Characterization of a Mutant Recombinant S100Protein Using Electrospray Ionization MassSpectrometry

Mark J. Raftery,* Craig A. Harrison and Carolyn L. GeczyCytokine Research Unit, School of Pathology, University of New South Wales, Kensington, New South Wales, 2052,Australia

Two recombinant proteins derived by thrombin cleavage of a fusion protein between glutathione-S-transferaseand CP10 (Chemotactic protein 10 kDa) were seperated by C4 reversed-phase high-performance liquidchromatography (RP-HPLC). Both proteins were recognised by a polyclonal antibody to native CP10following sodium dodecyl sulphate/polyacryamide gel electrophoresis (SDS/PAGE) and Western blotting. Themajor form ( ~ 90%) had a mass of 10 308 Da, by electrospray mass spectrometry (ESI-MS), which comparedwell with the theoretical mass of rCP10 (10 307.6 Da) whereas the minor component ( ~ 10%) had a mass of11 333 Da, 1025 mass units greater than expected. One sequence was obtained by N-terminal sequencing,suggesting that the N-terminus was not modified. The mass of peptides isolated after Asp-N digestion and C18RP-HPLC were determined by ESI-MS and each assigned a probable sequence based on the expected peptidemap of rCP10. The mutant protein produced one additional peak at 10.0 min with mass 1639 Da and thesequence DSHKEQQRGIPGNSS by Edman degradation. The first 5 amino acids corresponded to the last 5C-terminal amino acids of rCP10. Analysis of the cDNA sequence of the expression vector used to producerCP10 indicated that the 10 additional C-terminal amino acids were translated after the insertion of glutamineat the normal TAG stop codon. Another stop codon (TGA) located 27 base pairs downstream halts translation.The calculated mass of the mutant protein is 11 332.7 Da, in good agreement with the experimental mass.Readthrough occurs in strains of E.coli (eg JPA101) with the amber mutation supE, and this allowedsubstitution of glutamine at TAG codons in ~ 5–10% of transcripts. © 1997 by John Wiley & Sons, Ltd.

Received 29 November 1996; Revised 20 January 1997; Accepted 24 January 1997Rapid Commun. Mass Spectrom. 11, 405–409 (1997)No. of Figures: 4 No. of Tables: 2 No. of Refs: 19

CP10 (Chemotactic protein 10 kDa) is a potentchemotractant for murine and human polymorphonu-clear leukocytes in vivo and in vitro1 with optimumactivity at approx 10–12

M in vitro.2,3 The amino acidsequence was determined biochemically and from thederived complementary DNA (cDNA) sequence.2,4 It iscomposed of 88 amino acids, contains no post-transla-tional modifications and is a member of the S100 Ca2+

binding protein family.5 Initial studies indicated thatCP10 was produced in small quantities by activatedmurine spleen cells and isolation from supernatants wasa lengthy and complex procedure.4 To facilitate bio-chemical and structural characterization, a relativelylarge scale source of CP10 was obtained by chemicalsynthesis6 and, as a recombinant protein, using thepGEX expression system.7

The pGEX expression system produces the desiredrecombinant protein as a fusion with glutathione-S-transferase (GST) protein, enabling isolation frombacterial lysates by affinity chromatography under non-denaturing conditions. The fusion product is cleaved ata cloned consensus site between the two proteins usingeither thrombin or factor Xa.8 The GST/CP10 plasmid(DNA chain) was produced using a CP10-plasmid inwhich the start codon was mutated to a restriction sitesuitable for Bam HI endonuclease cleavage. Afterdigestion with Bam HI, the fragment was subclonedinto the pGEX-2T expression vector and this DNA

chain introduced (transfected) into E.coli. Expressionof the desired fusion protein was induced with iso-propyl-â-D-thiogalactopyronoside and the product wasthen isolated from the E.coli lysate by affinity chroma-tography using glutathione-agarose. RecombinantCP10, isolated following thrombin cleavage and C18reversed-phase high-performance liquid chromatog-raphy (RP-HPLC), has two additional amino acids(glycine and serine) at the N-terminus and a theoreticalmass of 10 307.6 Da.7

