J. Biochem. 112, 350-354 (1992)

Purification, Analysis, and Enzymatic Activity of Recombinant Human Synovial Fluid Phospholipase A2 and N-Terminal Variants Stefania Di Marco,* Fritz Mfirki,'* Hans Hofstetter,' Albert Schmitz,* Jan van Oostrum,* and Markus G. Griitter*' 'Biotechnology Department and "Department of Inflammation, Bone and Allergy, Pharmaceuticals Division, Ciba-Geigy, Ltd., K-681.5.45, CH-4002 Basel, Switzerland Received for publication, April 20, 1992

Recombinant human synovial fluid phospholipase A2 (rPLA2) and several variants with N-terminal sequences modified by addition or deletion of one or two amino acid residues (Ala or Met; Des-Asn 1 , Leu2) have been expressed in mammalian cells and in Escherichia coli, respectively, purified to homogeneity, and characterized. The observed values for the molecular mass of rPLA2 and variants are in complete agreement with the predicted values for a correctly folded structure containing seven disulfide bridges. Moreover, the relative proportions of the various types of secondary structures of the variants of rPLA2, as measured by CD spectroscopy, are similar to that found for native porcine pancreatic PLA2, indicating that the recombinant proteins are correctly folded. Enzymatic activities of rPLA2 with modified N-tennini decreased to 1.3-0.005% of the activity of the mature rPLA2> emphasizing a key role of the N-terminus for catalytic activity.

Phospholipases A2 (PLA2s) constitute a diverse family of enzymes that hydrolyze the sn-2 fatty acyl ester bond of phosphoglycerides liberating free fatty acids and lysophospholipids (2). Based on differences of the primary structure, the secretory forms of PLA2 are classified into group I and group IIPLA 2 (2). While group IPLA 2 of mammalian origin is mainly present in the pancreas, group II enzymes occur in several tissues, such as liver (3) and spleen (4) and are secreted from various cells, e.g. vascular smooth muscle cells (5), renal mesangial cells (6), and blood platelets (7, 8), in response to appropriate stimuli. The latter type of PLA2 is enriched in synovial fluids of patients with rheumatoid arthritis (9) and is thought to be involved in the pathogenesis of this inflammatory disease (10). Despite the structural differences between group I and group II PLA2, their respective active sites and hydrophobic regions are very similar {11). This is confirmed by a recent X-ray structure analysis of PLA2 from bovine pancreas, Crotalus atrox snake venom, and human synovial fluid, demonstrating that overall three-dimensional structures and functionally important residues from the active sites are virtually superimposable (12). The N-terminal region of pancreatic PLA2 is linked to the catalytic site by a system of hydrogen bonds (13) and seems critically involved in binding of the enzyme to the interface of aggregated substrate (14). Modifications of the N-terminus of the group I porcine pancreatic PLA2, e.g. by exchange of Ala-1 with other amino acid residues (15) or by transamination (14) reduced enzymatic activity to various extents. In the present report we extended this study to group II 1 To whom correspondence should be addressed. Abbreviations: CHO, Chinese hamster ovary; Dhfr, dihydrofolate reductase; FCS, fetal calf serum; MEM, modified Eagle's medium; PLAi, phospholipase A, [EC 3.1.1.4]; rPLA,, recombinant human synovial fluid phospholipase A,; TFA, trifluoroacetic acid.

human synovial fluid PLA2 by expressing several variants with modified N-termini by recombinant technique in Escherichia coli and determining their catalytic activities relative to unmodified enzyme expressed by Chinese hamster ovary (CHO) cells. The results are discussed and rationalized by modeling studies using the published structure of PLA2 (23). EXPERIMENTAL PROCEDURES Expression of Genes Coding for Recombinant Human Synovial Fluid PLA2 and Variants—The PLA2 gene (obtained from British Bio-technology, U.K.) was cloned into mammalian expression vector pCAL5mbDhfr (16) and used to transfect dihydrofolate reductase (Dhfr)" Chinese hamster ovary DUKX-Bl cells (17) by the calcium phosphate technique (28). After Dhfr selection, resistant colonies were pooled and grown into large-scale cultures in medium MEMa (Gibco) with 5% FCS (Gibco). For the expression of PLA2 in COS-7 (American type culture collection CRL 1651) cells, the PLA2 gene was transferred to the expression vector pBJ5 (29). COS-7 cells were transiently transfected using DEAE-Dextran. PLA2 gene was cloned into expression plasmid pPL.Mu. Ndel, a derivative of X P L -expression vector pPL.muSMCon (20). E. coli K-12 strain SG936 (21) was transformed with plasmids pPL.Mu.PLA2 (conferring resistance to ampicillin) and pcI857 (conferring resistance to kanamycin; 22) coding for the temperature sensitive ACI867 repressor. Heat-induction was accomplished as described by Buell et al. (20). Genes coding for Ala-rPLA2 and for [Des-Asn1, Leu']-rPLA,, were expressed accordingly. Purification and Analysis of Recombinant Human Synovial Fluid PLA2 and Variants—rPLA2 was purified by affinity chromatography on Heparin-Sepharose, modified after Horigome et al (23) and Hara et al (24). Pooled

