Protein Expression and Purification 59 (2008) 302–308

Contents lists available at ScienceDirect

Protein Expression and Purification

journal homepage: www.elsevier .com/locate /yprep

Expression and purification of the D4 region of PLD1 and characterizationof its interaction with PED-PEA15

Francesca Viparelli a,b,1, Nunzianna Doti a,c,1, Annamaria Sandomenico a,b, Daniela Marasco a,Nina A. Dathan a, Claudia Miele d, Francesco Beguinot d, Simona M. Monti a, Menotti Ruvo a,*

a Istituto di Biostrutture e Bioimmagini (IBB), CNR, via Mezzocannone, 16, 80134, Napoli, Italyb Dipartimento di Scienze Biologiche, Università di Napoli Federico II, via Mezzocannone, 16, 80134 Napoli, Italyc Dipartimento di Biochimica e Biofisica, Seconda Università di Napoli, Via S.M. di Costantinopoli 16, 80138 Napoli, Italyd Istituto di Endocrinologia ed Oncologia Sperimentale (IEOS), CNR, Via Pansini, 5, 80131 Napoli, Italy

a r t i c l e i n f o

Article history:Received 15 January 2008and in revised form 22 February 2008Available online 29 February 2008

Keywords:PhospholipasePhosphoprotein enriched in diabetes (PED)SPRProtein–protein interaction

1046-5928/$ - see front matter � 2008 Elsevier Inc. Adoi:10.1016/j.pep.2008.02.012

* Corresponding author. Fax: +39 081 2534574.E-mail address: menotti.ruvo@unina.it (M. Ruvo).

1 These authors contributed equally to this work.2 Abbreviations used: PLD, Phospholipase D;DAG, di

terase domain.

a b s t r a c t

PLD’s (Phospholipases D) are ubiquitously expressed proteins involved in many transphosphatidylationreactions. They have a bi-lobed structure composed by two similar domains which at their interfacereconstitute the catalytic site through the association of the two conserved HxKx4Dx6GSxN motifs.PLD1 interacts with the small phosphoprotein PED-PEA15 by an unknown mechanism that, by enhancingPLD1 stability, apparently increases its enzymatic activity; the minimum interacting region of PLD1 waspreviously identified as spanning residues 712–1074 (D4 region). Since the D4/PED-PEA15 interactionhas been claimed to be one of the multiple molecular events that can trigger type 2 diabetes, we purifiedthe two recombinant proteins to study in vitro this binding by both ELISA and SPR techniques. WhilstPED-PEA15 was easily expressed and purified, expression of recombinant D4 was more problematicand only the fusion protein with Thioredoxin A and a six Histidine Tag (Trx-His6-D4) demonstrated suf-ficient stability for further characterization. We have found that Trx-His6-D4 is present as two differentoligomeric forms, though only the monomeric variant is able to interact with PED-PEA15. All these find-ings may have important implications for both the mechanisms of phospholipase activity and PED-PEA15regulative functions.

� 2008 Elsevier Inc. All rights reserved.

Phospholipase D (PLD2) is a protein present in almost all organ-isms. It is implicated in a wide range of cellular processes, most ofwhich involve the production of phosphatidic acid from which orig-inates diacylglycerol (DAG), a potent protein kinase C activator, andlysophosphatidic acid, a strong mitogen also involved in lipid syn-thesis [1,2]. Two mammalian isoforms of PLD (PLD1 and PLD2) areknown and, although they share several common domains and prop-erties, they exhibit different catalytic and regulatory capacities [3,4].They both contain two invariant highly charged HxKx4Dx6GSxN mo-tifs (HKD), essential for PLD activity [5], which characterizes the lar-ger enzyme superfamily of phospholipases. The superfamily includesnucleases [6] and phosphodiesterases [7] and most members containtwo copies of the HKD motif that together form the enzyme activesite [6–8]. Structurally, PLDs are bi-lobed proteins, where each lobeis constituted by a phosphodiesterase domain (PD)-containing aHKD motif; intramolecular association of the PD is required to juxta-

ll rights reserved.

acylglycerol;PD, phosphodies-

pose the HKD motifs and reconstitute the catalytic core [6,8]. Thecrystallographic structure of a PLD from Streptomyces sp. strainPMF [8], has also shown that the two PD, in spite of a very poor pri-mary structure conservation, have a common 3D organization, sug-gesting the hypothesis that they are generated by a geneduplication event [9]. The structural analysis of the smaller bacterialendonucleases, Nuc, constituted by a single PD, has shown that itneeds to dimerize to become catalytically active [6], thus furtherconfirming the hypothesis that the catalytic activity is driven bythe intermolecular or intramolecular association of similar or identi-cal HKD-containing PD. Consistently, it has been reported that indi-vidually expressed recombinant N- and C-terminal halves of PLD1,each containing a PD, are catalytically inactive, but, when co-ex-pressed, PLD activity is reconstituted and a direct interaction be-tween the two halves can be observed [10,11]. Furthermore, theinteraction occurs only when N- and C-termini are co-expressedbut not when they are separately expressed and mixed in vitro. Theseobservations suggest that self-association likely occurs during trans-lation and may involve specific folding processes [3,11–13]. PLDrequirement of intramolecular association to form a catalytically ac-tive enzymes suggests that an intermolecular interaction may alsooccur between full length proteins. Indeed, co-localization studies

F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308 303

and co-immunoprecipitation of equal amounts of differentiallytagged PLD1s imply that there is an actual dimerization betweenthese molecules [12]. However, at present, the biological significanceof PLD dimerization is still unknown.

