Membrane Protein Proteolysis Assayed by FluorescenceQuenching: Assay of O-Sialoglycoprotein Endopeptidase

Ping Jiang and Alan Mellors1

Guelph–Waterloo Centre for Graduate Work in Chemistry and Biochemistry, University of Guelph, Guelph,Ontario N1G 2W1, Canada

Received November 20, 1998

The assay of the O-sialoglycoprotein endopeptidase ofPasteurella haemolytica has previously used the cleav-age of 125I-labeled glycophorin A, measured by SDS–PAGE, autoradiography, gel-slicing, and scintillationcounting. A new assay is based on the increased fluores-cence which results from proteolytic cleavage of a fluo-rescence-quenched micellar substrate, 4,4-difluor-5,7-di-methyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acidconjugated to glycophorin A (BODIPY-FL–glycophorinA). Micellar association of glycophorin A molecules re-sults in 97% fluorescence quenching despite a low molarratio of BODIPY-FL–glycophorin A. Proteolysis of themembrane protein causes greatly enhanced fluores-cence which is used for a rapid one-step proteolysis as-say. Direct monitoring of proteolysis in microcuvettes,or routine assay in microtiter plates can be used. Repro-ducibility is higher than with the radiolabeled substrateand the Km values for the two substrates are similar. Theassay is suitable for the O-sialoglycoprotein endopepti-dase activity of chromatographically purified enzyme orunpurified bacterial culture supernatants and can beused to monitor inhibition of the O-sialoglycoproteinendopeptidase by neutralizing antibodies. The O-sialo-glycoprotein endopeptidase assay employing BODIPY-FL–glycophorin A provides a rapid and nonradioactivemethod for the assay of this highly specific enzyme.© 1998 Academic Press

Several serotypes of the bovine lung pathogen, Pas-teurella haemolytica, secrete a protease which has ahigh specificity for O-sialoglycoproteins, many of whichare membrane glycoproteins (1, 2). This potential vir-ulence factor has been named O-sialoglycoprotein en-dopeptidase (glycoprotease) (EC 3.4.24.57) because its

only known substrates are glycoproteins with exten-sive clusters of negatively charged sugar residues asO-sialoglycan or sulfoglycan conjugates. Its substratesinclude the human cell surface glycoproteins, glyco-phorin A, CD34, CD43, CD44, and CD45 (2); the li-gands for P- and L-selectins (3, 4); the tumor antigenepitectin (5); the vascular adhesion protein VAP-1 (6);platelet glycoprotein Ib (a) (7); and cranin, a brainO-sialoglycoprotein (8). Recently CD24, a substratewith sulfoglycan clusters but no sialoglycans, has beenshown to be a substrate (9). No predominantlyN-linked sialoglycoprotein, nonsialated substrate orsmall-molecular-weight peptide has been found to becleaved by the enzyme (1, 10). Due to its narrow spec-ificity, the enzyme has been found to be useful forstructural and functional analysis of cell surface glyco-proteins and for the immunomagnetic isolation of hu-man bone-marrow stem cells (11). Glycophorin A is theonly substrate available in quantities sufficient for aroutine assay. It is a transmembrane sialoglycoproteincontaining 15 O-linked and 1 N-linked glycans, and itis cleaved by the glycoprotease mainly at Arg31–Asp32(1). However, this substrate presents practical difficul-ties for the assay of the enzyme. Glycophorin is 60%carbohydrate by weight and includes many negativelycharged sialate residues. These polar sugars imparthigh solubility to the substrate and the products of theproteolytic cleavage, so that they cannot be separatedby protein precipitation, but must be resolved by SDS–PAGE2 or chromatography. Furthermore, the sub-strate and the products aggregate, even during SDS–

1 To whom correspondence should be addressed. Fax: (519) 766-1499. E-mail: mellors@chembio.uoguelph.ca.

2 Abbreviations used: BSA, bovine serum albumin; BODIPY-FL,4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propri-onic acid; Chaps, 3-[3(3-cholamidopropyl)dimethylammonia]-1-pro-pane sulfonate; CM, carboxymethyl; FTC, fluoroscein thiocarbamoyl;FITC, fluoroscein-5-isothiocyanate; Hepes, N-2-hydroxyethylpipera-zine sulfonate; SDS–PAGE, sodium dodecylsulfate–polyacrylamidegel electrophoresis.

8 0003-2697/98 $25.00Copyright © 1998 by Academic Press

All rights of reproduction in any form reserved.