350

J. Biochem.

351

Recombinant. Human Synovial Fluid Phospholipase A2 and Variants culture supernatant8 from CHO cells (350 ml) was diluted with 350 ml deionized water and applied at aflowrate of 35 ml/h to an 8.0x1.0cm column of Heparin-Sepharose CL-6B (Pharmacia) equilibrated with buffer A (25 mM Tris-HCl containing 60 mM KC1 at pH 7.5). After rinsing with buffer A, elution was started with a linear salt gradient (KC1, 60 mM to 1.5 M) in 25 mM Tris-HCl buffer at pH 7.5. Peak fractions of PLA2 were pooled, concentrated using Centricon-10 microconcentrators (Amicon) and subjected to reversed phase (RP) HPLC as described below. For the purification of variants of rPLA2 expressed in E. coli, 15 g of cell paste, obtained from 5 liters of cell culture were suspended in 200 ml of 100 mM Tris buffer containing 5 mM benzamidine-HCl and 2.5 mM EDTA at pH 8.0 (lysis buffer) and passed three times through a French press (SLM Aminco) operating at 15,000 psi. The lysate was centrifuged at 10,000 X g for 60 min, the supernatant discarded and the rPLA2 variants isolated from the inclusion bodies containing pellet. The pellet was resuspended in 300 ml of 50 mM Tris-HCl buffer at pH 8.5 (washing buffer), centrifuged at 16,300 X g for 60 min and the resulting pellet resuspended in 100 ml of 100 mM Tris buffer containing 5 mM benzamidine-HCl, 2.5 mM EDTA, 6 M guanidine-HCl, and 1% /?-mercaptoethanol at pH8.0 (solubilization buffer). After solubilization the rPLA2 variants were refolded by the procedure described by van Scharrenburg et al. (25) except that the solution was dialyzed at 4°C for 2 days against 4 liters of 25 mM Tris buffer containing 5 mM CaCl2, 5 mM L-cysteine, and 0.9 M guanidine-HCl at pH 8.0 (refolding buffer). The protein solution was subsequently dialyzed overnight at 4*C against 50 mM sodium acetate buffer containing 200 mM NaCl at pH 4.5. Precipitated material was removed by centrifugation and the supernatant was separated on a Mono-S column (Pharmacia) followed by gel permeation chromatography on a Sephadex G-75 (Pharmacia) column as described by Kramer et aL (8). Peak fractions were pooled and concentrated by ultrafiltration (YM-5 Amicon membrane and Centricon-3 microconcentrators) and stored at — 20°C. Porcine pancreatic PLA2 (Boehringer) was denatured and refolded using the experimental conditions described above for variants of rPLA2 from E. coli. Protein concentrations were determined with the BioRad protein assay, according to the instructions of the manufacturer, with a bovine serum albumin standard. Protein samples were subjected to SDS-PAGE according to Laemmli (26). Proteins were stained with Coomassie Brilliant Blue R-250. SDS-PAGE was also performed using the PhastSystem (Pharmacia). The ready-to-use 20% mini-gels were stained by the silver-staining technique according to the instructions of the manufacturer (Pharmacia). Isoelectric focusing was perfomed with Servalyt-precotes PAG layer pH 3-10 gels (Serva). The protein, as well as the pi-marker proteins pH 3-11 (Serva) were visualized with Serva Blue R (Coomassie Brilliant Blue R-250, Serva). RP-HPLC was performed using a 4 X120 mm Nucleosil C-18 column equilibrated at room temperature with 0.1% trifluoroacetic acid (TFA) in water and developed at 1 ml/ min with a 15-min gradient (15-60% acetonitrile in 0.1% TFA). Protein solutions containing approximately 1-50 JX% Vol. 112, No. 3, 1992

