Molecular characterization of N-acylethanolamine-hydrolyzing acid amidase, a novel member of the choloylglycine hydrolase family with structural and functional similarity to acid ceramidase.

Bioactive N-acylethanolamines, including anandamide (an endocannabinoid) and N-palmitoylethanolamine (an anti-inflammatory and neuroprotective substance), are hydrolyzed to fatty acids and ethanolamine by fatty acid amide hydrolase. Moreover, we found another amidohydrolase catalyzing the same reaction only at acidic pH, and we purified it from rat lung (Ueda, N., Yamanaka, K., and Yamamoto, S. (2001) J. Biol. Chem. 276, 35552-35557). Here we report complementary DNA cloning and functional expression of the enzyme termed "N-acylethanolamine-hydrolyzing acid amidase (NAAA)" from human, rat, and mouse. The deduced primary structures revealed that NAAA had no homology to fatty acid amide hydrolase but belonged to the choloylglycine hydrolase family. Human NAAA was essentially identical to a gene product that had been noted to resemble acid ceramidase but lacked ceramide hydrolyzing activity. The recombinant human NAAA overexpressed in HEK293 cells hydrolyzed various N-acylethanolamines with N-palmitoylethanolamine as the most reactive substrate. Most interestingly, a very low ceramide hydrolyzing activity was also detected with NAAA, and N-lauroylethanolamine hydrolyzing activity was observed with acid ceramidase. By the use of tunicamycin and endoglycosidase, NAAA was found to be a glycoprotein. Furthermore, the enzyme was proteolytically processed to a shorter form at pH 4.5 but not at pH 7.4. Expression analysis of a green fluorescent protein-NAAA fusion protein showed a lysosome-like distribution in HEK293 cells. The organ distribution of the messenger RNA in rats revealed its wide distribution with the highest expression in lung. These results demonstrated that NAAA is a novel N-acylethanolamine-hydrolyzing enzyme that shows structural and functional similarity to acid ceramidase.

Anandamide and other NAEs are hydrolyzed to free fatty acids and ethanolamine, and this intracellular degradation is mostly attributed to the catalysis by fatty acid amide hydrolase (FAAH) (30 -33). This membrane-bound enzyme is widely distributed in mammalian organs and is characterized by an optimal pH value at 8.5-10 and high sensitivity to serine hydrolase inhibitors, including phenylmethylsulfonyl fluoride and methyl arachidonyl fluorophosphonate. Among the various NAEs hydrolyzed by FAAH, anandamide is the most reactive substrate, and N-palmitoylethanolamine is less reactive. Recent progress in the studies on FAAH, including cDNA cloning (34) and crystallography (35), revealed that FAAH belongs to the amidase signature family and possesses a catalytic triad of Ser-Ser-Lys. In addition, analysis of FAAH-deficient mice showed the central role of this enzyme in the degradation of anandamide and other NAEs in the brain (36 -38).
In addition to FAAH, we found another NAE-hydrolyzing amidohydrolase, which was active only at acidic pH, first in human megakaryoblastic (CMK) cells (39) and later in various rat tissues including lung, spleen, and macrophages (40). This enzyme, termed as N-acylethanolamine-hydrolyzing acid amidase (NAAA) in the present article, was distinguishable from FAAH by its optimal pH around 5, the preference of N-palmitoylethanolamine to other NAEs, activation by Triton X-100 (a non-ionic detergent) and dithiothreitol (DTT), and lower sensitivity to phenylmethylsulfonyl fluoride and methyl arachidonyl fluorophosphonate (39,40). We recently found that some ester and amide compounds such as N-cyclohexanecarbonylpentadecylamine are selective inhibitors of NAAA without inhibitory effects on FAAH (41,42). These results strongly suggested that NAAA was an enzyme protein different from FAAH.
The molecular characterization of NAAA was indispensable to the elucidation of its physiological role. Although we purified NAAA from rat lung to apparent homogeneity (40), cDNA cloning of NAAA has not been performed to date. Here we report for the first time the identification of cDNA encoding NAAA. The functional expression of the cDNA confirmed NAAA to be a second NAE-hydrolyzing enzyme. Furthermore, the deduced amino acid sequence unveiled similarity of NAAA to acid ceramidase (AC), an amidohydrolase-hydrolyzing ceramide to sphingosine and fatty acid, which urged us to compare catalytic properties of NAAA with those of AC. We also investigated possible glycosylation and proteolytic modification of NAAA and its intracellular localization. Our results may contribute to the further understanding of the catabolism of NAEs and ceramides in the context of lipid metabolism.
Enzyme Purification and Protein Sequencing-The purification of NAAA enzyme from rat lung was performed as described previously (40). Briefly, NAAA was solubilized in phosphate-buffered saline (PBS) at pH 7.4 by freezing and thawing from the 12,000 ϫ g pellet of the lung homogenates of 90 adult Wistar rats (250 -500-g body weight, Charles River Breeding Laboratories, Japan). After centrifugation at 105,000 ϫ g for 50 min, the supernatant was subjected to acid treatment at pH 4. Resultant insoluble proteins were removed by the centrifugation at 260,000 ϫ g for 40 min, and the supernatant was loaded onto phenyl-Sepharose CL-4B. NAAA was then eluted with 1% octyl glucoside and loaded on a HiTrap heparin column. The flow-through fractions were collected and applied onto a hydroxyapatite column (Bio-Gel HTP). NAAA was eluted with 20 mM Tris-HCl (pH 7.4) containing 10 mM sodium phosphate. The sample was finally loaded onto a HiTrap butyl FF equilibrated with 2 M ammonium sulfate, and NAAA was eluted with 20 mM Tris-HCl (pH 7.4) containing 1% octyl glucoside. Protein concentration was determined by the method of Bradford (45) with bovine serum albumin (BSA) as a standard.
