MC

S*Q2

R

tsItdwtezapitca

g

awtcteadeC

matp

du

Biochemical and Biophysical Research Communications 288, 1006–1010 (2001)

doi:10.1006/bbrc.2001.5881, available online at http://www.idealibrary.com on

0CA

olecular Characterization of Human and Bovineeruloplasmin Using MALDI-TOF Mass Spectrometry

tephane Boivin,*,†,1 M’hammed Aouffen,*,1 Alain Fournier,† and Mircea-Alexandru Mateescu*,2

Department of Chemistry and Biochemistry, Universite du Quebec a Montreal, C.P. 8888, Succ. Centre-ville, Montreal,uebec H3C 3P8, Canada; and †Centre de recherche en sante humaine, INRS-Institut Armand Frappier,45 boulevard Hymus, Pointe-Claire, Quebec H9R 1G6, Canada

eceived September 28, 2001

sis (7, 8). In humans, ceruloplasmin (hCP) is synthe-s1csmobispaggf(can1tpaudc(sake5«iwetpfiC

Using SDS–PAGE and MALDI-TOF mass spectrome-ry, we investigated the difference in the moleculartructure between human and bovine ceruloplasmin.n both cases, we found that the protein is present inwo majors forms of different molecular mass. Theifference between human and bovine ceruloplasminas more obvious when characterized by MALDI-TOF

han with the SDS–PAGE analysis. Furthermore, westablished that the N-glycoside content of both en-ymes is dissimilar and that the N-glycosyl moietiesre distributed in a distinctive fashion in two glyco-roteins. Finally, it appeared that both proteins exhib-

ted different cleavage patterns after treatment withrypsin. This study indicates that human and bovineeruloplasmin differ not only in sugar composition butlso in primary structure. © 2001 Academic Press

Key Words: ceruloplasmin; MALDI-TOF; fingerprints;lycosylation.

There is a growing interest for ceruloplasmin, an2-globulin which is a blue copper oxidase (EC 1.16.3.1)ith important physiological functions and potential

herapeutic applications. It plays an important role inopper transport, being responsible in vertebrates of upo 95% of the copper storage (1). This glycoproteinxhibits multiple functions such as ferroxidase andmine oxidase activities, copper mobilization, antioxi-ant properties (2–4) and, as previously described, ex-rts cardioprotective (5) and antifibrillatory (6) actions.eruloplasmin (CP) seems also involved in angiogene-

Abbreviations used: BSA, bovine serum albumin; CP, ceruloplas-in; KPi, potassium phosphate buffer; MALDI-TOF MS, matrix-

ssisted laser desorption/ionization time-of-flight mass spectrome-ry; PNGase F, N-glycosidase; SDS–PAGE, sodium dodecyl sulfate–olyacrylamide gel electrophoresis.

1 Equal contribution.2 To whom correspondence and reprint requests should be ad-

ressed. Fax: (514) 987 4054. E-mail: mateescu.m-alexandru@qam.ca.

1006006-291X/01 $35.00opyright © 2001 by Academic Pressll rights of reproduction in any form reserved.

ized in hepatocytes as a single polypeptide chain of046 amino acids and is secreted into the plasma as aopper-containing glycoprotein exhibiting four glyco-ylation sites (9). As estimated by SDS–PAGE, thisolecular arrangement accounts for a molecular mass

