covalent cross-linking of proteins without chemical reagents

Covalent cross-linking of proteins withoutchemical reagents

BRIGITTE L. SIMONS,1,2 MARY C. KING,2 TERRY CYR,1 MARY ALICE HEFFORD,1

AND HARVEY KAPLAN2

1Centre for Biologics Research, Biologics and Genetic Therapies Directorate, Health Canada,Ottawa, Canada K1A 0L22Department of Chemistry University of Ottawa, Ottawa, Ontario K1N 6N5

(RECEIVED November 1, 2001; FINAL REVISION March 21, 2002; ACCEPTED March 21, 2002)

Abstract

A facile method for the formation of zero-length covalent cross-links between protein molecules in thelyophilized state without the use of chemical reagents has been developed. The cross-linking process isperformed by simply sealing lyophilized protein under vacuum in a glass vessel and heating at 85°C for 24h. Under these conditions, approximately one-third of the total protein present becomes cross-linked, anddimer is the major product. Chemical and mass spectroscopic evidence obtained shows that zero-lengthcross-links are formed as a result of the condensation of interacting ammonium and carboxylate groups toform amide bonds between adjacent molecules. For the protein examined in the most detail, RNase A, thecross-linked dimer has only one amide cross-link and retains the enzymatic activity of the monomer. Thein vacuo cross-linking procedure appears to be general in its applicability because five different proteinstested gave substantial cross-linking, and co-lyophilization of lysozyme and RNase A also gave a heterog-eneous covalently cross-linked dimer.

Keywords: Protein cross-linking; lyophilized protein; zero-length cross-linking; covalent cross-linking; invacuo modification

Intermolecular covalent cross-linking of functionalgroups in proteins has proved to be a very useful approachin the study of structure–function relationships in proteins,especially in multiprotein complexes (Fancy 2000; Phizickyand Fields 1995; Lundblad 1994). This technique also hasmany practical applications, particularly in improving thestability of the quaternary structure of proteins. The subunitsof hemoglobin, for example, have been cross-linked to re-inforce active structure and produce a more stable oxygentransporter (Manning et al. 1991; Chang 1998; White andOlsen 1987; Jones et al. 1996). Reaction with homobifunc-tional and heterobifunctional reagents with variable spacingbetween the reactive groups has been the most widely used

strategy to achieve protein cross-links (Lundblad 1994). Inparticular, glutaraldehyde and bis-imido esters have beenwidely used because of their ability to form stable deriva-tives with the amino groups that are present in relativelyhigh abundance in most proteins.

Zero-length cross-linking, in which peptide chains arecovalently linked through existing functional groups with-out the incorporation of a spacer group, has been exten-sively used to immobilize proteins on non-protein matrices.This is usually accomplished by activation of carboxylgroups on a protein with a water soluble carbodiimide fol-lowed by coupling with an amino group on the support toform a stable amide bond (Lundblad 1994). Such cross-links between interacting proteins have been obtained byreaction with carbodiimide in the multi-protein electrontransport system (Mauk and Mauk 1989), but there are veryfew other reported cases in which such zero-length cross-links have been achieved. Zero-length cross-linking of thistype is most likely to occur when carboxyl groups are situ-

Reprint requests to: Harvey Kaplan, Department of Chemistry, Univer-sity of Ottawa, 10 Marie Curie, Ottawa, Ontario K1N 6N5; e-mail:[email protected]; fax: (613) 562-5180.

Article and publication are at http://www.proteinscience.org/cgi/doi/10.1110/ps.4390102.

Protein Science (2002), 11:1558–1564. Published by Cold Spring Harbor Laboratory Press. Copyright © 2002 The Protein Society1558

ated in proximity to amino groups, perhaps in salt linkages.Zero-length cross-linking can therefore provide informationabout local protein–protein interactions and has the advan-tage that no foreign group is introduced into the protein.

