Lorenzo’s oil inhibits ELOVL1 and lowers the level of sphingomyelin with a saturated very long-chain fatty acid

X-linked adrenoleukodystrophy (X-ALD) is a peroxisomal disorder caused by impaired degradation of very long-chain fatty acids (VLCFAs) due to mutations in the ABCD1 gene responsible for VLCFA transport into peroxisomes. Lorenzo’s oil, a 4:1 mixture of glyceryl trioleate and glyceryl trierucate, has been used to reduce the saturated VLCFA level in the plasma of X-ALD patients; however, the mechanism by which this occurs remains elusive. We report the biochemical characterization of Lorenzo’s oil activity toward ELOVL1, the primary enzyme responsible for the synthesis of saturated and monounsaturated VLCFAs. Oleic and erucic acids inhibited ELOVL1 and, moreover, their 4:1 mixture (the fatty acid (FA) composition of Lorenzo’s oil) exhibited the most potent inhibitory activity. The kinetics analysis revealed that this was a mixed (not a competitive) inhibition. At the cellular level, treatment with the 4:1 mixture reduced the level of sphingomyelin (SM) with a saturated VLCFA accompanied by an increased level of SM with a monounsaturated VLCFA, probably due to the incorporation of erucic acid into the FA elongation cycle. These results suggest that inhibition of ELOVL1 may be an underlying mechanism by which Lorenzo’s oil exerts its action.


INTRODUCTION
Fatty acids (FAs) are the major constituents of cellular lipids and are highly diverse in their chain length and degree of saturation. This structural diversity underlies the various functions of FAs. The most abundant cellular FAs are long-chain fatty acids (LCFAs) having a chain length of 11 to 20 carbons (C11-C20). FAs with >C20 are called very long-chain fatty acids (VLCFAs) and have essential physiological functions that cannot be compensated for by LCFAs, such as skin barrier formation, retinal function, myelin maintenance, and spermatogenesis (1)(2)(3)(4)(5). VLCFAs are components of the two important lipid classes, glycerolipids and sphingolipids, which are characterized by their lipid backbone that in the former is a glycerol and in the latter a sphingoid base.
Although the sphingoid base is structurally similar to the glycerol with a FA at the sn-1 position, VLCFAs linked to these backbones are strikingly different: those of glycerolipids are mostly polyunsaturated, whereas those of sphingolipids are saturated and monounsaturated (especially C24). In mammals, a polar head group, phosphocholine or sugar, is attached to the N-acylated sphingoid base (ceramide) forming sphingomyelin (SM) or glycosphingolipid, respectively. Sphingolipids containing VLCFAs have unique cellular functions such as microdomain/raft-mediated signal transduction and cell survival (6)(7)(8). Ceramides with !C26 VLCFAs are also present in the skin and essential for epidermal permeability barrier function (3,9,10).
X-linked adrenoleukodystrophy (X-ALD) is the most common peroxisomal disorder characterized by progressive demyelination and adrenal insufficiency (11,12).
The gene mutated in X-ALD is the ABCD1 gene (13) that encodes a peroxisomal ATP-binding cassette (ABC) protein responsible for transporting VLCFAs (as VLCFA-CoAs) into peroxisomes (14,15) where these FAs are broken down by !-oxidation (16). The defective ABCD1 protein impairs this normal VLCFA degradation process, resulting in the accumulation of VLCFAs, specifically C24:0 and C26:0, in the plasma and tissues (11,17). The accumulation of these VLCFAs is believed to play a crucial role in the pathogenesis of X-ALD such as inflammatory demyelination and oxidative damage (18,19). Therefore, reducing or preventing the accumulation of VLCFAs by either promoting their degradation or suppressing their synthesis may lead to the treatment of X-ALD.
VLCFAs are synthesized by the FA elongation cycle in the endoplasmic reticulum by sequential addition of C2 units from malonyl-CoA to long-chain acyl-CoA (5,20).
Each elongation cycle is composed of four reactions: condensation, reduction, dehydration, and reduction. The first condensation reaction is the rate-limiting step catalyzed by ELOVL family proteins. There are seven known ELOVL isozymes (ELOVL1-7) in mammals (5,20), each of which exhibits different specificity for chain length and degree of saturation of the acyl-CoA substrate (7). Among the seven isozymes, ELOVL1 is highly active toward saturated and monounsaturated C20-to C24-CoAs (7) and thus responsible for the synthesis of C22-to C26-VLCFAs such as those accumulated in X-ALD. Knockdown of ELOVL1 in X-ALD fibroblasts is reported to lower the level of C26:0 (21), suggesting ELOVL1 as a potential pharmacological target for the treatment of X-ALD (17,21,22). Lorenzo's oil, a 4:1 mixture of glyceryl trioleate (C18:1 n-9) and glyceryl trierucate (C22:1 n-9), was introduced in 1989 as a dietary treatment for X-ALD patients (23). After oral administration, the triglycerides are hydrolyzed by lipases in the digestive tract to liberate oleic and erucic acids, which are then absorbed from the intestine and transported via lymph and blood to the tissues where they exert their effects. Lorenzo's oil normalizes the levels of C24:0 and C26:0 in the plasma of X-ALD patients (23,24); however, it does not alter the clinical progression of patients with preexisting neurological or adrenal dysfunction (24,25) but may have a preventive effect in asymptomatic patients (26).
The development of Lorenzo's oil goes back to 1986, when oleic and erucic acids were found to lower the level of C26:0 in X-ALD fibroblasts (27). Since then, despite much research, the mechanism of action of Lorenzo's oil has remained ambiguous. Some possible mechanisms include: (1) prior conversion of oleic and erucic acids to their CoA esters followed by competitive inhibition of ELOVL isozymes, (2) direct inhibition of VLCFA synthesis by oleic and erucic acids themselves, (3) enhanced degradation/removal of saturated VLCFAs induced by oleic and erucic acids and/or their metabolites, and (4) alteration of lipid homeostasis mediated by oleic and erucic acids, followed by induction of various cellular responses including metabolic changes and gene expression (28). In the present study, we biochemically characterized the activity of Lorenzo's oil toward ELOVL1 using in vitro and cellular FA elongation assays to understand how Lorenzo's oil reduces saturated VLCFAs.

