Study of protein targets for covalent modification by the antitumoral and anti-inflammatory prostaglandin PGA1: focus on vimentin

JOURNAL OF MASS SPECTROMETRYJ. Mass Spectrom. 2007; 42: 1474–1484Published online in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/jms.1291

Study of protein targets for covalent modification bythe antitumoral and anti-inflammatory prostaglandinPGA1: focus on vimentin

Severine Gharbi,1† Beatriz Garzon,2† Javier Gayarre,2 John Timms1 andDolores Perez-Sala2∗

1 Cancer Proteomics Group, Ludwig Institute for Cancer Research, London, and Department of Gynaecological Oncology, University College London,UK2 Departamento de Ciencia de Proteınas, Centro de Investigaciones Biologicas, CSIC, Madrid, Spain

Received 12 April 2007; Accepted 30 July 2007

Prostaglandins with cyclopentenone structure (cyPG) display potent antiproliferative actions that haveelicited their study as potential anticancer agents. Several natural and synthetic analogs of the cyPGprostaglandin A1 (PGA1) have proven antitumoral efficacy in cancer cell lines and animal models. Inaddition, PGA1 has been used as an inhibitor of transcription factor NF-kB-mediated processes, includinginflammatory gene expression and viral replication. An important determinant for these effects is theability of cyPG to form Michael adducts with free thiol groups. The chemical nature of this interactionimplies that PGA1 could covalently modify cysteine residues in a large number of cellular proteinspotentially involved in its beneficial effects. However, only a few targets of PGA1 have been identified.In previous work, we have observed that a biotinylated analog of PGA1 that retains the cyclopentenonemoiety (PGA1-B) binds to multiple targets in fibroblasts. Here, we have addressed the identification ofthese targets through a proteomic approach. Cell fractionation followed by avidin affinity chromatographyyielded a fraction enriched in proteins modified by PGA1-B. Analysis of this fraction by SDS-PAGEand LC-MS/MS allowed the identification of the chaperone Hsp90, elongation and initiation factors forprotein synthesis and cytoskeletal proteins including actin, tubulin and vimentin. Furthermore, we havecharacterized the modification of vimentin both in vitro and in intact cells. Our observations indicate thatcysteine 328 is the main site for PGA1 addition. These results may contribute to a better understandingof the mechanism of action of PGA1 and the potential of cyPG-based therapeutic strategies. Copyright 2007 John Wiley & Sons, Ltd.

Supplementary electronic material for this paper is available in Wiley InterScience at http://www.interscience.wiley.com/jpages/1076-5174/suppmat/

KEYWORDS: cyclopentenone prostaglandins; proteomic identification; Michael addition; vimentin; cytoskeleton; inhibitionof proliferation

INTRODUCTION

Cyclopentenone prostaglandins (cyPG) are bioactiveprostanoids that are generated by the nonenzymatic dehy-dration of their parent prostaglandins (PG). Organic syn-thesis of these compounds has allowed the study of theirbiological actions.1 The cyPG prostaglandin A1 (PGA1) hasbeen shown to display potent antiviral, antiproliferative andanti-inflammatory effects. cyPG of the A series were soonnoted to inhibit viral replication.2 Since then, many studieshave established that various cyPG inhibit the replicationof several viruses, including those of vesicular stomatitis,3

influenza A4 and HIV.5 Santoro and coworkers established

ŁCorrespondence to: Dolores Perez-Sala, Departamento de Cienciade Proteınas, Centro de Investigaciones Biologicas, C.S.I.C., Ramirode Maeztu, 9, 28040 Madrid, Spain. E-mail: [email protected]†These authors contributed equally to this work.

a link between the antiviral effects of PGA1 and othercyPG, and their ability to inhibit NF-�B-mediated genetranscription and their activity as inducers of a heat shockresponse in cells.5 In fact, many viral genes require NK-�Bactivity for transcription. cyPG induce the activation of theheat shock factor and the expression of molecular chaper-ones, such as Hsp70 and Hsp27,6 which play a protectiverole against cellular stress preventing misfolded or dam-aged proteins from aggregation. The inhibition of NF-�Bby cyPG can occur at multiple levels, including the induc-tion of the inhibitory subunit I�B˛, the inhibition of the I�Bkinase (IKK),7 or the direct modification of NF-�B proteins.8,9

For these reasons, PGA1 and related cyPG have been widelyused as NF-�B inhibitors in diverse experimental settings.10,11

The inhibition of NF-�B is also at the basis of the anti-inflammatory effects of PGA1.7,10 Both, the anti-inflammatory

Copyright 2007 John Wiley & Sons, Ltd.

Protein modification by PGA1 1475

actions and the induction of heat shock proteins may con-tribute to the protective effects of PGA1 evidenced in variousanimal models of tissue injury, such as cerebral ischemia.12

An important property of PGA1 is its antiproliferativepotential. The antitumoral activity of the A series of cyPGhas been known for over 20 years. cyPG such as 7-PGA1 were found to increase the life span of Ehrlichascites tumor-bearing mice13 and to inhibit the growth ofseveral transformed cell lines.14 These and other findingselicited the use of PGA1 derivatives in preclinical studiesand the search for more stable analogs and suitable formsfor in vivo delivery.15,16 Therefore, the activity of severalPGA1 analogs such as 7-PGA1 methyl ester in a lipidmicrosphere formulation has been studied against humantumor xenographs in nude mice. This and related compoundshave shown cell-type dependent antitumoral activity.16

Structure–activity studies have established that the pres-ence of an ˛,ˇ-unsaturated carbonyl group in the struc-ture of cyPG is the key for their antiproliferative andanti-inflammatory activities. cyPG are therefore potent elec-trophiles that can form Michael adducts with nucleophilicresidues of proteins such as cysteine, histidine or lysine,although in the case of cyPG only cysteine adducts have beenreported. The stable binding of cyPG of the A series to cellu-lar proteins was early documented through the use of bothradioactive and biotinylated cyPG.17,18 Recent studies haveidentified several targets for covalent modification by A-typecyPG that may be involved in their cellular effects. The pro-teins identified include IKK,7 NF-�B subunits,9 and proteinsinvolved in cellular redox sensing and homeostasis suchas thioredoxin reductase19 and glutathione-S-transferase.20

In previous work we have observed that a biotinylatedanalog of PGA1 can selectively bind to several protein tar-gets in fibroblasts.21 However, to date, the identification ofPGA1 targets has not been addressed through proteomicapproaches. In this work we undertake the identification ofnovel targets for modification by PGA1 through a combi-nation of affinity chromatography and mass spectrometrytechniques. These results will broaden our knowledge aboutthe potential mechanisms involved in the antitumoral andanti-inflammatory effects of PGA1 and hopefully contributeto the identification of targets of therapeutic interest.