The purified rat lung NAAA was concentrated by a centrifugal filter device (Microcon YM-10), subjected to SDS-PAGE on a 10% gel, and electrotransferred to a hydrophobic polyvinylidene difluoride membrane (Hybond-P). After staining with Coomassie Brilliant Blue G-250, the band was excised, and its N-terminal sequence was determined with a Procise model 492 protein sequencer (Applied Biosystems, Foster City, CA).
cDNA Cloning-The data base search using the BLAST analysis program revealed that the determined N-terminal sequence of the purified rat NAAA was contained in a rat putative cDNA sequence (GenBank TM accession number XM_223237.2) and a mouse cDNA sequence (AK008776.1). We also found that a reported human amino acid sequence termed "acid ceramidase-like protein" (GenBank TM accession number Q02083) (46) was highly homologous to the rat and mouse sequences. We assumed these three sequences were as NAAA of rat, mouse, and human (see "Results").
The cDNAs containing the putative full-length coding region of NAAA and AC were generated by PCR. Total RNA was isolated with Trizol reagent from human CMK cells cultured in DMEM containing 10% FCS (PAA Laboratories) and 0.1 M 12-O-tetradecanoylphorbol 13-acetate for 4 days or from the lungs of rat (Wistar, Charles River Japan) or mouse (C57BL/6, Japan SLC). cDNAs were then prepared with the aid of oligo(dT) primer and Ready-To-Go You-Prime First-Strand Beads. For cDNA cloning of human NAAA, we prepared two cDNA fragments according to the cDNA sequence of acid ceramidaselike protein (46). One cDNA fragment was amplified by LA Taq polymerase with the sense primer 5Ј-GCGGAAGCTTGAGCCCGAGCCAT-GCGGAC-3Ј and the antisense primer 5Ј-CATCAGCAATAAGGGGAG-TCTTGGCCAACT-3Ј, and its digestion with HindIII and ApoI gave a 0.44-kb fragment containing nucleotides 27-463 (the numbers from Ref. 46). The other cDNA fragment was amplified by Pyrobest DNA polymerase with the sense primer 5Ј-AAGACTCCAGAGGCCACATTT-ACCATGGTC-3Ј and the antisense primer 5Ј-TTCTCTCGAGTTACTT-TCTACTCGGGTTTC-3Ј, and its digestion with ApoI and XhoI gave a 0.65-kb fragment containing nucleotides 464 -1118. The HindIII-ApoI fragment was first inserted between the HindIII and EcoRI sites of a eukaryotic expression vector pcDNA 3.1(ϩ), and the ApoI-XhoI fragment was then added between the EcoRI and XhoI sites to generate pcDNA 3.1(ϩ) harboring the putative full-length coding region of human NAAA cDNA. For cDNA cloning of rat and mouse NAAA, we first amplified cDNA fragments containing the entire open reading frame by Pyrobest DNA polymerase with the sense primer 5Ј-CTGGTACCGCT-CATTGGTCAGGTAGTGACC-3Ј (rat) or 5Ј-CAGGTACCATTCATTGG-TCCGGTGGTGGCC-3Ј (mouse), and the antisense primer 5Ј-TCGAA-TTCTTAGTCTTCCAAATGCAGCTC-3Ј (rat) or 5Ј-AGGAATTCCAGT-CTTAGACTTCTAAATGCAG-3Ј (mouse). These primers were designed on the basis of the above-mentioned cDNA sequences (GenBank TM accession numbers XM_223237.2 and AK008776.1) and the adjacent genomic sequence of rat (NW_047423.1). After the amplified cDNA fragments were digested with KpnI and EcoRI and inserted into pBluescript SK(ϩ), we performed the second PCR by Pyrobest DNA polymerase with these cDNA clones and internal primers (the sense primer, 5Ј-GTGGTACCAGCAGCCCGAGCCATGGGGAC-3Ј for rat or 5Ј-GTG-GTACCAGAAGCCGGAGCTATGGGGAC-3Ј for mouse; the antisense primer, 5Ј-AAGAATTCTCAGCTTGGGTTTCTGATC-3Ј for rat or 5Ј-CAGAATTCTCAGCTCGGGTTTCTGATC-3Ј for mouse). The amplified cDNA fragments were digested with KpnI and EcoRI and inserted into pcDNA 3.1(ϩ). For construction of cDNA for hexahistidine-tagged NAAA, a human NAAA cDNA lacking stop codon was generated by PCR using the sense primer 5Ј-GCGGAAGCTTGAGCCCGAGCCATGCGG-AC-3Ј and the antisense primer 5Ј-TCCTCGAGGATCCTTTCTACTCG-GGTTTCT-3Ј. This cDNA fragment was digested with HindIII and XhoI and inserted into pcDNA 3.1/myc-His C to generate a C-terminally hexahistidine-tagged NAAA protein.
