Photoaffinity Labeling Reveals a Role for the Unusual Triply Acylated Phospholipid N-Acylphosphatidylethanolamine in Lactate Homeostasis

Abstract

Most eukaryotic membranes comprise phospholipids bearing two hydrophobic tails, but N-acylphosphatidylethanolamine (NAPE) stands out as a long-known but poorly understood phospholipid with three hydrophobic groups. What little attention that NAPE has received has been devoted to understanding its metabolic functions as a precursor to N-acylethanolamine (NAE), a bioactive lipid that acts as an endocannabinoid. Yet, levels of NAPE increase during myocardial infarction and ischemia, suggesting potential signaling roles for this lipid. Here, we exploit photoaffinity labeling (PAL) to identify NAPE-interacting proteins and elucidate signaling functions of NAPE. By positioning diazirine and alkyne groups in metabolically distinct regions of the NAPE molecule, we ensured that our PAL probe reported on interactions of NAPE and not NAE. Our studies identified several NAPE interactors, including two single-pass transmembrane proteins, CD147/Basigin and CD44, both of which serve as chaperones for monocarboxylate transporters (MCTs) from the SLC16A family that mediate lactate flux across the plasma membrane. Functional studies revealed that NAPE stimulates lactate efflux by MCTs dependent upon CD147 and CD44, establishing NAPE as a bona fide signaling lipid and pointing to potential physiological roles in metabolic and energy homeostasis that may be pathologically relevant in ischemia.

Graphical Abstract

INTRODUCTION

The glycerolipids that comprise eukaryotic membranes all bear two acyl tails, with two notable exceptions. One is the four-tailed lipid cardiolipin (and its immediate metabolic precursors and products), a pseudo-dimeric phospholipid that plays several critical roles in the inner mitochondrial membrane.¹ The other is a poorly understood triply acylated phospholipid, N-acylphosphatidylethanolamine (NAPE). First identified in 1965 as a minor component of wheat flour², NAPE remained obscure until it was found to accumulate upon myocardial infarction and ischemia in mammals^3-6, stimulating interest in its biosynthesis and metabolism.

NAPE is synthesized by N-acylation of the abundant membrane phospholipid phosphatidylethanolamine (PE) by several N-acyltransferases, including calcium-dependent N-acyltransferase (Ca-NAT), PLA2G4E, and calcium-independent phospholipase A and acyltransferases (PLAATs), which obtain the third acyl chain from phosphatidylcholine (PC).^7,8 Catabolism of NAPE occurs via several pathways, including head group cleavage catalyzed by NAPE-specific phospholipase D (NAPE-PLD)^9-11 or phospholipase C (PLC)¹², as well as deacylation catalyzed by ABHD4¹³ or PLA2s^4,14. Although these pathways produce different short-lived intermediates, all lead to formation of the fatty acid amide N-acylethanolamine (NAE).¹⁵

In fact, studies of NAPE have been primarily motivated by a desire to understand its role as a biosynthetic intermediate toward the ultimate formation of NAEs, which have numerous important physiological functions as endocannabinoids that activate the CB1 and CB2 receptors and thus impact diverse systems including inflammation, pain, appetite, and mood.^16-20 Because of this focus on NAPE as a precursor to NAE, any potential physiological roles of this lipid in its own right have been largely overlooked. A single report ascribing a physiological function to NAPE was the observation of decreased food intake and reduction in weight in mice upon ingestion of large amounts of NAPE.²¹ However, a later study reported that NAPE-PLD was required for NAPE to render such effects of appetite regulation, implying that NAE and not NAPE is likely the actual effector in this instance.²²

Yet this unusual, triply acylated lipid exists at low basal levels in numerous tissues, and its production is stimulated in certain pathological settings (e.g., myocardial infarction, ischemia). Motivated by a dearth of knowledge of biological functions for NAPE, we sought to begin to understand physiological roles of this lipid by elucidating protein effectors that recognize it and whose function it might modulate. To achieve these goals, we employed a photoaffinity labeling approach to elucidate the protein interaction partners of NAPE. The resultant chemoproteomics studies elucidated the first description of the NAPE interactome. Validation of several of these lipid–protein interactions and subsequent mechanistic analysis of the interaction of NAPE with two effectors, CD147/BSG and CD44, revealed a role for this lipid in modulating lactate transport, pointing to potential molecular mechanisms underlying observations of increased NAPE levels in ischemia.

