a method for selective conjugation of an analyte to enzymes without unwanted enzyme–enzyme...

ANALYTICALBIOCHEMISTRY

Analytical Biochemistry 331 (2004) 40–45

www.elsevier.com/locate/yabio

A method for selective conjugation of an analyte to enzymeswithout unwanted enzyme–enzyme cross-linking

Vincent C. Lombardi¤ and David A. Schooley

Department of Biochemistry, University of Nevada, Reno, Reno, NV 89503, USA

Received 20 November 2003 Available online 10 June 2004

Abstract

The conjugation of a ligand to an enzyme is often a necessary step in the development of enzyme-linked immunoassays. Such con-jugation is typically accomplished by reacting an amine with a carboxyl functional group in the presence of an activator such as acarbodiimide. However, one enzyme's free carboxyl groups often react with another's free amino groups and a large amount ofcross-linking between enzyme molecules occurs; few discrete enzyme molecules conjugated only to the ligand of interest areproduced. Hence, it is necessary to carry out laborious chromatographic puriWcation steps or to make an activated ligand such as anN-hydroxysuccinimide ester. This too can be a diYcult task because N-hydroxysuccinimide esters are not stable in protic solventsand many biological ligands that would be of interest are poorly soluble in organic solvents. This diYculty may limit the quantityand yield of product. We describe a method that eliminates enzyme–enzyme cross-linking by blocking the solvent-accessible carboxylgroups of horseradish peroxidase and alkaline phosphatase, with dialysis being the only puriWcation step necessary. We are conse-quently able to produce enzyme–ligand conjugates in high purity and in large quantity with little eVort and in a relatively shortperiod of time. 2004 Elsevier Inc. All rights reserved.

Keywords: Horseradish peroxidase; Alkaline phosphatase; Conjugation; Enzyme immunoassay

Although there have been continuous eVorts to other enzyme amino groups under the reaction condi-
improve various aspects of assays that are commonlyused in clinical and research laboratories, preparing alarge quantity of an analyte–enzyme conjugate of highpurity by a simple and eYcient process is diYcult. Forexample, in a typical enzyme-linked immunosorbentassay (ELISA)1, analyte–enzyme conjugates are usuallyprepared by direct conjugation of the analytes and theenzymes. In a conjugation reaction, when carboxylgroups in the ligand are intended to be conjugated toamino groups in the enzyme, a large amount of cross-linking occurs between enzyme carboxyl groups and
¤ Corresponding author. Fax: 1-775-784-1419.E-mail address: [email protected] (V.C. Lombardi).1 Abbreviations used: ELISA, enzyme-linked immunosorbent assay;

HRP, horseradish peroxidase; TMB, tetramethylbenzidine; AP,alkaline phosphatase; PBS, phosphate-buVered saline; EIA, enzymeimmunoassay.

0003-2697/$ - see front matter 2004 Elsevier Inc. All rights reserved.doi:10.1016/j.ab.2004.04.018

tions (e.g., in the presence of carbodiimide). Thus, fewnon-cross-linked enzyme molecules are available to beconjugated to the analyte to form discrete analyte–enzyme conjugates [1,2]. Therefore, it is often necessaryto carry out laborious chromatographic puriWcationsteps to obtain the analyte–enzyme conjugates separatedfrom cross-linked materials. Accordingly, only a smallamount of the conjugate can be prepared at a time.Clearly, a need exists for a more eYcient method for pre-paring various analyte–enzyme conjugates that areessential components of numerous assays currently inuse including ELISAs.We describe a simple procedurefor blocking surface-accessible carboxyl groups in twoenzymes widely used in ELISA assays, horseradish per-oxidase and alkaline phosphatase, with Tris, ammonia,or taurine to block the ability of these carboxyl groups toconjugate with amino groups in conjugation reactions.These modiWcations had little to no eVect on the kineticproperties of the enzymes used, allowing subsequent

mail to : [email protected]

V.C. Lombardi, D.A. Schooley / Analytical Biochemistry 331 (2004) 40–45 41

conjugation of amino functionality in the enzymes toligands without cross-linking occurring between enzymemolecules.

