Accurate measurement of avidin and streptavidin in crude biofluids with a new, optimized biotin–fluorescein conjugate

Accurate measurement of avidin and streptavidin in crude bio£uids witha new, optimized biotin^£uorescein conjugate

Gerald Kada a, Heinz Falk b, Hermann J. Gruber a;*a Institute of Biophysics, J. Kepler University, Altenberger Str. 69, A-4040 Linz, Austriab Institute of Chemistry, J. Kepler University, Altenberger Str. 69, A-4040 Linz, Austria

Received 9 October 1998; received in revised form 18 December 1998; accepted 22 December 1998

Abstract

A new biotin^fluorescein conjugate with an ethylene diamine spacer was found to be the first fluorescent biotin derivativewhich truly mimicked D-biotin in terms of high affinity, fast association, and non-cooperative binding to avidin andstreptavidin tetramers. These exceptional properties were attributed to the small size/length of the new ligand since all larger/longer biotin derivatives are known for their mutual steric hindrance and anti-cooperative binding in 4:1 complexes withavidin and streptavidin tetramers. Specific binding of the new biotin^fluorescein conjugate towards avidin and streptavidinwas accompanied by 84^88% quenching of ligand fluorescence. In the accompanying study this effect was used for rapidestimation of avidin and streptavidin in a new `single tube assay'. In the present study the strong quenching effect was utilizedto accurately monitor stoichiometric titration of biotin-binding sites in samples with v 200 pM avidin or streptavidin. Theconcentration was calculated from the consumption of fluorescent ligand up to the distinct breakpoint in the fluorescencetitration profile which was marked by the abrupt appearance of strongly fluorescent ligands which were in excess. Due to thisprotocol the assay was not perturbed by background fluorescence or coloration in the unknown samples. The newfluorescence titration assay is particularly suited for quick checks on short notice because getting started only means to thawan aliquot of a standardized stock solution of fluorescent ligand. No calibration is required for the individual assay and theligand stock solution needs to be restandardized once per week (or once per year) when stored at 325³C (or at 370³C,respectively). ß 1999 Elsevier Science B.V. All rights reserved.

Keywords: Avidin; Biotin; Fluorescein; Fluorescence; Streptavidin

1. Introduction

Avidin and streptavidin are versatile tools in bio-science, diagnostics, and technology [1,2], thereforequantitative measurement of (strept)avidin is astandard task. Published assays can be groupedinto three categories: (i) radioligand binding meth-ods [3,4] with a sensitivity limit of 3 pM streptavidin[3], (ii) enzyme assays [5^7] with a detection limit of150 pM [5], and (iii) photometric/£uorimetric meth-ods [8^11] which are far less sensitive (0.05^1 WM).

Radioligand and enzyme assays also excel in spe-

0304-4165 / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved.PII: S 0 3 0 4 - 4 1 6 5 ( 9 8 ) 0 0 1 7 8 - 0

Abbreviations: biotin-4-FITC, 5-[2-(biotinoyl)-aminoethylthio-ureidyl]-£uorescein isothiocyanate; biotin-4-£uorescein, 5-(and6)-[2-(biotinoyl)-aminoethylaminocarbonyl]-£uorescein; biotin-NHS, N-hydroxysuccinimidyl ester of D-biotin; Boc, t-butyloxy-carbonyl group; Boc2O, di-tert-butyl pyrocarbonate; BSA, bo-vine serum albumin; DACA, p-dimethylaminocinnamaldehyde;EDTA, ethylenediamine-N,N,NP,NP-tetraacetic acid; `£uores-cein^biotin', 5-{[N-(5-{N-[6-(biotinoyl)-amino]-hexanoyl}-ami-no)-pentyl]-thioureidyl}-£uorescein; FITC, £uorescein isothio-cyanate; RT, room temperature; (strept)avidin, streptavidinand/or avidin

* Corresponding author. Fax: +43-732-2468-822;E-mail : [email protected]

BBAGEN 24764 4-3-99

Biochimica et Biophysica Acta 1427 (1999) 33^43

ci¢city, i.e., they are insensitive to a large excess ofother sample components. Their disadvantages are (i)expensive components with limited lifetime, (ii) labo-rious protocols, (iii) laborious calibration with mod-erate accuracy, (iv) long assay times between 2 and20 h, and (v) non-linear dose^response curves withmoderate precision.

Conventional photometric/£uorimetric assays, onthe other hand, are cheap, simple, fast, and muchmore accurate, their disadvantages being poor sensi-tivity1 for (strept)avidin and restriction to near ab-sence of other proteins and/or colored components[8^11].

Recently a new £uorimetric assay for (strept)avi-din has been proposed [12] which relies on the stoi-

chiometric binding of commercial `£uorescein^biotin'to the biotin-binding sites of (strept)avidin (Kd 6 0.4nM) and on the strong quenching of `£uorescein^biotin' in the bound state. This new assay can meas-ure V2 nM (strept)avidin with high accuracy incrude, colored samples. Thus it has ¢lled a gapamong all previous methods.

