the protein to allow the addition of two carbon units. The acyl chain is further displaced until it has reached a certain length and arrives at the lysine residue. This triggers elongation termination. Consequently, the length of the final VLCFA product is similar to the distance between the active site and the lysine (Denic and Weissman 2007). Artificially changing the lysine position results in the generation of VLCFAs of a specific length. This consequently allows genetically engineered yeast to produce altered VLCFA products that would not be generated naturally.
Ifa38 is the major 3-ketoreductase of the elongase system. The enzyme localizes to the ER and coimmunoprecipitates with Elo3 and Tsc13 as part of the elongation complex. Ifa38 mutants show a defect in 3-ketoreductase activity and has comparable phenotypes to other mutations in the elongase cycle including reduced VLCFA levels, elevated levels of medium-chain ceramides as well as elevated dihydrosphingosine (DHS) and phytosphingosine (PHS) levels. Another phenotype of the Ifa38 mutant is a reduced dehydratase activity resulting in an accumulation of a 3-OH-acyl intermediate Han et al 2002). It is possible that Ifa38 is involved in the stabilization or localization of the 3-hydroxyacyl-CoA dehydratase. Alternatively, the entire elongation complex could be stabilized by Ifa38 but supporting evidence is lacking. IFA38 is not an essential gene, suggesting the presence of proteins with similar enzymatic activity. Along that line, Ayr1, a bifunctional enzyme with a more promiscuous 3-ketoreductase activity potentially complements the IFA38 knockout (Han et al. 2002).
Phs1 is the 3-hydroxyacyl-CoA dehydratase that catalyzes the third step of the elongation cycle ( Figure 1; Kihara 2012). PHS1 is an essential gene in yeast, highlighting the importance of VLCFA elongation in cellular physiology. Phs1 was first reported as a potential regulator of LCB levels because of a strong deficiency in converting ceramides to IPC when Phs1 levels were reduced. In addition LCBs accumulated in PHS1 mutant cells resembling a phenotype of VLCFA metabolism mutations (Schuldiner et al. 2005). The IPC synthesis defect is most likely caused by a defect in CER or phosphatidylinositol (PI) transport from the ER to the Golgi. Interestingly, other PI transport requiring steps, such as MIPC to M(IP) 2 C conversion are more affected in PHS1 mutant cells then e.g. IPC-to-MIPC conversion. This suggests a possible connection between VLCFA synthesis and PI transport (Kihara et al. 2008).
TSC13 encodes the enoyl reductase that catalyzes the last step in the elongation cycle ( Figure 1). The Tsc13 protein family is conserved from yeast to mammals with the key glutamine 81 being conserved. A tsc13-1 mutant allele containing a substitution of this residue with lysine is responsible for a temperature sensitive phenotype in yeast (Kohlwein et al. 2001). TSC13 was identified in a csg2Δ suppressor screen for calcium sensitivity (TSC = temperature sensitive suppressors of Csg2 mutants; Kohlwein et al. 2001). LCBs and ceramide species with shorter chain lengths accumulate in TSC13 mutants. The TSC13 genȮe is also essential for yeast viability. The protein localizes to the ER. Supporting the concept of an elongation complex, Tsc13 co-immunoprecipitates with Ifa38, Elo2 and Elo3 (Han et al. 2002;Kohlwein et al. 2001). Interestingly, Tsc13 is enriched in the nuclear-vacuolar junction (NVJ), a subdomain of the ER. The molecular function of Tsc13 at this domain remains enigmatic. Localization of Tsc13 to the NVJ is not essential for viability and is independent of the other elongase cycle relating enzymes but depends on the Nvj1 protein (Kohlwein et al. 2001;Pan et al. 2000). However, it cannot be excluded that its NVJ enrichment is connected to the requirement of VLCFAs in stabilizing the nuclear pore complex (NPC) (Schneiter et al. 1996) or to PI containing VLCFAs present in the nuclear membrane (Schneiter et al. 2004). Interestingly, yeast cells in stationary phase bud off Tsc13 enriched vesicles into the vacuole from the nuclear membrane suggesting an involvement of Tsc13 in nuclear envelope recycling (Schneiter and Kohlwein 1997). Since Nvj1 expression is connected to piecemeal micro autophagy of the nucleus (PMN) this further supports a connection of Tsc13 with nuclear recycling (Kvam et al. 2005).