It has been found that human PED-PEA15 (Phosphoprotein En-riched in Diabetes-Astrocytes) binds to PLD1, lengthening its per-sistence in the cell and thus increasing its enzymatic activity[14–17]. PED-PEA15 is a widely distributed protein involved inseveral cellular processes [18,19], including the pathogenesis oftype II diabetes [15,16]. In fact, in PED-PEA15 overexpressing cellsthe increased activity of PLD1 causes PKC-a activation, induced byan increase of DAG levels. PKC-a activation, in turn, downregulatesinsulin induction of PKC-f function, impairing insulin-stimulatedglucose transport and Glut4 translocation [15,17]. To date, very lit-tle is known about the PLD1/PED-PEA15 interaction. However byusing a two-hybrid screen, the shortest PED-PEA15-interacting re-gion of PLD1 has been identified and named D4; this region, span-ning residues 712–1074, encompasses the entire second catalyticphosphodiesterase domain of the protein (residues 740–960),including one HKD motif [14]. To further extend our knowledgeon the PLD1/PED-PEA15 interaction, we have optimized theexpression and purification of the D4 domain of PLD1 and charac-terized its binding to PED-PEA15. The oligomeric status of D4 andthe implications this property has in the context of its interactionwith PED-PEA15 have also been investigated.

Table 1Purification tables

Totalvolume(mL)

Totalprotein(mg)

Targetprotein(mg)a

Targetproteinpurity (%)

Yield(%)

Purification of D4MLysate 20 67 1.5 4 100His-trap 2 2.5 1.1 44 73MonoQ-peak 1 0.5 0.17 0.16 95 11

Purification of D4DLysate 20 67 1.5 4 100His-trap 2 2.5 1.1 44 73MonoQ-peak 2 1 0.6 0.36 60 24Superdex-200-peak B 1 0.21 0.21 100 14

Purification of PEDLysateb 4 110 30 27 100His-trapb 2.5 21 19 90 63His-trap after TEV digestionc 3 16 6.5 40 22MonoQc 2 5.8 5.8 100 19

Purification of Trx-His6

Lysate 2 54 2.7 5 100His-trapd 2 1.4 1.2 86 44His-trap after TEV digestione 2 0.6 0.6 100 22

a Amount of the target protein was evaluated from SDS–PAGE gel.b The target protein is His6-GST-PED-PEA15.c The target protein is PED.d The target protein is Trx-His6-RAGE.e The target protein is Trx-His6.

Materials and methods

Materials

Oligonucleotides were synthesized by Sigma–Genosys (Sigma–Aldrich, Milano, Italy). pETM vectors were from EMBL (Heidelberg,Germany). Pfu DNA polymerase was from Stratagene (Milano,Italy). Restriction enzymes were from New England Biolabs (Milan-o, Italy). All molecular biology kits and Ni–NTA Magnetic AgaroseBeads were from Qiagen (Milano, Italy). Escherichia coli bacterialstrains were from Novagen (Milano, Italy). IPTG was from Inalco(Milano, Italy). Reagents for bacterial medium were from Becton-Dickinson (Milano, Italy). All reagents for SDS–PAGE, all chroma-tography columns and ÄKTA FPLC were from GE Healthcare(Milano, Italy). EZ-Link NHS-LC-Biotin Reagent was from PIERCE(Rockford, IL, US). TPCK-treated trypsin, STRV-HRP, Sigma-fastO-phenylenediamine dihydrochloride tablet sets were from Sig-ma–Aldrich (Steinheim, Germany). BIAcore system and related re-agents were from Pharmacia Biosensor AB (Uppsala, Sweden). Allother reagents and chemicals were commercially available by Sig-ma–Aldrich (Steinheim, Germany) or Fluka (Steinheim, Germany).

D4 cloning and expression

D4 was PCR amplified using 50-GCGCGCGAAGACCCATGGGGTCCCTTTCTTATCCTTTTCTGCTT-30 and 50-CGCGCGCTCGAGTCATTATTAAGTCCAAACCTCCATGGGCACTATG-30 oligonucleotides, thendouble digested with BbsI and XhoI restriction enzymes. Digestedfragments were extracted from 0.8% agarose gel using QIAquickGel Extraction Kit and cloned into five different pETM vectors(pETM-11, pETM-20, pETM-30, pETM-52 and pETM-60), previouslyNcoI–XhoI digested and dephosphorylated. Ligation products wereelectroporated into TOP10F’ cells and recombinant colonies wereisolated using QIAprep Spin Miniprep kit. To find out the best con-ditions for expression, 100 ng of the expression vectors werechemically transformed into E. coli competent cells. Recombinantcolonies were grown in 1.5 mL of LB medium up to 0.6 OD/mL at37 �C (OD was monitored at 600 nm) and expression was then in-duced with 1 mM IPTG for 16 h at 22 �C. Bacterial cultures were

harvested, lysed in 50 mM Tris–HCl, 1 mM PMSF, 0.05% TritonX-100, 1 mg/mL lysozyme, pH 8.0 buffer then soluble fractions wereadded to 30 lL of a 5% suspension of QIAgen Ni–NTA Magnetic Aga-rose Beads. Beads were washed with 300 lL of 10 mM potassiumphosphate, 250 mM NaCl, 20 mM imidazole, 8% glycerol, 0.2% TritonX-100, pH 7.8 buffer and then with 300 lL of 10 mM potassiumphosphate, 300 mM NaCl, 0.05% Triton X-100, pH 7.2 buffer. Mag-netic beads were eventually re-suspended in SDS-loading buffer(50 mM Tris–HCl, 5% 2-mercaptoethanol, 1% glycerol, 5% (w/v)SDS buffer, pH 6.8) and, after 50 at 100 �C, loaded on SDS–PAGE gel.