Electrospray ionization mass spectrometry (ESI-MS)has been used to characterize many recombinantproteins, including the S100 proteins calvasculin,9 calcy-clin9 and S100A3.10 In these examples, the proteinswere expressed in E.coli as full length proteins and theidentity of each was confirmed by ESI-MS aftercomparison of the theoretical and experimental masses.No mutant or post-translationally modified forms wereidentified. Matrix-assisted laser desorption ionizationmass spectrometry (MALDI-MS) has been used as analternative to sodium dodecyl sulphate/polyacrylamide-gel electrophoresis (SDS/PAGE), to monitor the extentof factor Xa cleavage of a fusion protein between GSTand HIV-1IIIB p26.11 We also used ESI-MS to character-ize an unusual post-translational modification of themurine S100 protein MRP14.12

A detailed account of the isolation and character-ization of recombinant CP10 (rCP10) has beenreported,7 but the mutant form was not characterized.Here we describe the isolation and characterization ofmodified rCP10 which contains 10 additional C-terminal amino acids.

* Correspondence to: M. J. Raftery

RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 11, 405–409 (1997)

CCC 0951–4198/97/040405–05 $17.50 © 1997 by John Wiley & Sons, Ltd.

EXPERIMENTAL

General

Reagents and chemicals were analytical grade (Sigma,St. Louis, MO, USA; BioRad, Hercules, CA, USA) andsolvents were HPLC grade (Mallinckrodt, ClaytonSouth, Vic., Australia). SDS/PAGE/Western blottingwere performed using a Mini Protean II apparatus(BioRad) with 15% gels and a Tris/Tricene buffersystem.13 Liquid chromatographic separations wereperformed using a non-metallic LC626 or LC625 HPLCsystem (Waters, Bedford, MA, USA) and UV absor-bance monitored at 214 nm and 280 nm with a Waters996 photodiode array or 490 UV/visible detector.

Recombinant CP10: expression, isolation andcharacterization

Recombinant CP10, produced in E.coli (strainJPA10114) as a fusion protein with glutathione-S-trans-ferase,7 was isolated after cell lysis (tris-buffered saline(TBS), 1% Triton X-100) using glutathione-agarosebeads (Sigma, St. Louis, MO, USA) and cleaved withthrombin (American Diagnostica Inc, New York, USA;30 NIH units; Tris 50 mM, pH 8; EDTA, 5 mM; 60 min;37 °C). The beads were washed with TBS (2 × 2 mL)and the combined eluate (100 µL) applied to a C4 RP-HPLC column (5 µ, 46 × 250 mm, 300 Å, Vydac,Hesperia, CA, USA) and proteins eluted with a lineargradient of 35 to 65% actetonitrile (with 0.1% tri-fluoroacetic acid (TFA)) at 1 mL/min over 30 min.Peaks with major UV absorbances were collectedmanually and concentrated using a Speedvac (Savant,Farmingdale, NY, USA) to a final concentration ofapprox. 50 ng/µL. Samples (approx. 200 ng) wereanalysed by SDS/PAGE; gels were either silver stainedor blotted onto polyvinylidene difluoride (PVDF)membrane (Millipore, Bedford, MA, USA). Mem-branes were blocked (5% non-fat skim milk), incubatedwith a polyclonal antibody to native CP10, and finallydetected with a horseradish peroxidase-conjugatedgoat anti-rabit H + L chain (BioRad) by enhancedchemiluminescence (Amersham, Buckinghamshire,UK).

Mass spectrometry

Electrospray ionization (ESI) mass spectra wereacquired using a single quadrupole mass spectrometerequipped with an electrospray ionization source (Plat-form, VG-Fisons Instruments, Manchester, UK). Sam-ples ( ~ 50 pmol, 10 µL) were injected into a solventflow (10 µL/min; 50:50 water:acetonitrile, 0.05% TFA)coupled directly to the ion source via a fused-silicacapillary (50 µm × 40 cm) interface. The sourcetemperature was 50 °C and nitrogen was used as thenebulizer gas. Sample droplets were introduced at apositive potential of approx. 3 kV and transferred tothe mass analyser, which was operated with a conevoltage of 50 V. The peak width at half height was 1mass unit. Spectra were acquired in the multi-channelacquisition mode over the range m/z 700 to 1800 in 5 s,and calibrated using horse heart myoglobin (Sigma).

