Accurate titration of avidin and streptavidin with biotin–fluorophore conjugates in complex, colored biofluids

Accurate titration of avidin and streptavidin with biotin–fluorophore conjugates in complex, colored biofluids

Biochimica et Biophysica Acta 1381 Ž1998. 203–212 Accurate titration of avidin and streptavidin with biotin–fluorophore conjugates in complex, colore...

275KB Sizes 0 Downloads 30 Views

Biochimica et Biophysica Acta 1381 Ž1998. 203–212

Accurate titration of avidin and streptavidin with biotin–fluorophore conjugates in complex, colored biofluids Hermann J. Gruber ) , Gerald Kada, Markus Marek, Karl Kaiser Institute of Biophysics, J. Kepler UniÕersity, Altenberger Str. 69, A-4040 Linz, Austria Received 17 November 1997; revised 27 January 1998; accepted 19 February 1998

Abstract A new fluorimetric assay is presented for the specific and reliable quantitation of G 2 nM avidin and streptavidin. The assay is based on pronounced changes in the fluorescence properties of commercial fluorescein–biotin, or of a newly synthesized biotin–polyŽethylene glycol. –pyrene conjugate, which occur upon binding to avidin and streptavidin. Accurate measurement of Žstrept.avidin in complex, colored biofluids, such as crude egg white or serum relies on a simple titration protocol. Only occasional recalibration of the reagent solution is required. Due to these merits the proposed assay is particularly suited for rapid measurement of few samples on short notice, for functional control of Žstrept.avidin-containing reagents after storage, and for the monitoring of Žstrept.avidin concentrations in large scale processes. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Avidin; Biotin; Fluorescein; Fluorescence; Streptavidin

1. Introduction Avidin and streptavidin are indispensable tools in bioscience and technology w1,2x, therefore quantitative measurement of functional Žstrept.avidin is a routine task. Assays can be grouped into three categories: radioligand binding w3,4x, enzyme assays w5–

Abbreviations: ANS, 2-anilinonaphthalene-6-sulfonic acid; Biotin–PEG 800 –pyrene, N-biotinoyl-N X-Ž1-pyrene.butanoylO,OX-bisŽ2-aminopropyl.-polyŽethylene glycol. 800 ; EDTA, ethylenediamine-N, N, N X , N X-tetraacetic acid; Fluorescein–biotin, 5 w N-Ž5- N-w6- Žbiotinoyl. aminox hexanoyl4 amino. pentylx -thioureidyl4fluorescein; PEG, polyŽethylene glycol.; ŽStrept.avidin, streptavidinror avidin ) Corresponding author. Fax: q43-732-2468-822; E-mail: [email protected]

7x, and photometricrfluorimetric methods w8–11x. The most advanced radioligand method w3x can determine 3–10 fmolr1 ml sample volume. For avoidance of radioisotopes, a solid phase enzyme assay has been developed w5x which is far more sensitive Ž 15–150 fmolr0.1 ml sample volume., general, and reliable than any of the homogeneous enzyme inhibition protocols Ž for a critical revision, see Refs. w6,7x.. Radioactive and enzyme assays are well suited for parallel processing of large sample numbers. Their disadvantages are Ži. need for expensive andror noncommercial components with limited lifetime, such as 125 I-biotinoyl-His or microplates coated with biotinoylated bovine serum albumin, Žii. long assay times between 2–20 h, Ž iii. acquaintance with heterogeneous assay protocols, and Ž iv. the need for simultaneous calibration, all of which represent significant

0304-4165r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 3 0 4 - 4 1 6 5 Ž 9 8 . 0 0 0 2 9 - 4