of rPLA2 and variants were injected. Protein was detected by UV absorbance (214 nm). N-terminal amino acid sequencing by automated Edmandegradation, using an ABI 477A protein sequencer was performed on samples collected after RP-HPLC chromatography. The mass was determined by mass spectrometry (MS) using the electrospray ionization technique (27). A PE/ SCIEX-API-HI mass spectrometer was equiped with an Ion Spray Interphase in the electrospray ionization mode. The RP-HPLC purified samples were introduced using the flow injection technique with a rate of 4 ^1/min. Spectra were averaged over the entire elution time of the samples. All samples were in CH3OH/H2O 1 : 1 (0.1% HCOOH). About 1 ng of protein was injected each time. Far-UV circular dichroism spectra of purified samples were measured by means of a JASCO J-720 spectropolarimeter using a circular quartz cell with a path length of 0.1 cm. Each measurement was the average of five repeated scans in steps of 0.1 nm at room temperature. All samples were extensively dialyzed against 10 mM K-phosphate buffer at pH 6.0, and were measured at a protein concentration of 0.06 mg/ml. The protein concentration was estimated after CD measurements. All spectra are background corrected. PLA2 Assays—The determination of PLA2 activity was modified after Marki and Franson (28). Briefly, enzyme was diluted with medium MEMo- with 5% FCS to produce a substrate hydrolysis up to 5% and incubated for 1 h at 37°C in 200 mM Tris-HCl buffer at pH 7.0 with 1.0 mM CaCl2 and [l- u C]oleate-labeled autoclaved E. coli as substrate (5 nmol phospholipid, 7,000 cpm) in an assay volume of 1.0 ml. The reaction was stopped by extraction of the released [14C]oleic acid, separation on a silicic acid column, and radiometry. For the assay of porcine pancreatic PLA2 the mixture (1 ml) containing enzyme diluted with 10 mM Tris-HCl buffer at pH 7.5 in 1 mg/ml fatty acid-free bovine serum albumin, 100 mM Tris/maleate buffer at pH 7.5, 1 mM sodium deoxycholate, 1 mM CaCl2, and 100 nmol (2 nCi) 1 -stearoyl-2- [ 1 - UC] arachidonyl-8n-glycero-3-pho8phochoUne (Amersham) was incubated for 10 min at 37"C. Extraction of released [ u C]arachidonic acid was performed as described above for [ u C]oleic acid. RESULTS Expression of Genes Coding for Recombinant Human Synovial Fluid PLA2 and Variants—rPLA2 with titers up to 500 ng/ml medium was transiently expressed (over a period of 3-5 days) by COS-7 cells, and at 0.7-1.4 ^g/ml for up to several weeks by CHO cells. The expression levels obtained for the various rPLA2 variants in E. coli were very similar. All rPLA2 variants were present as insoluble inclusion bodies (data not shown). Purification and Analysis of Recombinant Human Synovial Fluid PLA2 and Variants—By combining affinity chromatography on Heparin-Sepharose CL-6B and RPHPLC, the rPLA2 from CHO cell culture supernatants was purified to homogeneity, as judged by RP-HPLC (data not shown) and SDS-PAGE (Fig. 1A). An approximately 400-fold purification with a yield of 50-60% was repeatedly achieved in the affinity chromatography step.

S. Di Marco et al.

352 To purify the variants of rPLA2, from E. coli it was necessary to solubilize the inclusion bodies and then to refold the proteins. After refolding the soluble fractions were chromatographed on a Mono-S column. The different enzymes all eluted at about 1 M NaCl. A final gel-filtration step was included yielding homogeneous rPLA2 variants (Fig. IB). About 20 mg of pure protein, for each variant, were obtained from 5 g of starting material. Peaks of the purified samples were collected after RP-HPLC (elution with 30% acetonitrile) and used for N-terminal amino acid sequencing, for determination of the molecular mass and for enzymatic activity assay. The retention times were for the various molecules: 19.20 min forMet-rPLA 2 ; 18.15 for Ala-rPLA2; 17.45 and 18.15 min for [Des-Asn1, Leu2]-rPLA2 and Met-[Des-Asn'( Leu 2 ]rPLA 2 , respectively. The retention time for rPLA2 from mammalian cells was 20.65 min under the same conditions, distinctly different from all the E. coli variants. No protein was eluted at this retention time when variants of rPLAj were chromatographed. Figure 2 shows the N-terminal sequences of the variants and rPLA 2 . rPLA2 from E. coli has the expected N-terminal sequence but is preceded by a methionine without evidence of the presence of a trace of mature enzyme as verified by RP-HPLC. The expressed gene coding for Ala-rPLA2 resulted in a protein starting with the sequence

Ala-rPLA2, without evidence for a Met-Ala-rPLA2 sequence. An additional construct with a deletion instead of an addition, [Des-Asn1, Leu2] -rPLA2, resulted in a hetero-