Ionic interactions between an amino and a carboxyl groupoften occur in protein–protein interactions (Milligan 1996).The present investigation was undertaken to determinewhether advantage could be taken of this interaction to pro-mote the formation of zero-length amide cross-links. Therationale for undertaking this study was derived from theobservation that protonated amino groups in lyophilizedproteins reacted to a significant extent with iodomethane invacuo to form trimethylated amino derivatives (Vakos et al.2001). The most likely explanation for such an apparentlyunfavorable reaction is that it is driven by the removal ofone of the products, namely, hydrogen iodide, undervacuum. The condensation of ammonium and carboxylategroups to form an amide bond is thermodynamically unfa-vorable. However, if the reaction takes place in vacuo andwater is removed, the subsequent formation of an amidebond should become more favorable. In addition, heatingthe lyophilized protein under vacuum should facilitate theremoval of water and increase the rate of the condensationreaction, thereby promoting intermolecular cross-linking. Inthe present communication, we describe the results of sucha study in which proteins, lyophilized under various condi-tions, are simply placed in a glass container under vacuumand heated in an oven.

Results

Figure 1 shows the extent of cross-linking achieved whenRNase A, lyophilized at pH 7, is cycled through successive

24-h heating periods at 85°C under vacuum, redissolved,then relyophilized for a total of four heating periods. Astrong band with an apparent molecular mass of approxi-mately 28 kD, expected for the dimer, becomes evident afteronly one heating period of 24 h in vacuo and intensifies asthe heating time increases to 96 hr. Weaker bands that cor-respond to the expected masses of RNase A trimers (∼42kD) and tetramers (∼56 kD) are also clearly visible. Unfor-tunately, in this particular gel the lane with the molecularweight markers has run anomalously as a result of an edgeeffect and the expected masses due from the dimer andtrimers do not correspond exactly to the masses indicated bythese markers. Samples that had cycled through relyophili-zation before each heating trial show no increase in dimerformation compared with those kept continuously heatingunder vacuum for an equal length of time (Fig. 1, lane 7).Maximum dimer formation was obtained after 96 h of heat-ing in vacuo. The SDS-PAGE protein loading was high (20�g), and even in the untreated sample (Fig. 1, lane 2) asmall amount of dimeric RNase is visible, probably formedduring the initial lyophilization of the sample. A band witha slightly higher apparent molecular mass than monomericRNase A is also evident and is attributable to an RNase Bimpurity present in the preparation (as per product descrip-tion, Sigma-Aldrich, product number R4875).

Very similar results were obtained with lysozyme (Fig. 2)where, again, the dimer is the major product with a smalleramount of higher oligomer. As in the case of RNase A, asmall amount of dimeric lysozyme is visible in the untreatedsample (Fig. 2, lane 2). Lyophilized bovine serum albumin,hemoglobin, and chymotrypsin treated under the same con-ditions showed approximately the same extent of dimer for-mation (data not shown).

Fig. 1. SDS-PAGE of successive cycles of solubilization of 10 mg ofRNase A at pH 7.0, lyophilization, then heating in vacuo for 24 hr. Totalprotein load per lane is 20 �g. Lane 1: low range molecular weight marker;lane 2: lyophilized RNase A, no heating in vacuo; lane 3: Cycle 1; lane 4:Cycle 2; lane 5: Cycle 3; lane 6: Cycle 4; lane 7: lyophilized RNase Aheated continuously 96 h in vacuo with only one reconstitution.

Fig. 2. SDS-PAGE of successive cycles of solubilization of 10 mg oflysozyme at pH 7.0, lyophilization, then heating in vacuo for 24 hr. Totalprotein load per lane is 20 �g. Lane 1: low range molecular weight marker;lane 2: lyophilized lysozyme, no heating in vacuo; lane 3: Cycle 1; lane 4:Cycle 2; lane 5: Cycle 3; lane 6: Cycle 4; lane 7: lyophilized lysozymeheated continuously 96 h in vacuo with only one reconstitution.

In vacuo cross-linking of proteins

www.proteinscience.org 1559

The amount of RNase A dimer (Fig. 1) was estimated bythe pixel-counting application ImageQuaNT 5.1 and sizeexclusion chromatography (SEC) (Fig. 3). RNase A dimeryields varied from 20% to 30% of the total treated proteindepending on the length of the heating period. The elutionprofile of the RNase A heated in vacuo for 96 h (Fig. 3B)has peaks with molecular masses of 28,000 Da, and 14,000Da, corresponding to the RNase A dimer and monomer,respectively. From the total area under the peaks, theamount of dimer present was found to be approximately30% of the total protein, in agreement with the estimate bypixel counting of the gel photographs. This isolated dimerwas shown by an RNase in-gel activity assay to retain theactivity of the monomer.