H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

burdens when small sample numbers are to be assayed at irregular intervals. In principle, fluorimetric ligand binding assays are expected to combine sensitivity with convenience because usually protocols are simple, assay times are short, and standardized reagents can be stored frozen. Known fluorescence assays for functional Žstrept.avidin, however, are not at all competitive with radioligand and enzyme assays. The ANS method w11x is almost as insensitive as the spectrophotometric HABA assay w8x which can measure G 0.5 m M avidin, and both assays are not applicable to streptavidin and succinoyl–avidin. Quenching of endogeneous Trp fluorescence upon binding of D-biotin can be used to measure G 20 nM avidin w9x but for obvious reasons this method is limited to absence of other proteins or fluorophores. The most sensitive fluorescence assay presented so far relies on the 100–300% fluorescence increase in fluoresceinlabeled avidin w12x or streptavidin w13x which occurs upon binding of D-biotin, and which can be antagonized by unlabeled avidin w14,15x. While this effect has been optimized for specific detection of D-biotin and small biotin derivatives w16x no protocol for the reliable measurement of Žstrept.avidin in complex mixture has been developed so far. Unexpectedly, biotin–fluorophore conjugates have not yet been used as specific probes for Ž strept. avidin except for a fluorescence polarization method which is at best semiquantitative w17x. This lack must be attributed to the absence of basic knowledge on the thermodynamic and spectroscopic properties of the complexes formed by Žstrept. avidin and fluorescent biotins. In three recent studies, however, eleven biotin–fluorophore conjugates have been characterized with respect to their usefulness in fluorescence microscopy w18–20x. Fluorescein–biotin and biotin–PEG 800 – pyrene turned out to be the weakest fluorescence markers for immobilized Ž strept. avidin because in the bound state their fluorescence was dramatically quenched. On the other hand, it was obvious that these two fluorescent biotin derivatives should make for excellent bioanalytical probes to measure Žstrept.avidin concentration in bulk solution by the very same principle. In the present study, appropriate assay protocols for avidin and streptavidin were optimized for sensi-

tivity, accuracy, and convenience, yielding fluorescence assays which are nearly as sensitive, equally reliable, yet more convenient and rapid than radioligand and enzyme assays.

2. Materials and methods 2.1. Reagents and buffers P.a. grade materials were used as far as commercially available. Affinity-purified avidin, streptavidin, and D-biotin were obtained from Sigma Ž Munich, Germany. . ANS and fluorescein–biotin were purchased from Molecular Probes ŽLeiden, Netherlands. . All other materials were obtained from Merck ŽDarmstadt, Germany.. Biotin–PEG800 –pyrene was synthesized as described elsewhere w19x. Pig serum was prepared by a standard procedure w21x. Buffer A contained 100 mM NaCl, 50 mM NaH 2 PO4 , 1 mM EDTA, and the pH was adjusted to 7.5 with NaOH. 2.2. Standardization of stock solutions The primary standard was a 400-m M D-biotin stock solution prepared by dissolving 99% pure D-biotin in buffer A on a large scale. Aliquots were stored at y258C. Avidin and streptavidin were dissolved in buffer A at 2–5 m M Ž nominal concentration by weight. and stored at 48C for up to 1 week or at y258C for up to 1 month without re-freezing. The concentration of functional avidin Žs biotin binding sitesr4. was determined by titration with D-biotin w9x as shown in Fig. 1A. Stock solutions of fluorescein– biotin and biotin–PEG800 –pyrene were standardized as described in Section 3. 2.3. CumulatiÕe titration of aÕidin and streptaÕidin with fluorescent biotin deriÕatiÕes Typically, 2 ml of 50–100 nM Žstrept.avidin in buffer A Ž or in a mixture of buffer A with homogenized egg white or with pig serum. were stirred in a cuvette at 258C and 1–5 m l aliquots of biotin–fluorophore conjugate ŽF 16 m M effectiÕe concentration. were successively added from a Hamilton syringe at constant time intervals Žas stated.. Fluorescein–biotin was excited at 485 nm Ž5-nm slit. and the fluores-

H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212


cence was monitored at 525 nm Ž5-nm slit for standard experiments.. Biotin–PEG800 –pyrene was excited at 345 nm Ž 5-nm slit. and excitation spectra were recorded from 350 nm to 550 nm Ž 10-nm slit. for the simultaneous observation of pyrene monomer and pyrene excimer fluorescence. All measurements were performed on a standard fluorimeter Ž Shimadzu RF-540..

3. Results 3.1. Assay principles The well-known fluorescence assay for functional avidin w9x is shown in Fig. 1A. Hereby quenching of intrinsic Trp fluorescence serves to monitor binding of unmodified D-biotin. The amount of biotin-binding sites in the cuvette can directly be calculated from the consumption of D-biotin stock solution up to the breakpoint in the titration profile. However, this method is limited to near absence of other proteins which would cause a high Trp fluorescence background. Moreover, at - 20 nM avidin the Trp fluorescence is too weak for a standard fluorimeter. No such limitations are to be expected when using intensely fluorescent biotin derivatives for the monitoring of acceptor–ligand interaction, the new requirement being a distinct fluorescence change upon ligand binding. Among all known biotin–fluorophore conjugates, commercial fluorescein–biotin and a new biotin– PEG800 –pyrene conjugate exhibit the most dramatic losses in fluorescence after binding to avidin and streptavidin w18,19x. This is demonstrated in Figs. 1B and 2A, respectively. In the absence of avidin, fluorescence was linearly dependent on the concentration of the biotin–fluorophore conjugates Ž dotted lines in Figs. 1B and 2A. , and the same unperturbed fluorescence intensities were observed when biotin–saturated avidin was present Ž triangles in Fig. 1B. . In the presence of unblocked avidin, however, probe fluorescence was much lower, maximal quenching occurring at saturation of the tetravalent acceptor protein with four ligands Ž circles in Figs. 1B and 2A.. The steep linear rise at ) 4 ligandsravidin tetramer obviously reflected the high intensity of unbound excess ligands Žsee Section 4..