Asp-N digestion

Recombinant proteins (50 µg), isolated from C4 RP-HPLC, were digested in ammonium bicarbonate (250

µL, 50 mM, pH 8.0) using endoprotease Asp-N(sequencing grade, Boehringer Mannheim, Castle Hill,NSW, Australia) at an enzyme to substrate ratio ofapprox. 1:100 at 37 °C for 2 h. The pH of the digest waslowered to approx. 2 (1% TFA) and the mixtureapplied directly to a C18 RP-column (Vydac, 300 Å, 5µ, 4.6 × 250 mm). Peptides were eluted with a gradientof 5 to 75% acetonitrile (with 0.1% TFA) at 1 mL/minover 30 min. Fractions with major absorbance at 214 nmwere collected manually.

Automated N-terminal sequencing

Proteins and digest peptides (typically 250–500 pmol)were N-terminally sequenced using an Applied Bio-systems model 473 or 470A automated protein sequen-cer (Applied Biosystems, Burwood, Vic., Australia) ateither Sydney University Macromolecular AnalysisCentre (SUMAC) or at the School of Biochemistry, LaTrobe University, Bundoora, Victoria.

RESULTS AND DISCUSSION

The mutant form of rCP10 separated at a slightly higherapparent molecular weight on SDS/PAGE and as asmall, early eluting shoulder on C18 RP-HPLC.7 In thispaper we fully characterize the mutant form of rCP10and determine the likely mechanism of its productionin E.coli.

Isolation of recombinant CP10

Two forms of rCP10 were readily separated usinganalytical C4 RP-HPLC with a shallow acetonitrilegradient. The mutant form eluted at 13.6 min as adistinct peak followed by rCP10 itself at 14.1 min (Fig.1(a)). The disulphide-linked homodimer of rCP10eluted at 15.1 min followed by GST at 17.5 min (Fig.1(a)). The ratio of mutant protein to rCP10 wasapproximately 1:10 and this ratio did not vary over 10different preparations. The disulphide-linked homo-dimer, which varied in amount from batch to batch, wasprobably formed by oxidation of Cys43 during thrombincleavage and/or isolation using glutathione-agarose.Dimer formation was completely eliminated by addi-tion of dithiothreitol (DTT) (1 mM, 30 min, 37 °C)

Figure 1. (a) C4 RP-HPLC chromatogram of the thrombin cleavageproducts of rCP10 fusion protein: (b) SDS/PAGE analysis of isolatedproteins; lane 1, MW markers; lane 2, mutant rCP10; lane 3, rCP10;lane 4, CP10-homodimer; lane 5, GST (lanes 1 to 5 were silverstained); lane 6, mutant rCP10; lane 7, rCP10; lane 8, CP10-homodimer; lane 9, fusion protein (lanes 6 to 9 were detected with ananti-CP10 antibody and enhanced chemiluminescence).

406 CHARACTERIZATION OF A MUTANT RECOMBINANT PROTEIN

Table 1. Comparison of the ESI masses of proteins isolatedafter C4 RP-HPLC with the theoretical masses derived fromtheir cDNA sequences (see text for details)

Protein Calc. mass (DA) ESI mass (DA)

rCP10 10 307.6 10 308a

Mutant CP10 11 332.7 11 333a

rCP10Dimer 20 614.1 20 615a

GST 26 166.6 26 169b

Fusion protein 36 456.1 36 459b

Mutant fusion protein 37 481.2 37 484a

a A small peak at 21 640 Da was also observed, which was attributedto the rCP10-mutant rCP10 disulphide linked heterodimer (calc.mass 21 639.2 Da).b A peak 131 lower was also observed, probably due to partialremoval of the initiator Met after translation.

before C4 RP-HPLC (data not shown). S100 proteinswith free Cys residues readily form disulphide-linkedhomodimers. S100b, the disulphide-linked homodimerof S100â is a growth factor for glial cells whereas themonomer is inactive.15 GST, derived from the fusionprotein after cleavage with thrombin, was partiallywashed off the affinity column and subsequentlyisolated by C4 RP-HPLC. The yield of GST also variedin each preparation and was dependent on the age ofthe glutathione-agarose, suggesting some loss of speci-ficity of these beads with use.

