Enzymological analysis of the tumor suppressor A-C1 reveals a novel group of phospholipid-metabolizing enzymes.

A-C1 protein is the product of a tumor suppressor gene negatively regulating the oncogene Ras and belongs to the HRASLS (HRAS-like suppressor) subfamily. We recently found that four members of this subfamily expressed in human tissues function as phospholipid-metabolizing enzymes. Here we examined a possible enzyme activity of A-C1. The homogenates of COS-7 cells overexpressing recombinant A-C1s from human, mouse, and rat showed a phospholipase A1/2 (PLA1/2) activity toward phosphatidylcholine (PC). This finding was confirmed with the purified A-C1. The activity was Ca2+ independent, and dithiothreitol and Nonidet P-40 were indispensable for full activity. Phosphatidylethanolamine (PE) was also a substrate and the phospholipase A1 (PLA1) activity was dominant over the PLA2 activity. Furthermore, the protein exhibited acyltransferase activities transferring an acyl group of PCs to the amino group of PEs and the hydroxyl group of lyso PCs. As for tissue distribution in human, mouse, and rat, A-C1 mRNA was abundantly expressed in testis, skeletal muscle, brain, and heart. These results demonstrate that A-C1 is a novel phospholipid-metabolizing enzyme. Moreover, the fact that all five members of the HRASLS subfamily, including A-C1, show similar catalytic properties strongly suggests that these proteins constitute a new class of enzymes showing PLA1/2 and acyltransferase activities.


Construction of expression vectors
The cDNAs encoding C-terminally FLAG-tagged A-C1s of human, mouse, and rat were amplifi ed by PCR with Human Testis By the latter activity these proteins catalyzed N -acylation of the amino group of phosphatidylethanolamine (PE) and O -acylation of lysophospholipid, using PC as an acyl donor in both of the reactions. Enzymatic N -acylation of PE results in the formation of N -acyl-PE, known as precursors of anandamide and other bioactive N -acylethanolamines ( 21,22 ). Duncan et al. also reported PLA 2 activity of H-rev107 and referred to it as adipose-specifi c PLA 2 ( 23 ).
A-C1 was originally cloned by differential display between two mouse cell lines, embryonic fi broblast C3H10T1/2 and chondrogenic ATDC5, and was noted due to its structural similarity to the tumor suppressor H-rev107 ( 16 ). The expression of A-C1 in Ras-transformed NIH3T3 cells caused an increase in the number of fl at colonies and inhibition of cell growth ( 16 ). Therefore, this gene was considered to be a tumor suppressor gene. The human homolog was later cloned ( 24 ), and the methylation of the A-C1 gene in human gastric cancers was reported ( 25 ). Because all the other members of the HRASLS subfamily show phospholipidmetabolizing activities and because histidine-30 and cysteine-119 of A-C1 correspond to the catalytic dyad of LRAT ( 26 ) ( Fig. 1 ), we hypothesized that A-C1 also functions as an enzyme involved in phospholipid metabolism. Here, we investigate the catalytic properties of the A-C1 protein and compare them with those of the other members of the HRASLS subfamily (H-rev107, TIG3, HRASLS2, and iNAT). On the basis of the obtained results, we propose to redefi ne this class of tumor suppressor as a novel group of phospholipid-metabolizing enzymes. The deduced amino acid sequences of the members of the HRASLS subfamily. The deduced amino acid sequences of human members of the HRASLS subfamily were aligned using the program GENETYX-MAC (version 15). Closed and shaded boxes indicate identity in all fi ve and any three or four polypeptides, respectively. The highly conserved histidine and cysteine residues and the sequence NCEHFV are indicated by asterisks and an underline, respectively. in 100 l of 50 mM Tris-HCl (pH 8.0), 2 mM DTT, and 0.1% Nonidet P-40 at 37°C for 30 min. The reaction was terminated by the addition of 320 l of a mixture of chloroform-methanol (2:1; v/v) containing 5 mM 3(2)-t -butyl-4-hydroxyanisole. After centrifugation, 100 l of the lower fraction was spotted on a silica gel thin-layer plate (10 cm height) and developed at 4°C for 25 min either in chloroform-methanol-28% ammonium hydroxide (80:20:2; v/v) for the PE N -acylation assay or in chloroform-methanol-H 2 O (65:25:4; v/v) for PLA 1/2 and lyso PC O -acylation assays. The distribution of radioactivity on the plate was quantifi ed using a BAS1500 bioimaging analyzer (FUJIX Ltd., Tokyo).

