ANALYTICALBIOCHEMISTRY

Analytical Biochemistry 329 (2004) 199–206

www.elsevier.com/locate/yabio

Quantitative gas chromatography/mass spectrometry determination of C-mannosylation of tryptophan residues in glycoproteins

Jean-Pierre Zanetta,a,¤ Alexandre Pons,a Colette Richet,b Guillemette Huet,b

Philippe Timmerman,a Yves Leroy,a Anne Bohin,a Jean-Pierre Bohin,a

Pierre-André Trinel,c Daniel Poulain,c and Jan Hofsteenged

a CNRS Unité Mixte de Recherche 8576, Glycobiologie Structurale et Fonctionnelle,Université des Sciences et Technologies de Lille Bâtiment C9, 59655, Villeneuve d’Ascq Cedex, France

b INSERM U560, Place de Verdun, 59045 Lille Cedex, Francec INSERM E360, 1 Place de Verdun, 59045 Lille Cedex, France

d Friedrich Miescher-Institut, Maulbeerstrasse 66, PO Box 2543, CH-4002, Basel, Switzerland

Received 19 November 2003Available online 10 May 2004

Abstract

C-mannosylation of Trp residue is one of the most recently discovered types of glycosylation, but the identiWcation of these man-nosylated residues in proteins is rather tedious. In a previous paper [Biochemistry 42 (2003) 8342], it was reported that the completeanalysis of all constituents of glycoproteins (sialic acids, monosaccharides, and amino acids) could be determined on the same samplein three diVerent steps of gas chromatography/mass spectrometry of heptaXuorobutyrate derivatives. It was observed that during theacid-catalyzed methanolysis step used for liberation of monosaccharide from classical O- and N-glycans, Trp and His were quantita-tively transformed by the addition of a methanol molecule on their indole and imidazole groups, respectively. These derivatives werestable to acid hydrolysis used for the liberation of amino acids. Since monosaccharide derivatives were also stabilized as heptaXuo-robutyrate derivatives of O-methyl-glycosides, it was suggested that C-mannosides of Trp residues could quantitatively be recovered.Based on the analyses of standard compounds, peptides and RNase 2 from human urine, we report that C(2)-mannosylated Trpcould be quantitatively recovered and identiWed during the step of amino acid analysis. Analyses of diVerent samples indicated thatthis type of glycosylation is absent in bacteria and yeasts. 2004 Elsevier Inc. All rights reserved.

Keywords: Monosaccharide; Amino acids; Gas chromatography; Mass spectrometry; C-mannoside

1

C-mannosylation of Trp ((C-Man-)Trp) residues isone of the most recently discovered types of glycosyla-tion of proteins [1–3]. The biosynthesis of C-mannosyla-tion occurs through a dolichol-linked precursor [4], theexact mechanism giving rise to the formation of a C–Cbond between the �-mannose residue and the C(2)

¤ Corresponding author. Fax: +33-3-20-43-65-55.E-mail address: Jean-Pierre.Zanetta@univ-lille1.fr (J.-P. Zanetta).1 Abbreviations used: (C-Man-)Trp, C-mannosylation of Trp; HFB,

heptaXuorobutyrate; EI, electron impact; CI, chemical ionization;rMR, relative molar response; GC/MS, gas chromatography/massspectrometry; GPI, glycosyl-phosphatidylinositol; HFBAA, heptaXuo-robutyric anhydride; LB, Luria-Bertani; Nle, norleucine.

0003-2697/$ - see front matter 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.ab.2004.02.033

atom of the indole ring of Trp. The reaction is catalyzedby a microsome-associated transferase, whose activitycan be detected in various organisms and most organsfrom mice [4]. Based on several studies, a consensussequence for C-mannosylation, Trp–x–x–Trp, wasdeduced [5], the C-mannosylation occurring on the WrstTrp residue of the sequence, with the possibility that thesecond Trp residue might be replaced by anotheraromatic amino acid. Subsequently, however, exceptionsto this rule were found. For instance, human WbrinogenB�, tenascin, and N Cam L1 contain this motif, but arenot C-mannosylated. In these proteins a large hydro-phobic amino acid follows the Wrst Trp, a feature thatinhibits C-mannosylation (T. Smilda and J. Hofsteenge,

200 J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206

unpublished results). On the other hand, thrombospon-din type 1 repeats are C-mannosylated on Trp residueswithout a Trp or Phe at position +3 [6–8].

Therefore, speciWc techniques had to be developed todetermine the C-mannoside substitution. These methodsgenerally involved the proteolytic cleavage of the proteinand mass spectrometric analysis of peptides having anadditional mass corresponding to a substituted (C-Man-)Trp residue [9]. This approach is eYcient on puriWedmolecules, but the interpretation of the data becomesdiYcult for complex proteins. Since the C–C bond is notcleaved by classical techniques used in monosaccharideanalysis (i.e., acid-catalyzed methanolysis) and since Trpis in large part destroyed during acid hydrolysis used foramino acid determinations, the direct determination of(C-Man-)Trp residues in proteins or mixtures of proteinsremained impossible.