The effect of pH on the extent of cross-linking in vacuowas determined by preparing RNase A solutions with pHvalues varying from 3.0 to 10.0, then lyophilizing and heat-ing under vacuum. It was found that neutral to slightlyalkaline pH values, that is, pH 7.0–9.0, favor the formationof dimer (Fig. 4).

The effect of cations present as counter-ions on the extentof cross-linking in vacuo was determined by addition ofexcess LiCl or CsCl followed by dialysis. The pH of thesamples before lyophilization was adjusted with LiOH orCsOH. It was found that these counter-ions had no effectwhatsoever on the extent of dimerization (data not shown).

The effect of lyophilization of RNase A in the presence ofan excipient, trehalose, is shown in Figure 5. As the amountof trehalose present in the lyophilized sample increases, theamount of RNase A dimer produced decreases. At a 1 : 1w/w ratio, trehalose appears to prevent any dimer formation,because only a trace of dimer similar to that observed inuntreated samples is present.

The results obtained when equal amounts (w/w) of RNaseA and lysozyme were co-lyophilized are shown in Figure 6.Three bands are visible corresponding to the cross-linkeddimeric RNase, cross-linked dimeric lysozyme, and the het-erogeneously cross-linked lysozyme/RNase product. Ahigher extent of heterogeneous cross-linking than thatshown could be obtained using longer heating times; how-ever, the three cross-linked products could not be visualizedas distinct bands on SDS-PAGE because of the high inten-sity of the resulting bands. RNase A was also co-lyophilized

Fig. 3. Size-exclusion fast performance liquid chromatography (FPLC) ofin vacuo cross-linked RNase A. Separation of RNase A cross-linked prod-ucts achieved with two Superdex G75 HR 10/30 columns in tandem usinga mobile phase of 0.2 M Na2HPO4 and 0.15 M NaCl at pH 6.55. (A) RNaseA lyophilized at pH 7.0, no cross-linking; (B) RNase A lyophilized at pH7.0, then cross-linked in vacuo.

Fig. 4. The effect of pH on the extent of RNase A dimerization. RNase Asamples of 10 mg/mL were adjusted to pH values 3.0 to 10.0 with 1.0 NNaOH, lyophilized, then cross-linked under vacuum. A 10-�g sample ofthe treated protein was subjected to SDS-PAGE. Lane 1: RNase A at pH3; lane 2: RNase A at pH 4; lane 3: RNase A at pH 5; lane 4: RNase A atpH 6; lane 5: RNase A at pH 7; lane 6: RNase A at pH 8; lane 7: RNaseA at pH 9; lane 8: RNase A at pH 10.

Fig. 5. Cross-linking of RNase A in the presence of trehalose at pH 7.0.RNase A (10 mg) was co-lyophilized with different amounts of trehaloseand heated under vacuum for 96 h. A 20-�g sample of the treated proteinwas subjected to SDS-PAGE. Lane 1: RNase A alone cross-linked invacuo; lane 2: RNase A and trehalose in a 5 : 1 (w/w) ratio, cross-linked invacuo; lane 3: RNase A and trehalose in a 1 : 1 (w/w) ratio, cross-linked invacuo; lane 4: RNase A and trehalose in a 1 : 5 (w/w) ratio, cross-linked invacuo.

Simons et al.

1560 Protein Science, vol. 11

with high molecular weight polylysine or polyglutamic acid(1 : 1 w/w ) at pH 7.0. It was found that the high molecularweight fractions collected from SEC contained no free mo-nomeric or dimeric RNase and had a high RNase activity inan RNase in-gel assay (data not shown).

Two chemically modified RNase A samples were pre-pared, one in which the amino groups were dimethylated byreductive methylation and the other in which the carboxylgroups were modified by amidation with carbodiimide andglycinamide. The modified proteins were lyophilized andheated in vacuo under the same conditions as the unmodi-fied protein. In both cases, no dimerization was observed(Fig. 7).

The RNase A dimer was isolated by SEC and subjected toelectrospray mass spectrometry (Fig. 8). The major peak in

the spectrum occurs at 27,345 mass units corresponding totwice the mass of the monomer (13,682 ± 1 mass units)minus 18 mass units, that is, the loss of one water molecule,which shows that only one amide cross-link is present in thedimer. There is also a trace amount of a dimer peak at27,327 mass units resulting from the loss of two water mol-ecules and the formation of two amide cross-links. There isanother significant peak at 27,361 mass units and a minorpeak at 27,377 mass units, 16 and 32 mass units above themajor dimer peak, respectively. The mass spectrum ofRNase monomer used to prepare the dimer also shows acomponent 16 mass units greater than the monomer, whichis probably a result of the presence of some RNase withmethionine sulfoxide in the preparation. These two peaks,therefore, are probably a result of covalent dimers formedby condensations of molecules of RNase containing a resi-due of methionine sulfoxide.