Nevertheless, the new £uorimetric assay showed ageneral drawback which derived from the anti-coop-erative binding of commercial `£uorescein^biotin' toavidin and streptavidin (see Fig. 1A,B). Avidin andstreptavidin are homotetramers with pairwise ar-rangement of the four biotin-binding sites [13^17].As argued in the discussion of [12], the 14-atomspacer of `£uorescein^biotin' allowed for direct con-tact (and steric repulsion) between adjacently bound£uorescein residues (Fig. 1A,B) which caused incon-veniently slow association kinetics at 6 40 nM avi-din [12]. In the case of streptavidin an additionalcomplication was caused by the shorter `e¡ective'distance between neighbouring biotin-binding sites[13^17] and by the well-known pseudo-speci¢c inter-action of £uorescein residues with empty biotin-bind-ing sites [18^22]. Thus on a time scale of minutesonly one `£uorescein^biotin' was bivalently bound

Fig. 1. Model for the interaction of avidin tetramers (A,C) and streptavidin tetramers (B,D) with biotin^£uorescein conjugates. (A,B)Anti-cooperative binding of commercial `£uorescein^biotin' at adjacent biotin-binding sites. (C,D) Non-cooperative binding of theshort ligands biotin-4-FITC or biotin-4-£uorescein.

1 A single, interesting exception must be named: Schray et al.[31] have already synthesized biotin-4-FITC (see Fig. 2) by adi¡erent route and at 4 nM biotin-4-FITC they could correlate£uorescence polarization with avidin concentrations in the sub-nanomolar range. Unfortunately, the precision of this assay wasat best semiquantitative which explains why this assay has notbeen used since. As shown in our accompanying study [38], theerrors of the same assay were reduced to a few percent whenprecoating assay tubes with BSA.

BBAGEN 24764 4-3-99

G. Kada et al. / Biochimica et Biophysica Acta 1427 (1999) 33^4334

to a pair of biotin-binding sites (Fig. 1B, left side)while saturation with two ligands per pair of siteswas 10-fold slower (Fig. 1B, right side) [12].

The present study was based on the assumptionthat much shorter biotin^£uorescein conjugates (seeFig. 2) would be con¢ned to a single biotin-bindingsite (Fig. 1C,D), thereby eliminating any anti-coop-erative and/or bivalent binding (compare Fig. 1Aand B). Two new questions arose with new shortbiotin^£uorescein conjugates: (i) Is the a¤nity inthe desired subnanomolar range?; (ii) is £uorescencequenching in the (strept)avidin-bound state compa-rable to that of commercial `£uorescein^biotin'?

The results more than met all expectations: thebiotin^£uorescein conjugate with an ethylene diam-ine spacer (Fig. 2) turned out to be the ideal probefor biotin-binding sites: binding was strong, fast, andnon-cooperative, and the relative extent of quenchingin the bound state was the largest observed with anybiotin^£uorophore conjugate [23,24].

2. Materials and methods

2.1. Reagents and bu¡ers

P.a. grade materials were used as far as commer-cially available. A¤nity-puri¢ed avidin, streptavidin,and D-biotin were obtained from Sigma. Puri¢edBSA was purchased from Calbiochem-Behring.FITC, as well as 5-(and 6)-carboxy£uorescein succi-nimidyl ester were obtained from Molecular Probes.All other materials were bought from Merck. Biotin^NHS was synthesized as described [25]. Bu¡er A (100mM NaCl, 50 mM NaH2PO4, 1 mM EDTA, pHadjusted to 7.5 with NaOH) was used throughout.

2.2. Synthesis of biotin-4-FITC andbiotin-4-£uorescein

As a general precaution, all reactions were carriedout under an argon atmosphere. No argon was usedduring product workup, therefore oxidation of pri-mary amino groups was suppressed by addition ofacetic acid. In order to avoid premature cleavage ofBoc groups by acetic acid, toluene was included andthe rotavap was connected to a strong vacuumsource (1^10 Pa) in order to remove all solvents atroom temperature without heating. The 1.5 m tubing(8 mm inner diameter) between rotavap and coldtrap served to maintain a convenient vacuum gra-dient in the early stage of evaporation.

TLC solvents were 70:30:4 mixtures of chloro-form/methanol/x, with x = acetic acid, water, or con-centrated ammonia in solvents I, II, and III. Iodinevapor was used to stain all components, primaryamines were detected with ninhydrin [26], and biotinend groups with DACA [27].

Ethylene diamine (1.6 ml, 25.7 mmol) was dis-solved in 50 ml methanol and cooled to 37³C.Boc2O (4.6 g, 21 mmol, in 20 ml methanol) was

Fig. 2. Synthesis of biotin-4-£uorescein and biotin-4-FITC, andcomparison with commercial `£uorescein^biotin'.

BBAGEN 24764 4-3-99

G. Kada et al. / Biochimica et Biophysica Acta 1427 (1999) 33^43 35

added over a 1 h period. Subsequently the ice/saltbath was removed and stirring was continued for 1h at RT. Methanol was removed without heating (seeabove). The residue was dissolved in 200 ml chloro-form and a mixture of 60 ml toluene and 1.5 mlacetic acid was quickly added. The solvents wereremoved and the residue was azeotroped twice with60 ml toluene. The residue was chromatographed onsilica 60 (100 g, Merck) using TLC solvent I andcollecting 50 ml fractions. The di-Boc derivativewas in fraction 3 (RI

f = 0.83, traces only), Boc^NH^CH2^CH2^NH2 in fraction 6^12 (RI

f = 0.31,RII

f = 0.10, RIIIf = 0.48), and ethylene diamine elution

started at fraction 23 (RIf = 0.03). Product fractions

were combined, 200 ml toluene was added and thevolume was reduced to 100 ml. Another 100 ml tol-uene was added and the mixture was taken to dry-ness. The residue was azeotroped with 100 ml tolueneand dried at 1^10 Pa for 2 h. The residue was dis-solved in 20 ml chloroform and ¢ltered through pa-per to remove silica particles. The ¢ltrate was takento dryness and dried at 1^10 Pa overnight. Yield: 0.5g (2.3 mmol, 24% of theory) Boc^NH^CH2^CH2^NH�3 acetate. The product was pure by TLC. 1H-NMR (N, D6-DMSO, 200 MHz): 1.41 (s, t-butyl)1.94 (s, acetate) 2.96 (m; CH2-NH-tBoc) 3.32 (m,CH2^NH�3 ) 6.18 (bs, NH�3 ).