SDS/PAGE followed by silver staining or by Westernblotting of the proteins isolated from C4 RP-HPLC isshown in Fig. 1(b) (lanes 2–5 and 6–9 respectively).rCP10 (lane 3, Fig. 1(b)) had an apparent mol.wt. ofapprox. 8000 whereas the mutant protein (lane 2, Fig.1(b)) migrated with a slightly higher mol.wt., confirm-ing the previous report.7 The disulphide-linked homo-dimer (lane 4, Fig. 1(b)) had an apparent mol.wt. of20 000 whereas that for GST was 26 000 (lane 5, Fig.1(b)). The Western blot of the same proteins showedthat an anti-CP10 rabbit polyclonal antibody producedusing native CP10 reacted with both forms of rCP10(lane 6 and 7, Fig. 1(b)) indicating they possesscommon antigen epitopes. The antibody also reactedwith the disulphide-linked homodimer of rCP10 (lane8, Fig. 1(b)) and the fusion protein (lane 9, Fig. 1(b)).

Mass spectrometry

The mass of each protein isolated from C4 RP-HPLCwas determined using ESI-MS (Table 1) which enableda comparison with the theoretical masses derived fromthe known cDNA sequences to be made.4,16,17 Theexperimental mass of rCP10 was 10 308 (Table 1) whichcompares well with the theoretical mass (10 307.6 Da)whereas that of the mutant protein was 11 333 Da, 1025Da greater than that of rCP10. The experimentalmasses for the rCP10 homodimer and GST were 20 615and 26 169 Da respectively which compare well withtheir theoretical masses (Table 1), confirming theiridentity. The fusion protein was isolated directly fromthe glutathione-agarose beads before thrombin cleav-age by washing the beads with glutathione. Only onepeak corresponding to the fusion protein was separatedby C4 RP-HPLC (date not shown). One major series,together with a second minor series of multiply chargedions were observed after ESI-MS of this fraction.Transformation gave the masses of 36 459 and 37 484Da in an approx. 10:1 ratio (Fig. 2). The mass of themajor form corresponded to the theoretical mass of the

fusion protein (Table 1). The mass of the minor formwas 1025 Da greater than the theoretical mass of thefusion protein, and corresponded to the difference inmass observed between the two forms of rCP10,suggesting that these two forms were derived bythrombin cleavage of two fusion proteins. This indi-cated that the mutant form of rCP10 was produced as aresult of translation rather than by aberrant cleavage ofthe fusion protein by thrombin.

N-Terminal sequencing and peptide digest

The first 10 amino acids obtained after automatedN-terminal sequencing of the two forms of rCP10 wereGSPSELEKAL, which correspond to the predictedsequence of rCP10 and indicated an unmodifiedN-terminus.

To determine the location and identity of themodification, rCP10 and the mutant form were bothtreated with endoprotease Asp-N. The peptides isolatedcovered 90% of the rCP10 sequence. These peptideswere separated by C18 RP-HPLC (Fig. 3) and the massof each determined by ESI-MS (Table 2). Both digestsgave exactly the same C18 RP-HPLC trace except foran additional peak at 10.0 min (peptide 1) in the digestof the mutant protein (Fig. 3). Table 2 shows that themass of each co-eluting peptide is identical andcorresponds to the predicted digestion pattern ofrCP10. The mass of peptide 1 was measured as 1639 Da(Table 2), which does not correspond to any theoreticalAsp-N digest product of rCP10, suggesting that this

Figure 2. MaxEnt transformed mass spectrum of the two fusionproteins isolated after C4 RP-HPLC. The spectrum was acquired overthe mass range 1200 to 2000 Da in 5 s., using a cone voltage of 75 V.The peak width at half height was 1.5 u. The two forms of the fusionprotein had masses 36 459 and 37 484 ± 5 Da (see text for details).

Figure 3. Comparison of the C18 RP-HPLC chromatograms of theAsp-N digestion of rCP10 (Digest A) and mutant rCP10 (Digest B).All peptides were present in both digests except for one additionalpeptide (labelled 1) present in the digest of the mutant protein.