Recently it was reported that the complete analysis ofall classical constituents of glycoproteins (sialic acids,monosaccharides, and amino acids) could be determinedon the same sample using three diVerent steps of gaschromatography/mass spectrometry (GC/MS) as hep-taXuorobutyrate derivatives [10]. This methodologyinvolved Wrst a mild acid hydrolysis and second a GC/MS analysis of the liberated sialic acids as HFBderivatives of their methyl esters. The remaining mate-rial was submitted to acid-catalyzed methanolysis underanhydrous conditions, followed by GC/MS analysis ofthe HFB derivatives of the liberated O-methyl-glyco-sides. The material remaining after this step was sub-jected to a classical hydrolysis for peptide bond cleavageand the liberated compounds were analyzed as HFBderivatives of their isoamyl esters in the same GC/MSsystem [11]. Using this approach, it was observed that,during the acid-catalyzed methanolysis, Trp and Hiswere quantitatively transformed by the addition of amethanol molecule on their indole and imidazolegroups, respectively [10]. These derivatives were stable toacid hydrolysis. Since monosaccharide derivatives werealso stabilized as per-heptaXuorobutyrate derivatives ofO-methyl-glycosides, it was suggested that it should bepossible to quantitatively recover C-mannosides on Trpresidues.

Based on the analyses of puriWed (C-Man-)Trp, ofpeptides containing this amino acid, and of RNase 2from human urine, the Wne chemical formula of the per-HFB derivative of the isoamyl ester was determinedusing electron impact (EI) and chemical ionization (CI)mass spectrometry. Using speciWc reporter ions in the EImode of ionization, traces of this compound could easilybe identiWed and quantiWed in the third step of analysis,i.e., amino acid analysis. After determination of the rela-tive molar response of the (C-Man-)Trp derivative onstandard samples, this methodology was applied todiVerent samples, including mammalian, bacterial, andyeast materials.

Materials and methods

Chemicals

Diazogen was from Acros. HeptaXuorobutyricanhydride (HFBAA; puriss. grade) was from Fluka,Merck, or Acros. Standard amino acids were from Beck-man and Pierce. RNase 2 from human urine was isolatedas previously described [3,5]. Peptides from humanthrombospondin 1 were isolated as described by Hofste-enge et al. [7]. (C-Man-)Trp amino acid glycoside waspuriWed from human urine, essentially as described byGutsche et al. [12]. PuriWed MUC2 [13] was kindlyprovided by Dr. I. Carlstedt (Lund, Sweden).

Bacteria and yeast cultures

Five bacterial strains representative of three subdivi-sions of the proteobacteria were analyzed: Rodobactersphaeroides strain WS8, Azospirillum brasiliense strainCdRif ATCC29710, Xantomonas campestris pv. Citristrain N1, Erwinia chrisanthemi strain 3937, and Esche-richia coli K-12 strain DF214. These cells were grown at30 °C in LB broth without NaCl as previously described[14–18]. The cells were washed three times in water,followed by delipidation using chloroform methanolmixtures. The insoluble pellet (1 mg protein) was submit-ted to the complete procedure of GC/MS analysis.Candida albicans A serotype cells were cultured in 1-LErlenmeyer Xask containing 500 mL of Sabouraud’sbroth at 37 °C on an orbital shaker until conXuence [19].Cells were broken with a French press and centrifuged.The supernatant and the delipidized pellet (1 mg proteineach) were analyzed separately.

GC/MS analysis of sialic acids

The glycoprotein-bound sialic acids were analyzed aspreviously described [4]. BrieXy, the samples (1–20 �gprotein) were hydrolyzed (105 min at 80 °C in 2 M aceticacid), methyl-esteriWed using diazomethane, and trans-formed into heptaXuorobutyrate derivatives [20]. Thevolatile derivatives were analyzed by GC/MS in the EImode of ionization (see below). This step of liberationand analysis of sialic acid was not an absolute require-ment for the determination of C-mannosylation, but itcould provide important additional information on themonosaccharide composition of sialylated glycoproteins.

GC/MS analysis of monosaccharides

After analysis of sialic acids, the samples were driedunder a stream of nitrogen. Alternatively, if this analysiswas not needed, the dry samples were submitted to acid-catalyzed methanolysis (20 h at 80 °C in 0.5–1 mL anhy-drous methanol containing 0.5 M gaseous HCl [10,21]).

J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206 201

After drying under a stream of nitrogen, the sampleswere transformed into HFB derivatives and analyzed byGC/MS in the same system [10]. The acid-catalyzedaddition of MeOH suppressed the delocalization of thecycles of His and Trp, converting nitrogen of the cycleinto a secondary amino group, which subsequentlybecame quantitatively derivatized by HFBAA [10].