Boc^NH^CH2^CH2^NH�3 acetate (0.22 g, 1 mmol)was dissolved in 4 ml DMF and biotin^NHS (0.5 g,1.45 mmol, in 5 ml DMF) and triethylamine (0.2 ml,1.6 mmol, in 1 ml DMF) were added. After 1 h ofstirring at RT the reaction was terminated by addi-tion of water (2 ml) to destroy unreacted biotin^NHS and stirring was continued for 1 h. The sol-vents were removed and the residue dried at 1^10Pa overnight. For removal of biotin and N-hydrox-ysuccinimide the product was suspended in 50 mlchloroform and the suspension was washed withNa2CO3 solution (5 g in 50 ml water, saturatedwith NaCl). The chloroform suspension was ¢lteredthrough a glass frit and the solid product in the fritwas dissolved with ethanol. Removal of ethanol anddrying gave 165 mg biotin^NH^CH2^CH2^NH^Boc(0.43 mmol, 43% of theory). The product was pureby TLC (RI

f = 0.69, positive with DACA, positivewith ninhydrin after s 1 min delay at 100³C, dueto thermal deprotection). 1H-NMR (N, D6-DMSO,500 MHz): 1.22^1.59 (6H, m, CH2^CH2^CH2^

CH2^CO^NH, biotin side chain) 1.36 (9H, s, t-butyl)2.03 (2H, t, J = 7.2 Hz, CH2CONH, biotin sidechain) 2.56 (1H, d, Jgem = 12.5 Hz, SCH2(I), biotin)2.80 (1H, q, Jgem = 12.5 Hz, Jvic = 5 Hz, SCH2 (II),biotin) 2.93 (2H, q, J = 6 Hz, Boc^NHCH2) 3.02(2H, q, J = 6 Hz, biotin^NHCH2) 3.08 (1H, m,SCH, biotin) 4.12 (1H, m, NCH, biotin) 4.29 (1H,m NCH, biotin) 6.50 (2H, m, NHCONH, biotin)6.78 (1H, s, t-Boc^NH) 7.80 (1H, s biotin^NH).

For synthesis of biotin-4-£uorescein, 52 Wmol ofbiotin^NH^CH2^CH2^NH^Boc were dissolved in98^100% formic acid (2 ml) and water (40 Wl) wasadded. After stirring at RT (3 h) the solvent wasremoved and the residue was dried at 1^10 Pa (2 h).The intermediate was dissolved in DMF (1.3 ml),5-(and 6)-carboxy£uorescein succinimidyl ester (28.4mg, 60 Wmol) was added as a solid, and the suspen-sion was stirred while slowly adding 200 Wl triethyl-amine. After 2 h of stirring at RT the solvents wereremoved. The residue was dissolved in 3 ml of TLCsolvent I and 1 ml of this solution was applied to apreparative TLC plate (200U200U2 mm, silica 60,Merck) which was developed in TLC solvent I. Theproduct band was scraped o¡, crushed to powder,and extracted with plenty of water (10U25 ml).Evaporation of water gave 3.3 mg (4.4 Wmol, 25%of theory) of biotin-4-£uorescein. Purity was con-¢rmed by TLC (RI

f = 0.56, RIIf = 0.29, positive stain

with DACA) and by HPLC (RP-18 column, eluentsas in ref. [28]). 1H-NMR (N, CDCl3, 500 MHz):1.11^1.66 (6H, m, CH2^CH2^CH2^CH2^CO^NH,biotin side chain) 2.03 (2H, t, J = 7.2 Hz,CH2CONH, biotin side chain) 2.50 (1H, d,Jgem = 12.5 Hz, SCH2(I), biotin) 2.79 (1H, q,Jgem = 12.5 Hz, Jvic = 5 Hz, SCH2 (II), biotin) 2.91,2.98 (4H, m, Boc^NHCH2, biotin^NHCH2) 3.08(1H, m, SCH, biotin) 4.08 (1H, m, NCH, biotin)4.26 (1H, m NCH, biotin) 5.28, 5.48 (2U1H, s, t-Boc^NH, biotin^NH) 5.94^7.52 (6H, m, £uorescein,xanthene residue) 7.96^8.84 (3H, m, £uorescein,phthalide residue). MS (electrospray, NH3) m/z(%) = 644 (8; M), 363 (9), 349 (31), 255 (100), 241(34), 141 (70), 113 (63).