ANALYTICAL BIOCHEMISTRY 259, 8–15 (1998)ARTICLE NO. AB982637

PAGE analysis, to form dimeric and higher ordercomplexes, due to the strong hydrophobic interactionbetween the transmembrane region of this integralmembrane protein. To overcome these problems, theassay for glycoprotease activity has used 125I-labeledglycophorin A as a substrate, with analysis by SDS–PAGE, autoradiography, gel-slicing, and scintillationcounting (10).

Some nonspecific protease assay systems have ex-ploited the enhanced fluorescence seen when fluores-cence-quenched hyperconjugated fluorescein deriva-tives of bovine serum albumin (BSA), casein, and othergeneral protease substrates are proteolytically cleaved(12, 13). However, hyperconjugation with dye can leadto steric hindrance of proteolytic enzyme action, by thesubstitution of many lysine residues, the major targetsof the conjugation reaction. Fluorescein-labeled caseinhas been used in some protease assays (14–16) thoughthe pH sensitivity of the substrate can present techni-cal problems in the assay. Schade et al. used BODIPY-fluorescein thiocarbamoyl (FTC)–casein as a pH-inde-pendent protein substrate for a protease assay inwhich the loss of fluorescence polarization was moni-tored as the indicator of proteolysis (17). A more con-venient assay, not dependent on polarization, uses thefluorescence quenching observed in BODIPY-FL–casein protein substrate in which interaction betweenthe fluorophore residues results in almost completequenching of most of the dye fluorescence. Proteolysisof this substrate by a variety of proteases results inincreased fluorescence as dye–peptide conjugates arereleased (18). Proteolysis of BODIPY-FL–casein gavemuch greater increases in fluorescence than did prote-olysis of BODIPY-FL–BSA (13), an effect which may bedue partly to intermolecular fluorescence quenchingwithin the heterogeneous casein micelle, comprisedprincipally of a, b, g, and k proteins (19). Here wereport the rapid and sensitive assay of O-sialoglyco-protein endopeptidase, using micellar BODIPY-FL-la-beled glycophorin A as the substrate. This substratehas only five lysines (mole/mole) available for conjuga-tion; however, it readily aggregates, and the resultantintermolecular interactions between fluorophores con-tribute to fluorescence quenching within the micelle.Proteolysis of this micellar protein results in a highlyreproducible increase in fluorescence which correlateswell with proteolysis measured by the time-consumingradiochemical assay.

MATERIALS AND METHODS

Preparation of O-Sialoglycoprotease

All materials, unless otherwise stated, were obtainedfrom Sigma Chemical Co. (St. Louis, MO). P. haemolyticaA1 (ATCC 43270) was stored in a 50% glycerol culturestock at 270°C. A 4.5-hour brain–heart infusion broth

culture was used to inoculate RPMI 1640 (Gibco) contain-ing 0.2% 3-[(3-cholamidopropyl)dimethylammonio]-1-propane sulfonate (Chaps). After 3–4 h incubation at37°C with aeration, the culture was centrifuged (10,000g)and the supernatant was filtered through a 0.2-mm Mil-lipore filter (culture supernatant). The filtered culturesupernatant was concentrated 20-fold by using a Milli-pore Prep Scale cartridge (Mr cutoff 10,000), the finalstage of the concentration used diafiltration to replacethe aqueous phase with 15 mM N-2-hydroxyethylpipera-zine sulfonate (Hepes) buffer, pH 7.4, containing 0.2%Chaps to yield a concentrated culture supernatant frac-tion. Chromatographic purification of this O-sialoglyco-protein endopeptidase preparation was by cation-ex-change chromatography on a Macro-Prep CM column(Bio-Rad, Mississauga, Ontario, Canada).