Materials and methods

Molecular visualization was performed with thecomputer program Rasmol Windows Version 2.6-ucb;horseradish peroxidase coordinates (4ATJ) wereobtained from the Protein Data Bank. Horseradish per-oxidase (product code HRP4) and alkaline phosphatase(product code ALPI13G) were from Biozyme Laborato-ries. 1-[3-(Dimethylamino)propyl]-3-ethylcarbodiimidemethiodide, 1-cyclohexyl-3-(2-morpholinoethyl) carbo-diimide metho-p-toluenesulfonate, urea hydrogenperoxide, 1,2-phenylenediamine dihydrochloride, 2�-O-monosuccinyladenosine 3�:5�-cyclic mono phosphatesodium salt, and 2�-O-monosuccinylguanosine 3�:5�-cyclic monophosphate sodium salt were from Sigma–Aldrich. N-Hydroxysulfosuccinimide was from Pierce.4-Nitrophenylphosphate disodium salt was from Acros.The peroxidase substrate 3,3�,5,5�-tetramethylbenzidine(TMB) was from KPL. All other reagents were analyti-cal or HPLC grade. Antibodies were from AmericanQualex. Kinetics and microplate assay measurementswere made with a Molecular Devices Vmax kinetic micro-plate reader. All data were plotted using the computerprogram Prism Version 3.00.

Molecular visualization

Simple molecular visualization was used to investigatethe possibility of adversely aVecting the catalytic proper-ties of the HRP by removing surface-accessible carboxylgroups. We began by displaying the entire enzyme in aCPK color scheme and then removing all atoms, with theexception of the heme, which is believed to be the core ofthe catalytic center. Next, the color scheme of all acidicamino acids was changed to green and the substrate usedin determining the crystal structure (benzhydroxamicacid) to blue. Finally, we progressively added back theatoms that are in proximity to the heme unit in 2-Å incre-ments, thereby outlining the active site of the enzyme.

Enzyme blocking

Carboxyl blocking was achieved by Wrst dissolving4.5 mg of HRP in 1 ml of 0.1 M 2-(N-morpholino) ethanesulfonic acid buVer, 0.1 M Tris, pH 5. Next, 15 mg of 1-[3-(dimethylamino) propyl]-3-ethylcarbodiimide methio-dide was added and the mixture was reacted for 2 h atroom temperature with gentle mixing [3]. The modiWedenzyme was then dialyzed twice against 4 L ofphosphate-buVered saline (PBS) pH 7.4, overnight.Blocking of AP was carried out in a similar manner with

the exception that a more sterically hindered carbodii-mide (1-cyclohexyl-3-(2-morpholinoethyl)carbodiimidemetho-p-toluenesulfonate) was used in place of 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methio-dide and a Hepes-buVered saline (50 mM Hepes, MgCl210 mM, ZnCl2 0.1 mM, pH 7.4) replaced PBS. Additionalblocking methods were evaluated for HRP whereby amolar equivalent of either ammonium chloride or tau-rine was substituted for Tris; all other conditions were asdescribed above.

Formation of active ester of ligand and conjugation

2�-O-monosuccinyladenosine 3�:5�-cyclic monophos-phate sodium salt (1 mg) and 3.1 mg of N-hydroxysulfo-succinimide were added to the above horseradishperoxidase solution (approx. 4.9 mg in 1.5 ml PBS, pH7.4). Then 14 mg of 1-[3-(dimethylamino)propyl]-3-ethyl-carbodiimide methiodide was added and the solutionwas mixed on an orbital shaker overnight at 4 °C [4]. The2�-O-monosuccinylguanosine 3�:5�-cyclic monophos-phate conjugate was made in an identical manner. 1.5 mgof 2�-O-monosuccinyladenosine 3�:5�-cyclic monophos-phate sodium salt, 4 mg of N-hydroxysulfosuccinimide,and 13 mg of 1-cyclohexyl-3-(2-morpholinoethyl)carbo-diimide metho-p-toluenesulfonate were added to thealkaline phosphatase solution (approx. 5000 units in1.5 ml of Hepes-buVered saline, pH 7.4). This solutionwas also mixed on an orbital shaker overnight at 4 °C.

Kinetics of blocked alkaline phosphatase and horseradishperoxidase phosphatase

Kinetic measurements were made with a kineticmicroplate reader in 96-well microtiter plates. The aver-age of eight replications of 12 substrate concentrationswas used to calculate relative kinetic values. Alkalinephosphatase kinetics were determined as follows: modi-Wed enzyme stock was diluted 1:10,000 (2 �l/20 ml) inTris buVer (Tris–HCl 50 mM, MgCl2 10 mM, and ZnCl20.1 mM, adjusted to pH 9.0, with 10 M NaOH). The solu-tion was allowed to reach room temperature (23 °C). Toeach well was added 50�l of Tris buVer and then 50 �l ofsubstrate. Substrate concentrations were 10, 8, 6, 4, 2, 1,0.5, 0.3, 0.2, 0.1, 0.05, and 0 mg/ml of 4-nitrophenylphos-phate in Tris buVer made immediately before use. Theplate was read at 405 nm 67 times over 10 min at 9-sintervals. HRP kinetics were performed as follows: mod-iWed enzyme stock was diluted 1:40,000 (1 �l/40 ml) inphosphate/citrate buVer (50 mM phosphate, 33.8 mMcitrate made with sodium phosphate dibasic and citricacid anhydrous adjusted to pH 5.0 [5]). This solution wasallowed to reach room temperature (23 °C). Next, 10 mlof a 12.8-mg/ml solution of 1,2-phenylenediaminedihydrochloride was made in phosphate/citrate buVerand 100�l was added to each well of a 96-well microtiter