EXPERIMENTAL

MaterialsRecombinant hamster vimentin, accession number AH00-1833, was from Cytoskeleton Inc. Biotinylated PGA1

(PGA1-B, N-9-oxo-15S-hydroxy-prosta-10,13E-1-dien-1-oyl-N-biotinoyl-1,5-diaminohexane) was from Cayman Chem-ical. Horseradish peroxidase-conjugated streptavidin (HRP-streptavidin) and the enhanced chemiluminescence (ECL)detection system were from GE Healthcare. Antivimentinand antiactin antibodies were from Sigma. Anti-elongationfactor (EF)-1˛ was from Upstate. Antitubulin monoclonalantibody was the generous gift of Dr I. Barasoaın (C.I.B.,C.S.I.C., Madrid).

Cell treatments and fractionationNIH-3T3 fibroblasts and COS-7 cells were cultured inDMEM (Invitrogen) containing 10% fetal bovine serum, 2 mM

glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin.COS-7 cells were transfected with plasmids coding forwild-type human vimentin or the C328S mutant in fusionwith GFP, using Lipofectamine (Invitrogen), as previouslydescribed.22 For treatments, cells were incubated with PGA1-B for 2 h in serum-free medium. Total cell lysates wereobtained by disrupting cells in 50 mM Tris, pH 7.5, 0.1 mM

EDTA, 0.1 mM EGTA, 0.1 mM ˇ-mercaptoethanol, 0.2 mM

sodium vanadate, 50 mM NaF containing 2 µg/ml of each ofthe protease inhibitors: leupeptin, pepstatin A and aprotininand 1.3 mM Pefablock (Boehringer), by forced passages

through a 2612 G needle. Lysates were centrifuged at 1000g

for 5 min at 4 °C. Postnuclear supernatants were furthersubjected to ultracentrifugation at 100 000g for 30 min at4 °C to obtain soluble and membrane cellular fractions.Supernatants (S100) were collected and pellets (P100) wereresuspended in lysis buffer containing 1% NP-40 and 0.1%SDS. Protein concentration was determined by the BCAprotein assay from Pierce. Fifteen micrograms of proteinfrom each experimental condition was electrophoresed on12.5% polyacrylamide gels and transferred to Immobilon-Pmembranes (Millipore). Incorporation of PGA1-B into theproteins was assessed by Western blot and detected withHRP-streptavidin and ECL, as described.23

Enrichment of PGA1-B-modified proteins bychromatography on avidin beadsAliquots of P100 fractions from vehicle or PGA1-B-treatedcells containing 100 µg of protein were incubated batchwisewith 50 µl of a 1 : 1 slurry of Neutravidin beads (Pierce)equilibrated in lysis buffer containing 1% NP-40 and 0.1%SDS for 1 h at 4 °C. After incubation, the beads wereexhaustively washed with the same buffer and the proteinsretained were eluted by boiling in Laemmli buffer andseparated by SDS-PAGE on 8–12% polyacrylamide gradientslab gels (Invitrogen).

Protein identificationGel slices from the lanes containing the proteins eluted fromNeutravidin beads were washed twice in 50% acetonitrileand dried in a SpeedVac. Samples were reduced in 10 mM

dithiothreitol, 25 mM ammonium bicarbonate (AmBic) for45 min at 50 °C and then alkylated in 50 mM iodoacetamide,25 mM AmBic for 1 h at room temperature in the dark. Gelpieces were washed twice in 50% acetonitrile, and vacuum-dried. Proteins were proteolyzed with 50 ng of modifiedtrypsin (Promega, Southampton, United Kingdom) in 25 mM

AmBic overnight at 37 °C. Supernatant was collected,and peptides were further extracted in 5% trifluoroaceticacid, 50% acetonitrile. Peptide extracts were pooled andvacuum-dried and resuspended in 3 µl of 0.1% formicacid. Nano-HPLC-electrospray ionization-collision-induceddissociation MS/MS was performed on an Ultimate HPLCwith a PepMap C18 75-µm inner diameter column (bothLC Packings) at a flow rate of 300 nl/min, coupled to aQ-Tof1 (WatersMicromass) mass spectrometer. Spectra were

Copyright 2007 John Wiley & Sons, Ltd. J. Mass Spectrom. 2007; 42: 1474–1484DOI: 10.1002/jms

1476 S. Gharbi et al.

processed using MassLynx software version 2.4 (Waters) andmasses were searched against the IPI human database usingthe Mascot database search engine (Matrix Science). Positiveidentifications were made when MOWSE scores were abovethe significance threshold value (P D 0.05).

In vitro modification of vimentinPurified vimentin was incubated at room temperature for2 h in 5 mM Pipes, pH 7.0, 0.1 mM DTT, in the presenceof vehicle (DMSO) or the various Michael receptors, at theconcentrations indicated in the legends of the correspondingfigures. For analysis of the incorporation of PGA1-B tovimentin, aliquots from the incubation were subjected toSDS-PAGE and transferred to Immobilon-P membranes. Theincorporation of biotin was checked by incubation withHRP-conjugated streptavidin and ECL, and estimated bycomparison with a biotinylated BSA standard essentially asdescribed previously for actin.23

For mass spectrometry analysis, vimentin was digestedwith trypsin (Boehringer) at a vimentin to trypsin ratioof 80 : 1 for 20 h at 37 °C. The resulting peptide digestswere purified on C18 ZipTips (Millipore) following themanufacturer instructions. For MS, samples were mixedwith ˛-cyano-4-hydroxycinnamic acid (Bruker-Daltonics) in50% aqueous acetonitrile and 0.25% trifluoroacetic acid(99.5% purity; Sigma Chemical). This mixture was depositedonto a 600 µm AnchorChip prestructured MALDI probe(Bruker-Daltonics). MALDI-MS(/MS) spectra were recordedin an Ultraflex time-of-flight mass spectrometer (Bruker-Daltonics) equipped with a LIFT-MS/MS device.24 Spectrawere acquired in the positive-ion mode at 50 Hz laserfrequency, and 300 individual spectra were averaged.For fragment ion analysis in the tandem time-of-flight(TOF/TOF) mode, precursors were accelerated to 8 kV andselected in a timed ion gate. Fragment ions generated bylaser-induced decomposition of the precursor were furtheraccelerated by 19 kV in the LIFT cell and their masses wereanalyzed after passing the ion reflector. The analyses of massdata were performed using the FlexAnalysis and BioToolssoftware (Bruker-Daltonics).