VLCFAs as building blocks for lipids in Saccharomyces cerevisiae
In S. cerevisiae, VLCFAs are the building blocks for SLs and glycerophospholipids. They are components of GPI-anchored proteins and can be incorporated into storage lipids ( Figure 2). In the following paragraphs the role of VLCFAs in the generation of these molecules will be highlighted.
SLs are one of the three major lipid classes in all eukaryotic cells. They consist of an LCB, a FA and a polar head group. SL synthesis starts in the ER with the condensation of serine with palmitoyl-CoA to yield 3-ketodihydrosphingosine (KDHS) catalyzed by the ratelimiting enzyme SPT (Dickson et al. 2006;Olson et al. 2015b subunits Lcb1 and Lcb2 and the regulatory subunit Tsc3. The short lived intermediate KDHS is directly reduced to dihydrosphingosine (DHS) by Tsc10 (Beeler et al. 1998). DHS can be further hydroxylated at its C4 position by the hydroxylase Sur2. The resulting product is phytosphingosine (PHS) the major LCB in yeast cells. Yeast LCBs have varying chain lengths of 16-20 carbons (Lester and Dickson 2001). DHS and PHS are both substrates of the essential ceramide synthase in yeast. (Dickson et al. 2006;Schorling et al. 2001;Vallée and Riezman 2005). While mammalian cells have more variable ceramide FA composition, yeast cells primarily incorporate C26 VLCFAs in ceramides (Ejsing et al. 2009). Thus, VLCFAs are the primary FA of complex SLs in yeast cells. The most abundant VLCFA is 2-OH-C26:0 followed by C26:0 and small quantities of shorter FAs (Rattray et al. 1975). VLCFAs have varying hydroxylation sites. This also separates VLCFAs from shorter chain FAs that are usually not hydroxylated in wild-type cells (Dickson and Lester 2002) . Hydroxylation of the ceramide VLCFA depends on the Scs7 hydroxylase/desaturase that is specific for VLCFA. Thus, Scs7 and Sur2 are both required for ceramide hydroxylation (Haak et al. 1997). Ceramides can be further linked to a polar head group via a phosphodiester bond. Usually the head group is an inositolphosphoryl moiety. This moiety can be modified with a mannosyl group and a second inositolphosphate attached to the mannosyl making M(IP) 2 C, the most abundant SL in yeast. However, IPC and MIPC are also present in detectable quantities in yeast cells (Ejsing et al. 2009;Olsen and Jantzen 2001;Steiner et al. 1969).
Complex SLs are essential molecules in all eukaryotic cells that function as structural membrane components as well as signaling molecules. In S. cerevisiae SLs comprise approximately 10-20% of the total membrane lipid species (Oh et al. 1997). They mainly localize to the outer leaflet of the plasma membrane but low amounts of SLs are found in the nuclear membrane (Schneiter et al. 1996). SLs are involved in signaling, secretion, apoptosis, Ca2+ homeostasis and linked to many other processes (Dickson and Lester 2002;Dickson et al. 2006;Schneiter 1999). Phosphorylated LCBs may also play a role in the heat stress response and growth control (Skrzypek et al. 1999). The importance of VLCFAs as components of SLs is evident in mutants unable to synthesize SLs, for example, a temperature sensitive allele of the SPT (LCB1-100). At the restrictive temperature LCB1-100 cells die, unless they carry a suppressor mutation in the 1-acyl-sn-glycerol-3-phosphate acyltransferase SLC1 (named SLC1-1; AGPAT1 in mammals, Schneiter et al. 2004). LCB1-100 cells harboring the SLC1 mutation are capable of producing C26 FA containing glycerophospholipids which structurally mimic SLs (Munn and Riezman 1994;Zanolari et al. 2000). The cells bypass the essential requirement for ceramides by producing novel inositol glycerophospholipids that have SL like head group modifications and a VLCFA at the sn-2 position, highlighting the importance of these components.