42 V.C. Lombardi, D.A. Schooley / Analytical Biochemistry 331 (2004) 40–45

plate. Then 10 ml of a 12.8-mg/ml solution of ureahydrogen peroxide was made and serial dilutions of thiswere made as follows: 12.8, 6.4, 3.2, 1.6, 0.8, 0.4, 0.2, 0.1,0.05, 0.025, 0.0125, and 0.00625 mg/ml. Then 100�l ofeach was added to successive columns of a 96-wellmicrotiter plate. The plate was read at 450 nm 34 timesover 5 min at 9-s intervals. Kinetic values are reported asmillioptical density units/min (mOD/min). Data pointsfor the 12.8-mg/ml concentrations were discarded in theWnal calculation.

Test of cyclic nucleotide conjugates

HRP conjugates were tested according to the methodsof Horton et al. [6] with several modiWcations; speciWcally,a solution of goat anti-rabbit Fc antibody was preparedby suspending 27.5�l of antibody stock in 10ml of PBSbuVer. Then 90�l of this was added to each well of a 96-well F96 MaxiSorp Nunc-Immuno EIA plate and theplate was incubated for 4h at room temperature. The con-tents of the plate were then discarded and the plate wasblocked with 3% normal goat serum in PBS, 370�l perwell, for 1h. The contents were then discarded and theplate was next washed three times with 150�l of washingbuVer (0.05% solution of Tween 20 in PBS). Then, 75�l ofa 1:10,000 solution of rabbit anti-cAMP or anti-cGMPantisera in EIA buVer (0.15 M NaCl, Na2HPO4, 1 mMNa4EDTA, titrated to pH 7.4 with the same solutionexcept with NaH2PO4 replacing Na2HPO4) was added toeach well. Next, 25�l of eight standards was added to suc-cessive rows. Standards were 100 pmol/25�l to0.0001pmol/25�l diluted by factors of 10 in EIA buVer.Finally, 25�l of cyclic nucleotide–enzyme conjugate inEIA buVer was added to the microtiter plate as follows:columns 1–3 1:2000, 4–6 1:4000, 7–9 1:8000, and 10–121:16,000. The plates were then covered, mixed for 5 min ona microplate shaker, and incubated overnight at 4 °C. Thefollowing morning the contents of the plates were dis-carded and the plates were washed twice with washingbuVer. Next, 100�l of TMB peroxidase substrate wasadded and allowed to react for approximately 3min.Finally, the reaction was quenched with 100�l of 1 MH3PO4 and the plate read at 450 nm. AP conjugates weretested in an identical manner except 4-nitrophenylphos-phate (1 mg/ml prepared as previously described) replacedTMB and the reaction was not quenched.

Results

Substrate binding and catalysis

Fig. 1 shows the entire HRP enzyme in its dimericform and in CPK color scheme. Fig. 2 shows only theheme unit of the enzyme, which gives a starting point forexamining the active site. Fig. 3 displays all atoms of

nonhydrolysable substrate, benzhydroxamic acid (blue),and all atoms within 2Å of the heme unit. At this point,part of the enzyme active site is clearly visible. It is not untilall atoms are displayed within 6Å of the heme unit (Fig. 4)that we see any atoms belonging to acidic amino acids(green), only one of which is near the active site of themutant enzyme (carboxylic oxygen of Glu 42) and is notfound in the native enzyme [7]. At 10Å the active sitebecomes clearly deWned (Fig 5). Finally, the entire moleculeis displayed showing no acidic amino acids near theentrance to the active site (Fig. 6). This simple visualizationexperiment supports our later Wndings that no acidicamino acids are substantially involved in substrate binding.

Fig. 1. Entire HRP dimeric molecule in CPK color scheme. (For inter-pretation of the references to color in this Wgure legend, the reader isreferred to the web version of this paper.)

Fig. 2. Heme unit of HRP. (For interpretation of the references tocolor in this Wgure legend, the reader is referred to the web version ofthis paper.)