Western blotting
Samples (20 g protein) were separated by SDS-PAGE on 14% gel and electrotransferred to a hydrophobic polyvinylidene difluoride membrane (Hybond P). The membrane was blocked with PBS containing 5% dried milk and 0.1% Tween 20 (buffer B) and then incubated with anti-FLAG antibody (1:2,000 dilution) in buffer B at room temperature for 1 h, followed by incubation with HRP-labeled secondary antibody (1:4,000 dilution) in buffer B at room temperature for 1 h. FLAG-tagged proteins were visualized using an ECL Plus kit and analyzed using an LAS-1000plus lumino-imaging analyzer (FUJIX Ltd.).
Marathon-Ready™ cDNA, mouse brain cDNA, and rat testis cDNA, respectively, as templates. The mouse and rat cDNAs were prepared from 5 g of total RNA using Moloney murine leukemia virus reverse transcriptase and random hexamer. The primers used were the forward primers containing the Spe I site 5 ′ -CG-CACTAGTCCAAGATGGCGTTTAATGATTGCTTCAGT TTG-3 ′ (human A-C1), 5 ′ -CGCACTAGTCCAAGATGGCGGTAAATGAT-TGCTTC-3 ′ (mouse A-C1), and 5 ′ -CGCACTAGTCCAAGATGGC-GGTTAACGATTGCTTCAGTC-3 ′ (rat A-C1), and the reverse primers containing an in-frame FLAG sequence and the Not I site The cDNA encoding C-terminally FLAG-tagged human FAM84B was amplifi ed by PCR with human kidney cDNA contained in human MTC™ Panel I as a template. The primers used were the forward primer 5 ′ -CGCGGATCCGGAAAATGGGCAACCAGGTGGAGAA-ATTGA-3 ′ containing a Bam HI site and the reverse primer 5 ′ -CG-CGAATTCTCACTTATCGTCGTCATCCTTGTAATCGTGTGC-CACTGCCTCTCCGTCCTCC-3 ′ containing an in-frame FLAG sequence and an EcoR I site. PCR was carried out with KOD-Plus DNA polymerase for 30 cycles at 95°C for 20 s, 56°C for 20 s, and 72°C for 60 s in 5% (v/v) Me 2 SO. The obtained DNA fragments were subcloned into the corresponding restriction enzyme sites of pEF6/myc-His vector. All constructs were sequenced in both directions using an ABI 3130 Genetic Analyzer (Applied Biosystems Life Technologies; Carlsbad, CA).