For the synthesis of biotin-4-FITC, 31 Wmol ofbiotin^NH^CH2^CH2^NH^Boc were deprotected asdescribed above and dissolved in 0.5 ml DMF. Afteraddition of FITC (39 Wmol) and triethylamine (200Wl) the mixture was stirred for 1 h at RT. The sol-

BBAGEN 24764 4-3-99


vents were removed and one third of the crude prod-uct was chromatographed by preparative TLC asdescribed above. Yield: 2.1 mg (2.6 Wmol, 26% oftheory) biotin-4-FITC, pure by TLC (RI

f = 0.0.53,RII

f = 0.34, RIIIf = 0.02, positive stain with DACA)

and by HPLC (see above). 1H-NMR (N, CDCl3,500 MHz): 1.14^1.56 (6H, m, CH2^CH2^CH2^CH2^CO^NH, biotin side chain) 1.80 (2H, t,J = 7.2 Hz, CH2CONH, biotin side chain) 2.52 (1H,d, Jgem = 12.5 Hz, SCH2(I), biotin) 2.74 (1H, q,Jgem = 12.5 Hz, Jvic = 5 Hz, SCH2 (II), biotin) 3.02,3.08 (4H, m, Boc^NHCH2, biotin^NHCH2) 3.20(1H, m, SCH, biotin) 4.10 (1H, m, NCH, biotin)4.25 (1H, m NCH, biotin) 6.27^7.03 (6H, m, £uores-cein, xanthene residue) 7.88^8.68 (3H, m, £uorescein,phthalide residue) 9.54 (1H, broad signal, C(S)NH^£uorescein). MS (electrospray, NH3) m/z (%) = 675(28; M), 337 (49), 255 (23), 243 (60), 113 (100).

2.3. Standardization of stock solutions

The primary standard was a 400 WM D-biotinstock solution prepared by dissolving 99% pure D-biotin in bu¡er A. Aliquots were stored at 325³C.

Avidin and streptavidin were dissolved in bu¡er Aat 2 WM (nominal concentration by weight) andstored at 4³C for up to 1 week or at 325³C for upto 1 month without re-freezing. The functional con-centration of avidin or streptavidin was determinedby titration with D-biotin [9,12]. Typically, 80 Wl ofV2 WM (strept)avidin were added to 1920 Wl bu¡erA in a stirred cuvette and the decrease in tryptophan£uorescence was monitored while adding a 16 WM D-biotin stock solution in 5-Wl increments at 1-min in-tervals [12]. The breakpoint between progressivequenching and the subsequent plateau allowed tocalculate the amount D-biotin needed for saturationof all biotin-binding sites. The functional concentra-tion of (strept)avidin was de¢ned as [biotin3bindingsites]/4. Stock solutions with s 300 nM functional(strept)avidin were stable for several hours at RTbut more dilute stock solutions had to be used im-mediately to avoid loss of biotin-binding sites.

The stock solutions of biotin-4-FITC and biotin-4-£uorescein were prepared in DMSO (V0.3 to V0.6mM estimated concentration by weight) and storedat 370³C. These primary stock solutions were di-luted with 19 volumes of bu¡er A to give the work-

ing reagents (V15 to V30 WM estimated concentra-tion by weight) for the typical cumulative titrationexperiments (see Section 2.4). The nominal concen-tration of biotin-4-FITC and biotin-4-£uorescein inthese working reagents was calculated from their ab-sorbance at 495 nm, using molar extinction coe¤-cients of 71 000 M31 cm31 and 68 000 M31 cm31

[29]. The e¡ective concentration of biotin residuesin these working reagents was then determined bytitration of known avidin samples with these workingreagents (see Fig. 3). Aliquots of the working re-agents were stored at 325³C.

2.4. Cumulative titration of avidin and streptavidinwith biotin-4-£uorescein

Typically, a 2-ml sample of 2^100 nM (strept)avi-din in bu¡er A (or 2 ml of homogenized egg whitethat had been diluted with 9 volumes of bu¡er A)was stirred in a cuvette at 25³C and 1^10 Wl aliquotsof biotin^£uorophore conjugate (typically V16 WMe¡ective concentration) were successively added froma Hamilton syringe at constant time intervals (asstated). Unless stated otherwise, biotin-4-£uoresceinwas excited at 490 nm (5 nm slit) and the £uores-cence was monitored at 525 nm (5 nm slit for stand-ard experiments). Biotin-4-FITC was excited at 485nm (5 nm slit) and observed at 525 nm (5 nm slit).All measurements were performed on a standard £u-orimeter (Shimadzu RF-540). All £uorescence inten-sities were corrected for sample dilution to improvethe comparison between related series in the ¢gures.

2.5. Non-cumulative titration of avidin andstreptavidin at concentrations below 2 nM

Polystyrene tubes (4 ml) were ¢lled with bu¡er Acontaining 0.2 mg/ml of puri¢ed BSA, incubatedovernight, and rinsed with bu¡er A (two times).For each titration curve, 16 BSA-coated tubes wereprepared in which 1 ml of bu¡er A was mixed with6 100 Wl of a 5 to 10 nM (strept)avidin stock solu-tion (the accurate concentration of which was simul-taneously determined by cumulative titration with 30WM biotin-4-£uorescein as described above). Tothese (stept)avidin samples up to 75 Wl of 50 nMbiotin-4-£uorescein were added (freshly prepared bydilution of standardized 30 WM biotin-4-£uorescein

BBAGEN 24764 4-3-99


with bu¡er A) to give the ¢nal ligand/acceptor ratiosshown in Fig. 7. After the indicated incubation timesat 25³C, £uorescence was measured and corrected forthe small dilution factors as described above.