13,TM

13,ITS

11,000

11,IH

11,

rtO

I1.3T3

11,500

», B

I »•

i

B

1 I M,o« n

& 13.1H

Met-rPLA,

1 5 10 MNLVNFHRMIKLTTGK

Ala-rPLA,

ANLVNFHRMIKLTTGK

rPLA2 from CHO cells

13,715

IJ.ITO

13, t «

M,l«

11,215

11,370

M,1«

m/z

Fig. 1. A: SDS-PAGE (20%) of purified rPLA,. Proteins were stained by the silver-staining technique. Lane 1: rPLA, (0.1 fig); lane 2: marker proteins (0.6 fig), from top, phosphorylase b (92.5 kDa), bovine serum albumin (66.2 kDa), ovalbumin (45 kDa), carbonic anhydrase (31 kDa), soybean trypsin inhibitor (21.5 kDa), and lysozyme (14.4 kDa). B: SDS-PAGE (18%) of purified variants of rPLA,. Proteins were stained with Coomassie Brilliant Blue R-250. Lane 1: Met-rPLA, (3^g); lane 2: Ala-rPLA, (3^g); lane 3: Met[Dea-Asn1, Leu']-rPLA, and [Des-Asn1, Leu']-rPLA, (3^g); Iane4; marker proteins (6 fig) as in A.

[Des-Asn 1 , Leu 2 ]-rPLA 2

13,*»

Fig. 3. Electrospray ionlzation-MS spectra of rPLA, and Met-rPLA,. The observed values for the molecular mass of rPLA, and Met-rPLA, are 13,903.83±0.62 (A) and 14,035.11±0.43 (B), respectively, against calculated values of 13,903.89 and 14,035.09, respectively. The minor peaks correspond to adducts with phosphoric acid or sulfuric acid.

MVNFHRMIKLTTGK VNFHRMIKLTTGK NLVNFHRMIKLTTGK -3-106

PLA2 from platelets and rheumatoid synovial fluid

NLVNFHRMIKLTTGK

Fig. 2. N-termlnal amino acid sequences of variants of rPLA, from E. coli and rPLA, from CHO cells (this report), and PLA, from human platelets and rheumatoid synovial fluid (8).

X (nm) Fig. 4. CD spectra of the variants of rPLA, and porcine pancreatic PLA,. From top: , [Des-Asn1, Leu'J-rPLA, and Met1 [Des-Asn , Leu']-rPLA,; , Met-rPLA,; , Ala-rPLA,; , porcine pancreatic PLA,. J. Biochem.

Recombinant Hitman Synovial Fluid Phospholipase Ai and Variants TABLE I. Specific activities of variants of rPLA, from E. coli, of rPLAi from mammalian cells and of porcine pancreatic PLA,. [l-14C]Oleate-labeled E. coli was utilized as substrate forrPLAj and variants, and l-stearoly-2-[l-MC]arachidonyl-sn-glycero-3-phosphocholine for porcine pancreatic PLAj. Specific activity Enzyme (ji mol • min"' • mg~') Met-rPLA, Ala-rPLA, Met-[Des-Asn', Leu']-rPLA, [Des-Asn1, Leu2]-rPLA, rPLA, from CHO cells Porcine pancreatic PLA, Denatured and refolded porcine pancreatic PLAj

0.90±0.095 (n = 9) 0.79±0.0€ (n = 2) 0.13±0.004 (n = 3) 0.0037 68.7±7.3 (n = 13) 0.33±0.05 (n = 3) 0.40±0.03 (n = 3)