Biotinylated peptides from PGA1-B-modified vimentinwere enriched by adsorption onto SoftLink avidin (Promega).After extensive washing, bound peptides were eluted with10% acetic acid25 and used directly for MALDI-TOF MSanalysis.

RESULTS

Binding of PGA1-B to NIH-3T3 proteinsCovalent modification of cellular proteins is an importantmechanism for the biological effects of PGA1. Consistentwith our previous observations, treatment of intact NIH-3T3fibroblasts with PGA1-B resulted in the stable incorporationof this prostanoid into an ample number of polypeptides,as detected by Western blot (Fig. 1). In order to reducesample complexity for analysis, we performed subcellularfractionation. We observed that most of the biotin-labelledpolypeptides were present in the P100 fraction (Fig. 1).Endogenous biotinylated proteins were also more abundant

Figure 1. Detection of PGA1-B-modified proteins in totallysates and cellular fractions from NIH-3T3 fibroblasts. Cellswere incubated with vehicle or 60 µM PGA1-B for 2 h. Total celllysates and subcellular fractions, obtained as detailed in theexperimental section were analyzed by SDS-PAGE.Biotinylated proteins were visualized by Western blotting anddetected with HRP-streptavidin and ECL.

in this fraction. Therefore, this fraction was used forsubsequent analysis.

Identification of PGA1-B-modified proteinsIn order to further enrich PGA1-B-modified proteins weemployed an affinity strategy using Neutravidin beads.Proteins retained on the beads were eluted and analyzedby SDS-PAGE and colloidal Coomassie staining (Fig. 2(A)).The profiles of the densitometric analysis of the lanescorresponding to the eluates are shown in Fig. 2(B). Thebands for which most prominent differences were observedwere sliced from the lane containing the eluates from PGA1-B-treated cells and subjected to tryptic digestion and analysisby LC-MS/MS. Table 1 depicts a list of the identified proteins,along with their identification parameters. The sequencesof the peptides identified (Table S1) and the MASCOTsearch data of representative peptides from each proteinare provided as Supplementary material. Proteins identifiedby this method belonged to several functional categories.Several of the polypeptides corresponded to Hsp90 ˛ andˇ subunits. Two proteins that constitute the intermediatefilament network were also identified, namely, vimentinand GFAP. The presence of GFAP in NIH-3T3 fibroblastsis somewhat unexpected, given the fact that this protein isexpressed mainly in glial cells. Nevertheless, several reportsindicate that GFAP can also be expressed in fibroblastsof diverse origins.26 Other cytoskeletal proteins identifiedincluded actin and tubulin. These abundant proteins areknown to be the target of several agents reacting with cysteineresidues. In addition, two proteins were identified that areinvolved in protein synthesis, namely eukaryotic initiationfactor 4A-I and EF-1-˛ 1. Several cytokeratins were identifiedin the analysis of band 5 (MASCOT search data provided asSupplementary material).

The selectivity of the affinity approach was assessed byanalysis of the avidin-bound fractions by Western blot. Asit can be observed in Fig. 2(C), biotinylated proteins were



Figure 2. Enrichment of PGA1-B modified proteins by adsorption on Neutravidin-agarose. NIH-3T3 fibroblasts were cultured in thepresence of vehicle or 60 µM PGA1-B. Biotinylated proteins present in P100 fractions from these cells were adsorbed on Neutravidinbeads as detailed in the Experimental Section. Total and Neutravidin-bound fractions were analyzed by SDS-PAGE and Coomassiestaining (A) and densitometric profiles of the lanes corresponding to Neutravidin-bound fractions are shown in (B). Arrowheads markthe position of protein bands sliced for later digestion and LC-MS/MS analysis. Arrows mark the position of avidin oligomers.(C) Neutravidin-bound fractions were subjected to Western blot and detection with HRP-streptavidin or with antibodies againstsome of the proteins identified.

Table 1. Proteins identified in the Neutravidin eluates of P100 fractions from PGA1-B-treated NIH-3T3 fibroblasts

Bandno.

Protein accessionnumber Protein description

Proteinscore

Proteinmass

No. ofpeptidematches

Proteincoverage

(%)

1 P60709 Actin 267 41 710 9 28.3Actin-related hypothetical protein LOC345651 134 41 976 4 13.6

2 P60842 Eukaryotic initiation factor 4A-I 33 46 125 1 2.53 P68104 Elongation factor 1 � ˛ 1 84 50 109 7 164 Q8N532 TUBA6 protein (tubulin alpha) 88 36 997 5 17.75 P14136-1 Isoform 1 of Glial fibrillary acidic protein 73 49 850 2 4.4

P08670 Vimentin 103 53 488 7 13.96 P08238 Heat shock protein HSP 90-ˇ 224 83 081 12 17.8

Heat shock protein HSP 90-˛ 2 159 98 052 8 10.3Q58FF6 Heat shock protein 90Bd 53 58 228 3 5.7



retained by this method (left panel). In addition, tubulin,vimentin, EF-1˛ and actin were selectively enriched in theavidin-bound fraction from PGA1-B-treated cells comparedto the equivalent fraction obtained from cells treated withvehicle (right panel). Therefore, these proteins are likelycandidates for modification by PGA1-B in intact cells.