GC/MS analysis of amino acids

After monosaccharide analysis, the samples weredried under a stream of nitrogen and, after the additionof 1 nmol of norleucine (Nle) as the internal standard,the samples were submitted to acid hydrolysis (6 N HCl,20 h at 100 °C under a nitrogen atmosphere). After evap-oration under a stream of nitrogen, the samples weresubmitted to a short-time methanolysis (0.2 mL of 0.5 MHCl in anhydrous methanol, 1.5 h at 80 °C) to formmethyl esters and then trans-esteriWed with 0.2 mL of1.5 M HCl in redistilled isoamyl alcohol overnight at100 °C [11]. This two-step derivatization esteriWcationprocedure was required due to the quasi-insolubility ofsome derivatives in the isoamyl reagent. After dryingunder a stream of nitrogen and mild heating with a hairdryer, the mixture was derivatized with HFBAA andanalyzed by GC/MS in the same system. As reportedrecently [10], this method allowed qualitative and quan-titative determination of all amino acids as sharp peaksof HFB derivatives of their isoamyl esters and hexosam-ines as the HFB derivatives of their isoamyl-glycosides.His and Trp derivatives modiWed in the Wrst step ofmethanolysis were quantitatively recovered.

GC/MS analysis

For GC/MS analysis, the GC separation was per-formed on a Carlo Erba GC 8000 gas chromatographequipped with a 25-m £ 0.32-mm CP-Sil5 CB Low bleed/MS capillary column and 0.25-�m Wlm phase (Chrom-pack France, Les Ullis, France). The temperature of theRoss injector was 260 °C and the samples were analyzedusing the following temperature program: 90 °C for3 min, then 5 °C/min until 260 °C, followed by a plateauof 20 min at 260 °C (although the analyses could beshortened, we recommend this temperature program toclean the GC column from unrelated compounds, espe-cially very-long-chain fatty acid methyl esters). For rou-tine experiments, the column was coupled to a FinniganAutomass II mass spectrometer (mass detection limit1000 amu); for studies of masses larger than 1000, it wascoupled to a Nermag 10-10H mass spectrometer (massdetection limit 2000 amu). The analyses were performedroutinely in the EI mode (ionization energy 70 eV;source temperature 150 °C). To preserve the Wlament ofthe ionization source, the GC/MS records wereperformed 5 min after the injection of the sample. The

quantitation of the diVerent constituents was performedon the total ion count of the MS detector using the Xcal-ibur software (Thermoquest–Finnigan). For ascertainingthe mass of the diVerent derivatives, the MS analyseswere also performed in the CI mode in the presence ofammonia (ionization energy 150 eV; source temperature100 °C) with detection of positive ions.

Results and discussion

As previously reported [10], during the step of acid-catalyzed methanolysis used for the liberation of O- andN-glycans from glycoconjugates, the Trp residues (as theHis residues) were transformed by an acid-catalyzedaddition of a CH3OH molecule on the indole (imidazol)cycle. The structures of these derivatives (denoted asHis# and Trp# herein) were determined using their massspectra in the EI and CI modes of ionization [10]. Thesuppression of one double bond inhibited the delocaliza-tion of the double bonds in the cycle(s). This resulted instable compounds, quantitatively recovered in the aminoacid analysis step as the di-HFB derivatives of theisoamyl esters of methoxy derivatives of both Trp(molecular ion at m/z D 698) and His (molecular ion atm/z D 649). The exact position of the CH3O and H addi-tion on either the C(2) or the C(3) carbon atoms of theindole ring could not be determined. Nevertheless, theC(3) position seemed likely, because the delocalization ofthe N(1) nitrogen atom electron doublet would lead tothe appearance of a negative charge on the C(2) carbonatom susceptible to protonation, whereas the electron-attractive eVect of the methylene moiety would lead tothe appearance of a positive charge on the C(3) carbonatom susceptible to nucleophilic attack by CH3O

¡. Thisview was compatible with previous studies demonstrat-ing that the major by-product formed from Trp duringacid hydrolysis was �-3-oxindolyl-alanine [22,23].

The question whether, in (C-Man-)Trp#, the C-man-noside cycle was resistant to the hydrolysis step arose. Infact, as a result of the methanolysis and HFBAA steps,the C-mannose was present not as the free compoundbut rather as its per-HFB derivatives, these compoundsbeing more resistant than nonderivatized sugars to acidhydrolysis [10]. This was substantiated by the quantita-tive recovery of the isoamyl-glycoside of hexosaminesafter acid hydrolysis used for the cleavage of mostpeptide bonds and the high recovery of Man or Glc inthe GC/MS analysis of amino acids of glycoproteins.Since C-mannose was considered to be more stable toacid hydrolysis than Man-O-glycosides, it was suggestedthat (C-Man-)Trp# could possibly be recovered in goodyield.