Overexpression and purifi cation of recombinant proteins
COS-7 cells were grown at 37°C to 80% confl uency in 100 mm dishes containing Dulbecco's modifi ed Eagle's medium with 10% fetal calf serum in a humidifi ed 5% CO 2 and 95% air incubator. The expression vector harboring A-C1 or FAM84B cDNA was introduced into COS-7 cells using Lipofectamine 2000 according to the manufacturer's instruction. Forty-eight hours after transfection, cells were harvested and sonicated three times each for 3 s in 20 mM Tris-HCl (pH 7.4). For the purifi cation of recombinant FLAG-tagged human A-C1, cytosolic fractions were prepared from the cells grown in ten 100 mm dishes by centrifugation of the homogenates at 105,000 g for 55 min at 4°C and mixed with 1 ml of anti-FLAG M2 affi nity gel preequilibrated with 50 mM Tris-HCl (pH 7.4) containing 150 mM NaCl and 0.05% Nonidet P-40 (buffer A). After overnight incubation at 4°C under gentle mixing, the gel was packed into a column and washed three times each with 12 ml of buffer A. The FLAG-tagged protein was eluted with buffer A containing 0.1 mg/ml of FLAG peptide and every 0.5 ml fraction was collected. The purifi ed recombinant human iNAT was prepared as described previously ( 20 ). The protein concentration was determined by the method of Bradford with BSA as a standard (29). functional expression of rat A-C1 with pcDNA3.1(+) as an eukaryotic expression vector ( 17 ). We therefore constructed the pEF6/myc-His expression vector harboring the A-C1 cDNA of either human, mouse, or rat with a FLAG tag at the C terminus, and transiently expressed recombinant proteins in COS-7 cells. Based on the amino acid sequences, the molecular masses of the tagged proteins of A-C1s were calculated to be 19,745 (human), 19,778 (mouse), and 19,516 (rat) Da, respectively. When analyzed by Western blotting using anti-FLAG antibody, each cell homogenate exhibited an immunopositive band around 19-20 kDa ( Fig. 2B ). Although the band of human A-C1 consistently migrated a little faster than expected, the reason remained unclear. We next assayed the homogenates for PLA 1/2 activity. When the samples were incubated with 1,2-[ 14 C]dipalmitoyl-PC, followed by separation of the reaction products by TLC, the radioactive bands corresponding to palmitic acid and lyso PC were detected ( Fig. 2C ). The activities of the homogenates containing human, mouse, and rat A-C1 were 1.65, 0.47, and 0.75 nmol/min/mg of protein, respectively, whereas the endogenous activity of mock transfectant was 0.16 nmol/min/mg of protein ( Fig. 2D ). These results suggested that A-C1 possesses PLA 1/2 activity. We examined a possible secretion of recombinant human A-C1 into the culture medium by measuring PLA 1/2 activity. Consistent with the lack of the signal sequence for the secretory pathway in its primary structure, the activity was not detected in the culture medium of COS-7 cells expressing human A-C1 (data not shown). Although A-C1 was found as a tumor suppressor gene ( 16 ), its transient expression in COS-7 cells did not show an obvious effect on cell proliferation and viability (data not shown).

Characterization of the purifi ed human A-C1
To further analyze the enzymatic properties of A-C1, we prepared cytosolic fractions from the COS-7 cell homogenate by ultracentrifugation and purifi ed the C-terminally FLAG-tagged human A-C1 protein from the cytosol by anti-FLAG antibody-conjugated column chromatography. As analyzed by SDS-PAGE, a nearly homogenous protein band was seen around 19 kDa ( Fig. 3A ). The specifi c PLA 1/2 activity of the purifi ed protein was 182 nmol/min/ mg of protein, which was 110-fold higher than that of the A-C1-expressing cell homogenate. The optimal pH was around 8 ( Fig. 3B ). The PLA 1/2 activity increased up to 246 nmol/min/mg of protein, depending on the concentrations of the substrate PC, with an apparent K m at 80 M ( Fig. 3C ). We also examined the effects of several factors on the PLA 1/2 activity. The addition of 1 mM and 5 mM Ca 2+ reduced the activity by 9.9% and 21.7%, respectively. On the other hand, 1 mM EDTA increased the activity by 14.3% ( Fig. 3D ). In the absence of the sulfhydryl reducing reagent DTT, the activity was hardly detected ( Fig. 3E ). In agreement with this stimulatory effect of DTT, 5 mM iodoacetate, an irreversible sulfhydryl blocker, acted as an inhibitor. The standard reaction mix also contained 0.1% Nonidet P-40 (a nonionic detergent). Removal of the detergent decreased the activity by 94.5%. These effects of

Functional expression of A-C1 proteins
We previously cloned cDNA of A-C1 (tentatively termed RLP-2) from rat testis (GenBank™ accession number AB510983) ( 17 ). In the present study, we also cloned cDNAs of the counterparts from human testis and mouse brain (AB510981 and AB510982) based on the reported nucleotide sequences. Their sequences we determined were completely identical to those reported previously. The primary structures of A-C1 proteins were composed of 168 (human), 167 (mouse), and 167 (rat) amino acid residues, respectively ( Fig. 2A ). The alignment revealed their high homology to each other (85, 83, and 96% identity at amino acid level between human and mouse, between human and rat, and between mouse and rat, respectively). The putative catalytic dyad was completely conserved as histidine-30 and cysteine-119. We previously failed in the A -C1 showed a remarkable preference of sn-1 position over sn -2 position. A similar result was obtained using 1-[ 14 C]palmitoyl-2-palmitoyl-PC as a substrate (data not shown). 1-Palmitoyl-2-[ 14 C]arachidonoyl-PE was also an active substrate, and PLA 1 activity was again higher than PLA 2 activity ( Table 1 ). On the other hand, lysophospholipase activities for 1-[ 14 C]palmitoyl-lyso PC or 2-[ 14 C]palmitoyl-lyso PC were not detected.