Preparation of BODIPY-FL-Labeled Glycophorin A

Human GPA was purified from erythrocyte ghostsprepared from outdated pooled blood (20, 21). The iso-lation of GPA included extraction with deoxycholate,phenol precipitation, ethanol precipitation, and gel-filtration column chromatography. BODIPY-FL conju-gation with glycophorin A was carried out by MolecularProbes Inc. (Eugene, OR) by treating 10 mg of glyco-phorin A with the dye 4,4-difluor-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid, succinimi-dylester (BODIPY-FL C3-SE, Molecular Probes Inc.,U.S. Patent 4,774,339). The conjugation, purification ofproduct, and determinations of percentage quenchingand mole ratio conjugation were carried out accordingto the method of Jones et al. (18). The Beer–Lambertlaw was tested for the product by determining thatlight absorbance at 505 nm was directly proportional toprotein concentration up to the maximal concentrationtested, 170 mg BODIPY-FL–glycophorin A/ml. Muchlower BODIPY-FL–GPA concentrations were used todetermine fluorescence quenching, and in enzymaticassays based on the release of fluorescence quenching.The product was determined to be 97.6% quenched forBODIPY-FL fluorescence, with an average molar ratioof BODIPY-FL residue:glycophorin A of 4.7. The theo-retical limit for the molar ratio is 6:1, based on thepresence of five lysine target residues and a free N-terminal amino residue in each glycophorin molecule,molar mass 31,000. The BODIPY-FL-labeled glyco-phorin A was kept protected from light at 270°C andwas stable for months.

Preparation of 125I-Labeled Glycophorin A

The purified GPA was radiolabeled with 125I by thesolid-state method of Markwell (22), using ‘‘Iodo-beads’’ (Pierce, Rockford, IL). Two beads were incu-bated for 5 min with 200 mCi [125I]NaI (ICN Radio-

9ASSAY OF O-SIALOGLYCOPROTEIN ENDOPEPTIDASE

chemical Inc.) at room temperature, and then 2 mgGPA was added and incubated at room temperature for20–30 min. The labeled GPA was separated from thefree [125I]-iodide by Sephadex G-25 gel filtration col-umn chromatography. The purified 125I-labeled GPAwas chemically stable at 220°C but its short half-lifenecessitates frequent preparation.

Radiochemical Assay with 125I-Labeled GPASubstrate

Enzyme was incubated with 3.5 mg 125I-labeled GPAin 50 mM Hepes, pH 7.4, at 37°C for 15–60 min. Thesubstrate and products were then separated by SDS–PAGE on 12% gels, which were dried and autoradio-graphed to locate the radiolabeled bands. Glycopro-tease activity was measured by excising the monomerand dimer substrate bands from the gel and countingtotal radioactivity in a gamma counter with correctionfor variations in the recovery of radiolabeled materialin each assay (10).

Fluorimetry Assay for Glycoprotease Activity

Enzyme was incubated with BODIPY-FL–GPA (3.5mg of GPA protein) in 50 mM Hepes buffer, pH 7.4, at37°C for 10–15 min in microfuge tubes in a total vol-ume of 25 ml. The reaction was stopped by diluting thesystem with 0.5 ml of ice-cold 50 mM Hepes bufferimmediately followed by fluorescence measurements ina Hitachi F-2000 spectrophotofluorimeter, slit widths 2nm, in a 46 3 7.5 3 7.5-mm quartz cuvette at ambienttemperature. The optimal wavelengths were deter-mined by scanning to be excitation at 480 nm andemission at 515 nm. For the recording of real-timeincreases in fluorescence, enzyme was incubated with0.175 mg BODIPY-FL–GPA in 70 ml 50 mM Hepesbuffer, pH 7.4, in an ultramicrofluorescence cuvette at37°C. Fluorescence was monitored for 10 min in aFluorolog DM 3000 fluorimeter (SPEX Industries Inc.,Edison, NJ) at excitation and emission wavelengths480 and 515 nm, respectively, slit widths 1.8 nm. Ki-netic analysis of initial velocities of hydrolysis werecarried out by a linear regression fit to the Michaelis–Menten equation by an ‘‘Enzfitter’’ computer program(Elsevier Software).

Microplate Fluorescence Reader Assay

Enzyme was incubated with BODIPY-FL–GPA (3.5mg protein) in 50 mM Hepes buffer, pH 7.4, total vol-ume of 25 ml, at 37°C for 10 minutes in 96-well fluo-rimeter microplates (Corning–Costar, Cambridge,MA). Each well was then diluted 20-fold with Hepesbuffer and fluorescence was measured immediately ina FL 500 Fluorescence Plate Reader (Bio-Tek Inc, Wi-

nooki, VT). Excitation and emission filter settings were485 and 520 nm, respectively.