geneous N-terminus: Met-[Des-Asn1, Leu2]-rPLA2, as the largest fraction, and a minor fraction containing [Des-Asn1, Leu2]-rPLA2. The various molecules from E. coli showed an isoelectric point value higher than 10.65 (data not shown). The observed values for the molecular mass of the different molecules as determined by electrospray ionization-MS corresponded to the calculated mass: rPLA2 (Fig. 3A), Met-rPLA2 (Fig. 3B). Data from other variants are not shown. The difference between the observed and the calculated mass was less than 1 Da. Figure 4 shows that the far-UV CD spectra of the variants of rPLA2 and porcine pancreatic PLA2 are similar. A control refolding experiment was performed using porcine pancreatic PLA2 and the same refolding protocol as used for the variants of rPLA2 from E. coli, to verify that the refolding procedure adopted results in active enzymes. The enzymatic activity of denatured and refolded porcine pancreatic PLA2 was the same as the activity of the untreated protein (Table I). Enzymatic Activity—The relative specific activity of Met-rPLA2 was 1.3% of rPLA2 from mammalian cells, assayed under the same conditions and in parallel, as shown in Table I. For Ala-rPLA2 the relative specific activity was 1% and for Met-[Des-Asn1, Leu2]-rPLA2 and [Des-Asn1, Leu2]-rPLA2 0.2 and 0.005%, respectively. DISCUSSION The N-terminal amino acid sequences of native human synovial fluid PLA2 and the recombinant enzyme expressed by mammalian CHO cells are identical and start with Asn-1 (29, this report). As shown in Fig. 2, the sequence of the recombinant enzyme expressed by E. coli is preceded by an extra methionine residue. Protein synthesis in E. coli is under normal circumstances initiated at the translation initiation codon AUG, coding for methionine. However, post-translational modification occurs in a significant fraction of the proteins expressed by E. coli, since only 40% of the polypeptides of a cytosolic extract of this microorganism retain the N-terminal methionine (30). Excision of N-terminal methionine in proteins of E. coli is catalyzed by the cytoplasmic enzyme methionyl-aminopeptidase; the extent of excision is governed by the length of the side chain of the penultimate amino acid, as shown by Hirel et aL (31) and confirmed by Dalbfige et aL (32). These authors found complete methionine processing in proteins with an initiating short side chain amino acid, e.g. Ala, no processing with Vol. 112, No. 3, 1992

353

long side chain amino acid, e.g. Arg, and minimal processing (0 to 18%) with Asn, Asp, Leu, or lie in the penultimate position. In line with these findings, porcine and human pancreatic PLA2, containing a N-terminal Ala, are directly processed in vitro to the mature proteins when expressed in E. coli (33, 34), whereas cobra venom and human synovial fluid PLA2 both retain the initiating methionine preceding Asn-1 (34, this report). When we observed a very low enzymatic activity of Met-rPLA2 (1.3% of rPLA2, Table I), it was essential to verify that this activity was not due to contamination of Met-rPLA2 by a trace of rPLA2 released by methionylaminopeptidase. This possibility could definitely be excluded by purifying Met-rPLA2 by RP-HPLC prior to enzyme assay. As written above, RP-HPLC allows unambiguous separation of Met-rPLA2 from rPLA2 (as well as from the other recombinant proteins studied). The mass spectrum of Met-rPLA2 (Fig. 3B) provides further evidence for the absence of rPLA2 in the sample. In order to have confidence in the refolding procedure used for the rPLA2 variants from E. coli, we have denatured and refolded porcine pancreatic PLA2 under the same conditions. Reduced porcine pancreatic PLA2 was reoxidized into its native structure yielding full enzymatic activity (Table I). Further support that rPLA2 and the variants have a native -like fold is the close correspondence of the observed value for the molecular mass as determined by electrospray ionization MS, and the calculated theoretical mass (Fig. 3, A and B). The observed differences of less than 1 Da indicated that all cysteines must be involved in disulfide bridges. The formation of the correct intramolecular interactions, including disulfide bridges, is fully supported by CD spectroscopy. CD spectra of the variants (Fig. 4) show that the relative proportions of the various types of secondary structure are similar to that of natural porcine pancreatic PLA2, indicating that the recombinant proteins are correctly folded. The above considerations allow to conclude that the observed low enzymatic activities of the N-terminal variants of rPLA2 are due to homogeneous and correctly folded proteins and may therefore be entirely attributed to the modified N-termini. The essential role of the N-terminus of group IPLA 2 for interfacial binding to aggregated substrate and—consequently—catalytic activity has long been recognized (15, 14, 35). The present study provides supporting data from new N-terminal variants of a group IIPLA 2 from human synovial fluid. We observed that addition of a single amino acid, Met or Ala, reduced enzymatic activity to approximately 1% of the mature enzyme (Table I). An even more drastic reduction resulted from the deletion of the two N-terminal residues Asn-1 and Leu-2 (0.005%), or replacement of Asn-1 and Leu-2 by Met (0.2%). These results, together with the remarkable similarity of the three-dimensional structure of group II human synovial fluid PLA2 and group I PLAj (12) suggest an analogous, if not identical, functional role of the N-termini of these enzymes. The catalytic mechanism of PLA2 is believed to be initiated by the extraction of a proton from a water molecule by the Ntfl of the active site His-48 (36). The extracted proton is later deposited on the leaving lysophos-

354

S. Di Marco et aL

pholipid. The positive charge, acquired by the enzyme when it dissociates the proton from the attacking water molecule, is stabilized through a hydrogen bond between His-48 and Asp-99. The catalytic network is further extended by a water molecule that links the amino terminus to the 0