Production of HeLa stable transformants expressing 3xFLAG-ELOVL1
HeLa cells transfected with the pCE-puro 3xFLAG-ELOVL1 plasmid were subjected to selection with puromycin at 10 µg/ml. Among the isolated clones, clone #40 (HeLa-ELOVL1) expressed the highest level of 3xFLAG-ELOVL1 protein and was used for the experiment.

In vitro FA elongation assay
The assay was performed using the total membrane fraction as described previously (7).
Briefly, the total membrane fraction was incubated with C22:0-CoA and 0.075 µCi

Analysis of FA incorporation into HeLa cells
The [   BAS-2500 as described previously (7).

Immunoblotting
Immunoblotting was performed as described previously (7)

Oleic acid inhibits ELOVL1 activity in vitro
Of the seven mammalian ELOVLs, ELOVL1 is primarily responsible for the synthesis of C22-to C26-VLCFAs (7). We therefore generated HeLa cells transiently or stably expressing 3xFLAG-tagged ELOVL1 (3xFLAG-ELOVL1), thus overproducing ELOVL1. To confirm their increased ELOVL1 activities, we also generated HeLa cells expressing either vector control or 3xFLAG-tagged ELOVL1 mutant (negative control) in which the conserved HXXHH motif for ELOVL family members (where XX represents VF for ELOVL1) was replaced with HVFAA (i.e., 3xFLAG-ELOVL1(AA)).
The conserved motif is thought to be located in the active site, and its mutations in the yeast ELOVL homolog Fen1 and in ELOVL7 result in the loss of their activities (29,30). Both the mutant 3xFLAG-ELOVL1(AA) and the wild-type 3xFLAG-ELOVL1 proteins were expressed at a comparative level (Fig. 1A). The total membrane fractions were prepared from these cells and examined for their ELOVL1 activities in the in vitro FA elongation assay with [2-14 C]malonyl-CoA as the C2 donor and C22:0-CoA as the substrate. The reaction products were analyzed by TLC-autoradiography after alkaline hydrolysis to the corresponding FAs (Fig. 1B). As expected, the membrane fractions from those HeLa cells overproducing ELOVL1, but not from the cells expressing ELOVL1(AA), significantly increased ELOVL1 activity as compared to the vector control cells. Therefore, the total membrane fraction from HeLa cells stably overproducing ELOVL1 (HeLa-ELOVL1) was used in the following FA elongation assays, except where indicated, in which case the total membrane fraction from HEK 293T cells transiently overexpressing ELOVL1 (7) was used.
We first investigated the effect of FAs with different chain length and degree of saturation on ELOVL1 activity. Thus, the in vitro FA elongation assay was performed in the presence of oleic acid (C18:1, a major component of Lorenzo's oil), DHA (C22:6), EPA (C20:5), or ethanol (vehicle control). Oleic acid was found to inhibit ELOVL1 activity in a dose-dependent manner with ~60% inhibition at 20 µM, while neither DHA nor EPA exhibited any inhibition even at 20 µM ( Fig. 1C and D). The saturated analog of oleic acid, stearic acid (C18:0), had no effect on ELOVL1 activity even at 100 µM (Fig. S1). The longer-chain saturated FAs, including arachidic acid (C20:0) and behenic acid (C22:0), could not be evaluated due to their poor solubility in the assay medium. We analyzed these data by using a Lineweaver-Burk plot to determine the kinetic parameters (K m and V max ) of ELOVL1 toward C22:0-CoA. To exclude any non-specific effects seen at the higher concentration of C22:0-CoA, we used the data obtained at 1, 2, and 3 µM C22:0-CoA, yielding a linear increase of the ELOVL1 activity (Fig. 3B). The

Erucic acid is preferentially metabolized to sphingolipids
We next investigated the metabolism of oleic and erucic acids using HeLa cells.
The cells were incubated with 14 C-labeled oleic and erucic acids as well as with 14 C-labeled palmitic, stearic, arachidic, and behenic acids for comparison. Similar to to a much lesser extent to sphingolipids (hexosylceramide (HexCer) and sphingomyelin (SM)) (Fig. 4A) (7).