Discussion

Our hypothesis is that the covalent cross-linking observedby the in vacuo procedure described occurs by formation ofamide linkages that result from the condensation of a pro-tonated amino group with a negatively charged carboxyl,and this condensation reaction is driven by the removal ofwater. When either the amino groups or the carboxyl groupsare modified, by reductive methylation or amidation reac-tions, respectively, in vacuo cross-linking of the lyophilizedprotein does not occur. This result is consistent with theinvolvement of amino and carboxyl functions in the cross-linking. The observation that cross-linking is optimal whenproteins were lyophilized at neutral to slightly alkaline pHvalues, that is, pH 7 to 9.0, in which amino and carboxylgroups are expected to be in their positively and negativelycharged states, respectively, is also consistent with our hy-pothesis. The finding that RNase A could be covalentlycross-linked to polylysine and polyglutamic acid polymersby the in vacuo procedure provides further evidence that thecross-linking is occurring by the formation of amide bondsbetween interacting ammonium and carboxylate groups.

Lyophilization of proteins removes most of the solvationshell separating individual molecules in solution, which re-sults in direct interaction of the protein molecules. Excipi-ents such as trehalose have often been added to proteinsolutions before lyophilization to replace this solvent shelland thereby stabilize the protein (Zaks 1992). In the case ofRNase A, it was found that as the trehalose/RNase A ratioin the lyophilization mixture increases, the amount of dimerformation decreases. This result indicates that direct inter-action of the protein molecules is required for in vacuocross-linking and is also consistent with the formation ofamide bonds between interacting ammonium and carboxyl-ate groups.

Fig. 6. Heterogeneous cross-linking of RNase A and lysozyme. RNase A(10 mg) was co-lyophilized with lysozyme (10 mg) at pH 7.0, heated undervacuum for 48 hr, and 15 �g of the treated protein was subjected toSDS-PAGE. Lane 1: RNase A (pH 7.0) alone cross-linked in vacuo, 48 h;lane 2: lysozyme (pH 7.0) alone, cross-linked in vacuo, 48 h; lane 3: RNaseA (pH 7.0) and lysozyme (pH 7.0) co-lyophilized and cross-linked invacuo, 48 h. The lysozyme-RNase A dimer is indicated by the arrow.

Fig. 7. SDS-PAGE showing the effect of chemical modification of RNaseA (15 mg/mL) on in vacuo cross-linking. Total protein loaded per lane is10 �g. (A) Lane 1: RNase A lyophilized at pH 9.0 with no in vacuotreatment; lane 2: RNase A lyophilized at pH 9.0 and heated in vacuo, 48h; lane 3: reductively methylated RNase A lyophilized at pH 9.0, andheated in vacuo, 48 h. (B) Lane 1: RNase A with amidated carboxyl groupslyophilized at pH 7.0 and heated in vacuo, 48 h.



The evidence obtained from the conditions required forsuccessful in vacuo cross-linking and from the effects ofchemical modification provide strong support for an amidecross-link. Nevertheless, this evidence is not definitive con-cerning the nature of the chemical cross-link. Mass spec-troscopic analysis of the RNase A dimer from the in vacuoprocedure shows that the covalent dimerization results froma condensation reaction with the concomitant loss of a watermolecule. Given the evidence for the involvement of car-boxyl groups, there are only three possibilities as to howsuch cross-linking can occur with the loss of a water mol-ecule: (1) condensation of two protonated carboxyl groupsto form an anhydride; (2) condensation of a protonated car-boxyl with a hydroxyl group of a side chain to form an ester;or (3) condensation of a deprotonated carboxyl group witha protonated amino group to form an amide. For proteinslyophilized at neutral or slightly alkaline pH values, inwhich optimal in vacuo cross-linking occurs, the carboxylgroups are expected to be deprotonated and could not takepart in anhydride or ester formation with the loss of a watermolecule. Furthermore, covalent dimers formed by anhy-drides or ester linkages would not survive the conditions forpreparing and running SDS gels, namely, high temperaturesin the presence of mercaptoethanol at an alkaline pH value.Therefore, the result of the mass spectroscopic analysis ofthe RNase A dimer combined with all the other evidenceleaves no conclusion other than that the in vacuo dimeriza-tion occurs via amide bond formation. From the mass of thedimer, it can also be concluded that there are no other modi-fications that occur to the protein.