3. Results

3.1. Synthesis of biotin-4-£uorescein andbiotin-4-FITC

As argued in the Introduction, the ideal £uorescentprobe for (strept)avidin should be on the one handshort enough to be restricted to a single binding site(see Fig. 1C,D). On the other hand the spacer be-tween biotin and £uorescein should be long enoughto allow for an unperturbed ¢t of the biotin residuein its binding site. In this study, ethylene diamine waschosen as a compromise (Fig. 2), and the total linkerarm between biotin and £uorophore was longer byone heteroatom in the case of biotin-4-FITC.

The synthetic route to biotin-4-£uorescein and bio-tin-4-FITC followed the same general scheme used tosynthesize biotin^PEG^£uorophore conjugates withhigh purity [23,24,30]. The key intermediate biotin^NH^CH2^CH2^NH2 was prepared ¢rst and thedesired £uorophore was attached in the last step(Fig. 2). Any other amine-reactive marker functioncan be attached with great ease in the same way.Thus the route was more versatile than the reportedsynthesis of biotin-4-FITC [31] where FITC was ¢rstcoupled to ethylene diamine before biotin was added.

One advantageous peculiarity of the ethylenediamine spacer should be mentioned: when onemole of Boc2O was added to one mole of ethylenediamine the desired asymmetric product Boc^NH^CH2^CH2^NH2 was obtained in nearly 100% yieldand the traces of diamine and di-Boc derivative wereeasy to remove.

3.2. Assay principle and calibration

Both biotin-4-FITC and biotin-4-£uorescein ful-¢lled the basic requirements, i.e., stoichiometric bind-ing of four ligands per avidin tetramer, and concom-itant £uorescence quenching (see Fig. 3). When 80pmol of avidin (320 pmol of biotin-binding sites ac-cording to standardization with D-biotin, see Section2.3) were titrated with nominally 22.5 WM biotin-4-FITC (Fig. 3A, circles) the breakpoint at 13.5 Wlindicated speci¢c binding of nominally 304 pmol ofbiotin-4-FITC. The nominal number of 304 pmolobviously corresponded to an e¡ective number of320 pmol of ligands and sites, the minor discrepancy

Fig. 3. Titration of the biotin-binding sites in avidin with bio-tin-4-FITC and biotin-4-£uorescein. (A) Avidin (80 pmol, 2 ml)was titrated with nominally 22.5 WM biotin-4-FITC by cumula-tive additions at 1-min intervals (a). In a parallel control series,avidin had been presaturated with 8 nmol of D-biotin (E). Thebreakpoint at 13.5 Wl indicated an e¡ective ligand concentrationof 23.7 WM in the biotin-4-FITC stock solution. (B) Avidin(160 pmol, 2 ml) was titrated with nominally 16 WM biotin-4-£uorescein by cumulative additions at 1-min intervals (a). Inparallel control experiments, avidin was either omitted (U) orpresaturated with 1.6 nmol of D-biotin before titrating with bio-tin-4-£uorescein (E). The breakpoint at 42.5 Wl indicated an ef-fective ligand concentration of 15.1 WM in the biotin-4-£uores-cein stock solution.

BBAGEN 24764 4-3-99


being due to some uncertainty in the molar extinc-tion coe¤cient [12]. Thus the experiment shown inFig. 3A served to determine the e¡ective concentra-tion of the biotin-4-FITC stock solution (320 pmol/13.5 Wl = 23.7 WM). Absence of non-speci¢c bindingwas indicated by the steep linear £uorescence in-crease beyond 13.5 Wl (Fig. 3A, circles) which wasstrictly parallel to the control pro¢le obtained withbiotin-saturated avidin (Fig. 3A, squares).

In parallel, 160 pmol of avidin (640 pmol of bio-tin-binding sites) were titrated with nominally 16 WMbiotin-4-£uorescein (Fig. 3B, circles). The breakpointat 42.5 Wl indicated speci¢c binding of nominally 680pmol of biotin-4-£uorescein. In this case the nominalconcentration was a 6% overestimate of the e¡ectiveconcentration (640 pmol/42.5 Wl = 15.1 WM) in thebiotin-4-£uorescein stock solution. The minor incor-rectness of the nominal concentration was again dueto the uncertainty about the true molar extinctioncoe¤cient of the biotin^£uorescein conjugate. Ab-sence of non-speci¢c binding was evidenced by thesteep £uorescence rise with excess of ligand (Fig. 3B,circles) which was parallel to the control pro¢le (Fig.3B, squares).

The e¡ective ligand concentrations in DMSOstock solutions of biotin-4-FITC and biotin-4-£uo-rescein remained stable during 1 year of storage at370³C. Aqueous stock solutions were stable for oneday at RT and for at least one week at 325³C, afterlonger storage under these conditions the e¡ectiveligand concentrations were seen to decrease slowly,and these stock solutions were re-calibrated by titra-tion of known avidin samples as shown in Fig. 3.Analogous results as in Fig. 3 were also obtainedwhen streptavidin was titrated with biotin-4-FITCand biotin-4-£uorescein (not shown).

The extent of ligand £uorescence quenching byavidin and streptavidin is summarized in Table 1.The newly synthesized biotin-4-£uorescein emergedas the most responsive £uorescent probe for bothavidin and streptavidin. The high percentage of lig-and £uorescence quenching means that the break-points in the titration curves were much more pro-nounced with biotin-4-£uorescein (e.g., Fig. 3B) thanwith biotin-4-FITC (e.g., Fig. 3A) or with `£uores-cein^biotin' [12].