Different E. coli strains, namely BL21(DE3)pLysS, Rosetta(DE3),Rosetta(DE3)pLysS, C41(DE3), C43(DE3), BL21star(DE3), BL21star(DE3)pLysS, BL21-RIP(DE3), BL21-RIL(DE3) and different IPTGconcentrations (0.05, 0.1, 0.2, 0.5 and 1 mM) were tested.

D4 purification

About 1.1 g of wet weight cells were lysed in 20 mL of 50 mMTris–HCl, 200 mM NaCl, 20 mM imidazole, 1 mM PMSF, 0.05% Tri-ton X-100, 1 mg/mL lysozyme, 1 mM TCEP, pH 7.5 buffer for 300 atRT and sonicated for 6 min (200 0 on, 200 0 off). Soluble fraction wasrecovered after centrifugation for 200 at 4 �C at 15 krpm and loadedon a His-trap column (1 mL). Unbound proteins were washed awaywith 50 mM Tris–HCl, 200 mM NaCl, 20 mM imidazole, 1 mMTCEP, pH 7.5 buffer and Trx-His6-D4 was eluted in 50 mM Tris–HCl, 200 mM NaCl, 200 mM imidazole, 1 mM TCEP, pH 7.5. Imidaz-ole and NaCl excess was eliminated by dialysis against 50 mMTris–HCl, 50 mM NaCl, 1 mM TCEP, pH 7.5 buffer and the proteinwas further purified on a MonoQ HR 5/5 column (1 mL), using a0–100% gradient of 50 mM Tris–HCl, 1 mM TCEP, 1 M NaCl, pH7.5 buffer (see Table 1).

LC–MS analysis

LC–MS analyses were performed using 4 lL of a 0.1 mg/mL sam-ple and a LCQ DCA XP Ion Trap spectrometer (ThermoElectron, Mi-lan, Italy). This was equipped with an OPTON ESI source (operatingat a needle voltage of 4.2 kV and a temperature of 320 �C) and a

304 F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308

complete Surveyor HPLC system (including a MS pump, an auto-sampler and a photo diode array [PDA]). Analyses were performedusing a narrow bore 250 � 2 mm C4 Jupiter column, 300 Å, 3 lm(Phenomenex) and eluting with H2O/0.08% TFA (A) and CH3CN/0.05% TFA (B) over a gradient up to 70% B. During this time, thesample tray was kept at 20 �C, whereas the column was kept at25 �C. Mass spectra were recorded continuously at mass intervalsof 400–2000 amu in positive mode. Multicharge spectra were thendeconvoluted using the BioMass program implemented in the Bio-works 3.1 package provided by the manufacturer. Mass calibrationwas performed automatically by means of selected multiplecharged ions, in the presence of a calibrant (Ultramark, Thermo-Electron). All masses were reported as average values.

PED-PEA15 cloning, expression and purification

PED-PEA15 was cloned into a pETM-30 vector, using 50-CGCGCGCCATGGCTGAGTACGGGACCCTCCT-30 and 50-CGCGCGGATCCTTATCAGGCCTTCTTCGGTGGGGGAG-30 oligonucleotides for PCR.Amplified DNA was hydrolyzed using NcoI and BamHI enzymes,then ligated into pETM-30 plasmid; recombinant vector was iso-lated following a procedure analog to that used to clone D4.Expression of His6-GST-PED-PEA15 in BL21(DE3) star cells was in-duced with 1 mM IPTG, for 3 h at 37 �C; yield was about 300 mg/Lof culture. A cell pellet derived from 100 mL of culture (0.5 g) wasresuspended in 4 mL of 20 mM sodium phosphate, 500 mM NaCl,1 mM PMSF, 0.05% Triton X-100, 1 mg/mL lysozyme, pH 7.4 buffer,left for 300 at RT, then sonicated for 2 min (100 0 on, 100 0 off). Solublelysate was loaded on a His-trap column (1 mL), washed with20 mM sodium phosphate, 500 mM NaCl, 60 mM imidazole pH7.4 buffer and His6-GST-PED-PEA15 was eluted with 20 mM so-dium phosphate, 500 mM NaCl, 300 mM imidazole, pH 7.4 buffer.Purified protein was dialyzed against 20 mM sodium phosphate,150 mM NaCl, pH 7.4 buffer and hydrolyzed for 16 h at 22 �C inthe presence of His-tagged TEV protease (expressed and purifiedin our laboratories), using 0.01 mg of the enzyme for each mg ofPED-PEA15. Digested protein was loaded on a His-trap column,previously equilibrated in 50 mM Tris–HCl, pH 7.5 buffer andPED-PEA15 was recovered in the flow-through. PED-PEA15 waseventually purified on a MonoQ HR 5/5 column (1 mL) in a gradientfrom 0% to 100% of 50 mM Tris–HCl, 1 mM TCEP, 500 mM NaCl, pH7.5 buffer (see Table 1).