The yield of dimer obtained with RNase was approxi-mately 30%. A possible explanation for this apparent limitis the presence of counter-ions, especially Na+, in associa-

tion with carboxyl groups. This association could obstructthe salt-bridge interactions between protein molecules andreduce cross-linking efficiency. However, the presence ofsmall or large cations, such as Li+ or Cs+, had no effect onthe dimer yield. Thus, although the reason that dimer pro-duction in these experiments was limited to approximately30% of the total protein remains obscure, the presence ofcations associated with the lyophilized protein does not ap-pear to be a factor.

It is expected that in vacuo cross-linked proteins willretain their biological activity when returned to an aqueousenvironment because this is the case for most lyophilizedproteins. The RNase A dimer is active and RNase A cross-linked to high molecular weight polylysine is also active.Natural, non-covalent dimers such as the RNase dimer (Parkand Raines 2000) often have modulated activity. It is ex-pected that the zero-length cross-linking will also modulatethe biological activity and properties of the protein. How-ever, these specific effects will have to be investigated indetail for each protein or combination of proteins that arecross-linked.

Zero-length cross-linking of proteins has rarely beenachieved. The in vacuo cross-linking procedure provides asimple and convenient procedure for achieving such cross-linking. The result obtained with RNase A shows that thecovalent dimer is formed by only one amide cross-link.However, it has yet to be determined whether a single ho-mogeneous cross-link predominates or whether there aremultiple cross-linking sites formed between different car-boxylate and amino groups. We believe the former is themost likely possibility; if this is the case, proteolysis of thecovalent dimer should yield only one major cross-linkedpeptide and the latter will yield multiple cross-linked pep-

Fig. 8. Deconvoluted electrospray mass spectrum of the RNase A dimer produced by in vacuo cross-linking.

Simons et al.


tides. Further investigations with proteins that have a vari-ety of physical and structural properties are required to de-termine the generality of the results reported in this com-munication. The number and location of zero-length cross-links observed in a particular protein and their effects on itsactivity should provide novel information on structure–function relationships. In other cases, zero-length cross-linking of a protein or proteins may confer advantageousproperties that are useful for applications in medicine andindustry.

Materials and methods

Bovine pancreatic RNase A (Type I-A), lysozyme, poly-D-lysine,poly-D-glutamic acid, and D(+)-trehalose were purchased fromSigma-Aldrich. All other chemicals, reagents, and solvents werehigh purity preparations obtained from reliable commercialsources.

In vacuo cross-linking procedure

Lyophilized protein was obtained from the supplier, reconstitutedin distilled water to a concentration of 10 mg/mL, and the pH ofthe solution was adjusted to 7.0 with 1 N NaOH. The proteinsolution was placed in a glass tube and lyophilized. These glasstubes were sealed under a vacuum of approximately 33 × 10−3

mBar and then placed in an oven at 85°C for 24 h. The vacuumwas released and the protein sample reconstituted with 0.2 MNa2HPO4 and 0.15 M NaCl at pH 6.55 to give a final proteinconcentration of 10 mg/mL.

In some cases, the protein was reconstituted in distilled water(dH2O), instead of buffer, to a concentration of 10 mg/mL, analiquot was removed, and then the solution was relyophilized andheated again under vacuum at 85°C for an additional 24 h. Afterfour successive cycles of lyophilization, heating, and reconstitu-tion, the final lyophilized protein sample was reconstituted with0.2 M Na2HPO4 and 0.15 M NaCl at pH 6.55 to give a final proteinconcentration of 10 mg/mL.

The effect of pH, counter ions, or excipients

Protein solutions (10 mg/mL) at pH values varying from 3.0 to10.0 were prepared by the addition of 1 N NaOH or 1 N HCl witha micro-syringe, as required. The protein solutions were lyophi-lized and subjected to the in vacuo cross-linking procedure.