f about 132 kDa (4). Heterogeneity of ceruloplasminetween species (human, mice, rat, horse, bovine) andntra-species (differential glycosylation and alternativeplicing mRNA) has been demonstrated by several re-orts (10–12). Recently, Yang and colleagues (13) ex-mined the mRNA of hCP, and identified an isoformenerated by an alternative splicing from a single CPene. Moreover, Sternlieb identified a new 200 kDaorm of hCP which is recognized by a specific antibody14). On the other hand, Gitlin and coll. (15, 16) haveharacterized the rat and mouse CP as a 1062 aminocid polypeptide chain. The equine ceruloplasmin wasot sequenced but molecular weight of 120 kDa and15 kDa were estimated by SDS–PAGE and gel filtra-ion respectively (17). Surprisingly, for bovine cerulo-lasmin (bCP) there is not much information. Jenkinsnd colleagues (18) estimated by SDS–PAGE a molec-lar weight of 100 kDa. The propensity of hCP to lyticegradation was reported and purified hCP fragments,orresponding to 115, 70, 50, and 20 kDa were isolated19). When hCP was digested by V8 protease, a certainequence overlapping was observed between the 67nd 50 kDa fragments as well as between the 50 and 19Da fragments (4). This CP fragmentation can appearven during the storage. The formation of the 67 and0 kDa fragments is prevented by the addition of-amino caproic acid, suggesting a metalloprotenasemplication (20). Variations in molecular mass of hCPere observed during various purification steps andncountered with both commercial or fresh prepara-ions (20). It is yet unclear what is the origin of thisroteolysis as no contaminating proteases were identi-ed. Ryan and colleagues (10) demonstrated that ratP is more resistant to plasmin-mediated proteolysis

than human CP. Moreover, the rat 116- and 20-kDa CPfpwti

SswBsbNdc

M

fpstScG

wlapdofim2p

wewoom

arw

awbsGpPsA

tws

(twosMtiuf

R

(nw1bstpapttivltdstwuslCnsscmhOR

m(p

Vol. 288, No. 4, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

ragments appeared to be more resistant to furtherroteolysis, compared to the human counterpartshich are cleaved to smaller fragments. Variations in

he primary structures could explain these variationsn stability between human and rat CP (10).

The aim of this study was to further document, usingDS–PAGE and MALDI-TOF mass spectrometry, thetructural data related to the various moleculareights reported for human and bovine ceruloplasmin.esides the global mass estimate, we also evaluatedome of the glycosylation features of the two proteinsy comparing their behavior when treated with a-glycosidase F (PNGase F). Finally, we explored theistinct sensitivity of hCP and bCP toward trypsin andompared their tryptic digests.

ATERIALS AND METHODS

(a) Materials. Purified human ceruloplasmin was purchasedrom Calbiochem (San Diego, CA), N-glycosidase (PNGase F) wasrovided by New England LabSystem (Beverly, MA), Trypsin, bovineerum albumin (BSA), sinapinic acid (3,5-dimethoxy-4-hydroxy-rans-cinnaminic acid) matrix and «-amino caproic acid were fromigma-Aldrich Canada Ltd. (Oakville, Ontario, Canada). All chemi-als were “reagent grade” and used without further purification.elCode Carbohydrate Staining Kit was from Pierce (Rockford, IL),

(b) Purification of bovine ceruloplasmin. Bovine ceruloplasminas purified from serum according to the method described by Ca-

abrese et al. (21) and modified by Wang et al. (22) using single-stepffinity chromatography on an AE (aminoethyl)-agarose column, inresence of 5 mM «-amino caproic acid added to prevent proteolysisuring the purification steps. The obtained protein presented a ratiof A 610/A 280 $ 0.04, which is considered as satisfactory for the puri-ed CP. Protein concentrations were measured with the Bradfordethod using BSA as a standard (23). Ceruloplasmin was stored at20°C, in the elution buffer (100 mM potassium phosphate buffer,H 7.4), until use.

(c) Deglycosylation of ceruloplasmin. Both human and bovine CPere deglycosylated using N-glycosidase F, as described by Aouffen

t al. (24). Briefly, 1 mg aliquots (50 ml) of CP were incubated at 37°Cith 3000 U of PNGase F for 24 h. The incubation mix was appliedn an AE-agarose column followed by washings with three volumesf 10 mM KPi buffer. Then, deglycosylated CP was eluted with 100M KPi buffer, pH 7.4, and stored at 220°C before use.