RESULTS

Design and synthesis of a NAPE photoaffinity probe

Two functionalities are required for photoaffinity labeling (PAL): a photocrosslinking moiety to construct protein-ligand covalent bonds and a clickable moiety to biotinylate the protein-ligand complexes for subsequent enrichment. Many PAL methods employed diazirines for crosslinking and alkynes as clickable handles, and in many previous PAL studies of lipids, the same acyl chain was functionalized with both diazirine and alkyne functionalities.^23-27 However, because NAPE can undergo metabolism in several different ways, false discoveries may take place when a NAPE photoaffinity probe of similar design is used. However, the triply acylated nature of NAPE presents an opportunity to separate the diazirine and alkyne units into distinct portions of the molecule. Therefore, we elected to place the diazirine group on the N-acyl chain and alkynes on the sn-1 and sn-2 acyl tails, resulting in NAPE probe 1 (Scheme 1A). In this way, metabolism of 1 would largely produce lipid products that do not bear both a diazirine and an alkyne and thus not lead to false-positive hits.

Scheme 1.

Design and synthesis of the NAPE photoaffinity probe. (A) Three different possible metabolic pathways for degradation of NAPE probe 1 would produce an NAE analog containing diazirine but not an alkyne. (B) Synthetic route of NAPE photoaffinity probe 1.

Synthesis of 1 (Scheme 1B) began with diacylation of p-methoxybenzyl (PMB)-protected glycerol 2 with fatty acid 3 equipped with a terminal alkyne to give 1,2-diester 4,^28,29 which was subsequently deprotected with DDQ to yield 1,2-diacylglycerol (DAG) 5. Notably, PMB protection is key to successful preparation of 5, as using protecting groups that require acid or base for removal gave rise to undesired isomer 1,3-DAG as a major product through acyl shift following the deprotection, whereas DDQ deprotection afforded desired alcohol 5.^30-32 Subsequent conversion of 5 to phosphite intermediate 6 with phosphoramidite reagent followed by oxidation furnished phosphotriester 7.^33,34 Removal of the Boc protecting group revealed free amine 8, which was reacted with fatty acyl chloride 9 to yield 10.³⁵ Finally, base-mediated removal of the 2-cyanoethyl group produced NAPE probe 1.

Photoaffinity labeling identifies the NAPE interactome

Following the synthesis of 1, we first examined its stability in HeLa cells. We found that over a 2 h period, levels of 1 present in cells remained constant as assessed by LC–MS (Figure S1). We then assessed the ability of 1 to label proteins following UV irradiation and CuAAC tagging of the resultant lipid-protein conjugates (Figure 1A). We treated HeLa cells with 1 for varying amounts of time in the presence of LEI-401, a potent and specific NAPE-PLD inhibitor,²⁰ to minimize any potential intracellular degradation of 1 by NAPE-PLD, the most common pathway of NAPE metabolism. Subsequently, UV irradiation triggered lipid-protein covalent crosslinking, followed by CuAAC tagging of lipid-protein complexes with TAMRA-azide. Numerous TAMRA-labeled proteins were observed by in-gel fluorescence, and labeling intensity correlated with the concentration of 1 (Figure 1B) and was constant over a 1 h period (Figure S2). Virtually no labeling was observed in the absence of 1, UV irradiation, or the copper catalyst, highlighting the substantial and specific photoaffinity labeling afforded by 1 (Figure 1C).

Figure 1. A diazirine- and alkyne-containing NAPE analog targets numerous proteins by photoaffinity labeling.

(A) Schematic of PAL strategy for determining protein interactors of NAPE using bifunctional NAPE analog 1. (B) HeLa cells were treated with 1 at the indicated concentration for 1 h in the presence of the NAPE-PLD inhibitor LEI-401 (20 μM), followed by UV irradiation, lysis, CuAAC tagging with TAMRA-azide, SDS-PAGE, and analysis by in-gel fluorescence. Shown are TAMRA fluorescence (left) and Coomassie staining (right). (C) Experiments similar to those in (B) where key components (1, UV irradiation, or CuSO₄ in the CuAAC reaction) were omitted as negative controls to show specificity of the labeling. The concentration of 1 in these experiments was 50 μM. Molecular weight markers are shown in kDa.