GC/MS analyses of very heterogeneous mixtures ofglycoproteins from human origin (especially bronchialand colonic mucins), using the routine apparatus with a

202 J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206

mass limit of 1000 amu, presented a peak with intenseions at m/z D 880 and 667, together with minor ionspotentially compatible with the expected structure of(C-Man-)Trp# (M D 1644; Fig. 1). This suggested that(C-Man-)Trp# could be recovered using the proceduredepicted above. Therefore, studies were performed onstandard compounds to verify that the postulated(C-Man-)Trp# was actually this compound. Applyingan arbitrary relative molar response to this derivative of1 relative to Nle gave 3.01 mol of this compound for 1molecule of MUC2 (containing more than 5000 aminoacids). This was in good relation with the occurrence ofthe consensus sequence W–x–x–W (two sequences) andthe putative new consensus sequence [9] W–x–x–F (onesequence).

IdentiWcation of the (C-Man-)Trp# derivative

(C-Man-)Trp, peptides containing this amino acid,and RNase 2 from human urine were subjected to acid-catalyzed methanolysis under anhydrous conditions andformation of HFB derivatives, followed by acid hydroly-sis and formation of HFB derivatives of the isoamylesters. In all cases, peaks corresponding to the putative(C-Man-)Trp found in mucins were observed (data not

shown). Furthermore, the relative abundances of thesepeaks were in gross proportion to that expected fromprevious studies.

To determine the structure of the (C-Man-)Trp deriv-ative, these standard compounds were analyzed on amass spectrometer with a mass limit of 2000 amu. Thefragmentation EI mass spectrum obtained for (C-Man-)Trp# is shown in Fig. 2B. This analysis did not allowdetection of the molecular ion at m/z D 1644. The highestmass ion so far detected was at m/z D 1559 (i.e., M D 85),corresponding to the elimination of an isoamyl (M D 70)and a methyl group (M D 15). Although it was ofextremely low intensity, this ion was clearly distin-guished from the background, and chromatogramreconstitution for the ion perfectly Wtted with all otherions arising from this compound. In fact, all ions in theEI spectrum (the most representative ones are shown inFig. 2B and Table 1) could be explained by the proposedstructure shown in Fig. 2A. Therefore, it was concludedthat this compound corresponded to the (C-Man-)Trp#derivative. As discussed above, uncertainties in the posi-tion of the CH3O group and the conWguration of the C(3)carbon atom of the modiWed indole ring remained. Thepresence of the ion at m/z D 71 in the spectrum of lungmucins (Fig. 1C) indicated the presence of an isoamylester on the carboxyl group of the molecule, a deduction

Fig. 1. GC/MS in the EI mode of ionization on a Finnigan Automass II mass spectometer (mass limit of 1000 amu) of a mixture of glycoproteins(essentially mucins from lung mucus). (A) Chromatogram reconstitution for the ion at m/z D 880 characteristic of the proposed (C-Man-)Trp# deriv-ative. (B) Total ion counts. (C) Fragmentation mass spectrum of the derivative obtained with the routine GC/MS apparatus with a mass limit of1000 amu. Note in (A) and (B) the good separation of the compound from amino acids and from other constituents remaining after acid hydrolysis.Note also in (A) the eYciency of detection of the (C-Man-)Trp# derivative. When analyzing samples containing puriWed (C-Man-)Trp or the sameincluded in a protein or peptide sequence, a minor peak was observed (about 10% of the total), likely corresponding to the stereoisomer of thenucleophilic attack of CH3-O

¡ on the C(3) carbon atom of the indole ring during acid-catalyzed methanolysis. Note also, in addition to amino acids,hexosamine derivatives (recovered as per-HFB derivative of its isoamyl-glycoside) were present as major compounds.

J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206 203

reinforced by the presence of ions corresponding tolosses of mass of 70, 85, and 114/115 from the molecularion (Fig. 2B). The loss of mass of 15 (simultaneous to theloss of 70 D isoamyl group) suggested the presence of anO-methyl group, a conclusion reinforced by losses ofmass of 32 (CH3OH) for interpreting some ions (Fig. 2,legend). The presence of the C-hexoside group was evi-denced by the losses of mass of 947. Nevertheless,because of the absence of standard compounds with hex-oses other than mannose, this method will still not allowone to identify the nature of the C-glycoside bound tothe Trp residue. This point could probably be resolved inthe future, considering that speciWc ions from the hexo-side groups allow the determination of the exact natureof the hexose. Indeed, the EI analysis of HFB derivativesof O-alkyl-glycosides allowed a very easy discriminationbetween the diVerent monosaccharides [10,20,21], sug-gesting that such discrimination could be also possible inthe study of C-hexosides.

Interestingly, all fragment ions obtained on the rou-tine GC/MS apparatus with masses lower than 1000 amushowed intensities similar to those obtained using the

other spectrometer. Therefore, the very intense ions at m/z D 880 and 667 were considered as major reporter ionsof (C-Man-)Trp# (as illustrated in Fig. 1). The intenseion at m/z D 591 was not considered to be a usefulreporter ion, because of its presence in several unrelatedcompounds, including common contaminants of biolog-ical samples.