CHARACTERIZATION OF A MUTANT RECOMBINANT PROTEIN 407

Table 2. ESI masses of Asp-N peptides isolated by C18 RP-HPLC after digestion of rCP10 and mutant rCP10, togetherwith their theoretical masses (see text for details)

Peptide Fragment Calc. mass (Da) ESI mass (Da)rCP10/mutant rCP10

1 86–90/86–110 614.6/1639.7 not isolated/16392 15–33 2287.5 2288/22883 1–14 1457.7 1457/14574 34–59 3157.7 3158/31585 64–85 2432.8 2433/2433

peptide contained the modification. Automated N-ter-minal sequence analysis of peptide 1 indicated thesequence DSHKEQQRGIPGNSS. The calculated massfor this sequence is 1639.7 Da which compares well withthe experimental mass of 1639 Da (Table 2). The first 5amino acids (DSHKE) of this sequence are identical tothe last 5 C-terminal amino acids of rCP10. Theunmodified peptide (DSHKE) from the rCP10 digestwas not isolated after C18 RP-HPLC because thispeptide is very hydrophilic and was not retained on thecolumn. Thus the mutant form of rCP10 contains 100amino acids and has the same sequence as rCP10but also has 10 additional amino acids(QQRGIPGNSS) at the C-terminus. The calculatedmass of rCP10 incorporating the additional C-terminalamino acids is 11 332.7 Da, in good agreement with theexperimental mass (11 333 Da).

These data indicate that two forms of fusion proteinwere translated during protein synthesis, i.e. either twodifferent messenger RNAs (mRNAs) were translatedinto two different fusion proteins or a single mRNAwas translated into two distinct fusion proteins. It isunlikely that two fusion protein mRNAs were pro-duced because one plasmid was transfected into E.coliand the E.coli used to make the recombinant proteinwas derived from a single colony. Spontaneous muta-tion of the plasmid may have occurred but this cannotaccount for the constant ratio of mutant protein torCP10, which persisted over 10 preparations, suggestingthat the most likely source of mutant rCP10 occurred asa consequence of errant translation of a single fusionprotein mRNA.

Recombinant CP10-plasmid cDNA sequence

The nucleotide sequence of the plasmid used toproduce the recombinant proteins (plasmid pCP10-12),was verified using the chain termination method ofDNA sequencing.7 Figure 4(a) shows a partial nucleo-tide sequence derived from the expression vector usedto transfect E.coli. Two stop codons, TAG and TGA, arein frame. If translation proceeded normally it wouldstop at the first stop codon (TAG) to yield the last 9amino acids located at the C-terminus of rCP10 (Fig.

4(b)). If translation proceeded through the TAG codon,and a glutamine inserted as a consequence, then amutant protein identical to the one described here,would be formed. Translation would continue until thesecond stop codon (TGA), located 27 base pairsdownstream from the TAG codon, yielding fusionprotein (and subsequently mutant rCP10) with 10additional amino acids at the C-terminus (Fig. 4(c)).Based on the ratio of the two forms of rCP10 isolatedby C4 RP-HPLC, the mutation occurs in approx. 10%of transcripts (Fig. 1(a)).

E.coli can contain a number of suppressor mutationswhich allow protein synthesis termination (nonsense)codons to code for amino acids.18 In strains of E.coliwith suppressor mutations in transfer-RNA genes, thethree termination codons, TAG, TAA and TGA, caneach encode an amino acid. If E.coli has the suppressormutation supE, the normal stop codon TAG codes forglutamine in 5–10% of transcripts.18 Incorporation ofglutamine at a TAG stop codon is caused by a mutationin a glutamine transfer-RNA gene, i.e. the anticodon,which base-pairs with the codon on the mRNA, ismutated from CTG to CTA (G to A), allowing theusual chain termination codon TAG to code forglutamine.18 Suppressor genes are generally producedfrom transfer-RNA genes that are redundant, resultingin only partial formation of mutant proteins andbecause of this redundancy, suppressor mutations aregenerally not lethal. This mutation occurs in a numberof strains of E.coli used as hosts for expression vectorsfor production of recombinant proteins. The JPA101strain used to express rCP10 was derived from E.coliJM109 which has the supE44 gene.14 EukaryoticmRNA’s use the three termination codons with approx-imately equal preference, whereas bacteria (prokar-yotes) use the TAG stop codon 25 times less often thanTAA.19 Therefore, strains of E.coli with the supEgenotype are unsuitable for production of recombinantproteins from cDNA’s which use the TAG stopcodon.