To assess the requirement for NAPE-PLD inhibition, we compared labeling with 1 in the presence and absence of LEI-401 and surprisingly found that LEI-401 treatment decreased labeling (Figure S3). Because LEI-401 treatment increases levels of endogenous NAPE by preventing its degradation by NAPE-PLD²⁰ and because 1 achieves constant levels over a 2 h incubation (Figure S1), we hypothesize that 1 may not be a good NAPE-PLD substrate, and the net effect of LEI-401 in these experiments is to increase levels of a competitor (endogenous NAPE) for labeling of proteins with 1. We also assessed the localization of labeling 1 by fixation of cells treated with 1 followed by CuAAC tagging with a BODIPY-azide reagent and confocal microscopy. These studies revealed a complex intracellular localization consistent with a mixture of punctate, endosomal and post-Golgi vesicular compartments, plasma membrane, and other intracellular organelles (Figure S4). Consistent with the in-gel fluorescence studies (Figure S3), we found that addition of LEI-401 caused a modest decrease in labeling (Figure S4).

We next tagged the labeled proteins with biotin-azide to enable enrichment of these proteins using streptavidin-conjugated resin. To facilitate high-confidence labeling using stable isotope labeling with amino acids in cell culture (SILAC), we treated heavy isotopically labeled HeLa cells with 1, whereas cells treated with DMSO were cultured in light medium. After UV-activated crosslinking and lysis, lysates obtained from the two conditions were combined in a 1:1 ratio, followed by CuAAC tagging with biotin-azide, streptavidin affinity enrichment, and LC–MS/MS analysis for peptide identification and quantification. Enrichment ratios > 1.5 and P values < 0.05 were set as the threshold. In total, 189 proteins were identified in triplicate proteomics experiments (Figure 2A). Of the 189 proteins, 30 hits were found in at least two of the three repeats (Figure 2B and Table 1).

Figure 2. SILAC-based proteomics studies identify photoaffinity labeling targets of NAPE probe 1.

HeLa cells were treated either with 1 (50 μM, 1 h) (medium SILAC media) or DMSO (light SILAC media) in the presence of LEI-401 (20 μM), followed by UV irradiation, lysis, combination in 1:1 ratio, CuAAC tagging with biotin-azide, streptavidin-agarose enrichment, and analysis by LC–MS/MS-based proteomics. (A) Venn diagram showing the number of hit proteins, also the number of proteins unique to each of three replicates (A, B, C) with the number of proteins appearing in two or three replicates indicated in the intersections between circles. Hits were defined as having enrichment values of (heavy/light) > 1.5 and P values< 0.05. (B) Volcano plot showing the average enrichment fold change and P values of proteins from the experiment (n = 3). Proteins found in two or three replicates are indicated in orange and red, respectively. Identities of all such proteins are provided in Table 1.

Table 1. Identities of the 30 proteins that were significantly enriched in at least two of the three replicate proteomics experiments.^a.

CD147	CD44	SSBP1	PRKACA
RPL36	UBE2K	SSRP1	EIF4A3
PMPCB	RFC2	RRBP1	PPP2CA
RPS11	AP1B1	NUP155	BZW2
SAR1B	SEC23A	HNRNPA0	HLA-C
THOP1	LGALS3BP	STAG2	SMC4
GCLM	ANXA4	IGF2BP3	RANGAP1
FKBP1A	EIF3F

^aProteins found in three replicates have a gray background, and those found in two replicates have a white background.