In wild type yeast, PI is the only glycerophospholipid that contains VLCFAs. This unusual VLCFA-containing PI species comprises about 1% of total cellular PIs. It is found in the plasma membrane as well as the nuclear envelope. The C26-substitution is highly specific for the sn-1 position of the glycerol backbone. In liposomes composed of 1,2-dielaidoyl-sn-glycero-3-phosphoethanolamine (DEPE) it reduced the bilayer-to-hexagonal phase transition. This supports the idea of a possible involvement in membrane stabilization, especially of the highly curved membranes at the (NPC) (Řezanka and Sigler 2009;Schneiter et al. 2004) In addition, VLCFAs can also be found in storage lipids at lipid droplets. Interestingly, a FA can be attached to the hydroxyl group of a ceramide by the acyl transferases Lro1 and Dga1, yielding an acyl-ceramide (Voynova et al. 2012). In addition, after hyper activation of Elo2 and a concomitant increase in VLCFAs, VLCFAs are increased in triacylglycerols (Olson et al. 2015a).

The role of VLCFAs in GPI anchor biosynthesis
Another class of VLCFA-containing molecules are the GPI-anchor containing proteins. GPIs are complex glycophospholipids which are added post-translationally to the C-terminus of secreted proteins in the lumen of the ER. The anchoring machinery is used to attach proteins to membranes. The process of adding GPI to proteins is a complex 12 step pathway carried out by several essential enzymes. Yeast comprises about 60 GPI-anchored proteins with various functions, most of them involved in cell wall biosynthesis and stabilization (Pittet and Conzelmann 2007). The first step of GPI synthesis is the addition of Nacetylglucosamine (GlcNAc) to PI to yield GlcN-acyl-PI by an enzyme complex at the cytosolic surface of the ER. Next, the core carbohydrate structure of GPIs is formed by successive addition of mannose (Man) and phosphorylethanolamine (EtNH2-P). The core structure is highly conserved among eukaryotes and synthesis follows the same basic rules. GPIs usually contain four mannoses (Man1-4) and an optional Man5. They are linked to EtNH2-P on Man1, 2 and 3. The inositol is linked to two alternative lipid moieties, diacylglycerol (DAG) or ceramide that can be found on mature GPI anchored proteins.
While DAG classically consists of C16 to C18 FAs usually containing at least one monounsaturated fatty acyl (MUFA) chain (Ganesan et al. 2015), DAGs attached to proteins as part of a GPI anchor possess a C26:0 in sn-2. However, the lipid moieties of mature GPI-anchors of yeast contain a modified DAG possessing a C26:0 in sn-2 (Pittet and Conzelmann 2007). This process lacks mechanistic insights. The removal of the lipid moiety FA and attachment of the new VLCFA could be connected to the GPI-remodelase Gup1 (Sipos et al. 1997). In both anchor lipids, the C26 FA can be hydroxylated. However, ceramide harboring an OH-C26 are only introduced in the Golgi apparatus and not in the ER. After their biosynthesis the GPI-anchored proteins leave the ER in COPII coated vesicles and travel to the PM via the Golgi (Pittet and Conzelmann 2007). GPI anchored proteins, covalently linked to the cell wall frequently lose their lipid motif at a certain maturation stage. Cell wall attachment occurs via the remaining mannose residues of the anchor to β1,6-glucans of the cell wall (Yin et al. 2005). The role of the VLCFA for GPI containing proteins is still unclear. Gup1 the enzymes essential for the synthesis of C26:0-containing DAG anchors, harbors a membrane bound O-acyl-transferases (MBOAT) motif and is conserved among fungi and protozoa. gup1Δ mutants produce normal amounts of GPI proteins but most mature GPI anchors contain PI with C16 and C18 FAs or lyso-PI. In addition, ceramide incorporation into GPI anchors is disturbed. This affects the transport of GPI proteins from the ER to the Golgi. Overall gup1Δ cells have fragile cell walls and defects in bipolar bud site selection, with the latter being more likely connected to another function of the Gup1 protein (Bosson et al. 2006). Additionally, ceramide synthase mutants that are not able to attach VLCFAs to LCBs show severely altered lipid remodeling of GPI-anchored proteins (Pettus et al. 2002). Otherwise the VLCFA moiety is most likely important for the correct transport of the GPI anchor protein to their destination since there are several lines of evidence connecting VLCFAs with vesicular transport. However, the molecular role of VLCFAs in GPI anchored proteins is not fully understood (David et al. 1998;Obara et al. 2013).