Lorenzo's oil decreases the cellular level of SM with a saturated VLCFA
Since in eukaryotic cells saturated and monounsaturated VLCFAs are predominantly incorporated into sphingolipids, we suspected that inhibition of ELOVL1 responsible for VLCFA synthesis would generate a characteristic sphingolipid profile. significantly. At 20 µM, a significant increase was also detected for C26:1-SM. As a consequence, the ratio of monounsaturated to saturated SMs more than doubled at 20 µM of the 4:1 mixture compared to the untreated control (Fig. 5B). In contrast, treatment with either palmitic (C16:0) or stearic acid (C18:0) at 20 µM had almost no effect on the SM profile except for a slightly increased level of saturated SM (C18:0and C20:0-SMs or C18:0-, C20:0-, and C22:0-SMs, respectively) (Fig. S2). This is consistent with the aforementioned findings that stearic acid does not inhibit ELOVL1 activity (Fig. S1) and that palmitic and stearic acids are metabolized mainly to glycerophospholipids rather than to sphingolipids (Fig. 4).

Lorenzo's oil does not affect the expression level of ELOVL1 mRNA
We also analyzed the endogenous ELOVL1 mRNA level by RT-PCR in HeLa cells treated with a 4:1 mixture of oleic and erucic acids for 6 days and found no appreciable changes caused by the treatment (Fig. S3). The results suggest that Lorenzo's oil reduces the synthesis of saturated VLCFAs by inhibiting ELOVL1 at the protein level but not at the mRNA level.

DISCUSSION
It has been more than two decades since Lorenzo's oil was reported to lower the C26:0 level in X-ALD fibroblast and plasma (23,24,27); however, how Lorenzo's oil exerts such effects has remained largely speculative. The results of our present study elucidate for the first time the link between Lorenzo's oil and the reduction of the saturated VLCFA level.
We demonstrated that a 4:1 mixture of oleic and erucic acids potently inhibits the activity of ELOVL1 (Fig. 2) by a mixed inhibition mechanism, rather than by competitive inhibition (Fig. 3); however, at present it remains unclear how these monounsaturated FAs interact with ELOVL1. We can only speculate that since the ELOVL family members are multi-pass transmembrane proteins (5,20), oleic and erucic acids may interact with ELOVL1 by intercalating between the adjacent membrane-spanning regions of ELOVL1, thereby preventing the conformational change required for its catalytic activity. This may also explain our finding that monounsaturated oleic and erucic acids but not polyunsaturated DHA and EPA inhibited ELOVL1 activity ( Fig. 1C and D), since the more cis-double bonds in the FA chain increase its rigidity and space requirement due to the bent geometry of the cis-double bond and thus limit such lateral interactions. Lorenzo's oil appears to have the optimum ratio of oleic and erucic acids (i.e., 4:1) for ELOVL1 inhibition.
We also demonstrated that cells treated with a 4:1 mixture of oleic and erucic acids reduces the level of SM with either saturated C22-or C24-VLCFA (Fig. 5A) as a result of the inhibition of ELOVL1 activity, not transcriptional regulation of the ELOVL1 gene (Fig. S3).
In the cell-based assay, the reduction of C22:0-and C24:0-SM was accompanied by an increase in the level of C24:1-SM (Fig. 5A). This increase probably results from the conversion of some of oleic and erucic acids to the corresponding acyl-CoAs  . 4A and B). The preferential incorporation of [ 14 C]erucic/nervonic acid, but not [ 14 C]oleic acid, to sphingolipids seems to be consistent with the fact that saturated and monounsaturated VLCFAs are predominantly found in sphingolipids (10,31). This finding may help explain the increased levels of monounsaturated VLCFAs (C22:1 and C24:1) seen in the plasma of patients treated with Lorenzo's oil (23).
Moreover, the finding raises a possibility that the increased C22:1-CoA could compete with the saturated acyl-CoAs for the ELOVL1 active site, thereby contributing to the attenuation of saturated VLCFA synthesis; however, such a possibility is not likely for oleic acid since C18:1-CoA is not a substrate for ELOVL1 (7). Consequently, Lorenzo's oil may lower the pathogenic C24:0 and C26:0 levels by a combination of two mechanisms: direct ELOVL1 inhibition by oleic and erucic acids (mixed inhibition) and indirect ELOVL1 inhibition by C22:1-CoA (competitive inhibition).

20
All together, our findings seem to indicate that the beneficial effect of Lorenzo's oil is at least in part a result of its ELOVL1 inhibition. Although Lorenzo's oil normalizes plasma C24:0 and C26:0 levels in X-ALD patients, it does not arrest or ameliorate their neurological symptoms (23,24). Moreover, the treatment increases their plasma C22:1 and C24:1 levels (23), the effect of which on the pathology of X-ALD is currently unknown. More work is necessary to better understand the etiology of X-ALD and to develop novel treatments.