CD spectra were obtained on a Jasco J-715 circular dichroismspectrometer system, calibrated at 290 nm with an aqueous solu-tion of D(+)-10-camphor sulfonic acid. Data were collected at0.2 nm intervals with a 20 nm min�1 scan speed, a 2 nm band-width and a 16 s response, from 260 to 190 nm.

Trx-His6 expression and purification

About 100 ng of pETM-20 vector were chemically transformedin E. coli BL21(DE3) and a single colony was inoculated in 50 mLof LB. Expression of Trx-His6, fused to a stuffer protein (RAGE),was induced with 1 mM IPTG; after 3 h at 37 �C, cells were har-vested and re-suspended in 2 mL of 50 mM Tris–HCl, 300 mMNaCl, 10 mM imidazole, 1 mg/mL lysozyme, 1 mM PMSF, pH 8.0buffer. After 150 at RT, cells were sonicated for 20 (100 0 on, 100 0

off) and soluble fraction was loaded on 1 mL of chelating sepharosefast flow resin and washed three times with 50 mM Tris–HCl,300 mM NaCl, 40 mM imidazole, pH 8.0 buffer. The fusion proteinwas eluted with 50 mM Tris–HCl, 300 mM NaCl, 500 mM imidaz-ole, pH 8.0. Trx-His6 was then separated from the stuffer protein:digestion in presence of TEV protease (1:100 = TEV:fusion proteinw/w) was followed by a further purification on Chelating Sephar-ose Fast Flow resin, after which Trx-His6 was the only protein un-bound to the resin (see Table 1).

ELISA

PED-PEA15 was biotinylated using a Biotin-e-aminocaproic acidN-hydroxysuccinimide derivative (biotin-NHS), following the pro-ducer’s protocol; biotinylation of the protein was checked by LC–MS. Microtiter plate wells were coated with 0.5 lM solutions ofD4M and Trx-His6 for 16 h at 4 �C, then unbound sites wereblocked with a 0.1% w/v solution of BSA in PBS (137 mM NaCl,2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH 7.4). After wash-ing with PBS–Tween buffer (137 mM NaCl, 2.7 mM KCl, 10 mMNa2HPO4, 2 mM KH2PO4, 0.004% Tween 20, pH 7.4), a 1.5 lM solu-tion of biotinylated PED-PEA15 in PBS was added to coated wellsand allowed to associate for 1 h at 37 �C. After washing again withPBS–Tween, a 1 lg/mL solution of STRV-HRP was added, and theplate was left 1 h at 37 �C in the dark. Sigma-fast O-phenylenedi-amine dihydrochloride tablet set was used to reveal a potentialbinding by detecting absorbance at 490 nm.

PED-PEA15 coating on CM5 sensor surface

A CM5 sensor chip was used to bind PED-PEA15, using a stan-dard amine coupling procedure [20], using HBS-EP buffer (10 mMHEPES, 150 mM NaCl, 3 mM EDTA, 0.005% Surfactant P20, pH 7.4)as running buffer. 60 lL of a N-hydroxy-succinimide and N-ethyl-N0-(dimethyl-aminopropyl)-carbodiimide mixture were injectedat a flow rate of 5 ll/min to activate the sensor surface, followedby 30 and 25 lL injections of PED-PEA15 diluted to 0.1 mg/mL in10 mM sodium acetate pH 4.0 buffer. Unreacted activated groupswere blocked by a 15 lL injection of 1 M ethanolamine. Binding as-says were carried out at a flow rate of 30 ll/min in 50 mM sodiumphosphate, 150 mM NaCl, 1 mM TCEP, pH 7.4 buffer.

Results

D4 cloning, expression and purification

The cDNA encoding the D4 domain of PLD1 was amplified byPCR using two primers allowing its simultaneous cloning in fivedifferent E. coli pETM expression vectors, namely pETM-11,pETM-20, pETM-30, pETM-52 and pETM-60. This allowed theexpression of D4 fused respectively to a six-histidine tag (namedHis6), to Thioredoxin A-His6, to His6-Glutathione-S-transferase, toDisulfide Oxidoreductase A-His6 and Nus A-His6 at its N-terminus.An initial screening was performed by chemically transformingeach pETM expression vector into E. coli BL21(DE3) cells andexpression was then induced with 1 mM IPTG for 16 h at 22 �C in1.5 mL of LB. Bacterial cultures were harvested, lysed and purifiedon QIAgen Ni–NTA magnetic agarose beads, and buffer-solubilizedsamples were analyzed on SDS–PAGE together with 10 lL of totaland soluble fractions of each lysate (Fig. 1). Although yield wasnot very high with all five vectors, appreciable results were ob-tained when D4 was expressed fused to Thioredoxin A and His6

(the corresponding fusion protein was named Trx-His6-D4). A fur-ther screening transforming D4-cloned pETM-20 vector into sev-eral E. coli strains identified the BL21(DE3)pLysS strain as thebest expression host. Different IPTG concentrations were eventu-ally tested and induction with 0.1 mM IPTG produced the most sol-uble recombinant protein.