We next examined transacylation activities of the purifi ed A-C1. When [ 14 C]dipalmitoyl-PC and nonradioactive dioleoyl-PE were used as a donor substrate and an acceptor substrate, respectively, a radioactive band corresponding to N -palmitoyl-PE was detected on the TLC plate ( Fig.  4A ). In the absence of nonradioactive PE, this band was not detected (not shown). These results showed that A-C1 possesses an N -acyltransferase activity for PE. Moreover, when the protein was allowed to react with nonradioactive dipalmitoyl-PC and radioactive lyso PC (either 1-[ 14 C] palmitoyl-lyso PC or 2-[ 14 C]palmitoyl-lyso PC) as a donor substrate and an acceptor substrate, respectively, 14 C-labeled PC was formed from both of the lyso PCs ( Fig. 4B ). 2-[ 14 C]palmitoyl-lyso PC was a much more active substrate than 1-[ 14 C]palmitoyl-lyso PC, as shown in Fig. 4C . With the aid of our previous results ( 19,20 ), we compared catalytic properties of human A-C1, iNAT, H-rev107, HRASLS2, and TIG3, all of which belong to the HRASLS subfamily ( Fig. 4C ). All the purifi ed recombinant proteins showed PLA 1/2 activity, the highest being with H-rev107 and the lowest with iNAT. As for the PE N -acyltransferase activity, iNAT, HRASLS2, and A-C1 showed similar levels of activities, whereas H-rev107 and TIG3 were much less active. These fi ve proteins also showed O -acyltransferase activities toward both 1-[ 14 C]palmitoyl-lyso PC and 2-[ 14 C]palmitoyllyso PC. The latter lyso PC was consistently a more-active acceptor substrate than the former lyso PC. In particular, A-C1 and HRASLS2 showed high O -acyltransferase activities.

Metabolic labeling of A-C1-expressing cells
We metabolically labeled the COS-7 cells transiently expressing human A-C1 with [ 14 C]palmitic acid. When the extracted lipids were analyzed by TLC, we detected a clear radioactive band corresponding to N -palmitoyl-PE ( Fig.  5A, C ). This band was not observed in the control COS-7 cells. These results suggested that A-C1 actually functions as N -acyltransferase in the living cells. On the other hand, we did not see an obvious change in the levels of radioactive bands corresponding to free palmitic acid and lyso PC (expected products of PLA 1/2 ) ( Fig. 5A-C ). The levels of bands corresponding to PC and PE (potential substrates of Ca 2+ , DTT, iodoacetate, and Nonidet P-40 were similar to the catalytic properties of H-rev107, TIG3, and HRASLS2, which we reported previously ( 19 ).
Because [ 14 C]dipalmitoyl-PC that to this point we used as a substrate was radiolabeled on both sn -1 and sn -2 palmitoyl chains, we referred to the hydrolysis activity as PLA 1/2 activity. To distinguish PLA 1 activity from PLA 2 activity, we next used [ 14 C]PC radiolabeled only on the sn -2 palmitoyl chain (1-palmitoyl-2-[ 14 C]palmitoyl-PC). As shown in Table 1 ,  The purifi ed recombinant human A-C1 (0.13 g protein) was allowed to react with the indicated glycerophospholipids at 200 M. Mean values ± SD are shown ( n = 3). examined by semi-quantitative real-time PCR. The highest A-C1/GAPDH ratio was found in testis (53.5), followed by skeletal muscle (2.4), brain (1.1), and heart (0.7). Other human tissues showed lower levels. The dominant expression in these four tissues was also observed with mouse and rat. Low levels of its expression were detected in lung, stomach, kidney, and colon of mouse, and thymus, lung, and small intestine of rat. Such a relatively high expression in the limited tissues was similar to the dominant expression of iNAT in testis ( 17,20,30 ), but was different from ubiquitous expressions of H-rev107 and TIG3 ( 19 ).