Fig. 1. Titration of the biotin-binding sites in avidin with D-biotin or fluorescein–biotin. ŽA. Avidin Ž ;100 nM, 2 ml. was titrated with a 16 m M D-biotin standard by cumulative additions at 1 min-intervals. ŽB and C. A 2-ml sample from the same avidin stock solution was titrated with a fluorescein–biotin stock solution by cumulative additions at 1-min intervals Žcircles, solid lines.. Please note that ŽB. and ŽC. are alternate plots of the same experiment. The titration curve in ŽB. refers to an effectiÕe concentration of 12.6 m M while the titration profile in ŽC. refers to a nominal concentration of 16 m M in the fluorescein–biotin stock solution. In parallel control experiments avidin was either omitted Ždotted lines in B and C. or pre-saturated with an 80-fold excess of D-biotin before titrating with fluorescein–biotin Žtriangles in B and C..

Titration profiles with fluorescein–biotin ŽFig. 1B. were more pronounced than with biotin–PEG800 – pyrene ŽFig. 2A.. However, biotin–PEG800 –pyrene offered a unique second fluorescence parameter by which binding to avidin could independently but simultaneously be monitored: In the presence of avidin Žor streptavidin, see Fig. 5B. a new fluores-


H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

at the expected ratio of four ligandsravidin tetramer. Absence of nonspecific binding was again evidenced by the nearly constant signal at higher ligandravidin ratios, the small slope being equal to that of the corresponding control in the absence of avidin Ždotted line in Fig. 2B. . When dividing the excimer profile Ž circles in Fig. 2B. by the normal fluorescence profile Ž circles in Fig. 2A. we necessarily obtained a much more pronounced ‘synoptic titration curve’, with a steep rise for up to four ligandsravidin and a steep fall at higher ratios Žtrace a in Fig. 2C. . Accentuation could further be enhanced by plotting higher powers of the new function Žtraces b, c, and d in Fig. 2C.. It must be emphasized that the ‘synoptic titration profile’ Žtrace a in Fig. 2C. represents a ratio of measured data points—not of the fit lines in Fig. 2A and B. The peak maximum in Fig. 2C unequivocally defined stoichiometric binding of biotin–PEG800 –pyrene to all biotin-binding sites of the avidin-containing sample. 3.2. Standardization of fluorescent biotin stock solutions

Fig. 2. Titration of the biotin-binding sites in avidin with biotin– PEG800 –pyrene. Avidin Ž2 ml 80 nM. was titrated with 16 m M biotin–PEG800 –pyrene Ž effectiÕe concentration. by cumulative additions Ž2-min intervals. while monitoring pyrene monomer fluorescence at 390 nm ŽA. and pyrene excimer fluorescence at 480 nm ŽB.. Both data profiles Žnot the fitted lines. were combined in panel ŽC. by plotting the ratio of 480 nmr390 nm fluorescence with the power n Ž ns1, 2, 4, and 8 in profiles a, b, c, and d, respectively.. All curves in ŽC. were normalized for better comparison in a single plot. In a parallel control experiment Ždotted lines in ŽA. and in ŽB.. 2 ml of buffer without avidin was also titrated with biotin–PEG800 –pyrene Žonly the fits through the data points are shown..

cence signal appeared at 480 nm which is known to originate from excited state dimer Ž ‘excimer’. formation w19,22x. Obviously the binding of two, three, or four biotin termini to one avidin tetramer strongly favored association of the pyrene labels at the outer ends of the avidin-bound PEG800 chains, giving rise to the progressive signal increase in Fig. 2B. As in Fig. 2A, the breakpoint indicated acceptor saturation