In conclusion, we have identified and characterized amutant form of rCP10 derived from a mutant fusionprotein formed as a consequence of a glutamineinsertion at the normal stop codon, allowing transcrip-tion to proceed to the second stop codon 27 base pairsdownstream. This occurred in approx 10% of tran-scripts and was most likely due to the amber mutationsupE in the strain of E.coli used.

Acknowledgement

This work was supported in part by grants from the National Healthand Medical Research Council of Australia. Members of theCytokine Research Unit and Immunology group at The HeartResearch Institute, Missenden Rd, Camperdown, NSW are acknowl-edged for rCP10 preparations and helpful discussions.

Figure 4. Partial nucleotide sequence and derived C-terminal region of the proteins from the expression vector used to produce rCP10. Theexpression vector contains the cDNA sequence of CP10 and the pGEX-2T fusion vector gene. The nucleotide sequence contains two stop codons,TAG and TGA (A); The first stop codon is TAG and produces rCP10 (B). If this codon translates to a glutamine (because of the supE mutation)then the mutant protein, with 10 additional C-terminal amino acids would be expected (C).

408 CHARACTERIZATION OF A MUTANT RECOMBINANT PROTEIN

REFERENCES

1. J. M. Devery, N. J. King and C. L. Geczy, J. Immunol. 152,1888–1897 (1994).

2. M. Lackmann, C. J. Cornish, R. J. Simpson, R. L. Moritz and C.L. Geczy, J. Biol. Chem. 267, 7499–7504 (1992).

3. C. L. Geczy, Biochem. Biophys. Acta 1313, 246–253 (1996).4. M. Lackmann, P. Rajasekariah, S. E. Iismaa, C. J. Cornish, R. J.

Simpson, G. Reid, R. L. Moritz and C. L. Geczy, J. Immunol. 150,2981–2991 (1992).

5. G. Fano, S. Biocca, S. Fulle, M. A. Mariggio and S. Belia, Prog.Neurobiol. 46, 71–82 (1995).

6. P. F. Alewood, personal communication.7. S. E. Iismaa, S. Hu, M. Kocher, M. Lackmann, C. A. Harrison, S.

Thliveris and C. L. Geczy, DNA Cell Biol. 13, 183–192, (1994).8. D. B. Smith and K. S. Johnson, Gene 67, 31–40, (1988).9. M. Pedrocchi, B. W. Schafer, I. Durussel, J. A. Cox and C. W.

Heizmann, Biochemistry 33, 6732–8 (1994).10. U. G. Fohr, C. W. Heizmann, D. Engelkamp, B. W. Schafer and J.

A. Cox, J. Biol. Chem. 270, 21056–21061 (1995).

11. C. E. Parker, D. I. Papac and K. B. Tomer, Anal. Biochem. 239,25–34 (1996).

12. M. J. Raftery, C. A. Harrison, P. Alewood, A. Jones and C. L.Geczy, Biochem. J. 316, 285–293 (1996).

13. H. Schagger and G. von Jagow, Anal. Biochem. 166, 368–379(1987).

14. S. E. Iismaa, P. M. Ealing, K. F. Scott and J. M. Watson, Mol.Microbiol. 3, 1753–1764 (1989).

15. D. Kligman and D. R. Marshak, Proc. Natl. Acad. Sci. USA 82,7136–7139 (1985).

16. D. B. Smith, K. M. Davern, P. G. Board, W. U. Tiu, E. G. Garciaand G. F. Mitchell, Proc. Natl. Acad. Sci. USA 83, 8703–8707(1986).

17. GenBank, Accession number U13850, (1994).18. G. Eggertsson and D. Soll, Microbiol. Rev. 52, 354–374 (1988).19. J. Kohli and H. Grosjean, Mol. Gen. Genet. 182, 430–439

(1981).

CHARACTERIZATION OF A MUTANT RECOMBINANT PROTEIN 409