Protein solutions (10 mg/mL) were also prepared in the presenceof different cations by the addition of excess LiCl, NaCl, or CsClfollowed by dialysis against distilled water. Samples were treatedas described above except that the pH was adjusted to 7.0 with 1N LiOH, 1 N NaOH, or 1 N CsOH, as appropriate.

A solution of RNase A (10 mg/mL at pH 7.0) was lyophilizedin the presence of D-trehalose at w/w ratios of protein/trehalose of5 : 1, 1 : 1, and 1 : 5 and then subjected to the in vacuo cross-linking procedure. On completion, the excess trehalose was re-moved by dialysis.

Heterogeneous cross-linking in vacuo

A solution containing RNase A and lysozyme in equal amounts(10 mg/mL) was prepared and the pH was adjusted to 7.0 with 1

N NaOH before lyophilization. The in vacuo procedure was per-formed on the mixture of these two proteins as previously de-scribed.

Poly-D-lysine (Mr ∼340,000) or poly-D-glutamate (Mr ∼32,000)was mixed with RNase A in solution in a 5 : 1 w/w (protein/polymer) ratio. After adjusting the pH to 7.0, the mixture waslyophilized and subjected to the in vacuo cross-linking procedure.

Detection and quantification of cross-linked protein

Cross-linked products were detected by SDS-PAGE with the Bio-Rad Mini-Protean II electrophoresis system. Protein (5–20 �g)was loaded onto a 16.5% Tricine SDS-polyacrylamide gel. Afterelectrophoresis at 130 V for 90 min, the gel was stained withCoomassie Brilliant Blue G250. The relative amount of proteinpresent in each band was determined using the pixel-countingapplication in ImageQuaNT 5.1 (Molecular Dynamics).

SEC was performed using two Superdex G75 HR 10/30 col-umns (Amersham-Pharmacia) attached, in tandem, to a Pharmaciafast performance liquid chromatography (FPLC) system with de-tection at 210 nm. Mobile phase (0.2 M Na2HPO4 and 0.15 MNaCl at pH 6.55 at 4°C) was used at a flow rate of 0.05 mL/min.In general, 0.5 mL fractions were collected. Molecular weightstandards (phosphorylase b, 94 kDa; bovine serum albumin, 67kDa; ovalbumin, 43 kDa; bovine erythrocytes carbonic anhydrase,29 kDa; trypsin inhibitor, 20.1 kDa; �-lactalbumin, 14.4 kDa) usedin column calibration were purchased from Amersham PharmaciaBiotech. Pooled fractions containing RNase A dimer were concen-trated to 0.5 mL with Centriprep (Amicon) 3000 molecular weightcut-off concentrators.

RNase A in-gel activity assay

Cross-linked RNase A products were tested for catalytic activitywith an RNA agarose gel-based assay (Leland et al. 1998; Gaur etal. 2001). Cross-linked RNase A products (2 ng) were incubatedwith 5�g of total rat liver RNA in 100 mM Tris-HCl, pH 7.5,containing 10 mM DTT in a total reaction volume of 10 �L.Reaction was allowed to proceed for 10 min at 37°C and wasstopped by the addition of 1 �L of diethyl pyrocarbonate andfollowed by incubation on ice for 2 min. Samples were supple-mented with 2 �L of RNA gel loading buffer (10 mM Tris-HCl,pH 7.5, 50 mM EDTA), glycerol (30% v/v), xylene cyanol FF(0.25% w/v), and bromophenol blue (0.25% w/v) before loadingonto 1.5% agarose gel containing 2% formaldehyde and 0.05 Methidium bromide.

Chemical modification of proteins

Dimethylation of amino groups was performed on RNase A (15mg) according to the procedure described by Means and Feeney(1971). Excess reagent was removed by dialysis.

Amidation of carboxyl groups was performed on a solution ofRNase A (15 mg/mL in a total reaction volume of 1 mL) in 1.33M glycinamide at pH 4.75 with activation by carbodiimide, asdescribed by Means and Feeney (1971). On completion, excessreagent and by-products were removed by dialysis.