Analysis of PGA1 binding to vimentinVimentin is a major cytoskeletal protein shown to be the tar-get for modification by various thiol-reactive compounds.22,27

However, to date, there is little structural or functionalinformation on the modification of vimentin. By using aWestern-blot-based assay we have observed that PGA1-Bcan bind to purified vimentin in vitro in a concentration-dependent fashion (Fig. 3(A)). In contrast to what we foundfor other isolated proteins,28 we observed that the bindingof PGA1-B to vimentin did not vary substantially with theconcentration of reducing agent present in the incubation.The extent of binding under our experimental conditionswas approximately 0.1 mol of PGA1-B per mol of vimentin(Fig. 3(B)). Vimentin contains a single cysteine residue, cys-teine 328, which is conserved in vertebrates and constitutes

a likely site for the attachment of PGA1-B. As observed inFig. 3(C), pre-incubation with iodoacetamide reduced thebinding of PGA1-B to vimentin, thus supporting the require-ment for free thiol groups for PGA1-B addition. To identifythe site for covalent addition of PGA1, vimentin was incu-bated with PGA1 or PGA1-B as outlined above, and furthersubjected to trypsin digestion and MS analysis by MALDI-TOF MS (Fig. 3(D)). Notably, the peptide fingerprint ofcontrol vimentin consistently showed a lower number of pep-tides of high m/z when compared to the patterns obtainedfor PGA1- or PGA1-B-modified vimentin, which may reflectthat the chemical modification affects enzymatic cleavage.The peptide fingerprint analysis of PGA1-treated vimentinshowed the presence of a peptide with m/z 2656.35, whichwas not present in control vimentin and that was consistentwith the addition of PGA1 (average mass 336.47) to the trypticpeptide QVQSLTCEVDALKGTNESLER (residues 321–341,theoretical m/z 2320.14), which contains cysteine 328. Inaddition, a peptide of m/z 2981.09 was selectively detected inPGA1-B-treated vimentin, which is consistent with the addi-tion of PGA1-B (average mass 660.95) to the 321–341 peptide.A list of the peptides identified in the tryptic digests under

Figure 3. Binding of PGA1-B to vimentin in vitro. (A) Purified vimentin (5 µM) was incubated with the indicated concentrations ofPGA1-B and DTT for 2 h at room temperature Incorporation of PGA1-B was assessed by Western blotting with detection usingHRP-streptavidin. (B) Vimentin was incubated with 50 µM PGA1-B as in (A). An aliquot of the incubation mixture containing 20 pmolof vimentin was analyzed in parallel with aliquots of biotinylated BSA containing 0.3–3.2 pmol of biotin, as indicated. Results arerepresentative of three assays. (C) Vimentin was incubated with PGA1-B, as above after pre-incubation with vehicle or with 2 mM

iodoacetamide for 1 h. (D) Vimentin was treated with vehicle, PGA1 or PGA1-B and subjected to digestion with trypsin. The trypticpeptides were analyzed by MALDI-TOF MS. Spectra presented are representative of two assays with identical results. The positionof some of the peptides identified is shown. Asterisks mark the position of peptides the m/z of which is compatible with cyPGaddition. The equivalent positions of these peptides in control vimentin are indicated by arrows.



Table 2. Experimental and theoretical m/z of tryptic peptides from control, PGA1- and PGA1-B-modified vimentin

Experimental m/z Theoretical m/z

ControlPGA1-

vimentinPGA1-

B-vimentinUnmodified

m/zModified

m/z Position Sequence

733.35 733.36 71–77 SSMPGVR872.44 872.41 5–12 SVSSSSYR906.5 906.46 906.46 122–128 FLEQQNK914.49 914.46 914.45 29–36 SYVTTSTR932.49 932.46 402–409 LLEGEESR1002.50 1002.55 69–77 LRSSMPGVR1023.5 1023.46 1023.45 1023.51 273–281 QQYESVAAK1039.50 1039.46 14–23 MFGGPGTSNR (MSO)

(MSO)1046.51 1046.52 1046.53 188–195 LQEEMLQR1060.62 1060.56 401–409 KLLEGEESR1081.54 1081.51 1081.51 1081.5 313–320 QESNEYRR1088.59 1088.53 207–216 QDVDNASLAR1093.56 1093.52 1093.52 294–303 FADLSEAANR

1121.57 1121.58 381–389 EYQDLLNVK1125.62 1125.59 1125.56 1125.6 113–121 FANYIDKVR1179.59 1179.56 13–23 RMFGGPGTSNR1195.58 1195.56 13–23 RMFGGPGTSNR (MSO)

(MSO)1216.68 1216.66 1216.62 158–167 RQVDQLTNDK1252.61 1252.59 310–319 QAKQESNEYR1254.59 1254.56 1254.56 1254.56 145–154 LGDLYEEEMR1270.65 1270.56 145–154 LGDLYEEEMR (MSO)

(MSO)1287.68 1287.66 1287.64 1287.66 159–169 QVDQLTNDKAR

1295.64 1295.63 1295.66 390–400 MALDIEIATYR1303.71 1303.67 186–195 EKLQEEMLQR1309.62 1309.6 1309.6 282–291 NLQEAEEWYK

1347.56 1347.66 1–12 MSTRSVSSSSYR1393.74 1393.7 270–281 DVRQQYESVAAK1408.71 1408.69 310–320 QAKQESNEYRR1443.78 1444.72 1443.76 158–169 RQVDQLTNDKAR

1497.69 1497.70 1497.7 143–154 SRLGDLYEEEMR1527.82 1527.82 1527.82 378–389 HLREYQDLLNVK1533.83 1533.84 1533.85 223–235 VESLQEEIAFLKK1539.89 1539.9 1539.91 129–142 ILLAELEQLKGQGK1570.89 1570.89 1570.89 410–423 ISLPLPNFSSLNLR

1587.82 1587.78 1587.79 100–112 TNEKVELQELNDR1652.79 1652.79 1652.79 145–157 LGDLYEEEMRELR

1688.81 1688.83 1688.83 1688.82 170–183 VEVERDNLAEDIMR1734.81 1734.81 1734.81 364–377 LQDEIQNMKEEMAR1776.86 1776.87 1776.87 1776.86 294–309 FADLSEAANRNNDALR