RESULTS

Fluorescence Quenching within BODIPY-FL–GPA

To determine the fluorescence quenching for anaqueous solution of BODIPY-FL–GPA, the relative flu-orescence values were measured for BODIPY-FL–GPAand for a solution of free unconjugated BODIPY-FLwhich had the same absorbance at 505 nm. This com-parison showed that the BODIPY-FL–GPA was 97%quenched for fluorescence. The molar ratio ofBODIPY-FL fluorophore per mole protein was esti-mated from the absorbance of BODIPY-FL–GPA to be4.7 mol/mol: the anticipated limit is 6, the target lysineresidues plus the free N-terminal amino group for eachglycophorin molecule. BODIPY-FL–GPA, despite itslow molar ratio for conjugation, 4.7, shows significantfluorescence quenching (97%), apparently due in partto micellar association of the membrane protein andthe resultant intermolecular interactions between flu-orophores. This was confirmed by the addition of 1%w/v sodium dodecyl sulfate to an aqueous solution ofBODIPY-FL–GPA, which resulted in an increase influorescence to 250% of that of the control (Table 1) dueto breakup of the GPA micellar structure. A similarincrease was given by the zwitterionic detergent Chaps(1% w/v), which is used at lower concentrations in thepurification of the enzyme. At concentrations less than0.05% w/v, neither SDS nor Chaps increased the fluo-rescence of BODIPY-FL–GPA, and enzyme prepara-tions used in this study were dialyzed to reduce Chapsto below 0.01%.

Hydrolysis of BODIPY-FL–GPA by Proteases

Figure 1A shows that proteolysis of micellarBODIPY-FL–GPA results in increased fluorescence,

TABLE 1

Effect of Detergents on the Fluorescenceof BODIPY-FL–GPA

DetergentConcn(% w/v)

Fluorescence(% control)

None 100SDS 0.05 98 6 2SDS 0.10 120 6 2SDS 1.00 250 6 5Chaps 0.05 91 6 3Chaps 0.10 159 6 5Chaps 1.00 202 6 7

Note. The detergents were added to 3.5 mg BODIPY-FL–GPA in 25ml 50 mM Hepes buffer, pH 7.4, and equilibrated at 37°C for 15 min.The sample was diluted with 500 ml Hepes buffer and fluorescencewas measured as described under Materials and Methods. Means 6range for duplicate determinations are shown.

10 JIANG AND MELLORS

measurable by direct observation in real-time. Severalproteases were tested including the broad-specificityproteases, Streptomyces griseus protease and Tritira-chium album proteinase K, for comparison with theglycoprotein-specific O-sialoglycoprotein endopepti-dase (glycoprotease) from P. haemolytica. Within 450 s,fluorescence was increased 11-fold upon proteolysis byT. album proteinase K, 9.5-fold upon proteolysis by S.griseus protease, and 2.5-fold by glycoprotease action.More limited release of fluorescence quenching by theglycoprotease may be due to the hydropbobic associa-tion of larger product molecules: substrate and productaggregates can be seen even after SDS–PAGE analysis(23). Unlike hydrolysis by the broad specificity pro-teases from S. griseus and T. album, the P. haemolyticaglycoprotease shows limited hydrolysis of glycophorinA at specific sites (1). Negative controls (not shown)containing the heat-denatured P. haemolytica A1 cul-ture supernatant enzyme did not show any increase offluorescence. Figure 1B shows the real-time hydrolysisof BODIPY-FL–GPA by 1, 2, and 4 mg of partiallypurified glycoprotease enzyme protein. However, con-tinuous monitoring of the reaction rate of glycophorin

hydrolysis is impractical for routine enzyme assaysand more rapid methods were investigated.

O-Sialoglycoprotein Endopeptidase Hydrolysis ofBODIPY-FL–GPA

For routine assays, BODIPY-FL–GPA was incubatedwith the glycoprotease for 2–20 min at 37°C, and theincrease in fluorescence was measured at the end of theincubation. Figure 2 shows the increased fluorescenceseen when either 9 mg of concentrated culture super-natant (CCS) or 90 ng of CM gel-purified glycoproteaseenzyme were incubated with 70 mg of BODIPY-FL–GPA, in 0.5 ml 50 mM Hepes buffer, pH 7.5, and atvarious time intervals, a 25-ml aliquot was removedand diluted with 0.5 ml cold 50 mM Hepes buffer forfluorescence measurement. For reasons of economyand sensitivity, the concentration of substrate used islow, 4.5 mM compared with the Km of about 5 mMdetermined for 125I-labeled GPA (10). This low sub-strate concentration contributes to the decline in thehydrolysis rate over 20 min, but provides a sensitiveassay of proteolysis for both chromatographically puri-

FIG. 1. Direct monitoring of the proteolysis of BODIPY-FL–GPA by proteases. (A) Each protease (2 mg protein) was incubated with 0.18mg BODIPY-FL–GPA in 70 ml 50 mM Hepes buffer, pH 7.4, in an ultramicrocuvette at 37°C: (a) P. haemolytica O-sialoglycoproteinendopeptidase (glycoprotease); (b) S. griseus protease; (c) T. album proteinase K. (B) Glycoprotease: (a) 1 mg protein;(b) 2 mg protein; (c) 4 mgprotein.