Measurement of avidin in a very impure sample isshown in Fig. 4. Egg white from three eggs wascombined, homogenized, diluted with bu¡er, and ti-trated with a standardized stock solution of biotin-4-£uorescein. The breakpoint at 424 pmol of ligandindicated a concentration of 53 nM functional avidinin undiluted egg white, corresponding to 35.4 Wg avi-din/ml egg white. Thus the newly synthesized biotin-4-£uorescein appeared as a good substitute for com-mercial `£uorescein^biotin' [12] to measure avidin incrude, colored bio£uids.

3.3. Critical test at low concentrations of avidin andstreptavidin

Subsequently biotin-4-£uorescein was tested incomparison to the disadvantages of commercial `£u-orescein^biotin', i.e., inconveniently slow association

Table 1Fluorescence quenching of various biotin^£uorescein conjugatesin 4:1 complexes with avidin and streptavidin

Ligand Extent of £uorescence quenchinga by

avidin streptavidin

Biotin-4-£uorescein 84% 88%Biotin-4-FITC 70% 82%`Fluorescein^biotin'b) 84% 77%aFluorescence in presence of (strept)avidin was compared withabsence of (strept)avidin.bTaken from Fig. 1C and Fig. 5C in [12].

Fig. 4. Titration of homogenized egg white (2 ml of a 10-folddilution in bu¡er A) with biotin-4-£uorescein (10.0 WM e¡ectiveligand concentration) at 1-min intervals.

BBAGEN 24764 4-3-99


kinetics at 6 40 nM avidin and very slow, biphasicassociation at 6 100 nM streptavidin [12]. From Fig.5 it can be seen that even at 4 nM avidin (diamonds)the association kinetics of biotin-4-£uorescein wasfast enough to perform a cumulative titration.What is more important, the new biotin-4-£uoresceinexhibited the same kinetics of binding towards strept-avidin (Fig. 6) as towards avidin (Fig. 5), in sharpcontrast to commercial `£uorescein^biotin' [12].

At 9 2 nM avidin (Fig. 5, plus-signs) and strept-avidin (Fig. 6, plus-signs) the association kinetics wastoo slow to reach equilibrium within reasonable timeintervals in a cumulative titration. Nevertheless, evensubnanomolar concentrations of avidin and strept-avidin could be measured by non-cumulative titration(parallel incubation in 16 BSA-coated tubes) when

allowing 2 h for equilibration (Fig. 7). The lowersensitivity limit was found to be 0.2 nM, failure at0.1 nM (strept)avidin was related to sensitivity prob-lems of the £uorimeter.

In the course of the accompanying study [38] itwas found that inclusion of 0.1 mg/ml BSA in theassay bu¡er does not perturb the assay while elimi-nating the need for cumbersome overnight precoat-ing.

4. Discussion

4.1. Merits of the new ligand biotin-4-£uorescein

The above results demonstrated that biotin-4-£uo-rescein (Fig. 2) is an optimal speci¢c probe for thequantitative titration of biotin-binding sites in avidinand streptavidin: quenching in the bound state was

Fig. 5. Cumulative titration of known avidin samples (2 ml)with prestandardized stock solutions of biotin-4-£uorescein. Thefunctional concentrations of avidin and the time intervals be-tween consecutive ligand additions were as indicated in the ¢g-ure. The known e¡ective concentrations of the biotin-4-£uores-cein stock solutions were 5.6 WM (a,O), 705 nM (P), and 800nM (W,+). The £uorescence intensities were divided by the cor-responding avidin concentration and multiplied by the highestavidin concentration to facilitate the comparison within thegraph. The titration pro¢les for V34 nM avidin were verticallydisplaced to avoid overlap of data.

Fig. 6. Cumulative titration of known streptavidin samples(2 ml) with prestandardized biotin-4-£uorescein. The functionalconcentration of streptavidin and the time intervals betweenconsecutive ligand additions were as indicated in the ¢gure. Theknown e¡ective concentrations of the biotin-4-£uorescein stocksolutions were 5.6 Wm (a,O,P), 800 nM (W,+), and 200 nM(U). Fluorescence data were re-normalized and plotted as de-scribed in the legend to Fig. 5.

BBAGEN 24764 4-3-99


most pronounced (Table 1), the a¤nity was highenough to achieve stoichiometric ligand bindingeven at 0.2 nM (strept)avidin (Fig. 7), and the asso-ciation kinetics was fast enough for rapid binding atnanomolar concentrations (6 1 min at 8 nM (strept)-

avidin, see Figs. 5 and 6). In fact, no other biotin-£uorophore conjugate (except for the close analoguebiotin-4-FITC, Fig. 2) has shown comparable a¤nityand association rate [24].

Particularly interesting was the contrast betweenbiotin-4-£uorescein and commercial `£uorescein^bio-tin' (Fig. 2): The a¤nity of `£uorescein^biotin' wasweaker by one order of magnitude and associationkinetics slower by several orders of magnitude whichsigni¢cantly limited our previous (strept)avidin assay[12].

The model explanation (Fig. 1) for the strikingdi¡erence between biotin-4-£uorescein and `£uores-cein^biotin' focuses on the only signi¢cant structuraldi¡erence between these two, i.e. on spacer length(see Fig. 2). All relevant structural [15^17] and func-tional [13,14] features have been incorporated intothis model, including the well-known weak bindingof £uorescein residues to empty biotin binding siteswhich is accompanied by exceptional £uorescencequenching [12,18^22].