1808.9 1808.89 145–158 LGDLYEEEMRELRR1836.79 1836.80 450–465 DGQVINETSQHHDDLE

1838.97 1838.96 1838.96 424–439 ETNLESLPLVDTHSKR1895.93 1895.92 143–157 SRLGDLYEEEMRELR

1991.98 1991.98 292–309 SKFADLSEAANRNNDALR2126.05 2126.06 78–96 LLQDSVDFSLADAINTEFK

2200.95 2200.97 2200.97 2200.98 345–363 EMEENFALEAANYQDTIGR2222.14 2222.14 2222.15 104–121 VELQELNDRFANYIDKVR

2320.13 2320.14 321–341 QVQSLTCEVDALKGTNESLER2324.1 2324.1 2324.12 188–206 LQEEMLQREEAESTLQSFR

2423.05 2423.11 445–465 TVETRDGQVINETSQHHDDLE2439.19 2439.20 2439.21 100–119 TNEKVELQELNDRFANYIDK



Table 2. (Continued)

2497.18 2497.25 2497.25 2497.25 78–99 LLQDSVDFSLADAINTEFKN TR2581.23 2581.25 186–206 EKLQEEMLQREEAESTLQSF R2656.35 2320.14 321–341 QVQSLTCEVDALKGTNESLE R (CPGA1)

(C336.5) (CPGA1)2694.4 2694.41 2694.38 100–121 TNEKVELQELNDRFANYIDK VR

2980.55 2320.14 321–341 QVQSLTCEVDALKGTNESLE R (CPGA1-B)(C661.0) (CPGA1-B)

3393.62 3393.65 3393.53 188–216 LQEEMLQREEAESTLQSF RQDVDNASLAR3650.74 3650.78 3650.77 186–216 EKLQEEMLQREEAESTLQSF RQDVDNASLAR3909.00 3909.00 3908.89 236–269 LHDEEIQELQAQIQEQHVQI DVDVSKPDLTAALR4037.13 4037.12 4037.09 235–269 KLHDEEIQELQAQIQEQHVQ IDVDVSKPDLTAALR

(CPGA1 and CPGA1-B indicate m/z compatible with the addition of one molecule of the respective cyPG; MSO, m/z compatible withmethionine oxidation).

the various experimental conditions is provided in Table 2.In order to confirm the site of modification of vimentin byPGA1 or PGA1-B, the putatively modified peptides weresubjected to analysis by MALDI-TOF/TOF tandem MS. Inboth cases, the analysis of the precursor ions correspondingto the modified peptides yielded y- and b-ions compatiblewith the sequence of the 321–341 peptide, although a fullsequence could not be obtained (data not shown).

As a complementary approach, the tryptic digest fromPGA1-B-modified vimentin was subjected to purification onSoftLink avidin beads and elution with 10% acetic acid.MALDI-TOF analysis of the eluted peptides showed anoticeable increase of the relative intensity of the peak ofm/z 2981.09, thus indicating that this peak contains thebiotinylated PG (Fig. 4(A)). MS/MS analysis of this enrichedpeptide yielded the sequence expected for the 321–341tryptic peptide and showed an ion of m/z 2320.02, compatiblewith the retro-Michael fragmentation of the adduct (m/z2320.14). A mass increment compatible with the addition ofPGA1-B was observed on ions y21, y21-NH3 and y15 (markedwith asterisks in Fig. 4(B)). These results confirm the identityof the modified peptide and the presence of the modification.

Taken together, our observations indicate that PGA1

binds to cysteine 328 of vimentin. In order to assess theimportance of this residue for vimentin modification inintact cells COS-7 cells were transiently transfected withGFP-vimentin wild type and C328S constructs and culturedin the presence of PGA1-B. The results in Fig. 4(C) show thatsubstitution of cysteine 328 drastically reduces the retentionof GFP-vimentin on avidin beads. These results indicate thatcysteine 328 is important for PGA1-B binding to vimentin invivo.

DISCUSSION

CyPG exert potent biological actions that have raised inter-est in the use of these prostanoids or related compoundsas therapeutic agents. Addition to cysteine residues andthe subsequent alteration of protein function is an impor-tant mechanism in the anti-inflammatory and antiprolif-erative effects of these compounds. Interestingly, severalsynthetic compounds based on the structure of PGA1 orits derivatives have been studied as potential antitumoral

or neurotrophic agents.11,29,30 Moreover, several naturalcompounds with potent antitumoral or anti-inflammatoryactivity have been identified that possess a PGA1-likecyclopentenone moiety.31,32

We and others have shown that the binding of severalcyPG derivatives, including biotinylated 15d-PGJ2 (15d-PGJ2-B) and PGA1-B, to a number of cellular proteins isstable under denaturing conditions.9,19,21 This makes pro-teomic techniques a most advantageous approach for theidentification of cyPG targets. Using 2D-electrophoresis wehave observed that the subsets of cellular proteins mod-ified by cyPG with different structure do not completelyoverlap.21 This finding increases the interest of the use ofproteomic approaches in combination with drug design andpharmacological studies in order to increase the selectiv-ity and/or potency of the designed compounds. Proteomicstudies could also provide valuable information on themechanism of action and even on the potential adverseeffects of the compounds under evaluation. The results pre-sented herein identify several potential targets for PGA1-Baddition that were previously unknown. Full validation ofthese targets will require studies in intact cells involvingimmunoprecipitation and functional assays to assess theconsequences of the modification. Among the proteins iden-tified are components of the cellular stress defense system,like the cellular chaperone Hsp90 that regulates the activityof a wide number of proteins involved in apoptotic andsurvival pathways. Hsp90 and related heat shock proteinshave been identified as targets for modification by severalthiol-reactive agents, including nitric oxide33,34 and the cyPG15d-PGJ2.22 The polypeptide EF-1˛ (EF-1˛) is a highly abun-dant multifunctional protein. In its GTP-bound form EF-1˛catalyzes the binding of aminoacyl-tRNA to the A-site of theribosome. In addition, EF-1˛ has been reported to participatein several signal transduction cascades involved in stressresponses, apoptosis, cell transformation and cytoskeletalorganization (see Refs. 35,36 for review). The modificationof EF-1˛ cysteine residues by thiolation or nitrosylationhas been detected through proteomic approaches in sev-eral experimental systems.25,34,37 Nevertheless, the functionalconsequences of these modifications are not fully elucidated.The translation initiation factor eIF4A-I participates in therecruitment of the 50-unstranslated region of the mRNA to