VLCFAs as determining factors for membrane properties
Besides their role in lipid biosynthesis and in GPI anchoring of proteins, VLCFAs have been implicated in multiple intracellular pathways based on their specific biophysical properties. Saturation and length of the yeast VLCFAs make them unique in spanning more than one leaflet of membranes. Therefore, VLCFA cause very specific local membrane phenomena. Several studies link VLCFAs to highly curved membrane structures (David et al. 1998;Obara et al. 2013;Schneiter et al., 1996Schneiter et al., , 2004). This appears to be especially important in the endo-lysosomal pathway. For example, VPS21 was identified as a multi-copy suppressor of VLCFA biosynthesis deficient mutants. Vps21 is a Rab GTPase (Rab5 homolog) at the endosome required for endocytic transport and sorting of vacuolar hydrolases (Horazdovsky et al. 1994;Robinson et al., 1988aRobinson et al., , 1988bSinger-Krüger et al. 1994). It is also required for endosomal localization of the class C core vacuole/endosome tethering (CORVET) complex (Cabrera et al. 2013;Langemeyer et al. 2020;Peplowska et al. 2007). CORVET is the tether involved in fusion of vesicles to the late endosome/ multivesicular body (MVB). Double mutants of ELO3 and VPS21, or the CORVET subunit VPS3 show synthetic lethal phenotypes (Obara et al. 2013). In addition, yeast cells lacking the elongase genes ELO2 or ELO3 display morphological and vacuolar defects (Chung et al. 2003;Hurst et al. 2020) .
Other factors required for fusion of vesicles with their target membrane are soluble NSF attachment protein receptors (SNAREs). Snc1 and Snc2 are v-SNARES involved in the fusion of Golgi-derived secretory vesicles with the plasma membrane. Interestingly, the temperaturesensitive growth phenotype of cells with SNC1 and SNC2 gene deletions can be suppressed by loss of function mutations in ELO2 or ELO3 (David et al. 1998). These observations indicate a role for VLCFAs in vesicular transport that cannot be substituted by long chain FAs. However, this does not explain the role of VLCFAs on a molecular level. Possibly, VLCFAs exert their essential function by supporting the highly curved membranes involved in vesicular processes. Budding and fusion of vesicles requires local generation of highly curved membranes either at the neck of the budding vesicle or at the contact site of lipid mixing during fusion. Because of their high chain length VLCFA can span both leaflets of a lipid bilayer, enabling support of highly curved membranes. Invagination for intraluminal vesicles formation at the late endosome/MVB also requires highly curved membranes, possibly explaining VLCFAs involvement in this process. Highly curved membranes might be destabilized during elevated temperatures because of increased fluidity. Along that line, the requirement for VLCFAs is increased at higher temperatures (Obara et al. 2013). Free VLCFAs are very low abundant in yeast cells. VLCFAs are mainly incorporated into SLs. Therefore, the functions described above are most likely dependent on VLCFA-containing SLs. MIPC synthesis defective mutants csg1Δ, and csh1Δ, in combination with vps21Δ show growth defects at higher temperatures and impairment in vesicular transport pathways. Therefore, other parts of SLs such as the sugar head groups may also facilitate clustering of SL and therefore stabilize highly curved membranes in combination with VLCFAs (Obara et al. 2013). It is thus not surprising that SL homeostasis has been linked to endosomal transport pathways (Fröhlich et al. 2015;Voynova et al. 2015).