Concentrated stock solutions Ž; 0.5 mM. of fluorescein–biotin and biotin–PEG800 –pyrene were prepared in DMSO and diluted with buffer A to give a nominal concentration of 16 m M by UV-vis absorption, assuming ´ 495 s 71 000 My1 cmy1 at pH 9 for fluorescein–biotin and ´ 340 s 40 000 My1 cmy1 for biotin–PEG800 –pyrene w23x. The nominal concentration Ž dye content. is expected to equal the effectiÕe concentration Ž biotin derivative content. only if the biotin–fluorophore is chemically pure and if the estimate of the molar extinction coefficient is correct. This was obviously the case with the freshly prepared 16 m M stock solution of biotin–PEG800 –pyrene used in Fig. 2 since the same number of biotin-binding sites was determined in a parallel titration with 16 m M D-biotin Žnot shown. . The chemical purity of fluorescein–biotin, however, was ; 85% according to batch specifications, the impurity being fluorescent material without biotin residues. Consequently the nominal concentration Ž16 m M. was an overestimate of effectiÕe ligand concentration, as exemplified by the discrepancy between Fig. 1A Ž breakpoint at 770 pmol D-bio-

H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212


tin. and Fig. 1C Žcircles, parallel titration of an equal aliquot of avidin, breakpoint at nominally 980 pmol of fluorescein–biotin.. Thus the nominally 16 m M fluorescein–biotin stock solution had an effectiÕe ligand concentration of 12.5 m M. When the preliminary titration curve Ž Fig. 1C, circles. was replotted vs. effectiÕe ligand additions ŽFig. 1B, circles. the agreement with the unequivocal D-biotin titration curve Ž Fig. 1A. was restored. Aqueous stock solutions of fluorescein–biotin and biotin–PEG800 –pyrene were stable for at least 1 day at room temperature and for at least 1 week at y258C, thus frozen aliquots from a standardized stock solution could be used for 1 week without recalibration. On a longer time scale both fluorescent biotin derivatives showed slow decrease in the effective ligand concentration and standardization had to be repeated Žas shown in Fig. 1.. Less decay was observed in concentrated DMSO stock solutions during long term storage at y258C and at lower temperatures. 3.3. Measurement of aÕidin in complex biofluids and at low concentrations Even 50% reddish pig serum did not hamper quantitation of avidin with fluorescein–biotin because the same number of biotin-binding sites Ž520 pmol. was detected when replacing serum Žtriangles in Fig. 3A. by control buffer Žcircles in Fig. 3A. in a parallel titration. In this experiment the natural biotin content of serum could be ignored, typical concentrations ranging from 1 to 4 pmolrml of serum w24x. Biotin– PEG800 –pyrene was unsuitable for the measurement in serum since absolutely no pyrene fluorescence was detected when adding biotin–PEG800 –pyrene to serum, whether avidin was present or not Ž not shown. . In contrast, both fluorescein–biotin ŽFig. 3B. and biotin–PEG800 –pyrene ŽFig. 3C. could be used to measure avidin in crude chicken egg white. A total of 560 pmol or 570 pmol, respectively, of biotin-binding sites were determined in the 2-ml aliquots of 10-fold diluted egg white. The egg white samples had been drawn from a batch size of 90 chicken eggs. On average, one 57-g egg gave 36 ml of egg white containing 2.3 g of total protein Ž measured according to Bradford w25x, using BSA as standard protein. and 1.65 mg of functional avidin Ž according to Fig. 3B. .

Fig. 3. Measurement of avidin in serum and in egg white. ŽA. A total of 1 ml of pig serum Žtriangles. was mixed with 1 ml of buffer A containing 130 pmol avidin and titrated with fluorescein–biotin Ž12.5 m M effectiÕe concentration. at 1-min intervals. In a parallel control, 2 ml of buffer A containing 130 pmol avidin were used Žcircles, vertically displaced by q20 fluorescence units to avoid overlap of data.. ŽB. Homogenized chicken egg white Ž10-fold dilution in buffer A, 2 ml. was titrated with fluorescein–biotin Ž12.5 m M biotin termini. at 1-min intervals. ŽC. An equal egg white sample as in ŽB. was titrated with biotin–PEG800 –pyrene Ž14.6 m M effectiÕe concentration. at 2min intervals. The line in ŽC. represents a Gaussian fit through the data.