For human AC, the full-length coding region was amplified by Pyrobest DNA polymerase with the sense primer 5Ј-GCGGTACCAGGC-TAGAGCGATGCCGGGCC-3Ј and the antisense primer 5Ј-GACGGAT-CCTCACCAACCTATACAAGGGTC-3Ј. These primers were designed according to the reported nucleotide sequence of human AC (Gen-Bank TM accession number U70063) (47). The amplified 1.2-kb fragment was digested with KpnI and BamHI and inserted into pcDNA 3.1(ϩ). The inserted cDNAs for human, rat, and mouse NAAA, the C-terminally hexahistidine-tagged NAAA, and human AC were sequenced in both directions with the aid of ABI 310 and 377 DNA sequencers (Applied Biosystems).
Overexpression of Recombinant Enzymes-HEK293 cells were cultured at 37°C to 70% confluency in a poly-L-lysine-coated 100-mm dish containing DMEM with 10% FCS (Invitrogen) and 0.1 mM non-essential amino acids in a humidified 5% CO 2 , 95% air incubator. Transfection was performed according to the manufacturer's instructions; 8 g of the expression vector pcDNA 3.1(ϩ) harboring cDNA of NAAA or AC was mixed with 72 l of Lipofectamine in 1.6 ml of serum-free DMEM for 30 min and then added to the cells with 6.4 ml of serum-free DMEM. After incubation for 5 h, the medium was removed, and the transfected cells were further cultured in the presence of FCS for 43 h with one change of medium at 19 h. For tunicamycin treatment, the culture medium containing 5 g/ml tunicamycin or vehicle (0.05% Me 2 SO) was used after the change of medium at 19 h. The cells were harvested with the aid of trypsin, washed twice, suspended in 800 l of PBS, and sonicated three times each for 3 s. The resultant cell lysate was used as the enzyme preparation. Control HEK293 cells were prepared by the same method, except that the insert-free pcDNA 3.1(ϩ) vector was used for transfection.
Enzyme Assay-For the assay of NAE hydrolyzing activity, the enzyme was incubated with 200 M N-[ 14 C]acylethanolamine (1000 cpm/ nmol, dissolved in 10 l of Me 2 SO) at 37°C for 30 min in 100 l of 100 mM citrate-sodium phosphate buffer (pH 4.5) containing 3 mM DTT, 0.1% Nonidet P-40, 0.05% BSA, and 150 mM NaCl, unless otherwise noted. The reaction was terminated by the addition of 0.32 ml of a mixture of diethyl ether, methanol, 1 M citric acid (30:4:1, v/v). In the experiment of substrate specificity (Fig. 2), 5 mM 3(2)-t-butyl-4-hydroxyanisole was also included in the stop solution to suppress autoxidation of the compounds containing an unsaturated fatty acyl chain. After centrifugation, 100 l of the organic phase was spotted on a silica gel thin layer plate (10 cm height) and was developed at 4°C for 25 min with a mixture of chloroform, methanol, 28% ammonium hydroxide (80:20:2, v/v). Distribution of radioactivity on the plate was quantified by a BAS1500 bioimaging analyzer (Fujix, Tokyo, Japan). Oleamide hydrolyzing activity was assayed by the same procedure with 200 M [ 14 C]oleamide. 2-Arachidonoylglycerol hydrolyzing activity was examined by quantification of arachidonic acid produced from 100 M 2-arachidonoylglycerol using reverse phase-high performance liquid chromatography as described previously (48).
In Vitro Processing-HEK293 cells expressing hexahistidine-tagged NAAA were subjected to sonic disruption, and the lysates were centrifuged at 12,000 ϫ g for 30 min at 4°C. The obtained pellet was resuspended in PBS, and the NAAA enzyme was solubilized by two cycles of freezing and thawing, followed by further centrifugation at 105,000 ϫ g for 50 min at 4°C. The supernatant was then incubated with either citrate-sodium phosphate (pH 4.5) or Tris-HCl (pH 7.4) at 25 mM in the presence of 0.1% Nonidet P-40 at 37°C for 20 or 60 min. The samples were then subjected to the enzyme assay or Western blotting.
For PNGase F digestion, an aliquot of the sample (4 g of protein) was neutralized with 100 mM sodium phosphate (pH 7.5), denatured in 0.5% SDS and 1% ␤-mercaptoethanol at 100°C for 10 min, and digested with 50 units of PNGase F at 37°C for 2 h in the presence of 1% Nonidet P-40.
Western Blotting-After separation by SDS-PAGE on a 10% gel under denaturing conditions, proteins were electrotransferred to a Hybond-P membrane. The membrane was blocked with 5% nonfat dry milk in PBS containing 0.1% Tween 20 overnight and then incubated with anti-hexahistidine antibody (1:5000 dilution) in the blocking buffer at room temperature for 1 h, followed by the incubation with the horseradish peroxidase-linked secondary antibody (1:2000 dilution) in the blocking buffer at room temperature for 1 h. Finally, hexahistidinetagged NAAA was visualized using an ECL kit.
Construction of Green Fluorescent Protein (GFP)-Human NAAA Fusion Protein and Fluorescence Microscopy-The human NAAA cDNA lacking stop codon prepared as described above was inserted between the HindIII and BamHI sites of a eukaryotic expression vector pEGFP-N1. The resulting construct encoded a fusion protein in which enhanced GFP was fused to the C terminus of human NAAA. HEK293 cells were transfected with this plasmid as described above, except that the cells were plated onto 25-mm circular glass coverslips in a 6-well plate and one-eighth volumes of reagents were used throughout the transfection procedure. Forty three hours after the end of transfection, the cells were analyzed by fluorescence microscopy as described previously (49). Briefly, the culture medium was replaced with Ringer's buffer (155 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 2 mM Na 2 HPO 4 , 10 mM glucose, 10 mM HEPES (pH 7.2), and 0.5 mg/ml BSA), and the cells in a temperaturecontrolled chamber at 37°C were observed with an epifluorescence microscope (Nikon TE300, Tokyo, Japan). Fluorescent images of live cells were then captured using a cooled CCD camera controlled by Meta-Morph Imaging System (Universal Imaging Corp., West Chester, PA).