The structure of (C-Man-)Trp was also analyzed inthe chemical ionization mode in the presence of ammo-nia. The GC/MS analysis (and the direct introductionmethod) gave a very weak pseudo-molecular ion at m/z D 1662 ([M + NH4]+) and higher-intensity ions corre-sponding to fragmentations (data not shown). The frag-mentation of esters (here isoamyl esters) in the CI modeof ionization is a common feature. This point was docu-mented by the presence of an ion at m/z D 1573 corre-sponding to the loss of the isoamyl group. Therefore,both the EI and the CI studies indicated that thehigh-mass ions corresponded to those of the structure of(C-Man-)Trp# (M D 1644) proposed in Fig. 2A.

In the chromatograms of all samples containing(C-Man-)Trp, a minor peak representing less than 10%

Fig. 2. Proposed structure (A) and EI fragmentation mass spectrum (B) of the derivative of (C-Man-)Trp# obtained from the treatment of free(C-Man-)Trp. The addition of the O-methyl group was evidenced by a very minor ion at m/z D M ¡ 32 (m/z D 1612). The major addition (see legendof Fig. 1) of CH3O on the C(3) carbon atom of the indole ring was suggested by the simultaneous elimination of an isoamyl and a methyl group (ionat m/z D 1559; i.e., M ¡ (70 C 15)), the formation of which can be explained only by the formation of a six-atom ring.

204 J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206

of the major one and with an identical fragmentationpattern was observed (see, e.g., Fig. 1A). We assume thatit is the diastereoisomer resulting from the alternativeattack of CH3O

¡ on the C(3) atom. This is corroboratedby the absence of such a component in the Trp deriva-tive. Because this secondary peak was sometimes con-taminated by phthalate derivatives, it was notconsidered in studies on the determination of the relativemolar response of (C-Man-)Trp#.

Determination of the relative molar response of(C-Man-)Trp

The determination of the relative molar response(rMR) of (C-Man-)Trp# involved a quantitative analy-sis of two glycopeptides from human thrombospondin 1,containing both modiWed and unmodiWed Trp residues,and RNase 2 from human urine. As shown in Table 2,the analysis of the diVerent compounds allowed us todetermine the rMR of (C-Man-)Trp# to be very close to1.00, relative to Nle, the classical internal standard usedin amino acids analyses. Using this rMR, the composi-tions of the glycopeptides with respect to all other aminoacids were found to be in close agreement with the theo-retical values. The major diVerence was observed for Thr(known to be partially destroyed during acid hydrolysisof polypeptide chains) as also observed previously [10].The presence of additional Gly was observed, likelyoriginating from contamination. Nevertheless, for thesestandard glycopeptides, having both (C-Man-)Trp and

Trp, it was evident that the two types of compoundscould be recovered and quantiWed using this derivatiza-tion procedure. When degradation of these compoundsdid occur, this was less than 10% for both Trp# and(C-Man-)Trp#, as indicated by the calculated rMR.As a consequence, it was concluded that Trp# and(C-Man-)Trp# could be quantitatively determined usingan rMR factor of 1.000 relative to Nle. These quantita-tive determinations can routinely be performed withnanomolar amounts of initial compound, when 0.5% ofthe sample is injected into the GC/MS apparatus. In factquantitation of (C-Man-)Trp# could reliably be per-formed at the picogram level of the initial compound inthe sample.

Absence of (C-Man-)Trp in bacteria and yeast

The presence of major, speciWc reporter ions for(C-Man-)Trp# at m/z D 667 and 880 and its retentiontime relative to that of Nle allowed its speciWc identiWca-tion. Furthermore, based on the sensitivity of detection,it was calculated that (C-Man-)Trp# could be reliablyidentiWed at a signal/background ratio of 1 residue per5 £ 106 amino acid residues. This sensitivity oVered thepossibility to examine the presence of (C-Man-)Trp indiVerent organisms, including bacteria and yeasts.

Analysis of total homogenates of human and mam-malian cells indicated that (C-Man-)Trp was present inall samples so far analyzed starting from 10 �g of initialprotein material (data not shown). Therefore, the

Table 1Relative intensities and schematic explanation of fragment ions relevant to the structure of the (C-Man-)Trp# derivative shown in Fig. 2A obtainedin the EI mode of ionization

87, CH3(CH3)CHCH2CH2O; 115, CH3(CH3)CHCH2CH2OCñO; 169, CF3CF2CF2; 197, CF3CF2CF2CO; 213, CF3CF2CF2COO or CF3CF2CF2CONH2; 214, CF3CF2CF2COOH; 947, HFB-substituted (C-Man) residue. The values of the intensities of the ions are only indicative since they mayvary with the apparatus used and with experimental parameters such as the Wne adjustment of the column end into the ion source. Nevertheless, itcan be observed that the reporter ions deWned in the text are actually major fragment ions.