Additionally, VLCFAs have also been proposed to support highly curved membranes at (NPCs). The NPC spans the outer and inner membrane of the nucleus in eukaryotic cells. It is a giant protein-aqueous channel that serves as a gatekeeper by forming the site for entry and exit from the nucleus. It is anchored to the nuclear envelope by integral membrane proteins. At the NPC the nuclear membranes that are continuous with each other form a 180°turn (Wente 2000;Worman and Courvalin 2000). The connection between VLCFA and the structural integrity of the nuclear envelope is based on the observation that yeast acetyl-CoA carboxylase mutants unable to synthesize malonyl-CoA display severe perturbation of the nuclear envelope. This phenotype is independent of de novo synthesis of FAs and a structural role for VLCFA in stabilizing the highly curved membrane at the NPC was postulated (Schneiter et al. 1996). PI containing VLCFAs present in the nuclear membrane are required for maintaining the structure and function of the yeast nuclear membrane. The absence of this rare species of VLCFA-containing glycerophospholipid might cause the alterations of the nuclear membrane in the acc1/mtr7 mutant cells unable to synthesize VLCFA-containing PI (Schneiter et al. 2004) Finally, a recent publication has linked yeast VLCFA biosynthesis to ER membrane stress and induction of the unfolded protein response (UPR). Deletion of the VLCFA coenzyme-A synthase Fat1 results in an increase in saturated membrane lipids in general (see below). Together these results further support the essential role of VLCFAs in membrane homeostasis (Micoogullari et al. 2020).

Yeast cells control VLCFA levels
VLCFAs are building blocks for lipid biosynthesis and affect biophysical membrane properties. Therefore, it is not surprising that cells tightly control VLCFA levels in membranes. The levels of VLCFAs are the result of several parallel pathways namely the activation, biosynthesis, regulation, transport, storage and degradation of VLCFAs (Figure 3).
VLCFAs have to be activated by linkage to coenzyme A (CoA). Acyl-CoA derivatives are available for incorporation into complex lipids, acylation of proteins and degradation for energy production via β-oxidation (Watkins et al. 1998). Therefore, the regulation of acyl-CoA synthesis is a key factor in lipid homeostasis. The enzymes catalyzing this process are called acyl-CoA synthetases (ACSs). ACSs are an enzymatic family that catalyze the thio-esterification of FAs with CoA (Semenkovich 1997;Watkins 1997). This twostep process is energy dependent and requires ATP and CoA. In the first step an acyl-AMP intermediate is formed by transferring an AMP from ATP to a FA. This step is driven by hydrolysis of inorganic pyrophosphate (PPi) to inorganic phosphate (2Pi). In the second step acyl-AMP reacts with CoA to form an acyl-CoA molecule and AMP is released (Hisanaga et al. 2004). Special structural features of the ACS binding pockets allow for unique activity patterns. This allows a level of flexibility to support lipid related homeostasis during cellular stress or fluctuations in nutrient availability and during various points in the cell cycle . S. cerevisiae has six known ACSs. They can be identified by two signature motifs, the AMP signature and the ACS signature. All of them have varying expression patterns, localization, and unique features in metabolic and regulatory pathways. They are  s1 1 1 1 1 1 Tsc13  1 1 1 1 1 1 1 1 1 1  found in mitochondria, peroxisomes and the ER (Johnson et al. 1994;Watkins 1997). Faa1 through Faa4 are required for activation of long chain FAs . Pcs60/Fat2 is an oxalyl-CoA synthetase (Foster and Nakata 2014). VLCFAs are activated by the Fatty acid transporter 1 (Fat1). Yeast Fat1 was identified as a FA transporter based on structural and functional similarities to the murine Fatty Acid Transport Protein 1, FATP1 (Faergeman et al. 1997). Fat1 was shown to have VLCFA-CoA synthetase activity and the same amino acid sequence motif as other ACS including a FA binding region with similar features to other VLCFA activating enzyme motifs (Johnson et al. 1994;Watkins et al. 1998). Phenylalanine 528 and leucine 669 have been identified as essential for transport activity while serine 258 and glutamate 508 are essential residues for VLCFA activation . Fat1 is highly specific for activation of FAs longer then C20. Deletion of the FAT1 gene reduces activation of C24 FAs ∼70% and leads to accumulation of VLCFAs (Choi and Martin 1999). Interestingly, Fat1 has been described as a transmembrane protein of the plasma membrane (Obermeyer et al. 2007). However, several studies have identified Fat1 as a lipid droplet protein (Athenstaedt et al. 1999;Currie et al. 2014) which is in line with our own unpublished observations. This localization would argue against a direct role in FA transport across the plasma membrane and would exclude transmembrane domains, since these are impossible to incorporate into the monolayer of lipid droplets (Kory et al. 2016).