Expression of Trx-His6-D4 in 200 mL of bacterial cell culturewas induced under optimized conditions, yielding about 1.5 mgof protein per 1.1 g of wet weight cells. An initial recovery stepof Trx-His6-D4 from cell lysate on Ni-charged resin was followedby a polishing step on MonoQ anion-exchange column. Preliminaryattempts at purification showed that all purifications had to beperformed in the presence of TCEP (1 mM)-containing buffers, to

U T S P U T S P U T S P U T S P U T S P

100 kDa75 kDa

50 kDa

35 kDa

pETM11+D4 pETM20+D4 pETM30+D4 pETM52+D4 pETM60+D4

Fig. 1. Evaluation of the bacterial expression of the D4 domain of PLD1 fused to five different tags. The D4 cDNA was cloned into pETM-11, pETM-20, pETM-30, pETM-52 andpETM-60 vectors and its expression was induced in BL21(DE3) cells with 1 mM IPTG as described in the Materials and methods section. Ten microliters of soluble fractionfrom non-induced cell culture (U), 10 ll of total (T) and soluble (S) fractions from induced cell culture and micro-purified fraction (P, see Materials and methods) were loadedfor each vector; expected size for the fusion proteins is 43, 56, 68, 64 and 97 kDa, respectively.

F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308 305

limit covalent aggregation phenomena due to the 5 cysteines pres-ent in D4. Surprisingly, a protein of the size expected for Trx-His6-D4 (about 56 kDa) was eluted in two well separated fractions: oneeluted at 220 mM NaCl (peak 1) and the other at 280 mM NaCl(peak 2) (Fig. 2A and B). To check the identity of proteins elutedfrom the MonoQ column, the material under peak 2 was furtherseparated by size-exclusion chromatography. Three different peakswere individually separated and a protein of about 56 kDa wasrecovered in peak B (Fig. 2 C and D). 500 ng of each protein weresubsequently analyzed by LC–MS and their average masses calcu-lated. Results showed that proteins under both peak 1 from MonoQ

300

mA

U

250

200

150

100

50

0

100

80

60

40

20

0

mA

U

50 10 15 20

20 6 128 10

mL

mL4

Fig. 2. Purification of Trx-His6-D4. After an initial affinity purification on Ni-resin, Trx-Hirecovered (band at 56 kDa) under both the peaks 1 and 2, as evidenced by SDS–PAGE anal10 ll) were analyzed. The material corresponding to peak 2 was further fractionated by Sizfractions corresponding to peak A were loaded; in lanes 3 and 4, fractions corresponding to

(named D4M) and peak B from gel filtration (named D4D) had amass corresponding to the theoretical mass of 55525.8 amu calcu-lated for Trx-His6-D4. Peptide mass fingerprint analyses of bothproteins indicated the presence of both human PLD1 and TrxA,thus suggesting that they were Trx-His6-D4. Proteins in peaks Aand C, similarly analyzed, were bacterial proteins unrelated toTrx-His6-D4.

Following a careful calibration of a Superdex-200 HR 10/30 col-umn with 8 molecular weight standards in 20 mM sodium phos-phate, 1 mM TCEP, 150 mM NaCl, pH 7.5 buffer, we assessed thatD4M and D4D actually had apparent MW of about 54 kDa and

1 2 3 4 5 6

75 kDa50 kDa

35 kDa

14

100 kDa

75 kDa

50 kDa

35 kDa

1 2

s6-D4 was further purified by anion-exchange chromatography. (A) The protein wasysis (B), where fractions corresponding to peak 1 (lane 1, 10 ll) and to peak 2 (lane 2,e Exclusion Chromatography (C) and analyzed on SDS–PAGE gel. (D) In lanes 1 and 2,peak B were loaded; in lanes 5 and 6, fractions corresponding to peak C were loaded.

Ferritin, 460 kDaCatalase, 206 kDa

Aldolase, 170 kDa Transferrin, 81 kDa

Bovine serum albumin, 66 kDa

Ovalbumin, 43 kDa

Carbonic anhydrase, 29 kDa

CytochromeC, 12.4 kDa

D4D

0

0.5

1.0

1.5

2.0

2.5

3.0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7Kav

logM

W D4M

Fig. 3. Determination of D4M and D4D molecular weight by gel filtration. About 100 lg of eight molecular weight standards were analyzed on a Superdex-200 HR 10/30column, in 20 mM sodium phosphate, 1 mM TCEP, 150 mM NaCl, pH 7.5 buffer. Results were reported as log(MW) versus Kav (=(elution volume � void volume)/(totalvolume � void volume)); 100 lg of D4M and D4D were then analyzed and their apparent masses were extrapolated from the calibration curve (y = �3.76x + 3.24, R2 = 0.98).

Elution volume ml

0 2 4 6 8 10 12 140

2.0

4.0

6.0

8.0

0

2.0

4.0

6.0

8.0

0

2.0

4.0

6.0

8.010

A 490

, mA

UA 4

90, m

AU

A49

0, m

AU

0 2 4 6 8 10 12 14

0 2 4 6 8 10 12 14

Fig. 4. Effect of the ionic strength on the oligomerization equilibrium. About 150 lgof D4M were analyzed on a Superdex-200 HR 10/30 column under the followingconditions: (A), 50 mM sodium phosphate, 50 mM NaCl, pH 7.5; (B), 50 mM sodiumphosphate, 250 mM NaCl, pH 7.5; (C), 50 mM sodium phosphate, 1 M NaCl, pH 7.5.By increasing the buffer ionic strength, the peak relative to the dimeric D4 disap-pears, indicating that dimerization is mediated by ionic interactions.

306 F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308

112 kDa, respectively (Fig. 3). These results strongly suggested thatTrx-His6-D4 may exist in solution as two different oligomeric vari-ants (D4M, monomeric and D4D, dimeric).