11ASSAY OF O-SIALOGLYCOPROTEIN ENDOPEPTIDASE

fied and unpurified concentrated culture supernatantenzyme.

Comparison of the Hydrolysis of the FluorescentSubstrate with That of 125I-Labeled GPA

The O-sialoglycoprotein endopeptidase hydrolysis ofBODIPY-FL–GPA was directly compared with hydro-lysis of radiolabeled substrate GPA, under similar con-ditions. Both systems contained 3.5 mg of labeled GPA,either BODIPY-FL–GPA or 125I-labeled GPA, with con-centrated culture supernatant enzyme (0.10–0.16 mgprotein) in 25 ml 50 mM Hepes buffer, pH 7.4. Themixtures were incubated at 37°C for 10 min and thenthe fluorescence increase and the percentage degrada-tion of the radioactive substrate were measured, asdescribed in the methods. Figure 3 shows that there isa linear relationship between the increase of fluores-cence and substrate degradation measured by the ra-dioactivity assay. The standard errors of the means areshown for both assays (n 5 3) and are larger for theradiochemical assay, due to the inherent errors inSDS–PAGE analysis, gel slicing, and scintillationcounting. A major source of error in the radiochemicalmethod is the lack of reproducibility of the SDS–PAGEanalysis due to losses of the hydrophobic radiolabeledsubstrate by adhesion to the walls of pipets and tubes,as well as incomplete separation by electrophoresis.

These sources of variability are partially compensatedby correction for total recoveries of radiolabeled sub-strate and products as described under Materials andMethods. Similar comparisons between the fluorimet-ric assay and the radiochemical assay were carried outat several different concentrations of substrate (datanot shown) and confirmed the good correlation betweenthe two methods over a range of substrate concentra-tion.

Sensitivity of Glycoprotease Detection

Figure 4 shows the enzyme concentration depen-dence of the increase in fluorescence for the hydrolysisof BODIPY-FL–GPA by both the concentrated culturesupernatant enzyme and the chromatographically pu-rified enzyme. As little as 1 ng of chromatographicallypurified protein, containing the bulk of the glycopro-tease activity, could be detected (Fig. 4B). This lowerlimit for sensitivity corresponds to about 5% hydrolysisof the same concentration of radiolabeled substrate(3.5 mg per assay) in the 125I-labeled GPA assay sys-tem. The large standard error in the radiochemicalassay limits the sensitivity of this method, though sen-sitivity can be enhanced by extending the time of in-cubation of substrate with enzyme. To test the speci-ficity of the assay for the P. haemolytica A1

FIG. 2. The hydrolysis of BODIPY-FL–GPA by the glycoprotease,monitored by sampling. Aliquots (25 ml) were removed at intervalsfrom 0.5-ml reaction mixtures containing 70 mg BODIPY-FL–GPA in50 mM Hepes buffer, pH 7.4, and (‚) 90 ng purified enzyme; or (E) 9mg concentrated culture supernatant protein. Means 6 SE are shownfor independent triplicates.

FIG. 3. Comparison of glycoprotease action on BODIPY-FL–GPAmeasured fluorimetrically, with the equivalent hydrolysis of 125I-labeled GPA measured by SDS–PAGE. The 25-ml reaction mixturescontained either 4 mM BODIPY-FL–GPA or 4 mM 125I-labeled GPAand were incubated at 37°C for 10 minutes with 6 concentrations ofglycoprotease enzyme ( 0.10, 0.17, 0.32, 0.49, 0.87, and 1.65 mgprotein). Means 6 SE are shown by vertical and horizontal standarderror bars (n 5 3).

12 JIANG AND MELLORS

O-sialoglycoprotein endopeptidase activity, two relatedbacteria that are negative for the enzyme were tested(data not shown). P. haemolytica serotype A11 has anatypical organization of the glycoprotease gene anddoes not exhibit glycoprotease activity. Similarly theclosely related organisms formerly known as T bio-types of P. haemolytica and now classified as P. treha-losi do not possess glycoprotease activity (24). Whenculture supernatants from P. haemolytica serotype A11and P. trehalosi T3 were tested, no cleavage ofBODIPY-FL–GPA was observed.