Reverse Transcription (RT)-PCR-Total RNA was isolated with Trizol reagent from various organs of rats (Wistar, Charles River, Japan). cDNAs were then synthesized from total RNA (5 g) by using Moloney murine leukemia virus-reverse transcriptase and random hexamer and were subjected to PCR amplification. Primers for the rat NAAA gene were 5Ј-ATTCTGCACCAGTATTGTGGCCCAAGACTC-3Ј (sense) and 5Ј-TCCATTCAGAGGGTCAAGAGGCCAAATGTC-3Ј (antisense), and those for rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene were 5Ј-AAGGTCGGTGTGAACGGATTTGG-3Ј (sense) and 5Ј-ACAAACATGGGGGCATCAGC-3Ј (antisense). PCR conditions used were as follows: for NAAA, denaturation at 94°C for 30 s, annealing at 63°C for 30 s, and extension at 72°C for 48 s (26 cycles); for GAPDH, denaturation at 95°C for 30 s, annealing at 58°C for 15 s, and extension at 72°C for 30 s (27 cycles). PCR products, which were confirmed to be in the logarithmic phase, were electrophoresed on a 1.8% agarose gel and stained with ethidium bromide.

RESULTS
Identification and Cloning of NAAA cDNA-As described previously (40), we solubilized the NAAA enzyme from the 12,000 ϫ g pellet of rat lung homogenates by freezing and thawing, and we purified it by acid treatment and four chromatographic steps using phenyl-Sepharose, HiTrap heparin, hydroxyapatite, and HiTrap butyl. Through this purification procedure, the specific enzyme activity was increased 760-fold from 5.1 nmol/min/mg protein to 3.9 mol/min/mg protein with N-palmitoylethanolamine as substrate. In agreement with our previous results (40), the purified enzyme preparation gave a major protein band around 31 kDa as analyzed by SDS-PAGE. This 31-kDa protein was transferred to a hydrophobic polyvinylidene difluoride membrane (Hybond-P) and subjected to Edman degradation. The N-terminal sequence was thus determined as XTSIVAQDSQGRI (X; an unidentified amino acid).
We next searched the data base by using the BLAST analysis program, and we found that this sequence was contained in a putative cDNA clone of rat (GenBank TM accession number XM_223237.2, catalogued in UniGene Rn.101943) and a mouse cDNA sequence (AK008776.1, catalogued in UniGene Mm.28890). Further search revealed that a human protein (Q02083, catalogued in UniGene Hs.264330) was highly homologous to these rat and mouse sequences at the protein level. cDNA of this human protein was previously cloned from pla-centa and characterized by Hong et al. (46). They named the protein as acid ceramidase-like protein because of the similarity of its primary structure to AC, an enzyme hydrolyzing ceramide to sphingosine and fatty acid at acidic pH. However, they reported that this protein had no detectable ceramide hydrolyzing activity, and its function remained unclear. Our result therefore suggested that NAAA was identical to acid ceramidase-like protein. We carried out RT-PCR with PCR primers based on the nucleotide sequences, and we cloned cDNAs of putative human, rat, and mouse NAAA (hNAAA, rNAAA, and mNAAA, respectively) from human megakaryoblastic CMK cells and lungs of rat and mouse in which we had found NAAA activity (39,40).
We sequenced the coding region of hNAAA, which was found to be 99.8% identical at the nucleotide level to the acid ceramidase-like protein reported previously (46). The coding regions of rNAAA and mNAAA were 99.9% identical to the nucleotide sequences of XM_223237.2 and AK008776.1 (GenBank TM ), respectively. The amino acid sequences deduced from the cDNAs are aligned in Fig. 1A. The amino acid sequences were composed of 359 (human) and 362 (rat and mouse) residues. The molecular masses were calculated to be 40,065 (human), 40,312 (rat), and 40,074 (mouse). Their amino acid identity was 76.5% (between human and rat), 76.7% (between human and mouse), and 90.1% (between rat and mouse). There were six potential N-glycosylation sites in the human sequences (denoted with FIG. 1. Amino acid sequences deduced from cDNAs for human, rat, and mouse NAAA. A, alignment of the amino acid sequences deduced from hNAAA, rNAAA, and mNAAA. Asterisks, dots, and closed circles indicate identity shared by three polypeptides, identity shared by two polypeptides, and the potential N-glycosylation sites of the human sequence, respectively. B, comparison between amino acid sequences deduced from hNAAA and human AC. Asterisks and dots indicate identity and similarity, respectively. Dashes denote deletion of amino acid residues when compared with the other sequence. An arrowhead indicates the cleavage site to generate ␣and ␤-subunits of human AC. Underlines in A and B denote the sequence corresponding with the N-terminal sequence of the purified rat NAAA protein, which we determined. closed circles in Fig. 1A), five of which were also conserved in both of the rat and mouse sequences.