Since monomer and dimer were eluted at two distinct salt con-centrations, we investigated the occurrence of a monomer–dimerequilibrium under different condition of ionic strength. For thispurpose, 150 lg samples of D4M, equilibrated in three buffers withdifferent salt concentrations (50 mM, 250 mM and 1 M NaCl), wererepetitively injected on the gel filtration column previously equil-ibrated with buffer at the same ionic strength as the sample. Asshown in Fig. 4, at high salt concentration the oligomerizationequilibrium was completely shifted to the monomeric form, indi-cating that the inter-molecular forces at the subunits contact inter-face were largely electrostatic.

PED-PEA15 expression and purification

PED-PEA15 was cloned, expressed and purified according to theprocedure described in the Method section. The recombinant pro-tein was characterized by SDS–PAGE and LC–MS to assess purityand MW. The MW was in good agreement with that calculated(MWexp/MWtheor.: 15168.8 ± 1.3/15168.5 amu), considering theadditional 128 amu derived from the residual dipeptide glycine–alanine sequence introduced at the N-terminus during the cloning.A CD analysis of PED-PEA15 was performed and, as expected, aspectrum typical of a-helical structures was recorded, with a firstminimum at 208 nm and a second one at 221 nm. Indeed, previousNMR experiment had shown that PED-PEA15 has a death-effectordomain at its N-terminus, characterized by a typical structure of6 a-helices [21].

D4/PED-PEA15 ELISA

An ELISA was set up to confirm in vitro the ability of D4 to bindPED-PEA15. Since any attempt to cleave D4 from its fusion partnersfailed, the whole recombinant protein Trx-His6-D4 was used,whilst Trx-His6 alone, expressed and purified independently, wasused as a control in all experiments. For the assay, 0.5 lM biotinyl-ated PED-PEA15 was added over D4M-coated wells, and Trx-His6-coated wells were used as blank. Unfortunately, in this experiment,a non-specific interaction between biotinylated PED-PEA15 andTrx-His6 comparable to that observed for PED-PEA15/D4M was de-tected. Thus, to minimize this artifactual association, different buf-fers were tested during the association and washing steps.

In this set of experiments, biotinylated PED-PEA15 was dilutedin eight different buffers to test the effect of different concentra-

tions of both NaCl (150 and 300 mM) and Tween 20 (0, 0.004,0.02, 0.1% v/v) during the association phase, while a washing bufferwith higher salt concentration (PBS–NaCl buffer: 300 mM NaCl,2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, 0.004% Tween 20,pH 7.4) was used to rinse wells. As shown in Fig. 5A, in almostall cases, an enhancement of the D4M/PED-PEA15 binding signal

0%

20%

40%

60%

80%

100%

120%

-6.4 -6.3 -6.2 -6.1 -6.0 -5.9 -5.8 -5.7 -5.6log M [PED-PEA15]

(A/A

0)*10

0

- - - - - - - --

A 490

nm

0

0.2

0.4

0.6

0.8

1.0

1.2

1 2 3 4 5 6 7 8

D4MTrx+His6

Buffer

D4MTrx+His6

D4MTrx+His6

Fig. 5. ELISA-like assays were set up to determine the optimal conditions for binding of D4 to PED-PEA15. In (A) biotinylated PED-PEA15 was diluted up to 1 lM in buffers-containing different salt concentrations. The common components were 50 mM sodium phosphate, 1 mM TCEP, pH 7.5. Buffer 1: 150 mM NaCl; Buffer 2: 150 mM NaCl,0.004% Tween 20; Buffer 3: 150 mM NaCl, 0.02% Tween 20; Buffer 4: 150 mM NaCl, 0.1% Tween 20; Buffer 5: 300 mM NaCl; Buffer 6: 300 mM NaCl, 0.004% Tween 20; Buffer7: NaCl 300 mM, 0.02% Tween 20; Buffer 8: NaCl 300 mM, 0.1% Tween 20. The biotinylated protein diluted in the different buffers was added over wells coated with 0.5 lMsolutions of Trx-His6 and D4M. After addition of the streptavidin-derivatized secondary antibody, the absorbance at 490 nm was determined. Each data point was at least intriplicate. (B) Dose-dependent binding of PED-PEA15 to D4M. Results are reported as (A/A0) � 100 versus logarithm of molar concentration of the added biotinylated PED-PEA15. A is the average OD from the triplicate data at a given concentration and A0 is the maximum average OD. The binding assay was performed twice and the resultsaveraged.

F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308 307

was observed. In particular, adding 0.004% Tween 20 and 300 mMNaCl in the association buffer and rinsing the wells with the PBS–NaCl buffer, partially suppressed the association between Trx-His6

and biotinylated PED-PEA15 and also provided stronger signals(see Fig. 5A, lane 6).

The optimized ELISA protocol was used to perform a binding as-say between D4M and PED-PEA15. For this purpose, 100 lL of a0.25 lM solutions of D4M or Trx-His6 (blank control) were coatedon a 96-well plate, then 100 lL of increasing amounts of biotinyl-ated-PED-PEA15 were added. Triplicate data points were averagedand signals from Trx-His6-coated wells were subtracted from thecorresponding D4M-coated wells. KD extrapolated from the ELISAdata was �7 � 10�7 M (Fig. 5B).