Examination of potential NAPE functions using Gene Ontology analysis revealed several biological processes pathways related to protein synthesis and transport, and also metabolic regulation in response to stimuli or stress (Figure S5 and Tables S1 and S2), the latter of which may be related to the reported accumulation of NAPE in response to brain ischemia or heart infarction.^3-6

Validation of the interaction of NAPE with CD147/BSG, CD44, and BZW2

We subsequently sought to confirm select hits from the proteomics studies. A key component of such studies would be to assess if photoaffinity labeling of target proteins by probe 1 could be outcompeted by excess amounts of native NAPE. However, exogenous delivery of NAPE to requisite levels was not feasible due to the poor solubility of this very hydrophobic, triply acylated lipid in aqueous solutions. We therefore pursued an alternative strategy to elevate intracellular NAPE levels: overexpression of a NAPE-synthesizing enzyme. We selected Phospholipase A and acyltransferase 2 (PLAAT2), which produces NAPE from PE and the sn-1 acyl chain of PC and possess the greatest NAPE-synthesizing activity within the PLAAT family.³⁶

Thus, we generated a stable cell line bearing a HaloTag fusion of PLAAT2 (HT-PLAAT2) using the Sleeping Beauty tet-on vector system (pSBtet) such that addition of doxycycline would induce HT-PLAAT2 expression.³⁷ Analysis of lipid extracts from these cells indicated that HT-PLAAT2 expression over 48 h of doxycycline treatment caused substantial increases in intracellular NAPE (Figure 3). Addition of the NAPE-PLD inhibitor LEI-401 16 h prior to the endpoint of the 48-h doxycycline treatment further boosted the NAPE levels by approximately 3–5-fold (Figure 3). Interestingly, overexpression of HT-PLAAT2 also raised NAE levels, though NAPE-PLD inhibition with LEI-401 did not lead to downregulation of NAE, suggesting its formation via alternate NAPE degradation pathways (Figure S6). This strategy to induce de novo NAPE synthesis in situ not only solved the limitation of solubility but also yielded a panel of NAPE species based on acyl tail composition that could potentially exhibit different activities, as acyl chain composition can have major impacts on binding preferences and phospholipid functions.^38,39

Figure 3. A system for inducible production of NAPE in cells by doxycycline-mediated overexpression of PLAAT2.

HeLa-pSBtet-PLAAT2 cells were treated with doxycycline (Dox, 2.5 μg/mL; 48 h) and, for the final 16 h, either DMSO or LEI-401 (20 μM; 16 h), followed by generation of lipid extracts and lipid analysis by LC–MS (n=3). Quantification of levels of different NAPE species, indicated by total number of carbons and double bonds in the combined acyl tails. Note that NAPE levels were below the detection limit in the absence of doxycycline treatment.

Among the handful of proteomics hits present in all three replicates, we noted two cell-surface proteins, CD147/BSG and CD44, that had high enrichment factors and statistical significance (Figure 2B). As a result, we examined the labeling of these two proteins by 1 in HeLa-pSBtet-PLAAT2 cells. Treatment of cells with 1 followed by photocrosslinking, CuAAC tagging with biotin-azide, streptavidin enrichment, and blotting revealed substantial labeling of endogenous CD147 and CD44 (Figure 4A). Encouragingly, for both proteins, their extents of labeling were substantially diminished in the presence of elevated levels of intracellular NAPE induced by a combination of doxycycline and LEI-401 (Figure 4A). Consistent with earlier studies that showed that LEI-401 treatment decreased global photoaffinity labeling with 1 (Figures S3 and S4), we found that LEI-401 treatment of parental HeLa cells (i.e., those not overexpressing PLAAT2) caused a decrease in labeling of CD44 and CD147 by 1 (Figure S7). These results further support the conclusion that LEI-401 treatment increases endogenous NAPE, which can compete for 1 engagement with target proteins.

Figure 4. NAPE probe 1 labels CD147, CD44, and BZW2 in a manner that is competed by excess native NAPE.

HeLa-pSBtet-PLAAT2 cells were treated either with doxycycline (Dox, 2.5 μg/mL for 48 h) and, for the final 16 h, LEI-401 (20 μM) or DMSO. For the final 1 h of labeling, cells were treated with 1 (100 μM), followed by UV irradiation, lysis, CuAAC tagging with biotin-azide, streptavidin-agarose enrichment, and Western blot. (A) Blots of whole cell lysates (Input) or streptavidin-enriched fraction (IP: Streptavidin), probing for CD147, CD44, or GAPDH as a loading control. Quantification shown at right (n=3), with statistical significance determined using one-way ANOVA with Tukey post-hoc test. Error bars indicate standard deviation, and p values are indicated. (B) Similar experiments in cells transfected with FLAG-BZW2 and otherwise labeled and processed identically to in (A), with Western blot for FLAG and GAPDH.