(d) Trypsinolytic patterns. Complete trypsinolytic mapping waschieved by incubation of CP with trypsin (2500 U/mg protein) at aatio of 1:100 in 50 mM KPi, pH 7.4, for 24 h at 37°C; the reactionas stopped by freezing at 220°C.

(e) SDS–PAGE analysis. Gel electrophoresis was carried out withProtean II cell (Bio-Rad). Briefly, 15 ml of various CP preparationsere heated in the presence of b-mercaptoethanol and then resolvedy SDS–PAGE on a 10% polyacrylamide gel. The resulting gel wastained with Coomassie blue to reveal the protein bands and with theelCode carbohydrate stain to detect the carbohydrate groups of theroteins. The molecular mass of protein bands was calculated with aersonal Densitometer SI (Sunnyvale, CA) using an IPLab Gel Poftware (Vienna, VA) and molecular weight standards (Sigma-ldrich).

(f) MALDI-TOF analysis. Before the mass spectrometry analysis,he samples already solubilized in 100 mM phosphate buffer, pH 7.4,ere dialyzed against a solution of 25 mM Tris–HCl, pH 7.4. The

inapinic acid matrix was premixed (10 mg/ml) in 50% acetonitrile

1007

ACN) containing 0.1% aqueous trifluoroacetic acid (TFA). Calibra-ion was performed using BSA (66,400 Da). The samples were mixedith the chemical matrix in a ratio of 1:10 (1 ml of sample for 10 mlf matrix) and 1 ml of each sample were spotted onto the stainless-teel plate, dried, and analyzed by mass spectrometry using aALDI-TOF Voyager DE mass spectrometer (Perspective Biosys-

ems, Framingham, MA). All spectra were acquired in the positive-on filter mode. Bovine CP was compared to the human CP isoformnder various conditions, i.e., native, deglycosylated and trypsinizedorms.

ESULTS AND DISCUSSION

Human and bovine CP were analyzed by SDS–PAGE10%) to examine their purity (Fig. 1). For both formsative CP showed two bands with apparent moleculareights of 129 and 116 kDa for human CP and 124 and10 kDa for bovine CP. The presence of two distinctands for CP is a phenomenon discussed in severaltudies. Sato and colleagues (14) suggested that thewo bands correspond to two glycosylation levels ofrotein. However, other reports rather suggested that116-kDa fragment is generated from the 132-kDa

rotein following a proteolytic cleavage (19, 20). First,he densitometric analysis of the two bands revealedhat the heavier band is more abundant in bovine thann human CP, which exhibits the smaller band moreisible than that of bovine CP. The difference in mo-ecular weight between the two CP forms remains in-riguing. As suggested, a possible explanation for thisifference can be a variation in their respective glyco-ylation level. Thus, in order to estimate the role ofheir carbohydrate moieties, human and bovine CPere N-deglycosylated using PNGase F and analyzedsing SDS–PAGE. A shift in molecular mass corre-ponding to 4 and 2 kDa for bovine and human ceru-oplasmin respectively was observed (Fig. 1). BovineP was completely deglycosylated by PNGase F sinceo band was detected with the Pierce carbohydratetaining system (not shown). On the contrary, ashown with the staining visualization, human CP stillontained some carbohydrates after PNGase F treat-ent. Ryden and Bjork (25) reported two forms of

uman CP that differed only in carbohydrate content.ur data are in agreement with those reported byyden and Bjork suggesting that the two species of CP

FIG. 1. SDS–PAGE of bovine and human ceruloplasmins: (a)olecular weight standards (kDa), (b) native bovine ceruloplasmin,

c) N-deglycosylated bovine ceruloplasmin, (d) native human cerulo-lasmin, and (e) N-deglycosylated human ceruloplasmin.