The model has already been validated before [12]by explaining the strictly anti-cooperative binding of`£uorescein^biotin' to streptavidin where bivalentbinding (Fig. 1B, left side) preceded (and greatlyhampered) subsequent occupation of each bindingsite by one `£uorescein^biotin' molecule (Fig. 1B,right side). The model has also explained the lesspronounced anti-cooperative binding of `£uores-cein^biotin' to avidin (Fig. 1A) [12].

The same concept is able to rationalize the rapidsimultaneous association of four biotin-4-£uoresceinmolecules to tetrameric avidin (Fig. 1C) and strept-avidin (Fig. 1D) observed in this study (Figs. 5^7). Inother words, biotin-4-£uorescein and its close struc-tural relative biotin-4-FITC (Fig. 2) are the ¢rst bio-tin^£uorophore conjugates which can mimic D-biotinin terms of high a¤nity, rapid association, and non-cooperative binding. The only biotin-marker conju-gates to share these properties are small 125I-labeledhistamine or tyramine derivatives used in radioligandbinding assays [3]. All other biotin^£uorophore con-jugates [12,23,24] and biotinoyl-peptides [32^37] wereseen to bind in an anti-cooperative manner, i.e., thethird and the fourth ligand were associating moreslowly and dissociating much faster than the ¢rsttwo ligands per (strept)avidin tetramer.

In conclusion, it is obvious that the unique binding

Fig. 7. Non-cumulative titration of known avidin (A) andstreptavidin (B) samples at subnanomolar concentrations. Ineach series, 16 BSA-coated test tubes containing a constant(strept)avidin concentration (indicated in the ¢gure) and a vari-able known ligand concentration were incubated for the timeintervals indicated in the ¢gure. Fluorescence was measured at485 nm excitation (5 nm slit) and at 525 nm emission wave-length (10 nm slit). The known e¡ective concentration in thebiotin-4-£uorescein stock solution was 50 nM in all experi-ments. Titrations at 200 nM (strept)avidin were done in inde-pendent triplicates to demonstrate the reproducibility of stand-ardization and dilution steps. Fluorescence data were re-normalized and plotted as described in the legend to Fig. 5.

BBAGEN 24764 4-3-99


properties of the new ligand biotin-4-£uorescein mustbe attributed to its unusually short spacer (Fig.1C,D), whereas the anti-cooperative binding ofmost other biotin derivatives is due to long spacers[12,23,24] and/or bulky peptide residues [14].

4.2. Merits of the new £uorimetric (strept)avidinassay

Due to the fortunate properties of biotin-4-£uores-cein, the new (strept)avidin assay is 10-times moresensitive and much faster than the assay whichemployed commercial `£uorescein^biotin' [12]. Inaddition, the previous titration with commercial lig-and required widely di¡erent procedures for avidinas opposed to streptavidin, and a good understand-ing on the molecular level (see Fig. 1A,B). The newassay with biotin-4-£uorescein, however, is based onthe simple concept that each ligand exclusively inter-acts with a single binding site (Fig. 1C,D), and identi-cal protocols can be used for avidin as well as strept-avidin (see Figs. 5^7). Furthermore, the new biotin-4-£uorescein method appears advantageous over anyother known (strept)avidin assay, except when 6 200pM (strept)avidin is to be measured or when parallelprocessing of large sample numbers is required.

Sensitivity is second to the most advanced radio-ligand method only [3]. The sensitivity limit of thenon-cumulative £uorescence titration assay (0.2 nM,see Fig. 7) and of the rapid cumulative protocol (4nM, see Figs. 5 and 6) are quite similar to the lowerlimits of the best enzymatic method [5] when apply-ing 20 h or 2 h enzyme incubations, respectively,while the £uorescence method has additional advan-tages (see below).

Previously, radioligand and enzymatic assays werethe only methods for speci¢c measurement of (strept)-avidin in very impure, colored samples. The ¢rst £u-orimetric assay to break this `monopoly' was our pre-vious titration method with commercial `£uorescein^biotin' [12]. In the present study the same goal wasachieved by the same measuring principle: the (strep-t)avidin concentration was calculated from the con-sumption of £uorescent ligand up to the distinctbreakpoint in the £uorescence titration pro¢le ^ andnot from the actual magnitude of any £uorescencesignal. This explains the successful application tocrude avidin samples, such as egg white (see Fig. 4).

The disadvantage of subjecting each unknown to acomplete titration experiment is limitation to smallsample numbers. For this reason a new `single tubeassay' with biotin-4-£uorescein has been developed inwhich the extent of ligand quenching is correlatedwith (strept)avidin concentration [38]. The `singletube assay' is ideal for large sample numbers but itis much less robust and reliable for crude, coloredsamples and a complete calibration curve is requiredfor each £uorimeter session.

In contrast to the single tube £uorescence assayand to radioligand and enzyme assays only one cal-ibration per week (or per year) is required for the£uorescence titration method if standardized stocksolutions of biotin-4-£uorescein are stored at325³C (or at 370³C, respectively).

Getting started for an individual £uorescence titra-tion experiment only requires warming up the £uo-rimeter and thawing of an aliquot of the standar-dized biotin-4-£uorescein stock solution. Thus, inpractice, the presently proposed £uorescence titrationassay is the fastest method to measure (strept)avidinconcentrations on short notice, without compromiseconcerning sensitivity, accuracy, and reliability incrude biosamples.