The amino acid sequences of hNAAA, rNAAA, and mNAAA showed no homology with human, rat, and mouse FAAH (34,50) but revealed similarity to human AC (U70063) (47), rat AC (AF214647), and mouse AC (AF157500) (51), respectively. Those of hNAAA and human AC are aligned in Fig. 1B, which showed 33% identity and 70% similarity over their entire length. Likewise, the amino acid sequences of rNAAA and rat AC showed 33% identity and 70% similarity, and those of mNAAA and mouse AC showed 34% identity and 70% similarity. The data base search also revealed that NAAA as well as AC belongs to the choloylglycine hydrolase family, which includes several hydrolases cleaving carbon-nitrogen bonds, other than peptide bonds, in linear amides (Pfam accession number PF02275).
Human AC is a heterodimer composed of 13-(␣) and 40-kDa (␤) subunits that are derived from a common 55-kDa precursor encoded by a single cDNA (47). NAAA of human, rat, and mouse had a sequence (CTSIVAQDS) highly homologous to the N-terminal sequence of the ␤-subunit of AC (CTSIVAEDK). The N-terminal sequence of rat NAAA, which we determined (denoted with an underline in Fig. 1A), was in agreement with this sequence, suggesting that both NAAA and AC are subjected to proteolytic cleavage at the same site.
NAE Hydrolyzing Activities of Recombinant NAAA and AC-We overexpressed hNAAA cDNA in HEK293 cells by the Lipofection method. In order to detect its NAE hydrolyzing activity, the homogenates of the transfected cells were allowed to react with N-[ 14 C]palmitoylethanolamine at pH 4.5, and the produced [ 14 C]palmitic acid was separated by TLC. Our previous results showed that the NAE hydrolyzing activity of rat lung NAAA was potently enhanced by the addition of DTT and non-ionic detergent Triton X-100 (40). Therefore, we measured the enzyme activity in the presence of 3 mM DTT and 0.1% Nonidet P-40, composed of a major constituent of Triton X-100. Under these conditions, the homogenates of hNAAA-transfected cells hydrolyzed N-palmitoylethanolamine with a specific enzyme activity of 8.1 Ϯ 0.1 nmol/min/mg protein at 37°C (Fig. 2, A, lane 3, and B). On the other hand, the homogenates of HEK293 cells transfected with the insert-free vector were almost inactive (less than 0.03 nmol/min/mg protein) (Fig. 2, A,  lane 2, and B). The rat and mouse homologues (rNAAA and mNAAA) were also expressed in HEK293 cells by the same method, and the homogenates of the transfected cells exhibited the N-palmitoylethanolamine hydrolyzing activities.
When we examined substrate specificity using various NAEs with different long-chain fatty acids, N-palmitoylethanolamine was found to be the most reactive substrate (Fig. 2B). Relative activities in the presence of 0.1% Nonidet P-40 were as follows: N-palmitoylethanolamine, 100%; N-myristoylethanolamine, 27%; N-lauroylethanolamine, 4%; N-stearoylethanolamine, 4%; N-arachidonoylethanolamine (anandamide), 3%; and Noleoylethanolamine, 2%. Removal of Nonidet P-40 from the assay mixture resulted in a decrease in the enzyme activity for these NAEs. The endogenous activities toward all of the tested NAEs were almost undetectable as examined with the homogenates of HEK293 cells transfected with the insert-free vector (Fig. 2B). Oleamide (34) and 2-arachidonoylglycerol (48) were reported to be good substrates of FAAH. However, the homogenates of hNAAA-transfected cells hydrolyzed oleamide only at a low rate (4% of the N-palmitoylethanolamine hydrolysis) and were almost inactive with 2-arachidonoylglycerol. These catalytic properties of the recombinant NAAA were consistent with previous results with rat lung NAAA (40).
In relation to the above-mentioned considerable homology in the amino acid sequences between NAAA and AC, it was interesting to investigate whether or not AC also has NAE hydrolyzing activity. We prepared cDNA of human AC from CMK cells by RT-PCR, according to the published sequence (Gen-Bank TM accession number U70063) (47). The sequence that we determined was completely identical with that of U70063. AC was then overexpressed in HEK293 cells by the same method as the NAAA overexpression. The results indicated that recombinant AC has relatively low but significant NAE hydrolyzing activities (Fig. 2, A, lane 8, and B). Unlike NAAA, which hydrolyzed N-palmitoylethanolamine at the highest rate, AC preferred N-lauroylethanolamine to other N-acylethanolamines, including anandamide and N-palmitoylethanolamine. Most interestingly, in contrast to NAAA, N-lauroylethanolamine hydrolyzing activity of AC was decreased by the addition of 0.1% Nonidet P-40 (Fig. 2B).
We also compared the effects of DTT on the NAE hydrolyzing activities between NAAA and AC. As shown in Fig. 3A, the addition of DTT in a range of 30 M to 10 mM dose-dependently increased the N-palmitoylethanolamine hydrolyzing activity of recombinant NAAA up to 6.9-fold. In contrast, DTT showed only a weak stimulatory effect (up to 1.4-fold) on the N-lauroylethanolamine hydrolyzing activity of AC. These results indicated that both NAAA and AC have NAE hydrolyzing activity but were distinct in the substrate specificity and the effects of Nonidet P-40 and DTT. Previously, native NAAA (39, 40) and AC (52) were shown to have optimal pH at 5 and 3.8 -4.3, respectively. When the pH value in the reaction mix-ture was changed between 3 and 11, both NAE hydrolyzing activities of recombinant NAAA and AC were the highest at pH 4.5 and hardly detectable above pH 8 (Fig. 3B). We also performed kinetic analyses on the NAE hydrolysis of these two enzymes. Both NAAA (with N-palmitoylethanolamine as substrate) and AC (with N-lauroylethanolamine) displayed typical Michaelis-Menten kinetics, with apparent K m values of 97 and 55 M, respectively (Fig. 4).