Substrate Concentration

The BODIPY-FL–GPA substrate concentration wasvaried and the hydrolysis of this substrate, as for theradiolabeled GPA, followed Michaelis–Menten kinetics(Fig. 5). The Km for BODIPY-FL–GPA was estimatedto be 2.5 6 0.5 mM by linear regression. This Km valueis similar to that found for the radiolabeled substrate(5 mM) (10). In the routine assay, 3.5 mg of BODIPY-FL–GPA (4.5 mM) was chosen for optimal sensitivity.Higher substrate concentrations yield high back-ground fluorescence values with lower percentage in-creases in fluorescence, while lower concentrationsgive more deviation from linear initial velocities.

Neutralization of Glycoprotease by Calf Serum

The fluorescence assay is suitable for the detection ofenzyme neutralization by polyclonal antibodies from

the sera of cattle which have been exposed to P. hae-molytica infection. Bovine serum antibodies and mu-rine monoclonal antibodies are able to recognize theglycoprotease in immunoblots and are neutralizing forthe enzyme activity. Figure 6 shows that the calf poly-clonal sera pAb107 and pAb85 are neutralizing for theglycoprotease activity against the BODIPY-FL–GPAsubstrate, with strongest neutralization by pAb107 aspreviously demonstrated with the substrate [125I]-la-beled GPA (25). Thus the fluorescent assay can be usedfor rapid screening of serial dilutions of cattle sera, todetermine the immune status of the host and the se-lection of high titer antisera.

Microtiter Plate Format Assay

For the rapid and routine assay of large numbers ofsamples, fluorescence assays by microtiter plate readerare convenient. As described under Materials andMethods, the BODIPY-FL–GPA hydrolysis was carriedout in 96-well fluorimeter microtiter plates and theresults (Fig. 7) were essentially similar to those seenabove. The dependence of the hydrolysis on enzymeconcentration and on the concentration of substrateare shown: the latter were again used to estimate theKm of the glycoprotease enzyme (3.1 6 0.4 mM) to givea value close to previous estimates. Microtiter plate

FIG. 5. Substrate concentration dependence for the glycoproteasehydrolysis of BODIPY-FL–GPA. The 25-ml reaction mixture contain-ing 0.4 mg concentrated culture supernatant glycoprotease, incu-bated with varying amounts of BODIPY-FL–GPA in 50 mM Hepesbuffer at 37°C for 10 min. The reaction was stopped by dilution andthe fluorescence was measured. The inset shows the double recipro-cal plot for the same data. Means 6 SE (n 5 3).

FIG. 4. Hydrolysis of BODIPY-FL–GPA by P. haemolytica culturesupernatant and by chromatographically purified glycoprotease en-zyme. The assay (25 ml) contained 3.5 mg BODIPY-FL–GPA incu-bated with the glycoprotease extract in 50 mM Hepes, pH 7.4, for 10min, before dilution and fluorescence measurement. (A) concentratedculture supernatant; (B) carboxymethyl-agarose-purified glycopro-tease. Means 6 SE (n 5 3).

13ASSAY OF O-SIALOGLYCOPROTEIN ENDOPEPTIDASE

assay is especially suited for the determination of O-sialoglycoprotein endopeptidase activity neutralizationby serum antibodies, since logarithmic dilutions of seracan be applied to adjacent rows of wells containingenzyme.

DISCUSSION

The BODIPY-FL–GPA substrate was 97% fluores-cence-quenched, a high degree of quenching for a mol-ecule which has only 4.7 mol fluorophore/mol protein.High degrees of fluorescence quenching can be pro-duced by the hyperconjugation of fluorescein with largeproteins having many lysine residues per molecule. Forexample, for hyperconjugated fluorescein–FTC–bovineserum albumin with 59 lysines, the fluorophore molarratio was 25 and the reported percentage quenching offluorescence was 98%, for this soluble protein in whichfluorescence quenching is due to intramolecular inter-actions (13). Recently BODIPY-FL–bovine serum albu-min was prepared with a fluorophore molar ratio of 12and a fluorescence-quenching of 98% was observed(18). Thus the BODIPY-FL–fluorophore appears to bemore sensitive to quenching effects than the fluores-cein–FTC conjugate. These workers also prepared themore heterogeneous BODIPY-FL–casein substrate forwhich the average lysine mole ratio was about 5, andthey obtained a fluorescence quenching of 98% (18).However, whereas albumin is a soluble homogeneous