VLCFA elongation is controlled by phosphorylation of the rate limiting enzyme
The Elo proteins catalyze the rate limiting step in VLCFA synthesis and allow for structural diversity among FA classes. So far, little is known about the regulation of VLCFA biosynthetic enzymes in yeast. The regulation of Elo2 activity appears to be tightly linked to the regulation of SPT dependent LCB synthesis. Since LCBs and VLCFAs are both substrates for ceramide synthesis it is an attractive model to also regulate VLCFA biosynthesis according to need. When SL levels at the plasma membrane are low, the target of rapamycin complex 2 (TORC2) is activated and phosphorylates the yeast Akt homologue Ypk1 (Berchtold et al. 2012;Roelants et al. 2011). This leads to phosphorylation of the Orm proteins, negative regulators of the SPT, resulting in release of the inhibition and increased SL biosynthesis (Breslow et al. 2010;Gururaj et al. 2013;Han et al. 2010;Liu et al. 2012;Sun et al. 2012) . Similarly, Elo2 regulation also responds to changes in the PM. Its phosphorylation is controlled by the plasma membrane localized Rho1 guanine nucleotide exchange factor (GEF) Rom2. Here, depletion of SL levels leads to down-regulation of Rom2-signaling and de-phosphorylation of Elo2, which in turn increases VLCFA synthesis. The regulation of TORC2 and Rom2 appear to be reciprocal. Regulation of LCBs and VLCFAs are both regulated by plasma membrane SL levels. The exact signaling cascade downstream of Rom2 remains largely elusive, but appears to involve only the initial kinases of the upstream cell wall integrity pathway. Mck1, a GSK3-like kinase is responsible for Elo2 phosphorylation. Rom2 signal is potentially transduced via Rho GTP that targets TORC1, a negative regulator of Mck1 (Lee et al. 2012). Downregulation of TORC1 therefore leads to increased Elo2 phosphorylation. Depending on different models this might activate or deactivate VLCFA synthesis (Olson et al. 2015a;Zimmermann et al. 2013). TORC1 also controls Orm inhibition and therefore SL synthesis, again supporting the concept that the coordination of VLCFA and LCB regulation is coupled (Shimobayashi et al. 2013). TORC1 downregulation is also linked to the activation of cellular autophagy under conditions of low nutrient availability (He and Klionsky 2009). Increased VLCFA and LCB synthesis might help to provide the cell with enough membrane building blocks for autophagosome formation. VLCFA depletion also seems to constitutively activate autophagy via long chain base phosphorylation resulting in decreased survival (Zimmermann et al. 2013). The connection between all these processes is still unclear and the molecular mechanism of VLCFA regulation still needs to be determined. However, VLCFA regulation is essential for cellular survival with important implications for autophagy and cellular homeostasis (Micoogullari et al. 2020;Zimmermann et al. 2013).

VLCFA are transported between organelles
Besides VLCFA activation sufficient transport is a key factor in VLCFA homeostasis. Yeast does not import FAs greater than C20 in length. Synthesis and further metabolic consumption of VLCFAs both take place in the ER and therefore, there is little need for VLCFA transporters in the cell. Proteins delivering the acyl-CoA primers to the ER for FA elongation in yeast are the acyl-CoA binding proteins (ACBPs). ACBPs mediate cytosolic transport of acyl-CoAs and regulate intracellular acyl-CoA levels (Hiltunen et al. 2003). VLCFA transport between membranes and across membranes is essential for energy generation via β-oxidation in peroxisomes. The enzymes required for VLCFA translocation into peroxisomes are ATP-binding cassette (ABC) 2 transporters (Van Roermund et al. 2012). ABC2 transporters have at least four domains. Two substrate binding trans-membrane domains are necessary for acyl-CoA translocation and two nucleotide binding domains provide the required energy by ATP hydrolysis (Linton and Higgins 2007). Yeast has two half-size ABC proteins: the peroxisomal ABC transporter 1 (Pxa1) and Pxa2. They form a heterodimer to build a functional transporter (Roermund et al. 2008). The Pxa1-Pxa2 complex functions as an acyl-CoA flippase. While the hydrophobic part of the FA diffuses in the membrane Pxa1-Pxa2 flip the polar CoA head from the cytoplasmic leaflet to the luminal leaflet of the peroxisomal membrane (Hettema and Tabak 2000). Single PXA deletion mutants exhibit reduced β-oxidation and reduced growth on oleate containing medium. For the even stronger pxa1Δ and pxa2Δ double mutant phenotype a full rescue of mainly C22:0 and other VLCFA β-oxidation can be observed by expressing human ABCD2. C24:0 and C26:0 β-oxidation can be fully restored by ABCD1 expression (Van Roermund et al. 2011). It was suggested that VLCFA-CoAs are hydrolyzed by the ABC transporter during transport. The free CoA is released into the cytosol and the acyl chain is subsequently translocated into the peroxisome. In peroxisomes, VLCFAs get reesterified by peroxisomal Fat1 (Van Roermund et al. 2012). However, this model is again in contrast to the described lipid droplet localization of Fat1. Alternatively, Fat1 could be directly involved in VLCFA transport into the peroxisome interacting with Pxa1-Pxa2 ( Van Roermund et al. 2012). This raises the interesting possibility of VLCFA handover from lipid droplets to peroxisomes, potentially via a membrane contact site (Shai et al. 2016). In this model, peroxisomal accumulation of CoA can be easily prevented but also the requirement for a separate process of CoA import into the peroxisome would be necessary (Van Roermund et al. 2012).