Ceramide Hydrolyzing Activity of Recombinant NAAA-Although Hong et al. (46) reported that acid ceramidase-like protein did not show ceramide hydrolyzing activity, we examined whether recombinant NAAA hydrolyzed ceramide under our assay conditions with 3 mM DTT, 0.1% Nonidet P-40, 0.05% BSA, and 150 mM NaCl in the reaction mixture. As a positive control, the homogenates of the HEK293 cells overexpressing recombinant AC were allowed to react with N-[ 14 C]lauroylsphingosine (C 12 -ceramide) at pH 4.5, and the product [ 14 C]lauric acid was separated by TLC. The result showed the generation of lauric acid with a specific activity of 0.43 nmol/min/mg protein at 37°C (Fig. 5, A, lane 4, and B). Consistent with the previous report (52) revealing that human urine AC shows the highest reactivity with N-lauroylsphingosine among ceramides with different N-acyl groups, recombinant AC hydrolyzed N-[ 14 C]palmitoylsphingosine (C 16 -ceramide) at a lower rate (0.095 nmol/min/mg protein) (Fig. 5B). The homogenates of the HEK293 cells transfected with the insert-free vector showed an endogenous ceramide hydrolyzing activity (0.014 Ϯ 0.002 and 0.002 Ϯ 0.002 nmol/min/mg protein toward N-lauroylsphingosine and N-palmitoylsphingosine, respectively) (Fig. 5, A,  lane 2 and B). When the homogenates of the cells transfected with hNAAA were allowed to react with these two ceramides, the enzyme showed low but significant ceramide hydrolyzing activities (0.033 Ϯ 0.004 and 0.014 Ϯ 0.002 nmol/min/mg protein, respectively) (Fig. 5, A, lane 3 and B).
Glycosylation and Processing of NAAA-To analyze possible glycosylation and post-translational processing of NAAA, the C-terminally hexahistidine-tagged human NAAA was overexpressed in HEK293 cells. Upon Western blotting of cell homogenates with anti-hexahistidine antibody, a singlet band around 52 kDa and a doublet band around 39 -42 kDa were detected (Fig. 6A). The cell homogenates of HEK293 cells transfected with insert-free vector gave no detectable bands. When NAAA-transfected cells were cultured in the presence of tunicamycin, an inhibitor of N-type protein glycosylation, the 52-kDa band became hardly detectable, whereas the 39 -42-kDa doublet band was slightly enhanced. As described above, the molecular mass of human NAAA was presumed to be 40.1 kDa based on the deduced primary structure and that of the recombinant NAAA with a hexahistidine tag and a spacer peptide used in this experiment was calculated to be 43.4 kDa. Thus, in this expression system NAAA appeared to be expressed as both glycosylated (52 kDa) and unglycosylated forms (39)(40)(41)(42). Most interestingly, the tunicamycin treatment caused a considerable decrease in specific enzyme activity (Fig. 6B), suggesting that glycosylation is necessary for the full enzyme activity.
Because the N-terminal sequence of NAAA purified from rat lung began with Cys 131 (Fig. 1A), it was suggested that rat lung NAAA is proteolytically cleaved at this position. Therefore, it was likely that recombinant human NAAA expressed in HEK293 cells was also subjected to proteolytic cleavage at the corresponding position, resulting in the generation of a 29.7-kDa peptide backbone including the tag and spacer. However, the molecular mass of the unglycosylated form (39 -42 kDa) observed in Fig. 6A suggested that such a cleavage did not occur and, if any, only a short peptide (1-4 kb) was removed from the full-length NAAA (43.4 kDa). We then examined whether the recombinant NAAA could be cleaved in a cell-free system. The hexahistidine-tagged NAAA partially purified from HEK293 cells (a specific enzyme activity, 104 nmol/ min/mg protein) was used for this purpose. As analyzed by Western blotting with anti-hexahistidine antibody, NAAA in this partially purified enzyme preparation existed dominantly as the 52-kDa glycosylated form with a faint band at ϳ33 kDa (Fig. 6C, lane a). The unglycosylated band at 39 -42 kDa found in Fig. 6A was hardly detectable because this form of NAAA was not solubilized from the 12,000 ϫ g pellet by freezing and thawing. When the sample was incubated at pH 4.5 for 60 min, the 52-kDa form time-dependently decreased, whereas the 33-kDa form increased, suggesting that the 52-kDa form was converted to the 33-kDa form (Fig. 6C, lanes a-c). However, this conversion was hardly observed at pH 7.4 (Fig. 6C, lane d). Similar analyses were performed with the enzyme treated with PNGase F (an endoglycosidase cleaving all asparagine-linked oligosaccharides). Before the incubation at pH 4.5, a major band at ϳ40 kDa and a minor band at ϳ30 kDa were detected (Fig. 6C, lane e). The former band was then time-dependently converted to the latter band at pH 4.5 but not at pH 7.4 (Fig.  6C, lanes e-h). Therefore, these two bands were presumed to be the unglycosylated forms of the 52-and 33-kDa band, respec-tively. The molecular mass of the processed form (ϳ30 kDa) suggested that this cleavage occurred at or near Cys 131 , because the molecular mass of the resultant peptide (including a hexahistidine tag and a spacer) was calculated to be 29.7 kDa as mentioned above. This cleavage by acid treatment, however, did not cause an obvious change in specific enzyme activity (Fig. 6D). In this assay, the enzyme reaction was performed only for 5 min. During the 5-min incubation, the proteolytic cleavage did not proceed significantly (data not shown). These results revealed that a specific proteolytic cleavage of recombinant NAAA occurred at acidic pH, but this processing did not appear to be involved in the regulation of the enzyme activity.