protein, casein exists in solution as a large micellaraggregate of different proteins, mainly a, b, g, and kcaseins. Thus the high degree of fluorescence quench-ing observed for casein could be due to intermolecularas well as intramolecular interactions. BODIPY-FL–GPA which has a low fluorophore molar ratio, 4.7, isassociated in aqueous solution as micelles. GlycophorinA, a membrane protein with a single transmembranehydrophobic region, associates in aqueous solution toform micelles of 10 or more molecules per aggregate(26). The high degree of fluorescence quenching istherefore due in part to intermolecular interaction offluorophores. The importance of intermolecular inter-actions was confirmed by the increased fluorescencedue to loss of quenching when excess sodium dodecylsulfate was added, under conditions which yield oneglycophorin A molecule per SDS micelle. Low levels ofdetergent, including the zwitterionic detergent Chapsused in O-sialoglycoprotein endopeptidase prepara-tion, did not affect fluorescence quenching, but careshould be exercised to check that surface active agentsare not responsible for fluorescence changes during theenzymatic hydrolysis of BODIPY-FL-labeled micellarsubstrates. It is difficult to determine exactly howmuch of the observed decrease in fluorescence quench-ing is due to cleavage of the substrate compared withdecreases caused by reduced intra- or intermolecularassociations. While Table 1 shows that some fluores-

FIG. 7. Assay of the glycoprotease in microtiter plate wells. Theconcentrated culture supernatant glycoprotease enzyme was incu-bated with BODIPY-FL–GPA in 50 mM Hepes buffer, pH 7.4, in atotal volume of 25 ml, at 37°C for 10 min in the 1-ml wells of afluorimeter microtiter plate. Each well was then diluted 20-fold bythe addition of buffer and fluorescence was measured. (A) Substrateconcentration in 25 ml, enzyme protein 0.7 mg; (B) Protein concen-tration in 25 ml, substrate concentration 3 mg BODIPY-FL–GPA.Means 6 SE (n 5 3).

FIG. 6. Neutralization of glycoprotease by calf sera. The 10-foldserial dilutions of the serum samples were tested for neutralizingactivity. (1) Concentrated culture supernatant glycoprotease en-zyme; (2–5) enzyme with calf serum pAb107 at 10-, 100-, 1000-, and10,000 -fold dilution; (6–9) enzyme with calf serum pAb85 at 10-,100-, 1000-, and 10,000-fold dilution. Means 6 SE (n 5 3).

14 JIANG AND MELLORS

cence quenching is reduced by detergents in the ab-sence of cleavage, much more fluorescence quenching islifted by proteolytic cleavage (Fig. 1). Furthermore Fig.3 clearly demonstrates a direct relationship betweenproteolysis of a radiolabeled substrate and decreasedfluorescence quenching for the BODIPY-FL-labeledsubstrate, whether caused by proteolytic cleavage ordecreased associations of fluorophores.

The assay of glycophorin A hydrolysis by the O-sialoglycoprotein endopeptidase or by any other pro-tease has previously been a time-consuming procedure,involving many steps in the manipulation of radioac-tive substrate and products. The procedure describedin this study has the advantages of speed, safety, econ-omy, and a greater reproducibility. The sensitivity ofthe assay is comparable to the radioactive substrateassay. The fluorescent substrate is more stable thanthe short-lived radioactive glycophorin A. The hydro-lysis of the fluorescent substrate can be monitoreddirectly for accurate estimates of initial velocities. Al-ternatively, the assay can be carried out in microtiterplates, for large-scale routine testing of enzyme prep-arations, bacterial strains, or neutralizing antibodies.Other membrane proteins which associate as micelles,or soluble proteins which aggregate, may show similarfluorescence-quenching properties which can be ex-ploited in the study of the hydrolysis or the dissociationof the aggregated protein.

The use of fluorescence quenching to measure thehydrolysis of membrane proteins could be extended toother membrane protein substrates and to studies onthe rate of association and dissociation of membraneprotein aggregates. When other molecules such as un-labeled membrane protein, lipids, or soluble ligandsinteract with the BODIPY-FL-labeled membrane pro-tein, these are likely to reduce the intermolecular flu-orescence quenching. Therefore the association be-tween other molecules and the membrane proteincould be analyzed kinetically by this technique.