In practice it is possible to isolate 15–20 mg of pure avidin from 24 chicken eggs w26x. It is important to note that the simple cumulative titration protocol with fluorescein–biotin proved to be very impractical when measuring < 40 nM avidin because association kinetics was too slow. Such low avidin concentrations, however, were successfully measured with a noncumulative procedure, i.e., series of aliquots with the same avidin concentration and


H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

increasing ligand concentrations were incubated for 3.5 h at 258C before measuring fluorescence intensities. In this way correct titration profiles were obtained down to 2 nM avidin concentration Ž Fig. 4. . At 1 nM avidin concentration even longer incubation times Ž ; 8 h. were required to reach equilibrium, and below 1 nM avidin ligand binding was no longer stoichiometric because K D ; 0.2 nM for the 4:1 complex w19x. Measurement of avidin Ž or streptavidin, see below. with biotin–PEG800 –pyrene was found to be limited

to 40 nM Ž10 nM. acceptor protein concentration because K D ; 8 nM Ž2 nM. for the corresponding 4:1 complex w19x. Thus the lower detection limit for avidin and streptavidin was generally found at c min ; 5 K D , whether titration was performed with fluorescein–biotin or biotin–PEG800 –pyrene. Kinetic problems at low concentrations could always be overcome by appropriate incubation times. In no case signal intensities on a standard fluorimeter were the limiting parameter. 3.4. Adaptation to the measurement of streptaÕidin

Fig. 4. Noncumulative titration of dilute avidin samples with fluorescein–biotin. In each titration series, 15 test tubes containing 1125-m l sample volumes with a uniform avidin concentration and a variable ligand concentration were incubated at 258C for 3.5 h before fluorescence was measured at 485-nm excitation and 525-nm emission wavelength. For the most concentrated series, 750-m l aliquots of 94 nM avidin were mixed with 375 m l of buffer A Žto give equal final assay volumes. and F 350 m l of 1.6 m M fluorescein–biotin Ž effectiÕe concentration. was added while vortexing. Multipettes were used for all pipetting steps. For titration at any lower avidin concentration, the concentration of the corresponding ligand stock solution was also reduced by the same factor. Final avidin concentrations were 63 nM Žx., 31 nM Žq., 16 nM Ždiamonds., 7.8 nM Ždown triangles., 3.9 nM Žup triangles., 2.0 nM Žcircles., and 1.0 nM Žsquares.. For easier comparison of all titration profiles the signal intensities were divided by the corresponding avidin concentration and multiplied by 65 nM Žhighest avidin concentration..

The Trp fluorescence quenching assay Ž shown for avidin in Fig. 1A. is applicable to streptavidin as well w10x. As a consequence, potential users of fluorescein–biotin or biotin–PEG800 –pyrene do not depend upon avidin for the calibration procedure shown in Fig. 1 but can stick to streptavidin if desired. Such direct standardization of a pure streptavidin solution is shown in Fig. 5A. Titration of streptavidin with biotin–PEG800 – pyrene was equally straightforward Ž Fig. 5B. as in the case of avidin Ž Fig. 2C. . From the above results with avidin and biotin–PEG800 –pyrene it is inferred that this type of streptavidin assay will also work in crude biofluids, except for serum. Unexpectedly, titration curves of streptavidin with fluorescein–biotin Ž Fig. 5C. greatly differed from corresponding avidin titrations Ž Fig. 1B. with respect to shape and time dependence. When an equal 2-ml aliquot of 50 nM streptavidin Ž 100 pmol. as in Fig. 5A and B was titrated with fluorescein–biotin with a similar time protocol Ž40-s intervals. as used for avidin Ž1-min intervals. a single breakpoint between two linear segments was obtained at 211 pmol Ž circles in Fig. 5C.. This indicated binding of ; 2 fluorescein–biotin molecules per streptavidin tetramer. From visual inspection of a simultaneously recorded fluorescence time scan Žnot shown. it was obvious that in the range between 200–500 pmol ligand addition the system was far from equilibrium after each 40-s interval. Thus a parallel titration was performed with 10-min intervals between successive ligand additions Žsquares in Fig. 5C., resulting in almost perfect equilibration Ži.e., signal relaxation. after each ligand addition in all segments of the titration curve. As expected, two breakpoints between three linear seg-

H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212


instantaneously while the third and the fourth fluorescein–biotin exhibited very slow binding kinetics towards the streptavidin tetramer. The ‘slow titration’ profile Žsquares in Fig. 5C. was important for an understanding of the underlying molecular events Ž see Section 4. —but for practical purposes the ‘fast titration’ Žcircles in Fig. 5C. was much preferred because a clearer result Ž i.e., a single, more distinct breakpoint. was obtained in much shorter time. The reliability of this somewhat unusual nonequilibrium titration was further supported by detection of 100 pmol of bound ligand in a 2-ml aliquot of 25 nM streptavidin Ž triangles in Fig. 5C. which also correspond to two ligandsrstreptavidin