D4/PED-PEA15 surface plasmon resonance

To confirm the results obtained with the ELISA, the interactionbetween PED-PEA15 and Trx-His6-D4 was further assessed usinga BIAcore 3000 system. PED-PEA15 was efficiently bound to the

0

100

200

300

400

500

600

700

-10 10 30 50 70 90

Res

pons

e, R

U

Tim

Fig. 6. SPR analysis of the PED/PEA-15 and Trx-His6-D4 interaction. Solutions of both D4in 50 mM sodium phosphate, 150 mM NaCl, 1 mM TCEP, pH 7.4 buffer at 25 �C, at a flosensorgrams refer to the highest concentration used. By injecting increasing amounts o0.4 � 10�3 M�1 s�1) were extrapolated using BIAevaluation software, version 4.1 (BI10�7 M) for the PED-PEA15/D4M interaction was then calculated as the ratio between r

sensor surface of a CM5 chip, achieving an overall 11,000 RUimmobilization level. For this experiment, interference due to thepresence of the fusion partner were evaluated using a solution ofTrx-His6 in 50 mM sodium phosphate, 150 mM NaCl, 1 mM TCEP,pH 7.4 buffer as control. Importantly, in this case, no binding toimmobilized PED-PEA15 was detected. D4M and D4D were thenanalyzed by SPR, and only the monomeric variant was observedto bind to immobilized PED-PEA15 (Fig. 6). The KD extrapolatedby these experiments (data not shown) was 4 ± 1 � 10�7 M, a valuevery close to that derived from the ELISA.

Discussion

Several lines of evidence indicate that the PLD1/PED-PEA15interaction impairs the insulin-stimulated glucose transport intype II diabetes, validating this protein–protein interaction as animportant target. In order to investigate the in vitro binding be-tween the two proteins, we prepared recombinant PED-PEA15and D4 in bacteria: whilst PED-PEA15 was easily expressed and

110 130 150 170 190 210

D4M, 2 M

D4D, 2 M

D4M, 2 M

D4D, 2 M

e, s

M and D4D up to 2 lM were injected over the PED/PEA-15-immobilized sensor chipw rate of 30 lL/min. No binding was detected with the dimeric D4D. The reportedf D4M (data not shown), the rate constants kd (6.5 ± 0.8 � 10�3 s�1) and ka (1.8 ±

Acore Technologies, Inc.), assuming a 1:1 (Langmuir) binding. The KD (4 ± 1 �ate constants as follows: KD = kd/ka.

308 F. Viparelli et al. / Protein Expression and Purification 59 (2008) 302–308

purified, the preparation of D4 required several optimization steps.For this purpose, the D4 cDNA was cloned in a panel of vectors andthe expression attempted in different bacterial strains. Expressionlevels were quite low under all explored conditions, but, when ex-pressed fused to the TrxA-His6 tag, amounts of protein sufficientfor further investigation were recovered. The procedure for itspurification was also rather laborious. In fact, after an initial affin-ity step on a Nickel charged column, a second purification step byionic exchange chromatography was required to reach a sufficientlevel of purity. The final yield was also compromised by the pres-ence of two different oligomeric variants. Dimerization was notdriven by the formation of disulfide bonds between subunits, sincedimers also persisted under reducing conditions; moreover, dimer-ization was also seemingly specific for the fusion protein and pro-moted by the D4 domain alone, in fact the TrxA-His6 tag alone wasinvariably monomeric under all conditions studied. We found thatthe fusion protein Trx-His6-D4 was present only in the monomericform at salt concentrations higher than about 0.5 M, suggesting theoccurrence of electrostatic and polar interactions at the subunitsinterface. It must also be underlined that the equilibrium couldnot be reverted from monomer to dimer by lowering the salt con-centration. D4 is the C-terminal region of PLD1 that contains one ofthe two PD responsible for intramolecular association of PLD1extremities which is essential for its enzymatic activity. Here, weshow that D4 exists both as a monomer and a dimer, suggestingthat the C-terminal region of PLD1 might associate not only withthe corresponding N-terminus end of the same molecule, but alsowith a C-terminal end of a different PLD1. Thus, two different PLD1might dimerize through an intermolecular interaction of PD,resulting in inactive dimers, where N- and C-halves of each PLD1can no longer interact to constitute the intramolecular catalyticsite.

BIAcore analysis also showed that PED-PEA15 only binds themonomeric form of D4. We can exclude that PED-PEA15 bindsPLD1 through the HKD region, as this would impair PLD1 activity.Instead, PED-PEA15 binding reportedly induces an overallenhancement of the PLD1’s enzymatic activity which can only beachieved by the correct juxtaposition of the two PLD1 HKD motifs.Therefore, it can be deduced that the PED-PEA15 binding site onPLD1 is outside the catalytic interface and overlaps, at least par-tially, with the D4 dimerization site. We thus infer that dimeriza-tion could be a way of regulating PLD activity by forminginactive dimers or by altering their sub-cellular localization [12].On the basis of our data, we suggest that association ofPED-PEA15 could stabilize the active monomers, preventing theformation of inactive dimers or the translocation to cellular com-partments where the activity is reduced or suppressed. Therefore,one might hypothesize that in vivo PED-PEA15 binds the C-termi-nus of PLD1 forming long lasting complexes, thereby stabilizingthe active monomer and lengthening its persistence in the cell.

Acknowledgments

This work was supported by funds from the Ministero dell’Uni-versità e Ricerca (MUR) and partially by FIRB project number

RBNE03PX83_005 to M.R. We are also grateful to Prof. PietroFormisano for helpful discussion and to Mariano Amoroso for histechnical help.

References

[1] M. McDermott, M.J.O. Wakelam, A.J. Morris, Phospholipase D 1,2, Biochem, CellBiol. 82 (2004) 225–253.