Figure 4. Identification of the site for the addition of PGA1-B to vimentin. (A) MALDI-TOF MS analysis of tryptic digests fromPGA1-B-modified vimentin before and after avidin affinity purification. PGA1-B-modified vimentin was digested with trypsin asdetailed in the experimental section. An aliquot of the digestion (Input) was immediately used for purification on a C18 ZipTip andthen subjected to peptide mass fingerprint analysis. The rest was incubated batchwise with SoftLink avidin. After washing, retainedpeptides were eluted in a volume of 10% acetic acid equal to that of the starting aliquot and analyzed directly by MALDI-TOF MS(Eluate). The x-axis between m/z 2300 and 3200 is shown to illustrate the enrichment of the 2981.09 peptide (marked with anasterisk). (B) MALDI-TOF/TOF analysis of the putatively PGA1-B-modified peptide. The peptide with m/z 2981 from the avidinchromatography of the tryptic digest of PGA1-B-modified vimentin was analyzed by MS/MS. The m/z of the ions experimentallyfound is given. The peptide sequence obtained is shown. Asterisks denote the ions bearing the PGA1-B motif. (C) COS-7 cells weretransfected with GFP-vimentin wild type (wt) or C328S mutant constructs and treated with PGA1-B as above. Total cell lysates weresubjected to chromatography on Neutravidin-agarose. Total lysates and avidin-bound fractions were analyzed for the presence ofGFP-vimentin by SDS-PAGE and Western blot with antiGFP antibody. Results shown are representative of three assays with similarresults.



the small subunit of the ribosome. In addition, the activityand/or expression of this factor have been related to thegrowth status of the cell and it has been associated withcertain tumors.38 It will be interesting to explore whether themodification of these proteins by PGA1 may be involved inits antitumoral effects.

Several of the targets identified are cytoskeletal proteins.We have previously observed that both 15d-PGJ2-B andPGA1-B bind to a spot co-migrating with actin in 2Dgels.23 Moreover, actin can be modified by these two cyPGin vitro. Actin has also been shown to be the target for theaddition of the reactive aldehyde 4-hydroxy-2-nonenal.39

The modification by these electrophilic compounds appearsto map at the cysteine residue located in the carboxylend of the protein. Interestingly, the modification of actinhas been shown to correlate with disruption of the actincytoskeleton in neuroblastoma cells40 and with alteration offilament structure under in vitro polymerization conditions.23

Tubulin is a cysteine-rich protein that has been shownto be the target for thiolation, nitrosylation or oxidationunder various circumstances associated with oxidative ornitrosative stress.41,42 Alkylation of functional sulphydrylgroups in tubulin may lead to impairment of microtubulepolymerization and cell division.43 In fact, several agents thatspecifically target cysteine residues of ˇ-tubulin have beenassayed for their antitumor properties.43 We had previouslyidentified tubulin in a proteomic study of 15d-PGJ2-B-modified proteins in renal mesangial cells in associationwith morphological changes in the microtubule network.22

However, since cyPG can induce a potent oxidative stressin several cell types, it is important to take into accountthe possibility that some of the effects observed are causedindirectly by oxidative modifications.41

Vimentin is a constituent of the intermediate filamentnetwork and it has long been considered only as a structuralprotein. However, recent reports are uncovering expandingroles for vimentin in signal transduction processes. Forinstance, fragments of vimentin have been reported tosterically hinder the enzymatic dephosphorylation of ERK,44

and a role of vimentin phosphorylation in the regulation ofPAK activation has also been proposed.45 The modificationof vimentin by thiolation or oxidation has been reported invarious proteomic studies.27,46 However, the structural orfunctional consequences of these modifications are poorlyunderstood. In a recent study we identified vimentinas a potential target for the addition of 15d-PGJ2-B, inassociation with a profound alteration of the intermediatefilament network. Interestingly, a vimentin mutant lackingthe only cysteine residue of this protein appeared toconfer partial resistance against 15d-PGJ2-induced vimentindisorganization in transient transfections. For these reasonswe judged it interesting to carry out a more detailed study ofvimentin modification. Here we have shown that both PGA1

and PGA1-B can modify cysteine 328 of vimentin in vitro,as deduced from the peptide mass fingerprint. Moreover,the results from the transfection of vimentin wild typeand C328S constructs, indicate that cysteine 328 is alsothe main site for cyPG binding to vimentin in intact cells.Future studies using vimentin mutants will contribute to the

detailed characterization of the functional consequences ofvimentin modification.

In summary, the results reported herein constitute thefirst attempt to identify protein targets for modification byPGA1-B through proteomic techniques. The use of similarprocedures with structural analogs of cyPG of increasedstability or potency could help in the evaluation of thesecompounds as potential leads for the development oftherapeutic agents.

Supplementary materialSupplementary electronic material for this paper is availablein Wiley InterScience at: http://www.interscience.wiley.com/jpages/1076-5174/suppmat/.

AcknowledgementsThis work was supported by grants SAF2006-03489 (Ministerio deEducacion y Ciencia) and 0179/1 (Fundacion La Caixa) to D.P.-S. J.G.is the recipient of a pre-doctoral fellowship from the I3P Program(C.S.I.C., Fondo Social Europeo). B. G. is the recipient of a fellowshipfrom the FPI program (Ministerio de Educacion y Ciencia). The stayof J.G. at the laboratory of J.T. was supported by EMBO and FEBSshort term fellowships. The technical assistance of M.J. Carrasco isgratefully acknowledged. We thank Dr V. de los Rıos and A. Prieto,from C.I.B, for help with mass spectrometry.

REFERENCES1. Noyori R, Suzuki M. Organic synthesis of prostaglandins:

advancing biology. Science 1993; 259: 44.2. Santoro MG, Benedetto A, Carruba G, Garaci E, Jaffe BM.

Prostaglandin A compounds as antiviral agents. Science 1980;209: 1032.