Fig. 5. Titration of the biotin-binding sites in streptavidin with D-biotin, biotin–PEG 800 –pyrene, or fluorescein–biotin. ŽA. Streptavidin Ž50 nM, 2 ml. was titrated with 8 m M D-biotin by cumulative additions at 1-min intervals. ŽB. Streptavidin Ž50 nM, 2 ml. was titrated with biotin–PEG800 –pyrene Ž14.6 m M effective concentration. at 2-min intervals. The line represents a Gaussian fit. ŽC. Streptavidin Ž50 nM, 2 ml. was titrated with fluorescein–biotin Ž12.5 m M effective concentration., either at 10-min intervals Žsquares. or at 40-s intervals Žcircles.. Alternatively, 25 nM streptavidin in buffer A Žtriangles. or in 98% pig serum Ždiamonds. was titrated with the same reagent at 1-min intervals. Circles, triangles, and diamonds were vertically displaced by q20, q40, and q60 fluorescence units, respectively, to avoid overlap of data.

ments were now observed, the first at 186 pmol corresponding to ; 2, and the second at 408 pmol corresponding to ; 4 bound ligandsrstreptavidin tetramer Ž please note the good agreement with Fig. 5A and B.. Obviously ligands number 1 and 2 bound

Fig. 6. Cumulative nonequilibrium titrations of streptavidin with fluorescein–biotin. Final streptavidin concentrations in the 2-ml samples were 68 nM Žq., 34 nM Ždiamonds., 17 nM Ždown triangles., 8.5 nM Žup triangles., 4.3 nM Žcircles., 2.1 nM Žopen squares., or again 2.1 nM Žclosed squares, 160-s time intervals.. The 2-ml samples were stirred in the cuvette and 10-m l aliquots of an appropriate fluorescein–biotin stock solution were normally added at 40-s intervals, unless stated otherwise. Titration profiles Žexcept for q. were serially displaced by 0.3 fluorescence units in order to avoid overlap of the data. For easier comparison of all titration profiles the signal intensities were divided by the corresponding streptavidin concentration and multiplied by 68 nM Žhighest streptavidin concentration..


H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

tetramer. Most importantly, the same result was also obtained with 25 nM streptavidin in 98% pig serum Ždiamonds in Fig. 5C, breakpoint at 95 pmol. , thus the reduced accuracy of the fluorescein–biotin method in the case of streptavidin is compensated by unperturbed reliability, even in highly pigmented biofluids. Moreover, ‘fast titration’ of only two fluorescein– biotin sites per streptavidin tetramer allowed to apply simple cumulative titration to samples with as little as 4 nM streptavidin ŽFig. 6. , in contrast to avidin Ž Fig. 4.. At 2 nM streptavidin the kinetics of 2:1 complex formation was too slow for equilibration within 40-s intervals ŽFig. 6, open squares. but titration at 160-s intervals gave correct data ŽFig. 6, solid squares..

4. Discussion 4.1. Assay performance Among a series of 11 biotin–fluorophore conjugates, fluorescein–biotin and biotin–PEG800 –pyrene showed the most pronounced response when binding to avidin and streptavidin w18,19x, thus they were promising candidates as specific fluorescent probes. In the present study the interaction of Ž strept. avidin with fluorescein–biotin and biotin–PEG800 –pyrene was examined more closely, aiming at a new fluorimetric assay for Ž strept. avidin that could compete with radioligand and enzyme assays in practice. This goal was successfully reached: Both fluorescein–biotin and biotin–PEG800 –pyrene were suitable for convenient, rapid, and accurate measurement of Žstrept.avidin in crude biosamples, performance being far superior to previous fluorimetric assays. Fortunately, the commercial reagent fluorescein– biotin proved more versatile than biotin–PEG800 – pyrene: Fluorescein–biotin was advantageous in every respect when measuring avidin concentrations. In the case of streptavidin the relative preference for any of the two fluorescent ligands depended on goals and assay conditions: biotin–PEG800 –pyrene gave more significant data, due to the more pronounced fluorescence effect after acceptor binding Ž compare Fig. 5B and C., but applicability was limited to G 10 nM streptavidin and to absence of serum. Fluorescein–biotin titration of Žstrept.avidin is the first fluorimetric assay that compares well with the