ACKNOWLEDGMENTS

We acknowledge the excellent technical assistance of W. M. Clad-man and we thank Dr. R. P. Haugland for helpful suggestions.OMAFRA, NSERC Canada and Cedarlane Laboratories providedfinancial support.

REFERENCES

1. Abdullah, K. M., Udoh, E. A., Shewen, P. E., and Mellors, A.(1992) Infect. Immun. 60, 56–62.

2. Sutherland, D. R., Abdullah, K. M., Cyopick, P., and Mellors, A.(1992) J. Immunol. 148, 1458–1464.

3. Steininger, C. N., Eddy, C. A., Leimgruber, R. M., Mellors, A.,and Welply, J. K. (1992) Biochem. Biophys. Res. Commun. 188,760–766.

4. Norgard, K. E., Moore, K. L., Diaz, S., Stults, N. L., Ushiyama,S., McEver, R. P., Cummings, R. D., and Varki, A. (1993) J. Biol.Chem. 268, 12764–12774.

5. Hu, R.-H., Mellors, A. and Bhavanandan, V. P. (1994) Arch.Biochem. Biophys. 310, 300–309.

6. Salmi, M., and Jalkanen, S. (1996) J. Exp. Med. 183, 569-579.7. Yeo, E. L., and Sutherland, D. R. (1995) in Leukocyte Typing (V):

White Cell Differentiation Antigens (Schlossmann, S. F., Boum-sell, L. Gilks, W., Harlan, J. M., Kishimoto, T., Moritomo, C.,Ritz, J., Shaw, S., Silverstein, R., Springer, T., Tedder, T. F., andTodd, R. F., Eds.), pp. 1313–1315, Oxford Univ. Press, Oxford.

8. Smalheiser, N. R., and Kim, E. (1995) J. Biol. Chem. 270, 15425–15433.

9. Aigner, S., Sthoeger, Z. M., Fogel, M., Weber, E., Zarn, J., Rup-pert, M., Zeller, Y., Vestweber, D., Stahel, R., Sammar, M., andAltevogt, P. (1997) Blood 89, 3385–3395.

10. Mellors, A. and Lo, R. Y. C. (1995) Methods Enzymol. 248,728–740.

11. Mellors, A., and Sutherland, D. R. (1994) Trends Biotechnol. 12,15–18.

12. Homer, K. A., and Beighton, D. (1990) Anal. Biochem. 191,133–137.

13. Voss, E. W., Jr., Workman, C. J., and Mummert, M. E. (1996)BioTechniques 20, 286–291.

14. Spencer, R. D., Teledo, F. B., Williams, B. T., and Yoss, N. L.(1973) Clin. Chem. 19, 838–844.

15. Maeda, H. (1979) Anal. Biochem. 92, 222–227.16. Bolger, R., and Checovich, W. (1994) BioTechniques 17, 585–589.17. Schade, S. Z., Jolley, M. E., Sarauer, B. J., and Simonson, L. G.

(1996) Anal. Biochem. 243, 1–7.18. Jones, L. J., Upson, R. H., Haugland, R. P., Panchuk-Voloshina,

N., Zhou, M., and Haugland, R. P. (1997) Anal. Biochem., 251,144–152.

19. Farrell, H. M., Jr. (1988) Physical Equilibria: Protein Funda-mentals of Dairy Chemistry, pp. 461–463, Van Nostrand, NewYork.

20. Dodge, J. T., Michell, C., and Hanahan, D. J. (1963) Arch. Bio-chem. Biophys. 100, 119–130.

21. Segrest, J. P., Wilkinson, T. M., and Sheng, L. (1979) Biochim.Biophys. Acta 554, 533–537.

22. Markwell, M. A. K. (1982) Anal. Biochem. 125, 427–432.23. Watt, M. A. V., Lo, R. Y. C., and Mellors, A. (1997) Cell Stress

Chaperones 2, 180–190.24. Abdullah, K. M., Lo, R. Y. C., and Mellors, A. (1990) Biochem.

Soc. Trans. 18, 901–903.25. Watt, M. A. V., Mellors, A., and Lo, R. Y. C. (1997) FEMS

Microbiol. Lett. 147, 37–43.26. Chu, J. W. K., and Sharom, F. J. (1992) Cell. Immunol. 145,

223–239.

15ASSAY OF O-SIALOGLYCOPROTEIN ENDOPEPTIDASE