Intracellular Localization of NAAA-We determined the intracellular localization of human NAAA expressed in HEK293 cells using an NAAA-GFP fusion construct. By fluorescence microscopy, the expression of the fusion protein was associated with vesicular elements in the cytoplasm (Fig. 7A). In a time- lapse movie of living cells, these GFP-positive vesicles showed bidirectional linear movements (data not shown). The morphology and intracellular movement of these GFP-positive vesicles appeared to be typical of lysosomes. Upon tunicamycin treatment, the GFP signals on such lysosome-like vesicles were largely diminished. Instead, NAAA-GFP fusion proteins localized in a portion of the perinuclear Golgi region, which strongly suggested that this treatment caused retention of the enzyme in Golgi apparatus (Fig. 7B).
In order to clarify whether NAAA in lysosomes can actually utilize NAEs, we examined the capability of the intact HEK293 cells overexpressing NAAA to degrade N-[ 14 C]palmitoylethanolamine. As shown in Fig. 8, NAAA-expressing cells degraded N-palmitoylethanolamine much faster than control cells, and this degradation was inhibited by the addition of N-cyclohexanecarbonylpentadecylamine, a specific NAAA inhibitor (42). These results suggested the presence of passive or active transport of NAEs to lysosomes.
Organ Distribution of NAAA mRNA-The organ distribution of NAAA mRNA was investigated in the rat by RT-PCR (Fig. 9). NAAA mRNA was widely distributed among the various organs tested. The highest expression was observed in lung, followed by several organs including thymus, spleen, colon, and cecum. This distribution was in good agreement with that of NAAA activity as we reported previously (40). It should be noted that the mRNA distribution was largely different from that of FAAH, which was abundant in liver, brain, testis, and small intestine of rats (30). DISCUSSION Considering the biological actions of various NAEs, including anandamide (9,53) and N-palmitoylethanolamine (1,11,12), it is important to understand the metabolism of NAEs in detail. Intracellular hydrolysis of NAEs has been observed in a variety of mammalian organs and cell lines and has been attributed mostly to the catalysis by FAAH (30 -33). Analysis of FAAHdeficient mice revealed that this enzyme plays a crucial role for the degradation of anandamide and other NAEs in the brain (36 -38). Alternatively, we previously found a second NAEhydrolyzing amidohydrolase, referred to as NAAA here, in CMK cells and various rat tissues such as lung, spleen, and macrophages (39,40). The activity of this enzyme was the highest around pH 5 and was stimulated by DTT and Triton X-100. NAAA was most active with N-palmitoylethanolamine among the various NAEs tested. These catalytic properties were in contrast to those of FAAH which is most active at pH 8.5-10, is insensitive to DTT, and prefers anandamide to Npalmitoylethanolamine as substrate (39). Furthermore, NAAA was much less sensitive to phenylmethylsulfonyl fluoride and methyl arachidonyl fluorophosphonate, which are potent FAAH inhibitors (39).
Here we determined the N-terminal sequence of NAAA purified from rat lung, and for the first time we identified the cDNAs encoding NAAA from human, rat, and mouse. With human NAAA overexpressed in HEK293 cells, we examined the catalytic properties of the recombinant enzyme. The enzyme was activated by DTT (Fig. 3A) and Nonidet P-40 (essentially the same as Triton X-100) (Fig. 2B) and showed by far the highest activity with N-palmitoylethanolamine among the tested NAEs (Fig. 2B). Its optimal pH was 4.5 (Fig. 3B). These catalytic properties agreed well with those of the native enzyme, confirming the authenticity of NAAA cDNA. Thus, it was finally demonstrated that NAAA is the second NAE-hydrolyzing enzyme, catalytically and structurally distinguishable from FAAH.
Consistent with the catalytic differences between NAAA and FAAH, cDNA cloning of NAAA in the present study revealed no homology between the amino acid sequences of these two enzymes. FAAH was reported to have not only an amidohydrolase activity but also a high esterase activity for fatty acyl esters such as 2-arachidonoylglycerol (48) and methyl arachidonate (54), whereas rat lung NAAA (40) and recombinant NAAA (the present study) were almost inactive with ester compounds. Based on these differences in the structure and function between the two enzymes, the development of specific inhibitors of NAAA without acting on FAAH should be promising. Indeed, we have recently reported some ester and amide compounds structurally related to N-palmitoylethanolamine, such as Ncyclohexanecarbonylpentadecylamine, to be specific NAAA inhibitors without inhibitory effects on FAAH (41,42). More potent and selective inhibitors will be useful to elucidate the physiological roles of NAAA.
NAAA was essentially identical to a gene product, termed acid ceramidase-like protein, the deduced amino acid sequence of which had been noted to significantly resemble AC. NAAA, which we cloned from CMK cells, was different from acid ceramidase-like protein from the human placenta cDNA library only in one amino acid residue out of 359 residues. Specifically, Leu 334 of acid ceramidase-like protein was replaced by phenylalanine in NAAA. The human gene was reported to be located in the region of 4q21.1 (46). We found that NAAA belongs to the choloylglycine hydrolase family (Pfam PF02275), which includes AC, choloylglycine hydrolase (conjugated bile acid hydrolase), and penicillin V acylase (55,56). These enzymes cleave carbon-nitrogen bonds, other than peptide bonds, in linear amides.