Mass spectrometric analysis

The nanospray mass spectrum was obtained with a MicromassQ-Tof mass spectrometer. The mass spectroscopic data was de-



convoluted by use of MaxEnt1 software (Micromass Ltd.) to pro-vide the singly charged average masses. The RNase sample waspurified by standard ZipTip (Millipore) methodology. Analyteswere eluted with 75% methanol, 25% water, and 0.2% formic acid;the sample was centrifuged (6000 rpm for 2 min), and then 2 �Lwas loaded into a gold coated nanospray needle (New ObjectivesPicotip). The key variable mass spectroscopic voltages includedthe following: capillary (+950 V), cone (+47 V), and RF lens 1.05;the source temperature was 80°C, and the data for each scan werecollected for 5 sec over the range of 400 to 2500 Da by use of anNaTFA solution for external calibration.

Acknowledgments

This research was supported in part by a grant from the NaturalScience and Engineering Research Council of Canada (Grant No.OGP0090550) and, in part, by Health Canada. The in vacuo cross-linking procedure is the subject of Canadian and U.S. patent ap-plications.

The publication costs of this article were defrayed in part bypayment of page charges. This article must therefore be herebymarked “advertisement” in accordance with 18 USC section 1734solely to indicate this fact.

References

Chang, T.M. 1998. Modified hemoglobin blood substitutes: Present status andfuture perspectives. Biotechnol. Annu. Rev. 4: 75–112.

Fancy, D.A. 2000. Elucidation of protein–protein interactions using chemicalcross-linking or label transfer techniques. Curr. Opin. Chem. Biol. 4: 28–33.

Gaur, D., Swaminathan, S., and Batra, J.K. 2001. Interaction of human pancre-

atic ribonuclease with human ribonuclease inhibitor. Generation of inhibi-tor-resistant cytotoxic variants. J. Biol. Chem. 276: 24978–24984.

Jones, R.T., Shih, D.T., Fujita, T.S., Song, Y., Xiao, H., Head, C., and Kluger,R. 1996. A doubly cross-linked human hemoglobin. Effects of cross-linksbetween different subunits. J. Biol. Chem. 271: 675–680.

Leland, P.A., Schultz, L.W., Kim, B.M., and Raines, R.T. 1998. RibonucleaseA variants with potent cytotoxic activity. Proc. Natl. Acad. Sci. 95: 10407–10412.

Lundblad, R. 1994. Cross-linking of proteins. In Techniques in protein modi-fication. CRC Press. pp. 249–252. Boca Raton, FL.

Manning, L.R., Morgan, S., Beavis, R.C., Chait, B.T., Manning, J.M., Hess,J.R., Cross, M., Currell, D.L., Marini, M.A., and Winslow, R.M. 1991.Preparation, properties, and plasma retention of human hemoglobin deriva-tives: Comparison of uncrosslinked carboxymethylated hemoglobin withcrosslinked tetrameric hemoglobin. Proc. Natl. Acad. Sci. 88: 3329–3333.

Mauk, M.R. and Mauk, A.G. 1989. Crosslinking of cytochrome c and cyto-chrome b5 with a water-soluble carbodiimide. Reaction conditions, productanalysis and critique of the technique. Eur. J. Biochem. 186: 473–486.

Means, G.E. and Feeney, R.E. 1971. Akylating and similar reagents. In Chemi-cal modification of proteins. Holden-Day Press. pp. 130–132, 144–148,216–217, 222–223. San Francisco, CA.

Milligan, R.A. 1996. Protein–protein interactions in the rigor actomyosin com-plex. Proc. Natl. Acad. Sci. USA 93: 21–26.

Park, C.F. and Raines, R.T. 2000. Dimer formation by a “monomeric” protein.Protein Sci. 9: 2026–2033.

Phizicky, E.M. and Fields, S. 1995. Protein–protein interactions: Methods fordetection and analysis. Microbiol. Rev. 59: 94–123.

Vakos, H.T., Black, B., Dawson, B., Hefford, M.A., and Kaplan, H. 2001. Invacuo esterification of carboxyl groups in lyophilized proteins. J. ProteinChem. 20: 521–531.

White, F.L. and Olsen, K.W. 1987. Effects of crosslinking on the thermalstability of hemoglobin. I. The use of bis(3,5-dibromosalicyl) fumarate.Arch. Biochem. Biophys. 258: 51–57.

Zaks, A. 1992. Protein-water interactions: Role in protein structure and stability.In The stability of proteins (eds. T.J. Ahern and M.C. Manning), pp. 249–274. Plenum Press, New York.1

Simons et al.


covalent cross-linking of proteins without chemical reagents

Documents