plnlastfesp1tdeaMttmCioda

st(plc2cmtwtldictt

smotMb

c

Vol. 288, No. 4, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

resent some differences in their nature and in theevel of glycosylation. Because the SDS–PAGE tech-ique is not particularly accurate, a more precise mo-

ecular weight determination was utilized. A deepernalysis was thus performed using MALDI-TOF masspectrometry, a technique exhibiting a better resolu-ion than gel electrophoresis. It is a highly efficient toolor the measurement of molecular mass of biomol-cules. The spectra of bovine and human CP (Fig. 2)how two distinct molecular masses for each cerulo-lasmin form. Bovine CP exhibited peaks at 125.1 and06.5 kDa respectively, whereas human CP showedwo forms at 129.1 and 110.8 kDa. These MALDI-TOFata confirmed the presence of the two forms of differ-nt molecular masses already observed by SDS–PAGEnalysis. However, contrarily to SDS–PAGE, theALDI-TOF analysis revealed further differences be-

ween bovine and human ceruloplasmin. Thus, sub-raction of the small segment from the large one gave aass difference of 18.5 kDa for both bovine and humanP, suggesting that a defined small segment is specif-

cally generated by proteolysis. To identify localizationf carbohydrate moieties in both ceruloplasmin forms,eglycosylated bovine and human ceruloplasmin werenalyzed by MALDI-TOF spectrometry. Analysis of the

FIG. 2. MALDI-TOF mass spectra of (a) native bovine ceruloplaeruloplasmin and (d) N-deglycosylated human ceruloplasmin. Each

1008

pectra (Figs. 2b and 2d) indicated that the removal ofhe carbohydrates represented the same proportion4%) in mass loss for both segments of bovine cerulo-lasmin. Interestingly, in the case of the human ceru-oplasmin segments, there are differences in terms ofarbohydrate ratio (1.54% for the large segment and.27% for the small one). These data suggest that thearbohydrate moieties of human ceruloplasmin areore complex and seemingly not easily accessible on

he protein surface. These results are in agreementith data reported by Yamashita et al. (26), describing

he level and the complexity of the carbohydratesinked to human ceruloplasmin. The role of carbohy-rate in the structural and functional behaviors of CPs still poorly documented. However, we showed re-ently that deglycosylated bovine ceruloplasmin main-ains its enzymatic, antioxidant and cardioprotectiveherapeutic properties (24).

Peptide mapping is a standard procedure in proteinequence analysis as well as in protein to protein ho-ology assessment. We evaluated the tryptic profiles

f bovine and human CP taking the advantage ofhe high sensitivity and resolving characteristics ofALDI-TOF mass spectrometry. Trypsin digests of

oth proteins are presented in Fig. 3. It must be men-

n and (b) native human ceruloplasmin. (c) N-deglycosylated bovineectrum is representative of four independent analyses.

smisp

tdldb3HCpmcpacbdttt

steo

mkwFscamdtabPmf

A

itIb

d

Vol. 288, No. 4, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

ioned that sinapinic acid used as a matrix is useful toetect proteins up to 200 kDa, but is not efficient forow masses (,10 kDa). Therefore, following the trypsinigestion of both bovine and human CP, only the regionetween 9 to 145 kDa was analyzed. As shown in Fig., no peaks were observed in region the above 55 kDa.uman CP appears to be more sensitive than bovineP to trypsin proteolysis. Except for a few peptidesresent in both digests (at 18.2 and 23.3 kDa), theapping was completely different. These results indi-

ate that: (a) the fragment of 18 kDa is released byroteolysis and, (b) there are differences in molecularnd structural features between bovine and humaneruloplasmin. The difference between the human andovine ceruloplasmin fingerprints can be related toifferences in their primary structure. In fact, it seemshat bCP has two or three trypsin-sensitive sites lesshan hCP, which shows a larger diversity following aryptic profile digestion.