Lack of PLA 1/2 activity in FAM84B
FAM84A and FAM84B were found as human genes upregulated in some tumors ( 31,32 ). The deduced amino acid sequences of these two genes are homologous to those of the HRASLS subfamily members as shown in a phylogenetic tree ( Fig. 7A ) and by comparison with the sequence of human A-C1 ( Fig. 7B ). The two sequences exhibit 43.5% identity to each other ( Fig. 7B ). However, serine is substituted for the cysteine residue forming the catalytic dyad in these proteins. We cloned cDNA of FAM84B from human testis and constructed the expression PLA 1/2 ) were also unaltered. These fi ndings may be explained by a low PLA 1/2 activity of A-C1 in the living cells. Another possibility is that the produced [ 14 C]palmitic acid and [ 14 C]lyso PC are quickly incorporated into phospholipids.

Tissue distribution of A-C1
To examine tissue distribution of A-C1 in human, mouse, and rat, reverse transcriptase-PCR was employed ( Fig. 6 ). In human, the levels of A-C1 mRNA were by far the highest in testis and skeletal muscle, followed by brain and heart. The expression levels of human A-C1 were also The dominant PLA 1 activity over PLA 2 activity, as well as the preference of 2-acyl-lyso PC in the lyso PC O -acylation, suggested the involvement of A-C1 in the remodeling at the sn -1 position of glycerophospholipids in a CoA-independent manner. As for tissue distribution, mRNA of AC-1 was highly expressed in testes and skeletal muscles of human, rat, and mouse. In addition, its moderate expression was seen in heart and brain. Our results were in agreement with previous reports that A-C1 was predominantly expressed in skeletal muscle, heart, brain, and bone marrow of mouse ( 16 ), and skeletal muscle, testis, heart, and brain of human ( 24 ). Such a tissue distribution of A-C1 distinguishable from those of other members of the HRASLS subfamily suggests a unique physiological role of this protein.
The present study and our previous studies (17)(18)(19)(20) revealed that all proteins of the HRASLS subfamily possess phospholipid-metabolizing activities. Similarity among their enzymatic properties is in good agreement with high homology among their primary structures ( Fig. 1 ). The histidine and cysteine residues corresponding to the catalytic dyad of LRAT are completely conserved throughout the fi ve proteins. Because subtle structural differences of vector. Although transient expression of the FLAG-tagged recombinant protein in COS-7 cells was confi rmed by Western blotting using anti-FLAG antibody, the cell homogenates did not show a signifi cant PLA 1/2 activity. The same procedure was applied to cDNA cloning and expression of human FAM84A. However, we failed in its expression for unknown reasons.