Ion Intensity Ion Intensity

1559 0.08% M-15-70 1100 0.37% 1510-197-2131538 0.23% M-19-71-16 1020 0.27% 1217-1971510 0.07% M-19-71-16-28 941 0.15% 1154-2131490 0.04% 1559-69 881 80.08% M-115-197-224-2271461 0.07% M-169-14 880 78.10% M-115-197-224-2271297 0.21% 1510-213 863 3.28% 881-181243 0.15% 1510-70-197 862 10.90% 881-191241 0.18% M-18-19-169-197 861 16.35% 880-191217 0.14% M-213-214 684 7.02% 881-1971204 0.13% M-213-214 683 7.66% 880-1971180 0.14% M-70-197-197 668 84.56% 881-2131166 0.25% M-19-32-213-214 667 100.00% 881-2141155 1.93% M-19-43-213-214 666 40.99% 880-2141154 2.65% M-19-43-214-214 592 37.67% M-946*1153 1.39% M-19-43-214-215 591 47.59% M-9471138 0.52% 1154-16 356 4.71% Core + CH3O + 2O1128 0.42% 1538-197-213 355 10.44% Core + CH3O + 2O1115 0.27% M-32-70-213-214 282 23.67% Core1104 0.85% M-114-213-213 281 43.47% Core

J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206 205

method oVered the possibility to perform a simplescreening for the presence of (C-Man-)Trp in diVerentorganisms. Previous studies on RNase 2 expressed inbacteria [24] have shown that C-mannosylation of thisprotein did not occur in this bacterium, but the exactreasons for this and the occurrence of C-mannosylationin other bacteria have not been examined. Therefore, weinvestigated delipidized total homogenates of Wve diVer-ent bacteria for the presence of (C-Man-)Trp. R. sph-aeroides is a member of the alpha subdivision, whosegenetic analysis is highly developed because it is aremarkable model for the study of bacterial photosyn-thesis. R. sphaeroides shows a close relationship toorganisms that interact with eucaryotic hosts but arethemselves free-living organisms. A. brasilense, anothermember of the � subdivision, is a nitrogen-Wxing bacteriawhich grows in close association with the roots ofgrasses. X. campestris is a plant pathogen of the � subdi-vision. E. chrysanthemi is a pathogenic enterobacterium(� subdivision) responsible for the soft-rot disease of awide range of plants. E. coli K-12 is the well-knownmodel of enterobacteria.

In all these bacterial strains (C-Man-)Trp# was unde-tectable. Given the speciWcity and sensitivity describedabove, (C-Man-)Trp should have been easily identiWedover the background signals if it had been present.When looking for the W–x–x–W motif in protein databases, 542 occurrences were found in the complete E.coli 0157:47 genome (http://www.infobiogen.fr/services/analyseq/cgi-bin/patternp_in.pl) and 24,538 occurrences inthe human genome (http://alces.med.umn.edu/dbmotif.html). The lower occurrence of the W–x–x–W consensus

sequence (about 50 times lower in bacteria than inhuman) was not suYcient to explain the absence ofdetection of (C-Man-)Trp in the proteins of bacteria.Because in these speciWc experiments the quantity ofmaterial injected on the GC/MS apparatus was satu-rating with regard to amino acids, the limit of sensitivityof the method for detecting (C-Man-)Trp was at leastWve orders of magnitude higher than the quantity of(C-Man-)Trp expected from the putative occurrence ofthe consensus sequence proposed above. Therefore, itwas concluded that the absence of (C-Man-)Trp in bac-teria was due to the actual absence of this type of glyco-sylation in bacteria. These data reinforced the previousobservations that RNase 2 expressed in bacteria was notC-mannosylated [24].

We next analyzed C-mannosylation on yeast material.The analyses were performed on C. albicans because it isconsidered an extremely sophisticated yeast strain. Asfor bacteria, the GC/MS analyses indicated the completeabsence of C-mannosides (C-hexosides and C-pento-sides) in this strain. The results on this microorganism(able to synthesize huge amounts of mannosylatedcompounds and having machinery able to synthesize theprecursor for C-mannosylation, dolichol-P-Man)supported the view that, although the precursor of thesynthesis of (C-Man-)Trp is abundant in these organ-isms, the enzymes involved in the biosynthetic pathwayof C-mannosides were absent. This agrees with the previ-ous results that suggested that the yeasts Saccharomycescerevisiae and Schizosaccharomyces pombe did not pos-sess the enzyme involved in C-mannosylation ([24]; S.Hartmann and J. Hofsteenge, unpublished results).

Table 2Composition of standard glycopeptides

a The numbers in parentheses are residue numbers as deWned in [7].b Recovered as S-carboxy-methyl-cysteine. The experiments were performed starting from the following quantities: RNase 2, 21 �g; peptide Th-

RNase 2 (5–10) FT(C-Man)-WAQW, 1.3 nmol; peptide TSP1-(414–426), 4 nmol; peptide TSP1-(469–483), 6.5 nmol. Cman, (C-Man-)Trp#. Thesedata were obtained assigning the rMR of 1.000 for the major peak of this compound.