Acknowledgements

We are grateful to Prof. Norbert Mu«ller and Mar-tin Emsenhuber for measurement of NMR spectraand to D.I. Werner Ahrer for electrospray MS ex-periments. This work was supported by the AustrianResearch Funds (project P12097-PHY).

References

[1] M. Wilchek, E.A. Bayer, Methods Enzymol. 184 (1990) 5^13.


[3] E.V. Groman, J.M. Rothenberg, E.A. Bayer, M. Wilchek,Methods Enzymol. 184 (1990) 208^217.

[4] D.M. Mock, D.B. DuBois, Anal. Biochem. 153 (1986) 272^278.

[5] E.A. Bayer, H. Ben-Hur, M. Wilchek, Methods Enzymol.184 (1990) 217^223.

[6] R.S. Niedbala, F. Gergits III, K.J. Schray, J. Biochem. Bio-phys. Methods 13 (1986) 205^210.

BBAGEN 24764 4-3-99


[7] M.S. Barbarakis, W.G. Qaisi, S. Daunert, L.G. Bachas,Anal. Chem. 65 (1993) 457^460.

[8] N.M. Green, Methods Enzymol. 18A (1970) 418^424.[9] H.J. Lin, J.F. Kirsch, Methods Enzymol. 62 (1979) 287^289.

[10] G.P. Kurzban, G. Gitlin, E.A. Bayer, M. Wilchek, P.M.Horowitz, J. Protein Chem. 9 (1990) 673^682.

[11] D.M. Mock, P. Horowitz, Methods Enzymol. 184 (1990)234^240.

[12] H.J. Gruber, G. Kada, M. Marek, K. Kaiser, Biochim. Bio-phys. Acta 1381 (1998) 203^212.

[13] N.M. Green, Adv. Protein Chem. 29 (1975) 85^133.[14] N.M. Green, Methods Enzymol. 184 (1990) 51^67.[15] W.A. Hendrickson, A. Pa«hler, J.L. Smith, Y. Satow, E.A.

Merritt, R.P. Phizackerley, Proc. Natl. Acad. Sci. U.S.A. 86(1989) 2190^2194.

[16] O. Livnah, E.A. Bayer, M. Wilchek, J.L. Sussman, Proc.Natl. Acad. Sci. U.S.A. 90 (1993) 5076^5080.

[17] L. Pugliese, A. Coda, M. Malcovati, M. Bolognesi, J. Mol.Biol. 231 (1993) 698^710.

[18] M.S. Barbarakis, T. Smith-Palmer, L.G. Bachas, Talanta 40(1993) 1139^1145.

[19] N.G. Hentz, L.G. Bachas, Anal. Chem. 67 (1995) 1014^1018.

[20] M.H.H. Al-Hakiem, J. Landon, D.S. Smith, R.D. Nargessi,Anal. Biochem. 116 (1981) 264^267.

[21] R.D. Nargessi, D.S. Smith, Methods Enzymol. 122 (1986)67^72.

[22] S.V. Rao, K.W. Anderson, L.G. Bachas, Bioconjug. Chem.8 (1997) 94^98.

[23] H.J. Gruber, M. Marek, H. Schindler, K. Kaiser, Bioconjug.Chem. 8 (1997) 552^559.

[24] M. Marek, K. Kaiser, H.J. Gruber, Bioconjug. Chem. 8(1997) 560^566.


[26] F.M.A.H. Schuurmans-Stekhoven, G.I. Tesser, G. Ram-steyn, H.G.P. Swarts, J.J.H.H.M. De Pont, Biochim. Bio-phys. Acta 1109 (1992) 17^32.

[27] D.B. McCormick, J.A. Roth, Methods Enzymol. 18 (1970)383^385.

[28] R. Feldho¡, Techn. Protein Chem. 17 (1991) 55^63.[29] R.P. Haugland, Handbook of Fluorescent Probes and Re-

search Chemicals, 1992. Molecular Probes, Inc.[30] K. Kaiser, M. Marek, T. Haselgru«bler, H. Schindler, H.J.

Gruber, Bioconjug. Chem. 8 (1997) 545^551.[31] K.J. Schray, P.G. Artz, R.C. Hevey, Anal. Chem. 60 (1988)

853^855.[32] F.M. Finn, K.H. Hofmann, Methods Enzymol. 109 (1985)

418^445.[33] F.M. Finn, G. Titus, J.A. Montibeller, K. Hofmann, J. Biol.

Chem. 255 (1980) 5742^5746.[34] F.M. Finn, G. Titus, K. Hofmann, Biochemistry 23 (1984)

2554^2558.[35] K. Hofmann, G. Titus, J.A. Montibeller, F.M. Finn, Bio-

chemistry 21 (1982) 978^984.[36] S. Lavielle, G. Chassaing, A. Marquet, Biochim. Biophys.

Acta 759 (1983) 270^277.[37] H. Romovacek, F.M. Finn, K. Hofmann, Biochemistry 22

(1983) 904^909.[38] G. Kada, K. Kaiser, H. Falk, H.J. Gruber, Biochim. Bio-

phys. Acta 1427 (1999) 44^48 (this issue).

BBAGEN 24764 4-3-99


Accurate measurement of avidin and streptavidin in crude biofluids with a new, optimized biotin–fluorescein conjugate

Documents