most advanced enzymatic method w5x. Accuracy and sensitivity are similar—only when using overnight incubations the enzyme assay is 10 times more sensitive. Most importantly, the new fluorimetric assay is equally reliable and robust, as exemplified by unperturbed measurement of avidin and streptavidin in reddish serum by the standard protocol. Satisfactory performance is accompanied by several advantages over radioligand and enzyme assays: Nothing but a standardized stock solution of fluorescein–biotin is required, standardization is simple and valid for G 1 week Ždepending on storage conditions., costs are minimal, the simple protocol guarantees correct results at first try, and most importantly, assay times are usually rather short. Admittedly, the manual titration protocol is not applicable to a large number of samples. However, a similar through-put as in radioligand and enzyme assays could be achieved by use of microplates, automated pipetting, and automated fluorescence reading. 4.2. Mechanism of fluorescence quenching Titration of avidin with fluorescein–biotin or biotin–PEG800 –pyrene always gave a ‘hump’ region, followed by a steep linear fluorescence increase when ) 4 fluorescent biotins per avidin tetramer were present in the solution ŽFigs. 1B, 2A, 3A, B, obscured by compressed plots and lower precision in Fig. 4.. With biotin–PEG800 –pyrene the same type of titration curve was also obtained for streptavidin Ž see Fig. 5B. as for avidin ŽFig. 2. . The initial slope at F 0.5 ligands per tetramer Žcircles in Figs. 1B and 2A. always reflected the fluorescence yield of 1:1 complexes. The moderate fluorescence quenching in such 1:1 complexes Ž as compared to free fluorophore–biotin conjugates, dotted lines in Figs. 1B and 2A. is obviously due to some fluorophore–avidin contact. The ‘hump’ regions correspond to progressive decrease of normalized fluorescence yield Žfluorescence in presencerabsence of avidin. which accompanies specific binding of several fluorophores to one avidin tetramer. Mutual quenching of pyrene residues obviously is caused by excimer formation ŽFig. 2B. , while fluorescence resonance energy transfer can safely be assumed to be responsible for m utual quenching of Žstrept.avidin-bound fluorescein residues.

H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

The linear rise in fluorescence at ) 4 ligands per avidin tetramer was always strictly parallel to the linear titration profile in the absence of avidin Ž Figs. 1B and 2A, dotted lines., as expected for absence of nonspecific interaction. Moreover, the linear titration curve of avidin which had been specifically blocked with excess of D-biotin ŽFig. 1B, triangles. was indistinguishable from the corresponding control without avidin which further excludes nonspecific interaction. In conclusion, the ‘hump’ in the fluorescence titration profiles of avidin reflected progressive quenching of specifically bound fluorescent biotins while the steep linear rise at ) 4 ligands per avidin tetramer was due to completely unbound ligands which are in excess. Thus the breakpoint unequivocally corresponded to saturation of specific acceptor sites with fluorescent biotins. The anomalous fluorescein–biotin titration profiles of streptavidin can well be understood on the basis of known pairwise arrangement of the four biotin-binding sites on opposite sides of the tetrameric acceptor protein w27,28x, taking into account geometric differences between avidin and streptavidin w27–31x, and assuming a pseudo-specific interaction of fluorescein residues with empty biotin-binding sites in addition w12–16x. In each pair of streptavidin’s biotin-binding sites the two adjacent sites are sufficiently close to simultaneously bind the two biotin end groups of N, N Xbis-biotinoyl-1,12-diaminododecane w27,28x. Thus in a situation with one specifically bound fluorescein– biotin per pair of sites the 14-atom spacer of fluorescein–biotin allows for contact between the flexibly linked fluorescein residue and the adjacent empty biotin-binding pocket in streptavidin. Moreover fluorescein residues covalently linked to Ž strept. avidin are known to bind to empty biotin-binding sites, to fluoresce poorly when bound, and to be displaced upon specific binding of D-biotin, with a concomitant 100–300% increase in fluorescence yield w12–16x. As predicted on the basis of the above arguments, we indeed observed strong fluorescence quenching in 2:1 complexes of fluorescein–biotin with streptavidin and partial recovery of fluorescence yield upon binding of another two fluorescein–biotin molecules per streptavidin tetramer ŽFig. 5C, squares.. The partial occlusion of an empty biotin-binding site by the fluorescein moiety of an adjacently bound fluorescein–bio-


tin also explains why 2:1 complexes with streptavidin were formed on a much shorter time scale Ž Fig. 5C, circles. than 4:1 complexes Ž Fig. 5C, squares.. Absence of a similar interaction in 2:1 complexes between fluorescein–biotin and aÕidin must be attributed to the much longer effectiÕe pocket–pocket distance in avidin which could only be spanned by bis-biotin compounds with G 21 spacer atoms w27,28x. More recent crystallographic studies showed that the longer effectiÕe distance in avidin is due to ‘burial’ of the biotin-binding sites by peripheral loops and side chains which does not occur in streptavidin, whereas the linear distances of adjacently bound biotins are the same in avidin and streptavidin w29– 31x. These findings, therefore provide a rationale for synthesizing shorter fluorescein–biotin analogues which are confined to single biotin-binding sites in both avidin and streptavidin, offering new advantages for practical application.