RNase2 Th-RNase2 TSP1-(414–426)a TSP1-(469–483)a

Ala 6 6.05 1 1.10 0 1 1.05Gly 2 2.19 0 0.21 2 2.26 3 3.12Val 9 8.90 0 0 0Thr 12 10.25 1 0.85 0 0Ser 6 5.85 0 2 1.98 1 0.99Leu 5 4.98 0 0 0Ile 7 6.85 0 0 1 0.99Pro 12 12.00 0 1 1.02 3 3.00Met 4 3.42 0 0 0Phe 5 5.11 1 1.03 0 0Asp 21 22.55 0 1 1.03 2 2.11Lys 4 4.01 0 0 0Tyr 4 4.02 0 0 0Glu 14 14.00 1 1.00 1 1.00 0Arg 8 7.62 0 0 0His 5 4.87 0 1 0.98 0Trp 1 0.96 1 0.98 1 0.97 2 1.95Cysb 8 6.21 0 0 0CMan 1 1.02 1 1.01 2 1.98 1 1.02

206 J.-P. Zanetta et al. / Analytical Biochemistry 329 (2004) 199–206

Conclusion

The analytical approach described here allows thequalitative and quantitative determination of C-man-nosylation in puriWed glycoproteins and glycoproteinmixtures after acid hydrolysis. In fact, as discussed else-where [10], it allows the complete analysis of glycopro-teins (sialic acid diversity, monosaccharide (fatty acids[25]) composition, and amino acid composition) pro-vided the use of an initial additional step of analysis ofsialic acids on the same sample. Furthermore, thisapproach can provide also the complete analysis of GPI-anchored glycoproteins and proteoglycans [10,26], atlevels ranging from 1 to 100 pmol of initial puriWedglycoprotein material.

References

[1] J. Hofsteenge, D.R. Muller, T. de Beer, A. LoZer, W.J. Richter,J.F. Vliegenthart, New type of linkage between a carbohydrateand a protein: C-glycosylation of a speciWc tryptophan residue inhuman RNase Us, Biochemistry 33 (1994) 13524–13530.

[2] T. de Beer, J.F. Vliegenthart, A. LoZer, J. Hofsteenge, The hexo-pyranosyl residue that is C-glycosidically linked to the side chainof tryptophan-7 in human RNase Us is �-mannopyranose, Bio-chemistry 34 (1995) 11785–11789.

[3] A. LöZer, M.A. Doucey, A.M. Jansson, D.R. Muller, T. de Beer,D. Hess, M. Meldal, W.J. Richter, J.F. Vliegenthart, J. Hofsteenge,Spectroscopic and protein chemical analyses demonstrate thepresence of C-mannosylated tryptophan in intact human RNase 2and its isoforms, Biochemistry 35 (1996) 12005–12014.

[4] M.A. Doucey, D. Hess, R. Cacan, J. Hofsteenge, Protein C-mannosylation is enzyme-catalysed and uses dolichyl-phosphate-mannose as a precursor, Mol. Biol. Cell 9 (1998) 291–300.

[5] J. Krieg, S. Hartmann, A. Vicentini, W. Glasner, D. Hess, J. Hof-steenge, Recognition signal for C-mannosylation of Trp-7 inRNase 2 consists of sequence Trp-x-x-Trp, Mol. Biol. Cell 9 (1998)301–309.

[6] J. Hofsteenge, M. Blommers, D. Hess, A. Furmanek, O. Mir-oshnichenko, The four terminal components of the complementsystem are C-mannosylated on multiple tryptophan residues, J.Biol. Chem. 274 (1999) 32786–32794.

[7] J. Hofsteenge, K.G. Huwiler, B. Macek, D. Hess, J. Lawler, D.F.Mosher, J. Peter-Katalinic, C-mannosylation and O-fucosylationof the thrombospondin type 1 module, J. Biol. Chem. 276 (2001)6485–6498.

[8] A. Gonzalez de Peredo, D. Klein, B. Macek, D. Hess, J. Peter-Katalinic, J. Hofsteenge, C-mannosylation and O-fucosylation ofthrombospondin type 1 repeats, Mol. Cell. Proteomics 1 (2002)11–18.

[9] S. Hartmann, J. Hofsteenge, Properdin, the positive regulator ofcomplement, is highly C-mannosylated, J. Biol. Chem. 275 (2000)28569–28574.

[10] A. Pons, C. Richet, C. Robbe, A. Herrmann, P. Timmerman, G.Huet, Y. Leroy, I. Carlstedt, C. Capon, J.P. Zanetta, SequentialGC/MS analysis of sialic acids, monosaccharides and amino acidsof glycoproteins on a single sample as heptaXuorobutyrate deriva-tives, Biochemistry 42 (2003) 8342–8353.