DISCUSSION
A-C1 was discovered as a mouse protein that shows homology with H-rev107, a class II tumor suppressor, and that inhibits growth of Ras-transformed NIH3T3 cells ( 16 ). Its human homolog was also cloned ( 24 ). Our present studies revealed that this protein is capable of catalyzing PLA 1/2 -like hydrolysis, PE N -acylation, and lyso PC O -acylation. In these three reactions, PC was consistently used as an acyl donor, whereas water, the amino group of PE, and the hydroxyl group of lyso PC were used as acyl acceptors, respectively. The enzymatic properties were similar to those of other tumor suppressors belonging to the HRASLS subfamily (iNAT, H-rev107, HRASLS2, and TIG3) ( 19,20 ). All of these proteins required DTT and Nonidet P-40 for full activities, were sensitive to the inhibition by iodoacetate, and preferred esterolysis at the sn -1 position to that at sn -2 position. As compared with the other members, A-C1 showed a relatively high lyso PC O -acyltransferase activity, and its PE N -acylation activity was as high as those of iNAT and HRASLS2. Metabolic labeling of A-C1-expressing COS-7 cells with [ 14 C]palmitic acid revealed actual generation of N -palmitoyl-PE by A-C1 in the living cells.  6. The expression of A-C1 mRNA in human, mouse, and rat tissues. The expression of A-C1 mRNA in various human, mouse, and rat tissues was examined by reverse transcriptase-PCR. The housekeeping gene GAPDH was used as a control. The expression of A-C1 mRNA in human was also examined by semi-quantitative real-time PCR and was shown in terms of A-C1/GAPDH ratio. 1, brain; 2, thymus; 3, heart; 4, lung; 5, liver; 6, spleen; 7, stomach; 8, kidney; 9, small intestine; 10, colon; 11, testis; 12, skeletal muscle; 13, pancreas; 14, prostate; 15, ovary; 16, placenta; 17, peripheral leukocytes. Fig. 7. The phylogenetic tree of the LRAT family members and the deduced amino acid sequences of human FAM84A and FAM84B. A: The phylogenetic tree composed of human LRAT, FAM84A, FAM84B, and the HRASLS subfamily members was constructed using the program GENETYX-MAC (version 15). B: The deduced amino acid sequences of human A-C1, FAM84A, and FAM84B were aligned using the program GENETYX-MAC (version 15). Closed and shaded boxes indicate identity in all three or any two polypeptides, respectively. The highly conserved histidine and cysteine residues and the sequence NCEHFV of A-C1 are indicated by asterisks and an underline, respectively.
the members should explain different availabilities of acyl acceptor substrates, further investigation will be required to elucidate the structure-function relationship. Contribution of the catalytic activities to their tumor-suppressive activities currently remains unclear. The tumor-suppressing activity was mostly implicated in Ras-transformed cells ( 11,13,16,33 ), and the mutants of TIG3 addressed to the conserved asparagine and cysteine residues failed to induce the apoptosis of HtTA cervical cancer cells, which was caused by the wild-type ( 34 ). Because this cysteine residue functions as the catalytic center, it is possible that the phospholipid-metabolizing activity of the tumor suppressors regulates the function of Ras by altering the membrane structures of microdomains where Ras is specifi cally localized ( 35 ). All the HRASLS subfamily members contain the sequence NCEHFV (amino acids 118-123 in the case of A-C1) ( Fig. 1 ). Human FAM84A and FAM84B show homology to LRAT and the HRASLS subfamily members ( Fig. 6A ). Although both of the proteins contain a sequence similar to the sequence NCEHFV, serine is substituted for the cysteine residue. The lack of PLA 1/2 activity in FAM84B may be related to this substitution. As shown in the phylogenetic tree ( Fig. 6A ), the distinct evolution of LRAT, FAM84B, and the HRASLS subfamily members appears to explain the difference in their catalytic properties. However, we cannot rule out a possibility that FAM84B has another enzyme activity.
To date, various names have been used for each member of the HRASLS subfamily ( Table 2 ). According to the nomenclature proposed by the HUGO Gene Nomenclature Committee, HRASLS1-5 s are assigned to genes for A-C1, HRASLS2, H-rev107, TIG3, and iNAT, respectively. Considering that all these proteins possess PLA 1/2 and acyltransferase activities, here we propose to term the products of HRASLS1-5 genes as phospholipase A/acyltransferase (PLA/AT)-1 to -5, respectively ( Table 2 ).
In conclusion, we characterized for the fi rst time the tumor suppressor protein A-C1 as a phospholipidmetabolizing enzyme. Considering that fi ve human members of the HRASLS subfamily, including A-C1, share similar catalytic properties, these proteins appear to form a novel class of enzymes showing PLA 1/2 and acyltransferase activities.
The authors are grateful to Ms. Akiko Yamamoto for technical assistance, and acknowledge technical assistance from the Division of Research Instrument and Equipment and the Division of Radioisotope Research, Kagawa University. iNAT, HRLP5 PLA/AT-5