In conclusion, this MALDI-TOF MS study repre-ents an advanced evaluation of the molecular charac-eristics of ceruloplasmin as it allows a more precisestimate of the molecular mass as well as of the heter-geneity of both human and bovine ceruloplasmin. The

FIG. 3. MALDI-TOF mass spectra of fingerprints of (a) nativeigestion conditions were 24 h at 37°C in a 1:100 ratio. Each spectr

1009

olecular mass of human CP was established at 129.8Da and that of bovine CP at 125.1 kDa. These valuesere well correlated with the SDS–PAGE analysis.urthermore, for both bCP and hCP, we found amaller component of CP preparation obtained by ex-ision of a 19-kDa fragment following a proteolyticction on the native form. Deglycosylation and peptideapping data suggested that human and bovine CP

iffers from each other not only in terms of glycosyla-ion level and localization of carbohydrate moieties, butlso in their structure and proteolytic sites. The com-ination of MALDI-TOF mass spectrometry and SDS–AGE analysis appeared as a powerful tool for theolecular mass characterization of ceruloplasmin iso-

orms obtained from two distinct species.

CKNOWLEDGMENTS

We thank Dr. Joanne Paquin (UQAM) for helpful discussion. M.A.s the holder of a Ph.D. studentship from the Fonds pour la Forma-ion de Chercheurs et l’Aide a la Recherche (FCAR/University–ndustry Program) of the Gouvernement du Quebec. AF is supportedy the Fonds de la Recherche en Sante du Quebec (FRSQ).

vine ceruloplasmin and (b) native human ceruloplasmin. Trypticis representative of four independent analyses.

boum

REFERENCES

1

1

1

1

14. Sato, M., Schilsky, M. L., Stockert, R. J., Morell, A. G., and

1

1

1

1

1

2

2

2

2

2

2

2

Vol. 288, No. 4, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

1. Frieden, E., and Hsieh, H. S. (1976) Ceruloplasmin: The coppertransport protein with essential oxidase activity. Adv. Enzymol.44, 187–196.

2. Fleming, R. E., Whitman, I. P., and Gitlin, J. D. (1991) Inductionof ceruloplasmin gene expression in rat lung during inflamma-tion and hyperoxia. Am. J. Physiol. 260, L68–L74.

3. Gitlin, J. D. (1988) Transcriptional regulation of ceruloplasmingene expression during inflammation. J. Biol. Chem. 263, 6281–6287.

4. Takahashi, N., Ortel, T. L., and Putnam, F. W. (1984) Single-chain structure of human ceruloplasmin: The complete aminoacid sequence of the whole molecule. Proc. Natl. Acad. Sci. USA81, 390–394.

5. Chahine, R., Mateescu, M. A., Roger, S., Yamaguchi, N., deChamplain, J., and Nadeau, R. (1991) Protective effects of ceru-loplasmin against electrolysis-induced oxygen free radicals in ratheart. Can. J. Physiol. Pharmacol. 69, 1459–1464.

6. Atanasiu, R., Dumoulin, M.-J., Chahine, R., Mateescu, M.-A.,and Nadeau, R. (1995) Antiarrhythmic effects of ceruloplasminduring reperfusion in the ischemic isolated rat heart. Can.J. Physiol. Pharmacol. 73, 1253–1261.

7. Raju, K. S., Alessandri, G., Ziche, M., and Gullino, P. M. (1982)Ceruloplasmin, copper ions, and angiogenesis. J Natl. CancerInst. 69, 1183–1188.

8. Folkman, J., and Klagsbrun, M.(1987) Angiogenic factors. Sci-ence 235, 442–447.

9. Endo, M., Suzuki, K., Schmid, K., Fournet, B., Karamanos, Y.,Montreuil, J., Dorland, L., van Halbeek, H., and Vliegenthart,J. F. (1982) The structures and microheterogeneity of the carbo-hydrate chains of human plasma ceruloplasmin. A study employ-ing 500-MHz 1H NMR spectroscopy. J. Biol. Chem. 257, 8755–8760.