Acknowledgements This work was supported by the Austrian Research Funds Žproject S-6607. .

References w1x M. Wilchek, E.A. Bayer, Methods Enzymol. 184 Ž1990. 5–13. w2x M. Wilchek, E.A. Bayer, Methods Enzymol. 184 Ž1990. 14–45. w3x E.V. Groman, J.M. Rothenberg, E.A. Bayer, M. Wilchek, Methods Enzymol. 184 Ž1990. 208–217. w4x D.M. Mock, D.B. DuBois, Anal. Biochem. 153 Ž1986. 272–278. w5x E.A. Bayer, H. Ben-Hur, M. Wilchek, Methods Enzymol. 184 Ž1990. 217–223. w6x R.S. Niedbala, F. Gergits III, K.J. Schray, J. Biochem. Biophys. Methods 13 Ž1986. 205–210. w7x M.S. Barbarakis, W.G. Qaisi, S. Daunert, L.G. Bachas, Anal. Chem. 65 Ž1993. 457–460. w8x N.M. Green, Methods Enzymol. 18A Ž1970. 418–424. w9x H.J. Lin, J.F. Kirsch, Methods Enzymol. 62 Ž1979. 287–289. w10x G.P. Kurzban, G. Gitlin, E.A. Bayer, M. Wilchek, P.M. Horowitz, J. Protein Chem. 9 Ž1990. 673–682. w11x D.M. Mock, P. Horowitz, Methods Enzymol. 184 Ž1990. 234–240. w12x M.S. Barbarakis, T. Smith-Palmer, L.G. Bachas, Talanta 40 Ž1993. 1139–1145.


H.J. Gruber et al.r Biochimica et Biophysica Acta 1381 (1998) 203–212

w13x N.G. Hentz, L.G. Bachas, Anal. Chem. 67 Ž1995. 1014– 1018. w14x M.H.H. Al-Hakiem, J. Landon, D.S. Smith, R.D. Nargessi, Anal. Biochem. 116 Ž1981. 264–267. w15x R.D. Nargessi, D.S. Smith, Methods Enzymol. 122 Ž1986. 67–72. w16x S.V. Rao, K.W. Anderson, L.G. Bachas, Bioconjugate Chem. 8 Ž1997. 94–98. w17x K.J. Schray, P.G. Artz, R.C. Hevey, Anal. Chem. 60 Ž1988. 853–855. w18x H.J. Gruber, M. Marek, H. Schindler, K. Kaiser, Bioconjugate Chem. 8 Ž1997. 552–559. w19x M. Marek, K. Kaiser, H.J. Gruber, Bioconjugate Chem. 8 Ž1997. 560–566. w20x T. Schmidt, G.J. Schutz, ¨ H.J. Gruber, H. Schindler, Anal. Chem. 68 Ž1996. 4397–4401. w21x E. Harlow, D. Lane, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, New York, 1988, p. 119.

w22x H.-J. Galla, W. Hartmann, Chem. Phys. Lipids 27 Ž1980. 199–219. w23x Haugland, R.P., Handbook of Fluorescent Probes and Research Chemicals, 1992. w24x P.W. Chan, K. Bartlett, Clin. Chim. Acta 159 Ž1986. 185– 196. w25x M.M. Bradford, Anal. Biochem. 72 Ž1976. 248–254. w26x B. Fudem-Goldin, G.A. Orr, Methods Enzymol. 184 Ž1990. 167–173. w27x N.M. Green, Adv. Protein Chem. 29 Ž1975. 85–133. w28x N.M. Green, Methods Enzymol. 184 Ž1990. 51–67. w29x W.A. Hendrickson, A. Pahler, J.L. Smith, Y. Satow, E.A. ¨ Merritt, R.P. Phizackerley, Proc. Natl. Acad. Sci. U.S.A. 86 Ž1989. 2190–2194. w30x O. Livnah, E.A. Bayer, M. Wilchek, J.L. Sussman, Proc. Natl. Acad. Sci. U.S.A. 90 Ž1993. 5076–5080. w31x L. Pugliese, A. Coda, M. Malcovati, M. Bolognesi, J. Mol. Biol. 231 Ž1993. 698–710.