Fig. 3. Atoms within 2 Å of the heme unit. (For interpretation of thereferences to color in this Wgure legend, the reader is referred to theweb version of this paper.)



Vmax and Km values for the Tris-blocked enzyme(Table 1 and Figs. 7 and 8) show only slight deviationsfor the modiWed HRP compared to the unmodiWedenzyme. The Michaelis constant remained identicalwithin experimental error and the Vmax was decreased byapproximately 23%, an acceptable amount for our pur-pose. The eVect on AP was much greater. The Km wasincreased by 9.5% and Vmax was decreased by 43.4%(Table 1 and Figs. 9 and 10).

Total functional groups ligated

The mass of three carboxyl-blocked forms of HRPand one carboxyl-blocked amino-ligated HRP was mea-sured by matrix-assisted laser desorption ionizationmass spectrometry. The Tris- and ammonia-blockedenzyme showed an average of 4.2 and 4.0 moleculesadded, respectively, whereas HRP blocked with taurineshowed 3.4 molecules added. When Tris-blocked HRP


Fig. 6. Entire HRP dimeric molecule showing acidic residues (green)and substrate (blue). (For interpretation of the references to color inthis Wgure legend, the reader is referred to the web version of thispaper.)

Table 1Calculated kinetic values for unmodiWed and modiWed enzymes

Enzyme UnmodiWed ModiWed Percentage change

HRP Km (mOD) 1.849 1.841 0.004 (+)HRP Vmax (mOD min¡1) 1565 1212 22.6 (¡)AP Km (mOD) 0.8462 0.9295 9.5 (+)AP Vmax (mOD min¡1) 851.9 481.8 43.4 (¡)

was ligated with 2-O-monosuccinylguanosine-3�; 5�-cyclic monophosphate, the mass increase indicated thatan average of 5.0 molecules of the cyclic nucleotide hadconjugated to the enzyme (Table 2). No visible signs ofprecipitation were observed as a result of any of theblocking or ligation steps and all enzymes were catalyti-cally active (data not shown).

Fig. 7. Velocity vs concentration plot of unblocked HRP.

Fig. 8. Velocity vs concentration plot of blocked HRP.

Fig. 9. Velocity vs concentration plot of unblocked AP.

44 V.C. Lombardi, D.A. Schooley / Analytical Biochemistry 331 (2004) 40–45

Enzyme assays and conjugates

Competitive enzyme immunoassays were performedusing HRP–cAMP, HRP–cGMP, and AP–cAMP conju-gates blocked with Tris (Figs. 11–13, respectively). Vary-ing amounts of HRP conjugates were used to evaluatethe optimal concentration. Increased sensitivity wasobserved with decreasing conjugate concentration,although the linear range of the assay also decreased. Ata dilution of 1:16,000 we were able to measure cAMPlevels with an EC50 value of 1.27 pmol/25 �l and cGMP

Fig. 10. Velocity vs concentration plot of blocked AP.

Table 2Number of ligands added to HRP

a Mass of enzyme is an average mass of the three isoforms of HRP.b Indicates the mass increase from HRP–Tris.

Enzyme Massa Mass increase Ave. ligands added

HRP 43590.97 — —HRP–Tris 44180.64 589.67 4.03HRP–NH3 43732.16 141.19 4.24HRP–taurine 44073.30 482.33 3.37HRP–Tris–cGMP 46533.91 2353.27b 5.08

Fig. 11. Test of HRP–cGMP conjugate.

at an EC50 of 0.8 pmol/25 �l. As expected, a XuorescentHRP substrate gave even more sensitive results (data notshown). The AP conjugate with nitrophenyl phosphateas substrate was more sensitive than the HRP conjugatewith TMB as substrate and able to measure cAMP levelswith an EC50 value of 69 fmol/25 �l (Table 3).

Discussion

The aim of this work was to provide a way to enablethe facile conjugation of a ligand to an enzyme for use inimmunoassays. There are four types of functionalgroups found in native enzymes that lend themselves toconvenient conjugation chemistry. The groups areamines, carboxyl groups, sulfhydryl groups of cysteine,and sugar chains of glycosylation sites. Because cysteineis typically found as part of a disulWde bond and sugargroups are reactive only after oxidation, both functionalgroups are not free to react without modiWcation. Hence,only amine groups and carboxyl groups are usually freeto react [8,9]. Indeed, the most common type of conjuga-tion chemistry involves the condensation reaction

Fig. 12. Test of HRP–cAMP conjugate.