3. Pica F, De Marco A, De Cesare F, Santoro MG. Inhibition ofvesicular stomatitis virus replication by delta 12-prostaglandinJ2 is regulated at two separate levels and is associated withinduction of stress protein synthesis. Antiviral Research 1993; 20:193.

4. Pica F, Palamara AT, Rossi A, De Marco A, Amici C,Santoro MG. Delta(12)-prostaglandin J(2) is a potent inhibitorof influenza A virus replication. Antimicrobial Agents andChemotherapy 2000; 44: 200.

5. Rossi A, Elia G, Santoro G. Inhibition of nuclear factor �Bby prostaglandin A1: an effect associated with heat shocktranscription factor activation. Proceedings of the National Academyof Sciences of the United States of America 1997; 94: 746.

6. Ito H, Okamoto K, Kato K. Prostaglandins stimulate the stress-induced synthesis of hsp27 and alpha B crystallin. Journal ofCellular Physiology 1997; 170: 255.

7. Rossi A, Kapahi P, Natoli G, Takahashi T, Chen Y, Karin M,Santoro MG. Anti-inflammatory cyclopentenone prostaglandinsare direct inhibitors of I�B kinase. Nature 2000; 403: 103.

8. Straus DS, Pascual G, Li M, Welch JS, Ricote M, Hsiang CH,Sengchanthalangsy LL, Ghosh G, Glass CK. 15-deoxy-delta12,14-prostaglandin J2 inhibits multiple steps in the NF-kappa Bsignaling pathway. Proceedings of the National Academy of Sciencesof the United States of America 2000; 97: 4844.

9. Cernuda-Morollon E, Pineda-Molina E, Canada FJ, Perez-Sala D. 15-Deoxy-12,14-prostaglandin J2 inhibition of NF-�BDNA binding through covalent modification of the p50 subunit.Journal of Biological Chemistry 2001; 276: 35530.

10. Zernecke A, Erl W, Fraemohs L, Lietz M, Weber C. Suppressionof endothelial adhesion molecule up-regulation withcyclopentenone prostaglandins is dissociated from IkappaB-alpha kinase inhibition and cell death induction. FASEB Journal2003; 17: 1099.

11. Ciucci A, Gianferretti P, Piva R, Guyot T, Snape TJ, Roberts SM,Santoro MG. Induction of apoptosis in estrogen receptor-negative breast cancer cells by natural and synthetic



cyclopentenones: role of the IkappaB kinase/nuclear factor-kappaB pathway. Molecular Pharmacology 2006; 70: 1812.

12. Zhang HL, Huang ZH, Zhu Y, Liang ZQ, Han R, Wang XX,Chase TN, Qin ZH. Neuroprotective effects of prostaglandinA1 in animal models of focal ischemia. Brain Research 2005; 1039:203.

13. Kato T, Fukushima M, Kurozumi S, Noyori R. Antitumoractivity of delta 7-prostaglandin A1 and delta 12-prostaglandinJ2 in vitro and in vivo. Cancer Research 1986; 46: 3538.

14. Ikai K, Ujihara M, Kashihara M, Fukushima M. Inhibition of theproliferation of transformed epidermal cells in culture by variousprostaglandins. The Journal of Investigative Dermatology 1987; 89:69.

15. Suzuki M, Kiho T, Tomokiyo K, Furuta K, Fukushima S,Takeuchi Y, Nakanishi M, Noyori R. Rational design ofantitumor prostaglandins with high biological stability. Journalof Medicinal Chemistry 1998; 41: 3084.

16. Sasaki H, Niimi S, Akiyama M, Tanaka T, Hazato A,Kurozumi S, Fukushima S, Fukushima M. Antitumor activity of13,14-dihydro-15-deoxy-delta7-prostaglandin-A1-methyl esterintegrated into lipid microspheres against human ovariancarcinoma cells resistant to cisplatin in vivo. Cancer Research1999; 59: 3919.

17. Narumiya S, Ohno K, Fukushima M, Fujiwara M. Site andmechanism of growth inhibition by prostaglandins. III.Distribution and binding of prostaglandin A2 and delta12-prostaglandin J2 in nuclei. Journal of Pharmacology andExperimental Therapeutics 1987; 242: 306.

18. Parker J. Prostaglandin A2 protein interactions and inhibition ofcellular proliferation. Prostaglandins 1995; 50: 359.

19. Moos PJ, Edes K, Cassidy P, Massuda E, Fitzpatrick FA.Electrophilic prostaglandins and lipid aldehydes repress redox-sensitive transcription factors p53 and Hypoxia-inducible factorby impairing the selenoprotein thioredoxin reductase. Journal ofBiological Chemistry 2003; 278: 745.

20. Sanchez-Gomez FJ, Gayarre J, Avellano MI, Perez-Sala D. Directevidence for the covalent modification of glutathione-S-transferase P1-1 by electrophilic prostaglandins: implications forenzyme inactivation and cell survival. Archives of Biochemistryand Biophysics 2007; 457: 150.

21. Gayarre J, Stamatakis K, Renedo M, Perez-Sala D. Differentialselectivity of protein modification by the cyclopentenoneprostaglandins PGA1 and 15-deoxy-12,14-PGJ2: role ofglutathione. FEBS Letters 2005; 579: 5803.

22. Stamatakis K, Sanchez-Gomez FJ, Perez-Sala D. Identificationof novel protein targets for modification by 15-deoxy-12,14-prostaglandin J2 in mesangial cells reveals multiple interactionswith the cytoskeleton. Journal of the American Society of Nephrology2006; 17: 89.

23. Gayarre J, Sanchez D, Sanchez-Gomez FJ, Terron C, Llorca O,Perez-Sala D. Addition of electrophilic lipids to actinalters filament structure. Biochemical and Biophysical ResearchCommunications 2006; 349: 1387.

24. Suckau D, Resemann A, Schuerenberg M, Hufnagel P, Franzen J,Holle A. A novel MALDI LIFT-TOF/TOF mass spectrometer forproteomics. Analytical and Bioanalytical Chemistry 2003; 376: 952.

25. Hamnell-Pamment Y, Lind C, Palmberg C, Bergman T,Cotgreave IA. Determination of site-specificity of S-glutathionylated cellular proteins. Biochemical and BiophysicalResearch Communications 2005; 332: 362.