Human AC is a lysosomal enzyme that hydrolyzes ceramide to sphingosine and free fatty acid with an optimal pH at 3.8 -4.3 (52,57). The mature form of AC is a heterodimeric glycoprotein that is composed of unglycosylated ␣-subunit (molecular mass, ϳ13 kDa) and glycosylated ␤-subunit (ϳ40 kDa; peptide backbone, 28 kDa) derived from its single precursor polypeptide (52). Considering the similarity of the primary structure of NAAA to that of AC, it was likely that NAAA is also subjected to glycosylation and proteolytic cleavage during maturation. With the aid of tunicamycin and PNGase F, we could show that the recombinant NAAA is N-glycosylated (Fig.  6). In accordance with this finding, the deduced primary structure of human NAAA had six potential N-glycosylation sites (Asn-Xaa-(Ser/Thr)). The glycosylation appeared to be necessary for the full enzyme activity. The proteolytic cleavage of NAAA was strongly suggested because the N-terminal amino acid residue of the rat lung NAAA was determined to be Cys 131 (Fig. 1). Most interestingly, Cys 131 corresponded with Cys 143 of human AC, which is the N-terminal residue of the ␤-subunit. The cleavage of NAAA was then confirmed by Western blotting with the C-terminally hexahistidine-tagged NAAA, which was specifically cleaved at pH 4.5, but not at pH 7.4, in a cell-free system (Fig. 6). Judging from the molecular mass of the cleaved form, the cleavage appeared to occur at or near Cys 131 . In the case of AC, the mature form comprises an ␣␤-heterodimer through disulfide bond(s) (52). However, between the SDS-PAGE under reducing conditions and that under non-reducing conditions, the electrophoretic mobility of the cleaved form of NAAA did not change (data not shown), suggesting that unlike AC, NAAA does not form a heterodimer connected through disulfide bond(s).
Some lysosomal enzymes such as cathepsin L (58) and tripeptidyl-peptidase I (59) were also reported to be subjected to in vitro processing under acidic conditions and activated by the cleavage. In contrast, our results suggested that the cleavage of NAAA hardly affected the catalytic activity. Therefore, further analyses will be necessary to clarify the physiological significance of the cleavage of NAAA, which may include the alteration of stability and intracellular localization of the enzyme. Moreover, the cleavages of cathepsin L and tripeptidyl-peptidase I were self-catalyzed (59,60), raising the possibility that NAAA is also cleaved self-catalytically.
We showed that the organ distribution of NAAA mRNA in rats was similar to that of the enzyme activity reported previously (40), which was the highest in lung, followed by spleen, small intestine, thymus, and cecum. However, in human tissues the mRNA of acid ceramidase-like protein was shown to be widely distributed with higher expression in the liver and kidney (46). Thus, the organ distribution of NAAA appeared to be different between rat and human. In addition, the distribution of rat NAAA was considerably different from that of rat FAAH, suggesting distinct physiological roles between these two enzymes. As to the intracellular localization, NAAA-GFP fusion protein showed lysosome-like distribution in HEK293 cells (Fig. 7). This observation was consistent with the previous finding (46) that GFP fusion protein of acid ceramidase-like protein was present in a punctate pattern located throughout the cytoplasm of the COS-1 cells. The localization in lysosomes was also supported by the optimal pH of NAAA at 4.5. Tunicamycin treatment caused the retention of the enzyme in Golgilike compartments. This retention could result from the failure of the recognition of the enzyme by mannose 6-phosphate receptor. By analogy with the ceramide accumulation in Farber disease, the inherited deficiency of AC (47,61), it will be of interest to examine whether deficiency of NAAA might lead to accumulation of NAEs.
One of the intriguing findings in the present study is that AC hydrolyzed not only ceramide but also NAE with an optimal pH at 4.5. We also found that NAAA possessed a low but measurable ceramide hydrolyzing activity in addition to the NAE hydrolyzing activity. Thus, NAAA and AC shared the catalytic activities hydrolyzing the amide bonds of NAE and ceramide. However, several differences in the NAE hydrolyzing activity were observed as follows. First, the substrate specificity with respect to the N-acyl groups of NAEs was different; NAAA was by far the most active with N-palmitoylethanolamine followed by N-myristoylethanolamine, whereas AC hydrolyzed N-lauroylethanolamine at the highest rate with very low activity toward bioactive NAEs, including N-palmitoylethanolamine and anandamide. Most notably, the selectivity of AC for Nlauroylethanolamine corresponds to the observation that the ceramide hydrolyzing activity of AC was the highest with Nlauroylsphingosine among the various N-acylsphingosines (52). Second, DTT stimulated the NAE hydrolyzing activity of AC only weakly. Third, Nonidet P-40 caused inhibition of the Nlauroylethanolamine hydrolysis by AC. As described above, DTT and Nonidet P-40 acted as potent stimulators for NAAA. Most interestingly, N-oleoylethanolamine was earlier reported to inhibit ceramidase (62), which might be related to the NAE hydrolyzing activity of AC.
In conclusion, we identified the cDNA of NAAA and confirmed it as the second NAE-hydrolyzing enzyme. NAAA was revealed to be a glycoprotein localizing mainly in lysosomes. Its preference of N-palmitoylethanolamine as substrate suggests the physiological importance of NAAA in the regulation of this compound. Our studies also revealed the structural and functional resemblance between NAAA and AC.