Fig. 13. Test of AP–cAMP conjugate.


between free amine and free carboxyl groups. However,most, if not all, enzymes have both free amine and car-boxyl groups and if one tries to conjugate a ligand to anenzyme by this approach, the amines on one enzyme willreact with the carboxyl groups on another molecule.This results in multimers of enzymes cross-linked to eachother, which are often insoluble. We have resolved thisproblem by blocking the carboxyl functional group,leaving mainly amines as unique reactive functionalgroups. This allows clean conjugation of a ligand with acarboxyl functional group to the amines of the enzyme.Unfortunately, commercially prepared conjugates ofenzyme to ligand rarely if ever are supplied with detailsof their purity, method of preparation, or number ofligands attached per enzyme molecule. We examined theactive site of HRP by molecular visualization withRasmol. This analysis revealed no acidic amino acid res-idues within 6 Å of the active site. It is consequently nottoo surprising that blocking surface-accessible carboxylgroups as amides with Tris had little eVect on the kineticproperties of the enzyme. However, alkaline phospha-tase has several acidic amino acids as functional residuesin the active site. Accordingly, we choose to modify thelatter enzyme with a larger, more sterically hinderedwater-soluble carbodiimide to hopefully prevent conju-gation of active site residues. The Km of alkaline phos-phatase was barely aVected by conjugation with Tris, butthe Vmax was decreased by 48%. However, the modiWedenzyme still allowed highly sensitive assays; for thecAMP conjugate the sensitivity was signiWcantly betterthan that for a cAMP conjugate to modiWed HRP.

We have developed a method whereby the conjuga-tion of a ligand to an enzyme may be accomplished with-out unwanted cross-linking and without the need tosynthesize a preformed activated intermediate. Thismethod is easy to carry out and does not require anextensive knowledge of organic synthesis nor does itrequire protracted puriWcation procedures or specialized

equipment. This method also lends itself well to large-scale production and, therefore, may have advantagesover present technology. Results with two enzymes andtwo ligands are presented in this work; however, thisprocedure has widespread application to many enzymesand ligands.

Acknowledgments

We thank the NIH (GM48172) for their Wnancialsupport, Drs. Timothy Kingan, David Lightner, andWolfgang PXeiderer for their helpful discussions, andRichard Bjur for assistance in our patent application.

References

[1] M. Aslam, Bioconjugation: Protein Coupling Techniques for theBiomedical Sciences, Macmillan Reference, London, New York,1998, Grove's Dictionnaries; 228.

[2] G.T. Hermanson, Bioconjugate Techniques, Academic Press, SanDiego, 1996, p. 171.

[3] H. Yamada et al., Selective modiWcation of aspartic acid-101 inlysozyme by carbodiimide reaction, Biochemistry 20 (17) (1981)4836–4842.

[4] J.V. Staros, R.W. Wright, D.M. Swingle, Enhancement by N-hydroxysulfosuccinimide of water-soluble carbodiimide-mediatedcoupling reactions, Anal. Biochem. 156 (1) (1986) 220–222.

[5] G. Wolters et al., Solid-phase enzyme-immunoassay for detection ofhepatitis B surface antigen, J. Clin. Pathol. 29 (10) (1976) 873–879.

[6] J.K. Horton et al., Enzyme immunoassays for the estimation ofadenosine 3�, 5� cyclic monophosphate and guanosine 3�, 5� cyclicmonophosphate in biological Xuids, J. Immunol. Methods 155 (1)(1992) 31–40.

[7] A. Henriksen et al., Structural interactions between horseradishperoxidase C and the substrate benzhydroxamic acid determinedby X-ray crystallography, Biochemistry 37 (22) (1998) 8054–8060.

[8] J.M. Bobbitt, Periodate oxidation of carbohydrates, Adv. Carbo-hydr. Chem. 11 (1956) 1–41.

[9] W.W. Cleland, Dithiothreitol, a new protective reagent for SHgroups, Biochemistry 3 (3) (1964) 480–482.

Table 3EC50 (nM) values for diVerent concentrations of HRP and AP conjugate in a competitive immunoassay

Conjugate 1:2000 1:4000 1:8000 1:16000

HRP–cAMP EC50 D 452 EC50 D 218 EC50 D 71.2 EC50 D 50.8(R2 D 0.94) (R2 D 0.95) (R2 D 0.97) (R2 D 0.96)

HRP–cGMP EC50 D 123 EC50 D 80.0 EC50 D 96.0 EC50 D 31.0(R2 D 0.98) (R2 D 0.989) (R2 D 0.996) (R2 D 0.990)

AP–cAMP — EC50 D 2.76 — —(R2 D 0.98)

a method for selective conjugation of an analyte to enzymes without unwanted enzyme–enzyme...

Documents