VLCFAs are degraded in peroxisomes
Finally, intracellular VLCFA levels are also dependent on their degradation. In S. cerevisiae VLCFAs are exclusively degraded via β-oxidation in peroxisomes. This is in contrast to higher eukaryotes where VLCFA β-oxidation also takes place in mitochondria (Kunau et al. 1995). Yeast can utilize FAs as an energy source. Interestingly, C24:0 and C26:0 FAs cannot be used as a sole carbon source with C22:6 being the only VLCFA meeting this requirement in yeast ( Van Roermund et al. 2003). FA and VLCFA β-oxidation are mediated by three key enzymes Pox1/Fox1, Mfe2/Fox2, and Pot1/Fox3. These enzymes mediate the four-step process of FA degradation into acetyl-CoA ( Figure 4). First, dehydrogenation of the activated FA introducing a double bound between the α und β carbon atom, catalyzed by Fox1. Fox 1 is the only acyl-CoA oxidase in yeast and has no substrate specificity. Therefore, it catalyzes the dehydrogenation of shorter FA as well as VLCFAs. The trans-2-enoyl-CoA intermediate is hydrated to 3-hydroxyacyl-CoA followed by another step of dehydrogenation yielding 3-ketoacyl-CoA. Both reactions are catalyzed by the multifunctional enzyme Fox2 that possesses 2-enoyl-CoA hydratase activity as well as (3R)-specific 3-hydroxy acyl-CoA dehydrogenase activity. Finally, the 3-ketoacyl-CoA thiolase Fox3 catalyzes the cleavage of 3-ketoacyl-CoA esters into a C2-shortened acyl-CoA and acetyl-CoA. The acetyl CoA is further oxidized in the tricarboxylic acid cycle providing a major source of metabolic energy. In yeast, the remaining FA is entirely degraded in the peroxisome (Hiltunen et al. 2003;Van Roermund et al. 2003).

VLCFA metabolism is connected to human pathophysiology
While this review focuses on the metabolism of VLCFAs in the model organism yeast we briefly want to highlight the role of VLCFAs in human pathologies. A number of human diseases are caused by defective VLCFA metabolism (Moser and Moser 1996). X-linked adrenoleukodystrophy (X-ALD) is characterized by an accumulation of free VLCFAs. It is caused by loss-of-function mutations in the ABCD1 transporter that imports VLCFAs into peroxisomes. These defects result in reduced peroxisomal degradation of VLCFAs and a concomitant increase in elongation (Kemp et al. 2005). VLCFAs accumulate in plasma and tissue of patients, resulting in a fatal neurodegenerative phenotype that can include childhood-onset cerebral adrenoleukodystrophy (CCALD) and adrenomyeloneuropathy (AMN) (Hama et al. 2020). AMN is the milder phenotype and is characterized by a slowly progressive axonopathy. Besides mutations in ABCD1 additional genetic or environmental factors are required to trigger brain inflammation. The phenotypic expression of the disease can vary extremely between patients. The correlation between genotype and different phenotypes remains unclear and does not allow to link severity of the pathology of X-ALD to any particular cellular mis-regulation. Stronger mutations can result in milder phenotypes (Wiesinger et al. 2015). Accumulating VLCFAs comprise C24:0, C26:0 and C26:1. Hexacosenoyl (26:1)-CoA is the most abundant VLCFA-CoA species in ABCD1-KO HeLa cells and X-ALD fibroblasts. C26:1-CoA levels are about fiveto six-fold higher than 26:0-CoA but the 26:1 and 26:0 VLCFA levels are similar. Even without ABCD1 there is still a turnover of VLCFAs indicating an ABCD1 independent machinery that metabolizes VLCFA-CoA (Hama et al. 2020). Interestingly, deletion of the VLCFA-CoA-synthetase FAT1 in S. cerevisiae results in decreased peroxisomal β-oxidation and elevated cellular VLCFA levels mimicking an X-ALD like phenotype.