26. Danielyan L, Tolstonog G, Traub P, Salvetter J, Gleiter CH,Reisig D, Gebhardt R, Buniatian GH. Colocalization of glialfibrillary acidic protein, metallothionein, and MHC II in human,rat, NOD/SCID, and nude mouse skin keratinocytes andfibroblasts. The Journal of Investigative Dermatology 2007; 127:555.

27. Fratelli M, Demol H, Puype M, Casagrande S, Eberini I,Salmona M, Bonetto V, Mengozzi M, Duffieux F, Miclet E,Bachi A, Vandekerckhove J, Gianazza E, Ghezzi P. Identificationby redox proteomics of glutathionylated proteins in oxidatively

stressed human T lymphocytes. Proceedings of the NationalAcademy of Sciences of the United States of America 2002; 99: 3505.

28. Gayarre J, Avellano MI, Sanchez-Gomez FJ, Carrasco MJ,Canada FJ, Perez-Sala D. Modification of proteins bycyclopentenone prostaglandins is differentially modulated byGSH in vitro. Annals of the New York Academy of Sciences 2007;1096: 78.

29. Tanikawa M, Yamada K, Tominaga K, Morisaki H, Kaneko Y,Ikeda K, Suzuki M, Kiho T, Tomokiyo K, Furuta K, Noyori R,Nakanishi M. Potent prostaglandin A1 analogs that suppresstumor cell growth through induction of p21 and reduction ofcyclin E. Journal of Biological Chemistry 1998; 273: 18522.

30. Furuta K, Tomokiyo K, Satoh T, Watanabe Y, Suzuki M.Designed prostaglandins with neurotrophic activities.Chembiochem 2000; 1: 283.

31. Verbitski SM, Mullally JE, Fitzpatrick FA, Ireland CM.Punaglandins, chlorinated prostaglandins, function as potentMichael receptors to inhibit ubiquitin isopeptidase activity. Jour-nal of Medicinal Chemistry 2004; 47: 2062.

32. Chiang PC, Kung FL, Huang DM, Li TK, Fan JR, Pan SL,Shen YC, Guh JH. Induction of Fas clustering and apoptosisby coral prostanoid in human hormone-resistant prostate cancercells. European Journal of Pharmacology 2006; 542: 22.

33. Martinez-Ruiz A, Villanueva L, Gonzalez de Orduna C,Lopez-Ferrer D, Higueras MA, Tarin C, Rodriguez-Crespo I,Vazquez J, Lamas S. S-nitrosylation of Hsp90 promotes theinhibition of its ATPase and endothelial nitric oxide synthaseregulatory activities. Proceedings of the National Academy ofSciences of the United States of America 2005; 102: 8525.

34. Gao C, Guo H, Wei J, Mi Z, Wai PY, Kuo PC. Identification of S-nitrosylated proteins in endotoxin-stimulated RAW264.7 murinemacrophages. Nitric Oxide 2005; 12: 121.

35. Lamberti A, Caraglia M, Longo O, Marra M, Abbruzzese A,Arcari P. The translation elongation factor 1A in tumorigenesis,signal transduction and apoptosis: review article. Amino Acids2004; 26: 443.

36. Ejiri S. Moonlighting functions of polypeptide elongation factor1: from actin bundling to zinc finger protein R1-associatednuclear localization. Bioscience, Biotechnology, and Biochemistry2002; 66: 1.

37. Cumming RC, Andon NL, Haynes PA, Park M, Fischer WH,Schubert D. Protein disulfide bond formation in the cytoplasmduring oxidative stress. Journal of Biological Chemistry 2004; 279:21749.

38. Hernandez G, Vazquez-Pianzola P. Functional diversity of theeukaryotic translation initiation factors belonging to eIF4families. Mechanisms of Development 2005; 122: 865.

39. Aldini G, Dalle-Donne I, Vistoli G, Maffei Facino R, Carini M.Covalent modification of actin by 4-hydroxy-trans-2-nonenal(HNE): LC-ESI-MS/MS evidence for Cys374 Michael adduction.Journal of Mass Spectrometry 2005; 40: 946.

40. Aldini G, Carini M, Vistoli G, Shibata T, Kusano Y, Gam-beroni L, Dalle-Donne I, Milzani A, Uchida K. Identification ofactin as a 15-deoxy-delta(12,14)-prostaglandin J(2) target inneuroblastoma cells: mass spectrometric, computational, andfunctional approaches to investigate the effect on cytoskeletalderangement. Biochemistry 2007; 46: 2707.

41. Ishii T, Uchida K. Induction of reversible cysteine-targetedprotein oxidation by an endogenous electrophile 15-deoxy-delta(12,14)-prostaglandin J2. Chemical Research in Toxicology2004; 17: 1313.

42. Jaffrey SR, Erdjument-Bromage H, Ferris CD, Tempst P,Snyder SH. Protein S-nitrosylation: a physiological signal forneuronal nitric oxide. Nature Cell Biology 2001; 3: 193.

43. Casini A, Scozzafava A, Supuran CT. Cysteine-modifyingagents: a possible approach for effective anticancer and antiviraldrugs. Environmental Health Perspectives 2002; 110: 801.

44. Perlson E, Michaelevski I, Kowalsman N, Ben-Yaakov K,Shaked M, Seger R, Eisenstein M, Fainzilber M. Vimentinbinding to phosphorylated Erk sterically hinders enzymatic



dephosphorylation of the kinase. Journal of Molecular Biology2006; 364: 938.

45. Li QF, Spinelli AM, Wang R, Anfinogenova Y, Singer HA,Tang DD. Critical role of vimentin phosphorylation at ser-56 byp21-activated kinase in vimentin cytoskeleton signaling. Journalof Biological Chemistry 2006; 281: 34716.

46. West MB, Hill BG, Xuan YT, Bhatnagar A. Protein glutathiola-tion by nitric oxide: an intracellular mechanism regulating redoxprotein modification. FASEB Journal 2006; 20: 1715.


Documents

Study of protein targets for covalent modification by the antitumoral and anti-inflammatory prostaglandin PGA1: focus on vimentin