We also validated the NAPE-sensitive photocrosslinking of 1 with a third protein, BZW2, a soluble protein that was recently implicated in the formation of ER-plasma membrane contact sites,⁴⁰ in cells transfected with a FLAG-tagged BZW2 construct (Figure 4B). Overall, these data support strong and specific lipid-protein interactions between NAPE and these proteins and underscore the effectiveness of our chemoproteomic approach to interrogate the NAPE interactome.

Mapping of NAPE-binding sites in CD147 and CD44

We next sought to determine the sites within CD147 and CD44 bound by NAPE. In previous work, we had generated a panel of point mutants in an HA-tagged CD147 construct, targeting residues within or proximal to its sole transmembrane domain that have a high propensity toward photoaffinity labeling with diazirines.^41,42 Using this panel of mutants, we examined the ability of 1 to label each relative to the wild-type protein. Whereas most mutants were readily labeled by 1 comparably to the wild-type protein, 1 failed to label the E218A mutant, indicating the importance of this particular amino acid in the crosslinking process (Figure 5A). Intriguingly, in a previous study,⁴³ we found that E218 was the crosslinking site of the photoaffinity probe for phosphatidylethanol (PDAA), a phospholipid that share substantial structural similarity to NAPE in that it has two long acyl tails, the same overall negative charge coming from a phosphodiester, and an alkyl head group, albeit of much shorter length. These studies indicate that the NAPE-binding region of CD147 is in proximity to E218 and suggests that phospholipid binding may be an important feature of this protein.

Figure 5. Mapping of potential NAPE binding sites in CD147 and CD44.

Cells were transfected with the indicated wild-type (WT) or mutant variants of CD147-HA (A) or CD44-V5 (B). Cells were treated with 1 (100 μM, 1 h), followed by UV irradiation, lysis, CuAAC tagging with biotin-azide, streptavidin-agarose enrichment, and blotting of whole cell lysates (Input) or streptavidin-enriched fraction (IP: SA) for HA, V5, or GAPDH as indicated. Quantification shown below each representative set of blots (n=3), with statistical significance determined using one-way ANOVA with Tukey post-hoc test. Error bars indicate standard deviation, and p values are indicated.

Like CD147, CD44 is a single-pass transmembrane protein, and we therefore examined several point mutants of a V5-tagged CD44 construct, including at palmitoylation sites (C286A, C295A), phosphorylation sites (S291A, S316A, S325A), and motifs rich in positively charged residues (R292A/R293A/R294A (“3R3A”) and K298A/K299A/K300A (“3K3A”)). Because CD44 forms dimers,^44,45 to prevent heterodimerization of CD44 mutants with endogenous, wild-type CD44, we generated CD44-depleted cells (Figure S8A). Photocrosslinking studies of 1 in such cells transfected with one of several CD44 mutants revealed that mutation of palmitoylation sites (C295 and, to a lesser extent, C286), caused a partial decrease in labeling (Figure 5B). Interestingly, S-acylation of C295 and C286 cause an increased association of CD44 with liquid-ordered “raft” domains in the plasma membrane.⁴⁶ Importantly, a control experiment wherein we omitted the UV photocrosslinking step revealed no detectable labeling CD44 and CD147 with 1, ruling out the possibility of metabolism of 1 to alkyne-containing fatty acids that could in principle result in labeling of palmitoylated proteins such as CD44 (Figure S9). Overall, these studies are consistent with a model wherein the protein has greater access to 1, and presumably NAPE, in such regions, which may be enriched in these very hydrophobic, three-tailed lipids.