Mutations in other members of the ABCD transporter family are also linked to disease. ABCD5 deficiency causes retinal dystrophy with leukodystrophy (RDLKD) disease. RDLKD patients exhibit developmental delay as well as ataxia and spastic paraparesis. ACBD5 deficiency leads to accumulation of VLCFAs due to impaired peroxisomal β-oxidation. Peroxisomal β-oxidation activity is reduced for VLCFAs but not for FAs of a different length (Abu-Safieh et al. 2013;Ferdinandusse et al. 2017) Mutations in the VLCFA elongases have also been identified to cause different diseases. Mutation of ELOVL1 causes a disorder characterized by hypo-myelination, facial dysmorphia, and ichthyosis (Mueller et al. 2019). Mutations in ELOVL4 or ELOVL5 lead to spinocerebellar ataxia. ELOVL5 is the human orthologue of Elo2 also containing the highly conserved phosphorylation-site that is important for Elo2 regulation (Di Gregorio et al. 2014;Hayashi et al. 2020;Olson et al. 2015a).

Future perspectives
Although the basic enzymology involved in VLCFA elongation is well worked out, many questions related to the regulation of this process, the regulation of the enzymatic machinery, the coordination with cellular lipid metabolism, and the impact of VLCFA metabolism on cellular physiology are still unresolved. What is the exact localization of the VLCFA biosynthetic machinery in cells and how does this affect activity? How are VLCFAs distributed in particular lipids and proteins and how does this affect lipid metabolism and membrane structure? Which organelles are involved in VLCFA homeostasis besides the ER? How is VLCFA synthesis linked to biological processes such as secretion and endocytosis? For a long time, the complexity of FAs and lipid species was a major bottleneck in studying lipids. However, thanks to the important work of many groups we have overcome this limitation and mass spectrometry-based methods are easily available to study lipid and FA composition. And these methods are continuously developed and even the flux of metabolites through lipid metabolism can be analyzed in high throughput.
Another emerging technique is the determination of membrane protein complex structures by cryo-electron tomography (Gold et al. 2014). More and more complex membrane protein structures have become available in recent years which help to understand the molecular mechanisms of trans-membrane proteins. In the future, it will be interesting to see which proteins form the elongation complex for VLCFA biosynthesis. If this complex can be purified and analyzed by cryo-EM this might finally reveal the molecular mechanism of VLCFA elongation.
The molecular function of the VLCFA-CoA synthetase will also be an interesting question for the future. How does the protein localize to lipid droplets and which processes involving VLCFA are taking place here? Are lipid droplets important organelles in VLCFA metabolism and are Overview of the four-step β-oxidation cycle in S. cerevisiae. Enzymes and co-factors are illustrated for each step. Details about the four stepsdehydrogenation, hydration, dehydrogenation and thiolysisand enzymes are described in the text. NAD + , nicotinamide adenine dinucleotide; CoA, coenzyme A.
membrane contact sites between lipid droplets and other organelles important? Similarly, how does the localization of Tsc13 at the nuclear vacuole junction affect its activity and VLCFA biosynthesis? S. cerevisiae has been an important model organism to understand the mechanistic details of VLCFA elongation and it will continue to do so. Elucidating the molecular mechanisms of VLCFA biosynthesis, regulation and degradation in yeast will help us to understand the complex role of VLCFAsin human pathophysiology.