0. Ryan, T. P., Grover, T. A., and Aust, S. D. (1992) Rat ceruloplas-min: Resistance to proteolysis and kinetic comparison with hu-man ceruloplasmin. Arch. Biochem. Biophys. 293, 1–8.

1. Medda, R. Cara, N., and Floris, G. (1987) Horse plasma cerulo-plasmin molecular weight and subunit analysis. Prep. Biochem.17, 447–454.

2. Lowenstein, H. (1975) Immunochemical investigation on humanceruloplasmin. Partial explanation of the “heterogeneity.” Int. J.Peptide Protein Res. 7, 1–9.

3. Yang, F. M., Friedrichs, W. E., Cupples, R. L., Bonifacio, M. J.,Sandord, J. A., Horton, W. A., and Bowman, B. H. (1990) Humanceruloplasmin. Tissue-specific expression of transcripts pro-duced by alternative splicing. J. Biol. Chem. 265, 10780–10785.

1010

Sternlieb, I. (1990) Detection of multiple forms of human ceru-loplasmin. A novel M r 200,000 form. J. Biol. Chem. 265, 2533–2537.

5. Fleming, R. E., and Gitlin, J. D. (1990) Primary structure of ratceruloplasmin and analysis of tissue-specific gene expressionduring development. J. Biol. Chem. 265, 7701–7707.

6. Klomp, L. W., Farhangrazy, Z. S., Dugan, L. L., and Gitlin, J. D.(1996) Ceruloplasmin gene expression in the murine centralnervous system. J. Clin. Invest. 98, 207–215.

7. Medda, R., Cara, N., and Floris, G. (1987) Horse plasma cerulo-plasmin molecular weight and subunit analysis. Prep. Biochem.17, 447–454.

8. Dooley, D. M. Cote, C. E., Coolbaugh, T. S., and Jenkins P. L.(1981) Characterization of bovine ceruloplasmin. FEBS Lett.131, 363–365.

9. Moshkov, K. A., Lakatos, S., Hajdu, J., Zavodsky, P., andNeifakh, S. A. (1979) Proteolysis of human ceruloplasmin. Somepeptide bonds are particularly susceptible to proteolytic attack.Eur. J. Biochem. 94, 127–134.

0. Ehrenwald, E., and Fox, P. L. (1994) Isolation of nonlabile hu-man ceruloplasmin by chromatographic removal of a plasmametalloproteinase. Arch. Biochem. Biophys. 309, 392–395.

1. Calabrese, L., Mateescu, M. A., Carbonaro, M., and Mondovi, B.(1988) Reexamination of spectroscopic properties of ceruloplas-min freshly isolated with a novel very rapid single-step proce-dure. Biochem. Int. 16, 199–208.

2. Wang, X. T., Dumoulin, M. J., Befani, O., Mondovi, B., andMateescu, M. A. (1994) Joint chromatographic purification ofbovine serum ceruloplasmin and amineoxidase. Prep. Biochem.24, 237–250.

3. Bradford, M. M. (1976) A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein–dye binding. Anal. Biochem. 72, 248–254.

4. Aouffen, M., Paquin, J., De Grandpre, E., Nadeau, R., and Ma-teescu, M. A. (2001) Deglycosylated ceruloplasmin maintains itsenzymatic, antioxidant, cardioprotective, and neuronoprotectiveproperties. Biochem. Cell Biol. 79, 489–497.

5. Ryden, L., and Bjork, I. (1976) Reinvestigation of some physico-chemical and chemical properties of human ceruloplasmin (fer-roxidase). Biochemistry 15, 3411–3417.

6. Yamashita, K., Liang, C. J., Funakoshi, S., and Kobata, A. (1981)Structural studies of asparagine-linked sugar chains of humanceruloplasmin. Structural characteristics of the triantennarycomplex type sugar chains of human plasma glycoproteins.J. Biol. Chem. 256, 1283–1289.