NAPE activates monocarboxylate transporters and lactate levels

CD147 functions as a chaperone that facilitates the trafficking to and activation of monocarboxylate transporters from the SLC16A family, particularly MCT1 and MCT4, at the plasma membrane.⁴⁷ In the plasma membrane, these MCTs act as proton-coupled transporters that facilitate the import or export of small monocarboxylate metabolites such as lactate and pyruvate^48,49 and thus acts as a major regulator of intracellular lactate,^50-52 which is crucial for cell metabolism because lactate is a byproduct of glycolysis that can slow glycolytic rates and thus ATP production.^53-55 Therefore, clearance of intracellular lactate by MCTs is crucial to maintain an appropriate metabolic rate. Interestingly, E218 of CD147, the crosslinking site of 1, resides within the transmembrane domain and is within the region of CD147 that interacts with MCTs and is critical for the chaperone activity of CD147.^56,57 Further, CD44 was also reported to serve as an additional chaperone of MCTs.^58,59 Our discoveries that NAPE interacts with both CD147 and CD44 raise the possibility that these lipid–protein interactions are related to a common function of these proteins in modulating MCT-mediated lactate transport.

To test the hypothesis that NAPE modulates lactate transport by MCTs, we first examined whether elevations to NAPE levels affected the subcellular localization of GFP-tagged MCT1. We found that the localization of MCT1-GFP in PLAAT2-expressing HeLa cells either in the presence or absence of LEI-401 to further boost NAPE levels resulted in a similar localization of MCT1-GFP, namely substantially at the plasma membrane with a minor intracellular pool (Figure S10). We performed similar studies under conditions expected to produce a predominantly plasma membrane localization of MCT1-GFP by co-expression of its chaperones CD147-HA and CD44-V5, and again we found that LEI-401 treatment had no effect on the localization of MCT1-GFP (Figure S11). These results led us to hypothesize that the effect of NAPE on MCTs via CD147 and CD44 may occur via influencing transport activity rather than localization.

Therefore, we then measured lactate flux by separately collecting conditioned medium (for extracellular lactate measurement) and cell lysates (for intracellular lactate measurements) from cells with acute elevations to NAPE levels compared to control cells, followed by quantification of lactate levels by LC–MS. Remarkably, after only a short LEI-401 treatment to HeLa cells to inhibit NAPE-PLD and prevent NAPE degradation, the extracellular/intercellular lactate ratio was elevated by 29% compared to that from control cells, suggesting an acceleration of lactate export by NAPE (Figure 6A). Importantly, the inclusion of inhibitors of MCT1 and MCT4 activity (AZD3965⁶⁰ and AZD0095⁶¹, respectively) prevented the LEI-401-induced increase in lactate export. These studies indicate that accumulation of NAPE induced by LEI-401 affects lactate levels via the activity of MCT1/4.

Figure 6. NAPE promotes lactate export is via binding with CD147 and CD44 and regulation of monocarboxylate transporters.

(A) HeLa cells were treated with LEI-401 (10 μM) or DMSO for 45 min in the presence or absence of inhibitors of MCT1 and MCT4 (AZD3965 and AZD0095, used at 1 μM). Conditioned medium (for extracellular measurements) and cell pellets (for intracellular measurements) were collected and analyzed by LC–MS to quantify lactate levels, which were expressed as a ratio of extracellular to intracellular levels. (B) Similar studies to (A) except that cells were subjected to depletion of both CD147 and CD44 by shRNA and siRNA respectively, as indicated. (C) Similar studies to (A) except that cells were subjected to depletion of both CD147 and CD44 by shRNA and siRNA respectively, as well as rescue with a combination of either WT CD147-HA and CD44-V5 or mutant forms that do not photocrosslink to 1 (i.e., CD147^E218A-HA and CD44^C295A-V5), as indicated. Statistical significance was determined using one-way ANOVA with Tukey post-hoc test (n=6), with p values indicated on plots. ns, not significant. (D) Model for function of NAPE in lactate export via interactions with the monocarboxylate transporter (MCT) chaperones CD147 and CD44, which increases the lactate export activity of MCTs.