[2] J.M. Besterman, V. Duronio, P. Cuatrecasa, Rapid formation of diacylglycerolfrom phosphatidylcholine: a pathway for generation of a second messenger,Proc. Natl. Acad. Sci. USA 83 (1986) 6785–6789.

[3] T.C. Sung, Y. Zhang, A.J. Morris, M.A. Frohman, Structural analysis of humanphospholipase D1, J. Biol. Chem. 274 (1999) 3659–3666.

[4] T.C. Sung, Y.M. Altshuller, A.J. Morris, M.A. Frohman, Molecular analysis ofmammalian phospholipase D2, J. Biol. Chem. 274 (1999) 494–502.

[5] T.C. Sung, R. Roper, Y. Zhang, S.A. Rudge, R. Temel, S.M. Hammond, A.J. Morris,B. Moss, J. Engebrecht, M.A. Frohman, Mutagenesis of Phospholipase D definesa superfamily including a trans-golgi viral protein required for poxviruspathogenicity, EMBO J. 16 (1997) 4519–4539.

[6] J.A. Stuckey, J.E. Dixon, Crystal structure of a phospholipase D member, Nat.Struct. Biol. 6 (1999) 278–284.

[7] D.R. Davies, H. Interthal, J.J. Champoux, W.G. Hol, The crystal structure ofhuman tyrosyl-DNA phosphodiesterase, Tdp1, Structure 10 (2002) 237–248.

[8] I. Leiros, F. Secundo, C. Zambonelli, S. Servi, E. Hough, The first crystal structureof a phospholipase D, Structure 8 (2000) 655–667.

[9] C.P. Ponting, I.D. Kerr, A novel family of phospholipase D homologues thatincludes phospholipid synthases and putative endonucleases: identification ofduplicated repeats and potential active site residues, Protein Sci. 5 (1996) 914–922.

[10] Z. Xie, W.T. Ho, J. Exton, Association of the N- and C-terminal domains ofPhospholipase D, J. Biol. Chem. 275 (2000) 24962–24969.

[11] Z. Xie, W.T. Ho, J. Exton, Association of the N- and C-terminal domains ofPhospholipase D is required for catalytic activity, J. Biol. Chem. 273 (1998)34679–34682.

[12] Y. Kam, J.H. Exton, Dimerization of phospholipase D isozymes, Biochem.Biophys. Res. Commun. 290 (2002) 375–380.

[13] J.H. Exton, Phospholipase D-structure, regulation and function, Rev. Physiol.Biochem. Pharmacol. 144 (2002) 1–94.

[14] Y. Zhang, O. Redina, Y.M. Altshuller, M. Yamazaki, J. Ramos, H. Chneiweiss, Y.Kanaho, M.A. Frohman, Regulation of expression of Phospholipase D1 and D2by PEA-15, a novel protein that interacts with them, J. Biol. Chem. 275 (2000)35224–35232.

[15] G. Vigliotta, C. Miele, S. Santopietro, G. Portella, A. Perfetti, M.A. Maitan, A.Cassese, F. Oriente, A. Trencia, F. Fiory, C. Romano, C. Tiveron, L. Tatangelo, P.Formisano, F. Beguinot, Overexpression of the PED/PEA15 gene causes diabetesby impairing glucose-stimulated insulin secretion in addition to insulin action,Mol. Cell. Biol. 24 (2004) 5005–5015.

[16] G. Condorelli, G. Vigliotta, C. Iavarone, M. Caruso, C.G. Tocchetti, F. Andreozzi,A. Cafieri, M.F. Tecce, P. Formisano, L. Beguinot, F. Beguinot, PED/PEA-15 genecontrols glucose transport and is overexpressed in type 2 diabetes mellitus,EMBO J. 17 (1998) 3858–3866.

[17] G. Condorelli, G. Vigliotta, A. Trencia, M.A. Maitan, M. Caruso, C. Miele, F.Oriente, S. Santopietro, P. Formisano, F. Beguinot, Protein kinase C(PKC)-aactivation inhibits PKC-f and mediates the action of PED/PEA15 on glucosetransport in the L6 skeletal muscle cells, Diabetes 50 (2001) 1244–1252.

[18] C. Xiao, B.F. Yang, N. Asadi, F. Beguinot, C. Hao, Tumor necrosis factor-relatedapoptosis-inducing signaling complex and its modulation by c-Flip and PED/PEA-15 in glioma cells, J. Biol. Chem. 277 (28) (2002) 25020–25025.

[19] G. Condorelli, A. Trencia, G. Vigliotta, A. Perfetti, U. Goglia, A. Cassese, A.M.Musti, C. Miele, S. Santopietro, P. Formisano, F. Beguinot, Multiple members ofthe mitogen-activated protein kinase family are necessary for PED/PEA15 anti-apoptotic function, J. Biol. Chem. 277 (2002) 11013–11018.

[20] B. Johnsson, S. Lofas, G. Lindquist, Immobilization of proteins to acarboxylmethyl dextran modified gold surface for biospecific interactionanalysis in SPR sensors, Anal. Biochem. 198 (1991) 268–277.

[21] J.M. Hill, H. Vaidyanathan, J.W. Ramos, M.H. Ginsberg, M.H. Werner,Recognition of ERK MAP kinase by PEA-15 reveals a common docking sitewithin the death domain and death effector domain, EMBO J. 21 (2002) 6494–6504.