[11] J.P. Zanetta, G. Vincendon, Gas-liquid chromatography of theN(O)-heptaXuorobutyrates of the isoamyl esters of amino acids. I.

Separation and quantitative determination of the constituentamino acids of proteins, J. Chromatogr. 76 (1973) 91–99.

[12] B. Gutsche, C. Grun, D. Scheutzow, M. Herderich, Tryptophanglycoconjugates in food and human urine, Biochem. J. 343 (1999)11–19.

[13] A. Herrmann, J.R. Davies, G. Lindell, S. Martensson, N.H. Packer,D.M. Swallow, I. Carlstedt, Studies on the “insoluble” glycopro-tein complex from human colon. IdentiWcation of reduction-insen-sitive MUC2 oligomers and C-terminal cleavage, J. Biol. Chem.274 (1999) 15828–15836.

[14] P. Talaga, B. Stahl, J.M. Wieruszeski, F. Hillenkamp, S. Tsuyumu,G. Lippens, J.P. Bohin, Cell-associated glucans of Burkholderiasolanacearum and Xanthomonas campestris pv. citri: a new familyof periplasmic glucans, J. Bacteriol. 178 (1996) 2263–2271.

[15] S. Altabe, P. Talaga, J.M. Wieruszeski, G. Lippens, R. Ugalde, J.P.Bohin, Osmoregulated periplasmic glucans of Azospirillum brasil-iense, in: C. Elmerich, A. Kondorosi, W.E. Newton (Eds.), Biologi-cal Nitrogen Fixation for the 21st Century, Kluwer AcademicPublishers, Dordrecht, 1998, p. 390.

[16] J.M. Lacroix, E. Lanfroy, V. Cogez, Y. Lequette, A. Bohin, J.P.Bohin, The mdoC gene of Escherichia coli encodes a membraneprotein that is required for succinylation of osmoregulated peri-plasmic glucans, J. Bacteriol. 181 (1999) 3626–3631.

[17] V. Cogez, P. Talaga, J. Lemoine, J.P. Bohin, Osmoregulated peri-plasmic glucans of Erwinia chysanthemi, J. Bacteriol. 183 (2001)3127–3133.

[18] P. Talaga, V. Cogez, J.M. Wieruszeski, B. Stahl, J. Lemoine, G.Lippens, J.P. Bohin, Osmoregulated periplasmic glucans of thefree-living photosynthetic bacterium Rhodobacter sphaeroides,Eur. J. Biochem. 269 (2002) 2464–2472.

[19] P.A. Trinel, E. Maes, J.P. Zanetta, F. Delplace, B. Coddeville, T.Jouault, G. Strecker, D. Poulain, Candida albicans phospholipo-mannan, a new member of the fungal mannose inositol phospho-ceramide family, J. Biol. Chem. 277 (2002) 37260–37271.

[20] J.P. Zanetta, A. Pons, M. Iwersen, C. Mariller, Y. Leroy, P. Timm-erman, R. Schauer, Diversity of sialic acids revealed using gaschromatography/mass spectrometry of heptaXuorobutyrate deriv-atives, Glycobiology 11 (2001) 663–676.

[21] J.P. Zanetta, P. Timmerman, Y. Leroy, Gas-liquid chromatogra-phy of the heptaXuorobutyrate derivatives of the O-methyl glyco-sides on capillary columns: A method for the quantitativedetermination of monosaccharide composition of glycoproteinsand glycolipids, Glycobiology 9 (1999) 255–266.

[22] T. Nakai, T. Ohta, Beta-3-Oxindolylalanine: the main intermedi-ate in tryptophan degradation occurring in acid hydrolysis of pro-tein, Biochim. Biophys. Acta 420 (1976) 258–264.

[23] T. Ohta, T. Nakai, The reaction of tryptophan with cystine duringacid hydrolysis of proteins. Formation of tryptathionine as a tran-sient intermediate in a model system, Biochim. Biophys. Acta 533(1978) 440–445.

[24] J. Krieg, W. Glasner, A.. Vicentini, M.A. Doucey, A. LoZer, D.Hess, J. Hofsteenge, C-mannosylation of human RNase 2 is anintracellular process performed by a variety of cultured cells, J.Biol. Chem. 272 (1997) 26687–26692.

[25] A. Pons, J. Popa, J. Portoukalian, J. Bodennec, D. Ardail, O. Kol,M.J. Martin-Martin, P. Hueso, P. Timmerman, Y. Leroy, J.P. Zan-etta, Single-step gaschromatography-mass spectrometry analysisof glycolipid constituents as heptaXuorobutyrate derivatives witha special reference to the lipid portion, Anal. Biochem. 284 (2000)201–216.

[26] J.P. Zanetta, P. Timmerman, Y. Leroy, Determination ofconstituents of sulphated glycosaminoglycans using a meth-anolysis procedure and gas chromatography/mass spectrometryof heptaXuorobutyrate derivatives, Glycoconjug. J. 16 (1999)617–627.