To assess whether this higher lactate efflux rate is a result of the binding of NAPE to CD147 and CD44, both of these proteins were depleted from cells using a combination of shRNA knockdown of CD147 and siRNA knockdown of CD44. These studies revealed that in the absence of CD147 and CD44, LEI-401 treatment no longer caused an increase in extracellular/intracellular lactate levels (Figure 6B and Figure S8B-C). We then examined whether re-introduction of CD147 and CD44 into these CD147/CD44-depleted cells could rescue the defect in lactate transport, using either WT forms of these proteins or mutant variants that fail to photocrosslink to 1 and would be predicted to bind more poorly to endogenous NAPE (CD147^E218A and CD44^C295A, see Figure 5). We found that, whereas WT CD147/CD44 rescued the effects of CD147/CD44 knockdown on lactate transport, the mutant forms of these proteins did not (Figure 6C). Collectively, these studies support a model wherein NAPE–CD147 and NAPE–CD44 interactions causes activation of MCTs, which in turn boost the clearance of intracellular lactate (Figure 6D).

DISCUSSION

PAL has emerged as a powerful strategy for identification of elusive and often low-affinity non-covalent lipid–protein interactions. A persistent challenge for design of effective PAL probes for lipids is the rapid metabolism of most lipids. When the diazirine and alkyne groups are placed in proximity to one another, metabolism of a PAL probe may result in a diazirine- and alkyne-containing lipid with a different head group or tail composition from the synthetic probe. Thus, the ability to separate diazirine and alkyne groups into metabolically distinct regions of the lipid molecule is an ideal solution. NAPE presents an outstanding opportunity to exploit this design principle, because it is unusual in that it has three hydrophobic tails, and one of them, the N-acyl group, is separated from the others via the amide and phosphodiester, which are most commonly rapidly metabolized to produce NAE signaling molecules.

Biological functions for NAPE have proven elusive given its low cellular levels and rapid metabolism to NAE. Thus, we were motivated to use a PAL strategy to identify protein interactors of NAPE as a key to elucidating physiological functions for this lipid. Indeed, our NAPE PAL probe 1 featured a diazirine on the N-acyl group and alkynes on the O-acyl tails, such that metabolism to the corresponding NAE analogs would not permit photocrosslinking and enrichment of interactors. This approach proved effective, as several high-confidence NAPE interactors were identified using PAL-enabled chemoproteomics.

By taking advantage of two distinct tools for acute elevation of NAPE levels in cells, namely NAPE-PLD inhibition with LEI-401 and doxycycline-inducible expression of the NAPE-synthesizing enzyme PLAAT2, we validated the specificity of the interaction of 1 with select hits in competition experiments, including two single-pass transmembrane proteins, CD147 and CD44, indicating that 1 acted as a bona fide probe for NAPE.

Targeted mutagenesis of membrane-proximal and transmembrane residues within CD147 and CD44 allowed us to narrow down likely binding regions for NAPE on these proteins. Interestingly, both CD147 and CD44 act as chaperones for monocarboxylate transporters that transport lactate, pyruvate, and other anionic metabolites from the SLC16A family (e.g., MCT1, MCT4), prompting us to investigate CD147- and CD44-dependent roles for NAPE in modulating lactate flux. Taking advantage of the above-mentioned tools for acute perturbation of NAPE levels, as well as RNAi-mediated knockdown, we established that NAPE induces an increase in MCT-mediated lactate efflux from cells in a manner dependent upon CD147/CD44. Overall, these studies provide definitive evidence for a physiological signaling function for NAPE beyond its well documented role as metabolic precursor to NAE. These findings may have relevance in ischemia, which reprograms the main pathway of energy production from oxidative phosphorylation to glycolysis and therefore results in more lactate production. Because NAPE accumulates in the mammalian brain undergoing ischemia,^5,6 our findings suggest that NAPE may be a physiological response to ischemia to enable hypoxic cells to avoid aberrantly high lactate levels in the cytosol and restore metabolic and energy homeostasis.

Supplementary Material

The supporting information is available free of charge. Western blots validating global and target-specific NAPE probe labeling, confocal microscopy of NAPE probe labeling and MCT1-GFP localization, gene ontology analysis and list, lipid quantification by mass spectrometry, and RNAi knockdown validation (Figures S1-S11 and Table S1), materials and methods, synthetic methods, NMR spectra, and associated references (PDF). Raw proteomics data (Table S2) (XLSX).